diff --git a/babylm_hop_control_10M_seed0/runs/checkpoint-1600/config.json b/babylm_hop_control_10M_seed0/runs/checkpoint-1600/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fb0f9a53a507ffcdb60410deb71d60ef801bf350 --- /dev/null +++ b/babylm_hop_control_10M_seed0/runs/checkpoint-1600/config.json @@ -0,0 +1,36 @@ +{ + "_name_or_path": "meta-llama/Llama-3.2-3B", + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "eos_token_id": 128001, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 3072, + "initializer_range": 0.02, + "intermediate_size": 8192, + "max_position_embeddings": 131072, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 24, + "num_hidden_layers": 28, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": { + "factor": 32.0, + "high_freq_factor": 4.0, + "low_freq_factor": 1.0, + "original_max_position_embeddings": 8192, + "rope_type": "llama3" + }, + "rope_theta": 500000.0, + "tie_word_embeddings": true, + "torch_dtype": "float16", + "transformers_version": "4.45.1", + "use_cache": true, + "vocab_size": 128256 +} diff --git a/babylm_hop_control_10M_seed0/runs/checkpoint-1600/generation_config.json b/babylm_hop_control_10M_seed0/runs/checkpoint-1600/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..9c389b841a9a29ddd904e02d1eb0e08dcce82ad9 --- /dev/null +++ b/babylm_hop_control_10M_seed0/runs/checkpoint-1600/generation_config.json @@ -0,0 +1,9 @@ +{ + "_from_model_config": true, + "bos_token_id": 128000, + "do_sample": true, + "eos_token_id": 128001, + "temperature": 0.6, + "top_p": 0.9, + "transformers_version": "4.45.1" +} diff --git a/babylm_hop_control_10M_seed0/runs/checkpoint-1600/latest b/babylm_hop_control_10M_seed0/runs/checkpoint-1600/latest new file mode 100644 index 0000000000000000000000000000000000000000..10b09fac99bc80ff931649e8b3378aab683b28be --- /dev/null +++ b/babylm_hop_control_10M_seed0/runs/checkpoint-1600/latest @@ -0,0 +1 @@ +global_step1600 \ No newline at end of file diff --git a/babylm_hop_control_10M_seed0/runs/checkpoint-2100/config.json b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fb0f9a53a507ffcdb60410deb71d60ef801bf350 --- /dev/null +++ b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/config.json @@ -0,0 +1,36 @@ +{ + "_name_or_path": "meta-llama/Llama-3.2-3B", + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "eos_token_id": 128001, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 3072, + "initializer_range": 0.02, + "intermediate_size": 8192, + "max_position_embeddings": 131072, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 24, + "num_hidden_layers": 28, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": { + "factor": 32.0, + "high_freq_factor": 4.0, + "low_freq_factor": 1.0, + "original_max_position_embeddings": 8192, + "rope_type": "llama3" + }, + "rope_theta": 500000.0, + "tie_word_embeddings": true, + "torch_dtype": "float16", + "transformers_version": "4.45.1", + "use_cache": true, + "vocab_size": 128256 +} diff --git a/babylm_hop_control_10M_seed0/runs/checkpoint-2100/generation_config.json b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..9c389b841a9a29ddd904e02d1eb0e08dcce82ad9 --- /dev/null +++ b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/generation_config.json @@ -0,0 +1,9 @@ +{ + "_from_model_config": true, + "bos_token_id": 128000, + "do_sample": true, + "eos_token_id": 128001, + "temperature": 0.6, + "top_p": 0.9, + "transformers_version": "4.45.1" +} diff --git a/babylm_hop_control_10M_seed0/runs/checkpoint-2100/latest b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/latest new file mode 100644 index 0000000000000000000000000000000000000000..b05c33b7aff64e27ba9c8b2e7dc2454bea58ee7f --- /dev/null +++ b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/latest @@ -0,0 +1 @@ +global_step2100 \ No newline at end of file diff --git a/babylm_hop_control_10M_seed0/runs/checkpoint-2100/model.safetensors.index.json b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..ed64de846d720b9a7859dc20575fea8e8ca51940 --- /dev/null +++ b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/model.safetensors.index.json @@ -0,0 +1,262 @@ +{ + "metadata": { + "total_size": 7213504512 + }, + "weight_map": { + "lm_head.weight": "model-00002-of-00002.safetensors", + "model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.norm.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/babylm_hop_control_10M_seed0/runs/checkpoint-2100/special_tokens_map.json b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..6949c5975ee0e961ef61cf31010dce04df0a03f8 --- /dev/null +++ b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/special_tokens_map.json @@ -0,0 +1,23 @@ +{ + "bos_token": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/babylm_hop_control_10M_seed0/runs/checkpoint-2100/tokenizer_config.json b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..8991b49e9c2a43fc527dab9e09ad8171f0cc5943 --- /dev/null +++ b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/tokenizer_config.json @@ -0,0 +1,2086 @@ +{ + "added_tokens_decoder": { + "128000": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128001": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128002": { + "content": "<|reserved_special_token_0|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128003": { + "content": "<|reserved_special_token_1|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128004": { + "content": "<|finetune_right_pad_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128005": { + "content": "<|reserved_special_token_2|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128006": { + "content": "<|start_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128007": { + "content": "<|end_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128008": { + "content": "<|eom_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128009": { + "content": "<|eot_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128010": { + "content": "<|python_tag|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128011": { + "content": "<|reserved_special_token_3|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128012": { + "content": "<|reserved_special_token_4|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128013": { + "content": "<|reserved_special_token_5|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128014": { + "content": "<|reserved_special_token_6|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128015": { + "content": "<|reserved_special_token_7|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128016": { + "content": "<|reserved_special_token_8|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128017": { + "content": "<|reserved_special_token_9|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128018": { + "content": "<|reserved_special_token_10|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128019": { + "content": "<|reserved_special_token_11|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128020": { + "content": "<|reserved_special_token_12|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128021": { + "content": "<|reserved_special_token_13|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128022": { + "content": "<|reserved_special_token_14|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128023": { + "content": "<|reserved_special_token_15|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128024": { + "content": "<|reserved_special_token_16|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128025": { + "content": "<|reserved_special_token_17|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128026": { + "content": "<|reserved_special_token_18|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128027": { + "content": "<|reserved_special_token_19|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128028": { + "content": "<|reserved_special_token_20|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128029": { + "content": "<|reserved_special_token_21|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128030": { + "content": "<|reserved_special_token_22|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128031": { + "content": "<|reserved_special_token_23|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128032": { + "content": "<|reserved_special_token_24|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128033": { + "content": "<|reserved_special_token_25|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128034": { + "content": "<|reserved_special_token_26|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128035": { + "content": "<|reserved_special_token_27|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128036": { + "content": "<|reserved_special_token_28|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128037": { + "content": "<|reserved_special_token_29|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128038": { + "content": "<|reserved_special_token_30|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128039": { + "content": "<|reserved_special_token_31|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128040": { + "content": "<|reserved_special_token_32|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128041": { + "content": "<|reserved_special_token_33|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128042": { + "content": "<|reserved_special_token_34|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128043": { + "content": "<|reserved_special_token_35|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128044": { + "content": "<|reserved_special_token_36|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128045": { + "content": "<|reserved_special_token_37|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128046": { + "content": "<|reserved_special_token_38|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128047": { + "content": "<|reserved_special_token_39|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128048": { + "content": "<|reserved_special_token_40|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128049": { + "content": "<|reserved_special_token_41|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128050": { + "content": "<|reserved_special_token_42|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128051": { + "content": "<|reserved_special_token_43|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128052": { + "content": "<|reserved_special_token_44|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128053": { + "content": "<|reserved_special_token_45|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128054": { + "content": "<|reserved_special_token_46|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128055": { + "content": "<|reserved_special_token_47|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128056": { + "content": "<|reserved_special_token_48|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128057": { + "content": "<|reserved_special_token_49|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128058": { + "content": "<|reserved_special_token_50|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128059": { + "content": "<|reserved_special_token_51|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128060": { + "content": "<|reserved_special_token_52|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128061": { + "content": "<|reserved_special_token_53|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128062": { + "content": "<|reserved_special_token_54|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128063": { + "content": "<|reserved_special_token_55|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128064": { + "content": "<|reserved_special_token_56|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128065": { + "content": "<|reserved_special_token_57|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128066": { + "content": "<|reserved_special_token_58|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128067": { + "content": "<|reserved_special_token_59|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128068": { + "content": "<|reserved_special_token_60|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128069": { + "content": "<|reserved_special_token_61|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128070": { + "content": "<|reserved_special_token_62|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128071": { + "content": "<|reserved_special_token_63|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128072": { + "content": "<|reserved_special_token_64|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128073": { + "content": "<|reserved_special_token_65|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128074": { + "content": "<|reserved_special_token_66|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128075": { + "content": "<|reserved_special_token_67|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128076": { + "content": "<|reserved_special_token_68|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128077": { + "content": "<|reserved_special_token_69|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128078": { + "content": "<|reserved_special_token_70|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128079": { + "content": "<|reserved_special_token_71|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128080": { + "content": "<|reserved_special_token_72|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128081": { + "content": "<|reserved_special_token_73|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128082": { + "content": "<|reserved_special_token_74|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128083": { + "content": "<|reserved_special_token_75|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128084": { + "content": "<|reserved_special_token_76|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128085": { + "content": "<|reserved_special_token_77|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128086": { + "content": "<|reserved_special_token_78|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128087": { + "content": "<|reserved_special_token_79|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128088": { + "content": "<|reserved_special_token_80|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128089": { + "content": "<|reserved_special_token_81|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128090": { + "content": "<|reserved_special_token_82|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128091": { + "content": "<|reserved_special_token_83|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128092": { + "content": "<|reserved_special_token_84|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128093": { + "content": "<|reserved_special_token_85|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128094": { + "content": "<|reserved_special_token_86|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128095": { + "content": "<|reserved_special_token_87|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128096": { + "content": "<|reserved_special_token_88|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128097": { + "content": "<|reserved_special_token_89|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128098": { + "content": "<|reserved_special_token_90|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128099": { + "content": "<|reserved_special_token_91|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128100": { + "content": "<|reserved_special_token_92|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128101": { + "content": "<|reserved_special_token_93|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128102": { + "content": "<|reserved_special_token_94|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128103": { + "content": "<|reserved_special_token_95|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128104": { + "content": "<|reserved_special_token_96|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128105": { + "content": "<|reserved_special_token_97|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128106": { + "content": "<|reserved_special_token_98|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128107": { + "content": "<|reserved_special_token_99|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128108": { + "content": "<|reserved_special_token_100|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128109": { + "content": "<|reserved_special_token_101|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128110": { + "content": "<|reserved_special_token_102|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128111": { + "content": "<|reserved_special_token_103|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128112": { + "content": "<|reserved_special_token_104|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128113": { + "content": "<|reserved_special_token_105|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128114": { + "content": "<|reserved_special_token_106|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128115": { + "content": "<|reserved_special_token_107|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128116": { + "content": "<|reserved_special_token_108|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128117": { + "content": "<|reserved_special_token_109|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128118": { + "content": "<|reserved_special_token_110|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128119": { + "content": "<|reserved_special_token_111|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128120": { + "content": "<|reserved_special_token_112|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128121": { + "content": "<|reserved_special_token_113|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128122": { + "content": "<|reserved_special_token_114|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128123": { + "content": "<|reserved_special_token_115|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128124": { + "content": "<|reserved_special_token_116|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128125": { + "content": "<|reserved_special_token_117|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128126": { + "content": "<|reserved_special_token_118|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128127": { + "content": "<|reserved_special_token_119|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128128": { + "content": "<|reserved_special_token_120|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128129": { + "content": "<|reserved_special_token_121|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128130": { + "content": "<|reserved_special_token_122|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128131": { + "content": "<|reserved_special_token_123|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128132": { + "content": "<|reserved_special_token_124|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128133": { + "content": "<|reserved_special_token_125|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128134": { + "content": "<|reserved_special_token_126|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128135": { + "content": "<|reserved_special_token_127|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128136": { + "content": "<|reserved_special_token_128|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128137": { + "content": "<|reserved_special_token_129|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128138": { + "content": "<|reserved_special_token_130|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128139": { + "content": "<|reserved_special_token_131|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128140": { + "content": "<|reserved_special_token_132|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128141": { + "content": "<|reserved_special_token_133|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128142": { + "content": "<|reserved_special_token_134|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128143": { + "content": "<|reserved_special_token_135|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128144": { + "content": "<|reserved_special_token_136|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128145": { + "content": "<|reserved_special_token_137|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128146": { + "content": "<|reserved_special_token_138|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128147": { + "content": "<|reserved_special_token_139|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128148": { + "content": "<|reserved_special_token_140|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128149": { + "content": "<|reserved_special_token_141|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128150": { + "content": "<|reserved_special_token_142|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128151": { + "content": "<|reserved_special_token_143|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128152": { + "content": "<|reserved_special_token_144|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128153": { + "content": "<|reserved_special_token_145|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128154": { + "content": "<|reserved_special_token_146|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128155": { + "content": "<|reserved_special_token_147|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128156": { + "content": "<|reserved_special_token_148|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128157": { + "content": "<|reserved_special_token_149|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128158": { + "content": "<|reserved_special_token_150|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128159": { + "content": "<|reserved_special_token_151|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128160": { + "content": "<|reserved_special_token_152|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128161": { + "content": "<|reserved_special_token_153|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128162": { + "content": "<|reserved_special_token_154|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128163": { + "content": "<|reserved_special_token_155|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128164": { + "content": "<|reserved_special_token_156|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128165": { + "content": "<|reserved_special_token_157|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128166": { + "content": "<|reserved_special_token_158|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128167": { + "content": "<|reserved_special_token_159|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128168": { + "content": "<|reserved_special_token_160|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128169": { + "content": "<|reserved_special_token_161|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128170": { + "content": "<|reserved_special_token_162|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128171": { + "content": "<|reserved_special_token_163|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128172": { + "content": "<|reserved_special_token_164|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128173": { + "content": "<|reserved_special_token_165|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128174": { + "content": "<|reserved_special_token_166|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128175": { + "content": "<|reserved_special_token_167|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128176": { + "content": "<|reserved_special_token_168|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128177": { + "content": "<|reserved_special_token_169|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128178": { + "content": "<|reserved_special_token_170|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128179": { + "content": "<|reserved_special_token_171|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128180": { + "content": "<|reserved_special_token_172|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128181": { + "content": "<|reserved_special_token_173|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128182": { + "content": "<|reserved_special_token_174|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128183": { + "content": "<|reserved_special_token_175|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128184": { + "content": "<|reserved_special_token_176|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128185": { + "content": "<|reserved_special_token_177|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128186": { + "content": "<|reserved_special_token_178|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128187": { + "content": "<|reserved_special_token_179|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128188": { + "content": "<|reserved_special_token_180|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128189": { + "content": "<|reserved_special_token_181|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128190": { + "content": "<|reserved_special_token_182|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128191": { + "content": "<|reserved_special_token_183|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128192": { + "content": "<|reserved_special_token_184|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128193": { + "content": "<|reserved_special_token_185|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128194": { + "content": "<|reserved_special_token_186|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128195": { + "content": "<|reserved_special_token_187|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128196": { + "content": "<|reserved_special_token_188|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128197": { + "content": "<|reserved_special_token_189|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128198": { + "content": "<|reserved_special_token_190|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128199": { + "content": "<|reserved_special_token_191|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128200": { + "content": "<|reserved_special_token_192|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128201": { + "content": "<|reserved_special_token_193|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128202": { + "content": "<|reserved_special_token_194|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128203": { + "content": "<|reserved_special_token_195|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128204": { + "content": "<|reserved_special_token_196|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128205": { + "content": "<|reserved_special_token_197|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128206": { + "content": "<|reserved_special_token_198|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128207": { + "content": "<|reserved_special_token_199|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128208": { + "content": "<|reserved_special_token_200|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128209": { + "content": "<|reserved_special_token_201|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128210": { + "content": "<|reserved_special_token_202|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128211": { + "content": "<|reserved_special_token_203|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128212": { + "content": "<|reserved_special_token_204|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128213": { + "content": "<|reserved_special_token_205|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128214": { + "content": "<|reserved_special_token_206|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128215": { + "content": "<|reserved_special_token_207|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128216": { + "content": "<|reserved_special_token_208|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128217": { + "content": "<|reserved_special_token_209|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128218": { + "content": "<|reserved_special_token_210|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128219": { + "content": "<|reserved_special_token_211|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128220": { + "content": "<|reserved_special_token_212|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128221": { + "content": "<|reserved_special_token_213|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128222": { + "content": "<|reserved_special_token_214|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128223": { + "content": "<|reserved_special_token_215|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128224": { + "content": "<|reserved_special_token_216|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128225": { + "content": "<|reserved_special_token_217|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128226": { + "content": "<|reserved_special_token_218|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128227": { + "content": "<|reserved_special_token_219|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128228": { + "content": "<|reserved_special_token_220|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128229": { + "content": "<|reserved_special_token_221|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128230": { + "content": "<|reserved_special_token_222|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128231": { + "content": "<|reserved_special_token_223|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128232": { + "content": "<|reserved_special_token_224|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128233": { + "content": "<|reserved_special_token_225|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128234": { + "content": "<|reserved_special_token_226|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128235": { + "content": "<|reserved_special_token_227|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128236": { + "content": "<|reserved_special_token_228|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128237": { + "content": "<|reserved_special_token_229|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128238": { + "content": "<|reserved_special_token_230|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128239": { + "content": "<|reserved_special_token_231|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128240": { + "content": "<|reserved_special_token_232|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128241": { + "content": "<|reserved_special_token_233|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128242": { + "content": "<|reserved_special_token_234|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128243": { + "content": "<|reserved_special_token_235|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128244": { + "content": "<|reserved_special_token_236|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128245": { + "content": "<|reserved_special_token_237|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128246": { + "content": "<|reserved_special_token_238|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128247": { + "content": "<|reserved_special_token_239|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128248": { + "content": "<|reserved_special_token_240|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128249": { + "content": "<|reserved_special_token_241|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128250": { + "content": "<|reserved_special_token_242|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128251": { + "content": "<|reserved_special_token_243|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128252": { + "content": "<|reserved_special_token_244|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128253": { + "content": "<|reserved_special_token_245|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128254": { + "content": "<|reserved_special_token_246|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128255": { + "content": "<|reserved_special_token_247|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128256": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128257": { + "content": "🅂", + "lstrip": true, + "normalized": true, + "rstrip": false, + "single_word": false, + "special": false + }, + "128258": { + "content": "🄿", + "lstrip": true, + "normalized": true, + "rstrip": false, + "single_word": false, + "special": false + } + }, + "bos_token": "<|begin_of_text|>", + "clean_up_tokenization_spaces": true, + "eos_token": "<|end_of_text|>", + "model_input_names": [ + "input_ids", + "attention_mask" + ], + "model_max_length": 131072, + "pad_token": "[PAD]", + "tokenizer_class": "PreTrainedTokenizerFast" +} diff --git a/babylm_hop_control_10M_seed0/runs/checkpoint-2100/trainer_state.json b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..0e8ee721bc9aacd358206ab4df72c5e433e8cd9a --- /dev/null +++ b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/trainer_state.json @@ -0,0 +1,16413 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 2.613565650280025, + "eval_steps": 10, + "global_step": 2100, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0012445550715619166, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8073, + "step": 1 + }, + { + "epoch": 0.002489110143123833, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8285, + "step": 2 + }, + { + "epoch": 0.00373366521468575, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7975, + "step": 3 + }, + { + "epoch": 0.004978220286247666, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.798, + "step": 4 + }, + { + "epoch": 0.006222775357809583, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8443, + "step": 5 + }, + { + "epoch": 0.0074673304293715, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8074, + "step": 6 + }, + { + "epoch": 0.008711885500933417, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7588, + "step": 7 + }, + { + "epoch": 0.009956440572495333, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8136, + "step": 8 + }, + { + "epoch": 0.01120099564405725, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7397, + "step": 9 + }, + { + "epoch": 0.012445550715619166, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7745, + "step": 10 + }, + { + "epoch": 0.012445550715619166, + "eval_loss": 2.8210322856903076, + "eval_runtime": 43.0476, + "eval_samples_per_second": 23.23, + "eval_steps_per_second": 0.976, + "step": 10 + }, + { + "epoch": 0.013690105787181083, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8275, + "step": 11 + }, + { + "epoch": 0.014934660858743, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7762, + "step": 12 + }, + { + "epoch": 0.016179215930304917, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7799, + "step": 13 + }, + { + "epoch": 0.017423771001866834, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7906, + "step": 14 + }, + { + "epoch": 0.018668326073428748, + "grad_norm": 5.01304292678833, + "learning_rate": 2.0746887966804982e-08, + "loss": 2.7712, + "step": 15 + }, + { + "epoch": 0.019912881144990666, + "grad_norm": 4.976474761962891, + "learning_rate": 4.1493775933609963e-08, + "loss": 2.829, + "step": 16 + }, + { + "epoch": 0.021157436216552583, + "grad_norm": 5.539012908935547, + "learning_rate": 6.224066390041494e-08, + "loss": 2.813, + "step": 17 + }, + { + "epoch": 0.0224019912881145, + "grad_norm": 5.4900407791137695, + "learning_rate": 8.298755186721993e-08, + "loss": 2.8126, + "step": 18 + }, + { + "epoch": 0.023646546359676415, + "grad_norm": 5.457218647003174, + "learning_rate": 1.037344398340249e-07, + "loss": 2.7482, + "step": 19 + }, + { + "epoch": 0.024891101431238332, + "grad_norm": 5.231036186218262, + "learning_rate": 1.2448132780082988e-07, + "loss": 2.8323, + "step": 20 + }, + { + "epoch": 0.024891101431238332, + "eval_loss": 2.819715738296509, + "eval_runtime": 43.9884, + "eval_samples_per_second": 22.733, + "eval_steps_per_second": 0.955, + "step": 20 + }, + { + "epoch": 0.02613565650280025, + "grad_norm": 5.337130546569824, + "learning_rate": 1.4522821576763488e-07, + "loss": 2.776, + "step": 21 + }, + { + "epoch": 0.027380211574362167, + "grad_norm": 4.973108291625977, + "learning_rate": 1.6597510373443985e-07, + "loss": 2.8239, + "step": 22 + }, + { + "epoch": 0.02862476664592408, + "grad_norm": 4.8733906745910645, + "learning_rate": 1.8672199170124483e-07, + "loss": 2.8069, + "step": 23 + }, + { + "epoch": 0.029869321717486, + "grad_norm": 5.061434745788574, + "learning_rate": 2.074688796680498e-07, + "loss": 2.8063, + "step": 24 + }, + { + "epoch": 0.031113876789047916, + "grad_norm": 4.782018661499023, + "learning_rate": 2.2821576763485478e-07, + "loss": 2.7488, + "step": 25 + }, + { + "epoch": 0.03235843186060983, + "grad_norm": 4.394975662231445, + "learning_rate": 2.4896265560165975e-07, + "loss": 2.7683, + "step": 26 + }, + { + "epoch": 0.03360298693217175, + "grad_norm": 4.475763320922852, + "learning_rate": 2.6970954356846476e-07, + "loss": 2.8143, + "step": 27 + }, + { + "epoch": 0.03484754200373367, + "grad_norm": 4.362309455871582, + "learning_rate": 2.9045643153526976e-07, + "loss": 2.7875, + "step": 28 + }, + { + "epoch": 0.03609209707529558, + "grad_norm": 3.9444823265075684, + "learning_rate": 3.112033195020747e-07, + "loss": 2.766, + "step": 29 + }, + { + "epoch": 0.037336652146857496, + "grad_norm": 3.4614129066467285, + "learning_rate": 3.319502074688797e-07, + "loss": 2.7644, + "step": 30 + }, + { + "epoch": 0.037336652146857496, + "eval_loss": 2.772022008895874, + "eval_runtime": 45.8513, + "eval_samples_per_second": 21.81, + "eval_steps_per_second": 0.916, + "step": 30 + }, + { + "epoch": 0.038581207218419414, + "grad_norm": 3.449410915374756, + "learning_rate": 3.5269709543568466e-07, + "loss": 2.7653, + "step": 31 + }, + { + "epoch": 0.03982576228998133, + "grad_norm": 3.3115482330322266, + "learning_rate": 3.7344398340248966e-07, + "loss": 2.7032, + "step": 32 + }, + { + "epoch": 0.04107031736154325, + "grad_norm": 3.2057440280914307, + "learning_rate": 3.941908713692946e-07, + "loss": 2.7766, + "step": 33 + }, + { + "epoch": 0.042314872433105166, + "grad_norm": 3.4503021240234375, + "learning_rate": 4.149377593360996e-07, + "loss": 2.7214, + "step": 34 + }, + { + "epoch": 0.043559427504667084, + "grad_norm": 3.1477363109588623, + "learning_rate": 4.3568464730290456e-07, + "loss": 2.7333, + "step": 35 + }, + { + "epoch": 0.044803982576229, + "grad_norm": 2.8025119304656982, + "learning_rate": 4.5643153526970956e-07, + "loss": 2.7535, + "step": 36 + }, + { + "epoch": 0.04604853764779091, + "grad_norm": 2.967703342437744, + "learning_rate": 4.771784232365145e-07, + "loss": 2.7615, + "step": 37 + }, + { + "epoch": 0.04729309271935283, + "grad_norm": 2.9308114051818848, + "learning_rate": 4.979253112033195e-07, + "loss": 2.7115, + "step": 38 + }, + { + "epoch": 0.04853764779091475, + "grad_norm": 2.9203720092773438, + "learning_rate": 5.186721991701245e-07, + "loss": 2.701, + "step": 39 + }, + { + "epoch": 0.049782202862476664, + "grad_norm": 2.8226709365844727, + "learning_rate": 5.394190871369295e-07, + "loss": 2.6267, + "step": 40 + }, + { + "epoch": 0.049782202862476664, + "eval_loss": 2.6964941024780273, + "eval_runtime": 46.5181, + "eval_samples_per_second": 21.497, + "eval_steps_per_second": 0.903, + "step": 40 + }, + { + "epoch": 0.05102675793403858, + "grad_norm": 2.7212748527526855, + "learning_rate": 5.601659751037345e-07, + "loss": 2.6505, + "step": 41 + }, + { + "epoch": 0.0522713130056005, + "grad_norm": 2.658881425857544, + "learning_rate": 5.809128630705395e-07, + "loss": 2.6655, + "step": 42 + }, + { + "epoch": 0.053515868077162417, + "grad_norm": 2.4534730911254883, + "learning_rate": 6.016597510373444e-07, + "loss": 2.7098, + "step": 43 + }, + { + "epoch": 0.054760423148724334, + "grad_norm": 2.4222075939178467, + "learning_rate": 6.224066390041494e-07, + "loss": 2.6793, + "step": 44 + }, + { + "epoch": 0.056004978220286245, + "grad_norm": 2.5824973583221436, + "learning_rate": 6.431535269709543e-07, + "loss": 2.6212, + "step": 45 + }, + { + "epoch": 0.05724953329184816, + "grad_norm": 2.5982468128204346, + "learning_rate": 6.639004149377594e-07, + "loss": 2.5901, + "step": 46 + }, + { + "epoch": 0.05849408836341008, + "grad_norm": 2.213388442993164, + "learning_rate": 6.846473029045644e-07, + "loss": 2.6156, + "step": 47 + }, + { + "epoch": 0.059738643434972, + "grad_norm": 2.097372531890869, + "learning_rate": 7.053941908713693e-07, + "loss": 2.6154, + "step": 48 + }, + { + "epoch": 0.060983198506533914, + "grad_norm": 1.9788408279418945, + "learning_rate": 7.261410788381744e-07, + "loss": 2.5516, + "step": 49 + }, + { + "epoch": 0.06222775357809583, + "grad_norm": 2.052147150039673, + "learning_rate": 7.468879668049793e-07, + "loss": 2.6128, + "step": 50 + }, + { + "epoch": 0.06222775357809583, + "eval_loss": 2.6048905849456787, + "eval_runtime": 51.2137, + "eval_samples_per_second": 19.526, + "eval_steps_per_second": 0.82, + "step": 50 + }, + { + "epoch": 0.06347230864965775, + "grad_norm": 1.8841335773468018, + "learning_rate": 7.676348547717843e-07, + "loss": 2.5617, + "step": 51 + }, + { + "epoch": 0.06471686372121967, + "grad_norm": 2.0841684341430664, + "learning_rate": 7.883817427385892e-07, + "loss": 2.5627, + "step": 52 + }, + { + "epoch": 0.06596141879278158, + "grad_norm": 1.8119730949401855, + "learning_rate": 8.091286307053943e-07, + "loss": 2.5572, + "step": 53 + }, + { + "epoch": 0.0672059738643435, + "grad_norm": 1.7168290615081787, + "learning_rate": 8.298755186721992e-07, + "loss": 2.5465, + "step": 54 + }, + { + "epoch": 0.06845052893590542, + "grad_norm": 1.7731754779815674, + "learning_rate": 8.506224066390042e-07, + "loss": 2.543, + "step": 55 + }, + { + "epoch": 0.06969508400746734, + "grad_norm": 1.9654561281204224, + "learning_rate": 8.713692946058091e-07, + "loss": 2.5547, + "step": 56 + }, + { + "epoch": 0.07093963907902924, + "grad_norm": 1.8032375574111938, + "learning_rate": 8.921161825726142e-07, + "loss": 2.5319, + "step": 57 + }, + { + "epoch": 0.07218419415059116, + "grad_norm": 1.8775644302368164, + "learning_rate": 9.128630705394191e-07, + "loss": 2.5533, + "step": 58 + }, + { + "epoch": 0.07342874922215308, + "grad_norm": 1.6019998788833618, + "learning_rate": 9.336099585062241e-07, + "loss": 2.5402, + "step": 59 + }, + { + "epoch": 0.07467330429371499, + "grad_norm": 1.7785435914993286, + "learning_rate": 9.54356846473029e-07, + "loss": 2.4704, + "step": 60 + }, + { + "epoch": 0.07467330429371499, + "eval_loss": 2.541548252105713, + "eval_runtime": 51.1601, + "eval_samples_per_second": 19.546, + "eval_steps_per_second": 0.821, + "step": 60 + }, + { + "epoch": 0.07591785936527691, + "grad_norm": 1.5377200841903687, + "learning_rate": 9.751037344398341e-07, + "loss": 2.5028, + "step": 61 + }, + { + "epoch": 0.07716241443683883, + "grad_norm": 1.7212916612625122, + "learning_rate": 9.95850622406639e-07, + "loss": 2.4621, + "step": 62 + }, + { + "epoch": 0.07840696950840075, + "grad_norm": 1.632606029510498, + "learning_rate": 1.0165975103734441e-06, + "loss": 2.4868, + "step": 63 + }, + { + "epoch": 0.07965152457996266, + "grad_norm": 1.8643872737884521, + "learning_rate": 1.037344398340249e-06, + "loss": 2.4582, + "step": 64 + }, + { + "epoch": 0.08089607965152458, + "grad_norm": 1.8329377174377441, + "learning_rate": 1.058091286307054e-06, + "loss": 2.464, + "step": 65 + }, + { + "epoch": 0.0821406347230865, + "grad_norm": 1.9488356113433838, + "learning_rate": 1.078838174273859e-06, + "loss": 2.4275, + "step": 66 + }, + { + "epoch": 0.08338518979464841, + "grad_norm": 1.7663215398788452, + "learning_rate": 1.099585062240664e-06, + "loss": 2.516, + "step": 67 + }, + { + "epoch": 0.08462974486621033, + "grad_norm": 1.7675347328186035, + "learning_rate": 1.120331950207469e-06, + "loss": 2.4535, + "step": 68 + }, + { + "epoch": 0.08587429993777225, + "grad_norm": 1.7481404542922974, + "learning_rate": 1.141078838174274e-06, + "loss": 2.4089, + "step": 69 + }, + { + "epoch": 0.08711885500933417, + "grad_norm": 1.7963330745697021, + "learning_rate": 1.161825726141079e-06, + "loss": 2.4408, + "step": 70 + }, + { + "epoch": 0.08711885500933417, + "eval_loss": 2.466163396835327, + "eval_runtime": 50.4204, + "eval_samples_per_second": 19.833, + "eval_steps_per_second": 0.833, + "step": 70 + }, + { + "epoch": 0.08836341008089608, + "grad_norm": 2.053469657897949, + "learning_rate": 1.182572614107884e-06, + "loss": 2.408, + "step": 71 + }, + { + "epoch": 0.089607965152458, + "grad_norm": 1.900294303894043, + "learning_rate": 1.2033195020746888e-06, + "loss": 2.4066, + "step": 72 + }, + { + "epoch": 0.09085252022401992, + "grad_norm": 2.38637113571167, + "learning_rate": 1.224066390041494e-06, + "loss": 2.3548, + "step": 73 + }, + { + "epoch": 0.09209707529558182, + "grad_norm": 1.8274579048156738, + "learning_rate": 1.2448132780082988e-06, + "loss": 2.3844, + "step": 74 + }, + { + "epoch": 0.09334163036714374, + "grad_norm": 1.9040184020996094, + "learning_rate": 1.2655601659751037e-06, + "loss": 2.4406, + "step": 75 + }, + { + "epoch": 0.09458618543870566, + "grad_norm": 2.0351619720458984, + "learning_rate": 1.2863070539419086e-06, + "loss": 2.3624, + "step": 76 + }, + { + "epoch": 0.09583074051026758, + "grad_norm": 1.989876627922058, + "learning_rate": 1.307053941908714e-06, + "loss": 2.4097, + "step": 77 + }, + { + "epoch": 0.0970752955818295, + "grad_norm": 1.8507741689682007, + "learning_rate": 1.3278008298755188e-06, + "loss": 2.4482, + "step": 78 + }, + { + "epoch": 0.09831985065339141, + "grad_norm": 2.6118454933166504, + "learning_rate": 1.3485477178423237e-06, + "loss": 2.3178, + "step": 79 + }, + { + "epoch": 0.09956440572495333, + "grad_norm": 1.9467326402664185, + "learning_rate": 1.3692946058091288e-06, + "loss": 2.3175, + "step": 80 + }, + { + "epoch": 0.09956440572495333, + "eval_loss": 2.3719911575317383, + "eval_runtime": 52.3293, + "eval_samples_per_second": 19.11, + "eval_steps_per_second": 0.803, + "step": 80 + }, + { + "epoch": 0.10080896079651525, + "grad_norm": 2.4913370609283447, + "learning_rate": 1.3900414937759337e-06, + "loss": 2.3063, + "step": 81 + }, + { + "epoch": 0.10205351586807716, + "grad_norm": 2.4838919639587402, + "learning_rate": 1.4107883817427386e-06, + "loss": 2.3534, + "step": 82 + }, + { + "epoch": 0.10329807093963908, + "grad_norm": 2.798588991165161, + "learning_rate": 1.4315352697095435e-06, + "loss": 2.2732, + "step": 83 + }, + { + "epoch": 0.104542626011201, + "grad_norm": 2.688399076461792, + "learning_rate": 1.4522821576763488e-06, + "loss": 2.3112, + "step": 84 + }, + { + "epoch": 0.10578718108276292, + "grad_norm": 2.7367143630981445, + "learning_rate": 1.4730290456431537e-06, + "loss": 2.3055, + "step": 85 + }, + { + "epoch": 0.10703173615432483, + "grad_norm": 2.6406664848327637, + "learning_rate": 1.4937759336099586e-06, + "loss": 2.312, + "step": 86 + }, + { + "epoch": 0.10827629122588675, + "grad_norm": 2.703355312347412, + "learning_rate": 1.5145228215767635e-06, + "loss": 2.3305, + "step": 87 + }, + { + "epoch": 0.10952084629744867, + "grad_norm": 2.420084238052368, + "learning_rate": 1.5352697095435686e-06, + "loss": 2.2608, + "step": 88 + }, + { + "epoch": 0.11076540136901059, + "grad_norm": 2.999835729598999, + "learning_rate": 1.5560165975103735e-06, + "loss": 2.2389, + "step": 89 + }, + { + "epoch": 0.11200995644057249, + "grad_norm": 2.35611629486084, + "learning_rate": 1.5767634854771784e-06, + "loss": 2.2671, + "step": 90 + }, + { + "epoch": 0.11200995644057249, + "eval_loss": 2.2900290489196777, + "eval_runtime": 52.1871, + "eval_samples_per_second": 19.162, + "eval_steps_per_second": 0.805, + "step": 90 + }, + { + "epoch": 0.1132545115121344, + "grad_norm": 2.2686431407928467, + "learning_rate": 1.5975103734439833e-06, + "loss": 2.2405, + "step": 91 + }, + { + "epoch": 0.11449906658369632, + "grad_norm": 2.883517265319824, + "learning_rate": 1.6182572614107886e-06, + "loss": 2.2364, + "step": 92 + }, + { + "epoch": 0.11574362165525824, + "grad_norm": 2.4562909603118896, + "learning_rate": 1.6390041493775935e-06, + "loss": 2.2308, + "step": 93 + }, + { + "epoch": 0.11698817672682016, + "grad_norm": 2.7968456745147705, + "learning_rate": 1.6597510373443984e-06, + "loss": 2.1959, + "step": 94 + }, + { + "epoch": 0.11823273179838208, + "grad_norm": 2.8692259788513184, + "learning_rate": 1.6804979253112035e-06, + "loss": 2.215, + "step": 95 + }, + { + "epoch": 0.119477286869944, + "grad_norm": 2.8436100482940674, + "learning_rate": 1.7012448132780084e-06, + "loss": 2.1816, + "step": 96 + }, + { + "epoch": 0.12072184194150591, + "grad_norm": 2.4477386474609375, + "learning_rate": 1.7219917012448133e-06, + "loss": 2.1684, + "step": 97 + }, + { + "epoch": 0.12196639701306783, + "grad_norm": 2.610046863555908, + "learning_rate": 1.7427385892116182e-06, + "loss": 2.1656, + "step": 98 + }, + { + "epoch": 0.12321095208462975, + "grad_norm": 2.698709487915039, + "learning_rate": 1.7634854771784235e-06, + "loss": 2.2115, + "step": 99 + }, + { + "epoch": 0.12445550715619166, + "grad_norm": 2.476600408554077, + "learning_rate": 1.7842323651452284e-06, + "loss": 2.1619, + "step": 100 + }, + { + "epoch": 0.12445550715619166, + "eval_loss": 2.221313953399658, + "eval_runtime": 51.6214, + "eval_samples_per_second": 19.372, + "eval_steps_per_second": 0.814, + "step": 100 + }, + { + "epoch": 0.12570006222775357, + "grad_norm": 3.238584518432617, + "learning_rate": 1.8049792531120333e-06, + "loss": 2.1543, + "step": 101 + }, + { + "epoch": 0.1269446172993155, + "grad_norm": 2.3609964847564697, + "learning_rate": 1.8257261410788382e-06, + "loss": 2.1602, + "step": 102 + }, + { + "epoch": 0.1281891723708774, + "grad_norm": 3.3314859867095947, + "learning_rate": 1.8464730290456433e-06, + "loss": 2.1843, + "step": 103 + }, + { + "epoch": 0.12943372744243933, + "grad_norm": 2.42441725730896, + "learning_rate": 1.8672199170124482e-06, + "loss": 2.135, + "step": 104 + }, + { + "epoch": 0.13067828251400124, + "grad_norm": 3.2766177654266357, + "learning_rate": 1.8879668049792531e-06, + "loss": 2.158, + "step": 105 + }, + { + "epoch": 0.13192283758556317, + "grad_norm": 2.8646531105041504, + "learning_rate": 1.908713692946058e-06, + "loss": 2.1214, + "step": 106 + }, + { + "epoch": 0.13316739265712507, + "grad_norm": 2.8570213317871094, + "learning_rate": 1.929460580912863e-06, + "loss": 2.1345, + "step": 107 + }, + { + "epoch": 0.134411947728687, + "grad_norm": 2.5014989376068115, + "learning_rate": 1.9502074688796682e-06, + "loss": 2.1147, + "step": 108 + }, + { + "epoch": 0.1356565028002489, + "grad_norm": 2.818286895751953, + "learning_rate": 1.970954356846473e-06, + "loss": 2.0914, + "step": 109 + }, + { + "epoch": 0.13690105787181084, + "grad_norm": 3.0579800605773926, + "learning_rate": 1.991701244813278e-06, + "loss": 2.0879, + "step": 110 + }, + { + "epoch": 0.13690105787181084, + "eval_loss": 2.167081594467163, + "eval_runtime": 42.3907, + "eval_samples_per_second": 23.59, + "eval_steps_per_second": 0.991, + "step": 110 + }, + { + "epoch": 0.13814561294337274, + "grad_norm": 2.6039772033691406, + "learning_rate": 2.012448132780083e-06, + "loss": 2.1006, + "step": 111 + }, + { + "epoch": 0.13939016801493467, + "grad_norm": 3.1309447288513184, + "learning_rate": 2.0331950207468883e-06, + "loss": 2.0758, + "step": 112 + }, + { + "epoch": 0.14063472308649658, + "grad_norm": 2.6772613525390625, + "learning_rate": 2.053941908713693e-06, + "loss": 2.1542, + "step": 113 + }, + { + "epoch": 0.14187927815805848, + "grad_norm": 3.2689080238342285, + "learning_rate": 2.074688796680498e-06, + "loss": 2.0873, + "step": 114 + }, + { + "epoch": 0.1431238332296204, + "grad_norm": 2.7289857864379883, + "learning_rate": 2.095435684647303e-06, + "loss": 2.0697, + "step": 115 + }, + { + "epoch": 0.14436838830118232, + "grad_norm": 2.8635787963867188, + "learning_rate": 2.116182572614108e-06, + "loss": 2.0908, + "step": 116 + }, + { + "epoch": 0.14561294337274425, + "grad_norm": 2.805933713912964, + "learning_rate": 2.136929460580913e-06, + "loss": 2.0643, + "step": 117 + }, + { + "epoch": 0.14685749844430615, + "grad_norm": 2.562567710876465, + "learning_rate": 2.157676348547718e-06, + "loss": 2.0559, + "step": 118 + }, + { + "epoch": 0.14810205351586808, + "grad_norm": 3.000135898590088, + "learning_rate": 2.178423236514523e-06, + "loss": 2.0442, + "step": 119 + }, + { + "epoch": 0.14934660858742999, + "grad_norm": 2.6913654804229736, + "learning_rate": 2.199170124481328e-06, + "loss": 2.0551, + "step": 120 + }, + { + "epoch": 0.14934660858742999, + "eval_loss": 2.130500316619873, + "eval_runtime": 44.3089, + "eval_samples_per_second": 22.569, + "eval_steps_per_second": 0.948, + "step": 120 + }, + { + "epoch": 0.15059116365899192, + "grad_norm": 3.38159441947937, + "learning_rate": 2.219917012448133e-06, + "loss": 2.0359, + "step": 121 + }, + { + "epoch": 0.15183571873055382, + "grad_norm": 2.4632813930511475, + "learning_rate": 2.240663900414938e-06, + "loss": 2.0805, + "step": 122 + }, + { + "epoch": 0.15308027380211575, + "grad_norm": 3.2085115909576416, + "learning_rate": 2.2614107883817427e-06, + "loss": 2.0617, + "step": 123 + }, + { + "epoch": 0.15432482887367766, + "grad_norm": 2.7810094356536865, + "learning_rate": 2.282157676348548e-06, + "loss": 2.0467, + "step": 124 + }, + { + "epoch": 0.1555693839452396, + "grad_norm": 2.6023035049438477, + "learning_rate": 2.302904564315353e-06, + "loss": 2.0578, + "step": 125 + }, + { + "epoch": 0.1568139390168015, + "grad_norm": 3.073814630508423, + "learning_rate": 2.323651452282158e-06, + "loss": 2.0742, + "step": 126 + }, + { + "epoch": 0.15805849408836342, + "grad_norm": 2.622281312942505, + "learning_rate": 2.3443983402489627e-06, + "loss": 2.1668, + "step": 127 + }, + { + "epoch": 0.15930304915992533, + "grad_norm": 3.3766582012176514, + "learning_rate": 2.365145228215768e-06, + "loss": 2.0129, + "step": 128 + }, + { + "epoch": 0.16054760423148726, + "grad_norm": 3.1513423919677734, + "learning_rate": 2.385892116182573e-06, + "loss": 2.0215, + "step": 129 + }, + { + "epoch": 0.16179215930304916, + "grad_norm": 3.2852959632873535, + "learning_rate": 2.4066390041493776e-06, + "loss": 2.0557, + "step": 130 + }, + { + "epoch": 0.16179215930304916, + "eval_loss": 2.096493721008301, + "eval_runtime": 46.9447, + "eval_samples_per_second": 21.302, + "eval_steps_per_second": 0.895, + "step": 130 + }, + { + "epoch": 0.16303671437461106, + "grad_norm": 2.526399612426758, + "learning_rate": 2.4273858921161828e-06, + "loss": 2.065, + "step": 131 + }, + { + "epoch": 0.164281269446173, + "grad_norm": 3.3448667526245117, + "learning_rate": 2.448132780082988e-06, + "loss": 2.0472, + "step": 132 + }, + { + "epoch": 0.1655258245177349, + "grad_norm": 2.6260809898376465, + "learning_rate": 2.468879668049793e-06, + "loss": 1.9804, + "step": 133 + }, + { + "epoch": 0.16677037958929683, + "grad_norm": 3.6141812801361084, + "learning_rate": 2.4896265560165977e-06, + "loss": 2.0197, + "step": 134 + }, + { + "epoch": 0.16801493466085873, + "grad_norm": 2.4911234378814697, + "learning_rate": 2.5103734439834028e-06, + "loss": 1.9642, + "step": 135 + }, + { + "epoch": 0.16925948973242066, + "grad_norm": 2.928642749786377, + "learning_rate": 2.5311203319502074e-06, + "loss": 1.9489, + "step": 136 + }, + { + "epoch": 0.17050404480398257, + "grad_norm": 3.090965509414673, + "learning_rate": 2.5518672199170125e-06, + "loss": 1.994, + "step": 137 + }, + { + "epoch": 0.1717485998755445, + "grad_norm": 3.2897465229034424, + "learning_rate": 2.5726141078838172e-06, + "loss": 1.9746, + "step": 138 + }, + { + "epoch": 0.1729931549471064, + "grad_norm": 2.857083320617676, + "learning_rate": 2.5933609958506228e-06, + "loss": 1.9774, + "step": 139 + }, + { + "epoch": 0.17423771001866833, + "grad_norm": 3.803220510482788, + "learning_rate": 2.614107883817428e-06, + "loss": 1.9731, + "step": 140 + }, + { + "epoch": 0.17423771001866833, + "eval_loss": 2.064887046813965, + "eval_runtime": 44.0231, + "eval_samples_per_second": 22.715, + "eval_steps_per_second": 0.954, + "step": 140 + }, + { + "epoch": 0.17548226509023024, + "grad_norm": 2.9923534393310547, + "learning_rate": 2.6348547717842326e-06, + "loss": 1.9854, + "step": 141 + }, + { + "epoch": 0.17672682016179217, + "grad_norm": 3.3368566036224365, + "learning_rate": 2.6556016597510377e-06, + "loss": 2.0687, + "step": 142 + }, + { + "epoch": 0.17797137523335407, + "grad_norm": 3.3132379055023193, + "learning_rate": 2.6763485477178423e-06, + "loss": 1.9696, + "step": 143 + }, + { + "epoch": 0.179215930304916, + "grad_norm": 3.7914819717407227, + "learning_rate": 2.6970954356846475e-06, + "loss": 1.9926, + "step": 144 + }, + { + "epoch": 0.1804604853764779, + "grad_norm": 3.20161509513855, + "learning_rate": 2.717842323651452e-06, + "loss": 1.9193, + "step": 145 + }, + { + "epoch": 0.18170504044803984, + "grad_norm": 3.440420150756836, + "learning_rate": 2.7385892116182577e-06, + "loss": 1.991, + "step": 146 + }, + { + "epoch": 0.18294959551960174, + "grad_norm": 3.152684450149536, + "learning_rate": 2.7593360995850628e-06, + "loss": 1.9695, + "step": 147 + }, + { + "epoch": 0.18419415059116365, + "grad_norm": 3.0402464866638184, + "learning_rate": 2.7800829875518675e-06, + "loss": 1.9274, + "step": 148 + }, + { + "epoch": 0.18543870566272558, + "grad_norm": 2.778444290161133, + "learning_rate": 2.8008298755186726e-06, + "loss": 1.9198, + "step": 149 + }, + { + "epoch": 0.18668326073428748, + "grad_norm": 2.7946548461914062, + "learning_rate": 2.8215767634854773e-06, + "loss": 1.8958, + "step": 150 + }, + { + "epoch": 0.18668326073428748, + "eval_loss": 2.043288230895996, + "eval_runtime": 44.2268, + "eval_samples_per_second": 22.611, + "eval_steps_per_second": 0.95, + "step": 150 + }, + { + "epoch": 0.1879278158058494, + "grad_norm": 3.269259452819824, + "learning_rate": 2.8423236514522824e-06, + "loss": 1.9832, + "step": 151 + }, + { + "epoch": 0.18917237087741132, + "grad_norm": 3.2673771381378174, + "learning_rate": 2.863070539419087e-06, + "loss": 1.9345, + "step": 152 + }, + { + "epoch": 0.19041692594897325, + "grad_norm": 3.016599655151367, + "learning_rate": 2.883817427385892e-06, + "loss": 1.8969, + "step": 153 + }, + { + "epoch": 0.19166148102053515, + "grad_norm": 3.2771544456481934, + "learning_rate": 2.9045643153526977e-06, + "loss": 1.921, + "step": 154 + }, + { + "epoch": 0.19290603609209708, + "grad_norm": 3.008080244064331, + "learning_rate": 2.9253112033195024e-06, + "loss": 1.9367, + "step": 155 + }, + { + "epoch": 0.194150591163659, + "grad_norm": 3.2312428951263428, + "learning_rate": 2.9460580912863075e-06, + "loss": 1.9021, + "step": 156 + }, + { + "epoch": 0.19539514623522092, + "grad_norm": 3.294121503829956, + "learning_rate": 2.966804979253112e-06, + "loss": 1.9216, + "step": 157 + }, + { + "epoch": 0.19663970130678282, + "grad_norm": 2.771685838699341, + "learning_rate": 2.9875518672199173e-06, + "loss": 1.9435, + "step": 158 + }, + { + "epoch": 0.19788425637834475, + "grad_norm": 2.971971273422241, + "learning_rate": 3.008298755186722e-06, + "loss": 1.8851, + "step": 159 + }, + { + "epoch": 0.19912881144990666, + "grad_norm": 3.3144047260284424, + "learning_rate": 3.029045643153527e-06, + "loss": 1.853, + "step": 160 + }, + { + "epoch": 0.19912881144990666, + "eval_loss": 2.023491859436035, + "eval_runtime": 44.9425, + "eval_samples_per_second": 22.251, + "eval_steps_per_second": 0.935, + "step": 160 + }, + { + "epoch": 0.2003733665214686, + "grad_norm": 3.3733646869659424, + "learning_rate": 3.0497925311203326e-06, + "loss": 1.8836, + "step": 161 + }, + { + "epoch": 0.2016179215930305, + "grad_norm": 3.1801207065582275, + "learning_rate": 3.0705394190871373e-06, + "loss": 1.9438, + "step": 162 + }, + { + "epoch": 0.2028624766645924, + "grad_norm": 3.1199593544006348, + "learning_rate": 3.0912863070539424e-06, + "loss": 1.9219, + "step": 163 + }, + { + "epoch": 0.20410703173615433, + "grad_norm": 3.344089984893799, + "learning_rate": 3.112033195020747e-06, + "loss": 1.9174, + "step": 164 + }, + { + "epoch": 0.20535158680771623, + "grad_norm": 3.269702911376953, + "learning_rate": 3.132780082987552e-06, + "loss": 1.894, + "step": 165 + }, + { + "epoch": 0.20659614187927816, + "grad_norm": 3.021744966506958, + "learning_rate": 3.153526970954357e-06, + "loss": 1.8799, + "step": 166 + }, + { + "epoch": 0.20784069695084006, + "grad_norm": 3.0104167461395264, + "learning_rate": 3.174273858921162e-06, + "loss": 1.9116, + "step": 167 + }, + { + "epoch": 0.209085252022402, + "grad_norm": 3.3018341064453125, + "learning_rate": 3.1950207468879666e-06, + "loss": 1.8389, + "step": 168 + }, + { + "epoch": 0.2103298070939639, + "grad_norm": 3.0919857025146484, + "learning_rate": 3.215767634854772e-06, + "loss": 1.9522, + "step": 169 + }, + { + "epoch": 0.21157436216552583, + "grad_norm": 3.4702494144439697, + "learning_rate": 3.2365145228215773e-06, + "loss": 1.9204, + "step": 170 + }, + { + "epoch": 0.21157436216552583, + "eval_loss": 1.997478723526001, + "eval_runtime": 43.2143, + "eval_samples_per_second": 23.14, + "eval_steps_per_second": 0.972, + "step": 170 + }, + { + "epoch": 0.21281891723708773, + "grad_norm": 3.133046865463257, + "learning_rate": 3.257261410788382e-06, + "loss": 1.891, + "step": 171 + }, + { + "epoch": 0.21406347230864967, + "grad_norm": 3.1828863620758057, + "learning_rate": 3.278008298755187e-06, + "loss": 1.8816, + "step": 172 + }, + { + "epoch": 0.21530802738021157, + "grad_norm": 3.374898910522461, + "learning_rate": 3.2987551867219918e-06, + "loss": 1.8886, + "step": 173 + }, + { + "epoch": 0.2165525824517735, + "grad_norm": 3.6660516262054443, + "learning_rate": 3.319502074688797e-06, + "loss": 1.9442, + "step": 174 + }, + { + "epoch": 0.2177971375233354, + "grad_norm": 3.080636978149414, + "learning_rate": 3.3402489626556016e-06, + "loss": 1.866, + "step": 175 + }, + { + "epoch": 0.21904169259489734, + "grad_norm": 3.3684473037719727, + "learning_rate": 3.360995850622407e-06, + "loss": 1.8699, + "step": 176 + }, + { + "epoch": 0.22028624766645924, + "grad_norm": 3.818382740020752, + "learning_rate": 3.381742738589212e-06, + "loss": 1.8541, + "step": 177 + }, + { + "epoch": 0.22153080273802117, + "grad_norm": 3.0995497703552246, + "learning_rate": 3.402489626556017e-06, + "loss": 1.8453, + "step": 178 + }, + { + "epoch": 0.22277535780958307, + "grad_norm": 3.5696587562561035, + "learning_rate": 3.423236514522822e-06, + "loss": 1.9289, + "step": 179 + }, + { + "epoch": 0.22401991288114498, + "grad_norm": 3.480571985244751, + "learning_rate": 3.4439834024896267e-06, + "loss": 1.8175, + "step": 180 + }, + { + "epoch": 0.22401991288114498, + "eval_loss": 1.9842805862426758, + "eval_runtime": 43.7649, + "eval_samples_per_second": 22.849, + "eval_steps_per_second": 0.96, + "step": 180 + }, + { + "epoch": 0.2252644679527069, + "grad_norm": 3.229316473007202, + "learning_rate": 3.4647302904564318e-06, + "loss": 1.9048, + "step": 181 + }, + { + "epoch": 0.2265090230242688, + "grad_norm": 3.1970808506011963, + "learning_rate": 3.4854771784232365e-06, + "loss": 1.907, + "step": 182 + }, + { + "epoch": 0.22775357809583074, + "grad_norm": 3.5275299549102783, + "learning_rate": 3.5062240663900416e-06, + "loss": 1.8147, + "step": 183 + }, + { + "epoch": 0.22899813316739265, + "grad_norm": 3.367077589035034, + "learning_rate": 3.526970954356847e-06, + "loss": 1.8189, + "step": 184 + }, + { + "epoch": 0.23024268823895458, + "grad_norm": 3.3052914142608643, + "learning_rate": 3.5477178423236518e-06, + "loss": 1.8362, + "step": 185 + }, + { + "epoch": 0.23148724331051648, + "grad_norm": 3.4737954139709473, + "learning_rate": 3.568464730290457e-06, + "loss": 1.8747, + "step": 186 + }, + { + "epoch": 0.23273179838207841, + "grad_norm": 2.851705312728882, + "learning_rate": 3.5892116182572616e-06, + "loss": 1.7835, + "step": 187 + }, + { + "epoch": 0.23397635345364032, + "grad_norm": 3.615224599838257, + "learning_rate": 3.6099585062240667e-06, + "loss": 1.7855, + "step": 188 + }, + { + "epoch": 0.23522090852520225, + "grad_norm": 2.7468442916870117, + "learning_rate": 3.6307053941908714e-06, + "loss": 1.843, + "step": 189 + }, + { + "epoch": 0.23646546359676415, + "grad_norm": 3.3228273391723633, + "learning_rate": 3.6514522821576765e-06, + "loss": 1.8453, + "step": 190 + }, + { + "epoch": 0.23646546359676415, + "eval_loss": 1.9668089151382446, + "eval_runtime": 44.9228, + "eval_samples_per_second": 22.26, + "eval_steps_per_second": 0.935, + "step": 190 + }, + { + "epoch": 0.23771001866832608, + "grad_norm": 2.7882697582244873, + "learning_rate": 3.672199170124482e-06, + "loss": 1.8625, + "step": 191 + }, + { + "epoch": 0.238954573739888, + "grad_norm": 3.244044780731201, + "learning_rate": 3.6929460580912867e-06, + "loss": 1.8903, + "step": 192 + }, + { + "epoch": 0.24019912881144992, + "grad_norm": 2.8844549655914307, + "learning_rate": 3.713692946058092e-06, + "loss": 1.8929, + "step": 193 + }, + { + "epoch": 0.24144368388301182, + "grad_norm": 2.9207894802093506, + "learning_rate": 3.7344398340248965e-06, + "loss": 1.8458, + "step": 194 + }, + { + "epoch": 0.24268823895457373, + "grad_norm": 2.908562183380127, + "learning_rate": 3.7551867219917016e-06, + "loss": 1.828, + "step": 195 + }, + { + "epoch": 0.24393279402613566, + "grad_norm": 3.1195480823516846, + "learning_rate": 3.7759336099585063e-06, + "loss": 1.855, + "step": 196 + }, + { + "epoch": 0.24517734909769756, + "grad_norm": 3.3910183906555176, + "learning_rate": 3.7966804979253114e-06, + "loss": 1.7868, + "step": 197 + }, + { + "epoch": 0.2464219041692595, + "grad_norm": 2.921475887298584, + "learning_rate": 3.817427385892116e-06, + "loss": 1.7505, + "step": 198 + }, + { + "epoch": 0.2476664592408214, + "grad_norm": 3.5471713542938232, + "learning_rate": 3.838174273858922e-06, + "loss": 1.8445, + "step": 199 + }, + { + "epoch": 0.24891101431238333, + "grad_norm": 3.047698736190796, + "learning_rate": 3.858921161825726e-06, + "loss": 1.8037, + "step": 200 + }, + { + "epoch": 0.24891101431238333, + "eval_loss": 1.9400665760040283, + "eval_runtime": 49.1044, + "eval_samples_per_second": 20.365, + "eval_steps_per_second": 0.855, + "step": 200 + }, + { + "epoch": 0.25015556938394523, + "grad_norm": 3.4988913536071777, + "learning_rate": 3.879668049792531e-06, + "loss": 1.8061, + "step": 201 + }, + { + "epoch": 0.25140012445550713, + "grad_norm": 3.0373125076293945, + "learning_rate": 3.9004149377593365e-06, + "loss": 1.8081, + "step": 202 + }, + { + "epoch": 0.2526446795270691, + "grad_norm": 2.9764745235443115, + "learning_rate": 3.921161825726142e-06, + "loss": 1.8244, + "step": 203 + }, + { + "epoch": 0.253889234598631, + "grad_norm": 3.4413554668426514, + "learning_rate": 3.941908713692946e-06, + "loss": 1.8492, + "step": 204 + }, + { + "epoch": 0.2551337896701929, + "grad_norm": 2.9980437755584717, + "learning_rate": 3.962655601659751e-06, + "loss": 1.805, + "step": 205 + }, + { + "epoch": 0.2563783447417548, + "grad_norm": 3.228750228881836, + "learning_rate": 3.983402489626556e-06, + "loss": 1.7683, + "step": 206 + }, + { + "epoch": 0.25762289981331676, + "grad_norm": 3.008496046066284, + "learning_rate": 4.004149377593361e-06, + "loss": 1.8602, + "step": 207 + }, + { + "epoch": 0.25886745488487867, + "grad_norm": 3.338935375213623, + "learning_rate": 4.024896265560166e-06, + "loss": 1.7978, + "step": 208 + }, + { + "epoch": 0.26011200995644057, + "grad_norm": 3.1450204849243164, + "learning_rate": 4.045643153526971e-06, + "loss": 1.8243, + "step": 209 + }, + { + "epoch": 0.2613565650280025, + "grad_norm": 3.5714142322540283, + "learning_rate": 4.0663900414937765e-06, + "loss": 1.8386, + "step": 210 + }, + { + "epoch": 0.2613565650280025, + "eval_loss": 1.9310717582702637, + "eval_runtime": 42.4826, + "eval_samples_per_second": 23.539, + "eval_steps_per_second": 0.989, + "step": 210 + }, + { + "epoch": 0.26260112009956443, + "grad_norm": 3.3498189449310303, + "learning_rate": 4.087136929460581e-06, + "loss": 1.8055, + "step": 211 + }, + { + "epoch": 0.26384567517112634, + "grad_norm": 3.2423200607299805, + "learning_rate": 4.107883817427386e-06, + "loss": 1.8378, + "step": 212 + }, + { + "epoch": 0.26509023024268824, + "grad_norm": 3.0533974170684814, + "learning_rate": 4.128630705394191e-06, + "loss": 1.7699, + "step": 213 + }, + { + "epoch": 0.26633478531425014, + "grad_norm": 3.135615587234497, + "learning_rate": 4.149377593360996e-06, + "loss": 1.8019, + "step": 214 + }, + { + "epoch": 0.26757934038581205, + "grad_norm": 2.9863040447235107, + "learning_rate": 4.170124481327801e-06, + "loss": 1.7817, + "step": 215 + }, + { + "epoch": 0.268823895457374, + "grad_norm": 3.270598888397217, + "learning_rate": 4.190871369294606e-06, + "loss": 1.7542, + "step": 216 + }, + { + "epoch": 0.2700684505289359, + "grad_norm": 3.2279393672943115, + "learning_rate": 4.211618257261411e-06, + "loss": 1.7964, + "step": 217 + }, + { + "epoch": 0.2713130056004978, + "grad_norm": 3.0626120567321777, + "learning_rate": 4.232365145228216e-06, + "loss": 1.7687, + "step": 218 + }, + { + "epoch": 0.2725575606720597, + "grad_norm": 3.070702075958252, + "learning_rate": 4.253112033195021e-06, + "loss": 1.8075, + "step": 219 + }, + { + "epoch": 0.2738021157436217, + "grad_norm": 3.3623268604278564, + "learning_rate": 4.273858921161826e-06, + "loss": 1.8517, + "step": 220 + }, + { + "epoch": 0.2738021157436217, + "eval_loss": 1.9212934970855713, + "eval_runtime": 46.8503, + "eval_samples_per_second": 21.345, + "eval_steps_per_second": 0.896, + "step": 220 + }, + { + "epoch": 0.2750466708151836, + "grad_norm": 3.005740165710449, + "learning_rate": 4.294605809128631e-06, + "loss": 1.816, + "step": 221 + }, + { + "epoch": 0.2762912258867455, + "grad_norm": 3.5257251262664795, + "learning_rate": 4.315352697095436e-06, + "loss": 1.8191, + "step": 222 + }, + { + "epoch": 0.2775357809583074, + "grad_norm": 3.361103057861328, + "learning_rate": 4.336099585062241e-06, + "loss": 1.8378, + "step": 223 + }, + { + "epoch": 0.27878033602986935, + "grad_norm": 3.787623405456543, + "learning_rate": 4.356846473029046e-06, + "loss": 1.7736, + "step": 224 + }, + { + "epoch": 0.28002489110143125, + "grad_norm": 3.45717453956604, + "learning_rate": 4.3775933609958506e-06, + "loss": 1.8195, + "step": 225 + }, + { + "epoch": 0.28126944617299315, + "grad_norm": 3.1698601245880127, + "learning_rate": 4.398340248962656e-06, + "loss": 1.7756, + "step": 226 + }, + { + "epoch": 0.28251400124455506, + "grad_norm": 3.345258951187134, + "learning_rate": 4.419087136929461e-06, + "loss": 1.7272, + "step": 227 + }, + { + "epoch": 0.28375855631611696, + "grad_norm": 2.905773639678955, + "learning_rate": 4.439834024896266e-06, + "loss": 1.84, + "step": 228 + }, + { + "epoch": 0.2850031113876789, + "grad_norm": 3.2004425525665283, + "learning_rate": 4.460580912863071e-06, + "loss": 1.8164, + "step": 229 + }, + { + "epoch": 0.2862476664592408, + "grad_norm": 2.965275287628174, + "learning_rate": 4.481327800829876e-06, + "loss": 1.7718, + "step": 230 + }, + { + "epoch": 0.2862476664592408, + "eval_loss": 1.903228521347046, + "eval_runtime": 51.1663, + "eval_samples_per_second": 19.544, + "eval_steps_per_second": 0.821, + "step": 230 + }, + { + "epoch": 0.2874922215308027, + "grad_norm": 2.8130710124969482, + "learning_rate": 4.502074688796681e-06, + "loss": 1.7696, + "step": 231 + }, + { + "epoch": 0.28873677660236463, + "grad_norm": 2.9987664222717285, + "learning_rate": 4.5228215767634855e-06, + "loss": 1.775, + "step": 232 + }, + { + "epoch": 0.2899813316739266, + "grad_norm": 3.0936989784240723, + "learning_rate": 4.543568464730291e-06, + "loss": 1.7775, + "step": 233 + }, + { + "epoch": 0.2912258867454885, + "grad_norm": 3.0936989784240723, + "learning_rate": 4.543568464730291e-06, + "loss": 1.7154, + "step": 234 + }, + { + "epoch": 0.2924704418170504, + "grad_norm": 2.8632843494415283, + "learning_rate": 4.564315352697096e-06, + "loss": 1.7771, + "step": 235 + }, + { + "epoch": 0.2937149968886123, + "grad_norm": 3.3314013481140137, + "learning_rate": 4.585062240663901e-06, + "loss": 1.7243, + "step": 236 + }, + { + "epoch": 0.29495955196017426, + "grad_norm": 2.8522872924804688, + "learning_rate": 4.605809128630706e-06, + "loss": 1.7543, + "step": 237 + }, + { + "epoch": 0.29620410703173616, + "grad_norm": 3.319157600402832, + "learning_rate": 4.626556016597511e-06, + "loss": 1.7652, + "step": 238 + }, + { + "epoch": 0.29744866210329807, + "grad_norm": 3.123116970062256, + "learning_rate": 4.647302904564316e-06, + "loss": 1.6933, + "step": 239 + }, + { + "epoch": 0.29869321717485997, + "grad_norm": 3.1735854148864746, + "learning_rate": 4.66804979253112e-06, + "loss": 1.7437, + "step": 240 + }, + { + "epoch": 0.29869321717485997, + "eval_loss": 1.891274333000183, + "eval_runtime": 50.0054, + "eval_samples_per_second": 19.998, + "eval_steps_per_second": 0.84, + "step": 240 + }, + { + "epoch": 0.29993777224642193, + "grad_norm": 3.0996642112731934, + "learning_rate": 4.6887966804979255e-06, + "loss": 1.7594, + "step": 241 + }, + { + "epoch": 0.30118232731798383, + "grad_norm": 3.255063533782959, + "learning_rate": 4.709543568464731e-06, + "loss": 1.7645, + "step": 242 + }, + { + "epoch": 0.30242688238954574, + "grad_norm": 2.722285747528076, + "learning_rate": 4.730290456431536e-06, + "loss": 1.7549, + "step": 243 + }, + { + "epoch": 0.30367143746110764, + "grad_norm": 3.1903417110443115, + "learning_rate": 4.751037344398341e-06, + "loss": 1.7382, + "step": 244 + }, + { + "epoch": 0.30491599253266954, + "grad_norm": 2.8352959156036377, + "learning_rate": 4.771784232365146e-06, + "loss": 1.7497, + "step": 245 + }, + { + "epoch": 0.3061605476042315, + "grad_norm": 3.158536434173584, + "learning_rate": 4.792531120331951e-06, + "loss": 1.7467, + "step": 246 + }, + { + "epoch": 0.3074051026757934, + "grad_norm": 2.8456921577453613, + "learning_rate": 4.813278008298755e-06, + "loss": 1.7472, + "step": 247 + }, + { + "epoch": 0.3086496577473553, + "grad_norm": 3.1215128898620605, + "learning_rate": 4.83402489626556e-06, + "loss": 1.7577, + "step": 248 + }, + { + "epoch": 0.3098942128189172, + "grad_norm": 2.928015947341919, + "learning_rate": 4.8547717842323655e-06, + "loss": 1.6932, + "step": 249 + }, + { + "epoch": 0.3111387678904792, + "grad_norm": 3.001044511795044, + "learning_rate": 4.875518672199171e-06, + "loss": 1.7276, + "step": 250 + }, + { + "epoch": 0.3111387678904792, + "eval_loss": 1.8855507373809814, + "eval_runtime": 46.0143, + "eval_samples_per_second": 21.732, + "eval_steps_per_second": 0.913, + "step": 250 + }, + { + "epoch": 0.3123833229620411, + "grad_norm": 2.988994598388672, + "learning_rate": 4.896265560165976e-06, + "loss": 1.739, + "step": 251 + }, + { + "epoch": 0.313627878033603, + "grad_norm": 3.0091233253479004, + "learning_rate": 4.91701244813278e-06, + "loss": 1.7118, + "step": 252 + }, + { + "epoch": 0.3148724331051649, + "grad_norm": 3.2735323905944824, + "learning_rate": 4.937759336099586e-06, + "loss": 1.7832, + "step": 253 + }, + { + "epoch": 0.31611698817672684, + "grad_norm": 3.055468797683716, + "learning_rate": 4.95850622406639e-06, + "loss": 1.7151, + "step": 254 + }, + { + "epoch": 0.31736154324828875, + "grad_norm": 3.0889883041381836, + "learning_rate": 4.979253112033195e-06, + "loss": 1.7411, + "step": 255 + }, + { + "epoch": 0.31860609831985065, + "grad_norm": 2.8418147563934326, + "learning_rate": 5e-06, + "loss": 1.7575, + "step": 256 + }, + { + "epoch": 0.31985065339141255, + "grad_norm": 3.09739351272583, + "learning_rate": 4.99769372693727e-06, + "loss": 1.7988, + "step": 257 + }, + { + "epoch": 0.3210952084629745, + "grad_norm": 3.245515823364258, + "learning_rate": 4.995387453874539e-06, + "loss": 1.7706, + "step": 258 + }, + { + "epoch": 0.3223397635345364, + "grad_norm": 3.250432014465332, + "learning_rate": 4.993081180811809e-06, + "loss": 1.7839, + "step": 259 + }, + { + "epoch": 0.3235843186060983, + "grad_norm": 3.1325368881225586, + "learning_rate": 4.990774907749078e-06, + "loss": 1.754, + "step": 260 + }, + { + "epoch": 0.3235843186060983, + "eval_loss": 1.869175672531128, + "eval_runtime": 49.0199, + "eval_samples_per_second": 20.4, + "eval_steps_per_second": 0.857, + "step": 260 + }, + { + "epoch": 0.3248288736776602, + "grad_norm": 3.1627941131591797, + "learning_rate": 4.988468634686347e-06, + "loss": 1.7123, + "step": 261 + }, + { + "epoch": 0.3260734287492221, + "grad_norm": 2.848202705383301, + "learning_rate": 4.986162361623617e-06, + "loss": 1.7231, + "step": 262 + }, + { + "epoch": 0.3273179838207841, + "grad_norm": 2.9585494995117188, + "learning_rate": 4.983856088560886e-06, + "loss": 1.7301, + "step": 263 + }, + { + "epoch": 0.328562538892346, + "grad_norm": 3.160170316696167, + "learning_rate": 4.981549815498156e-06, + "loss": 1.7283, + "step": 264 + }, + { + "epoch": 0.3298070939639079, + "grad_norm": 2.6879520416259766, + "learning_rate": 4.979243542435424e-06, + "loss": 1.7494, + "step": 265 + }, + { + "epoch": 0.3310516490354698, + "grad_norm": 3.1313576698303223, + "learning_rate": 4.976937269372694e-06, + "loss": 1.7224, + "step": 266 + }, + { + "epoch": 0.33229620410703176, + "grad_norm": 3.0477867126464844, + "learning_rate": 4.974630996309964e-06, + "loss": 1.6951, + "step": 267 + }, + { + "epoch": 0.33354075917859366, + "grad_norm": 3.133974313735962, + "learning_rate": 4.972324723247233e-06, + "loss": 1.7362, + "step": 268 + }, + { + "epoch": 0.33478531425015556, + "grad_norm": 3.4925222396850586, + "learning_rate": 4.970018450184502e-06, + "loss": 1.7053, + "step": 269 + }, + { + "epoch": 0.33602986932171747, + "grad_norm": 2.848466634750366, + "learning_rate": 4.9677121771217715e-06, + "loss": 1.7336, + "step": 270 + }, + { + "epoch": 0.33602986932171747, + "eval_loss": 1.8723454475402832, + "eval_runtime": 47.9758, + "eval_samples_per_second": 20.844, + "eval_steps_per_second": 0.875, + "step": 270 + }, + { + "epoch": 0.3372744243932794, + "grad_norm": 3.405003309249878, + "learning_rate": 4.965405904059041e-06, + "loss": 1.7178, + "step": 271 + }, + { + "epoch": 0.33851897946484133, + "grad_norm": 2.982916831970215, + "learning_rate": 4.96309963099631e-06, + "loss": 1.7255, + "step": 272 + }, + { + "epoch": 0.33976353453640323, + "grad_norm": 3.1712088584899902, + "learning_rate": 4.96079335793358e-06, + "loss": 1.7328, + "step": 273 + }, + { + "epoch": 0.34100808960796514, + "grad_norm": 3.325481414794922, + "learning_rate": 4.958487084870849e-06, + "loss": 1.7268, + "step": 274 + }, + { + "epoch": 0.3422526446795271, + "grad_norm": 3.0751800537109375, + "learning_rate": 4.956180811808119e-06, + "loss": 1.7916, + "step": 275 + }, + { + "epoch": 0.343497199751089, + "grad_norm": 2.9451465606689453, + "learning_rate": 4.953874538745388e-06, + "loss": 1.7588, + "step": 276 + }, + { + "epoch": 0.3447417548226509, + "grad_norm": 3.1492040157318115, + "learning_rate": 4.9515682656826574e-06, + "loss": 1.7312, + "step": 277 + }, + { + "epoch": 0.3459863098942128, + "grad_norm": 3.165748119354248, + "learning_rate": 4.949261992619927e-06, + "loss": 1.7645, + "step": 278 + }, + { + "epoch": 0.3472308649657747, + "grad_norm": 3.059307336807251, + "learning_rate": 4.946955719557196e-06, + "loss": 1.6783, + "step": 279 + }, + { + "epoch": 0.34847542003733667, + "grad_norm": 2.940147638320923, + "learning_rate": 4.944649446494466e-06, + "loss": 1.7079, + "step": 280 + }, + { + "epoch": 0.34847542003733667, + "eval_loss": 1.8577951192855835, + "eval_runtime": 50.1741, + "eval_samples_per_second": 19.931, + "eval_steps_per_second": 0.837, + "step": 280 + }, + { + "epoch": 0.3497199751088986, + "grad_norm": 2.6892037391662598, + "learning_rate": 4.942343173431734e-06, + "loss": 1.7104, + "step": 281 + }, + { + "epoch": 0.3509645301804605, + "grad_norm": 3.136791467666626, + "learning_rate": 4.940036900369004e-06, + "loss": 1.763, + "step": 282 + }, + { + "epoch": 0.3522090852520224, + "grad_norm": 2.6873879432678223, + "learning_rate": 4.937730627306274e-06, + "loss": 1.722, + "step": 283 + }, + { + "epoch": 0.35345364032358434, + "grad_norm": 3.1375961303710938, + "learning_rate": 4.9354243542435426e-06, + "loss": 1.7048, + "step": 284 + }, + { + "epoch": 0.35469819539514624, + "grad_norm": 2.920725107192993, + "learning_rate": 4.933118081180812e-06, + "loss": 1.7014, + "step": 285 + }, + { + "epoch": 0.35594275046670815, + "grad_norm": 2.750018835067749, + "learning_rate": 4.930811808118081e-06, + "loss": 1.6283, + "step": 286 + }, + { + "epoch": 0.35718730553827005, + "grad_norm": 3.399036169052124, + "learning_rate": 4.928505535055351e-06, + "loss": 1.7265, + "step": 287 + }, + { + "epoch": 0.358431860609832, + "grad_norm": 3.1734066009521484, + "learning_rate": 4.92619926199262e-06, + "loss": 1.7007, + "step": 288 + }, + { + "epoch": 0.3596764156813939, + "grad_norm": 3.038865089416504, + "learning_rate": 4.92389298892989e-06, + "loss": 1.7357, + "step": 289 + }, + { + "epoch": 0.3609209707529558, + "grad_norm": 3.0309486389160156, + "learning_rate": 4.921586715867159e-06, + "loss": 1.7487, + "step": 290 + }, + { + "epoch": 0.3609209707529558, + "eval_loss": 1.8484960794448853, + "eval_runtime": 50.374, + "eval_samples_per_second": 19.852, + "eval_steps_per_second": 0.834, + "step": 290 + }, + { + "epoch": 0.3621655258245177, + "grad_norm": 2.867432117462158, + "learning_rate": 4.9192804428044285e-06, + "loss": 1.6988, + "step": 291 + }, + { + "epoch": 0.3634100808960797, + "grad_norm": 3.31636905670166, + "learning_rate": 4.916974169741698e-06, + "loss": 1.7003, + "step": 292 + }, + { + "epoch": 0.3646546359676416, + "grad_norm": 3.1181511878967285, + "learning_rate": 4.914667896678967e-06, + "loss": 1.6762, + "step": 293 + }, + { + "epoch": 0.3658991910392035, + "grad_norm": 2.978194236755371, + "learning_rate": 4.912361623616237e-06, + "loss": 1.6435, + "step": 294 + }, + { + "epoch": 0.3671437461107654, + "grad_norm": 3.0623116493225098, + "learning_rate": 4.910055350553506e-06, + "loss": 1.7038, + "step": 295 + }, + { + "epoch": 0.3683883011823273, + "grad_norm": 2.83353853225708, + "learning_rate": 4.907749077490776e-06, + "loss": 1.7237, + "step": 296 + }, + { + "epoch": 0.36963285625388925, + "grad_norm": 2.8587100505828857, + "learning_rate": 4.905442804428044e-06, + "loss": 1.6927, + "step": 297 + }, + { + "epoch": 0.37087741132545116, + "grad_norm": 2.7930493354797363, + "learning_rate": 4.903136531365314e-06, + "loss": 1.7381, + "step": 298 + }, + { + "epoch": 0.37212196639701306, + "grad_norm": 2.8992788791656494, + "learning_rate": 4.900830258302584e-06, + "loss": 1.6485, + "step": 299 + }, + { + "epoch": 0.37336652146857496, + "grad_norm": 2.8896944522857666, + "learning_rate": 4.898523985239853e-06, + "loss": 1.7143, + "step": 300 + }, + { + "epoch": 0.37336652146857496, + "eval_loss": 1.8331259489059448, + "eval_runtime": 51.0214, + "eval_samples_per_second": 19.6, + "eval_steps_per_second": 0.823, + "step": 300 + }, + { + "epoch": 0.3746110765401369, + "grad_norm": 2.885444164276123, + "learning_rate": 4.896217712177122e-06, + "loss": 1.6914, + "step": 301 + }, + { + "epoch": 0.3758556316116988, + "grad_norm": 3.2996826171875, + "learning_rate": 4.893911439114391e-06, + "loss": 1.7353, + "step": 302 + }, + { + "epoch": 0.37710018668326073, + "grad_norm": 2.7231531143188477, + "learning_rate": 4.891605166051661e-06, + "loss": 1.6837, + "step": 303 + }, + { + "epoch": 0.37834474175482263, + "grad_norm": 2.9413955211639404, + "learning_rate": 4.88929889298893e-06, + "loss": 1.6448, + "step": 304 + }, + { + "epoch": 0.3795892968263846, + "grad_norm": 2.6486589908599854, + "learning_rate": 4.8869926199262e-06, + "loss": 1.6826, + "step": 305 + }, + { + "epoch": 0.3808338518979465, + "grad_norm": 2.977836847305298, + "learning_rate": 4.884686346863469e-06, + "loss": 1.6737, + "step": 306 + }, + { + "epoch": 0.3820784069695084, + "grad_norm": 2.635324716567993, + "learning_rate": 4.8823800738007384e-06, + "loss": 1.7112, + "step": 307 + }, + { + "epoch": 0.3833229620410703, + "grad_norm": 3.0650839805603027, + "learning_rate": 4.880073800738008e-06, + "loss": 1.7123, + "step": 308 + }, + { + "epoch": 0.3845675171126322, + "grad_norm": 3.059629440307617, + "learning_rate": 4.877767527675277e-06, + "loss": 1.662, + "step": 309 + }, + { + "epoch": 0.38581207218419417, + "grad_norm": 2.7559597492218018, + "learning_rate": 4.875461254612546e-06, + "loss": 1.7418, + "step": 310 + }, + { + "epoch": 0.38581207218419417, + "eval_loss": 1.8277366161346436, + "eval_runtime": 42.6671, + "eval_samples_per_second": 23.437, + "eval_steps_per_second": 0.984, + "step": 310 + }, + { + "epoch": 0.38705662725575607, + "grad_norm": 2.7987024784088135, + "learning_rate": 4.873154981549816e-06, + "loss": 1.7326, + "step": 311 + }, + { + "epoch": 0.388301182327318, + "grad_norm": 2.999824047088623, + "learning_rate": 4.8708487084870856e-06, + "loss": 1.7344, + "step": 312 + }, + { + "epoch": 0.3895457373988799, + "grad_norm": 2.942995071411133, + "learning_rate": 4.868542435424355e-06, + "loss": 1.7055, + "step": 313 + }, + { + "epoch": 0.39079029247044184, + "grad_norm": 2.823906183242798, + "learning_rate": 4.8662361623616235e-06, + "loss": 1.6819, + "step": 314 + }, + { + "epoch": 0.39203484754200374, + "grad_norm": 2.7798759937286377, + "learning_rate": 4.863929889298894e-06, + "loss": 1.6694, + "step": 315 + }, + { + "epoch": 0.39327940261356564, + "grad_norm": 2.599005699157715, + "learning_rate": 4.861623616236163e-06, + "loss": 1.6924, + "step": 316 + }, + { + "epoch": 0.39452395768512755, + "grad_norm": 2.687539577484131, + "learning_rate": 4.859317343173432e-06, + "loss": 1.685, + "step": 317 + }, + { + "epoch": 0.3957685127566895, + "grad_norm": 2.9679677486419678, + "learning_rate": 4.857011070110701e-06, + "loss": 1.718, + "step": 318 + }, + { + "epoch": 0.3970130678282514, + "grad_norm": 2.8611817359924316, + "learning_rate": 4.854704797047971e-06, + "loss": 1.7247, + "step": 319 + }, + { + "epoch": 0.3982576228998133, + "grad_norm": 2.7461249828338623, + "learning_rate": 4.85239852398524e-06, + "loss": 1.6434, + "step": 320 + }, + { + "epoch": 0.3982576228998133, + "eval_loss": 1.8234667778015137, + "eval_runtime": 44.3433, + "eval_samples_per_second": 22.551, + "eval_steps_per_second": 0.947, + "step": 320 + }, + { + "epoch": 0.3995021779713752, + "grad_norm": 2.7142515182495117, + "learning_rate": 4.8500922509225095e-06, + "loss": 1.6855, + "step": 321 + }, + { + "epoch": 0.4007467330429372, + "grad_norm": 2.87727689743042, + "learning_rate": 4.847785977859779e-06, + "loss": 1.705, + "step": 322 + }, + { + "epoch": 0.4019912881144991, + "grad_norm": 2.9598982334136963, + "learning_rate": 4.845479704797048e-06, + "loss": 1.6482, + "step": 323 + }, + { + "epoch": 0.403235843186061, + "grad_norm": 2.8654463291168213, + "learning_rate": 4.843173431734318e-06, + "loss": 1.5816, + "step": 324 + }, + { + "epoch": 0.4044803982576229, + "grad_norm": 2.983837842941284, + "learning_rate": 4.840867158671587e-06, + "loss": 1.6962, + "step": 325 + }, + { + "epoch": 0.4057249533291848, + "grad_norm": 2.730914354324341, + "learning_rate": 4.838560885608857e-06, + "loss": 1.6991, + "step": 326 + }, + { + "epoch": 0.40696950840074675, + "grad_norm": 2.6901161670684814, + "learning_rate": 4.836254612546126e-06, + "loss": 1.6839, + "step": 327 + }, + { + "epoch": 0.40821406347230865, + "grad_norm": 2.6661486625671387, + "learning_rate": 4.8339483394833955e-06, + "loss": 1.6894, + "step": 328 + }, + { + "epoch": 0.40945861854387056, + "grad_norm": 2.64243221282959, + "learning_rate": 4.831642066420665e-06, + "loss": 1.6533, + "step": 329 + }, + { + "epoch": 0.41070317361543246, + "grad_norm": 2.586725950241089, + "learning_rate": 4.8293357933579335e-06, + "loss": 1.6486, + "step": 330 + }, + { + "epoch": 0.41070317361543246, + "eval_loss": 1.82283616065979, + "eval_runtime": 46.5937, + "eval_samples_per_second": 21.462, + "eval_steps_per_second": 0.901, + "step": 330 + }, + { + "epoch": 0.4119477286869944, + "grad_norm": 2.899968147277832, + "learning_rate": 4.827029520295204e-06, + "loss": 1.7136, + "step": 331 + }, + { + "epoch": 0.4131922837585563, + "grad_norm": 3.067671537399292, + "learning_rate": 4.824723247232473e-06, + "loss": 1.672, + "step": 332 + }, + { + "epoch": 0.4144368388301182, + "grad_norm": 2.8144404888153076, + "learning_rate": 4.822416974169742e-06, + "loss": 1.7173, + "step": 333 + }, + { + "epoch": 0.41568139390168013, + "grad_norm": 3.0551373958587646, + "learning_rate": 4.820110701107011e-06, + "loss": 1.6843, + "step": 334 + }, + { + "epoch": 0.4169259489732421, + "grad_norm": 2.7585153579711914, + "learning_rate": 4.817804428044281e-06, + "loss": 1.6834, + "step": 335 + }, + { + "epoch": 0.418170504044804, + "grad_norm": 2.8691210746765137, + "learning_rate": 4.81549815498155e-06, + "loss": 1.6814, + "step": 336 + }, + { + "epoch": 0.4194150591163659, + "grad_norm": 2.871523380279541, + "learning_rate": 4.8131918819188194e-06, + "loss": 1.6222, + "step": 337 + }, + { + "epoch": 0.4206596141879278, + "grad_norm": 2.8928768634796143, + "learning_rate": 4.810885608856089e-06, + "loss": 1.6984, + "step": 338 + }, + { + "epoch": 0.42190416925948976, + "grad_norm": 3.285256862640381, + "learning_rate": 4.808579335793358e-06, + "loss": 1.7029, + "step": 339 + }, + { + "epoch": 0.42314872433105166, + "grad_norm": 2.785504102706909, + "learning_rate": 4.806273062730628e-06, + "loss": 1.6857, + "step": 340 + }, + { + "epoch": 0.42314872433105166, + "eval_loss": 1.8053412437438965, + "eval_runtime": 48.0656, + "eval_samples_per_second": 20.805, + "eval_steps_per_second": 0.874, + "step": 340 + }, + { + "epoch": 0.42439327940261357, + "grad_norm": 3.0227086544036865, + "learning_rate": 4.803966789667897e-06, + "loss": 1.6903, + "step": 341 + }, + { + "epoch": 0.42563783447417547, + "grad_norm": 2.616769790649414, + "learning_rate": 4.8016605166051665e-06, + "loss": 1.6144, + "step": 342 + }, + { + "epoch": 0.4268823895457374, + "grad_norm": 2.9519753456115723, + "learning_rate": 4.799354243542436e-06, + "loss": 1.6157, + "step": 343 + }, + { + "epoch": 0.42812694461729933, + "grad_norm": 2.764512062072754, + "learning_rate": 4.797047970479705e-06, + "loss": 1.6248, + "step": 344 + }, + { + "epoch": 0.42937149968886124, + "grad_norm": 2.999178886413574, + "learning_rate": 4.794741697416975e-06, + "loss": 1.6764, + "step": 345 + }, + { + "epoch": 0.43061605476042314, + "grad_norm": 3.0472185611724854, + "learning_rate": 4.792435424354243e-06, + "loss": 1.6421, + "step": 346 + }, + { + "epoch": 0.43186060983198504, + "grad_norm": 2.6638400554656982, + "learning_rate": 4.790129151291514e-06, + "loss": 1.6304, + "step": 347 + }, + { + "epoch": 0.433105164903547, + "grad_norm": 2.945295810699463, + "learning_rate": 4.787822878228783e-06, + "loss": 1.6181, + "step": 348 + }, + { + "epoch": 0.4343497199751089, + "grad_norm": 2.915125846862793, + "learning_rate": 4.7855166051660525e-06, + "loss": 1.6789, + "step": 349 + }, + { + "epoch": 0.4355942750466708, + "grad_norm": 2.5674026012420654, + "learning_rate": 4.783210332103321e-06, + "loss": 1.6723, + "step": 350 + }, + { + "epoch": 0.4355942750466708, + "eval_loss": 1.8109745979309082, + "eval_runtime": 43.9943, + "eval_samples_per_second": 22.73, + "eval_steps_per_second": 0.955, + "step": 350 + }, + { + "epoch": 0.4368388301182327, + "grad_norm": 2.9533393383026123, + "learning_rate": 4.7809040590405905e-06, + "loss": 1.6271, + "step": 351 + }, + { + "epoch": 0.43808338518979467, + "grad_norm": 2.7677664756774902, + "learning_rate": 4.778597785977861e-06, + "loss": 1.6633, + "step": 352 + }, + { + "epoch": 0.4393279402613566, + "grad_norm": 2.527456521987915, + "learning_rate": 4.776291512915129e-06, + "loss": 1.5644, + "step": 353 + }, + { + "epoch": 0.4405724953329185, + "grad_norm": 2.9778528213500977, + "learning_rate": 4.773985239852399e-06, + "loss": 1.6166, + "step": 354 + }, + { + "epoch": 0.4418170504044804, + "grad_norm": 2.5554800033569336, + "learning_rate": 4.771678966789668e-06, + "loss": 1.5573, + "step": 355 + }, + { + "epoch": 0.44306160547604234, + "grad_norm": 2.7415409088134766, + "learning_rate": 4.769372693726938e-06, + "loss": 1.6564, + "step": 356 + }, + { + "epoch": 0.44430616054760425, + "grad_norm": 3.0275440216064453, + "learning_rate": 4.767066420664207e-06, + "loss": 1.67, + "step": 357 + }, + { + "epoch": 0.44555071561916615, + "grad_norm": 2.720694065093994, + "learning_rate": 4.7647601476014765e-06, + "loss": 1.6374, + "step": 358 + }, + { + "epoch": 0.44679527069072805, + "grad_norm": 2.97589111328125, + "learning_rate": 4.762453874538746e-06, + "loss": 1.6596, + "step": 359 + }, + { + "epoch": 0.44803982576228996, + "grad_norm": 2.8005781173706055, + "learning_rate": 4.760147601476015e-06, + "loss": 1.6281, + "step": 360 + }, + { + "epoch": 0.44803982576228996, + "eval_loss": 1.8139524459838867, + "eval_runtime": 44.5817, + "eval_samples_per_second": 22.431, + "eval_steps_per_second": 0.942, + "step": 360 + }, + { + "epoch": 0.4492843808338519, + "grad_norm": 3.3947463035583496, + "learning_rate": 4.757841328413285e-06, + "loss": 1.7149, + "step": 361 + }, + { + "epoch": 0.4505289359054138, + "grad_norm": 2.991457939147949, + "learning_rate": 4.755535055350554e-06, + "loss": 1.5971, + "step": 362 + }, + { + "epoch": 0.4517734909769757, + "grad_norm": 2.653313398361206, + "learning_rate": 4.753228782287823e-06, + "loss": 1.6465, + "step": 363 + }, + { + "epoch": 0.4530180460485376, + "grad_norm": 2.716360569000244, + "learning_rate": 4.750922509225093e-06, + "loss": 1.6775, + "step": 364 + }, + { + "epoch": 0.4542626011200996, + "grad_norm": 2.733598470687866, + "learning_rate": 4.748616236162362e-06, + "loss": 1.6087, + "step": 365 + }, + { + "epoch": 0.4555071561916615, + "grad_norm": 2.581547737121582, + "learning_rate": 4.746309963099631e-06, + "loss": 1.6339, + "step": 366 + }, + { + "epoch": 0.4567517112632234, + "grad_norm": 2.6841800212860107, + "learning_rate": 4.7440036900369e-06, + "loss": 1.6193, + "step": 367 + }, + { + "epoch": 0.4579962663347853, + "grad_norm": 2.658369302749634, + "learning_rate": 4.741697416974171e-06, + "loss": 1.6287, + "step": 368 + }, + { + "epoch": 0.45924082140634725, + "grad_norm": 2.7204179763793945, + "learning_rate": 4.739391143911439e-06, + "loss": 1.6755, + "step": 369 + }, + { + "epoch": 0.46048537647790916, + "grad_norm": 2.6627941131591797, + "learning_rate": 4.737084870848709e-06, + "loss": 1.6142, + "step": 370 + }, + { + "epoch": 0.46048537647790916, + "eval_loss": 1.8019787073135376, + "eval_runtime": 42.7804, + "eval_samples_per_second": 23.375, + "eval_steps_per_second": 0.982, + "step": 370 + }, + { + "epoch": 0.46172993154947106, + "grad_norm": 2.650705337524414, + "learning_rate": 4.734778597785978e-06, + "loss": 1.641, + "step": 371 + }, + { + "epoch": 0.46297448662103297, + "grad_norm": 2.7031619548797607, + "learning_rate": 4.7324723247232475e-06, + "loss": 1.6596, + "step": 372 + }, + { + "epoch": 0.4642190416925949, + "grad_norm": 3.3050339221954346, + "learning_rate": 4.730166051660517e-06, + "loss": 1.6664, + "step": 373 + }, + { + "epoch": 0.46546359676415683, + "grad_norm": 2.841404676437378, + "learning_rate": 4.727859778597786e-06, + "loss": 1.6474, + "step": 374 + }, + { + "epoch": 0.46670815183571873, + "grad_norm": 3.0990259647369385, + "learning_rate": 4.725553505535056e-06, + "loss": 1.6008, + "step": 375 + }, + { + "epoch": 0.46795270690728064, + "grad_norm": 2.987772226333618, + "learning_rate": 4.723247232472325e-06, + "loss": 1.6374, + "step": 376 + }, + { + "epoch": 0.46919726197884254, + "grad_norm": 3.3345255851745605, + "learning_rate": 4.720940959409595e-06, + "loss": 1.6405, + "step": 377 + }, + { + "epoch": 0.4704418170504045, + "grad_norm": 3.0810067653656006, + "learning_rate": 4.718634686346864e-06, + "loss": 1.6229, + "step": 378 + }, + { + "epoch": 0.4716863721219664, + "grad_norm": 3.246685266494751, + "learning_rate": 4.716328413284133e-06, + "loss": 1.6262, + "step": 379 + }, + { + "epoch": 0.4729309271935283, + "grad_norm": 2.6052889823913574, + "learning_rate": 4.714022140221403e-06, + "loss": 1.6241, + "step": 380 + }, + { + "epoch": 0.4729309271935283, + "eval_loss": 1.7957485914230347, + "eval_runtime": 43.6626, + "eval_samples_per_second": 22.903, + "eval_steps_per_second": 0.962, + "step": 380 + }, + { + "epoch": 0.4741754822650902, + "grad_norm": 3.747142791748047, + "learning_rate": 4.711715867158672e-06, + "loss": 1.6764, + "step": 381 + }, + { + "epoch": 0.47542003733665217, + "grad_norm": 2.8508121967315674, + "learning_rate": 4.709409594095941e-06, + "loss": 1.6597, + "step": 382 + }, + { + "epoch": 0.47666459240821407, + "grad_norm": 3.5615711212158203, + "learning_rate": 4.70710332103321e-06, + "loss": 1.6089, + "step": 383 + }, + { + "epoch": 0.477909147479776, + "grad_norm": 2.5886309146881104, + "learning_rate": 4.704797047970481e-06, + "loss": 1.5996, + "step": 384 + }, + { + "epoch": 0.4791537025513379, + "grad_norm": 3.3900864124298096, + "learning_rate": 4.702490774907749e-06, + "loss": 1.7137, + "step": 385 + }, + { + "epoch": 0.48039825762289984, + "grad_norm": 2.913641929626465, + "learning_rate": 4.700184501845019e-06, + "loss": 1.6512, + "step": 386 + }, + { + "epoch": 0.48164281269446174, + "grad_norm": 2.810722827911377, + "learning_rate": 4.697878228782288e-06, + "loss": 1.6939, + "step": 387 + }, + { + "epoch": 0.48288736776602365, + "grad_norm": 3.2701401710510254, + "learning_rate": 4.6955719557195575e-06, + "loss": 1.6455, + "step": 388 + }, + { + "epoch": 0.48413192283758555, + "grad_norm": 2.972931385040283, + "learning_rate": 4.693265682656827e-06, + "loss": 1.6272, + "step": 389 + }, + { + "epoch": 0.48537647790914745, + "grad_norm": 3.2157890796661377, + "learning_rate": 4.690959409594096e-06, + "loss": 1.6409, + "step": 390 + }, + { + "epoch": 0.48537647790914745, + "eval_loss": 1.7859243154525757, + "eval_runtime": 45.2093, + "eval_samples_per_second": 22.119, + "eval_steps_per_second": 0.929, + "step": 390 + }, + { + "epoch": 0.4866210329807094, + "grad_norm": 2.783360719680786, + "learning_rate": 4.688653136531366e-06, + "loss": 1.6303, + "step": 391 + }, + { + "epoch": 0.4878655880522713, + "grad_norm": 2.6783013343811035, + "learning_rate": 4.686346863468635e-06, + "loss": 1.7259, + "step": 392 + }, + { + "epoch": 0.4891101431238332, + "grad_norm": 2.899916172027588, + "learning_rate": 4.6840405904059046e-06, + "loss": 1.6434, + "step": 393 + }, + { + "epoch": 0.4903546981953951, + "grad_norm": 2.604377269744873, + "learning_rate": 4.681734317343174e-06, + "loss": 1.6838, + "step": 394 + }, + { + "epoch": 0.4915992532669571, + "grad_norm": 3.0830864906311035, + "learning_rate": 4.6794280442804426e-06, + "loss": 1.6053, + "step": 395 + }, + { + "epoch": 0.492843808338519, + "grad_norm": 3.1381475925445557, + "learning_rate": 4.677121771217713e-06, + "loss": 1.6132, + "step": 396 + }, + { + "epoch": 0.4940883634100809, + "grad_norm": 3.097426652908325, + "learning_rate": 4.674815498154982e-06, + "loss": 1.6237, + "step": 397 + }, + { + "epoch": 0.4953329184816428, + "grad_norm": 2.9942305088043213, + "learning_rate": 4.672509225092252e-06, + "loss": 1.6554, + "step": 398 + }, + { + "epoch": 0.49657747355320475, + "grad_norm": 2.834199905395508, + "learning_rate": 4.67020295202952e-06, + "loss": 1.6389, + "step": 399 + }, + { + "epoch": 0.49782202862476665, + "grad_norm": 3.183389902114868, + "learning_rate": 4.66789667896679e-06, + "loss": 1.6518, + "step": 400 + }, + { + "epoch": 0.49782202862476665, + "eval_loss": 1.781000018119812, + "eval_runtime": 45.4127, + "eval_samples_per_second": 22.02, + "eval_steps_per_second": 0.925, + "step": 400 + }, + { + "epoch": 0.49906658369632856, + "grad_norm": 2.8186683654785156, + "learning_rate": 4.66559040590406e-06, + "loss": 1.669, + "step": 401 + }, + { + "epoch": 0.5003111387678905, + "grad_norm": 2.618595600128174, + "learning_rate": 4.6632841328413285e-06, + "loss": 1.6767, + "step": 402 + }, + { + "epoch": 0.5015556938394524, + "grad_norm": 3.080432653427124, + "learning_rate": 4.660977859778598e-06, + "loss": 1.7096, + "step": 403 + }, + { + "epoch": 0.5028002489110143, + "grad_norm": 2.6780874729156494, + "learning_rate": 4.658671586715867e-06, + "loss": 1.6576, + "step": 404 + }, + { + "epoch": 0.5040448039825762, + "grad_norm": 2.8194704055786133, + "learning_rate": 4.656365313653137e-06, + "loss": 1.6405, + "step": 405 + }, + { + "epoch": 0.5052893590541382, + "grad_norm": 3.2195825576782227, + "learning_rate": 4.654059040590406e-06, + "loss": 1.6631, + "step": 406 + }, + { + "epoch": 0.5065339141257, + "grad_norm": 2.7719528675079346, + "learning_rate": 4.651752767527676e-06, + "loss": 1.6036, + "step": 407 + }, + { + "epoch": 0.507778469197262, + "grad_norm": 3.186394453048706, + "learning_rate": 4.649446494464945e-06, + "loss": 1.6508, + "step": 408 + }, + { + "epoch": 0.5090230242688238, + "grad_norm": 2.674922227859497, + "learning_rate": 4.6471402214022145e-06, + "loss": 1.6033, + "step": 409 + }, + { + "epoch": 0.5102675793403858, + "grad_norm": 3.2772562503814697, + "learning_rate": 4.644833948339484e-06, + "loss": 1.5916, + "step": 410 + }, + { + "epoch": 0.5102675793403858, + "eval_loss": 1.784346342086792, + "eval_runtime": 42.2178, + "eval_samples_per_second": 23.687, + "eval_steps_per_second": 0.995, + "step": 410 + }, + { + "epoch": 0.5115121344119478, + "grad_norm": 2.9295618534088135, + "learning_rate": 4.642527675276753e-06, + "loss": 1.6693, + "step": 411 + }, + { + "epoch": 0.5127566894835096, + "grad_norm": 2.8492650985717773, + "learning_rate": 4.640221402214023e-06, + "loss": 1.658, + "step": 412 + }, + { + "epoch": 0.5140012445550716, + "grad_norm": 2.921211004257202, + "learning_rate": 4.637915129151292e-06, + "loss": 1.6559, + "step": 413 + }, + { + "epoch": 0.5152457996266335, + "grad_norm": 2.583395004272461, + "learning_rate": 4.635608856088562e-06, + "loss": 1.6869, + "step": 414 + }, + { + "epoch": 0.5164903546981954, + "grad_norm": 2.79984712600708, + "learning_rate": 4.63330258302583e-06, + "loss": 1.6007, + "step": 415 + }, + { + "epoch": 0.5177349097697573, + "grad_norm": 2.872856616973877, + "learning_rate": 4.6309963099631e-06, + "loss": 1.6132, + "step": 416 + }, + { + "epoch": 0.5189794648413192, + "grad_norm": 2.7482125759124756, + "learning_rate": 4.62869003690037e-06, + "loss": 1.5805, + "step": 417 + }, + { + "epoch": 0.5202240199128811, + "grad_norm": 2.9227206707000732, + "learning_rate": 4.6263837638376384e-06, + "loss": 1.6362, + "step": 418 + }, + { + "epoch": 0.5214685749844431, + "grad_norm": 2.858302354812622, + "learning_rate": 4.624077490774908e-06, + "loss": 1.5934, + "step": 419 + }, + { + "epoch": 0.522713130056005, + "grad_norm": 3.2696762084960938, + "learning_rate": 4.621771217712177e-06, + "loss": 1.6215, + "step": 420 + }, + { + "epoch": 0.522713130056005, + "eval_loss": 1.7770200967788696, + "eval_runtime": 46.0922, + "eval_samples_per_second": 21.696, + "eval_steps_per_second": 0.911, + "step": 420 + }, + { + "epoch": 0.5239576851275669, + "grad_norm": 2.525740623474121, + "learning_rate": 4.619464944649447e-06, + "loss": 1.6193, + "step": 421 + }, + { + "epoch": 0.5252022401991289, + "grad_norm": 2.9050614833831787, + "learning_rate": 4.617158671586716e-06, + "loss": 1.69, + "step": 422 + }, + { + "epoch": 0.5264467952706907, + "grad_norm": 2.959547519683838, + "learning_rate": 4.6148523985239856e-06, + "loss": 1.6874, + "step": 423 + }, + { + "epoch": 0.5276913503422527, + "grad_norm": 2.6331565380096436, + "learning_rate": 4.612546125461255e-06, + "loss": 1.5795, + "step": 424 + }, + { + "epoch": 0.5289359054138145, + "grad_norm": 2.7323524951934814, + "learning_rate": 4.610239852398524e-06, + "loss": 1.5882, + "step": 425 + }, + { + "epoch": 0.5301804604853765, + "grad_norm": 2.8021790981292725, + "learning_rate": 4.607933579335794e-06, + "loss": 1.5852, + "step": 426 + }, + { + "epoch": 0.5314250155569384, + "grad_norm": 2.984299659729004, + "learning_rate": 4.605627306273063e-06, + "loss": 1.6168, + "step": 427 + }, + { + "epoch": 0.5326695706285003, + "grad_norm": 2.92749285697937, + "learning_rate": 4.603321033210333e-06, + "loss": 1.5768, + "step": 428 + }, + { + "epoch": 0.5339141257000622, + "grad_norm": 2.7681751251220703, + "learning_rate": 4.601014760147602e-06, + "loss": 1.5831, + "step": 429 + }, + { + "epoch": 0.5351586807716241, + "grad_norm": 2.965458631515503, + "learning_rate": 4.5987084870848715e-06, + "loss": 1.6412, + "step": 430 + }, + { + "epoch": 0.5351586807716241, + "eval_loss": 1.766748070716858, + "eval_runtime": 51.1098, + "eval_samples_per_second": 19.566, + "eval_steps_per_second": 0.822, + "step": 430 + }, + { + "epoch": 0.536403235843186, + "grad_norm": 2.681853771209717, + "learning_rate": 4.59640221402214e-06, + "loss": 1.6302, + "step": 431 + }, + { + "epoch": 0.537647790914748, + "grad_norm": 2.698493480682373, + "learning_rate": 4.5940959409594095e-06, + "loss": 1.6774, + "step": 432 + }, + { + "epoch": 0.5388923459863099, + "grad_norm": 3.1715168952941895, + "learning_rate": 4.59178966789668e-06, + "loss": 1.5617, + "step": 433 + }, + { + "epoch": 0.5401369010578718, + "grad_norm": 2.98128080368042, + "learning_rate": 4.589483394833948e-06, + "loss": 1.5901, + "step": 434 + }, + { + "epoch": 0.5413814561294338, + "grad_norm": 3.3887572288513184, + "learning_rate": 4.587177121771218e-06, + "loss": 1.6448, + "step": 435 + }, + { + "epoch": 0.5426260112009956, + "grad_norm": 2.647975206375122, + "learning_rate": 4.584870848708487e-06, + "loss": 1.6209, + "step": 436 + }, + { + "epoch": 0.5438705662725576, + "grad_norm": 2.833264112472534, + "learning_rate": 4.5825645756457575e-06, + "loss": 1.6875, + "step": 437 + }, + { + "epoch": 0.5451151213441194, + "grad_norm": 3.0966858863830566, + "learning_rate": 4.580258302583026e-06, + "loss": 1.6291, + "step": 438 + }, + { + "epoch": 0.5463596764156814, + "grad_norm": 2.768793821334839, + "learning_rate": 4.5779520295202955e-06, + "loss": 1.6183, + "step": 439 + }, + { + "epoch": 0.5476042314872434, + "grad_norm": 3.4009342193603516, + "learning_rate": 4.575645756457565e-06, + "loss": 1.6602, + "step": 440 + }, + { + "epoch": 0.5476042314872434, + "eval_loss": 1.76938796043396, + "eval_runtime": 46.2093, + "eval_samples_per_second": 21.641, + "eval_steps_per_second": 0.909, + "step": 440 + }, + { + "epoch": 0.5488487865588052, + "grad_norm": 2.769113063812256, + "learning_rate": 4.573339483394834e-06, + "loss": 1.6453, + "step": 441 + }, + { + "epoch": 0.5500933416303672, + "grad_norm": 3.3592801094055176, + "learning_rate": 4.571033210332104e-06, + "loss": 1.6602, + "step": 442 + }, + { + "epoch": 0.551337896701929, + "grad_norm": 2.6817564964294434, + "learning_rate": 4.568726937269373e-06, + "loss": 1.5714, + "step": 443 + }, + { + "epoch": 0.552582451773491, + "grad_norm": 3.093724250793457, + "learning_rate": 4.566420664206643e-06, + "loss": 1.6924, + "step": 444 + }, + { + "epoch": 0.5538270068450529, + "grad_norm": 2.766580104827881, + "learning_rate": 4.564114391143912e-06, + "loss": 1.5714, + "step": 445 + }, + { + "epoch": 0.5550715619166148, + "grad_norm": 2.868870973587036, + "learning_rate": 4.5618081180811814e-06, + "loss": 1.5823, + "step": 446 + }, + { + "epoch": 0.5563161169881767, + "grad_norm": 2.8286170959472656, + "learning_rate": 4.559501845018451e-06, + "loss": 1.6123, + "step": 447 + }, + { + "epoch": 0.5575606720597387, + "grad_norm": 2.66096568107605, + "learning_rate": 4.5571955719557194e-06, + "loss": 1.6044, + "step": 448 + }, + { + "epoch": 0.5588052271313005, + "grad_norm": 2.7411820888519287, + "learning_rate": 4.55488929889299e-06, + "loss": 1.6361, + "step": 449 + }, + { + "epoch": 0.5600497822028625, + "grad_norm": 2.7591240406036377, + "learning_rate": 4.552583025830259e-06, + "loss": 1.6257, + "step": 450 + }, + { + "epoch": 0.5600497822028625, + "eval_loss": 1.758321762084961, + "eval_runtime": 52.9135, + "eval_samples_per_second": 18.899, + "eval_steps_per_second": 0.794, + "step": 450 + }, + { + "epoch": 0.5612943372744243, + "grad_norm": 2.733456611633301, + "learning_rate": 4.550276752767528e-06, + "loss": 1.7159, + "step": 451 + }, + { + "epoch": 0.5625388923459863, + "grad_norm": 2.7318639755249023, + "learning_rate": 4.547970479704797e-06, + "loss": 1.6062, + "step": 452 + }, + { + "epoch": 0.5637834474175483, + "grad_norm": 2.8235232830047607, + "learning_rate": 4.5456642066420666e-06, + "loss": 1.6616, + "step": 453 + }, + { + "epoch": 0.5650280024891101, + "grad_norm": 2.499028444290161, + "learning_rate": 4.543357933579336e-06, + "loss": 1.5795, + "step": 454 + }, + { + "epoch": 0.5662725575606721, + "grad_norm": 2.7547316551208496, + "learning_rate": 4.541051660516605e-06, + "loss": 1.6204, + "step": 455 + }, + { + "epoch": 0.5675171126322339, + "grad_norm": 2.7086215019226074, + "learning_rate": 4.538745387453875e-06, + "loss": 1.5611, + "step": 456 + }, + { + "epoch": 0.5687616677037959, + "grad_norm": 2.911731719970703, + "learning_rate": 4.536439114391144e-06, + "loss": 1.645, + "step": 457 + }, + { + "epoch": 0.5700062227753578, + "grad_norm": 2.5553250312805176, + "learning_rate": 4.534132841328414e-06, + "loss": 1.6093, + "step": 458 + }, + { + "epoch": 0.5712507778469197, + "grad_norm": 2.795919895172119, + "learning_rate": 4.531826568265683e-06, + "loss": 1.6, + "step": 459 + }, + { + "epoch": 0.5724953329184816, + "grad_norm": 2.92104434967041, + "learning_rate": 4.5295202952029525e-06, + "loss": 1.6444, + "step": 460 + }, + { + "epoch": 0.5724953329184816, + "eval_loss": 1.760049819946289, + "eval_runtime": 48.811, + "eval_samples_per_second": 20.487, + "eval_steps_per_second": 0.86, + "step": 460 + }, + { + "epoch": 0.5737398879900436, + "grad_norm": 2.6339287757873535, + "learning_rate": 4.527214022140222e-06, + "loss": 1.6133, + "step": 461 + }, + { + "epoch": 0.5749844430616055, + "grad_norm": 2.531398057937622, + "learning_rate": 4.524907749077491e-06, + "loss": 1.5931, + "step": 462 + }, + { + "epoch": 0.5762289981331674, + "grad_norm": 2.8858656883239746, + "learning_rate": 4.522601476014761e-06, + "loss": 1.6287, + "step": 463 + }, + { + "epoch": 0.5774735532047293, + "grad_norm": 2.7309634685516357, + "learning_rate": 4.520295202952029e-06, + "loss": 1.6102, + "step": 464 + }, + { + "epoch": 0.5787181082762912, + "grad_norm": 2.7758548259735107, + "learning_rate": 4.5179889298893e-06, + "loss": 1.6292, + "step": 465 + }, + { + "epoch": 0.5799626633478532, + "grad_norm": 2.6181721687316895, + "learning_rate": 4.515682656826569e-06, + "loss": 1.6113, + "step": 466 + }, + { + "epoch": 0.581207218419415, + "grad_norm": 2.6215691566467285, + "learning_rate": 4.513376383763838e-06, + "loss": 1.5158, + "step": 467 + }, + { + "epoch": 0.582451773490977, + "grad_norm": 2.727487087249756, + "learning_rate": 4.511070110701107e-06, + "loss": 1.6212, + "step": 468 + }, + { + "epoch": 0.583696328562539, + "grad_norm": 2.6248037815093994, + "learning_rate": 4.5087638376383765e-06, + "loss": 1.6117, + "step": 469 + }, + { + "epoch": 0.5849408836341008, + "grad_norm": 2.714291572570801, + "learning_rate": 4.506457564575646e-06, + "loss": 1.612, + "step": 470 + }, + { + "epoch": 0.5849408836341008, + "eval_loss": 1.7569347620010376, + "eval_runtime": 51.8475, + "eval_samples_per_second": 19.287, + "eval_steps_per_second": 0.81, + "step": 470 + }, + { + "epoch": 0.5861854387056628, + "grad_norm": 2.752577066421509, + "learning_rate": 4.504151291512915e-06, + "loss": 1.5969, + "step": 471 + }, + { + "epoch": 0.5874299937772246, + "grad_norm": 2.688563346862793, + "learning_rate": 4.501845018450185e-06, + "loss": 1.6085, + "step": 472 + }, + { + "epoch": 0.5886745488487866, + "grad_norm": 2.8581559658050537, + "learning_rate": 4.499538745387454e-06, + "loss": 1.6688, + "step": 473 + }, + { + "epoch": 0.5899191039203485, + "grad_norm": 2.8520970344543457, + "learning_rate": 4.497232472324724e-06, + "loss": 1.6034, + "step": 474 + }, + { + "epoch": 0.5911636589919104, + "grad_norm": 2.8052237033843994, + "learning_rate": 4.494926199261993e-06, + "loss": 1.5656, + "step": 475 + }, + { + "epoch": 0.5924082140634723, + "grad_norm": 2.619663953781128, + "learning_rate": 4.4926199261992624e-06, + "loss": 1.5543, + "step": 476 + }, + { + "epoch": 0.5936527691350342, + "grad_norm": 2.604964017868042, + "learning_rate": 4.490313653136532e-06, + "loss": 1.5671, + "step": 477 + }, + { + "epoch": 0.5948973242065961, + "grad_norm": 2.985004186630249, + "learning_rate": 4.488007380073801e-06, + "loss": 1.5681, + "step": 478 + }, + { + "epoch": 0.5961418792781581, + "grad_norm": 2.8348851203918457, + "learning_rate": 4.485701107011071e-06, + "loss": 1.6214, + "step": 479 + }, + { + "epoch": 0.5973864343497199, + "grad_norm": 2.580967664718628, + "learning_rate": 4.483394833948339e-06, + "loss": 1.5753, + "step": 480 + }, + { + "epoch": 0.5973864343497199, + "eval_loss": 1.7542102336883545, + "eval_runtime": 49.5092, + "eval_samples_per_second": 20.198, + "eval_steps_per_second": 0.848, + "step": 480 + }, + { + "epoch": 0.5986309894212819, + "grad_norm": 3.0104517936706543, + "learning_rate": 4.4810885608856096e-06, + "loss": 1.6282, + "step": 481 + }, + { + "epoch": 0.5998755444928439, + "grad_norm": 2.9261820316314697, + "learning_rate": 4.478782287822879e-06, + "loss": 1.5811, + "step": 482 + }, + { + "epoch": 0.6011200995644057, + "grad_norm": 2.9920477867126465, + "learning_rate": 4.4764760147601476e-06, + "loss": 1.5905, + "step": 483 + }, + { + "epoch": 0.6023646546359677, + "grad_norm": 3.046887159347534, + "learning_rate": 4.474169741697417e-06, + "loss": 1.6587, + "step": 484 + }, + { + "epoch": 0.6036092097075295, + "grad_norm": 2.9660050868988037, + "learning_rate": 4.471863468634686e-06, + "loss": 1.5888, + "step": 485 + }, + { + "epoch": 0.6048537647790915, + "grad_norm": 2.770362377166748, + "learning_rate": 4.469557195571957e-06, + "loss": 1.5925, + "step": 486 + }, + { + "epoch": 0.6060983198506534, + "grad_norm": 2.621699094772339, + "learning_rate": 4.467250922509225e-06, + "loss": 1.5671, + "step": 487 + }, + { + "epoch": 0.6073428749222153, + "grad_norm": 2.749340295791626, + "learning_rate": 4.464944649446495e-06, + "loss": 1.5672, + "step": 488 + }, + { + "epoch": 0.6085874299937772, + "grad_norm": 3.091601848602295, + "learning_rate": 4.462638376383764e-06, + "loss": 1.5965, + "step": 489 + }, + { + "epoch": 0.6098319850653391, + "grad_norm": 2.6989245414733887, + "learning_rate": 4.4603321033210335e-06, + "loss": 1.5845, + "step": 490 + }, + { + "epoch": 0.6098319850653391, + "eval_loss": 1.7461379766464233, + "eval_runtime": 52.879, + "eval_samples_per_second": 18.911, + "eval_steps_per_second": 0.794, + "step": 490 + }, + { + "epoch": 0.611076540136901, + "grad_norm": 2.8998405933380127, + "learning_rate": 4.458025830258303e-06, + "loss": 1.5764, + "step": 491 + }, + { + "epoch": 0.612321095208463, + "grad_norm": 2.6035947799682617, + "learning_rate": 4.455719557195572e-06, + "loss": 1.5988, + "step": 492 + }, + { + "epoch": 0.6135656502800249, + "grad_norm": 2.712857484817505, + "learning_rate": 4.453413284132842e-06, + "loss": 1.6276, + "step": 493 + }, + { + "epoch": 0.6148102053515868, + "grad_norm": 2.73052978515625, + "learning_rate": 4.451107011070111e-06, + "loss": 1.5907, + "step": 494 + }, + { + "epoch": 0.6160547604231488, + "grad_norm": 2.8072566986083984, + "learning_rate": 4.448800738007381e-06, + "loss": 1.5771, + "step": 495 + }, + { + "epoch": 0.6172993154947106, + "grad_norm": 2.6016132831573486, + "learning_rate": 4.446494464944649e-06, + "loss": 1.5983, + "step": 496 + }, + { + "epoch": 0.6185438705662726, + "grad_norm": 2.9697721004486084, + "learning_rate": 4.4441881918819195e-06, + "loss": 1.5974, + "step": 497 + }, + { + "epoch": 0.6197884256378344, + "grad_norm": 2.6966018676757812, + "learning_rate": 4.441881918819189e-06, + "loss": 1.6042, + "step": 498 + }, + { + "epoch": 0.6210329807093964, + "grad_norm": 2.871964931488037, + "learning_rate": 4.439575645756458e-06, + "loss": 1.5668, + "step": 499 + }, + { + "epoch": 0.6222775357809583, + "grad_norm": 2.791602611541748, + "learning_rate": 4.437269372693727e-06, + "loss": 1.5222, + "step": 500 + }, + { + "epoch": 0.6222775357809583, + "eval_loss": 1.756259799003601, + "eval_runtime": 53.0959, + "eval_samples_per_second": 18.834, + "eval_steps_per_second": 0.791, + "step": 500 + }, + { + "epoch": 0.6235220908525202, + "grad_norm": 2.6534979343414307, + "learning_rate": 4.434963099630996e-06, + "loss": 1.5618, + "step": 501 + }, + { + "epoch": 0.6247666459240822, + "grad_norm": 3.0166008472442627, + "learning_rate": 4.432656826568267e-06, + "loss": 1.639, + "step": 502 + }, + { + "epoch": 0.6260112009956441, + "grad_norm": 2.7879786491394043, + "learning_rate": 4.430350553505535e-06, + "loss": 1.5521, + "step": 503 + }, + { + "epoch": 0.627255756067206, + "grad_norm": 2.695650339126587, + "learning_rate": 4.428044280442805e-06, + "loss": 1.6086, + "step": 504 + }, + { + "epoch": 0.6285003111387679, + "grad_norm": 2.8340883255004883, + "learning_rate": 4.425738007380074e-06, + "loss": 1.583, + "step": 505 + }, + { + "epoch": 0.6297448662103298, + "grad_norm": 2.872870683670044, + "learning_rate": 4.4234317343173434e-06, + "loss": 1.5633, + "step": 506 + }, + { + "epoch": 0.6309894212818917, + "grad_norm": 2.840087652206421, + "learning_rate": 4.421125461254613e-06, + "loss": 1.603, + "step": 507 + }, + { + "epoch": 0.6322339763534537, + "grad_norm": 2.86578631401062, + "learning_rate": 4.418819188191882e-06, + "loss": 1.6031, + "step": 508 + }, + { + "epoch": 0.6334785314250155, + "grad_norm": 2.419501543045044, + "learning_rate": 4.416512915129152e-06, + "loss": 1.5975, + "step": 509 + }, + { + "epoch": 0.6347230864965775, + "grad_norm": 2.6204769611358643, + "learning_rate": 4.414206642066421e-06, + "loss": 1.608, + "step": 510 + }, + { + "epoch": 0.6347230864965775, + "eval_loss": 1.7449734210968018, + "eval_runtime": 44.3185, + "eval_samples_per_second": 22.564, + "eval_steps_per_second": 0.948, + "step": 510 + }, + { + "epoch": 0.6359676415681393, + "grad_norm": 2.837980270385742, + "learning_rate": 4.4119003690036905e-06, + "loss": 1.625, + "step": 511 + }, + { + "epoch": 0.6372121966397013, + "grad_norm": 2.589782238006592, + "learning_rate": 4.40959409594096e-06, + "loss": 1.5976, + "step": 512 + }, + { + "epoch": 0.6384567517112633, + "grad_norm": 2.596064567565918, + "learning_rate": 4.407287822878229e-06, + "loss": 1.5568, + "step": 513 + }, + { + "epoch": 0.6397013067828251, + "grad_norm": 3.031325578689575, + "learning_rate": 4.404981549815499e-06, + "loss": 1.5744, + "step": 514 + }, + { + "epoch": 0.6409458618543871, + "grad_norm": 2.581284761428833, + "learning_rate": 4.402675276752768e-06, + "loss": 1.5822, + "step": 515 + }, + { + "epoch": 0.642190416925949, + "grad_norm": 2.7712416648864746, + "learning_rate": 4.400369003690037e-06, + "loss": 1.606, + "step": 516 + }, + { + "epoch": 0.6434349719975109, + "grad_norm": 2.5830018520355225, + "learning_rate": 4.398062730627306e-06, + "loss": 1.5845, + "step": 517 + }, + { + "epoch": 0.6446795270690728, + "grad_norm": 2.6646499633789062, + "learning_rate": 4.3957564575645765e-06, + "loss": 1.6323, + "step": 518 + }, + { + "epoch": 0.6459240821406347, + "grad_norm": 2.614750862121582, + "learning_rate": 4.393450184501845e-06, + "loss": 1.5447, + "step": 519 + }, + { + "epoch": 0.6471686372121966, + "grad_norm": 2.6826741695404053, + "learning_rate": 4.3911439114391145e-06, + "loss": 1.6081, + "step": 520 + }, + { + "epoch": 0.6471686372121966, + "eval_loss": 1.7463606595993042, + "eval_runtime": 45.7383, + "eval_samples_per_second": 21.864, + "eval_steps_per_second": 0.918, + "step": 520 + }, + { + "epoch": 0.6484131922837586, + "grad_norm": 2.9165496826171875, + "learning_rate": 4.388837638376384e-06, + "loss": 1.5853, + "step": 521 + }, + { + "epoch": 0.6496577473553204, + "grad_norm": 2.789862871170044, + "learning_rate": 4.386531365313653e-06, + "loss": 1.6115, + "step": 522 + }, + { + "epoch": 0.6509023024268824, + "grad_norm": 2.690093994140625, + "learning_rate": 4.384225092250923e-06, + "loss": 1.5625, + "step": 523 + }, + { + "epoch": 0.6521468574984443, + "grad_norm": 2.660036087036133, + "learning_rate": 4.381918819188192e-06, + "loss": 1.6065, + "step": 524 + }, + { + "epoch": 0.6533914125700062, + "grad_norm": 2.5939362049102783, + "learning_rate": 4.379612546125462e-06, + "loss": 1.5593, + "step": 525 + }, + { + "epoch": 0.6546359676415682, + "grad_norm": 2.6998841762542725, + "learning_rate": 4.377306273062731e-06, + "loss": 1.6168, + "step": 526 + }, + { + "epoch": 0.65588052271313, + "grad_norm": 2.5756912231445312, + "learning_rate": 4.3750000000000005e-06, + "loss": 1.528, + "step": 527 + }, + { + "epoch": 0.657125077784692, + "grad_norm": 2.6959640979766846, + "learning_rate": 4.37269372693727e-06, + "loss": 1.5987, + "step": 528 + }, + { + "epoch": 0.6583696328562539, + "grad_norm": 2.773494005203247, + "learning_rate": 4.370387453874539e-06, + "loss": 1.6018, + "step": 529 + }, + { + "epoch": 0.6596141879278158, + "grad_norm": 2.8194501399993896, + "learning_rate": 4.368081180811809e-06, + "loss": 1.6082, + "step": 530 + }, + { + "epoch": 0.6596141879278158, + "eval_loss": 1.7327930927276611, + "eval_runtime": 46.7893, + "eval_samples_per_second": 21.372, + "eval_steps_per_second": 0.898, + "step": 530 + }, + { + "epoch": 0.6608587429993777, + "grad_norm": 2.721564292907715, + "learning_rate": 4.365774907749078e-06, + "loss": 1.5549, + "step": 531 + }, + { + "epoch": 0.6621032980709396, + "grad_norm": 2.6635236740112305, + "learning_rate": 4.363468634686347e-06, + "loss": 1.6016, + "step": 532 + }, + { + "epoch": 0.6633478531425016, + "grad_norm": 2.6541085243225098, + "learning_rate": 4.361162361623616e-06, + "loss": 1.5574, + "step": 533 + }, + { + "epoch": 0.6645924082140635, + "grad_norm": 2.9270925521850586, + "learning_rate": 4.3588560885608864e-06, + "loss": 1.5722, + "step": 534 + }, + { + "epoch": 0.6658369632856254, + "grad_norm": 2.6041259765625, + "learning_rate": 4.356549815498156e-06, + "loss": 1.6181, + "step": 535 + }, + { + "epoch": 0.6670815183571873, + "grad_norm": 2.7029967308044434, + "learning_rate": 4.354243542435424e-06, + "loss": 1.5576, + "step": 536 + }, + { + "epoch": 0.6683260734287492, + "grad_norm": 2.805567979812622, + "learning_rate": 4.351937269372694e-06, + "loss": 1.6279, + "step": 537 + }, + { + "epoch": 0.6695706285003111, + "grad_norm": 2.6513259410858154, + "learning_rate": 4.349630996309963e-06, + "loss": 1.5626, + "step": 538 + }, + { + "epoch": 0.6708151835718731, + "grad_norm": 2.7320876121520996, + "learning_rate": 4.347324723247233e-06, + "loss": 1.6, + "step": 539 + }, + { + "epoch": 0.6720597386434349, + "grad_norm": 2.9504754543304443, + "learning_rate": 4.345018450184502e-06, + "loss": 1.6755, + "step": 540 + }, + { + "epoch": 0.6720597386434349, + "eval_loss": 1.733923077583313, + "eval_runtime": 45.4809, + "eval_samples_per_second": 21.987, + "eval_steps_per_second": 0.923, + "step": 540 + }, + { + "epoch": 0.6733042937149969, + "grad_norm": 2.7561354637145996, + "learning_rate": 4.3427121771217715e-06, + "loss": 1.5624, + "step": 541 + }, + { + "epoch": 0.6745488487865589, + "grad_norm": 2.612337827682495, + "learning_rate": 4.340405904059041e-06, + "loss": 1.5806, + "step": 542 + }, + { + "epoch": 0.6757934038581207, + "grad_norm": 3.2763278484344482, + "learning_rate": 4.33809963099631e-06, + "loss": 1.5939, + "step": 543 + }, + { + "epoch": 0.6770379589296827, + "grad_norm": 2.632627248764038, + "learning_rate": 4.33579335793358e-06, + "loss": 1.564, + "step": 544 + }, + { + "epoch": 0.6782825140012445, + "grad_norm": 2.9776413440704346, + "learning_rate": 4.333487084870848e-06, + "loss": 1.5624, + "step": 545 + }, + { + "epoch": 0.6795270690728065, + "grad_norm": 3.083935260772705, + "learning_rate": 4.331180811808119e-06, + "loss": 1.5957, + "step": 546 + }, + { + "epoch": 0.6807716241443684, + "grad_norm": 2.8425586223602295, + "learning_rate": 4.328874538745388e-06, + "loss": 1.5559, + "step": 547 + }, + { + "epoch": 0.6820161792159303, + "grad_norm": 3.0011188983917236, + "learning_rate": 4.3265682656826575e-06, + "loss": 1.5308, + "step": 548 + }, + { + "epoch": 0.6832607342874922, + "grad_norm": 2.7645163536071777, + "learning_rate": 4.324261992619926e-06, + "loss": 1.6114, + "step": 549 + }, + { + "epoch": 0.6845052893590542, + "grad_norm": 2.8784844875335693, + "learning_rate": 4.321955719557196e-06, + "loss": 1.5591, + "step": 550 + }, + { + "epoch": 0.6845052893590542, + "eval_loss": 1.7328442335128784, + "eval_runtime": 45.5776, + "eval_samples_per_second": 21.941, + "eval_steps_per_second": 0.922, + "step": 550 + }, + { + "epoch": 0.685749844430616, + "grad_norm": 3.1323604583740234, + "learning_rate": 4.319649446494466e-06, + "loss": 1.5626, + "step": 551 + }, + { + "epoch": 0.686994399502178, + "grad_norm": 2.7367660999298096, + "learning_rate": 4.317343173431734e-06, + "loss": 1.5914, + "step": 552 + }, + { + "epoch": 0.6882389545737398, + "grad_norm": 3.056155204772949, + "learning_rate": 4.315036900369004e-06, + "loss": 1.6014, + "step": 553 + }, + { + "epoch": 0.6894835096453018, + "grad_norm": 2.7824926376342773, + "learning_rate": 4.312730627306273e-06, + "loss": 1.5508, + "step": 554 + }, + { + "epoch": 0.6907280647168638, + "grad_norm": 2.629610538482666, + "learning_rate": 4.310424354243543e-06, + "loss": 1.5385, + "step": 555 + }, + { + "epoch": 0.6919726197884256, + "grad_norm": 2.753021240234375, + "learning_rate": 4.308118081180812e-06, + "loss": 1.6211, + "step": 556 + }, + { + "epoch": 0.6932171748599876, + "grad_norm": 2.8113718032836914, + "learning_rate": 4.3058118081180815e-06, + "loss": 1.5904, + "step": 557 + }, + { + "epoch": 0.6944617299315494, + "grad_norm": 2.849471092224121, + "learning_rate": 4.303505535055351e-06, + "loss": 1.6126, + "step": 558 + }, + { + "epoch": 0.6957062850031114, + "grad_norm": 2.6744682788848877, + "learning_rate": 4.30119926199262e-06, + "loss": 1.5743, + "step": 559 + }, + { + "epoch": 0.6969508400746733, + "grad_norm": 2.7900655269622803, + "learning_rate": 4.29889298892989e-06, + "loss": 1.5865, + "step": 560 + }, + { + "epoch": 0.6969508400746733, + "eval_loss": 1.724618673324585, + "eval_runtime": 44.2181, + "eval_samples_per_second": 22.615, + "eval_steps_per_second": 0.95, + "step": 560 + }, + { + "epoch": 0.6981953951462352, + "grad_norm": 2.7363393306732178, + "learning_rate": 4.296586715867159e-06, + "loss": 1.5347, + "step": 561 + }, + { + "epoch": 0.6994399502177971, + "grad_norm": 2.696748733520508, + "learning_rate": 4.2942804428044286e-06, + "loss": 1.5463, + "step": 562 + }, + { + "epoch": 0.7006845052893591, + "grad_norm": 2.5793895721435547, + "learning_rate": 4.291974169741698e-06, + "loss": 1.6231, + "step": 563 + }, + { + "epoch": 0.701929060360921, + "grad_norm": 2.576230764389038, + "learning_rate": 4.289667896678967e-06, + "loss": 1.5663, + "step": 564 + }, + { + "epoch": 0.7031736154324829, + "grad_norm": 2.718759298324585, + "learning_rate": 4.287361623616236e-06, + "loss": 1.5316, + "step": 565 + }, + { + "epoch": 0.7044181705040448, + "grad_norm": 2.651751756668091, + "learning_rate": 4.285055350553506e-06, + "loss": 1.5777, + "step": 566 + }, + { + "epoch": 0.7056627255756067, + "grad_norm": 2.931264638900757, + "learning_rate": 4.282749077490776e-06, + "loss": 1.5647, + "step": 567 + }, + { + "epoch": 0.7069072806471687, + "grad_norm": 2.8062546253204346, + "learning_rate": 4.280442804428044e-06, + "loss": 1.5685, + "step": 568 + }, + { + "epoch": 0.7081518357187305, + "grad_norm": 2.7551019191741943, + "learning_rate": 4.278136531365314e-06, + "loss": 1.5659, + "step": 569 + }, + { + "epoch": 0.7093963907902925, + "grad_norm": 2.799724578857422, + "learning_rate": 4.275830258302583e-06, + "loss": 1.5917, + "step": 570 + }, + { + "epoch": 0.7093963907902925, + "eval_loss": 1.7217484712600708, + "eval_runtime": 44.5312, + "eval_samples_per_second": 22.456, + "eval_steps_per_second": 0.943, + "step": 570 + }, + { + "epoch": 0.7106409458618543, + "grad_norm": 2.8496406078338623, + "learning_rate": 4.273523985239853e-06, + "loss": 1.5607, + "step": 571 + }, + { + "epoch": 0.7118855009334163, + "grad_norm": 2.5612425804138184, + "learning_rate": 4.271217712177122e-06, + "loss": 1.5496, + "step": 572 + }, + { + "epoch": 0.7131300560049783, + "grad_norm": 2.6605045795440674, + "learning_rate": 4.268911439114391e-06, + "loss": 1.5371, + "step": 573 + }, + { + "epoch": 0.7143746110765401, + "grad_norm": 2.823802947998047, + "learning_rate": 4.266605166051661e-06, + "loss": 1.5699, + "step": 574 + }, + { + "epoch": 0.7156191661481021, + "grad_norm": 2.8212928771972656, + "learning_rate": 4.26429889298893e-06, + "loss": 1.5529, + "step": 575 + }, + { + "epoch": 0.716863721219664, + "grad_norm": 2.7193386554718018, + "learning_rate": 4.2619926199262e-06, + "loss": 1.5723, + "step": 576 + }, + { + "epoch": 0.7181082762912259, + "grad_norm": 2.8348941802978516, + "learning_rate": 4.259686346863469e-06, + "loss": 1.5212, + "step": 577 + }, + { + "epoch": 0.7193528313627878, + "grad_norm": 2.7177276611328125, + "learning_rate": 4.2573800738007385e-06, + "loss": 1.5688, + "step": 578 + }, + { + "epoch": 0.7205973864343497, + "grad_norm": 2.675734043121338, + "learning_rate": 4.255073800738008e-06, + "loss": 1.5345, + "step": 579 + }, + { + "epoch": 0.7218419415059116, + "grad_norm": 2.640942335128784, + "learning_rate": 4.252767527675277e-06, + "loss": 1.5874, + "step": 580 + }, + { + "epoch": 0.7218419415059116, + "eval_loss": 1.72821044921875, + "eval_runtime": 45.9409, + "eval_samples_per_second": 21.767, + "eval_steps_per_second": 0.914, + "step": 580 + }, + { + "epoch": 0.7230864965774736, + "grad_norm": 2.785066604614258, + "learning_rate": 4.250461254612546e-06, + "loss": 1.5685, + "step": 581 + }, + { + "epoch": 0.7243310516490354, + "grad_norm": 2.8149311542510986, + "learning_rate": 4.248154981549816e-06, + "loss": 1.5106, + "step": 582 + }, + { + "epoch": 0.7255756067205974, + "grad_norm": 2.5781238079071045, + "learning_rate": 4.245848708487086e-06, + "loss": 1.5895, + "step": 583 + }, + { + "epoch": 0.7268201617921594, + "grad_norm": 2.6275103092193604, + "learning_rate": 4.243542435424355e-06, + "loss": 1.601, + "step": 584 + }, + { + "epoch": 0.7280647168637212, + "grad_norm": 2.5282299518585205, + "learning_rate": 4.241236162361624e-06, + "loss": 1.5376, + "step": 585 + }, + { + "epoch": 0.7293092719352832, + "grad_norm": 2.7331771850585938, + "learning_rate": 4.238929889298893e-06, + "loss": 1.5855, + "step": 586 + }, + { + "epoch": 0.730553827006845, + "grad_norm": 2.6941936016082764, + "learning_rate": 4.236623616236163e-06, + "loss": 1.5937, + "step": 587 + }, + { + "epoch": 0.731798382078407, + "grad_norm": 2.924062728881836, + "learning_rate": 4.234317343173432e-06, + "loss": 1.5914, + "step": 588 + }, + { + "epoch": 0.7330429371499689, + "grad_norm": 2.687530279159546, + "learning_rate": 4.232011070110701e-06, + "loss": 1.5598, + "step": 589 + }, + { + "epoch": 0.7342874922215308, + "grad_norm": 2.7230913639068604, + "learning_rate": 4.229704797047971e-06, + "loss": 1.6048, + "step": 590 + }, + { + "epoch": 0.7342874922215308, + "eval_loss": 1.7240839004516602, + "eval_runtime": 46.2617, + "eval_samples_per_second": 21.616, + "eval_steps_per_second": 0.908, + "step": 590 + }, + { + "epoch": 0.7355320472930927, + "grad_norm": 2.833481788635254, + "learning_rate": 4.22739852398524e-06, + "loss": 1.545, + "step": 591 + }, + { + "epoch": 0.7367766023646546, + "grad_norm": 2.7287282943725586, + "learning_rate": 4.2250922509225096e-06, + "loss": 1.5795, + "step": 592 + }, + { + "epoch": 0.7380211574362165, + "grad_norm": 2.847174882888794, + "learning_rate": 4.222785977859779e-06, + "loss": 1.5469, + "step": 593 + }, + { + "epoch": 0.7392657125077785, + "grad_norm": 2.9214589595794678, + "learning_rate": 4.220479704797048e-06, + "loss": 1.5791, + "step": 594 + }, + { + "epoch": 0.7405102675793404, + "grad_norm": 2.7506580352783203, + "learning_rate": 4.218173431734318e-06, + "loss": 1.6578, + "step": 595 + }, + { + "epoch": 0.7417548226509023, + "grad_norm": 2.7006874084472656, + "learning_rate": 4.215867158671587e-06, + "loss": 1.5598, + "step": 596 + }, + { + "epoch": 0.7429993777224643, + "grad_norm": 2.795417547225952, + "learning_rate": 4.213560885608857e-06, + "loss": 1.5862, + "step": 597 + }, + { + "epoch": 0.7442439327940261, + "grad_norm": 2.7418057918548584, + "learning_rate": 4.211254612546125e-06, + "loss": 1.5717, + "step": 598 + }, + { + "epoch": 0.7454884878655881, + "grad_norm": 2.884915590286255, + "learning_rate": 4.2089483394833955e-06, + "loss": 1.5828, + "step": 599 + }, + { + "epoch": 0.7467330429371499, + "grad_norm": 2.752328395843506, + "learning_rate": 4.206642066420665e-06, + "loss": 1.576, + "step": 600 + }, + { + "epoch": 0.7467330429371499, + "eval_loss": 1.7281302213668823, + "eval_runtime": 45.1978, + "eval_samples_per_second": 22.125, + "eval_steps_per_second": 0.929, + "step": 600 + }, + { + "epoch": 0.7479775980087119, + "grad_norm": 2.9103147983551025, + "learning_rate": 4.2043357933579335e-06, + "loss": 1.4959, + "step": 601 + }, + { + "epoch": 0.7492221530802738, + "grad_norm": 2.5317649841308594, + "learning_rate": 4.202029520295203e-06, + "loss": 1.5092, + "step": 602 + }, + { + "epoch": 0.7504667081518357, + "grad_norm": 2.7383415699005127, + "learning_rate": 4.199723247232473e-06, + "loss": 1.4696, + "step": 603 + }, + { + "epoch": 0.7517112632233977, + "grad_norm": 2.890305519104004, + "learning_rate": 4.197416974169742e-06, + "loss": 1.511, + "step": 604 + }, + { + "epoch": 0.7529558182949595, + "grad_norm": 2.7425734996795654, + "learning_rate": 4.195110701107011e-06, + "loss": 1.5658, + "step": 605 + }, + { + "epoch": 0.7542003733665215, + "grad_norm": 2.657726764678955, + "learning_rate": 4.192804428044281e-06, + "loss": 1.5648, + "step": 606 + }, + { + "epoch": 0.7554449284380834, + "grad_norm": 3.181201696395874, + "learning_rate": 4.19049815498155e-06, + "loss": 1.583, + "step": 607 + }, + { + "epoch": 0.7566894835096453, + "grad_norm": 2.665555715560913, + "learning_rate": 4.1881918819188195e-06, + "loss": 1.5439, + "step": 608 + }, + { + "epoch": 0.7579340385812072, + "grad_norm": 2.7328009605407715, + "learning_rate": 4.185885608856089e-06, + "loss": 1.5264, + "step": 609 + }, + { + "epoch": 0.7591785936527692, + "grad_norm": 2.7027499675750732, + "learning_rate": 4.183579335793358e-06, + "loss": 1.5327, + "step": 610 + }, + { + "epoch": 0.7591785936527692, + "eval_loss": 1.7161823511123657, + "eval_runtime": 42.4811, + "eval_samples_per_second": 23.54, + "eval_steps_per_second": 0.989, + "step": 610 + }, + { + "epoch": 0.760423148724331, + "grad_norm": 2.6982595920562744, + "learning_rate": 4.181273062730628e-06, + "loss": 1.585, + "step": 611 + }, + { + "epoch": 0.761667703795893, + "grad_norm": 2.780768394470215, + "learning_rate": 4.178966789667897e-06, + "loss": 1.5754, + "step": 612 + }, + { + "epoch": 0.7629122588674548, + "grad_norm": 2.7140064239501953, + "learning_rate": 4.176660516605167e-06, + "loss": 1.5649, + "step": 613 + }, + { + "epoch": 0.7641568139390168, + "grad_norm": 2.60182523727417, + "learning_rate": 4.174354243542435e-06, + "loss": 1.5879, + "step": 614 + }, + { + "epoch": 0.7654013690105788, + "grad_norm": 2.6529550552368164, + "learning_rate": 4.1720479704797054e-06, + "loss": 1.5418, + "step": 615 + }, + { + "epoch": 0.7666459240821406, + "grad_norm": 2.7936999797821045, + "learning_rate": 4.169741697416975e-06, + "loss": 1.5993, + "step": 616 + }, + { + "epoch": 0.7678904791537026, + "grad_norm": 2.7894914150238037, + "learning_rate": 4.1674354243542434e-06, + "loss": 1.5962, + "step": 617 + }, + { + "epoch": 0.7691350342252644, + "grad_norm": 2.8094332218170166, + "learning_rate": 4.165129151291513e-06, + "loss": 1.5898, + "step": 618 + }, + { + "epoch": 0.7703795892968264, + "grad_norm": 2.9678053855895996, + "learning_rate": 4.162822878228783e-06, + "loss": 1.5602, + "step": 619 + }, + { + "epoch": 0.7716241443683883, + "grad_norm": 2.7238247394561768, + "learning_rate": 4.1605166051660526e-06, + "loss": 1.554, + "step": 620 + }, + { + "epoch": 0.7716241443683883, + "eval_loss": 1.720330834388733, + "eval_runtime": 47.0576, + "eval_samples_per_second": 21.251, + "eval_steps_per_second": 0.893, + "step": 620 + }, + { + "epoch": 0.7728686994399502, + "grad_norm": 2.6484885215759277, + "learning_rate": 4.158210332103321e-06, + "loss": 1.5464, + "step": 621 + }, + { + "epoch": 0.7741132545115121, + "grad_norm": 2.722929000854492, + "learning_rate": 4.1559040590405906e-06, + "loss": 1.5471, + "step": 622 + }, + { + "epoch": 0.7753578095830741, + "grad_norm": 2.842886209487915, + "learning_rate": 4.15359778597786e-06, + "loss": 1.5512, + "step": 623 + }, + { + "epoch": 0.776602364654636, + "grad_norm": 2.647134780883789, + "learning_rate": 4.151291512915129e-06, + "loss": 1.5264, + "step": 624 + }, + { + "epoch": 0.7778469197261979, + "grad_norm": 2.763277053833008, + "learning_rate": 4.148985239852399e-06, + "loss": 1.5102, + "step": 625 + }, + { + "epoch": 0.7790914747977598, + "grad_norm": 2.9078922271728516, + "learning_rate": 4.146678966789668e-06, + "loss": 1.5538, + "step": 626 + }, + { + "epoch": 0.7803360298693217, + "grad_norm": 2.60658860206604, + "learning_rate": 4.144372693726938e-06, + "loss": 1.486, + "step": 627 + }, + { + "epoch": 0.7815805849408837, + "grad_norm": 2.5513789653778076, + "learning_rate": 4.142066420664207e-06, + "loss": 1.5323, + "step": 628 + }, + { + "epoch": 0.7828251400124455, + "grad_norm": 3.0412704944610596, + "learning_rate": 4.1397601476014765e-06, + "loss": 1.5835, + "step": 629 + }, + { + "epoch": 0.7840696950840075, + "grad_norm": 2.9483704566955566, + "learning_rate": 4.137453874538745e-06, + "loss": 1.5849, + "step": 630 + }, + { + "epoch": 0.7840696950840075, + "eval_loss": 1.7138603925704956, + "eval_runtime": 47.7618, + "eval_samples_per_second": 20.937, + "eval_steps_per_second": 0.879, + "step": 630 + }, + { + "epoch": 0.7853142501555694, + "grad_norm": 2.784747362136841, + "learning_rate": 4.135147601476015e-06, + "loss": 1.5625, + "step": 631 + }, + { + "epoch": 0.7865588052271313, + "grad_norm": 2.775791883468628, + "learning_rate": 4.132841328413285e-06, + "loss": 1.5827, + "step": 632 + }, + { + "epoch": 0.7878033602986932, + "grad_norm": 2.867126226425171, + "learning_rate": 4.130535055350554e-06, + "loss": 1.5174, + "step": 633 + }, + { + "epoch": 0.7890479153702551, + "grad_norm": 2.7829818725585938, + "learning_rate": 4.128228782287823e-06, + "loss": 1.5853, + "step": 634 + }, + { + "epoch": 0.790292470441817, + "grad_norm": 2.630842924118042, + "learning_rate": 4.125922509225092e-06, + "loss": 1.5853, + "step": 635 + }, + { + "epoch": 0.791537025513379, + "grad_norm": 2.8100547790527344, + "learning_rate": 4.1236162361623625e-06, + "loss": 1.5162, + "step": 636 + }, + { + "epoch": 0.7927815805849409, + "grad_norm": 2.5983335971832275, + "learning_rate": 4.121309963099631e-06, + "loss": 1.5926, + "step": 637 + }, + { + "epoch": 0.7940261356565028, + "grad_norm": 2.826624870300293, + "learning_rate": 4.1190036900369005e-06, + "loss": 1.5527, + "step": 638 + }, + { + "epoch": 0.7952706907280647, + "grad_norm": 2.863215684890747, + "learning_rate": 4.11669741697417e-06, + "loss": 1.5514, + "step": 639 + }, + { + "epoch": 0.7965152457996266, + "grad_norm": 2.7492613792419434, + "learning_rate": 4.114391143911439e-06, + "loss": 1.5101, + "step": 640 + }, + { + "epoch": 0.7965152457996266, + "eval_loss": 1.7091425657272339, + "eval_runtime": 46.5279, + "eval_samples_per_second": 21.492, + "eval_steps_per_second": 0.903, + "step": 640 + }, + { + "epoch": 0.7977598008711886, + "grad_norm": 2.9577860832214355, + "learning_rate": 4.112084870848709e-06, + "loss": 1.5496, + "step": 641 + }, + { + "epoch": 0.7990043559427504, + "grad_norm": 2.612668991088867, + "learning_rate": 4.109778597785978e-06, + "loss": 1.5204, + "step": 642 + }, + { + "epoch": 0.8002489110143124, + "grad_norm": 2.5427377223968506, + "learning_rate": 4.107472324723248e-06, + "loss": 1.5038, + "step": 643 + }, + { + "epoch": 0.8014934660858744, + "grad_norm": 2.7057406902313232, + "learning_rate": 4.105166051660517e-06, + "loss": 1.5831, + "step": 644 + }, + { + "epoch": 0.8027380211574362, + "grad_norm": 2.664353609085083, + "learning_rate": 4.1028597785977864e-06, + "loss": 1.555, + "step": 645 + }, + { + "epoch": 0.8039825762289982, + "grad_norm": 2.5452728271484375, + "learning_rate": 4.100553505535056e-06, + "loss": 1.5847, + "step": 646 + }, + { + "epoch": 0.80522713130056, + "grad_norm": 2.5370709896087646, + "learning_rate": 4.098247232472325e-06, + "loss": 1.5396, + "step": 647 + }, + { + "epoch": 0.806471686372122, + "grad_norm": 2.730109930038452, + "learning_rate": 4.095940959409595e-06, + "loss": 1.5234, + "step": 648 + }, + { + "epoch": 0.8077162414436839, + "grad_norm": 2.586108446121216, + "learning_rate": 4.093634686346864e-06, + "loss": 1.5885, + "step": 649 + }, + { + "epoch": 0.8089607965152458, + "grad_norm": 2.557053804397583, + "learning_rate": 4.091328413284133e-06, + "loss": 1.483, + "step": 650 + }, + { + "epoch": 0.8089607965152458, + "eval_loss": 1.7204149961471558, + "eval_runtime": 48.0041, + "eval_samples_per_second": 20.832, + "eval_steps_per_second": 0.875, + "step": 650 + }, + { + "epoch": 0.8102053515868077, + "grad_norm": 2.673672676086426, + "learning_rate": 4.089022140221402e-06, + "loss": 1.4837, + "step": 651 + }, + { + "epoch": 0.8114499066583696, + "grad_norm": 2.6465933322906494, + "learning_rate": 4.086715867158672e-06, + "loss": 1.5219, + "step": 652 + }, + { + "epoch": 0.8126944617299315, + "grad_norm": 2.6897783279418945, + "learning_rate": 4.084409594095941e-06, + "loss": 1.5116, + "step": 653 + }, + { + "epoch": 0.8139390168014935, + "grad_norm": 2.644369602203369, + "learning_rate": 4.08210332103321e-06, + "loss": 1.5141, + "step": 654 + }, + { + "epoch": 0.8151835718730553, + "grad_norm": 2.7475595474243164, + "learning_rate": 4.07979704797048e-06, + "loss": 1.5509, + "step": 655 + }, + { + "epoch": 0.8164281269446173, + "grad_norm": 2.72043514251709, + "learning_rate": 4.077490774907749e-06, + "loss": 1.6098, + "step": 656 + }, + { + "epoch": 0.8176726820161793, + "grad_norm": 2.648069381713867, + "learning_rate": 4.075184501845019e-06, + "loss": 1.5368, + "step": 657 + }, + { + "epoch": 0.8189172370877411, + "grad_norm": 2.644893169403076, + "learning_rate": 4.072878228782288e-06, + "loss": 1.492, + "step": 658 + }, + { + "epoch": 0.8201617921593031, + "grad_norm": 2.9441068172454834, + "learning_rate": 4.0705719557195575e-06, + "loss": 1.5904, + "step": 659 + }, + { + "epoch": 0.8214063472308649, + "grad_norm": 2.752913475036621, + "learning_rate": 4.068265682656827e-06, + "loss": 1.5544, + "step": 660 + }, + { + "epoch": 0.8214063472308649, + "eval_loss": 1.709177017211914, + "eval_runtime": 48.0401, + "eval_samples_per_second": 20.816, + "eval_steps_per_second": 0.874, + "step": 660 + }, + { + "epoch": 0.8226509023024269, + "grad_norm": 2.6959197521209717, + "learning_rate": 4.065959409594096e-06, + "loss": 1.5553, + "step": 661 + }, + { + "epoch": 0.8238954573739888, + "grad_norm": 2.8757998943328857, + "learning_rate": 4.063653136531366e-06, + "loss": 1.554, + "step": 662 + }, + { + "epoch": 0.8251400124455507, + "grad_norm": 2.7723867893218994, + "learning_rate": 4.061346863468635e-06, + "loss": 1.5499, + "step": 663 + }, + { + "epoch": 0.8263845675171126, + "grad_norm": 2.7620463371276855, + "learning_rate": 4.059040590405905e-06, + "loss": 1.5962, + "step": 664 + }, + { + "epoch": 0.8276291225886746, + "grad_norm": 2.8601174354553223, + "learning_rate": 4.056734317343174e-06, + "loss": 1.5218, + "step": 665 + }, + { + "epoch": 0.8288736776602365, + "grad_norm": 2.81356143951416, + "learning_rate": 4.054428044280443e-06, + "loss": 1.5382, + "step": 666 + }, + { + "epoch": 0.8301182327317984, + "grad_norm": 2.904679298400879, + "learning_rate": 4.052121771217712e-06, + "loss": 1.5314, + "step": 667 + }, + { + "epoch": 0.8313627878033603, + "grad_norm": 2.798236846923828, + "learning_rate": 4.049815498154982e-06, + "loss": 1.5989, + "step": 668 + }, + { + "epoch": 0.8326073428749222, + "grad_norm": 2.8923728466033936, + "learning_rate": 4.047509225092252e-06, + "loss": 1.5475, + "step": 669 + }, + { + "epoch": 0.8338518979464842, + "grad_norm": 2.6134703159332275, + "learning_rate": 4.04520295202952e-06, + "loss": 1.5232, + "step": 670 + }, + { + "epoch": 0.8338518979464842, + "eval_loss": 1.7080007791519165, + "eval_runtime": 52.6531, + "eval_samples_per_second": 18.992, + "eval_steps_per_second": 0.798, + "step": 670 + }, + { + "epoch": 0.835096453018046, + "grad_norm": 2.6788907051086426, + "learning_rate": 4.04289667896679e-06, + "loss": 1.5437, + "step": 671 + }, + { + "epoch": 0.836341008089608, + "grad_norm": 2.872429847717285, + "learning_rate": 4.04059040590406e-06, + "loss": 1.5368, + "step": 672 + }, + { + "epoch": 0.8375855631611698, + "grad_norm": 2.714857578277588, + "learning_rate": 4.038284132841329e-06, + "loss": 1.5404, + "step": 673 + }, + { + "epoch": 0.8388301182327318, + "grad_norm": 2.6955931186676025, + "learning_rate": 4.035977859778598e-06, + "loss": 1.5501, + "step": 674 + }, + { + "epoch": 0.8400746733042938, + "grad_norm": 2.7090582847595215, + "learning_rate": 4.0336715867158674e-06, + "loss": 1.5068, + "step": 675 + }, + { + "epoch": 0.8413192283758556, + "grad_norm": 2.8406383991241455, + "learning_rate": 4.031365313653137e-06, + "loss": 1.5529, + "step": 676 + }, + { + "epoch": 0.8425637834474176, + "grad_norm": 2.965557098388672, + "learning_rate": 4.029059040590406e-06, + "loss": 1.5847, + "step": 677 + }, + { + "epoch": 0.8438083385189795, + "grad_norm": 2.664083242416382, + "learning_rate": 4.026752767527676e-06, + "loss": 1.5701, + "step": 678 + }, + { + "epoch": 0.8450528935905414, + "grad_norm": 2.7362401485443115, + "learning_rate": 4.024446494464945e-06, + "loss": 1.5496, + "step": 679 + }, + { + "epoch": 0.8462974486621033, + "grad_norm": 2.695342540740967, + "learning_rate": 4.0221402214022145e-06, + "loss": 1.5009, + "step": 680 + }, + { + "epoch": 0.8462974486621033, + "eval_loss": 1.7063637971878052, + "eval_runtime": 52.7749, + "eval_samples_per_second": 18.948, + "eval_steps_per_second": 0.796, + "step": 680 + }, + { + "epoch": 0.8475420037336652, + "grad_norm": 2.615994453430176, + "learning_rate": 4.019833948339484e-06, + "loss": 1.5281, + "step": 681 + }, + { + "epoch": 0.8487865588052271, + "grad_norm": 3.065732002258301, + "learning_rate": 4.017527675276753e-06, + "loss": 1.5616, + "step": 682 + }, + { + "epoch": 0.8500311138767891, + "grad_norm": 2.812166690826416, + "learning_rate": 4.015221402214022e-06, + "loss": 1.5573, + "step": 683 + }, + { + "epoch": 0.8512756689483509, + "grad_norm": 2.745252847671509, + "learning_rate": 4.012915129151292e-06, + "loss": 1.5426, + "step": 684 + }, + { + "epoch": 0.8525202240199129, + "grad_norm": 2.862264633178711, + "learning_rate": 4.010608856088562e-06, + "loss": 1.4973, + "step": 685 + }, + { + "epoch": 0.8537647790914747, + "grad_norm": 2.752469062805176, + "learning_rate": 4.00830258302583e-06, + "loss": 1.5416, + "step": 686 + }, + { + "epoch": 0.8550093341630367, + "grad_norm": 2.5692927837371826, + "learning_rate": 4.0059963099631e-06, + "loss": 1.5021, + "step": 687 + }, + { + "epoch": 0.8562538892345987, + "grad_norm": 2.896444320678711, + "learning_rate": 4.003690036900369e-06, + "loss": 1.5794, + "step": 688 + }, + { + "epoch": 0.8574984443061605, + "grad_norm": 2.7448434829711914, + "learning_rate": 4.0013837638376385e-06, + "loss": 1.5138, + "step": 689 + }, + { + "epoch": 0.8587429993777225, + "grad_norm": 2.8280975818634033, + "learning_rate": 3.999077490774908e-06, + "loss": 1.5071, + "step": 690 + }, + { + "epoch": 0.8587429993777225, + "eval_loss": 1.7048547267913818, + "eval_runtime": 52.8015, + "eval_samples_per_second": 18.939, + "eval_steps_per_second": 0.795, + "step": 690 + }, + { + "epoch": 0.8599875544492844, + "grad_norm": 2.701293706893921, + "learning_rate": 3.996771217712177e-06, + "loss": 1.5457, + "step": 691 + }, + { + "epoch": 0.8612321095208463, + "grad_norm": 2.673452138900757, + "learning_rate": 3.994464944649447e-06, + "loss": 1.5355, + "step": 692 + }, + { + "epoch": 0.8624766645924082, + "grad_norm": 2.7728021144866943, + "learning_rate": 3.992158671586716e-06, + "loss": 1.5486, + "step": 693 + }, + { + "epoch": 0.8637212196639701, + "grad_norm": 2.747614860534668, + "learning_rate": 3.989852398523986e-06, + "loss": 1.5771, + "step": 694 + }, + { + "epoch": 0.864965774735532, + "grad_norm": 2.6240482330322266, + "learning_rate": 3.987546125461255e-06, + "loss": 1.5338, + "step": 695 + }, + { + "epoch": 0.866210329807094, + "grad_norm": 2.8253114223480225, + "learning_rate": 3.9852398523985245e-06, + "loss": 1.5029, + "step": 696 + }, + { + "epoch": 0.8674548848786559, + "grad_norm": 2.7980496883392334, + "learning_rate": 3.982933579335794e-06, + "loss": 1.5239, + "step": 697 + }, + { + "epoch": 0.8686994399502178, + "grad_norm": 2.937443971633911, + "learning_rate": 3.980627306273063e-06, + "loss": 1.5406, + "step": 698 + }, + { + "epoch": 0.8699439950217797, + "grad_norm": 2.714353322982788, + "learning_rate": 3.978321033210332e-06, + "loss": 1.5724, + "step": 699 + }, + { + "epoch": 0.8711885500933416, + "grad_norm": 2.7986929416656494, + "learning_rate": 3.976014760147602e-06, + "loss": 1.5687, + "step": 700 + }, + { + "epoch": 0.8711885500933416, + "eval_loss": 1.6998120546340942, + "eval_runtime": 46.4448, + "eval_samples_per_second": 21.531, + "eval_steps_per_second": 0.904, + "step": 700 + }, + { + "epoch": 0.8724331051649036, + "grad_norm": 2.755740165710449, + "learning_rate": 3.973708487084872e-06, + "loss": 1.5047, + "step": 701 + }, + { + "epoch": 0.8736776602364654, + "grad_norm": 2.760399341583252, + "learning_rate": 3.97140221402214e-06, + "loss": 1.5406, + "step": 702 + }, + { + "epoch": 0.8749222153080274, + "grad_norm": 2.780214786529541, + "learning_rate": 3.96909594095941e-06, + "loss": 1.551, + "step": 703 + }, + { + "epoch": 0.8761667703795893, + "grad_norm": 2.669801950454712, + "learning_rate": 3.966789667896679e-06, + "loss": 1.5578, + "step": 704 + }, + { + "epoch": 0.8774113254511512, + "grad_norm": 2.729586362838745, + "learning_rate": 3.964483394833948e-06, + "loss": 1.4808, + "step": 705 + }, + { + "epoch": 0.8786558805227132, + "grad_norm": 2.608551263809204, + "learning_rate": 3.962177121771218e-06, + "loss": 1.4649, + "step": 706 + }, + { + "epoch": 0.879900435594275, + "grad_norm": 3.0885748863220215, + "learning_rate": 3.959870848708487e-06, + "loss": 1.5602, + "step": 707 + }, + { + "epoch": 0.881144990665837, + "grad_norm": 2.659917116165161, + "learning_rate": 3.957564575645757e-06, + "loss": 1.5341, + "step": 708 + }, + { + "epoch": 0.8823895457373989, + "grad_norm": 2.718696355819702, + "learning_rate": 3.955258302583026e-06, + "loss": 1.5232, + "step": 709 + }, + { + "epoch": 0.8836341008089608, + "grad_norm": 2.902811288833618, + "learning_rate": 3.9529520295202955e-06, + "loss": 1.517, + "step": 710 + }, + { + "epoch": 0.8836341008089608, + "eval_loss": 1.7022311687469482, + "eval_runtime": 43.1291, + "eval_samples_per_second": 23.186, + "eval_steps_per_second": 0.974, + "step": 710 + }, + { + "epoch": 0.8848786558805227, + "grad_norm": 2.816802978515625, + "learning_rate": 3.950645756457565e-06, + "loss": 1.5097, + "step": 711 + }, + { + "epoch": 0.8861232109520847, + "grad_norm": 2.8189518451690674, + "learning_rate": 3.948339483394834e-06, + "loss": 1.5842, + "step": 712 + }, + { + "epoch": 0.8873677660236465, + "grad_norm": 2.879767417907715, + "learning_rate": 3.946033210332104e-06, + "loss": 1.5484, + "step": 713 + }, + { + "epoch": 0.8886123210952085, + "grad_norm": 2.8263540267944336, + "learning_rate": 3.943726937269373e-06, + "loss": 1.5259, + "step": 714 + }, + { + "epoch": 0.8898568761667703, + "grad_norm": 2.655139923095703, + "learning_rate": 3.941420664206642e-06, + "loss": 1.482, + "step": 715 + }, + { + "epoch": 0.8911014312383323, + "grad_norm": 2.63429856300354, + "learning_rate": 3.939114391143912e-06, + "loss": 1.512, + "step": 716 + }, + { + "epoch": 0.8923459863098943, + "grad_norm": 2.9241650104522705, + "learning_rate": 3.9368081180811815e-06, + "loss": 1.5347, + "step": 717 + }, + { + "epoch": 0.8935905413814561, + "grad_norm": 2.8512847423553467, + "learning_rate": 3.934501845018451e-06, + "loss": 1.5791, + "step": 718 + }, + { + "epoch": 0.8948350964530181, + "grad_norm": 2.5280821323394775, + "learning_rate": 3.9321955719557195e-06, + "loss": 1.4823, + "step": 719 + }, + { + "epoch": 0.8960796515245799, + "grad_norm": 2.6618733406066895, + "learning_rate": 3.929889298892989e-06, + "loss": 1.4708, + "step": 720 + }, + { + "epoch": 0.8960796515245799, + "eval_loss": 1.6904946565628052, + "eval_runtime": 49.3149, + "eval_samples_per_second": 20.278, + "eval_steps_per_second": 0.852, + "step": 720 + }, + { + "epoch": 0.8973242065961419, + "grad_norm": 2.8406248092651367, + "learning_rate": 3.927583025830259e-06, + "loss": 1.5425, + "step": 721 + }, + { + "epoch": 0.8985687616677038, + "grad_norm": 2.807971715927124, + "learning_rate": 3.925276752767528e-06, + "loss": 1.5617, + "step": 722 + }, + { + "epoch": 0.8998133167392657, + "grad_norm": 2.6703925132751465, + "learning_rate": 3.922970479704797e-06, + "loss": 1.5544, + "step": 723 + }, + { + "epoch": 0.9010578718108276, + "grad_norm": 2.785083770751953, + "learning_rate": 3.920664206642067e-06, + "loss": 1.5696, + "step": 724 + }, + { + "epoch": 0.9023024268823896, + "grad_norm": 2.7778568267822266, + "learning_rate": 3.918357933579336e-06, + "loss": 1.5262, + "step": 725 + }, + { + "epoch": 0.9035469819539514, + "grad_norm": 2.6343681812286377, + "learning_rate": 3.9160516605166055e-06, + "loss": 1.5509, + "step": 726 + }, + { + "epoch": 0.9047915370255134, + "grad_norm": 2.6923940181732178, + "learning_rate": 3.913745387453875e-06, + "loss": 1.5425, + "step": 727 + }, + { + "epoch": 0.9060360920970753, + "grad_norm": 2.732797861099243, + "learning_rate": 3.911439114391144e-06, + "loss": 1.4524, + "step": 728 + }, + { + "epoch": 0.9072806471686372, + "grad_norm": 2.6365227699279785, + "learning_rate": 3.909132841328414e-06, + "loss": 1.5617, + "step": 729 + }, + { + "epoch": 0.9085252022401992, + "grad_norm": 2.646639823913574, + "learning_rate": 3.906826568265683e-06, + "loss": 1.5122, + "step": 730 + }, + { + "epoch": 0.9085252022401992, + "eval_loss": 1.6920651197433472, + "eval_runtime": 48.5914, + "eval_samples_per_second": 20.58, + "eval_steps_per_second": 0.864, + "step": 730 + }, + { + "epoch": 0.909769757311761, + "grad_norm": 2.5834903717041016, + "learning_rate": 3.9045202952029526e-06, + "loss": 1.5095, + "step": 731 + }, + { + "epoch": 0.911014312383323, + "grad_norm": 2.4998092651367188, + "learning_rate": 3.902214022140222e-06, + "loss": 1.4376, + "step": 732 + }, + { + "epoch": 0.9122588674548848, + "grad_norm": 2.5143818855285645, + "learning_rate": 3.899907749077491e-06, + "loss": 1.5222, + "step": 733 + }, + { + "epoch": 0.9135034225264468, + "grad_norm": 2.7795114517211914, + "learning_rate": 3.897601476014761e-06, + "loss": 1.5498, + "step": 734 + }, + { + "epoch": 0.9147479775980087, + "grad_norm": 2.819859027862549, + "learning_rate": 3.895295202952029e-06, + "loss": 1.5288, + "step": 735 + }, + { + "epoch": 0.9159925326695706, + "grad_norm": 2.663306951522827, + "learning_rate": 3.892988929889299e-06, + "loss": 1.4517, + "step": 736 + }, + { + "epoch": 0.9172370877411326, + "grad_norm": 2.775521993637085, + "learning_rate": 3.890682656826569e-06, + "loss": 1.529, + "step": 737 + }, + { + "epoch": 0.9184816428126945, + "grad_norm": 2.761237859725952, + "learning_rate": 3.888376383763838e-06, + "loss": 1.4781, + "step": 738 + }, + { + "epoch": 0.9197261978842564, + "grad_norm": 2.5410993099212646, + "learning_rate": 3.886070110701107e-06, + "loss": 1.4205, + "step": 739 + }, + { + "epoch": 0.9209707529558183, + "grad_norm": 2.6289987564086914, + "learning_rate": 3.8837638376383765e-06, + "loss": 1.5253, + "step": 740 + }, + { + "epoch": 0.9209707529558183, + "eval_loss": 1.700616717338562, + "eval_runtime": 46.1038, + "eval_samples_per_second": 21.69, + "eval_steps_per_second": 0.911, + "step": 740 + }, + { + "epoch": 0.9222153080273802, + "grad_norm": 2.694406270980835, + "learning_rate": 3.881457564575646e-06, + "loss": 1.5335, + "step": 741 + }, + { + "epoch": 0.9234598630989421, + "grad_norm": 2.597160577774048, + "learning_rate": 3.879151291512915e-06, + "loss": 1.5026, + "step": 742 + }, + { + "epoch": 0.9247044181705041, + "grad_norm": 2.625049114227295, + "learning_rate": 3.876845018450185e-06, + "loss": 1.4885, + "step": 743 + }, + { + "epoch": 0.9259489732420659, + "grad_norm": 2.5983152389526367, + "learning_rate": 3.874538745387454e-06, + "loss": 1.5015, + "step": 744 + }, + { + "epoch": 0.9271935283136279, + "grad_norm": 2.6912894248962402, + "learning_rate": 3.872232472324724e-06, + "loss": 1.5179, + "step": 745 + }, + { + "epoch": 0.9284380833851898, + "grad_norm": 2.699328899383545, + "learning_rate": 3.869926199261993e-06, + "loss": 1.5571, + "step": 746 + }, + { + "epoch": 0.9296826384567517, + "grad_norm": 2.7117769718170166, + "learning_rate": 3.8676199261992625e-06, + "loss": 1.5225, + "step": 747 + }, + { + "epoch": 0.9309271935283137, + "grad_norm": 2.785900354385376, + "learning_rate": 3.865313653136532e-06, + "loss": 1.5571, + "step": 748 + }, + { + "epoch": 0.9321717485998755, + "grad_norm": 2.6901021003723145, + "learning_rate": 3.863007380073801e-06, + "loss": 1.5108, + "step": 749 + }, + { + "epoch": 0.9334163036714375, + "grad_norm": 2.55047607421875, + "learning_rate": 3.860701107011071e-06, + "loss": 1.4934, + "step": 750 + }, + { + "epoch": 0.9334163036714375, + "eval_loss": 1.693824291229248, + "eval_runtime": 45.7403, + "eval_samples_per_second": 21.863, + "eval_steps_per_second": 0.918, + "step": 750 + }, + { + "epoch": 0.9346608587429994, + "grad_norm": 2.7620902061462402, + "learning_rate": 3.858394833948339e-06, + "loss": 1.5129, + "step": 751 + }, + { + "epoch": 0.9359054138145613, + "grad_norm": 2.798583984375, + "learning_rate": 3.856088560885609e-06, + "loss": 1.4986, + "step": 752 + }, + { + "epoch": 0.9371499688861232, + "grad_norm": 2.5885400772094727, + "learning_rate": 3.853782287822879e-06, + "loss": 1.4779, + "step": 753 + }, + { + "epoch": 0.9383945239576851, + "grad_norm": 2.8280391693115234, + "learning_rate": 3.851476014760148e-06, + "loss": 1.6156, + "step": 754 + }, + { + "epoch": 0.939639079029247, + "grad_norm": 2.7616288661956787, + "learning_rate": 3.849169741697417e-06, + "loss": 1.5051, + "step": 755 + }, + { + "epoch": 0.940883634100809, + "grad_norm": 2.598370313644409, + "learning_rate": 3.8468634686346865e-06, + "loss": 1.5569, + "step": 756 + }, + { + "epoch": 0.9421281891723708, + "grad_norm": 2.649433135986328, + "learning_rate": 3.844557195571956e-06, + "loss": 1.515, + "step": 757 + }, + { + "epoch": 0.9433727442439328, + "grad_norm": 2.568361759185791, + "learning_rate": 3.842250922509225e-06, + "loss": 1.4926, + "step": 758 + }, + { + "epoch": 0.9446172993154948, + "grad_norm": 2.6539297103881836, + "learning_rate": 3.839944649446495e-06, + "loss": 1.4813, + "step": 759 + }, + { + "epoch": 0.9458618543870566, + "grad_norm": 2.6790902614593506, + "learning_rate": 3.837638376383764e-06, + "loss": 1.5056, + "step": 760 + }, + { + "epoch": 0.9458618543870566, + "eval_loss": 1.6913493871688843, + "eval_runtime": 44.5061, + "eval_samples_per_second": 22.469, + "eval_steps_per_second": 0.944, + "step": 760 + }, + { + "epoch": 0.9471064094586186, + "grad_norm": 2.6265010833740234, + "learning_rate": 3.8353321033210336e-06, + "loss": 1.5421, + "step": 761 + }, + { + "epoch": 0.9483509645301804, + "grad_norm": 2.5852062702178955, + "learning_rate": 3.833025830258303e-06, + "loss": 1.5145, + "step": 762 + }, + { + "epoch": 0.9495955196017424, + "grad_norm": 2.6579225063323975, + "learning_rate": 3.830719557195572e-06, + "loss": 1.4988, + "step": 763 + }, + { + "epoch": 0.9508400746733043, + "grad_norm": 2.7944133281707764, + "learning_rate": 3.828413284132842e-06, + "loss": 1.5199, + "step": 764 + }, + { + "epoch": 0.9520846297448662, + "grad_norm": 2.5813052654266357, + "learning_rate": 3.826107011070111e-06, + "loss": 1.5072, + "step": 765 + }, + { + "epoch": 0.9533291848164281, + "grad_norm": 2.551501512527466, + "learning_rate": 3.823800738007381e-06, + "loss": 1.5348, + "step": 766 + }, + { + "epoch": 0.95457373988799, + "grad_norm": 2.755141496658325, + "learning_rate": 3.821494464944649e-06, + "loss": 1.4787, + "step": 767 + }, + { + "epoch": 0.955818294959552, + "grad_norm": 2.521984338760376, + "learning_rate": 3.819188191881919e-06, + "loss": 1.4778, + "step": 768 + }, + { + "epoch": 0.9570628500311139, + "grad_norm": 2.6301770210266113, + "learning_rate": 3.816881918819189e-06, + "loss": 1.4703, + "step": 769 + }, + { + "epoch": 0.9583074051026758, + "grad_norm": 2.972749710083008, + "learning_rate": 3.814575645756458e-06, + "loss": 1.5516, + "step": 770 + }, + { + "epoch": 0.9583074051026758, + "eval_loss": 1.689251184463501, + "eval_runtime": 44.3352, + "eval_samples_per_second": 22.555, + "eval_steps_per_second": 0.947, + "step": 770 + }, + { + "epoch": 0.9595519601742377, + "grad_norm": 2.7247636318206787, + "learning_rate": 3.812269372693727e-06, + "loss": 1.5659, + "step": 771 + }, + { + "epoch": 0.9607965152457997, + "grad_norm": 2.6853489875793457, + "learning_rate": 3.809963099630997e-06, + "loss": 1.5149, + "step": 772 + }, + { + "epoch": 0.9620410703173615, + "grad_norm": 2.6767218112945557, + "learning_rate": 3.8076568265682662e-06, + "loss": 1.5573, + "step": 773 + }, + { + "epoch": 0.9632856253889235, + "grad_norm": 2.5949296951293945, + "learning_rate": 3.8053505535055352e-06, + "loss": 1.517, + "step": 774 + }, + { + "epoch": 0.9645301804604853, + "grad_norm": 2.841171979904175, + "learning_rate": 3.8030442804428046e-06, + "loss": 1.515, + "step": 775 + }, + { + "epoch": 0.9657747355320473, + "grad_norm": 2.7451512813568115, + "learning_rate": 3.800738007380074e-06, + "loss": 1.516, + "step": 776 + }, + { + "epoch": 0.9670192906036092, + "grad_norm": 2.6852564811706543, + "learning_rate": 3.798431734317343e-06, + "loss": 1.5187, + "step": 777 + }, + { + "epoch": 0.9682638456751711, + "grad_norm": 2.72956919670105, + "learning_rate": 3.796125461254613e-06, + "loss": 1.5344, + "step": 778 + }, + { + "epoch": 0.9695084007467331, + "grad_norm": 2.672933578491211, + "learning_rate": 3.7938191881918823e-06, + "loss": 1.5143, + "step": 779 + }, + { + "epoch": 0.9707529558182949, + "grad_norm": 2.6565003395080566, + "learning_rate": 3.7915129151291518e-06, + "loss": 1.4759, + "step": 780 + }, + { + "epoch": 0.9707529558182949, + "eval_loss": 1.6836074590682983, + "eval_runtime": 44.0331, + "eval_samples_per_second": 22.71, + "eval_steps_per_second": 0.954, + "step": 780 + }, + { + "epoch": 0.9719975108898569, + "grad_norm": 2.665921449661255, + "learning_rate": 3.7892066420664208e-06, + "loss": 1.4985, + "step": 781 + }, + { + "epoch": 0.9732420659614188, + "grad_norm": 2.671757221221924, + "learning_rate": 3.7869003690036906e-06, + "loss": 1.5014, + "step": 782 + }, + { + "epoch": 0.9744866210329807, + "grad_norm": 2.7320780754089355, + "learning_rate": 3.78459409594096e-06, + "loss": 1.5052, + "step": 783 + }, + { + "epoch": 0.9757311761045426, + "grad_norm": 3.0772581100463867, + "learning_rate": 3.782287822878229e-06, + "loss": 1.5666, + "step": 784 + }, + { + "epoch": 0.9769757311761046, + "grad_norm": 2.651129722595215, + "learning_rate": 3.7799815498154984e-06, + "loss": 1.4954, + "step": 785 + }, + { + "epoch": 0.9782202862476664, + "grad_norm": 2.6942505836486816, + "learning_rate": 3.777675276752768e-06, + "loss": 1.5186, + "step": 786 + }, + { + "epoch": 0.9794648413192284, + "grad_norm": 2.767474412918091, + "learning_rate": 3.775369003690037e-06, + "loss": 1.4699, + "step": 787 + }, + { + "epoch": 0.9807093963907902, + "grad_norm": 2.728269100189209, + "learning_rate": 3.7730627306273067e-06, + "loss": 1.5176, + "step": 788 + }, + { + "epoch": 0.9819539514623522, + "grad_norm": 2.697984218597412, + "learning_rate": 3.770756457564576e-06, + "loss": 1.5327, + "step": 789 + }, + { + "epoch": 0.9831985065339142, + "grad_norm": 2.8172037601470947, + "learning_rate": 3.768450184501845e-06, + "loss": 1.4996, + "step": 790 + }, + { + "epoch": 0.9831985065339142, + "eval_loss": 1.7013198137283325, + "eval_runtime": 44.5233, + "eval_samples_per_second": 22.46, + "eval_steps_per_second": 0.943, + "step": 790 + }, + { + "epoch": 0.984443061605476, + "grad_norm": 2.8202197551727295, + "learning_rate": 3.7661439114391146e-06, + "loss": 1.5193, + "step": 791 + }, + { + "epoch": 0.985687616677038, + "grad_norm": 2.764653444290161, + "learning_rate": 3.763837638376384e-06, + "loss": 1.5544, + "step": 792 + }, + { + "epoch": 0.9869321717485999, + "grad_norm": 2.6528818607330322, + "learning_rate": 3.761531365313654e-06, + "loss": 1.5328, + "step": 793 + }, + { + "epoch": 0.9881767268201618, + "grad_norm": 2.643666982650757, + "learning_rate": 3.759225092250923e-06, + "loss": 1.5279, + "step": 794 + }, + { + "epoch": 0.9894212818917237, + "grad_norm": 2.631007432937622, + "learning_rate": 3.7569188191881922e-06, + "loss": 1.5135, + "step": 795 + }, + { + "epoch": 0.9906658369632856, + "grad_norm": 2.6622400283813477, + "learning_rate": 3.7546125461254617e-06, + "loss": 1.5099, + "step": 796 + }, + { + "epoch": 0.9919103920348475, + "grad_norm": 2.5758702754974365, + "learning_rate": 3.7523062730627307e-06, + "loss": 1.4691, + "step": 797 + }, + { + "epoch": 0.9931549471064095, + "grad_norm": 2.822147846221924, + "learning_rate": 3.7500000000000005e-06, + "loss": 1.4846, + "step": 798 + }, + { + "epoch": 0.9943995021779714, + "grad_norm": 3.166783571243286, + "learning_rate": 3.74769372693727e-06, + "loss": 1.5586, + "step": 799 + }, + { + "epoch": 0.9956440572495333, + "grad_norm": 2.931654691696167, + "learning_rate": 3.745387453874539e-06, + "loss": 1.4773, + "step": 800 + }, + { + "epoch": 0.9956440572495333, + "eval_loss": 1.6904431581497192, + "eval_runtime": 43.927, + "eval_samples_per_second": 22.765, + "eval_steps_per_second": 0.956, + "step": 800 + }, + { + "epoch": 0.9968886123210952, + "grad_norm": 2.97959303855896, + "learning_rate": 3.7430811808118084e-06, + "loss": 1.4555, + "step": 801 + }, + { + "epoch": 0.9981331673926571, + "grad_norm": 2.796096086502075, + "learning_rate": 3.7407749077490778e-06, + "loss": 1.5243, + "step": 802 + }, + { + "epoch": 0.9993777224642191, + "grad_norm": 2.8529677391052246, + "learning_rate": 3.7384686346863468e-06, + "loss": 1.4919, + "step": 803 + }, + { + "epoch": 1.000622277535781, + "grad_norm": 2.9269766807556152, + "learning_rate": 3.7361623616236166e-06, + "loss": 1.5188, + "step": 804 + }, + { + "epoch": 1.0018668326073428, + "grad_norm": 2.8332459926605225, + "learning_rate": 3.733856088560886e-06, + "loss": 1.499, + "step": 805 + }, + { + "epoch": 1.0031113876789048, + "grad_norm": 2.783031702041626, + "learning_rate": 3.7315498154981555e-06, + "loss": 1.4383, + "step": 806 + }, + { + "epoch": 1.0043559427504667, + "grad_norm": 3.097334623336792, + "learning_rate": 3.7292435424354245e-06, + "loss": 1.5, + "step": 807 + }, + { + "epoch": 1.0056004978220285, + "grad_norm": 3.1136531829833984, + "learning_rate": 3.726937269372694e-06, + "loss": 1.5161, + "step": 808 + }, + { + "epoch": 1.0068450528935906, + "grad_norm": 2.8723881244659424, + "learning_rate": 3.7246309963099637e-06, + "loss": 1.4716, + "step": 809 + }, + { + "epoch": 1.0080896079651525, + "grad_norm": 2.798966407775879, + "learning_rate": 3.7223247232472327e-06, + "loss": 1.4874, + "step": 810 + }, + { + "epoch": 1.0080896079651525, + "eval_loss": 1.6889101266860962, + "eval_runtime": 42.0407, + "eval_samples_per_second": 23.786, + "eval_steps_per_second": 0.999, + "step": 810 + }, + { + "epoch": 1.0093341630367143, + "grad_norm": 2.731884241104126, + "learning_rate": 3.720018450184502e-06, + "loss": 1.443, + "step": 811 + }, + { + "epoch": 1.0105787181082764, + "grad_norm": 2.651594877243042, + "learning_rate": 3.7177121771217716e-06, + "loss": 1.4537, + "step": 812 + }, + { + "epoch": 1.0118232731798382, + "grad_norm": 2.8403851985931396, + "learning_rate": 3.7154059040590406e-06, + "loss": 1.4801, + "step": 813 + }, + { + "epoch": 1.0130678282514, + "grad_norm": 2.8779821395874023, + "learning_rate": 3.71309963099631e-06, + "loss": 1.4716, + "step": 814 + }, + { + "epoch": 1.0143123833229621, + "grad_norm": 2.8755605220794678, + "learning_rate": 3.71079335793358e-06, + "loss": 1.471, + "step": 815 + }, + { + "epoch": 1.015556938394524, + "grad_norm": 2.9748878479003906, + "learning_rate": 3.708487084870849e-06, + "loss": 1.4988, + "step": 816 + }, + { + "epoch": 1.0168014934660858, + "grad_norm": 2.9609413146972656, + "learning_rate": 3.7061808118081183e-06, + "loss": 1.5184, + "step": 817 + }, + { + "epoch": 1.0180460485376477, + "grad_norm": 2.8808226585388184, + "learning_rate": 3.7038745387453877e-06, + "loss": 1.5161, + "step": 818 + }, + { + "epoch": 1.0192906036092098, + "grad_norm": 3.0764009952545166, + "learning_rate": 3.7015682656826576e-06, + "loss": 1.4991, + "step": 819 + }, + { + "epoch": 1.0205351586807716, + "grad_norm": 2.802974224090576, + "learning_rate": 3.6992619926199266e-06, + "loss": 1.4967, + "step": 820 + }, + { + "epoch": 1.0205351586807716, + "eval_loss": 1.6831101179122925, + "eval_runtime": 44.5974, + "eval_samples_per_second": 22.423, + "eval_steps_per_second": 0.942, + "step": 820 + }, + { + "epoch": 1.0217797137523335, + "grad_norm": 2.8415472507476807, + "learning_rate": 3.696955719557196e-06, + "loss": 1.4636, + "step": 821 + }, + { + "epoch": 1.0230242688238955, + "grad_norm": 2.830817461013794, + "learning_rate": 3.6946494464944654e-06, + "loss": 1.4648, + "step": 822 + }, + { + "epoch": 1.0242688238954574, + "grad_norm": 2.993065357208252, + "learning_rate": 3.6923431734317344e-06, + "loss": 1.4923, + "step": 823 + }, + { + "epoch": 1.0255133789670192, + "grad_norm": 2.913902521133423, + "learning_rate": 3.690036900369004e-06, + "loss": 1.4825, + "step": 824 + }, + { + "epoch": 1.0267579340385813, + "grad_norm": 2.984265089035034, + "learning_rate": 3.6877306273062737e-06, + "loss": 1.457, + "step": 825 + }, + { + "epoch": 1.0280024891101431, + "grad_norm": 3.063187599182129, + "learning_rate": 3.6854243542435427e-06, + "loss": 1.5537, + "step": 826 + }, + { + "epoch": 1.029247044181705, + "grad_norm": 2.876314878463745, + "learning_rate": 3.683118081180812e-06, + "loss": 1.4555, + "step": 827 + }, + { + "epoch": 1.030491599253267, + "grad_norm": 3.039470911026001, + "learning_rate": 3.6808118081180815e-06, + "loss": 1.4862, + "step": 828 + }, + { + "epoch": 1.031736154324829, + "grad_norm": 2.8783342838287354, + "learning_rate": 3.678505535055351e-06, + "loss": 1.5264, + "step": 829 + }, + { + "epoch": 1.0329807093963908, + "grad_norm": 2.829578161239624, + "learning_rate": 3.67619926199262e-06, + "loss": 1.5054, + "step": 830 + }, + { + "epoch": 1.0329807093963908, + "eval_loss": 1.6823173761367798, + "eval_runtime": 45.0076, + "eval_samples_per_second": 22.218, + "eval_steps_per_second": 0.933, + "step": 830 + }, + { + "epoch": 1.0342252644679526, + "grad_norm": 2.9113621711730957, + "learning_rate": 3.6738929889298898e-06, + "loss": 1.4211, + "step": 831 + }, + { + "epoch": 1.0354698195395147, + "grad_norm": 2.901256561279297, + "learning_rate": 3.671586715867159e-06, + "loss": 1.5566, + "step": 832 + }, + { + "epoch": 1.0367143746110765, + "grad_norm": 2.689814805984497, + "learning_rate": 3.669280442804428e-06, + "loss": 1.4062, + "step": 833 + }, + { + "epoch": 1.0379589296826384, + "grad_norm": 2.848036766052246, + "learning_rate": 3.6669741697416976e-06, + "loss": 1.524, + "step": 834 + }, + { + "epoch": 1.0392034847542004, + "grad_norm": 2.7704179286956787, + "learning_rate": 3.6646678966789675e-06, + "loss": 1.5119, + "step": 835 + }, + { + "epoch": 1.0404480398257623, + "grad_norm": 2.9936625957489014, + "learning_rate": 3.6623616236162365e-06, + "loss": 1.5188, + "step": 836 + }, + { + "epoch": 1.0416925948973241, + "grad_norm": 2.752682685852051, + "learning_rate": 3.660055350553506e-06, + "loss": 1.4848, + "step": 837 + }, + { + "epoch": 1.0429371499688862, + "grad_norm": 2.7587647438049316, + "learning_rate": 3.6577490774907753e-06, + "loss": 1.4784, + "step": 838 + }, + { + "epoch": 1.044181705040448, + "grad_norm": 2.6882312297821045, + "learning_rate": 3.6554428044280443e-06, + "loss": 1.4574, + "step": 839 + }, + { + "epoch": 1.04542626011201, + "grad_norm": 2.847754955291748, + "learning_rate": 3.6531365313653137e-06, + "loss": 1.5025, + "step": 840 + }, + { + "epoch": 1.04542626011201, + "eval_loss": 1.6905474662780762, + "eval_runtime": 47.5605, + "eval_samples_per_second": 21.026, + "eval_steps_per_second": 0.883, + "step": 840 + }, + { + "epoch": 1.046670815183572, + "grad_norm": 2.7890496253967285, + "learning_rate": 3.6508302583025836e-06, + "loss": 1.5182, + "step": 841 + }, + { + "epoch": 1.0479153702551338, + "grad_norm": 2.761654853820801, + "learning_rate": 3.648523985239853e-06, + "loss": 1.4691, + "step": 842 + }, + { + "epoch": 1.0491599253266957, + "grad_norm": 2.8627774715423584, + "learning_rate": 3.646217712177122e-06, + "loss": 1.4785, + "step": 843 + }, + { + "epoch": 1.0504044803982575, + "grad_norm": 2.8348138332366943, + "learning_rate": 3.6439114391143914e-06, + "loss": 1.4418, + "step": 844 + }, + { + "epoch": 1.0516490354698196, + "grad_norm": 2.774764060974121, + "learning_rate": 3.641605166051661e-06, + "loss": 1.4872, + "step": 845 + }, + { + "epoch": 1.0528935905413814, + "grad_norm": 2.9172654151916504, + "learning_rate": 3.63929889298893e-06, + "loss": 1.473, + "step": 846 + }, + { + "epoch": 1.0541381456129433, + "grad_norm": 2.7001171112060547, + "learning_rate": 3.6369926199261997e-06, + "loss": 1.422, + "step": 847 + }, + { + "epoch": 1.0553827006845053, + "grad_norm": 2.8742928504943848, + "learning_rate": 3.634686346863469e-06, + "loss": 1.4152, + "step": 848 + }, + { + "epoch": 1.0566272557560672, + "grad_norm": 2.9559481143951416, + "learning_rate": 3.632380073800738e-06, + "loss": 1.5244, + "step": 849 + }, + { + "epoch": 1.057871810827629, + "grad_norm": 2.7938740253448486, + "learning_rate": 3.6300738007380075e-06, + "loss": 1.4555, + "step": 850 + }, + { + "epoch": 1.057871810827629, + "eval_loss": 1.6855521202087402, + "eval_runtime": 47.3414, + "eval_samples_per_second": 21.123, + "eval_steps_per_second": 0.887, + "step": 850 + }, + { + "epoch": 1.0591163658991911, + "grad_norm": 3.011599540710449, + "learning_rate": 3.627767527675277e-06, + "loss": 1.5024, + "step": 851 + }, + { + "epoch": 1.060360920970753, + "grad_norm": 2.720555305480957, + "learning_rate": 3.625461254612546e-06, + "loss": 1.5183, + "step": 852 + }, + { + "epoch": 1.0616054760423148, + "grad_norm": 2.9110827445983887, + "learning_rate": 3.623154981549816e-06, + "loss": 1.4608, + "step": 853 + }, + { + "epoch": 1.0628500311138769, + "grad_norm": 3.001163959503174, + "learning_rate": 3.6208487084870852e-06, + "loss": 1.4797, + "step": 854 + }, + { + "epoch": 1.0640945861854387, + "grad_norm": 2.862942934036255, + "learning_rate": 3.6185424354243547e-06, + "loss": 1.4596, + "step": 855 + }, + { + "epoch": 1.0653391412570006, + "grad_norm": 2.907052755355835, + "learning_rate": 3.6162361623616237e-06, + "loss": 1.4452, + "step": 856 + }, + { + "epoch": 1.0665836963285624, + "grad_norm": 3.063230276107788, + "learning_rate": 3.6139298892988935e-06, + "loss": 1.4864, + "step": 857 + }, + { + "epoch": 1.0678282514001245, + "grad_norm": 3.0228397846221924, + "learning_rate": 3.611623616236163e-06, + "loss": 1.4971, + "step": 858 + }, + { + "epoch": 1.0690728064716863, + "grad_norm": 3.149158477783203, + "learning_rate": 3.609317343173432e-06, + "loss": 1.4578, + "step": 859 + }, + { + "epoch": 1.0703173615432482, + "grad_norm": 2.7961742877960205, + "learning_rate": 3.6070110701107014e-06, + "loss": 1.4246, + "step": 860 + }, + { + "epoch": 1.0703173615432482, + "eval_loss": 1.6836330890655518, + "eval_runtime": 48.4106, + "eval_samples_per_second": 20.657, + "eval_steps_per_second": 0.868, + "step": 860 + }, + { + "epoch": 1.0715619166148103, + "grad_norm": 3.029447078704834, + "learning_rate": 3.6047047970479708e-06, + "loss": 1.4864, + "step": 861 + }, + { + "epoch": 1.072806471686372, + "grad_norm": 2.9505438804626465, + "learning_rate": 3.6023985239852398e-06, + "loss": 1.46, + "step": 862 + }, + { + "epoch": 1.074051026757934, + "grad_norm": 2.899418592453003, + "learning_rate": 3.6000922509225096e-06, + "loss": 1.4937, + "step": 863 + }, + { + "epoch": 1.075295581829496, + "grad_norm": 2.889408588409424, + "learning_rate": 3.597785977859779e-06, + "loss": 1.4639, + "step": 864 + }, + { + "epoch": 1.0765401369010579, + "grad_norm": 3.0129477977752686, + "learning_rate": 3.595479704797048e-06, + "loss": 1.515, + "step": 865 + }, + { + "epoch": 1.0777846919726197, + "grad_norm": 3.0222294330596924, + "learning_rate": 3.5931734317343175e-06, + "loss": 1.49, + "step": 866 + }, + { + "epoch": 1.0790292470441818, + "grad_norm": 3.229149580001831, + "learning_rate": 3.590867158671587e-06, + "loss": 1.5106, + "step": 867 + }, + { + "epoch": 1.0802738021157436, + "grad_norm": 3.0588462352752686, + "learning_rate": 3.5885608856088567e-06, + "loss": 1.5002, + "step": 868 + }, + { + "epoch": 1.0815183571873055, + "grad_norm": 3.0674571990966797, + "learning_rate": 3.5862546125461257e-06, + "loss": 1.4592, + "step": 869 + }, + { + "epoch": 1.0827629122588673, + "grad_norm": 2.9173855781555176, + "learning_rate": 3.583948339483395e-06, + "loss": 1.4813, + "step": 870 + }, + { + "epoch": 1.0827629122588673, + "eval_loss": 1.677014946937561, + "eval_runtime": 49.4597, + "eval_samples_per_second": 20.218, + "eval_steps_per_second": 0.849, + "step": 870 + }, + { + "epoch": 1.0840074673304294, + "grad_norm": 2.8278772830963135, + "learning_rate": 3.5816420664206646e-06, + "loss": 1.4805, + "step": 871 + }, + { + "epoch": 1.0852520224019913, + "grad_norm": 2.8501267433166504, + "learning_rate": 3.5793357933579336e-06, + "loss": 1.471, + "step": 872 + }, + { + "epoch": 1.086496577473553, + "grad_norm": 2.922009229660034, + "learning_rate": 3.5770295202952034e-06, + "loss": 1.4801, + "step": 873 + }, + { + "epoch": 1.0877411325451152, + "grad_norm": 2.9171924591064453, + "learning_rate": 3.574723247232473e-06, + "loss": 1.473, + "step": 874 + }, + { + "epoch": 1.088985687616677, + "grad_norm": 2.992713212966919, + "learning_rate": 3.572416974169742e-06, + "loss": 1.4611, + "step": 875 + }, + { + "epoch": 1.0902302426882389, + "grad_norm": 2.7770748138427734, + "learning_rate": 3.5701107011070113e-06, + "loss": 1.474, + "step": 876 + }, + { + "epoch": 1.091474797759801, + "grad_norm": 2.6337552070617676, + "learning_rate": 3.5678044280442807e-06, + "loss": 1.4982, + "step": 877 + }, + { + "epoch": 1.0927193528313628, + "grad_norm": 2.8013854026794434, + "learning_rate": 3.5654981549815497e-06, + "loss": 1.5154, + "step": 878 + }, + { + "epoch": 1.0939639079029246, + "grad_norm": 2.731400728225708, + "learning_rate": 3.5631918819188195e-06, + "loss": 1.5056, + "step": 879 + }, + { + "epoch": 1.0952084629744867, + "grad_norm": 2.8960869312286377, + "learning_rate": 3.560885608856089e-06, + "loss": 1.4578, + "step": 880 + }, + { + "epoch": 1.0952084629744867, + "eval_loss": 1.6756274700164795, + "eval_runtime": 50.0986, + "eval_samples_per_second": 19.961, + "eval_steps_per_second": 0.838, + "step": 880 + }, + { + "epoch": 1.0964530180460486, + "grad_norm": 2.736156940460205, + "learning_rate": 3.5585793357933584e-06, + "loss": 1.4014, + "step": 881 + }, + { + "epoch": 1.0976975731176104, + "grad_norm": 2.8206899166107178, + "learning_rate": 3.5562730627306274e-06, + "loss": 1.519, + "step": 882 + }, + { + "epoch": 1.0989421281891725, + "grad_norm": 2.8125741481781006, + "learning_rate": 3.553966789667897e-06, + "loss": 1.493, + "step": 883 + }, + { + "epoch": 1.1001866832607343, + "grad_norm": 2.7721848487854004, + "learning_rate": 3.5516605166051667e-06, + "loss": 1.4805, + "step": 884 + }, + { + "epoch": 1.1014312383322962, + "grad_norm": 2.8858790397644043, + "learning_rate": 3.5493542435424357e-06, + "loss": 1.4339, + "step": 885 + }, + { + "epoch": 1.102675793403858, + "grad_norm": 2.897688865661621, + "learning_rate": 3.547047970479705e-06, + "loss": 1.4663, + "step": 886 + }, + { + "epoch": 1.10392034847542, + "grad_norm": 2.8516011238098145, + "learning_rate": 3.5447416974169745e-06, + "loss": 1.4946, + "step": 887 + }, + { + "epoch": 1.105164903546982, + "grad_norm": 2.801990270614624, + "learning_rate": 3.5424354243542435e-06, + "loss": 1.4764, + "step": 888 + }, + { + "epoch": 1.1064094586185438, + "grad_norm": 2.8651299476623535, + "learning_rate": 3.5401291512915133e-06, + "loss": 1.471, + "step": 889 + }, + { + "epoch": 1.1076540136901059, + "grad_norm": 2.8247878551483154, + "learning_rate": 3.5378228782287828e-06, + "loss": 1.5494, + "step": 890 + }, + { + "epoch": 1.1076540136901059, + "eval_loss": 1.6814671754837036, + "eval_runtime": 49.7658, + "eval_samples_per_second": 20.094, + "eval_steps_per_second": 0.844, + "step": 890 + }, + { + "epoch": 1.1088985687616677, + "grad_norm": 2.750497579574585, + "learning_rate": 3.535516605166052e-06, + "loss": 1.4578, + "step": 891 + }, + { + "epoch": 1.1101431238332296, + "grad_norm": 2.7336409091949463, + "learning_rate": 3.533210332103321e-06, + "loss": 1.4486, + "step": 892 + }, + { + "epoch": 1.1113876789047916, + "grad_norm": 3.0764975547790527, + "learning_rate": 3.5309040590405906e-06, + "loss": 1.4661, + "step": 893 + }, + { + "epoch": 1.1126322339763535, + "grad_norm": 2.7930431365966797, + "learning_rate": 3.5285977859778605e-06, + "loss": 1.4378, + "step": 894 + }, + { + "epoch": 1.1138767890479153, + "grad_norm": 2.944173574447632, + "learning_rate": 3.5262915129151295e-06, + "loss": 1.4639, + "step": 895 + }, + { + "epoch": 1.1151213441194774, + "grad_norm": 3.0577549934387207, + "learning_rate": 3.523985239852399e-06, + "loss": 1.4668, + "step": 896 + }, + { + "epoch": 1.1163658991910392, + "grad_norm": 2.805238962173462, + "learning_rate": 3.5216789667896683e-06, + "loss": 1.4531, + "step": 897 + }, + { + "epoch": 1.117610454262601, + "grad_norm": 2.8995230197906494, + "learning_rate": 3.5193726937269373e-06, + "loss": 1.5016, + "step": 898 + }, + { + "epoch": 1.1188550093341632, + "grad_norm": 2.8598949909210205, + "learning_rate": 3.5170664206642067e-06, + "loss": 1.4276, + "step": 899 + }, + { + "epoch": 1.120099564405725, + "grad_norm": 2.8928749561309814, + "learning_rate": 3.5147601476014766e-06, + "loss": 1.4888, + "step": 900 + }, + { + "epoch": 1.120099564405725, + "eval_loss": 1.6829349994659424, + "eval_runtime": 46.7111, + "eval_samples_per_second": 21.408, + "eval_steps_per_second": 0.899, + "step": 900 + }, + { + "epoch": 1.1213441194772868, + "grad_norm": 2.9442391395568848, + "learning_rate": 3.5124538745387456e-06, + "loss": 1.4636, + "step": 901 + }, + { + "epoch": 1.1225886745488487, + "grad_norm": 3.1975884437561035, + "learning_rate": 3.510147601476015e-06, + "loss": 1.4654, + "step": 902 + }, + { + "epoch": 1.1238332296204108, + "grad_norm": 2.915771722793579, + "learning_rate": 3.5078413284132844e-06, + "loss": 1.501, + "step": 903 + }, + { + "epoch": 1.1250777846919726, + "grad_norm": 2.880690097808838, + "learning_rate": 3.505535055350554e-06, + "loss": 1.4629, + "step": 904 + }, + { + "epoch": 1.1263223397635345, + "grad_norm": 2.868868350982666, + "learning_rate": 3.503228782287823e-06, + "loss": 1.4926, + "step": 905 + }, + { + "epoch": 1.1275668948350965, + "grad_norm": 3.1113288402557373, + "learning_rate": 3.5009225092250927e-06, + "loss": 1.4572, + "step": 906 + }, + { + "epoch": 1.1288114499066584, + "grad_norm": 2.904810667037964, + "learning_rate": 3.498616236162362e-06, + "loss": 1.4151, + "step": 907 + }, + { + "epoch": 1.1300560049782202, + "grad_norm": 2.906148672103882, + "learning_rate": 3.496309963099631e-06, + "loss": 1.4713, + "step": 908 + }, + { + "epoch": 1.1313005600497823, + "grad_norm": 2.9063987731933594, + "learning_rate": 3.4940036900369005e-06, + "loss": 1.5141, + "step": 909 + }, + { + "epoch": 1.1325451151213441, + "grad_norm": 2.7462973594665527, + "learning_rate": 3.4916974169741704e-06, + "loss": 1.4396, + "step": 910 + }, + { + "epoch": 1.1325451151213441, + "eval_loss": 1.6758365631103516, + "eval_runtime": 41.9838, + "eval_samples_per_second": 23.819, + "eval_steps_per_second": 1.0, + "step": 910 + }, + { + "epoch": 1.133789670192906, + "grad_norm": 2.854579448699951, + "learning_rate": 3.4893911439114394e-06, + "loss": 1.4471, + "step": 911 + }, + { + "epoch": 1.135034225264468, + "grad_norm": 2.8700456619262695, + "learning_rate": 3.487084870848709e-06, + "loss": 1.4237, + "step": 912 + }, + { + "epoch": 1.13627878033603, + "grad_norm": 2.883131742477417, + "learning_rate": 3.4847785977859782e-06, + "loss": 1.4014, + "step": 913 + }, + { + "epoch": 1.1375233354075918, + "grad_norm": 2.9069855213165283, + "learning_rate": 3.4824723247232472e-06, + "loss": 1.396, + "step": 914 + }, + { + "epoch": 1.1387678904791536, + "grad_norm": 2.9044485092163086, + "learning_rate": 3.4801660516605166e-06, + "loss": 1.4516, + "step": 915 + }, + { + "epoch": 1.1400124455507157, + "grad_norm": 2.9939963817596436, + "learning_rate": 3.4778597785977865e-06, + "loss": 1.4347, + "step": 916 + }, + { + "epoch": 1.1412570006222775, + "grad_norm": 2.762651205062866, + "learning_rate": 3.475553505535056e-06, + "loss": 1.4278, + "step": 917 + }, + { + "epoch": 1.1425015556938394, + "grad_norm": 2.84051775932312, + "learning_rate": 3.473247232472325e-06, + "loss": 1.5154, + "step": 918 + }, + { + "epoch": 1.1437461107654014, + "grad_norm": 2.7700562477111816, + "learning_rate": 3.4709409594095943e-06, + "loss": 1.4296, + "step": 919 + }, + { + "epoch": 1.1449906658369633, + "grad_norm": 2.8805646896362305, + "learning_rate": 3.4686346863468638e-06, + "loss": 1.4716, + "step": 920 + }, + { + "epoch": 1.1449906658369633, + "eval_loss": 1.672538161277771, + "eval_runtime": 44.9663, + "eval_samples_per_second": 22.239, + "eval_steps_per_second": 0.934, + "step": 920 + }, + { + "epoch": 1.1462352209085251, + "grad_norm": 2.8342535495758057, + "learning_rate": 3.4663284132841328e-06, + "loss": 1.4461, + "step": 921 + }, + { + "epoch": 1.1474797759800872, + "grad_norm": 3.000023603439331, + "learning_rate": 3.4640221402214026e-06, + "loss": 1.4244, + "step": 922 + }, + { + "epoch": 1.148724331051649, + "grad_norm": 2.7098913192749023, + "learning_rate": 3.461715867158672e-06, + "loss": 1.3943, + "step": 923 + }, + { + "epoch": 1.149968886123211, + "grad_norm": 2.8422343730926514, + "learning_rate": 3.459409594095941e-06, + "loss": 1.456, + "step": 924 + }, + { + "epoch": 1.151213441194773, + "grad_norm": 2.7740139961242676, + "learning_rate": 3.4571033210332105e-06, + "loss": 1.4299, + "step": 925 + }, + { + "epoch": 1.1524579962663348, + "grad_norm": 2.792193651199341, + "learning_rate": 3.4547970479704803e-06, + "loss": 1.4255, + "step": 926 + }, + { + "epoch": 1.1537025513378967, + "grad_norm": 2.786365032196045, + "learning_rate": 3.4524907749077493e-06, + "loss": 1.4674, + "step": 927 + }, + { + "epoch": 1.1549471064094585, + "grad_norm": 2.791139841079712, + "learning_rate": 3.4501845018450187e-06, + "loss": 1.4219, + "step": 928 + }, + { + "epoch": 1.1561916614810206, + "grad_norm": 2.829876184463501, + "learning_rate": 3.447878228782288e-06, + "loss": 1.5115, + "step": 929 + }, + { + "epoch": 1.1574362165525824, + "grad_norm": 2.8779923915863037, + "learning_rate": 3.4455719557195576e-06, + "loss": 1.4484, + "step": 930 + }, + { + "epoch": 1.1574362165525824, + "eval_loss": 1.6719067096710205, + "eval_runtime": 44.4657, + "eval_samples_per_second": 22.489, + "eval_steps_per_second": 0.945, + "step": 930 + }, + { + "epoch": 1.1586807716241443, + "grad_norm": 2.7856647968292236, + "learning_rate": 3.4432656826568266e-06, + "loss": 1.4423, + "step": 931 + }, + { + "epoch": 1.1599253266957064, + "grad_norm": 2.8183937072753906, + "learning_rate": 3.4409594095940964e-06, + "loss": 1.4492, + "step": 932 + }, + { + "epoch": 1.1611698817672682, + "grad_norm": 2.874630928039551, + "learning_rate": 3.438653136531366e-06, + "loss": 1.4546, + "step": 933 + }, + { + "epoch": 1.16241443683883, + "grad_norm": 2.7180988788604736, + "learning_rate": 3.436346863468635e-06, + "loss": 1.4402, + "step": 934 + }, + { + "epoch": 1.1636589919103921, + "grad_norm": 2.682368516921997, + "learning_rate": 3.4340405904059043e-06, + "loss": 1.4442, + "step": 935 + }, + { + "epoch": 1.164903546981954, + "grad_norm": 2.8584909439086914, + "learning_rate": 3.4317343173431737e-06, + "loss": 1.4707, + "step": 936 + }, + { + "epoch": 1.1661481020535158, + "grad_norm": 3.074317693710327, + "learning_rate": 3.4294280442804427e-06, + "loss": 1.5241, + "step": 937 + }, + { + "epoch": 1.167392657125078, + "grad_norm": 2.9097986221313477, + "learning_rate": 3.4271217712177125e-06, + "loss": 1.4518, + "step": 938 + }, + { + "epoch": 1.1686372121966397, + "grad_norm": 2.9661238193511963, + "learning_rate": 3.424815498154982e-06, + "loss": 1.4665, + "step": 939 + }, + { + "epoch": 1.1698817672682016, + "grad_norm": 2.8246936798095703, + "learning_rate": 3.4225092250922514e-06, + "loss": 1.4789, + "step": 940 + }, + { + "epoch": 1.1698817672682016, + "eval_loss": 1.6716506481170654, + "eval_runtime": 43.1065, + "eval_samples_per_second": 23.198, + "eval_steps_per_second": 0.974, + "step": 940 + }, + { + "epoch": 1.1711263223397634, + "grad_norm": 2.9993770122528076, + "learning_rate": 3.4202029520295204e-06, + "loss": 1.5287, + "step": 941 + }, + { + "epoch": 1.1723708774113255, + "grad_norm": 2.7746829986572266, + "learning_rate": 3.41789667896679e-06, + "loss": 1.4105, + "step": 942 + }, + { + "epoch": 1.1736154324828874, + "grad_norm": 2.7988786697387695, + "learning_rate": 3.4155904059040596e-06, + "loss": 1.4543, + "step": 943 + }, + { + "epoch": 1.1748599875544492, + "grad_norm": 2.8851349353790283, + "learning_rate": 3.4132841328413286e-06, + "loss": 1.4527, + "step": 944 + }, + { + "epoch": 1.1761045426260113, + "grad_norm": 2.8439807891845703, + "learning_rate": 3.410977859778598e-06, + "loss": 1.4383, + "step": 945 + }, + { + "epoch": 1.1773490976975731, + "grad_norm": 2.811246156692505, + "learning_rate": 3.4086715867158675e-06, + "loss": 1.4004, + "step": 946 + }, + { + "epoch": 1.178593652769135, + "grad_norm": 2.8452935218811035, + "learning_rate": 3.4063653136531365e-06, + "loss": 1.4989, + "step": 947 + }, + { + "epoch": 1.179838207840697, + "grad_norm": 2.7633206844329834, + "learning_rate": 3.4040590405904063e-06, + "loss": 1.4644, + "step": 948 + }, + { + "epoch": 1.1810827629122589, + "grad_norm": 2.777296543121338, + "learning_rate": 3.4017527675276758e-06, + "loss": 1.4398, + "step": 949 + }, + { + "epoch": 1.1823273179838207, + "grad_norm": 3.0879335403442383, + "learning_rate": 3.3994464944649448e-06, + "loss": 1.4901, + "step": 950 + }, + { + "epoch": 1.1823273179838207, + "eval_loss": 1.6718789339065552, + "eval_runtime": 43.7598, + "eval_samples_per_second": 22.852, + "eval_steps_per_second": 0.96, + "step": 950 + }, + { + "epoch": 1.1835718730553828, + "grad_norm": 3.030923366546631, + "learning_rate": 3.397140221402214e-06, + "loss": 1.4992, + "step": 951 + }, + { + "epoch": 1.1848164281269447, + "grad_norm": 2.7215065956115723, + "learning_rate": 3.3948339483394836e-06, + "loss": 1.4709, + "step": 952 + }, + { + "epoch": 1.1860609831985065, + "grad_norm": 2.738738775253296, + "learning_rate": 3.3925276752767534e-06, + "loss": 1.4838, + "step": 953 + }, + { + "epoch": 1.1873055382700684, + "grad_norm": 2.8057045936584473, + "learning_rate": 3.3902214022140224e-06, + "loss": 1.4879, + "step": 954 + }, + { + "epoch": 1.1885500933416304, + "grad_norm": 2.846531867980957, + "learning_rate": 3.387915129151292e-06, + "loss": 1.4585, + "step": 955 + }, + { + "epoch": 1.1897946484131923, + "grad_norm": 2.8825955390930176, + "learning_rate": 3.3856088560885613e-06, + "loss": 1.5483, + "step": 956 + }, + { + "epoch": 1.1910392034847541, + "grad_norm": 2.720550775527954, + "learning_rate": 3.3833025830258303e-06, + "loss": 1.4356, + "step": 957 + }, + { + "epoch": 1.1922837585563162, + "grad_norm": 2.7653985023498535, + "learning_rate": 3.3809963099630997e-06, + "loss": 1.4679, + "step": 958 + }, + { + "epoch": 1.193528313627878, + "grad_norm": 2.868544101715088, + "learning_rate": 3.3786900369003696e-06, + "loss": 1.4492, + "step": 959 + }, + { + "epoch": 1.1947728686994399, + "grad_norm": 2.877669334411621, + "learning_rate": 3.3763837638376386e-06, + "loss": 1.493, + "step": 960 + }, + { + "epoch": 1.1947728686994399, + "eval_loss": 1.6707913875579834, + "eval_runtime": 44.0558, + "eval_samples_per_second": 22.698, + "eval_steps_per_second": 0.953, + "step": 960 + }, + { + "epoch": 1.196017423771002, + "grad_norm": 3.0898303985595703, + "learning_rate": 3.374077490774908e-06, + "loss": 1.4336, + "step": 961 + }, + { + "epoch": 1.1972619788425638, + "grad_norm": 2.859119415283203, + "learning_rate": 3.3717712177121774e-06, + "loss": 1.4862, + "step": 962 + }, + { + "epoch": 1.1985065339141256, + "grad_norm": 2.9307544231414795, + "learning_rate": 3.3694649446494464e-06, + "loss": 1.4977, + "step": 963 + }, + { + "epoch": 1.1997510889856877, + "grad_norm": 2.785691976547241, + "learning_rate": 3.3671586715867163e-06, + "loss": 1.4203, + "step": 964 + }, + { + "epoch": 1.2009956440572496, + "grad_norm": 2.762620449066162, + "learning_rate": 3.3648523985239857e-06, + "loss": 1.5147, + "step": 965 + }, + { + "epoch": 1.2022401991288114, + "grad_norm": 2.8887553215026855, + "learning_rate": 3.362546125461255e-06, + "loss": 1.4435, + "step": 966 + }, + { + "epoch": 1.2034847542003733, + "grad_norm": 2.8161838054656982, + "learning_rate": 3.360239852398524e-06, + "loss": 1.4436, + "step": 967 + }, + { + "epoch": 1.2047293092719353, + "grad_norm": 3.0834996700286865, + "learning_rate": 3.3579335793357935e-06, + "loss": 1.4861, + "step": 968 + }, + { + "epoch": 1.2059738643434972, + "grad_norm": 3.0328774452209473, + "learning_rate": 3.3556273062730634e-06, + "loss": 1.5064, + "step": 969 + }, + { + "epoch": 1.207218419415059, + "grad_norm": 2.977429151535034, + "learning_rate": 3.3533210332103324e-06, + "loss": 1.4148, + "step": 970 + }, + { + "epoch": 1.207218419415059, + "eval_loss": 1.672300100326538, + "eval_runtime": 46.0257, + "eval_samples_per_second": 21.727, + "eval_steps_per_second": 0.913, + "step": 970 + }, + { + "epoch": 1.208462974486621, + "grad_norm": 2.9340717792510986, + "learning_rate": 3.3510147601476018e-06, + "loss": 1.4726, + "step": 971 + }, + { + "epoch": 1.209707529558183, + "grad_norm": 2.8549611568450928, + "learning_rate": 3.348708487084871e-06, + "loss": 1.4798, + "step": 972 + }, + { + "epoch": 1.2109520846297448, + "grad_norm": 2.74188232421875, + "learning_rate": 3.34640221402214e-06, + "loss": 1.4231, + "step": 973 + }, + { + "epoch": 1.2121966397013069, + "grad_norm": 2.8730149269104004, + "learning_rate": 3.3440959409594096e-06, + "loss": 1.4135, + "step": 974 + }, + { + "epoch": 1.2134411947728687, + "grad_norm": 2.844428062438965, + "learning_rate": 3.3417896678966795e-06, + "loss": 1.4164, + "step": 975 + }, + { + "epoch": 1.2146857498444306, + "grad_norm": 2.920804738998413, + "learning_rate": 3.3394833948339485e-06, + "loss": 1.4898, + "step": 976 + }, + { + "epoch": 1.2159303049159926, + "grad_norm": 2.9339325428009033, + "learning_rate": 3.337177121771218e-06, + "loss": 1.5112, + "step": 977 + }, + { + "epoch": 1.2171748599875545, + "grad_norm": 2.7855703830718994, + "learning_rate": 3.3348708487084873e-06, + "loss": 1.4209, + "step": 978 + }, + { + "epoch": 1.2184194150591163, + "grad_norm": 2.9347097873687744, + "learning_rate": 3.332564575645757e-06, + "loss": 1.4345, + "step": 979 + }, + { + "epoch": 1.2196639701306782, + "grad_norm": 2.9517159461975098, + "learning_rate": 3.330258302583026e-06, + "loss": 1.4873, + "step": 980 + }, + { + "epoch": 1.2196639701306782, + "eval_loss": 1.6751104593276978, + "eval_runtime": 45.8657, + "eval_samples_per_second": 21.803, + "eval_steps_per_second": 0.916, + "step": 980 + }, + { + "epoch": 1.2209085252022402, + "grad_norm": 3.043290853500366, + "learning_rate": 3.3279520295202956e-06, + "loss": 1.5121, + "step": 981 + }, + { + "epoch": 1.222153080273802, + "grad_norm": 2.9060018062591553, + "learning_rate": 3.325645756457565e-06, + "loss": 1.4888, + "step": 982 + }, + { + "epoch": 1.223397635345364, + "grad_norm": 3.186314344406128, + "learning_rate": 3.323339483394834e-06, + "loss": 1.4944, + "step": 983 + }, + { + "epoch": 1.224642190416926, + "grad_norm": 3.2327520847320557, + "learning_rate": 3.3210332103321034e-06, + "loss": 1.515, + "step": 984 + }, + { + "epoch": 1.2258867454884879, + "grad_norm": 2.9379491806030273, + "learning_rate": 3.3187269372693733e-06, + "loss": 1.4475, + "step": 985 + }, + { + "epoch": 1.2271313005600497, + "grad_norm": 2.732428789138794, + "learning_rate": 3.3164206642066423e-06, + "loss": 1.4595, + "step": 986 + }, + { + "epoch": 1.2283758556316118, + "grad_norm": 2.745485782623291, + "learning_rate": 3.3141143911439117e-06, + "loss": 1.4716, + "step": 987 + }, + { + "epoch": 1.2296204107031736, + "grad_norm": 3.172178268432617, + "learning_rate": 3.311808118081181e-06, + "loss": 1.4901, + "step": 988 + }, + { + "epoch": 1.2308649657747355, + "grad_norm": 2.8245432376861572, + "learning_rate": 3.3095018450184506e-06, + "loss": 1.4738, + "step": 989 + }, + { + "epoch": 1.2321095208462975, + "grad_norm": 2.8529610633850098, + "learning_rate": 3.3071955719557196e-06, + "loss": 1.4549, + "step": 990 + }, + { + "epoch": 1.2321095208462975, + "eval_loss": 1.6769033670425415, + "eval_runtime": 45.7162, + "eval_samples_per_second": 21.874, + "eval_steps_per_second": 0.919, + "step": 990 + }, + { + "epoch": 1.2333540759178594, + "grad_norm": 2.7896978855133057, + "learning_rate": 3.3048892988929894e-06, + "loss": 1.4207, + "step": 991 + }, + { + "epoch": 1.2345986309894212, + "grad_norm": 3.011547088623047, + "learning_rate": 3.302583025830259e-06, + "loss": 1.4615, + "step": 992 + }, + { + "epoch": 1.235843186060983, + "grad_norm": 2.838911771774292, + "learning_rate": 3.300276752767528e-06, + "loss": 1.4539, + "step": 993 + }, + { + "epoch": 1.2370877411325452, + "grad_norm": 2.731318235397339, + "learning_rate": 3.2979704797047972e-06, + "loss": 1.4452, + "step": 994 + }, + { + "epoch": 1.238332296204107, + "grad_norm": 2.7334372997283936, + "learning_rate": 3.2956642066420667e-06, + "loss": 1.4658, + "step": 995 + }, + { + "epoch": 1.2395768512756689, + "grad_norm": 2.8925108909606934, + "learning_rate": 3.2933579335793357e-06, + "loss": 1.4368, + "step": 996 + }, + { + "epoch": 1.240821406347231, + "grad_norm": 2.8550665378570557, + "learning_rate": 3.2910516605166055e-06, + "loss": 1.4428, + "step": 997 + }, + { + "epoch": 1.2420659614187928, + "grad_norm": 2.953028440475464, + "learning_rate": 3.288745387453875e-06, + "loss": 1.5585, + "step": 998 + }, + { + "epoch": 1.2433105164903546, + "grad_norm": 2.833862781524658, + "learning_rate": 3.286439114391144e-06, + "loss": 1.5013, + "step": 999 + }, + { + "epoch": 1.2445550715619167, + "grad_norm": 3.0078632831573486, + "learning_rate": 3.2841328413284134e-06, + "loss": 1.421, + "step": 1000 + }, + { + "epoch": 1.2445550715619167, + "eval_loss": 1.6693010330200195, + "eval_runtime": 46.1434, + "eval_samples_per_second": 21.672, + "eval_steps_per_second": 0.91, + "step": 1000 + }, + { + "epoch": 1.2457996266334785, + "grad_norm": 3.1139776706695557, + "learning_rate": 3.281826568265683e-06, + "loss": 1.4494, + "step": 1001 + }, + { + "epoch": 1.2470441817050404, + "grad_norm": 2.6980462074279785, + "learning_rate": 3.2795202952029526e-06, + "loss": 1.4689, + "step": 1002 + }, + { + "epoch": 1.2482887367766025, + "grad_norm": 2.797846794128418, + "learning_rate": 3.2772140221402216e-06, + "loss": 1.3867, + "step": 1003 + }, + { + "epoch": 1.2495332918481643, + "grad_norm": 2.949923515319824, + "learning_rate": 3.274907749077491e-06, + "loss": 1.4953, + "step": 1004 + }, + { + "epoch": 1.2507778469197262, + "grad_norm": 3.0071775913238525, + "learning_rate": 3.2726014760147605e-06, + "loss": 1.507, + "step": 1005 + }, + { + "epoch": 1.252022401991288, + "grad_norm": 2.9245765209198, + "learning_rate": 3.2702952029520295e-06, + "loss": 1.497, + "step": 1006 + }, + { + "epoch": 1.25326695706285, + "grad_norm": 2.755397081375122, + "learning_rate": 3.2679889298892993e-06, + "loss": 1.4847, + "step": 1007 + }, + { + "epoch": 1.254511512134412, + "grad_norm": 2.8560867309570312, + "learning_rate": 3.2656826568265687e-06, + "loss": 1.4446, + "step": 1008 + }, + { + "epoch": 1.255756067205974, + "grad_norm": 2.805844783782959, + "learning_rate": 3.2633763837638377e-06, + "loss": 1.4877, + "step": 1009 + }, + { + "epoch": 1.2570006222775358, + "grad_norm": 3.004056453704834, + "learning_rate": 3.261070110701107e-06, + "loss": 1.4706, + "step": 1010 + }, + { + "epoch": 1.2570006222775358, + "eval_loss": 1.662219762802124, + "eval_runtime": 42.45, + "eval_samples_per_second": 23.557, + "eval_steps_per_second": 0.989, + "step": 1010 + }, + { + "epoch": 1.2582451773490977, + "grad_norm": 2.9530556201934814, + "learning_rate": 3.2587638376383766e-06, + "loss": 1.4623, + "step": 1011 + }, + { + "epoch": 1.2594897324206595, + "grad_norm": 2.8305587768554688, + "learning_rate": 3.2564575645756456e-06, + "loss": 1.4659, + "step": 1012 + }, + { + "epoch": 1.2607342874922216, + "grad_norm": 2.895482063293457, + "learning_rate": 3.2541512915129154e-06, + "loss": 1.4267, + "step": 1013 + }, + { + "epoch": 1.2619788425637835, + "grad_norm": 2.9198293685913086, + "learning_rate": 3.251845018450185e-06, + "loss": 1.4226, + "step": 1014 + }, + { + "epoch": 1.2632233976353453, + "grad_norm": 2.9762983322143555, + "learning_rate": 3.2495387453874543e-06, + "loss": 1.4666, + "step": 1015 + }, + { + "epoch": 1.2644679527069074, + "grad_norm": 2.905694007873535, + "learning_rate": 3.2472324723247233e-06, + "loss": 1.445, + "step": 1016 + }, + { + "epoch": 1.2657125077784692, + "grad_norm": 2.8624324798583984, + "learning_rate": 3.244926199261993e-06, + "loss": 1.4685, + "step": 1017 + }, + { + "epoch": 1.266957062850031, + "grad_norm": 2.8351657390594482, + "learning_rate": 3.2426199261992625e-06, + "loss": 1.4477, + "step": 1018 + }, + { + "epoch": 1.268201617921593, + "grad_norm": 3.0314555168151855, + "learning_rate": 3.2403136531365315e-06, + "loss": 1.4744, + "step": 1019 + }, + { + "epoch": 1.269446172993155, + "grad_norm": 2.848743438720703, + "learning_rate": 3.238007380073801e-06, + "loss": 1.3832, + "step": 1020 + }, + { + "epoch": 1.269446172993155, + "eval_loss": 1.6641864776611328, + "eval_runtime": 44.6957, + "eval_samples_per_second": 22.374, + "eval_steps_per_second": 0.94, + "step": 1020 + }, + { + "epoch": 1.2706907280647168, + "grad_norm": 2.8528616428375244, + "learning_rate": 3.2357011070110704e-06, + "loss": 1.486, + "step": 1021 + }, + { + "epoch": 1.271935283136279, + "grad_norm": 3.064131259918213, + "learning_rate": 3.2333948339483394e-06, + "loss": 1.4449, + "step": 1022 + }, + { + "epoch": 1.2731798382078408, + "grad_norm": 3.1047396659851074, + "learning_rate": 3.2310885608856092e-06, + "loss": 1.4796, + "step": 1023 + }, + { + "epoch": 1.2744243932794026, + "grad_norm": 2.897571086883545, + "learning_rate": 3.2287822878228787e-06, + "loss": 1.4762, + "step": 1024 + }, + { + "epoch": 1.2756689483509644, + "grad_norm": 2.914445161819458, + "learning_rate": 3.2264760147601477e-06, + "loss": 1.4315, + "step": 1025 + }, + { + "epoch": 1.2769135034225265, + "grad_norm": 2.913105010986328, + "learning_rate": 3.224169741697417e-06, + "loss": 1.4314, + "step": 1026 + }, + { + "epoch": 1.2781580584940884, + "grad_norm": 2.7426509857177734, + "learning_rate": 3.2218634686346865e-06, + "loss": 1.4006, + "step": 1027 + }, + { + "epoch": 1.2794026135656502, + "grad_norm": 3.15177845954895, + "learning_rate": 3.2195571955719564e-06, + "loss": 1.4426, + "step": 1028 + }, + { + "epoch": 1.2806471686372123, + "grad_norm": 2.9131808280944824, + "learning_rate": 3.2172509225092254e-06, + "loss": 1.5067, + "step": 1029 + }, + { + "epoch": 1.2818917237087741, + "grad_norm": 2.9975693225860596, + "learning_rate": 3.2149446494464948e-06, + "loss": 1.4612, + "step": 1030 + }, + { + "epoch": 1.2818917237087741, + "eval_loss": 1.6711934804916382, + "eval_runtime": 50.154, + "eval_samples_per_second": 19.939, + "eval_steps_per_second": 0.837, + "step": 1030 + }, + { + "epoch": 1.283136278780336, + "grad_norm": 2.9257564544677734, + "learning_rate": 3.212638376383764e-06, + "loss": 1.4383, + "step": 1031 + }, + { + "epoch": 1.2843808338518978, + "grad_norm": 2.9188551902770996, + "learning_rate": 3.210332103321033e-06, + "loss": 1.5097, + "step": 1032 + }, + { + "epoch": 1.28562538892346, + "grad_norm": 2.9625136852264404, + "learning_rate": 3.208025830258303e-06, + "loss": 1.4127, + "step": 1033 + }, + { + "epoch": 1.2868699439950217, + "grad_norm": 2.6977076530456543, + "learning_rate": 3.2057195571955725e-06, + "loss": 1.4163, + "step": 1034 + }, + { + "epoch": 1.2881144990665838, + "grad_norm": 2.711514949798584, + "learning_rate": 3.2034132841328415e-06, + "loss": 1.4292, + "step": 1035 + }, + { + "epoch": 1.2893590541381457, + "grad_norm": 2.8522849082946777, + "learning_rate": 3.201107011070111e-06, + "loss": 1.4386, + "step": 1036 + }, + { + "epoch": 1.2906036092097075, + "grad_norm": 3.0120058059692383, + "learning_rate": 3.1988007380073803e-06, + "loss": 1.4429, + "step": 1037 + }, + { + "epoch": 1.2918481642812694, + "grad_norm": 2.865572929382324, + "learning_rate": 3.1964944649446493e-06, + "loss": 1.4766, + "step": 1038 + }, + { + "epoch": 1.2930927193528314, + "grad_norm": 2.8616085052490234, + "learning_rate": 3.194188191881919e-06, + "loss": 1.4391, + "step": 1039 + }, + { + "epoch": 1.2943372744243933, + "grad_norm": 2.7028727531433105, + "learning_rate": 3.1918819188191886e-06, + "loss": 1.4141, + "step": 1040 + }, + { + "epoch": 1.2943372744243933, + "eval_loss": 1.6724774837493896, + "eval_runtime": 50.8274, + "eval_samples_per_second": 19.674, + "eval_steps_per_second": 0.826, + "step": 1040 + }, + { + "epoch": 1.2955818294959551, + "grad_norm": 2.8322980403900146, + "learning_rate": 3.189575645756458e-06, + "loss": 1.4789, + "step": 1041 + }, + { + "epoch": 1.2968263845675172, + "grad_norm": 2.979027032852173, + "learning_rate": 3.187269372693727e-06, + "loss": 1.4945, + "step": 1042 + }, + { + "epoch": 1.298070939639079, + "grad_norm": 3.0695791244506836, + "learning_rate": 3.1849630996309964e-06, + "loss": 1.5108, + "step": 1043 + }, + { + "epoch": 1.299315494710641, + "grad_norm": 2.7639617919921875, + "learning_rate": 3.1826568265682663e-06, + "loss": 1.4381, + "step": 1044 + }, + { + "epoch": 1.3005600497822027, + "grad_norm": 2.865220308303833, + "learning_rate": 3.1803505535055353e-06, + "loss": 1.4787, + "step": 1045 + }, + { + "epoch": 1.3018046048537648, + "grad_norm": 2.9786012172698975, + "learning_rate": 3.1780442804428047e-06, + "loss": 1.4356, + "step": 1046 + }, + { + "epoch": 1.3030491599253267, + "grad_norm": 2.867659568786621, + "learning_rate": 3.175738007380074e-06, + "loss": 1.4831, + "step": 1047 + }, + { + "epoch": 1.3042937149968887, + "grad_norm": 2.7092227935791016, + "learning_rate": 3.173431734317343e-06, + "loss": 1.4189, + "step": 1048 + }, + { + "epoch": 1.3055382700684506, + "grad_norm": 2.85172176361084, + "learning_rate": 3.1711254612546125e-06, + "loss": 1.4252, + "step": 1049 + }, + { + "epoch": 1.3067828251400124, + "grad_norm": 2.9820401668548584, + "learning_rate": 3.1688191881918824e-06, + "loss": 1.4924, + "step": 1050 + }, + { + "epoch": 1.3067828251400124, + "eval_loss": 1.672357201576233, + "eval_runtime": 53.7158, + "eval_samples_per_second": 18.616, + "eval_steps_per_second": 0.782, + "step": 1050 + }, + { + "epoch": 1.3080273802115743, + "grad_norm": 2.9170854091644287, + "learning_rate": 3.166512915129152e-06, + "loss": 1.4698, + "step": 1051 + }, + { + "epoch": 1.3092719352831363, + "grad_norm": 2.9447641372680664, + "learning_rate": 3.164206642066421e-06, + "loss": 1.4571, + "step": 1052 + }, + { + "epoch": 1.3105164903546982, + "grad_norm": 2.895512342453003, + "learning_rate": 3.1619003690036902e-06, + "loss": 1.4373, + "step": 1053 + }, + { + "epoch": 1.31176104542626, + "grad_norm": 2.82749605178833, + "learning_rate": 3.15959409594096e-06, + "loss": 1.4487, + "step": 1054 + }, + { + "epoch": 1.3130056004978221, + "grad_norm": 2.9615845680236816, + "learning_rate": 3.157287822878229e-06, + "loss": 1.4917, + "step": 1055 + }, + { + "epoch": 1.314250155569384, + "grad_norm": 2.8264551162719727, + "learning_rate": 3.1549815498154985e-06, + "loss": 1.4197, + "step": 1056 + }, + { + "epoch": 1.3154947106409458, + "grad_norm": 2.8582863807678223, + "learning_rate": 3.152675276752768e-06, + "loss": 1.4882, + "step": 1057 + }, + { + "epoch": 1.3167392657125077, + "grad_norm": 3.049182653427124, + "learning_rate": 3.150369003690037e-06, + "loss": 1.5096, + "step": 1058 + }, + { + "epoch": 1.3179838207840697, + "grad_norm": 2.803922653198242, + "learning_rate": 3.1480627306273063e-06, + "loss": 1.4287, + "step": 1059 + }, + { + "epoch": 1.3192283758556316, + "grad_norm": 2.8965165615081787, + "learning_rate": 3.145756457564576e-06, + "loss": 1.4786, + "step": 1060 + }, + { + "epoch": 1.3192283758556316, + "eval_loss": 1.6656724214553833, + "eval_runtime": 53.0223, + "eval_samples_per_second": 18.86, + "eval_steps_per_second": 0.792, + "step": 1060 + }, + { + "epoch": 1.3204729309271936, + "grad_norm": 3.2148702144622803, + "learning_rate": 3.143450184501845e-06, + "loss": 1.4997, + "step": 1061 + }, + { + "epoch": 1.3217174859987555, + "grad_norm": 3.013700246810913, + "learning_rate": 3.1411439114391146e-06, + "loss": 1.4835, + "step": 1062 + }, + { + "epoch": 1.3229620410703173, + "grad_norm": 2.879260540008545, + "learning_rate": 3.138837638376384e-06, + "loss": 1.42, + "step": 1063 + }, + { + "epoch": 1.3242065961418792, + "grad_norm": 2.6866135597229004, + "learning_rate": 3.1365313653136535e-06, + "loss": 1.4037, + "step": 1064 + }, + { + "epoch": 1.3254511512134413, + "grad_norm": 2.8592655658721924, + "learning_rate": 3.1342250922509225e-06, + "loss": 1.443, + "step": 1065 + }, + { + "epoch": 1.326695706285003, + "grad_norm": 3.1136722564697266, + "learning_rate": 3.1319188191881923e-06, + "loss": 1.469, + "step": 1066 + }, + { + "epoch": 1.327940261356565, + "grad_norm": 2.8858542442321777, + "learning_rate": 3.1296125461254617e-06, + "loss": 1.4525, + "step": 1067 + }, + { + "epoch": 1.329184816428127, + "grad_norm": 3.01088809967041, + "learning_rate": 3.1273062730627307e-06, + "loss": 1.4017, + "step": 1068 + }, + { + "epoch": 1.3304293714996889, + "grad_norm": 3.185218334197998, + "learning_rate": 3.125e-06, + "loss": 1.4936, + "step": 1069 + }, + { + "epoch": 1.3316739265712507, + "grad_norm": 3.168509006500244, + "learning_rate": 3.12269372693727e-06, + "loss": 1.4512, + "step": 1070 + }, + { + "epoch": 1.3316739265712507, + "eval_loss": 1.66367506980896, + "eval_runtime": 51.2724, + "eval_samples_per_second": 19.504, + "eval_steps_per_second": 0.819, + "step": 1070 + }, + { + "epoch": 1.3329184816428126, + "grad_norm": 3.036905527114868, + "learning_rate": 3.120387453874539e-06, + "loss": 1.4057, + "step": 1071 + }, + { + "epoch": 1.3341630367143746, + "grad_norm": 3.114818572998047, + "learning_rate": 3.1180811808118084e-06, + "loss": 1.4687, + "step": 1072 + }, + { + "epoch": 1.3354075917859365, + "grad_norm": 2.929135799407959, + "learning_rate": 3.115774907749078e-06, + "loss": 1.5013, + "step": 1073 + }, + { + "epoch": 1.3366521468574986, + "grad_norm": 3.1101112365722656, + "learning_rate": 3.113468634686347e-06, + "loss": 1.4222, + "step": 1074 + }, + { + "epoch": 1.3378967019290604, + "grad_norm": 2.9812445640563965, + "learning_rate": 3.1111623616236163e-06, + "loss": 1.4476, + "step": 1075 + }, + { + "epoch": 1.3391412570006223, + "grad_norm": 2.7879486083984375, + "learning_rate": 3.108856088560886e-06, + "loss": 1.4228, + "step": 1076 + }, + { + "epoch": 1.340385812072184, + "grad_norm": 2.9183881282806396, + "learning_rate": 3.1065498154981555e-06, + "loss": 1.4768, + "step": 1077 + }, + { + "epoch": 1.3416303671437462, + "grad_norm": 3.136688709259033, + "learning_rate": 3.1042435424354245e-06, + "loss": 1.4867, + "step": 1078 + }, + { + "epoch": 1.342874922215308, + "grad_norm": 3.087040424346924, + "learning_rate": 3.101937269372694e-06, + "loss": 1.4912, + "step": 1079 + }, + { + "epoch": 1.3441194772868699, + "grad_norm": 2.981447696685791, + "learning_rate": 3.0996309963099634e-06, + "loss": 1.4133, + "step": 1080 + }, + { + "epoch": 1.3441194772868699, + "eval_loss": 1.6594599485397339, + "eval_runtime": 50.0245, + "eval_samples_per_second": 19.99, + "eval_steps_per_second": 0.84, + "step": 1080 + }, + { + "epoch": 1.345364032358432, + "grad_norm": 3.1392548084259033, + "learning_rate": 3.0973247232472324e-06, + "loss": 1.4131, + "step": 1081 + }, + { + "epoch": 1.3466085874299938, + "grad_norm": 2.813328742980957, + "learning_rate": 3.0950184501845022e-06, + "loss": 1.4461, + "step": 1082 + }, + { + "epoch": 1.3478531425015556, + "grad_norm": 2.930349111557007, + "learning_rate": 3.0927121771217716e-06, + "loss": 1.4423, + "step": 1083 + }, + { + "epoch": 1.3490976975731175, + "grad_norm": 2.8514888286590576, + "learning_rate": 3.0904059040590406e-06, + "loss": 1.4499, + "step": 1084 + }, + { + "epoch": 1.3503422526446796, + "grad_norm": 2.8733901977539062, + "learning_rate": 3.08809963099631e-06, + "loss": 1.4425, + "step": 1085 + }, + { + "epoch": 1.3515868077162414, + "grad_norm": 2.8833565711975098, + "learning_rate": 3.0857933579335795e-06, + "loss": 1.4808, + "step": 1086 + }, + { + "epoch": 1.3528313627878035, + "grad_norm": 2.9774460792541504, + "learning_rate": 3.0834870848708485e-06, + "loss": 1.5147, + "step": 1087 + }, + { + "epoch": 1.3540759178593653, + "grad_norm": 2.9540138244628906, + "learning_rate": 3.0811808118081183e-06, + "loss": 1.4577, + "step": 1088 + }, + { + "epoch": 1.3553204729309272, + "grad_norm": 2.9065780639648438, + "learning_rate": 3.0788745387453878e-06, + "loss": 1.4564, + "step": 1089 + }, + { + "epoch": 1.356565028002489, + "grad_norm": 3.2054378986358643, + "learning_rate": 3.076568265682657e-06, + "loss": 1.4435, + "step": 1090 + }, + { + "epoch": 1.356565028002489, + "eval_loss": 1.6565171480178833, + "eval_runtime": 51.0475, + "eval_samples_per_second": 19.59, + "eval_steps_per_second": 0.823, + "step": 1090 + }, + { + "epoch": 1.357809583074051, + "grad_norm": 2.9422333240509033, + "learning_rate": 3.074261992619926e-06, + "loss": 1.4844, + "step": 1091 + }, + { + "epoch": 1.359054138145613, + "grad_norm": 2.738215684890747, + "learning_rate": 3.071955719557196e-06, + "loss": 1.4349, + "step": 1092 + }, + { + "epoch": 1.3602986932171748, + "grad_norm": 3.032463312149048, + "learning_rate": 3.0696494464944655e-06, + "loss": 1.4133, + "step": 1093 + }, + { + "epoch": 1.3615432482887369, + "grad_norm": 2.8251936435699463, + "learning_rate": 3.0673431734317345e-06, + "loss": 1.4751, + "step": 1094 + }, + { + "epoch": 1.3627878033602987, + "grad_norm": 2.8674609661102295, + "learning_rate": 3.065036900369004e-06, + "loss": 1.464, + "step": 1095 + }, + { + "epoch": 1.3640323584318605, + "grad_norm": 2.8118209838867188, + "learning_rate": 3.0627306273062733e-06, + "loss": 1.476, + "step": 1096 + }, + { + "epoch": 1.3652769135034224, + "grad_norm": 3.152803897857666, + "learning_rate": 3.0604243542435423e-06, + "loss": 1.4209, + "step": 1097 + }, + { + "epoch": 1.3665214685749845, + "grad_norm": 2.8795511722564697, + "learning_rate": 3.058118081180812e-06, + "loss": 1.462, + "step": 1098 + }, + { + "epoch": 1.3677660236465463, + "grad_norm": 2.897827625274658, + "learning_rate": 3.0558118081180816e-06, + "loss": 1.4194, + "step": 1099 + }, + { + "epoch": 1.3690105787181084, + "grad_norm": 2.9295172691345215, + "learning_rate": 3.053505535055351e-06, + "loss": 1.4742, + "step": 1100 + }, + { + "epoch": 1.3690105787181084, + "eval_loss": 1.6628323793411255, + "eval_runtime": 52.0227, + "eval_samples_per_second": 19.222, + "eval_steps_per_second": 0.807, + "step": 1100 + }, + { + "epoch": 1.3702551337896702, + "grad_norm": 2.79770565032959, + "learning_rate": 3.05119926199262e-06, + "loss": 1.3923, + "step": 1101 + }, + { + "epoch": 1.371499688861232, + "grad_norm": 2.8151590824127197, + "learning_rate": 3.0488929889298894e-06, + "loss": 1.4396, + "step": 1102 + }, + { + "epoch": 1.372744243932794, + "grad_norm": 2.9383020401000977, + "learning_rate": 3.0465867158671593e-06, + "loss": 1.476, + "step": 1103 + }, + { + "epoch": 1.373988799004356, + "grad_norm": 2.8010449409484863, + "learning_rate": 3.0442804428044283e-06, + "loss": 1.4579, + "step": 1104 + }, + { + "epoch": 1.3752333540759178, + "grad_norm": 2.8714241981506348, + "learning_rate": 3.0419741697416977e-06, + "loss": 1.4921, + "step": 1105 + }, + { + "epoch": 1.3764779091474797, + "grad_norm": 3.0715506076812744, + "learning_rate": 3.039667896678967e-06, + "loss": 1.4102, + "step": 1106 + }, + { + "epoch": 1.3777224642190418, + "grad_norm": 2.8336658477783203, + "learning_rate": 3.037361623616236e-06, + "loss": 1.4065, + "step": 1107 + }, + { + "epoch": 1.3789670192906036, + "grad_norm": 2.833709716796875, + "learning_rate": 3.035055350553506e-06, + "loss": 1.3485, + "step": 1108 + }, + { + "epoch": 1.3802115743621655, + "grad_norm": 2.970330238342285, + "learning_rate": 3.0327490774907754e-06, + "loss": 1.3942, + "step": 1109 + }, + { + "epoch": 1.3814561294337273, + "grad_norm": 2.932877779006958, + "learning_rate": 3.0304428044280444e-06, + "loss": 1.4281, + "step": 1110 + }, + { + "epoch": 1.3814561294337273, + "eval_loss": 1.6577523946762085, + "eval_runtime": 42.2764, + "eval_samples_per_second": 23.654, + "eval_steps_per_second": 0.993, + "step": 1110 + }, + { + "epoch": 1.3827006845052894, + "grad_norm": 3.1917965412139893, + "learning_rate": 3.028136531365314e-06, + "loss": 1.4662, + "step": 1111 + }, + { + "epoch": 1.3839452395768512, + "grad_norm": 2.867098331451416, + "learning_rate": 3.0258302583025832e-06, + "loss": 1.4471, + "step": 1112 + }, + { + "epoch": 1.3851897946484133, + "grad_norm": 2.8024351596832275, + "learning_rate": 3.023523985239853e-06, + "loss": 1.4211, + "step": 1113 + }, + { + "epoch": 1.3864343497199751, + "grad_norm": 2.7674360275268555, + "learning_rate": 3.021217712177122e-06, + "loss": 1.3762, + "step": 1114 + }, + { + "epoch": 1.387678904791537, + "grad_norm": 2.873400926589966, + "learning_rate": 3.0189114391143915e-06, + "loss": 1.4825, + "step": 1115 + }, + { + "epoch": 1.3889234598630988, + "grad_norm": 2.9294228553771973, + "learning_rate": 3.016605166051661e-06, + "loss": 1.4855, + "step": 1116 + }, + { + "epoch": 1.390168014934661, + "grad_norm": 2.8885910511016846, + "learning_rate": 3.01429889298893e-06, + "loss": 1.4196, + "step": 1117 + }, + { + "epoch": 1.3914125700062228, + "grad_norm": 2.7579641342163086, + "learning_rate": 3.0119926199261993e-06, + "loss": 1.4174, + "step": 1118 + }, + { + "epoch": 1.3926571250777848, + "grad_norm": 2.8475749492645264, + "learning_rate": 3.009686346863469e-06, + "loss": 1.4271, + "step": 1119 + }, + { + "epoch": 1.3939016801493467, + "grad_norm": 2.8472487926483154, + "learning_rate": 3.007380073800738e-06, + "loss": 1.4649, + "step": 1120 + }, + { + "epoch": 1.3939016801493467, + "eval_loss": 1.6544252634048462, + "eval_runtime": 45.034, + "eval_samples_per_second": 22.205, + "eval_steps_per_second": 0.933, + "step": 1120 + }, + { + "epoch": 1.3951462352209085, + "grad_norm": 2.916067123413086, + "learning_rate": 3.0050738007380076e-06, + "loss": 1.4134, + "step": 1121 + }, + { + "epoch": 1.3963907902924704, + "grad_norm": 2.8867766857147217, + "learning_rate": 3.002767527675277e-06, + "loss": 1.4378, + "step": 1122 + }, + { + "epoch": 1.3976353453640322, + "grad_norm": 2.844597101211548, + "learning_rate": 3.000461254612546e-06, + "loss": 1.4521, + "step": 1123 + }, + { + "epoch": 1.3988799004355943, + "grad_norm": 2.850982427597046, + "learning_rate": 2.998154981549816e-06, + "loss": 1.3948, + "step": 1124 + }, + { + "epoch": 1.4001244555071561, + "grad_norm": 2.813525438308716, + "learning_rate": 2.9958487084870853e-06, + "loss": 1.457, + "step": 1125 + }, + { + "epoch": 1.4013690105787182, + "grad_norm": 2.9029061794281006, + "learning_rate": 2.9935424354243547e-06, + "loss": 1.4657, + "step": 1126 + }, + { + "epoch": 1.40261356565028, + "grad_norm": 2.8818368911743164, + "learning_rate": 2.9912361623616237e-06, + "loss": 1.4283, + "step": 1127 + }, + { + "epoch": 1.403858120721842, + "grad_norm": 2.8317973613739014, + "learning_rate": 2.988929889298893e-06, + "loss": 1.4019, + "step": 1128 + }, + { + "epoch": 1.4051026757934038, + "grad_norm": 2.9725966453552246, + "learning_rate": 2.986623616236163e-06, + "loss": 1.4425, + "step": 1129 + }, + { + "epoch": 1.4063472308649658, + "grad_norm": 2.977820634841919, + "learning_rate": 2.984317343173432e-06, + "loss": 1.4729, + "step": 1130 + }, + { + "epoch": 1.4063472308649658, + "eval_loss": 1.6550770998001099, + "eval_runtime": 47.3909, + "eval_samples_per_second": 21.101, + "eval_steps_per_second": 0.886, + "step": 1130 + }, + { + "epoch": 1.4075917859365277, + "grad_norm": 2.845229148864746, + "learning_rate": 2.9820110701107014e-06, + "loss": 1.4531, + "step": 1131 + }, + { + "epoch": 1.4088363410080897, + "grad_norm": 2.883481979370117, + "learning_rate": 2.979704797047971e-06, + "loss": 1.4103, + "step": 1132 + }, + { + "epoch": 1.4100808960796516, + "grad_norm": 3.0005011558532715, + "learning_rate": 2.97739852398524e-06, + "loss": 1.4381, + "step": 1133 + }, + { + "epoch": 1.4113254511512134, + "grad_norm": 2.88545560836792, + "learning_rate": 2.9750922509225093e-06, + "loss": 1.4285, + "step": 1134 + }, + { + "epoch": 1.4125700062227753, + "grad_norm": 2.893871307373047, + "learning_rate": 2.972785977859779e-06, + "loss": 1.4578, + "step": 1135 + }, + { + "epoch": 1.4138145612943374, + "grad_norm": 3.0424110889434814, + "learning_rate": 2.970479704797048e-06, + "loss": 1.474, + "step": 1136 + }, + { + "epoch": 1.4150591163658992, + "grad_norm": 2.8640952110290527, + "learning_rate": 2.9681734317343175e-06, + "loss": 1.4857, + "step": 1137 + }, + { + "epoch": 1.416303671437461, + "grad_norm": 3.0472874641418457, + "learning_rate": 2.965867158671587e-06, + "loss": 1.4251, + "step": 1138 + }, + { + "epoch": 1.4175482265090231, + "grad_norm": 2.9340946674346924, + "learning_rate": 2.9635608856088564e-06, + "loss": 1.4469, + "step": 1139 + }, + { + "epoch": 1.418792781580585, + "grad_norm": 2.9423937797546387, + "learning_rate": 2.9612546125461254e-06, + "loss": 1.4835, + "step": 1140 + }, + { + "epoch": 1.418792781580585, + "eval_loss": 1.647971749305725, + "eval_runtime": 44.6901, + "eval_samples_per_second": 22.376, + "eval_steps_per_second": 0.94, + "step": 1140 + }, + { + "epoch": 1.4200373366521468, + "grad_norm": 2.9077932834625244, + "learning_rate": 2.958948339483395e-06, + "loss": 1.4231, + "step": 1141 + }, + { + "epoch": 1.4212818917237087, + "grad_norm": 2.8576467037200928, + "learning_rate": 2.9566420664206646e-06, + "loss": 1.4136, + "step": 1142 + }, + { + "epoch": 1.4225264467952707, + "grad_norm": 2.790850877761841, + "learning_rate": 2.9543357933579336e-06, + "loss": 1.4181, + "step": 1143 + }, + { + "epoch": 1.4237710018668326, + "grad_norm": 2.889728307723999, + "learning_rate": 2.952029520295203e-06, + "loss": 1.4474, + "step": 1144 + }, + { + "epoch": 1.4250155569383947, + "grad_norm": 2.7928731441497803, + "learning_rate": 2.949723247232473e-06, + "loss": 1.4235, + "step": 1145 + }, + { + "epoch": 1.4262601120099565, + "grad_norm": 2.8135132789611816, + "learning_rate": 2.947416974169742e-06, + "loss": 1.4327, + "step": 1146 + }, + { + "epoch": 1.4275046670815184, + "grad_norm": 2.851341485977173, + "learning_rate": 2.9451107011070113e-06, + "loss": 1.4637, + "step": 1147 + }, + { + "epoch": 1.4287492221530802, + "grad_norm": 2.839298725128174, + "learning_rate": 2.9428044280442807e-06, + "loss": 1.4882, + "step": 1148 + }, + { + "epoch": 1.4299937772246423, + "grad_norm": 2.8126413822174072, + "learning_rate": 2.9404981549815497e-06, + "loss": 1.4219, + "step": 1149 + }, + { + "epoch": 1.4312383322962041, + "grad_norm": 2.892277479171753, + "learning_rate": 2.938191881918819e-06, + "loss": 1.4558, + "step": 1150 + }, + { + "epoch": 1.4312383322962041, + "eval_loss": 1.6500873565673828, + "eval_runtime": 45.6914, + "eval_samples_per_second": 21.886, + "eval_steps_per_second": 0.919, + "step": 1150 + }, + { + "epoch": 1.432482887367766, + "grad_norm": 2.9965860843658447, + "learning_rate": 2.935885608856089e-06, + "loss": 1.4712, + "step": 1151 + }, + { + "epoch": 1.433727442439328, + "grad_norm": 2.886841297149658, + "learning_rate": 2.9335793357933584e-06, + "loss": 1.4475, + "step": 1152 + }, + { + "epoch": 1.4349719975108899, + "grad_norm": 2.8532493114471436, + "learning_rate": 2.9312730627306274e-06, + "loss": 1.4436, + "step": 1153 + }, + { + "epoch": 1.4362165525824517, + "grad_norm": 2.873619794845581, + "learning_rate": 2.928966789667897e-06, + "loss": 1.4505, + "step": 1154 + }, + { + "epoch": 1.4374611076540136, + "grad_norm": 2.8811511993408203, + "learning_rate": 2.9266605166051663e-06, + "loss": 1.4196, + "step": 1155 + }, + { + "epoch": 1.4387056627255757, + "grad_norm": 2.8095715045928955, + "learning_rate": 2.9243542435424353e-06, + "loss": 1.46, + "step": 1156 + }, + { + "epoch": 1.4399502177971375, + "grad_norm": 2.8669135570526123, + "learning_rate": 2.922047970479705e-06, + "loss": 1.4121, + "step": 1157 + }, + { + "epoch": 1.4411947728686996, + "grad_norm": 2.9843199253082275, + "learning_rate": 2.9197416974169746e-06, + "loss": 1.4847, + "step": 1158 + }, + { + "epoch": 1.4424393279402614, + "grad_norm": 2.92252516746521, + "learning_rate": 2.9174354243542436e-06, + "loss": 1.4334, + "step": 1159 + }, + { + "epoch": 1.4436838830118233, + "grad_norm": 2.8623733520507812, + "learning_rate": 2.915129151291513e-06, + "loss": 1.4292, + "step": 1160 + }, + { + "epoch": 1.4436838830118233, + "eval_loss": 1.6537351608276367, + "eval_runtime": 45.0702, + "eval_samples_per_second": 22.188, + "eval_steps_per_second": 0.932, + "step": 1160 + }, + { + "epoch": 1.4449284380833851, + "grad_norm": 2.9631102085113525, + "learning_rate": 2.912822878228783e-06, + "loss": 1.4855, + "step": 1161 + }, + { + "epoch": 1.4461729931549472, + "grad_norm": 2.9020369052886963, + "learning_rate": 2.9105166051660522e-06, + "loss": 1.4229, + "step": 1162 + }, + { + "epoch": 1.447417548226509, + "grad_norm": 2.7908735275268555, + "learning_rate": 2.9082103321033212e-06, + "loss": 1.4329, + "step": 1163 + }, + { + "epoch": 1.4486621032980709, + "grad_norm": 2.8533918857574463, + "learning_rate": 2.9059040590405907e-06, + "loss": 1.447, + "step": 1164 + }, + { + "epoch": 1.449906658369633, + "grad_norm": 2.7187347412109375, + "learning_rate": 2.90359778597786e-06, + "loss": 1.4167, + "step": 1165 + }, + { + "epoch": 1.4511512134411948, + "grad_norm": 2.906304359436035, + "learning_rate": 2.901291512915129e-06, + "loss": 1.4229, + "step": 1166 + }, + { + "epoch": 1.4523957685127566, + "grad_norm": 2.8748841285705566, + "learning_rate": 2.898985239852399e-06, + "loss": 1.4472, + "step": 1167 + }, + { + "epoch": 1.4536403235843185, + "grad_norm": 2.925084114074707, + "learning_rate": 2.8966789667896684e-06, + "loss": 1.4257, + "step": 1168 + }, + { + "epoch": 1.4548848786558806, + "grad_norm": 2.7901270389556885, + "learning_rate": 2.8943726937269374e-06, + "loss": 1.4698, + "step": 1169 + }, + { + "epoch": 1.4561294337274424, + "grad_norm": 2.980588436126709, + "learning_rate": 2.8920664206642068e-06, + "loss": 1.4812, + "step": 1170 + }, + { + "epoch": 1.4561294337274424, + "eval_loss": 1.6486413478851318, + "eval_runtime": 51.0519, + "eval_samples_per_second": 19.588, + "eval_steps_per_second": 0.823, + "step": 1170 + }, + { + "epoch": 1.4573739887990045, + "grad_norm": 3.0290639400482178, + "learning_rate": 2.889760147601476e-06, + "loss": 1.4307, + "step": 1171 + }, + { + "epoch": 1.4586185438705663, + "grad_norm": 2.8323185443878174, + "learning_rate": 2.887453874538745e-06, + "loss": 1.412, + "step": 1172 + }, + { + "epoch": 1.4598630989421282, + "grad_norm": 2.880540609359741, + "learning_rate": 2.885147601476015e-06, + "loss": 1.4255, + "step": 1173 + }, + { + "epoch": 1.46110765401369, + "grad_norm": 2.817723512649536, + "learning_rate": 2.8828413284132845e-06, + "loss": 1.426, + "step": 1174 + }, + { + "epoch": 1.462352209085252, + "grad_norm": 2.81062912940979, + "learning_rate": 2.880535055350554e-06, + "loss": 1.4441, + "step": 1175 + }, + { + "epoch": 1.463596764156814, + "grad_norm": 2.9940171241760254, + "learning_rate": 2.878228782287823e-06, + "loss": 1.4418, + "step": 1176 + }, + { + "epoch": 1.4648413192283758, + "grad_norm": 2.8164620399475098, + "learning_rate": 2.8759225092250923e-06, + "loss": 1.4671, + "step": 1177 + }, + { + "epoch": 1.4660858742999379, + "grad_norm": 2.830596446990967, + "learning_rate": 2.873616236162362e-06, + "loss": 1.4699, + "step": 1178 + }, + { + "epoch": 1.4673304293714997, + "grad_norm": 2.7963128089904785, + "learning_rate": 2.871309963099631e-06, + "loss": 1.4217, + "step": 1179 + }, + { + "epoch": 1.4685749844430616, + "grad_norm": 2.8172154426574707, + "learning_rate": 2.8690036900369006e-06, + "loss": 1.4675, + "step": 1180 + }, + { + "epoch": 1.4685749844430616, + "eval_loss": 1.6465628147125244, + "eval_runtime": 45.7709, + "eval_samples_per_second": 21.848, + "eval_steps_per_second": 0.918, + "step": 1180 + }, + { + "epoch": 1.4698195395146234, + "grad_norm": 2.886220693588257, + "learning_rate": 2.86669741697417e-06, + "loss": 1.4212, + "step": 1181 + }, + { + "epoch": 1.4710640945861855, + "grad_norm": 2.925135374069214, + "learning_rate": 2.864391143911439e-06, + "loss": 1.4341, + "step": 1182 + }, + { + "epoch": 1.4723086496577473, + "grad_norm": 3.0038998126983643, + "learning_rate": 2.862084870848709e-06, + "loss": 1.5181, + "step": 1183 + }, + { + "epoch": 1.4735532047293094, + "grad_norm": 2.871967315673828, + "learning_rate": 2.8597785977859783e-06, + "loss": 1.4243, + "step": 1184 + }, + { + "epoch": 1.4747977598008712, + "grad_norm": 2.966698408126831, + "learning_rate": 2.8574723247232473e-06, + "loss": 1.4968, + "step": 1185 + }, + { + "epoch": 1.476042314872433, + "grad_norm": 2.8052289485931396, + "learning_rate": 2.8551660516605167e-06, + "loss": 1.3491, + "step": 1186 + }, + { + "epoch": 1.477286869943995, + "grad_norm": 2.932821035385132, + "learning_rate": 2.852859778597786e-06, + "loss": 1.437, + "step": 1187 + }, + { + "epoch": 1.478531425015557, + "grad_norm": 2.780231475830078, + "learning_rate": 2.850553505535056e-06, + "loss": 1.3966, + "step": 1188 + }, + { + "epoch": 1.4797759800871189, + "grad_norm": 2.8993380069732666, + "learning_rate": 2.848247232472325e-06, + "loss": 1.4059, + "step": 1189 + }, + { + "epoch": 1.4810205351586807, + "grad_norm": 3.020855665206909, + "learning_rate": 2.8459409594095944e-06, + "loss": 1.4047, + "step": 1190 + }, + { + "epoch": 1.4810205351586807, + "eval_loss": 1.649813175201416, + "eval_runtime": 48.4672, + "eval_samples_per_second": 20.633, + "eval_steps_per_second": 0.867, + "step": 1190 + }, + { + "epoch": 1.4822650902302428, + "grad_norm": 2.9288201332092285, + "learning_rate": 2.843634686346864e-06, + "loss": 1.4481, + "step": 1191 + }, + { + "epoch": 1.4835096453018046, + "grad_norm": 2.8591952323913574, + "learning_rate": 2.841328413284133e-06, + "loss": 1.4509, + "step": 1192 + }, + { + "epoch": 1.4847542003733665, + "grad_norm": 2.869105339050293, + "learning_rate": 2.8390221402214022e-06, + "loss": 1.4371, + "step": 1193 + }, + { + "epoch": 1.4859987554449283, + "grad_norm": 2.9339916706085205, + "learning_rate": 2.836715867158672e-06, + "loss": 1.4562, + "step": 1194 + }, + { + "epoch": 1.4872433105164904, + "grad_norm": 3.259225606918335, + "learning_rate": 2.834409594095941e-06, + "loss": 1.4495, + "step": 1195 + }, + { + "epoch": 1.4884878655880522, + "grad_norm": 3.1184468269348145, + "learning_rate": 2.8321033210332105e-06, + "loss": 1.4542, + "step": 1196 + }, + { + "epoch": 1.4897324206596143, + "grad_norm": 2.828624963760376, + "learning_rate": 2.82979704797048e-06, + "loss": 1.4254, + "step": 1197 + }, + { + "epoch": 1.4909769757311762, + "grad_norm": 2.80483078956604, + "learning_rate": 2.827490774907749e-06, + "loss": 1.4673, + "step": 1198 + }, + { + "epoch": 1.492221530802738, + "grad_norm": 2.8064558506011963, + "learning_rate": 2.8251845018450188e-06, + "loss": 1.4469, + "step": 1199 + }, + { + "epoch": 1.4934660858742999, + "grad_norm": 2.9313275814056396, + "learning_rate": 2.822878228782288e-06, + "loss": 1.5039, + "step": 1200 + }, + { + "epoch": 1.4934660858742999, + "eval_loss": 1.6433771848678589, + "eval_runtime": 45.402, + "eval_samples_per_second": 22.025, + "eval_steps_per_second": 0.925, + "step": 1200 + }, + { + "epoch": 1.494710640945862, + "grad_norm": 2.7521519660949707, + "learning_rate": 2.8205719557195576e-06, + "loss": 1.4288, + "step": 1201 + }, + { + "epoch": 1.4959551960174238, + "grad_norm": 2.8780877590179443, + "learning_rate": 2.8182656826568266e-06, + "loss": 1.4388, + "step": 1202 + }, + { + "epoch": 1.4971997510889856, + "grad_norm": 2.8828067779541016, + "learning_rate": 2.815959409594096e-06, + "loss": 1.4212, + "step": 1203 + }, + { + "epoch": 1.4984443061605477, + "grad_norm": 2.959860324859619, + "learning_rate": 2.813653136531366e-06, + "loss": 1.4021, + "step": 1204 + }, + { + "epoch": 1.4996888612321095, + "grad_norm": 3.159862518310547, + "learning_rate": 2.811346863468635e-06, + "loss": 1.4543, + "step": 1205 + }, + { + "epoch": 1.5009334163036714, + "grad_norm": 2.742307424545288, + "learning_rate": 2.8090405904059043e-06, + "loss": 1.4301, + "step": 1206 + }, + { + "epoch": 1.5021779713752332, + "grad_norm": 2.9725406169891357, + "learning_rate": 2.8067343173431737e-06, + "loss": 1.412, + "step": 1207 + }, + { + "epoch": 1.5034225264467953, + "grad_norm": 2.866685152053833, + "learning_rate": 2.8044280442804427e-06, + "loss": 1.4234, + "step": 1208 + }, + { + "epoch": 1.5046670815183572, + "grad_norm": 2.8023900985717773, + "learning_rate": 2.802121771217712e-06, + "loss": 1.4927, + "step": 1209 + }, + { + "epoch": 1.5059116365899192, + "grad_norm": 2.830970048904419, + "learning_rate": 2.799815498154982e-06, + "loss": 1.4373, + "step": 1210 + }, + { + "epoch": 1.5059116365899192, + "eval_loss": 1.651065468788147, + "eval_runtime": 45.9527, + "eval_samples_per_second": 21.761, + "eval_steps_per_second": 0.914, + "step": 1210 + }, + { + "epoch": 1.507156191661481, + "grad_norm": 2.934990167617798, + "learning_rate": 2.7975092250922514e-06, + "loss": 1.485, + "step": 1211 + }, + { + "epoch": 1.508400746733043, + "grad_norm": 2.8635311126708984, + "learning_rate": 2.7952029520295204e-06, + "loss": 1.4389, + "step": 1212 + }, + { + "epoch": 1.5096453018046048, + "grad_norm": 2.868851661682129, + "learning_rate": 2.79289667896679e-06, + "loss": 1.4481, + "step": 1213 + }, + { + "epoch": 1.5108898568761666, + "grad_norm": 2.9756863117218018, + "learning_rate": 2.7905904059040597e-06, + "loss": 1.4425, + "step": 1214 + }, + { + "epoch": 1.5121344119477287, + "grad_norm": 2.8925342559814453, + "learning_rate": 2.7882841328413287e-06, + "loss": 1.4497, + "step": 1215 + }, + { + "epoch": 1.5133789670192908, + "grad_norm": 2.886101007461548, + "learning_rate": 2.785977859778598e-06, + "loss": 1.4772, + "step": 1216 + }, + { + "epoch": 1.5146235220908526, + "grad_norm": 2.8732335567474365, + "learning_rate": 2.7836715867158675e-06, + "loss": 1.4351, + "step": 1217 + }, + { + "epoch": 1.5158680771624145, + "grad_norm": 2.938364028930664, + "learning_rate": 2.7813653136531365e-06, + "loss": 1.3822, + "step": 1218 + }, + { + "epoch": 1.5171126322339763, + "grad_norm": 2.8434407711029053, + "learning_rate": 2.779059040590406e-06, + "loss": 1.3868, + "step": 1219 + }, + { + "epoch": 1.5183571873055381, + "grad_norm": 2.9862983226776123, + "learning_rate": 2.776752767527676e-06, + "loss": 1.4333, + "step": 1220 + }, + { + "epoch": 1.5183571873055381, + "eval_loss": 1.6455049514770508, + "eval_runtime": 47.7092, + "eval_samples_per_second": 20.96, + "eval_steps_per_second": 0.88, + "step": 1220 + }, + { + "epoch": 1.5196017423771002, + "grad_norm": 3.222745656967163, + "learning_rate": 2.774446494464945e-06, + "loss": 1.4321, + "step": 1221 + }, + { + "epoch": 1.520846297448662, + "grad_norm": 2.8184573650360107, + "learning_rate": 2.7721402214022142e-06, + "loss": 1.4346, + "step": 1222 + }, + { + "epoch": 1.5220908525202241, + "grad_norm": 2.867112636566162, + "learning_rate": 2.7698339483394837e-06, + "loss": 1.4359, + "step": 1223 + }, + { + "epoch": 1.523335407591786, + "grad_norm": 3.063075304031372, + "learning_rate": 2.767527675276753e-06, + "loss": 1.3919, + "step": 1224 + }, + { + "epoch": 1.5245799626633478, + "grad_norm": 2.830925941467285, + "learning_rate": 2.765221402214022e-06, + "loss": 1.4133, + "step": 1225 + }, + { + "epoch": 1.5258245177349097, + "grad_norm": 2.937262535095215, + "learning_rate": 2.762915129151292e-06, + "loss": 1.4079, + "step": 1226 + }, + { + "epoch": 1.5270690728064715, + "grad_norm": 2.9446723461151123, + "learning_rate": 2.7606088560885613e-06, + "loss": 1.4169, + "step": 1227 + }, + { + "epoch": 1.5283136278780336, + "grad_norm": 2.8828818798065186, + "learning_rate": 2.7583025830258303e-06, + "loss": 1.3707, + "step": 1228 + }, + { + "epoch": 1.5295581829495957, + "grad_norm": 2.8671786785125732, + "learning_rate": 2.7559963099630998e-06, + "loss": 1.4999, + "step": 1229 + }, + { + "epoch": 1.5308027380211575, + "grad_norm": 3.20247745513916, + "learning_rate": 2.753690036900369e-06, + "loss": 1.3893, + "step": 1230 + }, + { + "epoch": 1.5308027380211575, + "eval_loss": 1.6418426036834717, + "eval_runtime": 50.6885, + "eval_samples_per_second": 19.728, + "eval_steps_per_second": 0.829, + "step": 1230 + }, + { + "epoch": 1.5320472930927194, + "grad_norm": 2.7745611667633057, + "learning_rate": 2.751383763837638e-06, + "loss": 1.3804, + "step": 1231 + }, + { + "epoch": 1.5332918481642812, + "grad_norm": 2.93796706199646, + "learning_rate": 2.749077490774908e-06, + "loss": 1.4082, + "step": 1232 + }, + { + "epoch": 1.534536403235843, + "grad_norm": 2.9986002445220947, + "learning_rate": 2.7467712177121775e-06, + "loss": 1.5093, + "step": 1233 + }, + { + "epoch": 1.5357809583074051, + "grad_norm": 2.8558406829833984, + "learning_rate": 2.7444649446494465e-06, + "loss": 1.3856, + "step": 1234 + }, + { + "epoch": 1.537025513378967, + "grad_norm": 2.8558406829833984, + "learning_rate": 2.7444649446494465e-06, + "loss": 1.4818, + "step": 1235 + }, + { + "epoch": 1.538270068450529, + "grad_norm": 2.7725794315338135, + "learning_rate": 2.742158671586716e-06, + "loss": 1.4256, + "step": 1236 + }, + { + "epoch": 1.539514623522091, + "grad_norm": 2.9687700271606445, + "learning_rate": 2.7398523985239857e-06, + "loss": 1.4623, + "step": 1237 + }, + { + "epoch": 1.5407591785936527, + "grad_norm": 2.709195613861084, + "learning_rate": 2.737546125461255e-06, + "loss": 1.4665, + "step": 1238 + }, + { + "epoch": 1.5420037336652146, + "grad_norm": 2.9164133071899414, + "learning_rate": 2.735239852398524e-06, + "loss": 1.4623, + "step": 1239 + }, + { + "epoch": 1.5432482887367764, + "grad_norm": 3.0910398960113525, + "learning_rate": 2.7329335793357936e-06, + "loss": 1.4771, + "step": 1240 + }, + { + "epoch": 1.5432482887367764, + "eval_loss": 1.6488943099975586, + "eval_runtime": 50.98, + "eval_samples_per_second": 19.616, + "eval_steps_per_second": 0.824, + "step": 1240 + }, + { + "epoch": 1.5444928438083385, + "grad_norm": 2.9204413890838623, + "learning_rate": 2.730627306273063e-06, + "loss": 1.4769, + "step": 1241 + }, + { + "epoch": 1.5457373988799006, + "grad_norm": 2.967817783355713, + "learning_rate": 2.728321033210332e-06, + "loss": 1.4374, + "step": 1242 + }, + { + "epoch": 1.5469819539514624, + "grad_norm": 2.891555070877075, + "learning_rate": 2.726014760147602e-06, + "loss": 1.4116, + "step": 1243 + }, + { + "epoch": 1.5482265090230243, + "grad_norm": 2.882413148880005, + "learning_rate": 2.7237084870848713e-06, + "loss": 1.4304, + "step": 1244 + }, + { + "epoch": 1.5494710640945861, + "grad_norm": 2.720031499862671, + "learning_rate": 2.7214022140221403e-06, + "loss": 1.4035, + "step": 1245 + }, + { + "epoch": 1.550715619166148, + "grad_norm": 2.890763282775879, + "learning_rate": 2.7190959409594097e-06, + "loss": 1.4276, + "step": 1246 + }, + { + "epoch": 1.55196017423771, + "grad_norm": 2.702901840209961, + "learning_rate": 2.716789667896679e-06, + "loss": 1.3971, + "step": 1247 + }, + { + "epoch": 1.553204729309272, + "grad_norm": 2.8923733234405518, + "learning_rate": 2.714483394833948e-06, + "loss": 1.4242, + "step": 1248 + }, + { + "epoch": 1.554449284380834, + "grad_norm": 2.909700632095337, + "learning_rate": 2.712177121771218e-06, + "loss": 1.4599, + "step": 1249 + }, + { + "epoch": 1.5556938394523958, + "grad_norm": 2.9248416423797607, + "learning_rate": 2.7098708487084874e-06, + "loss": 1.4367, + "step": 1250 + }, + { + "epoch": 1.5556938394523958, + "eval_loss": 1.6400885581970215, + "eval_runtime": 52.6283, + "eval_samples_per_second": 19.001, + "eval_steps_per_second": 0.798, + "step": 1250 + }, + { + "epoch": 1.5569383945239577, + "grad_norm": 2.9018962383270264, + "learning_rate": 2.707564575645757e-06, + "loss": 1.482, + "step": 1251 + }, + { + "epoch": 1.5581829495955195, + "grad_norm": 2.928920030593872, + "learning_rate": 2.705258302583026e-06, + "loss": 1.4326, + "step": 1252 + }, + { + "epoch": 1.5594275046670814, + "grad_norm": 2.8409273624420166, + "learning_rate": 2.7029520295202956e-06, + "loss": 1.4206, + "step": 1253 + }, + { + "epoch": 1.5606720597386434, + "grad_norm": 2.7638025283813477, + "learning_rate": 2.700645756457565e-06, + "loss": 1.3835, + "step": 1254 + }, + { + "epoch": 1.5619166148102055, + "grad_norm": 3.0424954891204834, + "learning_rate": 2.698339483394834e-06, + "loss": 1.4695, + "step": 1255 + }, + { + "epoch": 1.5631611698817673, + "grad_norm": 3.0798399448394775, + "learning_rate": 2.6960332103321035e-06, + "loss": 1.4698, + "step": 1256 + }, + { + "epoch": 1.5644057249533292, + "grad_norm": 2.9255073070526123, + "learning_rate": 2.693726937269373e-06, + "loss": 1.4153, + "step": 1257 + }, + { + "epoch": 1.565650280024891, + "grad_norm": 3.1047182083129883, + "learning_rate": 2.691420664206642e-06, + "loss": 1.4391, + "step": 1258 + }, + { + "epoch": 1.5668948350964529, + "grad_norm": 2.8893303871154785, + "learning_rate": 2.6891143911439118e-06, + "loss": 1.3855, + "step": 1259 + }, + { + "epoch": 1.568139390168015, + "grad_norm": 2.886103868484497, + "learning_rate": 2.686808118081181e-06, + "loss": 1.4359, + "step": 1260 + }, + { + "epoch": 1.568139390168015, + "eval_loss": 1.640094518661499, + "eval_runtime": 49.7531, + "eval_samples_per_second": 20.099, + "eval_steps_per_second": 0.844, + "step": 1260 + }, + { + "epoch": 1.5693839452395768, + "grad_norm": 3.337623357772827, + "learning_rate": 2.6845018450184506e-06, + "loss": 1.5257, + "step": 1261 + }, + { + "epoch": 1.5706285003111389, + "grad_norm": 3.0222229957580566, + "learning_rate": 2.6821955719557196e-06, + "loss": 1.4221, + "step": 1262 + }, + { + "epoch": 1.5718730553827007, + "grad_norm": 2.967501401901245, + "learning_rate": 2.679889298892989e-06, + "loss": 1.5488, + "step": 1263 + }, + { + "epoch": 1.5731176104542626, + "grad_norm": 3.064924716949463, + "learning_rate": 2.677583025830259e-06, + "loss": 1.4424, + "step": 1264 + }, + { + "epoch": 1.5743621655258244, + "grad_norm": 2.851073741912842, + "learning_rate": 2.675276752767528e-06, + "loss": 1.4266, + "step": 1265 + }, + { + "epoch": 1.5756067205973865, + "grad_norm": 2.922013282775879, + "learning_rate": 2.6729704797047973e-06, + "loss": 1.4069, + "step": 1266 + }, + { + "epoch": 1.5768512756689483, + "grad_norm": 2.7312874794006348, + "learning_rate": 2.6706642066420667e-06, + "loss": 1.3591, + "step": 1267 + }, + { + "epoch": 1.5780958307405104, + "grad_norm": 3.37092924118042, + "learning_rate": 2.6683579335793357e-06, + "loss": 1.4512, + "step": 1268 + }, + { + "epoch": 1.5793403858120723, + "grad_norm": 2.899977684020996, + "learning_rate": 2.6660516605166056e-06, + "loss": 1.4177, + "step": 1269 + }, + { + "epoch": 1.580584940883634, + "grad_norm": 2.962331771850586, + "learning_rate": 2.663745387453875e-06, + "loss": 1.4522, + "step": 1270 + }, + { + "epoch": 1.580584940883634, + "eval_loss": 1.643125057220459, + "eval_runtime": 49.7864, + "eval_samples_per_second": 20.086, + "eval_steps_per_second": 0.844, + "step": 1270 + }, + { + "epoch": 1.581829495955196, + "grad_norm": 3.224000930786133, + "learning_rate": 2.661439114391144e-06, + "loss": 1.4427, + "step": 1271 + }, + { + "epoch": 1.5830740510267578, + "grad_norm": 3.090118169784546, + "learning_rate": 2.6591328413284134e-06, + "loss": 1.4308, + "step": 1272 + }, + { + "epoch": 1.5843186060983199, + "grad_norm": 2.877392530441284, + "learning_rate": 2.656826568265683e-06, + "loss": 1.4277, + "step": 1273 + }, + { + "epoch": 1.5855631611698817, + "grad_norm": 3.2313079833984375, + "learning_rate": 2.6545202952029527e-06, + "loss": 1.4342, + "step": 1274 + }, + { + "epoch": 1.5868077162414438, + "grad_norm": 3.0011425018310547, + "learning_rate": 2.6522140221402217e-06, + "loss": 1.4545, + "step": 1275 + }, + { + "epoch": 1.5880522713130056, + "grad_norm": 2.802729845046997, + "learning_rate": 2.649907749077491e-06, + "loss": 1.4286, + "step": 1276 + }, + { + "epoch": 1.5892968263845675, + "grad_norm": 3.168306350708008, + "learning_rate": 2.6476014760147605e-06, + "loss": 1.4726, + "step": 1277 + }, + { + "epoch": 1.5905413814561293, + "grad_norm": 2.8712313175201416, + "learning_rate": 2.6452952029520295e-06, + "loss": 1.424, + "step": 1278 + }, + { + "epoch": 1.5917859365276914, + "grad_norm": 3.0059421062469482, + "learning_rate": 2.642988929889299e-06, + "loss": 1.4461, + "step": 1279 + }, + { + "epoch": 1.5930304915992533, + "grad_norm": 2.808736562728882, + "learning_rate": 2.640682656826569e-06, + "loss": 1.4507, + "step": 1280 + }, + { + "epoch": 1.5930304915992533, + "eval_loss": 1.6343430280685425, + "eval_runtime": 52.0843, + "eval_samples_per_second": 19.2, + "eval_steps_per_second": 0.806, + "step": 1280 + }, + { + "epoch": 1.5942750466708153, + "grad_norm": 2.65305757522583, + "learning_rate": 2.638376383763838e-06, + "loss": 1.3912, + "step": 1281 + }, + { + "epoch": 1.5955196017423772, + "grad_norm": 2.9121081829071045, + "learning_rate": 2.6360701107011072e-06, + "loss": 1.3855, + "step": 1282 + }, + { + "epoch": 1.596764156813939, + "grad_norm": 3.035590410232544, + "learning_rate": 2.6337638376383766e-06, + "loss": 1.4341, + "step": 1283 + }, + { + "epoch": 1.5980087118855009, + "grad_norm": 2.829501152038574, + "learning_rate": 2.6314575645756456e-06, + "loss": 1.4061, + "step": 1284 + }, + { + "epoch": 1.5992532669570627, + "grad_norm": 2.9735665321350098, + "learning_rate": 2.629151291512915e-06, + "loss": 1.3729, + "step": 1285 + }, + { + "epoch": 1.6004978220286248, + "grad_norm": 3.0455307960510254, + "learning_rate": 2.626845018450185e-06, + "loss": 1.4475, + "step": 1286 + }, + { + "epoch": 1.6017423771001866, + "grad_norm": 3.0120978355407715, + "learning_rate": 2.6245387453874543e-06, + "loss": 1.4747, + "step": 1287 + }, + { + "epoch": 1.6029869321717487, + "grad_norm": 2.8382294178009033, + "learning_rate": 2.6222324723247233e-06, + "loss": 1.4107, + "step": 1288 + }, + { + "epoch": 1.6042314872433105, + "grad_norm": 2.9768004417419434, + "learning_rate": 2.6199261992619928e-06, + "loss": 1.5052, + "step": 1289 + }, + { + "epoch": 1.6054760423148724, + "grad_norm": 2.9081623554229736, + "learning_rate": 2.6176199261992626e-06, + "loss": 1.4094, + "step": 1290 + }, + { + "epoch": 1.6054760423148724, + "eval_loss": 1.6426081657409668, + "eval_runtime": 49.497, + "eval_samples_per_second": 20.203, + "eval_steps_per_second": 0.849, + "step": 1290 + }, + { + "epoch": 1.6067205973864342, + "grad_norm": 2.9973902702331543, + "learning_rate": 2.6153136531365316e-06, + "loss": 1.4339, + "step": 1291 + }, + { + "epoch": 1.6079651524579963, + "grad_norm": 2.91597843170166, + "learning_rate": 2.613007380073801e-06, + "loss": 1.4457, + "step": 1292 + }, + { + "epoch": 1.6092097075295582, + "grad_norm": 2.8896684646606445, + "learning_rate": 2.6107011070110704e-06, + "loss": 1.4149, + "step": 1293 + }, + { + "epoch": 1.6104542626011202, + "grad_norm": 2.8653416633605957, + "learning_rate": 2.6083948339483394e-06, + "loss": 1.4667, + "step": 1294 + }, + { + "epoch": 1.611698817672682, + "grad_norm": 3.1000678539276123, + "learning_rate": 2.606088560885609e-06, + "loss": 1.4776, + "step": 1295 + }, + { + "epoch": 1.612943372744244, + "grad_norm": 2.879951000213623, + "learning_rate": 2.6037822878228787e-06, + "loss": 1.4861, + "step": 1296 + }, + { + "epoch": 1.6141879278158058, + "grad_norm": 2.852196455001831, + "learning_rate": 2.6014760147601477e-06, + "loss": 1.4013, + "step": 1297 + }, + { + "epoch": 1.6154324828873676, + "grad_norm": 2.9815022945404053, + "learning_rate": 2.599169741697417e-06, + "loss": 1.5116, + "step": 1298 + }, + { + "epoch": 1.6166770379589297, + "grad_norm": 2.8925321102142334, + "learning_rate": 2.5968634686346866e-06, + "loss": 1.4761, + "step": 1299 + }, + { + "epoch": 1.6179215930304915, + "grad_norm": 2.8847744464874268, + "learning_rate": 2.594557195571956e-06, + "loss": 1.4389, + "step": 1300 + }, + { + "epoch": 1.6179215930304915, + "eval_loss": 1.6421130895614624, + "eval_runtime": 53.2733, + "eval_samples_per_second": 18.771, + "eval_steps_per_second": 0.788, + "step": 1300 + }, + { + "epoch": 1.6191661481020536, + "grad_norm": 2.8704094886779785, + "learning_rate": 2.592250922509225e-06, + "loss": 1.423, + "step": 1301 + }, + { + "epoch": 1.6204107031736155, + "grad_norm": 3.0075790882110596, + "learning_rate": 2.589944649446495e-06, + "loss": 1.4343, + "step": 1302 + }, + { + "epoch": 1.6216552582451773, + "grad_norm": 2.8250370025634766, + "learning_rate": 2.5876383763837643e-06, + "loss": 1.3954, + "step": 1303 + }, + { + "epoch": 1.6228998133167392, + "grad_norm": 2.8975305557250977, + "learning_rate": 2.5853321033210333e-06, + "loss": 1.441, + "step": 1304 + }, + { + "epoch": 1.6241443683883012, + "grad_norm": 3.0084152221679688, + "learning_rate": 2.5830258302583027e-06, + "loss": 1.4447, + "step": 1305 + }, + { + "epoch": 1.625388923459863, + "grad_norm": 2.7868149280548096, + "learning_rate": 2.5807195571955725e-06, + "loss": 1.4517, + "step": 1306 + }, + { + "epoch": 1.6266334785314251, + "grad_norm": 2.963439464569092, + "learning_rate": 2.5784132841328415e-06, + "loss": 1.4578, + "step": 1307 + }, + { + "epoch": 1.627878033602987, + "grad_norm": 3.0648446083068848, + "learning_rate": 2.576107011070111e-06, + "loss": 1.5124, + "step": 1308 + }, + { + "epoch": 1.6291225886745488, + "grad_norm": 2.768556594848633, + "learning_rate": 2.5738007380073804e-06, + "loss": 1.4145, + "step": 1309 + }, + { + "epoch": 1.6303671437461107, + "grad_norm": 2.780244827270508, + "learning_rate": 2.5714944649446494e-06, + "loss": 1.4298, + "step": 1310 + }, + { + "epoch": 1.6303671437461107, + "eval_loss": 1.6375432014465332, + "eval_runtime": 43.006, + "eval_samples_per_second": 23.253, + "eval_steps_per_second": 0.977, + "step": 1310 + }, + { + "epoch": 1.6316116988176725, + "grad_norm": 2.7755584716796875, + "learning_rate": 2.5691881918819188e-06, + "loss": 1.44, + "step": 1311 + }, + { + "epoch": 1.6328562538892346, + "grad_norm": 2.8583791255950928, + "learning_rate": 2.5668819188191886e-06, + "loss": 1.4375, + "step": 1312 + }, + { + "epoch": 1.6341008089607967, + "grad_norm": 2.8298938274383545, + "learning_rate": 2.564575645756458e-06, + "loss": 1.4171, + "step": 1313 + }, + { + "epoch": 1.6353453640323585, + "grad_norm": 2.79626727104187, + "learning_rate": 2.562269372693727e-06, + "loss": 1.3764, + "step": 1314 + }, + { + "epoch": 1.6365899191039204, + "grad_norm": 2.743941307067871, + "learning_rate": 2.5599630996309965e-06, + "loss": 1.3855, + "step": 1315 + }, + { + "epoch": 1.6378344741754822, + "grad_norm": 2.950843572616577, + "learning_rate": 2.557656826568266e-06, + "loss": 1.4521, + "step": 1316 + }, + { + "epoch": 1.639079029247044, + "grad_norm": 3.0037457942962646, + "learning_rate": 2.555350553505535e-06, + "loss": 1.4246, + "step": 1317 + }, + { + "epoch": 1.6403235843186061, + "grad_norm": 2.9074928760528564, + "learning_rate": 2.5530442804428047e-06, + "loss": 1.4269, + "step": 1318 + }, + { + "epoch": 1.641568139390168, + "grad_norm": 2.915112257003784, + "learning_rate": 2.550738007380074e-06, + "loss": 1.4262, + "step": 1319 + }, + { + "epoch": 1.64281269446173, + "grad_norm": 2.7829935550689697, + "learning_rate": 2.548431734317343e-06, + "loss": 1.4143, + "step": 1320 + }, + { + "epoch": 1.64281269446173, + "eval_loss": 1.6351630687713623, + "eval_runtime": 45.9445, + "eval_samples_per_second": 21.765, + "eval_steps_per_second": 0.914, + "step": 1320 + }, + { + "epoch": 1.644057249533292, + "grad_norm": 2.8766672611236572, + "learning_rate": 2.5461254612546126e-06, + "loss": 1.3596, + "step": 1321 + }, + { + "epoch": 1.6453018046048538, + "grad_norm": 2.9509501457214355, + "learning_rate": 2.543819188191882e-06, + "loss": 1.3766, + "step": 1322 + }, + { + "epoch": 1.6465463596764156, + "grad_norm": 2.993206262588501, + "learning_rate": 2.541512915129152e-06, + "loss": 1.4226, + "step": 1323 + }, + { + "epoch": 1.6477909147479775, + "grad_norm": 2.872122049331665, + "learning_rate": 2.539206642066421e-06, + "loss": 1.4096, + "step": 1324 + }, + { + "epoch": 1.6490354698195395, + "grad_norm": 3.1725332736968994, + "learning_rate": 2.5369003690036903e-06, + "loss": 1.4616, + "step": 1325 + }, + { + "epoch": 1.6502800248911016, + "grad_norm": 2.920024871826172, + "learning_rate": 2.5345940959409597e-06, + "loss": 1.4581, + "step": 1326 + }, + { + "epoch": 1.6515245799626634, + "grad_norm": 2.9344615936279297, + "learning_rate": 2.5322878228782287e-06, + "loss": 1.4416, + "step": 1327 + }, + { + "epoch": 1.6527691350342253, + "grad_norm": 2.8990018367767334, + "learning_rate": 2.5299815498154986e-06, + "loss": 1.4658, + "step": 1328 + }, + { + "epoch": 1.6540136901057871, + "grad_norm": 2.9203572273254395, + "learning_rate": 2.527675276752768e-06, + "loss": 1.4132, + "step": 1329 + }, + { + "epoch": 1.655258245177349, + "grad_norm": 3.0077338218688965, + "learning_rate": 2.525369003690037e-06, + "loss": 1.4959, + "step": 1330 + }, + { + "epoch": 1.655258245177349, + "eval_loss": 1.6384899616241455, + "eval_runtime": 44.8728, + "eval_samples_per_second": 22.285, + "eval_steps_per_second": 0.936, + "step": 1330 + }, + { + "epoch": 1.656502800248911, + "grad_norm": 2.98171329498291, + "learning_rate": 2.5230627306273064e-06, + "loss": 1.3821, + "step": 1331 + }, + { + "epoch": 1.657747355320473, + "grad_norm": 2.848473310470581, + "learning_rate": 2.520756457564576e-06, + "loss": 1.4157, + "step": 1332 + }, + { + "epoch": 1.658991910392035, + "grad_norm": 2.8811120986938477, + "learning_rate": 2.518450184501845e-06, + "loss": 1.4507, + "step": 1333 + }, + { + "epoch": 1.6602364654635968, + "grad_norm": 2.8421835899353027, + "learning_rate": 2.5161439114391147e-06, + "loss": 1.415, + "step": 1334 + }, + { + "epoch": 1.6614810205351587, + "grad_norm": 2.8382010459899902, + "learning_rate": 2.513837638376384e-06, + "loss": 1.3882, + "step": 1335 + }, + { + "epoch": 1.6627255756067205, + "grad_norm": 3.009395122528076, + "learning_rate": 2.5115313653136535e-06, + "loss": 1.3916, + "step": 1336 + }, + { + "epoch": 1.6639701306782824, + "grad_norm": 2.8583426475524902, + "learning_rate": 2.5092250922509225e-06, + "loss": 1.4196, + "step": 1337 + }, + { + "epoch": 1.6652146857498444, + "grad_norm": 2.8368632793426514, + "learning_rate": 2.506918819188192e-06, + "loss": 1.3973, + "step": 1338 + }, + { + "epoch": 1.6664592408214065, + "grad_norm": 2.9369115829467773, + "learning_rate": 2.5046125461254618e-06, + "loss": 1.4193, + "step": 1339 + }, + { + "epoch": 1.6677037958929684, + "grad_norm": 2.870661973953247, + "learning_rate": 2.5023062730627308e-06, + "loss": 1.4212, + "step": 1340 + }, + { + "epoch": 1.6677037958929684, + "eval_loss": 1.6332666873931885, + "eval_runtime": 45.9731, + "eval_samples_per_second": 21.752, + "eval_steps_per_second": 0.914, + "step": 1340 + }, + { + "epoch": 1.6689483509645302, + "grad_norm": 2.908878803253174, + "learning_rate": 2.5e-06, + "loss": 1.4083, + "step": 1341 + }, + { + "epoch": 1.670192906036092, + "grad_norm": 2.833280563354492, + "learning_rate": 2.4976937269372696e-06, + "loss": 1.3911, + "step": 1342 + }, + { + "epoch": 1.671437461107654, + "grad_norm": 2.7213969230651855, + "learning_rate": 2.495387453874539e-06, + "loss": 1.3997, + "step": 1343 + }, + { + "epoch": 1.672682016179216, + "grad_norm": 2.9431257247924805, + "learning_rate": 2.4930811808118085e-06, + "loss": 1.4476, + "step": 1344 + }, + { + "epoch": 1.6739265712507778, + "grad_norm": 2.8822503089904785, + "learning_rate": 2.490774907749078e-06, + "loss": 1.468, + "step": 1345 + }, + { + "epoch": 1.6751711263223399, + "grad_norm": 2.836256504058838, + "learning_rate": 2.488468634686347e-06, + "loss": 1.4392, + "step": 1346 + }, + { + "epoch": 1.6764156813939017, + "grad_norm": 2.8763248920440674, + "learning_rate": 2.4861623616236163e-06, + "loss": 1.4834, + "step": 1347 + }, + { + "epoch": 1.6776602364654636, + "grad_norm": 2.884368658065796, + "learning_rate": 2.4838560885608857e-06, + "loss": 1.4323, + "step": 1348 + }, + { + "epoch": 1.6789047915370254, + "grad_norm": 2.9247467517852783, + "learning_rate": 2.481549815498155e-06, + "loss": 1.4483, + "step": 1349 + }, + { + "epoch": 1.6801493466085873, + "grad_norm": 3.007827043533325, + "learning_rate": 2.4792435424354246e-06, + "loss": 1.4802, + "step": 1350 + }, + { + "epoch": 1.6801493466085873, + "eval_loss": 1.6364624500274658, + "eval_runtime": 45.8512, + "eval_samples_per_second": 21.81, + "eval_steps_per_second": 0.916, + "step": 1350 + }, + { + "epoch": 1.6813939016801493, + "grad_norm": 2.8283681869506836, + "learning_rate": 2.476937269372694e-06, + "loss": 1.3845, + "step": 1351 + }, + { + "epoch": 1.6826384567517114, + "grad_norm": 3.0330536365509033, + "learning_rate": 2.4746309963099634e-06, + "loss": 1.3922, + "step": 1352 + }, + { + "epoch": 1.6838830118232733, + "grad_norm": 3.1264121532440186, + "learning_rate": 2.472324723247233e-06, + "loss": 1.4498, + "step": 1353 + }, + { + "epoch": 1.6851275668948351, + "grad_norm": 2.756582021713257, + "learning_rate": 2.470018450184502e-06, + "loss": 1.3814, + "step": 1354 + }, + { + "epoch": 1.686372121966397, + "grad_norm": 2.8777999877929688, + "learning_rate": 2.4677121771217713e-06, + "loss": 1.3932, + "step": 1355 + }, + { + "epoch": 1.6876166770379588, + "grad_norm": 2.9220948219299316, + "learning_rate": 2.4654059040590407e-06, + "loss": 1.3972, + "step": 1356 + }, + { + "epoch": 1.6888612321095209, + "grad_norm": 2.818316698074341, + "learning_rate": 2.46309963099631e-06, + "loss": 1.3438, + "step": 1357 + }, + { + "epoch": 1.6901057871810827, + "grad_norm": 2.9452595710754395, + "learning_rate": 2.4607933579335795e-06, + "loss": 1.4417, + "step": 1358 + }, + { + "epoch": 1.6913503422526448, + "grad_norm": 3.1954009532928467, + "learning_rate": 2.458487084870849e-06, + "loss": 1.4563, + "step": 1359 + }, + { + "epoch": 1.6925948973242066, + "grad_norm": 2.8596904277801514, + "learning_rate": 2.4561808118081184e-06, + "loss": 1.4252, + "step": 1360 + }, + { + "epoch": 1.6925948973242066, + "eval_loss": 1.6394789218902588, + "eval_runtime": 49.4542, + "eval_samples_per_second": 20.221, + "eval_steps_per_second": 0.849, + "step": 1360 + }, + { + "epoch": 1.6938394523957685, + "grad_norm": 3.1115806102752686, + "learning_rate": 2.453874538745388e-06, + "loss": 1.4094, + "step": 1361 + }, + { + "epoch": 1.6950840074673303, + "grad_norm": 3.2042553424835205, + "learning_rate": 2.451568265682657e-06, + "loss": 1.412, + "step": 1362 + }, + { + "epoch": 1.6963285625388922, + "grad_norm": 3.1195313930511475, + "learning_rate": 2.4492619926199267e-06, + "loss": 1.4928, + "step": 1363 + }, + { + "epoch": 1.6975731176104543, + "grad_norm": 2.992426633834839, + "learning_rate": 2.4469557195571957e-06, + "loss": 1.5128, + "step": 1364 + }, + { + "epoch": 1.6988176726820163, + "grad_norm": 2.8993239402770996, + "learning_rate": 2.444649446494465e-06, + "loss": 1.4387, + "step": 1365 + }, + { + "epoch": 1.7000622277535782, + "grad_norm": 2.923661708831787, + "learning_rate": 2.4423431734317345e-06, + "loss": 1.4179, + "step": 1366 + }, + { + "epoch": 1.70130678282514, + "grad_norm": 2.822744131088257, + "learning_rate": 2.440036900369004e-06, + "loss": 1.4615, + "step": 1367 + }, + { + "epoch": 1.7025513378967019, + "grad_norm": 2.769036293029785, + "learning_rate": 2.437730627306273e-06, + "loss": 1.4654, + "step": 1368 + }, + { + "epoch": 1.7037958929682637, + "grad_norm": 2.798102855682373, + "learning_rate": 2.4354243542435428e-06, + "loss": 1.351, + "step": 1369 + }, + { + "epoch": 1.7050404480398258, + "grad_norm": 2.8862292766571045, + "learning_rate": 2.4331180811808118e-06, + "loss": 1.4554, + "step": 1370 + }, + { + "epoch": 1.7050404480398258, + "eval_loss": 1.6349775791168213, + "eval_runtime": 46.8676, + "eval_samples_per_second": 21.337, + "eval_steps_per_second": 0.896, + "step": 1370 + }, + { + "epoch": 1.7062850031113876, + "grad_norm": 2.9363768100738525, + "learning_rate": 2.4308118081180816e-06, + "loss": 1.4002, + "step": 1371 + }, + { + "epoch": 1.7075295581829497, + "grad_norm": 2.893996000289917, + "learning_rate": 2.4285055350553506e-06, + "loss": 1.3996, + "step": 1372 + }, + { + "epoch": 1.7087741132545116, + "grad_norm": 2.9740145206451416, + "learning_rate": 2.42619926199262e-06, + "loss": 1.4479, + "step": 1373 + }, + { + "epoch": 1.7100186683260734, + "grad_norm": 2.7912323474884033, + "learning_rate": 2.4238929889298895e-06, + "loss": 1.4292, + "step": 1374 + }, + { + "epoch": 1.7112632233976353, + "grad_norm": 2.780423164367676, + "learning_rate": 2.421586715867159e-06, + "loss": 1.3784, + "step": 1375 + }, + { + "epoch": 1.712507778469197, + "grad_norm": 2.824495792388916, + "learning_rate": 2.4192804428044283e-06, + "loss": 1.4788, + "step": 1376 + }, + { + "epoch": 1.7137523335407592, + "grad_norm": 2.8135807514190674, + "learning_rate": 2.4169741697416977e-06, + "loss": 1.3823, + "step": 1377 + }, + { + "epoch": 1.7149968886123212, + "grad_norm": 2.7912282943725586, + "learning_rate": 2.4146678966789667e-06, + "loss": 1.3853, + "step": 1378 + }, + { + "epoch": 1.716241443683883, + "grad_norm": 2.9185361862182617, + "learning_rate": 2.4123616236162366e-06, + "loss": 1.3905, + "step": 1379 + }, + { + "epoch": 1.717485998755445, + "grad_norm": 2.9129741191864014, + "learning_rate": 2.4100553505535056e-06, + "loss": 1.369, + "step": 1380 + }, + { + "epoch": 1.717485998755445, + "eval_loss": 1.6308677196502686, + "eval_runtime": 46.1789, + "eval_samples_per_second": 21.655, + "eval_steps_per_second": 0.91, + "step": 1380 + }, + { + "epoch": 1.7187305538270068, + "grad_norm": 3.103837013244629, + "learning_rate": 2.407749077490775e-06, + "loss": 1.4413, + "step": 1381 + }, + { + "epoch": 1.7199751088985686, + "grad_norm": 3.030013084411621, + "learning_rate": 2.4054428044280444e-06, + "loss": 1.4523, + "step": 1382 + }, + { + "epoch": 1.7212196639701307, + "grad_norm": 2.939122200012207, + "learning_rate": 2.403136531365314e-06, + "loss": 1.4349, + "step": 1383 + }, + { + "epoch": 1.7224642190416926, + "grad_norm": 2.780301570892334, + "learning_rate": 2.4008302583025833e-06, + "loss": 1.4054, + "step": 1384 + }, + { + "epoch": 1.7237087741132546, + "grad_norm": 2.8379433155059814, + "learning_rate": 2.3985239852398527e-06, + "loss": 1.4338, + "step": 1385 + }, + { + "epoch": 1.7249533291848165, + "grad_norm": 2.77071475982666, + "learning_rate": 2.3962177121771217e-06, + "loss": 1.396, + "step": 1386 + }, + { + "epoch": 1.7261978842563783, + "grad_norm": 2.815927267074585, + "learning_rate": 2.3939114391143915e-06, + "loss": 1.4436, + "step": 1387 + }, + { + "epoch": 1.7274424393279402, + "grad_norm": 3.002138614654541, + "learning_rate": 2.3916051660516605e-06, + "loss": 1.4325, + "step": 1388 + }, + { + "epoch": 1.728686994399502, + "grad_norm": 2.901615619659424, + "learning_rate": 2.3892988929889304e-06, + "loss": 1.3761, + "step": 1389 + }, + { + "epoch": 1.729931549471064, + "grad_norm": 2.9884517192840576, + "learning_rate": 2.3869926199261994e-06, + "loss": 1.4541, + "step": 1390 + }, + { + "epoch": 1.729931549471064, + "eval_loss": 1.634454369544983, + "eval_runtime": 47.1589, + "eval_samples_per_second": 21.205, + "eval_steps_per_second": 0.891, + "step": 1390 + }, + { + "epoch": 1.7311761045426262, + "grad_norm": 2.942561626434326, + "learning_rate": 2.384686346863469e-06, + "loss": 1.4087, + "step": 1391 + }, + { + "epoch": 1.732420659614188, + "grad_norm": 3.0410168170928955, + "learning_rate": 2.3823800738007382e-06, + "loss": 1.4453, + "step": 1392 + }, + { + "epoch": 1.7336652146857499, + "grad_norm": 2.8488197326660156, + "learning_rate": 2.3800738007380077e-06, + "loss": 1.4642, + "step": 1393 + }, + { + "epoch": 1.7349097697573117, + "grad_norm": 2.8784000873565674, + "learning_rate": 2.377767527675277e-06, + "loss": 1.4334, + "step": 1394 + }, + { + "epoch": 1.7361543248288736, + "grad_norm": 2.818960428237915, + "learning_rate": 2.3754612546125465e-06, + "loss": 1.4579, + "step": 1395 + }, + { + "epoch": 1.7373988799004356, + "grad_norm": 2.9459662437438965, + "learning_rate": 2.3731549815498155e-06, + "loss": 1.4524, + "step": 1396 + }, + { + "epoch": 1.7386434349719975, + "grad_norm": 2.829437732696533, + "learning_rate": 2.3708487084870853e-06, + "loss": 1.4765, + "step": 1397 + }, + { + "epoch": 1.7398879900435595, + "grad_norm": 2.8156938552856445, + "learning_rate": 2.3685424354243543e-06, + "loss": 1.3722, + "step": 1398 + }, + { + "epoch": 1.7411325451151214, + "grad_norm": 2.8707408905029297, + "learning_rate": 2.3662361623616238e-06, + "loss": 1.4209, + "step": 1399 + }, + { + "epoch": 1.7423771001866832, + "grad_norm": 3.141056776046753, + "learning_rate": 2.363929889298893e-06, + "loss": 1.4622, + "step": 1400 + }, + { + "epoch": 1.7423771001866832, + "eval_loss": 1.6302117109298706, + "eval_runtime": 45.7354, + "eval_samples_per_second": 21.865, + "eval_steps_per_second": 0.918, + "step": 1400 + }, + { + "epoch": 1.743621655258245, + "grad_norm": 2.8620946407318115, + "learning_rate": 2.3616236162361626e-06, + "loss": 1.4243, + "step": 1401 + }, + { + "epoch": 1.744866210329807, + "grad_norm": 2.927598237991333, + "learning_rate": 2.359317343173432e-06, + "loss": 1.441, + "step": 1402 + }, + { + "epoch": 1.746110765401369, + "grad_norm": 3.0190820693969727, + "learning_rate": 2.3570110701107015e-06, + "loss": 1.407, + "step": 1403 + }, + { + "epoch": 1.747355320472931, + "grad_norm": 2.891575574874878, + "learning_rate": 2.3547047970479705e-06, + "loss": 1.4215, + "step": 1404 + }, + { + "epoch": 1.748599875544493, + "grad_norm": 2.9068562984466553, + "learning_rate": 2.3523985239852403e-06, + "loss": 1.4028, + "step": 1405 + }, + { + "epoch": 1.7498444306160548, + "grad_norm": 2.8855676651000977, + "learning_rate": 2.3500922509225093e-06, + "loss": 1.4561, + "step": 1406 + }, + { + "epoch": 1.7510889856876166, + "grad_norm": 2.9122142791748047, + "learning_rate": 2.3477859778597787e-06, + "loss": 1.3914, + "step": 1407 + }, + { + "epoch": 1.7523335407591785, + "grad_norm": 2.9551825523376465, + "learning_rate": 2.345479704797048e-06, + "loss": 1.4017, + "step": 1408 + }, + { + "epoch": 1.7535780958307405, + "grad_norm": 2.846301555633545, + "learning_rate": 2.3431734317343176e-06, + "loss": 1.3984, + "step": 1409 + }, + { + "epoch": 1.7548226509023024, + "grad_norm": 3.2388343811035156, + "learning_rate": 2.340867158671587e-06, + "loss": 1.4908, + "step": 1410 + }, + { + "epoch": 1.7548226509023024, + "eval_loss": 1.6310688257217407, + "eval_runtime": 43.6166, + "eval_samples_per_second": 22.927, + "eval_steps_per_second": 0.963, + "step": 1410 + }, + { + "epoch": 1.7560672059738645, + "grad_norm": 2.8560962677001953, + "learning_rate": 2.3385608856088564e-06, + "loss": 1.4064, + "step": 1411 + }, + { + "epoch": 1.7573117610454263, + "grad_norm": 2.8909051418304443, + "learning_rate": 2.336254612546126e-06, + "loss": 1.4395, + "step": 1412 + }, + { + "epoch": 1.7585563161169881, + "grad_norm": 2.973435401916504, + "learning_rate": 2.333948339483395e-06, + "loss": 1.424, + "step": 1413 + }, + { + "epoch": 1.75980087118855, + "grad_norm": 3.0337705612182617, + "learning_rate": 2.3316420664206643e-06, + "loss": 1.4188, + "step": 1414 + }, + { + "epoch": 1.7610454262601118, + "grad_norm": 2.7573161125183105, + "learning_rate": 2.3293357933579337e-06, + "loss": 1.3815, + "step": 1415 + }, + { + "epoch": 1.762289981331674, + "grad_norm": 2.7978484630584717, + "learning_rate": 2.327029520295203e-06, + "loss": 1.4007, + "step": 1416 + }, + { + "epoch": 1.763534536403236, + "grad_norm": 2.9547295570373535, + "learning_rate": 2.3247232472324725e-06, + "loss": 1.3924, + "step": 1417 + }, + { + "epoch": 1.7647790914747978, + "grad_norm": 2.941542863845825, + "learning_rate": 2.322416974169742e-06, + "loss": 1.4434, + "step": 1418 + }, + { + "epoch": 1.7660236465463597, + "grad_norm": 2.855360984802246, + "learning_rate": 2.3201107011070114e-06, + "loss": 1.3852, + "step": 1419 + }, + { + "epoch": 1.7672682016179215, + "grad_norm": 2.87410044670105, + "learning_rate": 2.317804428044281e-06, + "loss": 1.4014, + "step": 1420 + }, + { + "epoch": 1.7672682016179215, + "eval_loss": 1.6320465803146362, + "eval_runtime": 47.5665, + "eval_samples_per_second": 21.023, + "eval_steps_per_second": 0.883, + "step": 1420 + }, + { + "epoch": 1.7685127566894834, + "grad_norm": 2.84712290763855, + "learning_rate": 2.31549815498155e-06, + "loss": 1.4049, + "step": 1421 + }, + { + "epoch": 1.7697573117610454, + "grad_norm": 2.924795627593994, + "learning_rate": 2.3131918819188192e-06, + "loss": 1.3823, + "step": 1422 + }, + { + "epoch": 1.7710018668326073, + "grad_norm": 2.9930474758148193, + "learning_rate": 2.3108856088560886e-06, + "loss": 1.4226, + "step": 1423 + }, + { + "epoch": 1.7722464219041694, + "grad_norm": 2.975088596343994, + "learning_rate": 2.308579335793358e-06, + "loss": 1.448, + "step": 1424 + }, + { + "epoch": 1.7734909769757312, + "grad_norm": 2.958648443222046, + "learning_rate": 2.3062730627306275e-06, + "loss": 1.421, + "step": 1425 + }, + { + "epoch": 1.774735532047293, + "grad_norm": 3.057056427001953, + "learning_rate": 2.303966789667897e-06, + "loss": 1.4383, + "step": 1426 + }, + { + "epoch": 1.775980087118855, + "grad_norm": 2.8630926609039307, + "learning_rate": 2.3016605166051663e-06, + "loss": 1.3638, + "step": 1427 + }, + { + "epoch": 1.777224642190417, + "grad_norm": 2.9939122200012207, + "learning_rate": 2.2993542435424358e-06, + "loss": 1.363, + "step": 1428 + }, + { + "epoch": 1.7784691972619788, + "grad_norm": 2.8689146041870117, + "learning_rate": 2.2970479704797048e-06, + "loss": 1.4288, + "step": 1429 + }, + { + "epoch": 1.779713752333541, + "grad_norm": 3.0293314456939697, + "learning_rate": 2.294741697416974e-06, + "loss": 1.5169, + "step": 1430 + }, + { + "epoch": 1.779713752333541, + "eval_loss": 1.6306153535842896, + "eval_runtime": 46.5436, + "eval_samples_per_second": 21.485, + "eval_steps_per_second": 0.902, + "step": 1430 + }, + { + "epoch": 1.7809583074051027, + "grad_norm": 2.9598498344421387, + "learning_rate": 2.2924354243542436e-06, + "loss": 1.4094, + "step": 1431 + }, + { + "epoch": 1.7822028624766646, + "grad_norm": 2.7908709049224854, + "learning_rate": 2.290129151291513e-06, + "loss": 1.4045, + "step": 1432 + }, + { + "epoch": 1.7834474175482264, + "grad_norm": 2.9249887466430664, + "learning_rate": 2.2878228782287825e-06, + "loss": 1.4238, + "step": 1433 + }, + { + "epoch": 1.7846919726197883, + "grad_norm": 2.9420599937438965, + "learning_rate": 2.285516605166052e-06, + "loss": 1.3984, + "step": 1434 + }, + { + "epoch": 1.7859365276913504, + "grad_norm": 2.89164137840271, + "learning_rate": 2.2832103321033213e-06, + "loss": 1.4121, + "step": 1435 + }, + { + "epoch": 1.7871810827629122, + "grad_norm": 3.031681537628174, + "learning_rate": 2.2809040590405907e-06, + "loss": 1.4333, + "step": 1436 + }, + { + "epoch": 1.7884256378344743, + "grad_norm": 2.9249119758605957, + "learning_rate": 2.2785977859778597e-06, + "loss": 1.4223, + "step": 1437 + }, + { + "epoch": 1.7896701929060361, + "grad_norm": 2.9735820293426514, + "learning_rate": 2.2762915129151296e-06, + "loss": 1.4092, + "step": 1438 + }, + { + "epoch": 1.790914747977598, + "grad_norm": 2.992908000946045, + "learning_rate": 2.2739852398523986e-06, + "loss": 1.4088, + "step": 1439 + }, + { + "epoch": 1.7921593030491598, + "grad_norm": 2.9289679527282715, + "learning_rate": 2.271678966789668e-06, + "loss": 1.455, + "step": 1440 + }, + { + "epoch": 1.7921593030491598, + "eval_loss": 1.6259137392044067, + "eval_runtime": 48.3913, + "eval_samples_per_second": 20.665, + "eval_steps_per_second": 0.868, + "step": 1440 + }, + { + "epoch": 1.793403858120722, + "grad_norm": 2.9257709980010986, + "learning_rate": 2.2693726937269374e-06, + "loss": 1.3674, + "step": 1441 + }, + { + "epoch": 1.7946484131922837, + "grad_norm": 2.86993408203125, + "learning_rate": 2.267066420664207e-06, + "loss": 1.4333, + "step": 1442 + }, + { + "epoch": 1.7958929682638458, + "grad_norm": 2.9061667919158936, + "learning_rate": 2.2647601476014763e-06, + "loss": 1.439, + "step": 1443 + }, + { + "epoch": 1.7971375233354077, + "grad_norm": 2.8412883281707764, + "learning_rate": 2.2624538745387457e-06, + "loss": 1.3728, + "step": 1444 + }, + { + "epoch": 1.7983820784069695, + "grad_norm": 3.0139172077178955, + "learning_rate": 2.2601476014760147e-06, + "loss": 1.4635, + "step": 1445 + }, + { + "epoch": 1.7996266334785314, + "grad_norm": 3.016864538192749, + "learning_rate": 2.2578413284132845e-06, + "loss": 1.4729, + "step": 1446 + }, + { + "epoch": 1.8008711885500932, + "grad_norm": 2.9747543334960938, + "learning_rate": 2.2555350553505535e-06, + "loss": 1.4484, + "step": 1447 + }, + { + "epoch": 1.8021157436216553, + "grad_norm": 3.039520740509033, + "learning_rate": 2.253228782287823e-06, + "loss": 1.4188, + "step": 1448 + }, + { + "epoch": 1.8033602986932171, + "grad_norm": 3.0893354415893555, + "learning_rate": 2.2509225092250924e-06, + "loss": 1.4346, + "step": 1449 + }, + { + "epoch": 1.8046048537647792, + "grad_norm": 2.9363458156585693, + "learning_rate": 2.248616236162362e-06, + "loss": 1.4305, + "step": 1450 + }, + { + "epoch": 1.8046048537647792, + "eval_loss": 1.6296759843826294, + "eval_runtime": 49.8814, + "eval_samples_per_second": 20.048, + "eval_steps_per_second": 0.842, + "step": 1450 + }, + { + "epoch": 1.805849408836341, + "grad_norm": 2.941295623779297, + "learning_rate": 2.2463099630996312e-06, + "loss": 1.3816, + "step": 1451 + }, + { + "epoch": 1.807093963907903, + "grad_norm": 3.3323709964752197, + "learning_rate": 2.2440036900369006e-06, + "loss": 1.4556, + "step": 1452 + }, + { + "epoch": 1.8083385189794647, + "grad_norm": 2.942725658416748, + "learning_rate": 2.2416974169741696e-06, + "loss": 1.4102, + "step": 1453 + }, + { + "epoch": 1.8095830740510268, + "grad_norm": 3.136852979660034, + "learning_rate": 2.2393911439114395e-06, + "loss": 1.4187, + "step": 1454 + }, + { + "epoch": 1.8108276291225887, + "grad_norm": 2.959941864013672, + "learning_rate": 2.2370848708487085e-06, + "loss": 1.4008, + "step": 1455 + }, + { + "epoch": 1.8120721841941507, + "grad_norm": 3.136758804321289, + "learning_rate": 2.2347785977859783e-06, + "loss": 1.4283, + "step": 1456 + }, + { + "epoch": 1.8133167392657126, + "grad_norm": 2.973801612854004, + "learning_rate": 2.2324723247232473e-06, + "loss": 1.4911, + "step": 1457 + }, + { + "epoch": 1.8145612943372744, + "grad_norm": 2.96920108795166, + "learning_rate": 2.2301660516605168e-06, + "loss": 1.4012, + "step": 1458 + }, + { + "epoch": 1.8158058494088363, + "grad_norm": 3.128460645675659, + "learning_rate": 2.227859778597786e-06, + "loss": 1.4134, + "step": 1459 + }, + { + "epoch": 1.8170504044803981, + "grad_norm": 2.8887293338775635, + "learning_rate": 2.2255535055350556e-06, + "loss": 1.3961, + "step": 1460 + }, + { + "epoch": 1.8170504044803981, + "eval_loss": 1.6299411058425903, + "eval_runtime": 47.4462, + "eval_samples_per_second": 21.077, + "eval_steps_per_second": 0.885, + "step": 1460 + }, + { + "epoch": 1.8182949595519602, + "grad_norm": 3.1253349781036377, + "learning_rate": 2.2232472324723246e-06, + "loss": 1.462, + "step": 1461 + }, + { + "epoch": 1.819539514623522, + "grad_norm": 3.0835256576538086, + "learning_rate": 2.2209409594095944e-06, + "loss": 1.4127, + "step": 1462 + }, + { + "epoch": 1.820784069695084, + "grad_norm": 3.2318263053894043, + "learning_rate": 2.2186346863468634e-06, + "loss": 1.3559, + "step": 1463 + }, + { + "epoch": 1.822028624766646, + "grad_norm": 3.0926361083984375, + "learning_rate": 2.2163284132841333e-06, + "loss": 1.3967, + "step": 1464 + }, + { + "epoch": 1.8232731798382078, + "grad_norm": 2.927354097366333, + "learning_rate": 2.2140221402214023e-06, + "loss": 1.4169, + "step": 1465 + }, + { + "epoch": 1.8245177349097697, + "grad_norm": 3.00901460647583, + "learning_rate": 2.2117158671586717e-06, + "loss": 1.4099, + "step": 1466 + }, + { + "epoch": 1.8257622899813317, + "grad_norm": 3.255394458770752, + "learning_rate": 2.209409594095941e-06, + "loss": 1.4229, + "step": 1467 + }, + { + "epoch": 1.8270068450528936, + "grad_norm": 3.0004467964172363, + "learning_rate": 2.2071033210332106e-06, + "loss": 1.4089, + "step": 1468 + }, + { + "epoch": 1.8282514001244556, + "grad_norm": 2.873079299926758, + "learning_rate": 2.20479704797048e-06, + "loss": 1.4043, + "step": 1469 + }, + { + "epoch": 1.8294959551960175, + "grad_norm": 2.924314498901367, + "learning_rate": 2.2024907749077494e-06, + "loss": 1.389, + "step": 1470 + }, + { + "epoch": 1.8294959551960175, + "eval_loss": 1.6256897449493408, + "eval_runtime": 45.593, + "eval_samples_per_second": 21.933, + "eval_steps_per_second": 0.921, + "step": 1470 + }, + { + "epoch": 1.8307405102675793, + "grad_norm": 2.975888252258301, + "learning_rate": 2.2001845018450184e-06, + "loss": 1.3983, + "step": 1471 + }, + { + "epoch": 1.8319850653391412, + "grad_norm": 2.9005095958709717, + "learning_rate": 2.1978782287822883e-06, + "loss": 1.4247, + "step": 1472 + }, + { + "epoch": 1.833229620410703, + "grad_norm": 2.952568292617798, + "learning_rate": 2.1955719557195573e-06, + "loss": 1.4195, + "step": 1473 + }, + { + "epoch": 1.834474175482265, + "grad_norm": 3.21946120262146, + "learning_rate": 2.1932656826568267e-06, + "loss": 1.4282, + "step": 1474 + }, + { + "epoch": 1.8357187305538272, + "grad_norm": 2.847078800201416, + "learning_rate": 2.190959409594096e-06, + "loss": 1.3733, + "step": 1475 + }, + { + "epoch": 1.836963285625389, + "grad_norm": 2.9820823669433594, + "learning_rate": 2.1886531365313655e-06, + "loss": 1.4494, + "step": 1476 + }, + { + "epoch": 1.8382078406969509, + "grad_norm": 2.9088616371154785, + "learning_rate": 2.186346863468635e-06, + "loss": 1.3263, + "step": 1477 + }, + { + "epoch": 1.8394523957685127, + "grad_norm": 2.9320101737976074, + "learning_rate": 2.1840405904059044e-06, + "loss": 1.3931, + "step": 1478 + }, + { + "epoch": 1.8406969508400746, + "grad_norm": 2.8825459480285645, + "learning_rate": 2.1817343173431734e-06, + "loss": 1.4188, + "step": 1479 + }, + { + "epoch": 1.8419415059116366, + "grad_norm": 2.9527266025543213, + "learning_rate": 2.1794280442804432e-06, + "loss": 1.4245, + "step": 1480 + }, + { + "epoch": 1.8419415059116366, + "eval_loss": 1.623984456062317, + "eval_runtime": 45.3346, + "eval_samples_per_second": 22.058, + "eval_steps_per_second": 0.926, + "step": 1480 + }, + { + "epoch": 1.8431860609831985, + "grad_norm": 2.951578140258789, + "learning_rate": 2.177121771217712e-06, + "loss": 1.3723, + "step": 1481 + }, + { + "epoch": 1.8444306160547606, + "grad_norm": 2.841216564178467, + "learning_rate": 2.1748154981549816e-06, + "loss": 1.3893, + "step": 1482 + }, + { + "epoch": 1.8456751711263224, + "grad_norm": 2.933821201324463, + "learning_rate": 2.172509225092251e-06, + "loss": 1.4196, + "step": 1483 + }, + { + "epoch": 1.8469197261978842, + "grad_norm": 3.2382006645202637, + "learning_rate": 2.1702029520295205e-06, + "loss": 1.3512, + "step": 1484 + }, + { + "epoch": 1.848164281269446, + "grad_norm": 2.972015380859375, + "learning_rate": 2.16789667896679e-06, + "loss": 1.3762, + "step": 1485 + }, + { + "epoch": 1.849408836341008, + "grad_norm": 2.8245530128479004, + "learning_rate": 2.1655904059040593e-06, + "loss": 1.3758, + "step": 1486 + }, + { + "epoch": 1.85065339141257, + "grad_norm": 3.108802556991577, + "learning_rate": 2.1632841328413287e-06, + "loss": 1.4454, + "step": 1487 + }, + { + "epoch": 1.851897946484132, + "grad_norm": 3.196927547454834, + "learning_rate": 2.160977859778598e-06, + "loss": 1.4548, + "step": 1488 + }, + { + "epoch": 1.853142501555694, + "grad_norm": 2.9764368534088135, + "learning_rate": 2.158671586715867e-06, + "loss": 1.4335, + "step": 1489 + }, + { + "epoch": 1.8543870566272558, + "grad_norm": 2.8558847904205322, + "learning_rate": 2.1563653136531366e-06, + "loss": 1.4468, + "step": 1490 + }, + { + "epoch": 1.8543870566272558, + "eval_loss": 1.6222788095474243, + "eval_runtime": 50.0333, + "eval_samples_per_second": 19.987, + "eval_steps_per_second": 0.839, + "step": 1490 + }, + { + "epoch": 1.8556316116988176, + "grad_norm": 2.875429153442383, + "learning_rate": 2.154059040590406e-06, + "loss": 1.4258, + "step": 1491 + }, + { + "epoch": 1.8568761667703795, + "grad_norm": 2.993828773498535, + "learning_rate": 2.1517527675276754e-06, + "loss": 1.3982, + "step": 1492 + }, + { + "epoch": 1.8581207218419415, + "grad_norm": 2.9214375019073486, + "learning_rate": 2.149446494464945e-06, + "loss": 1.4978, + "step": 1493 + }, + { + "epoch": 1.8593652769135034, + "grad_norm": 2.9349868297576904, + "learning_rate": 2.1471402214022143e-06, + "loss": 1.4385, + "step": 1494 + }, + { + "epoch": 1.8606098319850655, + "grad_norm": 2.926352024078369, + "learning_rate": 2.1448339483394837e-06, + "loss": 1.3913, + "step": 1495 + }, + { + "epoch": 1.8618543870566273, + "grad_norm": 2.8497722148895264, + "learning_rate": 2.142527675276753e-06, + "loss": 1.3915, + "step": 1496 + }, + { + "epoch": 1.8630989421281892, + "grad_norm": 2.8016490936279297, + "learning_rate": 2.140221402214022e-06, + "loss": 1.3867, + "step": 1497 + }, + { + "epoch": 1.864343497199751, + "grad_norm": 2.919970989227295, + "learning_rate": 2.1379151291512916e-06, + "loss": 1.3784, + "step": 1498 + }, + { + "epoch": 1.8655880522713129, + "grad_norm": 3.095557689666748, + "learning_rate": 2.135608856088561e-06, + "loss": 1.4297, + "step": 1499 + }, + { + "epoch": 1.866832607342875, + "grad_norm": 3.001741647720337, + "learning_rate": 2.1333025830258304e-06, + "loss": 1.4259, + "step": 1500 + }, + { + "epoch": 1.866832607342875, + "eval_loss": 1.6227651834487915, + "eval_runtime": 48.9527, + "eval_samples_per_second": 20.428, + "eval_steps_per_second": 0.858, + "step": 1500 + }, + { + "epoch": 1.868077162414437, + "grad_norm": 2.906398057937622, + "learning_rate": 2.1309963099631e-06, + "loss": 1.4209, + "step": 1501 + }, + { + "epoch": 1.8693217174859988, + "grad_norm": 3.006981134414673, + "learning_rate": 2.1286900369003692e-06, + "loss": 1.4008, + "step": 1502 + }, + { + "epoch": 1.8705662725575607, + "grad_norm": 2.848041534423828, + "learning_rate": 2.1263837638376387e-06, + "loss": 1.4402, + "step": 1503 + }, + { + "epoch": 1.8718108276291225, + "grad_norm": 2.9804189205169678, + "learning_rate": 2.124077490774908e-06, + "loss": 1.3883, + "step": 1504 + }, + { + "epoch": 1.8730553827006844, + "grad_norm": 2.9567763805389404, + "learning_rate": 2.1217712177121775e-06, + "loss": 1.452, + "step": 1505 + }, + { + "epoch": 1.8742999377722465, + "grad_norm": 2.842329978942871, + "learning_rate": 2.1194649446494465e-06, + "loss": 1.3862, + "step": 1506 + }, + { + "epoch": 1.8755444928438083, + "grad_norm": 2.8646347522735596, + "learning_rate": 2.117158671586716e-06, + "loss": 1.4487, + "step": 1507 + }, + { + "epoch": 1.8767890479153704, + "grad_norm": 2.9858622550964355, + "learning_rate": 2.1148523985239854e-06, + "loss": 1.4656, + "step": 1508 + }, + { + "epoch": 1.8780336029869322, + "grad_norm": 3.1869702339172363, + "learning_rate": 2.1125461254612548e-06, + "loss": 1.4804, + "step": 1509 + }, + { + "epoch": 1.879278158058494, + "grad_norm": 2.987708806991577, + "learning_rate": 2.110239852398524e-06, + "loss": 1.4597, + "step": 1510 + }, + { + "epoch": 1.879278158058494, + "eval_loss": 1.6230274438858032, + "eval_runtime": 43.1679, + "eval_samples_per_second": 23.165, + "eval_steps_per_second": 0.973, + "step": 1510 + }, + { + "epoch": 1.880522713130056, + "grad_norm": 2.9133615493774414, + "learning_rate": 2.1079335793357936e-06, + "loss": 1.438, + "step": 1511 + }, + { + "epoch": 1.8817672682016178, + "grad_norm": 2.992630958557129, + "learning_rate": 2.1056273062730626e-06, + "loss": 1.3445, + "step": 1512 + }, + { + "epoch": 1.8830118232731798, + "grad_norm": 2.8437397480010986, + "learning_rate": 2.1033210332103325e-06, + "loss": 1.4519, + "step": 1513 + }, + { + "epoch": 1.884256378344742, + "grad_norm": 2.7574918270111084, + "learning_rate": 2.1010147601476015e-06, + "loss": 1.398, + "step": 1514 + }, + { + "epoch": 1.8855009334163038, + "grad_norm": 2.9313275814056396, + "learning_rate": 2.098708487084871e-06, + "loss": 1.4079, + "step": 1515 + }, + { + "epoch": 1.8867454884878656, + "grad_norm": 2.9336822032928467, + "learning_rate": 2.0964022140221403e-06, + "loss": 1.4588, + "step": 1516 + }, + { + "epoch": 1.8879900435594275, + "grad_norm": 2.7718148231506348, + "learning_rate": 2.0940959409594097e-06, + "loss": 1.398, + "step": 1517 + }, + { + "epoch": 1.8892345986309893, + "grad_norm": 2.9897115230560303, + "learning_rate": 2.091789667896679e-06, + "loss": 1.4558, + "step": 1518 + }, + { + "epoch": 1.8904791537025514, + "grad_norm": 2.8779306411743164, + "learning_rate": 2.0894833948339486e-06, + "loss": 1.3617, + "step": 1519 + }, + { + "epoch": 1.8917237087741132, + "grad_norm": 3.035209894180298, + "learning_rate": 2.0871771217712176e-06, + "loss": 1.4149, + "step": 1520 + }, + { + "epoch": 1.8917237087741132, + "eval_loss": 1.6238139867782593, + "eval_runtime": 47.5895, + "eval_samples_per_second": 21.013, + "eval_steps_per_second": 0.883, + "step": 1520 + }, + { + "epoch": 1.8929682638456753, + "grad_norm": 2.9098663330078125, + "learning_rate": 2.0848708487084874e-06, + "loss": 1.4138, + "step": 1521 + }, + { + "epoch": 1.8942128189172371, + "grad_norm": 2.8637537956237793, + "learning_rate": 2.0825645756457564e-06, + "loss": 1.3726, + "step": 1522 + }, + { + "epoch": 1.895457373988799, + "grad_norm": 2.981131076812744, + "learning_rate": 2.0802583025830263e-06, + "loss": 1.4156, + "step": 1523 + }, + { + "epoch": 1.8967019290603608, + "grad_norm": 2.9190938472747803, + "learning_rate": 2.0779520295202953e-06, + "loss": 1.3988, + "step": 1524 + }, + { + "epoch": 1.8979464841319227, + "grad_norm": 2.937696695327759, + "learning_rate": 2.0756457564575647e-06, + "loss": 1.3746, + "step": 1525 + }, + { + "epoch": 1.8991910392034848, + "grad_norm": 3.0422637462615967, + "learning_rate": 2.073339483394834e-06, + "loss": 1.4064, + "step": 1526 + }, + { + "epoch": 1.9004355942750468, + "grad_norm": 2.920274496078491, + "learning_rate": 2.0710332103321035e-06, + "loss": 1.416, + "step": 1527 + }, + { + "epoch": 1.9016801493466087, + "grad_norm": 2.917694091796875, + "learning_rate": 2.0687269372693725e-06, + "loss": 1.455, + "step": 1528 + }, + { + "epoch": 1.9029247044181705, + "grad_norm": 2.961634635925293, + "learning_rate": 2.0664206642066424e-06, + "loss": 1.3617, + "step": 1529 + }, + { + "epoch": 1.9041692594897324, + "grad_norm": 3.014674186706543, + "learning_rate": 2.0641143911439114e-06, + "loss": 1.4464, + "step": 1530 + }, + { + "epoch": 1.9041692594897324, + "eval_loss": 1.6272473335266113, + "eval_runtime": 49.0723, + "eval_samples_per_second": 20.378, + "eval_steps_per_second": 0.856, + "step": 1530 + }, + { + "epoch": 1.9054138145612942, + "grad_norm": 3.0149717330932617, + "learning_rate": 2.0618081180811812e-06, + "loss": 1.4478, + "step": 1531 + }, + { + "epoch": 1.9066583696328563, + "grad_norm": 2.9122891426086426, + "learning_rate": 2.0595018450184502e-06, + "loss": 1.4161, + "step": 1532 + }, + { + "epoch": 1.9079029247044181, + "grad_norm": 2.873610496520996, + "learning_rate": 2.0571955719557197e-06, + "loss": 1.4102, + "step": 1533 + }, + { + "epoch": 1.9091474797759802, + "grad_norm": 3.173130512237549, + "learning_rate": 2.054889298892989e-06, + "loss": 1.4761, + "step": 1534 + }, + { + "epoch": 1.910392034847542, + "grad_norm": 3.0038845539093018, + "learning_rate": 2.0525830258302585e-06, + "loss": 1.411, + "step": 1535 + }, + { + "epoch": 1.911636589919104, + "grad_norm": 3.0535807609558105, + "learning_rate": 2.050276752767528e-06, + "loss": 1.3858, + "step": 1536 + }, + { + "epoch": 1.9128811449906657, + "grad_norm": 2.9695370197296143, + "learning_rate": 2.0479704797047974e-06, + "loss": 1.4343, + "step": 1537 + }, + { + "epoch": 1.9141257000622276, + "grad_norm": 2.9697799682617188, + "learning_rate": 2.0456642066420664e-06, + "loss": 1.4218, + "step": 1538 + }, + { + "epoch": 1.9153702551337897, + "grad_norm": 2.903224468231201, + "learning_rate": 2.043357933579336e-06, + "loss": 1.4614, + "step": 1539 + }, + { + "epoch": 1.9166148102053517, + "grad_norm": 2.789318084716797, + "learning_rate": 2.041051660516605e-06, + "loss": 1.3156, + "step": 1540 + }, + { + "epoch": 1.9166148102053517, + "eval_loss": 1.6211179494857788, + "eval_runtime": 47.0639, + "eval_samples_per_second": 21.248, + "eval_steps_per_second": 0.892, + "step": 1540 + }, + { + "epoch": 1.9178593652769136, + "grad_norm": 3.066250801086426, + "learning_rate": 2.0387453874538746e-06, + "loss": 1.4439, + "step": 1541 + }, + { + "epoch": 1.9191039203484754, + "grad_norm": 2.8898353576660156, + "learning_rate": 2.036439114391144e-06, + "loss": 1.3716, + "step": 1542 + }, + { + "epoch": 1.9203484754200373, + "grad_norm": 2.975554943084717, + "learning_rate": 2.0341328413284135e-06, + "loss": 1.4008, + "step": 1543 + }, + { + "epoch": 1.9215930304915991, + "grad_norm": 3.0215322971343994, + "learning_rate": 2.031826568265683e-06, + "loss": 1.3787, + "step": 1544 + }, + { + "epoch": 1.9228375855631612, + "grad_norm": 3.074124574661255, + "learning_rate": 2.0295202952029523e-06, + "loss": 1.4528, + "step": 1545 + }, + { + "epoch": 1.924082140634723, + "grad_norm": 2.8828608989715576, + "learning_rate": 2.0272140221402213e-06, + "loss": 1.3934, + "step": 1546 + }, + { + "epoch": 1.9253266957062851, + "grad_norm": 2.8467960357666016, + "learning_rate": 2.024907749077491e-06, + "loss": 1.414, + "step": 1547 + }, + { + "epoch": 1.926571250777847, + "grad_norm": 2.9690022468566895, + "learning_rate": 2.02260147601476e-06, + "loss": 1.4024, + "step": 1548 + }, + { + "epoch": 1.9278158058494088, + "grad_norm": 2.913491725921631, + "learning_rate": 2.02029520295203e-06, + "loss": 1.4166, + "step": 1549 + }, + { + "epoch": 1.9290603609209707, + "grad_norm": 2.9665236473083496, + "learning_rate": 2.017988929889299e-06, + "loss": 1.3949, + "step": 1550 + }, + { + "epoch": 1.9290603609209707, + "eval_loss": 1.6198350191116333, + "eval_runtime": 48.1578, + "eval_samples_per_second": 20.765, + "eval_steps_per_second": 0.872, + "step": 1550 + }, + { + "epoch": 1.9303049159925325, + "grad_norm": 3.0438289642333984, + "learning_rate": 2.0156826568265684e-06, + "loss": 1.4364, + "step": 1551 + }, + { + "epoch": 1.9315494710640946, + "grad_norm": 2.9459996223449707, + "learning_rate": 2.013376383763838e-06, + "loss": 1.4102, + "step": 1552 + }, + { + "epoch": 1.9327940261356567, + "grad_norm": 2.898162364959717, + "learning_rate": 2.0110701107011073e-06, + "loss": 1.3856, + "step": 1553 + }, + { + "epoch": 1.9340385812072185, + "grad_norm": 3.0183422565460205, + "learning_rate": 2.0087638376383767e-06, + "loss": 1.4351, + "step": 1554 + }, + { + "epoch": 1.9352831362787803, + "grad_norm": 3.0031955242156982, + "learning_rate": 2.006457564575646e-06, + "loss": 1.4383, + "step": 1555 + }, + { + "epoch": 1.9365276913503422, + "grad_norm": 3.0006372928619385, + "learning_rate": 2.004151291512915e-06, + "loss": 1.4132, + "step": 1556 + }, + { + "epoch": 1.937772246421904, + "grad_norm": 2.8013830184936523, + "learning_rate": 2.0018450184501845e-06, + "loss": 1.3738, + "step": 1557 + }, + { + "epoch": 1.9390168014934661, + "grad_norm": 2.8243088722229004, + "learning_rate": 1.999538745387454e-06, + "loss": 1.3414, + "step": 1558 + }, + { + "epoch": 1.940261356565028, + "grad_norm": 2.864657163619995, + "learning_rate": 1.9972324723247234e-06, + "loss": 1.3452, + "step": 1559 + }, + { + "epoch": 1.94150591163659, + "grad_norm": 2.888096570968628, + "learning_rate": 1.994926199261993e-06, + "loss": 1.3919, + "step": 1560 + }, + { + "epoch": 1.94150591163659, + "eval_loss": 1.6243816614151, + "eval_runtime": 48.9206, + "eval_samples_per_second": 20.441, + "eval_steps_per_second": 0.859, + "step": 1560 + }, + { + "epoch": 1.9427504667081519, + "grad_norm": 2.871358871459961, + "learning_rate": 1.9926199261992622e-06, + "loss": 1.4041, + "step": 1561 + }, + { + "epoch": 1.9439950217797137, + "grad_norm": 2.8249785900115967, + "learning_rate": 1.9903136531365317e-06, + "loss": 1.3993, + "step": 1562 + }, + { + "epoch": 1.9452395768512756, + "grad_norm": 2.8441622257232666, + "learning_rate": 1.988007380073801e-06, + "loss": 1.4237, + "step": 1563 + }, + { + "epoch": 1.9464841319228374, + "grad_norm": 2.849748373031616, + "learning_rate": 1.98570110701107e-06, + "loss": 1.4004, + "step": 1564 + }, + { + "epoch": 1.9477286869943995, + "grad_norm": 3.0251729488372803, + "learning_rate": 1.9833948339483395e-06, + "loss": 1.3856, + "step": 1565 + }, + { + "epoch": 1.9489732420659616, + "grad_norm": 3.016796827316284, + "learning_rate": 1.981088560885609e-06, + "loss": 1.4412, + "step": 1566 + }, + { + "epoch": 1.9502177971375234, + "grad_norm": 2.8642990589141846, + "learning_rate": 1.9787822878228783e-06, + "loss": 1.3923, + "step": 1567 + }, + { + "epoch": 1.9514623522090853, + "grad_norm": 3.1414499282836914, + "learning_rate": 1.9764760147601478e-06, + "loss": 1.4632, + "step": 1568 + }, + { + "epoch": 1.952706907280647, + "grad_norm": 3.001513957977295, + "learning_rate": 1.974169741697417e-06, + "loss": 1.4113, + "step": 1569 + }, + { + "epoch": 1.953951462352209, + "grad_norm": 3.0168347358703613, + "learning_rate": 1.9718634686346866e-06, + "loss": 1.4621, + "step": 1570 + }, + { + "epoch": 1.953951462352209, + "eval_loss": 1.6229702234268188, + "eval_runtime": 47.8157, + "eval_samples_per_second": 20.914, + "eval_steps_per_second": 0.878, + "step": 1570 + }, + { + "epoch": 1.955196017423771, + "grad_norm": 2.82951283454895, + "learning_rate": 1.969557195571956e-06, + "loss": 1.4078, + "step": 1571 + }, + { + "epoch": 1.9564405724953329, + "grad_norm": 2.948712110519409, + "learning_rate": 1.9672509225092255e-06, + "loss": 1.3425, + "step": 1572 + }, + { + "epoch": 1.957685127566895, + "grad_norm": 2.8607540130615234, + "learning_rate": 1.9649446494464945e-06, + "loss": 1.3507, + "step": 1573 + }, + { + "epoch": 1.9589296826384568, + "grad_norm": 2.852407455444336, + "learning_rate": 1.962638376383764e-06, + "loss": 1.4256, + "step": 1574 + }, + { + "epoch": 1.9601742377100186, + "grad_norm": 2.7342076301574707, + "learning_rate": 1.9603321033210333e-06, + "loss": 1.334, + "step": 1575 + }, + { + "epoch": 1.9614187927815805, + "grad_norm": 2.900651454925537, + "learning_rate": 1.9580258302583027e-06, + "loss": 1.3887, + "step": 1576 + }, + { + "epoch": 1.9626633478531423, + "grad_norm": 2.8871212005615234, + "learning_rate": 1.955719557195572e-06, + "loss": 1.4563, + "step": 1577 + }, + { + "epoch": 1.9639079029247044, + "grad_norm": 2.9238333702087402, + "learning_rate": 1.9534132841328416e-06, + "loss": 1.4203, + "step": 1578 + }, + { + "epoch": 1.9651524579962665, + "grad_norm": 2.9352502822875977, + "learning_rate": 1.951107011070111e-06, + "loss": 1.4425, + "step": 1579 + }, + { + "epoch": 1.9663970130678283, + "grad_norm": 2.7428557872772217, + "learning_rate": 1.9488007380073804e-06, + "loss": 1.3578, + "step": 1580 + }, + { + "epoch": 1.9663970130678283, + "eval_loss": 1.6171667575836182, + "eval_runtime": 52.2103, + "eval_samples_per_second": 19.153, + "eval_steps_per_second": 0.804, + "step": 1580 + }, + { + "epoch": 1.9676415681393902, + "grad_norm": 2.86552095413208, + "learning_rate": 1.9464944649446494e-06, + "loss": 1.4064, + "step": 1581 + }, + { + "epoch": 1.968886123210952, + "grad_norm": 2.848663330078125, + "learning_rate": 1.944188191881919e-06, + "loss": 1.4168, + "step": 1582 + }, + { + "epoch": 1.9701306782825139, + "grad_norm": 2.948513984680176, + "learning_rate": 1.9418819188191883e-06, + "loss": 1.4197, + "step": 1583 + }, + { + "epoch": 1.971375233354076, + "grad_norm": 2.9728989601135254, + "learning_rate": 1.9395756457564577e-06, + "loss": 1.4375, + "step": 1584 + }, + { + "epoch": 1.9726197884256378, + "grad_norm": 2.849546194076538, + "learning_rate": 1.937269372693727e-06, + "loss": 1.4057, + "step": 1585 + }, + { + "epoch": 1.9738643434971999, + "grad_norm": 2.9513893127441406, + "learning_rate": 1.9349630996309965e-06, + "loss": 1.3849, + "step": 1586 + }, + { + "epoch": 1.9751088985687617, + "grad_norm": 3.021533250808716, + "learning_rate": 1.932656826568266e-06, + "loss": 1.4118, + "step": 1587 + }, + { + "epoch": 1.9763534536403236, + "grad_norm": 2.8262035846710205, + "learning_rate": 1.9303505535055354e-06, + "loss": 1.3507, + "step": 1588 + }, + { + "epoch": 1.9775980087118854, + "grad_norm": 2.976349115371704, + "learning_rate": 1.9280442804428044e-06, + "loss": 1.3975, + "step": 1589 + }, + { + "epoch": 1.9788425637834475, + "grad_norm": 2.988787889480591, + "learning_rate": 1.925738007380074e-06, + "loss": 1.3816, + "step": 1590 + }, + { + "epoch": 1.9788425637834475, + "eval_loss": 1.6183580160140991, + "eval_runtime": 47.1659, + "eval_samples_per_second": 21.202, + "eval_steps_per_second": 0.89, + "step": 1590 + }, + { + "epoch": 1.9800871188550093, + "grad_norm": 2.9827136993408203, + "learning_rate": 1.9234317343173432e-06, + "loss": 1.4502, + "step": 1591 + }, + { + "epoch": 1.9813316739265714, + "grad_norm": 2.9239609241485596, + "learning_rate": 1.9211254612546126e-06, + "loss": 1.4047, + "step": 1592 + }, + { + "epoch": 1.9825762289981332, + "grad_norm": 2.916612386703491, + "learning_rate": 1.918819188191882e-06, + "loss": 1.4188, + "step": 1593 + }, + { + "epoch": 1.983820784069695, + "grad_norm": 2.894780158996582, + "learning_rate": 1.9165129151291515e-06, + "loss": 1.3911, + "step": 1594 + }, + { + "epoch": 1.985065339141257, + "grad_norm": 2.8818533420562744, + "learning_rate": 1.914206642066421e-06, + "loss": 1.4115, + "step": 1595 + }, + { + "epoch": 1.9863098942128188, + "grad_norm": 2.8122730255126953, + "learning_rate": 1.9119003690036903e-06, + "loss": 1.4147, + "step": 1596 + }, + { + "epoch": 1.9875544492843809, + "grad_norm": 3.0528125762939453, + "learning_rate": 1.9095940959409593e-06, + "loss": 1.3903, + "step": 1597 + }, + { + "epoch": 1.9887990043559427, + "grad_norm": 2.944221019744873, + "learning_rate": 1.907287822878229e-06, + "loss": 1.4044, + "step": 1598 + }, + { + "epoch": 1.9900435594275048, + "grad_norm": 2.785045862197876, + "learning_rate": 1.9049815498154984e-06, + "loss": 1.4238, + "step": 1599 + }, + { + "epoch": 1.9912881144990666, + "grad_norm": 2.9685428142547607, + "learning_rate": 1.9026752767527676e-06, + "loss": 1.4523, + "step": 1600 + }, + { + "epoch": 1.9912881144990666, + "eval_loss": 1.6160162687301636, + "eval_runtime": 47.5378, + "eval_samples_per_second": 21.036, + "eval_steps_per_second": 0.884, + "step": 1600 + }, + { + "epoch": 1.9925326695706285, + "grad_norm": 3.050858736038208, + "learning_rate": 1.900369003690037e-06, + "loss": 1.3857, + "step": 1601 + }, + { + "epoch": 1.9937772246421903, + "grad_norm": 2.978673219680786, + "learning_rate": 1.8980627306273065e-06, + "loss": 1.3948, + "step": 1602 + }, + { + "epoch": 1.9950217797137524, + "grad_norm": 2.8683884143829346, + "learning_rate": 1.8957564575645759e-06, + "loss": 1.383, + "step": 1603 + }, + { + "epoch": 1.9962663347853142, + "grad_norm": 2.836437225341797, + "learning_rate": 1.8934501845018453e-06, + "loss": 1.4177, + "step": 1604 + }, + { + "epoch": 1.9975108898568763, + "grad_norm": 2.8872523307800293, + "learning_rate": 1.8911439114391145e-06, + "loss": 1.3652, + "step": 1605 + }, + { + "epoch": 1.9987554449284382, + "grad_norm": 2.9714925289154053, + "learning_rate": 1.888837638376384e-06, + "loss": 1.4065, + "step": 1606 + }, + { + "epoch": 2.0, + "grad_norm": 2.8581159114837646, + "learning_rate": 1.8865313653136534e-06, + "loss": 1.4139, + "step": 1607 + }, + { + "epoch": 2.001244555071562, + "grad_norm": 2.9979677200317383, + "learning_rate": 1.8842250922509226e-06, + "loss": 1.3277, + "step": 1608 + }, + { + "epoch": 2.0024891101431237, + "grad_norm": 3.1065833568573, + "learning_rate": 1.881918819188192e-06, + "loss": 1.4148, + "step": 1609 + }, + { + "epoch": 2.0037336652146855, + "grad_norm": 3.072094678878784, + "learning_rate": 1.8796125461254614e-06, + "loss": 1.359, + "step": 1610 + }, + { + "epoch": 2.0037336652146855, + "eval_loss": 1.623691201210022, + "eval_runtime": 43.8246, + "eval_samples_per_second": 22.818, + "eval_steps_per_second": 0.958, + "step": 1610 + }, + { + "epoch": 2.004978220286248, + "grad_norm": 3.1043107509613037, + "learning_rate": 1.8773062730627308e-06, + "loss": 1.2892, + "step": 1611 + }, + { + "epoch": 2.0062227753578097, + "grad_norm": 3.19337797164917, + "learning_rate": 1.8750000000000003e-06, + "loss": 1.3369, + "step": 1612 + }, + { + "epoch": 2.0074673304293715, + "grad_norm": 3.238434314727783, + "learning_rate": 1.8726937269372695e-06, + "loss": 1.3339, + "step": 1613 + }, + { + "epoch": 2.0087118855009334, + "grad_norm": 3.206210136413574, + "learning_rate": 1.8703874538745389e-06, + "loss": 1.3078, + "step": 1614 + }, + { + "epoch": 2.0099564405724952, + "grad_norm": 3.3336570262908936, + "learning_rate": 1.8680811808118083e-06, + "loss": 1.3766, + "step": 1615 + }, + { + "epoch": 2.011200995644057, + "grad_norm": 3.3670783042907715, + "learning_rate": 1.8657749077490777e-06, + "loss": 1.3779, + "step": 1616 + }, + { + "epoch": 2.0124455507156194, + "grad_norm": 3.2298123836517334, + "learning_rate": 1.863468634686347e-06, + "loss": 1.3725, + "step": 1617 + }, + { + "epoch": 2.013690105787181, + "grad_norm": 3.251560688018799, + "learning_rate": 1.8611623616236164e-06, + "loss": 1.3197, + "step": 1618 + }, + { + "epoch": 2.014934660858743, + "grad_norm": 3.377767562866211, + "learning_rate": 1.8588560885608858e-06, + "loss": 1.3178, + "step": 1619 + }, + { + "epoch": 2.016179215930305, + "grad_norm": 3.141448736190796, + "learning_rate": 1.856549815498155e-06, + "loss": 1.3402, + "step": 1620 + }, + { + "epoch": 2.016179215930305, + "eval_loss": 1.6248869895935059, + "eval_runtime": 52.3566, + "eval_samples_per_second": 19.1, + "eval_steps_per_second": 0.802, + "step": 1620 + }, + { + "epoch": 2.0174237710018668, + "grad_norm": 3.2168662548065186, + "learning_rate": 1.8542435424354244e-06, + "loss": 1.3757, + "step": 1621 + }, + { + "epoch": 2.0186683260734286, + "grad_norm": 3.399155616760254, + "learning_rate": 1.8519372693726939e-06, + "loss": 1.3949, + "step": 1622 + }, + { + "epoch": 2.0199128811449905, + "grad_norm": 3.3192570209503174, + "learning_rate": 1.8496309963099633e-06, + "loss": 1.3109, + "step": 1623 + }, + { + "epoch": 2.0211574362165527, + "grad_norm": 3.4673702716827393, + "learning_rate": 1.8473247232472327e-06, + "loss": 1.4209, + "step": 1624 + }, + { + "epoch": 2.0224019912881146, + "grad_norm": 3.3889009952545166, + "learning_rate": 1.845018450184502e-06, + "loss": 1.3648, + "step": 1625 + }, + { + "epoch": 2.0236465463596764, + "grad_norm": 3.3962509632110596, + "learning_rate": 1.8427121771217713e-06, + "loss": 1.432, + "step": 1626 + }, + { + "epoch": 2.0248911014312383, + "grad_norm": 3.2030723094940186, + "learning_rate": 1.8404059040590408e-06, + "loss": 1.3882, + "step": 1627 + }, + { + "epoch": 2.0261356565028, + "grad_norm": 3.262387990951538, + "learning_rate": 1.83809963099631e-06, + "loss": 1.372, + "step": 1628 + }, + { + "epoch": 2.027380211574362, + "grad_norm": 3.2742042541503906, + "learning_rate": 1.8357933579335796e-06, + "loss": 1.3777, + "step": 1629 + }, + { + "epoch": 2.0286247666459243, + "grad_norm": 3.147533893585205, + "learning_rate": 1.8334870848708488e-06, + "loss": 1.3462, + "step": 1630 + }, + { + "epoch": 2.0286247666459243, + "eval_loss": 1.6222952604293823, + "eval_runtime": 54.0648, + "eval_samples_per_second": 18.496, + "eval_steps_per_second": 0.777, + "step": 1630 + }, + { + "epoch": 2.029869321717486, + "grad_norm": 3.2916760444641113, + "learning_rate": 1.8311808118081182e-06, + "loss": 1.3186, + "step": 1631 + }, + { + "epoch": 2.031113876789048, + "grad_norm": 3.2263071537017822, + "learning_rate": 1.8288745387453877e-06, + "loss": 1.3411, + "step": 1632 + }, + { + "epoch": 2.03235843186061, + "grad_norm": 3.3505237102508545, + "learning_rate": 1.8265682656826569e-06, + "loss": 1.3726, + "step": 1633 + }, + { + "epoch": 2.0336029869321717, + "grad_norm": 3.3842458724975586, + "learning_rate": 1.8242619926199265e-06, + "loss": 1.3931, + "step": 1634 + }, + { + "epoch": 2.0348475420037335, + "grad_norm": 3.261672019958496, + "learning_rate": 1.8219557195571957e-06, + "loss": 1.3413, + "step": 1635 + }, + { + "epoch": 2.0360920970752954, + "grad_norm": 3.310851573944092, + "learning_rate": 1.819649446494465e-06, + "loss": 1.3712, + "step": 1636 + }, + { + "epoch": 2.0373366521468577, + "grad_norm": 3.1508402824401855, + "learning_rate": 1.8173431734317346e-06, + "loss": 1.3332, + "step": 1637 + }, + { + "epoch": 2.0385812072184195, + "grad_norm": 3.197026014328003, + "learning_rate": 1.8150369003690038e-06, + "loss": 1.3787, + "step": 1638 + }, + { + "epoch": 2.0398257622899814, + "grad_norm": 3.2075631618499756, + "learning_rate": 1.812730627306273e-06, + "loss": 1.3838, + "step": 1639 + }, + { + "epoch": 2.041070317361543, + "grad_norm": 3.1499602794647217, + "learning_rate": 1.8104243542435426e-06, + "loss": 1.3945, + "step": 1640 + }, + { + "epoch": 2.041070317361543, + "eval_loss": 1.6184145212173462, + "eval_runtime": 50.1434, + "eval_samples_per_second": 19.943, + "eval_steps_per_second": 0.838, + "step": 1640 + }, + { + "epoch": 2.042314872433105, + "grad_norm": 3.1790080070495605, + "learning_rate": 1.8081180811808118e-06, + "loss": 1.3261, + "step": 1641 + }, + { + "epoch": 2.043559427504667, + "grad_norm": 3.137925148010254, + "learning_rate": 1.8058118081180815e-06, + "loss": 1.3328, + "step": 1642 + }, + { + "epoch": 2.044803982576229, + "grad_norm": 3.2554683685302734, + "learning_rate": 1.8035055350553507e-06, + "loss": 1.3293, + "step": 1643 + }, + { + "epoch": 2.046048537647791, + "grad_norm": 3.3206231594085693, + "learning_rate": 1.8011992619926199e-06, + "loss": 1.3415, + "step": 1644 + }, + { + "epoch": 2.047293092719353, + "grad_norm": 3.390183210372925, + "learning_rate": 1.7988929889298895e-06, + "loss": 1.3855, + "step": 1645 + }, + { + "epoch": 2.0485376477909147, + "grad_norm": 3.5058743953704834, + "learning_rate": 1.7965867158671587e-06, + "loss": 1.3517, + "step": 1646 + }, + { + "epoch": 2.0497822028624766, + "grad_norm": 3.34566593170166, + "learning_rate": 1.7942804428044284e-06, + "loss": 1.343, + "step": 1647 + }, + { + "epoch": 2.0510267579340384, + "grad_norm": 3.3177366256713867, + "learning_rate": 1.7919741697416976e-06, + "loss": 1.3202, + "step": 1648 + }, + { + "epoch": 2.0522713130056003, + "grad_norm": 3.260061025619507, + "learning_rate": 1.7896678966789668e-06, + "loss": 1.39, + "step": 1649 + }, + { + "epoch": 2.0535158680771626, + "grad_norm": 3.246882677078247, + "learning_rate": 1.7873616236162364e-06, + "loss": 1.3591, + "step": 1650 + }, + { + "epoch": 2.0535158680771626, + "eval_loss": 1.6233465671539307, + "eval_runtime": 54.5853, + "eval_samples_per_second": 18.32, + "eval_steps_per_second": 0.769, + "step": 1650 + }, + { + "epoch": 2.0547604231487244, + "grad_norm": 3.2423501014709473, + "learning_rate": 1.7850553505535056e-06, + "loss": 1.3653, + "step": 1651 + }, + { + "epoch": 2.0560049782202863, + "grad_norm": 3.2285611629486084, + "learning_rate": 1.7827490774907748e-06, + "loss": 1.356, + "step": 1652 + }, + { + "epoch": 2.057249533291848, + "grad_norm": 3.3764569759368896, + "learning_rate": 1.7804428044280445e-06, + "loss": 1.3982, + "step": 1653 + }, + { + "epoch": 2.05849408836341, + "grad_norm": 3.390240430831909, + "learning_rate": 1.7781365313653137e-06, + "loss": 1.3606, + "step": 1654 + }, + { + "epoch": 2.059738643434972, + "grad_norm": 3.301675319671631, + "learning_rate": 1.7758302583025833e-06, + "loss": 1.3707, + "step": 1655 + }, + { + "epoch": 2.060983198506534, + "grad_norm": 3.2020740509033203, + "learning_rate": 1.7735239852398525e-06, + "loss": 1.2709, + "step": 1656 + }, + { + "epoch": 2.062227753578096, + "grad_norm": 3.5533366203308105, + "learning_rate": 1.7712177121771217e-06, + "loss": 1.3734, + "step": 1657 + }, + { + "epoch": 2.063472308649658, + "grad_norm": 3.539336681365967, + "learning_rate": 1.7689114391143914e-06, + "loss": 1.39, + "step": 1658 + }, + { + "epoch": 2.0647168637212197, + "grad_norm": 3.355268716812134, + "learning_rate": 1.7666051660516606e-06, + "loss": 1.3949, + "step": 1659 + }, + { + "epoch": 2.0659614187927815, + "grad_norm": 3.1635212898254395, + "learning_rate": 1.7642988929889302e-06, + "loss": 1.3321, + "step": 1660 + }, + { + "epoch": 2.0659614187927815, + "eval_loss": 1.6189814805984497, + "eval_runtime": 50.8958, + "eval_samples_per_second": 19.648, + "eval_steps_per_second": 0.825, + "step": 1660 + }, + { + "epoch": 2.0672059738643433, + "grad_norm": 3.4177732467651367, + "learning_rate": 1.7619926199261994e-06, + "loss": 1.3824, + "step": 1661 + }, + { + "epoch": 2.068450528935905, + "grad_norm": 3.3664519786834717, + "learning_rate": 1.7596863468634687e-06, + "loss": 1.3587, + "step": 1662 + }, + { + "epoch": 2.0696950840074675, + "grad_norm": 3.2475030422210693, + "learning_rate": 1.7573800738007383e-06, + "loss": 1.3446, + "step": 1663 + }, + { + "epoch": 2.0709396390790293, + "grad_norm": 3.354497194290161, + "learning_rate": 1.7550738007380075e-06, + "loss": 1.366, + "step": 1664 + }, + { + "epoch": 2.072184194150591, + "grad_norm": 3.247157573699951, + "learning_rate": 1.752767527675277e-06, + "loss": 1.3831, + "step": 1665 + }, + { + "epoch": 2.073428749222153, + "grad_norm": 3.3339738845825195, + "learning_rate": 1.7504612546125463e-06, + "loss": 1.3362, + "step": 1666 + }, + { + "epoch": 2.074673304293715, + "grad_norm": 3.289401054382324, + "learning_rate": 1.7481549815498156e-06, + "loss": 1.3549, + "step": 1667 + }, + { + "epoch": 2.0759178593652767, + "grad_norm": 3.5802483558654785, + "learning_rate": 1.7458487084870852e-06, + "loss": 1.3826, + "step": 1668 + }, + { + "epoch": 2.077162414436839, + "grad_norm": 3.389045476913452, + "learning_rate": 1.7435424354243544e-06, + "loss": 1.3561, + "step": 1669 + }, + { + "epoch": 2.078406969508401, + "grad_norm": 3.104553461074829, + "learning_rate": 1.7412361623616236e-06, + "loss": 1.313, + "step": 1670 + }, + { + "epoch": 2.078406969508401, + "eval_loss": 1.6231248378753662, + "eval_runtime": 47.7675, + "eval_samples_per_second": 20.935, + "eval_steps_per_second": 0.879, + "step": 1670 + }, + { + "epoch": 2.0796515245799627, + "grad_norm": 3.247082233428955, + "learning_rate": 1.7389298892988932e-06, + "loss": 1.3544, + "step": 1671 + }, + { + "epoch": 2.0808960796515246, + "grad_norm": 3.2196743488311768, + "learning_rate": 1.7366236162361625e-06, + "loss": 1.3013, + "step": 1672 + }, + { + "epoch": 2.0821406347230864, + "grad_norm": 3.4537713527679443, + "learning_rate": 1.7343173431734319e-06, + "loss": 1.3486, + "step": 1673 + }, + { + "epoch": 2.0833851897946483, + "grad_norm": 3.1996660232543945, + "learning_rate": 1.7320110701107013e-06, + "loss": 1.3524, + "step": 1674 + }, + { + "epoch": 2.08462974486621, + "grad_norm": 3.4143941402435303, + "learning_rate": 1.7297047970479705e-06, + "loss": 1.4062, + "step": 1675 + }, + { + "epoch": 2.0858742999377724, + "grad_norm": 3.329657793045044, + "learning_rate": 1.7273985239852401e-06, + "loss": 1.336, + "step": 1676 + }, + { + "epoch": 2.0871188550093343, + "grad_norm": 3.490600347518921, + "learning_rate": 1.7250922509225094e-06, + "loss": 1.3845, + "step": 1677 + }, + { + "epoch": 2.088363410080896, + "grad_norm": 3.2943880558013916, + "learning_rate": 1.7227859778597788e-06, + "loss": 1.3952, + "step": 1678 + }, + { + "epoch": 2.089607965152458, + "grad_norm": 3.3676629066467285, + "learning_rate": 1.7204797047970482e-06, + "loss": 1.3816, + "step": 1679 + }, + { + "epoch": 2.09085252022402, + "grad_norm": 3.3164591789245605, + "learning_rate": 1.7181734317343174e-06, + "loss": 1.4124, + "step": 1680 + }, + { + "epoch": 2.09085252022402, + "eval_loss": 1.6207176446914673, + "eval_runtime": 48.9133, + "eval_samples_per_second": 20.444, + "eval_steps_per_second": 0.859, + "step": 1680 + }, + { + "epoch": 2.0920970752955816, + "grad_norm": 3.3276405334472656, + "learning_rate": 1.7158671586715868e-06, + "loss": 1.4032, + "step": 1681 + }, + { + "epoch": 2.093341630367144, + "grad_norm": 3.2198235988616943, + "learning_rate": 1.7135608856088563e-06, + "loss": 1.3483, + "step": 1682 + }, + { + "epoch": 2.094586185438706, + "grad_norm": 3.1618118286132812, + "learning_rate": 1.7112546125461257e-06, + "loss": 1.3478, + "step": 1683 + }, + { + "epoch": 2.0958307405102676, + "grad_norm": 3.1534817218780518, + "learning_rate": 1.708948339483395e-06, + "loss": 1.3727, + "step": 1684 + }, + { + "epoch": 2.0970752955818295, + "grad_norm": 3.230872631072998, + "learning_rate": 1.7066420664206643e-06, + "loss": 1.3604, + "step": 1685 + }, + { + "epoch": 2.0983198506533913, + "grad_norm": 3.088921308517456, + "learning_rate": 1.7043357933579337e-06, + "loss": 1.3425, + "step": 1686 + }, + { + "epoch": 2.099564405724953, + "grad_norm": 3.3279967308044434, + "learning_rate": 1.7020295202952032e-06, + "loss": 1.3365, + "step": 1687 + }, + { + "epoch": 2.100808960796515, + "grad_norm": 3.342949628829956, + "learning_rate": 1.6997232472324724e-06, + "loss": 1.3641, + "step": 1688 + }, + { + "epoch": 2.1020535158680773, + "grad_norm": 3.5477232933044434, + "learning_rate": 1.6974169741697418e-06, + "loss": 1.376, + "step": 1689 + }, + { + "epoch": 2.103298070939639, + "grad_norm": 3.363581895828247, + "learning_rate": 1.6951107011070112e-06, + "loss": 1.3251, + "step": 1690 + }, + { + "epoch": 2.103298070939639, + "eval_loss": 1.6247185468673706, + "eval_runtime": 45.6567, + "eval_samples_per_second": 21.903, + "eval_steps_per_second": 0.92, + "step": 1690 + }, + { + "epoch": 2.104542626011201, + "grad_norm": 3.526014566421509, + "learning_rate": 1.6928044280442806e-06, + "loss": 1.3873, + "step": 1691 + }, + { + "epoch": 2.105787181082763, + "grad_norm": 3.36954665184021, + "learning_rate": 1.6904981549815499e-06, + "loss": 1.3708, + "step": 1692 + }, + { + "epoch": 2.1070317361543247, + "grad_norm": 3.2767856121063232, + "learning_rate": 1.6881918819188193e-06, + "loss": 1.3022, + "step": 1693 + }, + { + "epoch": 2.1082762912258866, + "grad_norm": 3.5752694606781006, + "learning_rate": 1.6858856088560887e-06, + "loss": 1.3486, + "step": 1694 + }, + { + "epoch": 2.109520846297449, + "grad_norm": 3.247654676437378, + "learning_rate": 1.6835793357933581e-06, + "loss": 1.3567, + "step": 1695 + }, + { + "epoch": 2.1107654013690107, + "grad_norm": 3.328662633895874, + "learning_rate": 1.6812730627306275e-06, + "loss": 1.3694, + "step": 1696 + }, + { + "epoch": 2.1120099564405725, + "grad_norm": 3.2084357738494873, + "learning_rate": 1.6789667896678968e-06, + "loss": 1.3467, + "step": 1697 + }, + { + "epoch": 2.1132545115121344, + "grad_norm": 3.4850761890411377, + "learning_rate": 1.6766605166051662e-06, + "loss": 1.3635, + "step": 1698 + }, + { + "epoch": 2.1144990665836962, + "grad_norm": 3.40693998336792, + "learning_rate": 1.6743542435424356e-06, + "loss": 1.3589, + "step": 1699 + }, + { + "epoch": 2.115743621655258, + "grad_norm": 3.430428981781006, + "learning_rate": 1.6720479704797048e-06, + "loss": 1.4017, + "step": 1700 + }, + { + "epoch": 2.115743621655258, + "eval_loss": 1.623623251914978, + "eval_runtime": 46.6136, + "eval_samples_per_second": 21.453, + "eval_steps_per_second": 0.901, + "step": 1700 + }, + { + "epoch": 2.11698817672682, + "grad_norm": 3.4609930515289307, + "learning_rate": 1.6697416974169742e-06, + "loss": 1.3514, + "step": 1701 + }, + { + "epoch": 2.1182327317983822, + "grad_norm": 3.4114904403686523, + "learning_rate": 1.6674354243542437e-06, + "loss": 1.4017, + "step": 1702 + }, + { + "epoch": 2.119477286869944, + "grad_norm": 3.4334566593170166, + "learning_rate": 1.665129151291513e-06, + "loss": 1.3339, + "step": 1703 + }, + { + "epoch": 2.120721841941506, + "grad_norm": 3.3242125511169434, + "learning_rate": 1.6628228782287825e-06, + "loss": 1.3739, + "step": 1704 + }, + { + "epoch": 2.1219663970130678, + "grad_norm": 3.2655038833618164, + "learning_rate": 1.6605166051660517e-06, + "loss": 1.3473, + "step": 1705 + }, + { + "epoch": 2.1232109520846296, + "grad_norm": 3.2794158458709717, + "learning_rate": 1.6582103321033211e-06, + "loss": 1.3431, + "step": 1706 + }, + { + "epoch": 2.1244555071561915, + "grad_norm": 3.4481866359710693, + "learning_rate": 1.6559040590405906e-06, + "loss": 1.3343, + "step": 1707 + }, + { + "epoch": 2.1257000622277538, + "grad_norm": 3.250974178314209, + "learning_rate": 1.6535977859778598e-06, + "loss": 1.3255, + "step": 1708 + }, + { + "epoch": 2.1269446172993156, + "grad_norm": 3.2407114505767822, + "learning_rate": 1.6512915129151294e-06, + "loss": 1.3587, + "step": 1709 + }, + { + "epoch": 2.1281891723708775, + "grad_norm": 3.4453704357147217, + "learning_rate": 1.6489852398523986e-06, + "loss": 1.3937, + "step": 1710 + }, + { + "epoch": 2.1281891723708775, + "eval_loss": 1.6269443035125732, + "eval_runtime": 42.294, + "eval_samples_per_second": 23.644, + "eval_steps_per_second": 0.993, + "step": 1710 + }, + { + "epoch": 2.1294337274424393, + "grad_norm": 3.4088101387023926, + "learning_rate": 1.6466789667896678e-06, + "loss": 1.389, + "step": 1711 + }, + { + "epoch": 2.130678282514001, + "grad_norm": 3.3215279579162598, + "learning_rate": 1.6443726937269375e-06, + "loss": 1.3685, + "step": 1712 + }, + { + "epoch": 2.131922837585563, + "grad_norm": 3.3293471336364746, + "learning_rate": 1.6420664206642067e-06, + "loss": 1.3714, + "step": 1713 + }, + { + "epoch": 2.133167392657125, + "grad_norm": 3.232369899749756, + "learning_rate": 1.6397601476014763e-06, + "loss": 1.3614, + "step": 1714 + }, + { + "epoch": 2.134411947728687, + "grad_norm": 3.3835854530334473, + "learning_rate": 1.6374538745387455e-06, + "loss": 1.3509, + "step": 1715 + }, + { + "epoch": 2.135656502800249, + "grad_norm": 3.2008566856384277, + "learning_rate": 1.6351476014760147e-06, + "loss": 1.2935, + "step": 1716 + }, + { + "epoch": 2.136901057871811, + "grad_norm": 3.4421873092651367, + "learning_rate": 1.6328413284132844e-06, + "loss": 1.3442, + "step": 1717 + }, + { + "epoch": 2.1381456129433727, + "grad_norm": 3.379856586456299, + "learning_rate": 1.6305350553505536e-06, + "loss": 1.3917, + "step": 1718 + }, + { + "epoch": 2.1393901680149345, + "grad_norm": 3.5044374465942383, + "learning_rate": 1.6282287822878228e-06, + "loss": 1.4052, + "step": 1719 + }, + { + "epoch": 2.1406347230864964, + "grad_norm": 3.3533782958984375, + "learning_rate": 1.6259225092250924e-06, + "loss": 1.3435, + "step": 1720 + }, + { + "epoch": 2.1406347230864964, + "eval_loss": 1.6230002641677856, + "eval_runtime": 45.9961, + "eval_samples_per_second": 21.741, + "eval_steps_per_second": 0.913, + "step": 1720 + }, + { + "epoch": 2.1418792781580587, + "grad_norm": 3.356597661972046, + "learning_rate": 1.6236162361623616e-06, + "loss": 1.3176, + "step": 1721 + }, + { + "epoch": 2.1431238332296205, + "grad_norm": 3.204878807067871, + "learning_rate": 1.6213099630996313e-06, + "loss": 1.3553, + "step": 1722 + }, + { + "epoch": 2.1443683883011824, + "grad_norm": 3.2749500274658203, + "learning_rate": 1.6190036900369005e-06, + "loss": 1.327, + "step": 1723 + }, + { + "epoch": 2.145612943372744, + "grad_norm": 3.3262641429901123, + "learning_rate": 1.6166974169741697e-06, + "loss": 1.3453, + "step": 1724 + }, + { + "epoch": 2.146857498444306, + "grad_norm": 3.351846218109131, + "learning_rate": 1.6143911439114393e-06, + "loss": 1.334, + "step": 1725 + }, + { + "epoch": 2.148102053515868, + "grad_norm": 3.2748796939849854, + "learning_rate": 1.6120848708487085e-06, + "loss": 1.3178, + "step": 1726 + }, + { + "epoch": 2.1493466085874298, + "grad_norm": 3.261385917663574, + "learning_rate": 1.6097785977859782e-06, + "loss": 1.317, + "step": 1727 + }, + { + "epoch": 2.150591163658992, + "grad_norm": 3.4880056381225586, + "learning_rate": 1.6074723247232474e-06, + "loss": 1.377, + "step": 1728 + }, + { + "epoch": 2.151835718730554, + "grad_norm": 3.3214828968048096, + "learning_rate": 1.6051660516605166e-06, + "loss": 1.3757, + "step": 1729 + }, + { + "epoch": 2.1530802738021158, + "grad_norm": 3.3787684440612793, + "learning_rate": 1.6028597785977862e-06, + "loss": 1.3458, + "step": 1730 + }, + { + "epoch": 2.1530802738021158, + "eval_loss": 1.6261134147644043, + "eval_runtime": 45.3233, + "eval_samples_per_second": 22.064, + "eval_steps_per_second": 0.927, + "step": 1730 + }, + { + "epoch": 2.1543248288736776, + "grad_norm": 3.423396587371826, + "learning_rate": 1.6005535055350554e-06, + "loss": 1.3543, + "step": 1731 + }, + { + "epoch": 2.1555693839452394, + "grad_norm": 3.432394504547119, + "learning_rate": 1.5982472324723247e-06, + "loss": 1.3548, + "step": 1732 + }, + { + "epoch": 2.1568139390168013, + "grad_norm": 3.3628666400909424, + "learning_rate": 1.5959409594095943e-06, + "loss": 1.3619, + "step": 1733 + }, + { + "epoch": 2.1580584940883636, + "grad_norm": 3.3048722743988037, + "learning_rate": 1.5936346863468635e-06, + "loss": 1.3565, + "step": 1734 + }, + { + "epoch": 2.1593030491599254, + "grad_norm": 3.21997332572937, + "learning_rate": 1.5913284132841331e-06, + "loss": 1.3353, + "step": 1735 + }, + { + "epoch": 2.1605476042314873, + "grad_norm": 3.272420644760132, + "learning_rate": 1.5890221402214023e-06, + "loss": 1.3155, + "step": 1736 + }, + { + "epoch": 2.161792159303049, + "grad_norm": 3.246156692504883, + "learning_rate": 1.5867158671586716e-06, + "loss": 1.3636, + "step": 1737 + }, + { + "epoch": 2.163036714374611, + "grad_norm": 3.3257296085357666, + "learning_rate": 1.5844095940959412e-06, + "loss": 1.3748, + "step": 1738 + }, + { + "epoch": 2.164281269446173, + "grad_norm": 3.298217535018921, + "learning_rate": 1.5821033210332104e-06, + "loss": 1.3171, + "step": 1739 + }, + { + "epoch": 2.1655258245177347, + "grad_norm": 3.4946815967559814, + "learning_rate": 1.57979704797048e-06, + "loss": 1.3971, + "step": 1740 + }, + { + "epoch": 2.1655258245177347, + "eval_loss": 1.6242249011993408, + "eval_runtime": 46.2522, + "eval_samples_per_second": 21.621, + "eval_steps_per_second": 0.908, + "step": 1740 + }, + { + "epoch": 2.166770379589297, + "grad_norm": 3.1652512550354004, + "learning_rate": 1.5774907749077492e-06, + "loss": 1.3332, + "step": 1741 + }, + { + "epoch": 2.168014934660859, + "grad_norm": 3.292405843734741, + "learning_rate": 1.5751845018450185e-06, + "loss": 1.3105, + "step": 1742 + }, + { + "epoch": 2.1692594897324207, + "grad_norm": 3.3213112354278564, + "learning_rate": 1.572878228782288e-06, + "loss": 1.372, + "step": 1743 + }, + { + "epoch": 2.1705040448039825, + "grad_norm": 3.1998512744903564, + "learning_rate": 1.5705719557195573e-06, + "loss": 1.36, + "step": 1744 + }, + { + "epoch": 2.1717485998755444, + "grad_norm": 3.449044942855835, + "learning_rate": 1.5682656826568267e-06, + "loss": 1.3262, + "step": 1745 + }, + { + "epoch": 2.172993154947106, + "grad_norm": 3.435952663421631, + "learning_rate": 1.5659594095940962e-06, + "loss": 1.3473, + "step": 1746 + }, + { + "epoch": 2.1742377100186685, + "grad_norm": 3.211683511734009, + "learning_rate": 1.5636531365313654e-06, + "loss": 1.3676, + "step": 1747 + }, + { + "epoch": 2.1754822650902303, + "grad_norm": 3.3525032997131348, + "learning_rate": 1.561346863468635e-06, + "loss": 1.3792, + "step": 1748 + }, + { + "epoch": 2.176726820161792, + "grad_norm": 3.3060429096221924, + "learning_rate": 1.5590405904059042e-06, + "loss": 1.3104, + "step": 1749 + }, + { + "epoch": 2.177971375233354, + "grad_norm": 3.1698927879333496, + "learning_rate": 1.5567343173431734e-06, + "loss": 1.3142, + "step": 1750 + }, + { + "epoch": 2.177971375233354, + "eval_loss": 1.6239022016525269, + "eval_runtime": 47.4352, + "eval_samples_per_second": 21.081, + "eval_steps_per_second": 0.885, + "step": 1750 + }, + { + "epoch": 2.179215930304916, + "grad_norm": 3.3830013275146484, + "learning_rate": 1.554428044280443e-06, + "loss": 1.38, + "step": 1751 + }, + { + "epoch": 2.1804604853764777, + "grad_norm": 3.144509792327881, + "learning_rate": 1.5521217712177123e-06, + "loss": 1.3447, + "step": 1752 + }, + { + "epoch": 2.18170504044804, + "grad_norm": 3.2411561012268066, + "learning_rate": 1.5498154981549817e-06, + "loss": 1.3709, + "step": 1753 + }, + { + "epoch": 2.182949595519602, + "grad_norm": 3.19698429107666, + "learning_rate": 1.5475092250922511e-06, + "loss": 1.3068, + "step": 1754 + }, + { + "epoch": 2.1841941505911637, + "grad_norm": 3.301417589187622, + "learning_rate": 1.5452029520295203e-06, + "loss": 1.3219, + "step": 1755 + }, + { + "epoch": 2.1854387056627256, + "grad_norm": 3.3205950260162354, + "learning_rate": 1.5428966789667897e-06, + "loss": 1.333, + "step": 1756 + }, + { + "epoch": 2.1866832607342874, + "grad_norm": 3.3267645835876465, + "learning_rate": 1.5405904059040592e-06, + "loss": 1.3536, + "step": 1757 + }, + { + "epoch": 2.1879278158058493, + "grad_norm": 3.37038254737854, + "learning_rate": 1.5382841328413286e-06, + "loss": 1.3481, + "step": 1758 + }, + { + "epoch": 2.189172370877411, + "grad_norm": 3.5459911823272705, + "learning_rate": 1.535977859778598e-06, + "loss": 1.3798, + "step": 1759 + }, + { + "epoch": 2.1904169259489734, + "grad_norm": 3.40732741355896, + "learning_rate": 1.5336715867158672e-06, + "loss": 1.3672, + "step": 1760 + }, + { + "epoch": 2.1904169259489734, + "eval_loss": 1.6204931735992432, + "eval_runtime": 45.5831, + "eval_samples_per_second": 21.938, + "eval_steps_per_second": 0.921, + "step": 1760 + }, + { + "epoch": 2.1916614810205353, + "grad_norm": 3.218372106552124, + "learning_rate": 1.5313653136531366e-06, + "loss": 1.3486, + "step": 1761 + }, + { + "epoch": 2.192906036092097, + "grad_norm": 3.370919942855835, + "learning_rate": 1.529059040590406e-06, + "loss": 1.3927, + "step": 1762 + }, + { + "epoch": 2.194150591163659, + "grad_norm": 3.3146965503692627, + "learning_rate": 1.5267527675276755e-06, + "loss": 1.3539, + "step": 1763 + }, + { + "epoch": 2.195395146235221, + "grad_norm": 3.410720109939575, + "learning_rate": 1.5244464944649447e-06, + "loss": 1.3744, + "step": 1764 + }, + { + "epoch": 2.1966397013067827, + "grad_norm": 3.2189249992370605, + "learning_rate": 1.5221402214022141e-06, + "loss": 1.36, + "step": 1765 + }, + { + "epoch": 2.197884256378345, + "grad_norm": 3.149064064025879, + "learning_rate": 1.5198339483394836e-06, + "loss": 1.2923, + "step": 1766 + }, + { + "epoch": 2.199128811449907, + "grad_norm": 3.1745471954345703, + "learning_rate": 1.517527675276753e-06, + "loss": 1.3394, + "step": 1767 + }, + { + "epoch": 2.2003733665214686, + "grad_norm": 3.3207221031188965, + "learning_rate": 1.5152214022140222e-06, + "loss": 1.3309, + "step": 1768 + }, + { + "epoch": 2.2016179215930305, + "grad_norm": 3.3948538303375244, + "learning_rate": 1.5129151291512916e-06, + "loss": 1.3425, + "step": 1769 + }, + { + "epoch": 2.2028624766645923, + "grad_norm": 3.4169349670410156, + "learning_rate": 1.510608856088561e-06, + "loss": 1.3285, + "step": 1770 + }, + { + "epoch": 2.2028624766645923, + "eval_loss": 1.6240391731262207, + "eval_runtime": 47.9216, + "eval_samples_per_second": 20.867, + "eval_steps_per_second": 0.876, + "step": 1770 + }, + { + "epoch": 2.204107031736154, + "grad_norm": 3.359754800796509, + "learning_rate": 1.5083025830258305e-06, + "loss": 1.3369, + "step": 1771 + }, + { + "epoch": 2.205351586807716, + "grad_norm": 3.280728340148926, + "learning_rate": 1.5059963099630997e-06, + "loss": 1.3381, + "step": 1772 + }, + { + "epoch": 2.2065961418792783, + "grad_norm": 3.4326274394989014, + "learning_rate": 1.503690036900369e-06, + "loss": 1.383, + "step": 1773 + }, + { + "epoch": 2.20784069695084, + "grad_norm": 3.3060855865478516, + "learning_rate": 1.5013837638376385e-06, + "loss": 1.36, + "step": 1774 + }, + { + "epoch": 2.209085252022402, + "grad_norm": 3.369748115539551, + "learning_rate": 1.499077490774908e-06, + "loss": 1.3699, + "step": 1775 + }, + { + "epoch": 2.210329807093964, + "grad_norm": 3.203705072402954, + "learning_rate": 1.4967712177121774e-06, + "loss": 1.3314, + "step": 1776 + }, + { + "epoch": 2.2115743621655257, + "grad_norm": 3.255790948867798, + "learning_rate": 1.4944649446494466e-06, + "loss": 1.3791, + "step": 1777 + }, + { + "epoch": 2.2128189172370876, + "grad_norm": 3.2817792892456055, + "learning_rate": 1.492158671586716e-06, + "loss": 1.3321, + "step": 1778 + }, + { + "epoch": 2.21406347230865, + "grad_norm": 3.3101930618286133, + "learning_rate": 1.4898523985239854e-06, + "loss": 1.4065, + "step": 1779 + }, + { + "epoch": 2.2153080273802117, + "grad_norm": 3.360100030899048, + "learning_rate": 1.4875461254612546e-06, + "loss": 1.3083, + "step": 1780 + }, + { + "epoch": 2.2153080273802117, + "eval_loss": 1.625471591949463, + "eval_runtime": 46.2605, + "eval_samples_per_second": 21.617, + "eval_steps_per_second": 0.908, + "step": 1780 + }, + { + "epoch": 2.2165525824517736, + "grad_norm": 3.3503854274749756, + "learning_rate": 1.485239852398524e-06, + "loss": 1.374, + "step": 1781 + }, + { + "epoch": 2.2177971375233354, + "grad_norm": 3.196617603302002, + "learning_rate": 1.4829335793357935e-06, + "loss": 1.3226, + "step": 1782 + }, + { + "epoch": 2.2190416925948973, + "grad_norm": 3.3946502208709717, + "learning_rate": 1.4806273062730627e-06, + "loss": 1.3445, + "step": 1783 + }, + { + "epoch": 2.220286247666459, + "grad_norm": 3.252145528793335, + "learning_rate": 1.4783210332103323e-06, + "loss": 1.3514, + "step": 1784 + }, + { + "epoch": 2.2215308027380214, + "grad_norm": 3.452378511428833, + "learning_rate": 1.4760147601476015e-06, + "loss": 1.3386, + "step": 1785 + }, + { + "epoch": 2.2227753578095832, + "grad_norm": 3.3612773418426514, + "learning_rate": 1.473708487084871e-06, + "loss": 1.3572, + "step": 1786 + }, + { + "epoch": 2.224019912881145, + "grad_norm": 3.295044422149658, + "learning_rate": 1.4714022140221404e-06, + "loss": 1.328, + "step": 1787 + }, + { + "epoch": 2.225264467952707, + "grad_norm": 3.2037711143493652, + "learning_rate": 1.4690959409594096e-06, + "loss": 1.2999, + "step": 1788 + }, + { + "epoch": 2.226509023024269, + "grad_norm": 3.3932483196258545, + "learning_rate": 1.4667896678966792e-06, + "loss": 1.3606, + "step": 1789 + }, + { + "epoch": 2.2277535780958306, + "grad_norm": 3.4563357830047607, + "learning_rate": 1.4644833948339484e-06, + "loss": 1.3875, + "step": 1790 + }, + { + "epoch": 2.2277535780958306, + "eval_loss": 1.622439980506897, + "eval_runtime": 47.9981, + "eval_samples_per_second": 20.834, + "eval_steps_per_second": 0.875, + "step": 1790 + }, + { + "epoch": 2.2289981331673925, + "grad_norm": 3.3003604412078857, + "learning_rate": 1.4621771217712176e-06, + "loss": 1.3373, + "step": 1791 + }, + { + "epoch": 2.2302426882389548, + "grad_norm": 3.398839235305786, + "learning_rate": 1.4598708487084873e-06, + "loss": 1.3432, + "step": 1792 + }, + { + "epoch": 2.2314872433105166, + "grad_norm": 3.3929128646850586, + "learning_rate": 1.4575645756457565e-06, + "loss": 1.347, + "step": 1793 + }, + { + "epoch": 2.2327317983820785, + "grad_norm": 3.21239972114563, + "learning_rate": 1.4552583025830261e-06, + "loss": 1.347, + "step": 1794 + }, + { + "epoch": 2.2339763534536403, + "grad_norm": 3.242029905319214, + "learning_rate": 1.4529520295202953e-06, + "loss": 1.3256, + "step": 1795 + }, + { + "epoch": 2.235220908525202, + "grad_norm": 3.3526346683502197, + "learning_rate": 1.4506457564575645e-06, + "loss": 1.315, + "step": 1796 + }, + { + "epoch": 2.236465463596764, + "grad_norm": 3.351020574569702, + "learning_rate": 1.4483394833948342e-06, + "loss": 1.357, + "step": 1797 + }, + { + "epoch": 2.2377100186683263, + "grad_norm": 3.402653694152832, + "learning_rate": 1.4460332103321034e-06, + "loss": 1.3695, + "step": 1798 + }, + { + "epoch": 2.238954573739888, + "grad_norm": 3.2951724529266357, + "learning_rate": 1.4437269372693726e-06, + "loss": 1.3366, + "step": 1799 + }, + { + "epoch": 2.24019912881145, + "grad_norm": 3.2638256549835205, + "learning_rate": 1.4414206642066422e-06, + "loss": 1.3147, + "step": 1800 + }, + { + "epoch": 2.24019912881145, + "eval_loss": 1.6255451440811157, + "eval_runtime": 49.0934, + "eval_samples_per_second": 20.369, + "eval_steps_per_second": 0.856, + "step": 1800 + }, + { + "epoch": 2.241443683883012, + "grad_norm": 3.536435604095459, + "learning_rate": 1.4391143911439114e-06, + "loss": 1.3609, + "step": 1801 + }, + { + "epoch": 2.2426882389545737, + "grad_norm": 3.219020366668701, + "learning_rate": 1.436808118081181e-06, + "loss": 1.3291, + "step": 1802 + }, + { + "epoch": 2.2439327940261355, + "grad_norm": 3.1892435550689697, + "learning_rate": 1.4345018450184503e-06, + "loss": 1.3384, + "step": 1803 + }, + { + "epoch": 2.2451773490976974, + "grad_norm": 3.2276196479797363, + "learning_rate": 1.4321955719557195e-06, + "loss": 1.3181, + "step": 1804 + }, + { + "epoch": 2.2464219041692597, + "grad_norm": 3.317469596862793, + "learning_rate": 1.4298892988929891e-06, + "loss": 1.3088, + "step": 1805 + }, + { + "epoch": 2.2476664592408215, + "grad_norm": 3.4198920726776123, + "learning_rate": 1.4275830258302584e-06, + "loss": 1.2905, + "step": 1806 + }, + { + "epoch": 2.2489110143123834, + "grad_norm": 3.4940414428710938, + "learning_rate": 1.425276752767528e-06, + "loss": 1.3419, + "step": 1807 + }, + { + "epoch": 2.2501555693839452, + "grad_norm": 3.4554691314697266, + "learning_rate": 1.4229704797047972e-06, + "loss": 1.3558, + "step": 1808 + }, + { + "epoch": 2.251400124455507, + "grad_norm": 3.490118980407715, + "learning_rate": 1.4206642066420664e-06, + "loss": 1.3639, + "step": 1809 + }, + { + "epoch": 2.252644679527069, + "grad_norm": 3.452986717224121, + "learning_rate": 1.418357933579336e-06, + "loss": 1.343, + "step": 1810 + }, + { + "epoch": 2.252644679527069, + "eval_loss": 1.6230783462524414, + "eval_runtime": 43.9128, + "eval_samples_per_second": 22.772, + "eval_steps_per_second": 0.956, + "step": 1810 + }, + { + "epoch": 2.253889234598631, + "grad_norm": 3.3851468563079834, + "learning_rate": 1.4160516605166053e-06, + "loss": 1.3067, + "step": 1811 + }, + { + "epoch": 2.255133789670193, + "grad_norm": 3.373572826385498, + "learning_rate": 1.4137453874538745e-06, + "loss": 1.3826, + "step": 1812 + }, + { + "epoch": 2.256378344741755, + "grad_norm": 3.5561487674713135, + "learning_rate": 1.411439114391144e-06, + "loss": 1.3662, + "step": 1813 + }, + { + "epoch": 2.2576228998133168, + "grad_norm": 3.4163501262664795, + "learning_rate": 1.4091328413284133e-06, + "loss": 1.3664, + "step": 1814 + }, + { + "epoch": 2.2588674548848786, + "grad_norm": 3.2873945236206055, + "learning_rate": 1.406826568265683e-06, + "loss": 1.3277, + "step": 1815 + }, + { + "epoch": 2.2601120099564405, + "grad_norm": 3.2029595375061035, + "learning_rate": 1.4045202952029522e-06, + "loss": 1.3202, + "step": 1816 + }, + { + "epoch": 2.2613565650280023, + "grad_norm": 3.376417398452759, + "learning_rate": 1.4022140221402214e-06, + "loss": 1.3264, + "step": 1817 + }, + { + "epoch": 2.2626011200995646, + "grad_norm": 3.4385898113250732, + "learning_rate": 1.399907749077491e-06, + "loss": 1.3571, + "step": 1818 + }, + { + "epoch": 2.2638456751711264, + "grad_norm": 3.342299461364746, + "learning_rate": 1.3976014760147602e-06, + "loss": 1.3385, + "step": 1819 + }, + { + "epoch": 2.2650902302426883, + "grad_norm": 3.355975389480591, + "learning_rate": 1.3952952029520298e-06, + "loss": 1.346, + "step": 1820 + }, + { + "epoch": 2.2650902302426883, + "eval_loss": 1.6243224143981934, + "eval_runtime": 46.6848, + "eval_samples_per_second": 21.42, + "eval_steps_per_second": 0.9, + "step": 1820 + }, + { + "epoch": 2.26633478531425, + "grad_norm": 3.4472744464874268, + "learning_rate": 1.392988929889299e-06, + "loss": 1.404, + "step": 1821 + }, + { + "epoch": 2.267579340385812, + "grad_norm": 3.3384392261505127, + "learning_rate": 1.3906826568265683e-06, + "loss": 1.3492, + "step": 1822 + }, + { + "epoch": 2.268823895457374, + "grad_norm": 3.3477087020874023, + "learning_rate": 1.388376383763838e-06, + "loss": 1.3565, + "step": 1823 + }, + { + "epoch": 2.270068450528936, + "grad_norm": 3.3641061782836914, + "learning_rate": 1.3860701107011071e-06, + "loss": 1.344, + "step": 1824 + }, + { + "epoch": 2.271313005600498, + "grad_norm": 3.435539960861206, + "learning_rate": 1.3837638376383765e-06, + "loss": 1.3709, + "step": 1825 + }, + { + "epoch": 2.27255756067206, + "grad_norm": 3.349618434906006, + "learning_rate": 1.381457564575646e-06, + "loss": 1.3265, + "step": 1826 + }, + { + "epoch": 2.2738021157436217, + "grad_norm": 3.402650833129883, + "learning_rate": 1.3791512915129152e-06, + "loss": 1.3537, + "step": 1827 + }, + { + "epoch": 2.2750466708151835, + "grad_norm": 3.389064311981201, + "learning_rate": 1.3768450184501846e-06, + "loss": 1.3843, + "step": 1828 + }, + { + "epoch": 2.2762912258867454, + "grad_norm": 3.3653721809387207, + "learning_rate": 1.374538745387454e-06, + "loss": 1.376, + "step": 1829 + }, + { + "epoch": 2.277535780958307, + "grad_norm": 3.254830837249756, + "learning_rate": 1.3722324723247232e-06, + "loss": 1.344, + "step": 1830 + }, + { + "epoch": 2.277535780958307, + "eval_loss": 1.621465802192688, + "eval_runtime": 52.2643, + "eval_samples_per_second": 19.134, + "eval_steps_per_second": 0.804, + "step": 1830 + }, + { + "epoch": 2.2787803360298695, + "grad_norm": 3.4339189529418945, + "learning_rate": 1.3699261992619929e-06, + "loss": 1.3539, + "step": 1831 + }, + { + "epoch": 2.2800248911014314, + "grad_norm": 3.5471079349517822, + "learning_rate": 1.367619926199262e-06, + "loss": 1.4031, + "step": 1832 + }, + { + "epoch": 2.281269446172993, + "grad_norm": 3.4783706665039062, + "learning_rate": 1.3653136531365315e-06, + "loss": 1.3504, + "step": 1833 + }, + { + "epoch": 2.282514001244555, + "grad_norm": 3.360304355621338, + "learning_rate": 1.363007380073801e-06, + "loss": 1.3215, + "step": 1834 + }, + { + "epoch": 2.283758556316117, + "grad_norm": 3.447568416595459, + "learning_rate": 1.3607011070110701e-06, + "loss": 1.3286, + "step": 1835 + }, + { + "epoch": 2.2850031113876788, + "grad_norm": 3.311884641647339, + "learning_rate": 1.3583948339483396e-06, + "loss": 1.3288, + "step": 1836 + }, + { + "epoch": 2.286247666459241, + "grad_norm": 3.3999180793762207, + "learning_rate": 1.356088560885609e-06, + "loss": 1.2837, + "step": 1837 + }, + { + "epoch": 2.287492221530803, + "grad_norm": 3.360581636428833, + "learning_rate": 1.3537822878228784e-06, + "loss": 1.3233, + "step": 1838 + }, + { + "epoch": 2.2887367766023647, + "grad_norm": 3.3952338695526123, + "learning_rate": 1.3514760147601478e-06, + "loss": 1.3702, + "step": 1839 + }, + { + "epoch": 2.2899813316739266, + "grad_norm": 3.5232338905334473, + "learning_rate": 1.349169741697417e-06, + "loss": 1.4087, + "step": 1840 + }, + { + "epoch": 2.2899813316739266, + "eval_loss": 1.6189563274383545, + "eval_runtime": 51.0863, + "eval_samples_per_second": 19.575, + "eval_steps_per_second": 0.822, + "step": 1840 + }, + { + "epoch": 2.2912258867454884, + "grad_norm": 3.4118707180023193, + "learning_rate": 1.3468634686346865e-06, + "loss": 1.3838, + "step": 1841 + }, + { + "epoch": 2.2924704418170503, + "grad_norm": 3.511007070541382, + "learning_rate": 1.3445571955719559e-06, + "loss": 1.3846, + "step": 1842 + }, + { + "epoch": 2.293714996888612, + "grad_norm": 3.44270920753479, + "learning_rate": 1.3422509225092253e-06, + "loss": 1.339, + "step": 1843 + }, + { + "epoch": 2.2949595519601744, + "grad_norm": 3.322871446609497, + "learning_rate": 1.3399446494464945e-06, + "loss": 1.289, + "step": 1844 + }, + { + "epoch": 2.2962041070317363, + "grad_norm": 3.3219494819641113, + "learning_rate": 1.337638376383764e-06, + "loss": 1.3613, + "step": 1845 + }, + { + "epoch": 2.297448662103298, + "grad_norm": 3.3900601863861084, + "learning_rate": 1.3353321033210334e-06, + "loss": 1.3516, + "step": 1846 + }, + { + "epoch": 2.29869321717486, + "grad_norm": 3.4780476093292236, + "learning_rate": 1.3330258302583028e-06, + "loss": 1.3779, + "step": 1847 + }, + { + "epoch": 2.299937772246422, + "grad_norm": 3.485691547393799, + "learning_rate": 1.330719557195572e-06, + "loss": 1.3, + "step": 1848 + }, + { + "epoch": 2.3011823273179837, + "grad_norm": 3.456273317337036, + "learning_rate": 1.3284132841328414e-06, + "loss": 1.3946, + "step": 1849 + }, + { + "epoch": 2.302426882389546, + "grad_norm": 3.2921931743621826, + "learning_rate": 1.3261070110701108e-06, + "loss": 1.3338, + "step": 1850 + }, + { + "epoch": 2.302426882389546, + "eval_loss": 1.6193547248840332, + "eval_runtime": 51.2837, + "eval_samples_per_second": 19.499, + "eval_steps_per_second": 0.819, + "step": 1850 + }, + { + "epoch": 2.303671437461108, + "grad_norm": 3.311541795730591, + "learning_rate": 1.3238007380073803e-06, + "loss": 1.344, + "step": 1851 + }, + { + "epoch": 2.3049159925326697, + "grad_norm": 3.2640180587768555, + "learning_rate": 1.3214944649446495e-06, + "loss": 1.3345, + "step": 1852 + }, + { + "epoch": 2.3061605476042315, + "grad_norm": 3.3187389373779297, + "learning_rate": 1.319188191881919e-06, + "loss": 1.3325, + "step": 1853 + }, + { + "epoch": 2.3074051026757934, + "grad_norm": 3.478851795196533, + "learning_rate": 1.3168819188191883e-06, + "loss": 1.3748, + "step": 1854 + }, + { + "epoch": 2.308649657747355, + "grad_norm": 3.32222580909729, + "learning_rate": 1.3145756457564575e-06, + "loss": 1.3538, + "step": 1855 + }, + { + "epoch": 2.309894212818917, + "grad_norm": 3.2635533809661865, + "learning_rate": 1.3122693726937272e-06, + "loss": 1.3459, + "step": 1856 + }, + { + "epoch": 2.3111387678904793, + "grad_norm": 3.335123300552368, + "learning_rate": 1.3099630996309964e-06, + "loss": 1.3621, + "step": 1857 + }, + { + "epoch": 2.312383322962041, + "grad_norm": 3.4098165035247803, + "learning_rate": 1.3076568265682658e-06, + "loss": 1.3431, + "step": 1858 + }, + { + "epoch": 2.313627878033603, + "grad_norm": 3.2582387924194336, + "learning_rate": 1.3053505535055352e-06, + "loss": 1.3591, + "step": 1859 + }, + { + "epoch": 2.314872433105165, + "grad_norm": 3.2456133365631104, + "learning_rate": 1.3030442804428044e-06, + "loss": 1.3367, + "step": 1860 + }, + { + "epoch": 2.314872433105165, + "eval_loss": 1.6218940019607544, + "eval_runtime": 54.4652, + "eval_samples_per_second": 18.36, + "eval_steps_per_second": 0.771, + "step": 1860 + }, + { + "epoch": 2.3161169881767267, + "grad_norm": 3.220767021179199, + "learning_rate": 1.3007380073800739e-06, + "loss": 1.2744, + "step": 1861 + }, + { + "epoch": 2.3173615432482886, + "grad_norm": 3.4657704830169678, + "learning_rate": 1.2984317343173433e-06, + "loss": 1.366, + "step": 1862 + }, + { + "epoch": 2.318606098319851, + "grad_norm": 3.399174928665161, + "learning_rate": 1.2961254612546125e-06, + "loss": 1.3775, + "step": 1863 + }, + { + "epoch": 2.3198506533914127, + "grad_norm": 3.4700374603271484, + "learning_rate": 1.2938191881918821e-06, + "loss": 1.3456, + "step": 1864 + }, + { + "epoch": 2.3210952084629746, + "grad_norm": 3.2533438205718994, + "learning_rate": 1.2915129151291513e-06, + "loss": 1.2934, + "step": 1865 + }, + { + "epoch": 2.3223397635345364, + "grad_norm": 3.4083049297332764, + "learning_rate": 1.2892066420664208e-06, + "loss": 1.3759, + "step": 1866 + }, + { + "epoch": 2.3235843186060983, + "grad_norm": 3.3690836429595947, + "learning_rate": 1.2869003690036902e-06, + "loss": 1.313, + "step": 1867 + }, + { + "epoch": 2.32482887367766, + "grad_norm": 3.4750704765319824, + "learning_rate": 1.2845940959409594e-06, + "loss": 1.3658, + "step": 1868 + }, + { + "epoch": 2.326073428749222, + "grad_norm": 3.458674192428589, + "learning_rate": 1.282287822878229e-06, + "loss": 1.4026, + "step": 1869 + }, + { + "epoch": 2.3273179838207843, + "grad_norm": 3.5049021244049072, + "learning_rate": 1.2799815498154982e-06, + "loss": 1.3638, + "step": 1870 + }, + { + "epoch": 2.3273179838207843, + "eval_loss": 1.622696876525879, + "eval_runtime": 49.0313, + "eval_samples_per_second": 20.395, + "eval_steps_per_second": 0.857, + "step": 1870 + }, + { + "epoch": 2.328562538892346, + "grad_norm": 3.3939497470855713, + "learning_rate": 1.2776752767527675e-06, + "loss": 1.3198, + "step": 1871 + }, + { + "epoch": 2.329807093963908, + "grad_norm": 3.3821396827697754, + "learning_rate": 1.275369003690037e-06, + "loss": 1.3682, + "step": 1872 + }, + { + "epoch": 2.33105164903547, + "grad_norm": 3.6034491062164307, + "learning_rate": 1.2730627306273063e-06, + "loss": 1.3476, + "step": 1873 + }, + { + "epoch": 2.3322962041070316, + "grad_norm": 3.3996806144714355, + "learning_rate": 1.270756457564576e-06, + "loss": 1.3239, + "step": 1874 + }, + { + "epoch": 2.3335407591785935, + "grad_norm": 3.2999489307403564, + "learning_rate": 1.2684501845018451e-06, + "loss": 1.311, + "step": 1875 + }, + { + "epoch": 2.334785314250156, + "grad_norm": 3.292593002319336, + "learning_rate": 1.2661439114391144e-06, + "loss": 1.3333, + "step": 1876 + }, + { + "epoch": 2.3360298693217176, + "grad_norm": 3.323923349380493, + "learning_rate": 1.263837638376384e-06, + "loss": 1.3154, + "step": 1877 + }, + { + "epoch": 2.3372744243932795, + "grad_norm": 3.2100625038146973, + "learning_rate": 1.2615313653136532e-06, + "loss": 1.3248, + "step": 1878 + }, + { + "epoch": 2.3385189794648413, + "grad_norm": 3.3764309883117676, + "learning_rate": 1.2592250922509224e-06, + "loss": 1.3574, + "step": 1879 + }, + { + "epoch": 2.339763534536403, + "grad_norm": 3.3943352699279785, + "learning_rate": 1.256918819188192e-06, + "loss": 1.3958, + "step": 1880 + }, + { + "epoch": 2.339763534536403, + "eval_loss": 1.6206164360046387, + "eval_runtime": 47.5352, + "eval_samples_per_second": 21.037, + "eval_steps_per_second": 0.884, + "step": 1880 + }, + { + "epoch": 2.341008089607965, + "grad_norm": 3.475947141647339, + "learning_rate": 1.2546125461254613e-06, + "loss": 1.3491, + "step": 1881 + }, + { + "epoch": 2.342252644679527, + "grad_norm": 3.3503031730651855, + "learning_rate": 1.2523062730627309e-06, + "loss": 1.3198, + "step": 1882 + }, + { + "epoch": 2.343497199751089, + "grad_norm": 3.290644884109497, + "learning_rate": 1.25e-06, + "loss": 1.3237, + "step": 1883 + }, + { + "epoch": 2.344741754822651, + "grad_norm": 3.2442216873168945, + "learning_rate": 1.2476937269372695e-06, + "loss": 1.3149, + "step": 1884 + }, + { + "epoch": 2.345986309894213, + "grad_norm": 3.2475531101226807, + "learning_rate": 1.245387453874539e-06, + "loss": 1.3265, + "step": 1885 + }, + { + "epoch": 2.3472308649657747, + "grad_norm": 3.16241192817688, + "learning_rate": 1.2430811808118082e-06, + "loss": 1.3237, + "step": 1886 + }, + { + "epoch": 2.3484754200373366, + "grad_norm": 3.6679129600524902, + "learning_rate": 1.2407749077490776e-06, + "loss": 1.3583, + "step": 1887 + }, + { + "epoch": 2.3497199751088984, + "grad_norm": 3.5134646892547607, + "learning_rate": 1.238468634686347e-06, + "loss": 1.3543, + "step": 1888 + }, + { + "epoch": 2.3509645301804607, + "grad_norm": 3.2337441444396973, + "learning_rate": 1.2361623616236164e-06, + "loss": 1.3, + "step": 1889 + }, + { + "epoch": 2.3522090852520225, + "grad_norm": 3.4864821434020996, + "learning_rate": 1.2338560885608856e-06, + "loss": 1.3786, + "step": 1890 + }, + { + "epoch": 2.3522090852520225, + "eval_loss": 1.6194350719451904, + "eval_runtime": 53.3692, + "eval_samples_per_second": 18.737, + "eval_steps_per_second": 0.787, + "step": 1890 + }, + { + "epoch": 2.3534536403235844, + "grad_norm": 3.4141929149627686, + "learning_rate": 1.231549815498155e-06, + "loss": 1.3584, + "step": 1891 + }, + { + "epoch": 2.3546981953951462, + "grad_norm": 3.3861517906188965, + "learning_rate": 1.2292435424354245e-06, + "loss": 1.371, + "step": 1892 + }, + { + "epoch": 2.355942750466708, + "grad_norm": 3.345811128616333, + "learning_rate": 1.226937269372694e-06, + "loss": 1.326, + "step": 1893 + }, + { + "epoch": 2.35718730553827, + "grad_norm": 3.3748316764831543, + "learning_rate": 1.2246309963099633e-06, + "loss": 1.3422, + "step": 1894 + }, + { + "epoch": 2.358431860609832, + "grad_norm": 3.4072210788726807, + "learning_rate": 1.2223247232472325e-06, + "loss": 1.3706, + "step": 1895 + }, + { + "epoch": 2.359676415681394, + "grad_norm": 3.3167407512664795, + "learning_rate": 1.220018450184502e-06, + "loss": 1.3553, + "step": 1896 + }, + { + "epoch": 2.360920970752956, + "grad_norm": 3.3539159297943115, + "learning_rate": 1.2177121771217714e-06, + "loss": 1.3306, + "step": 1897 + }, + { + "epoch": 2.3621655258245178, + "grad_norm": 3.4795944690704346, + "learning_rate": 1.2154059040590408e-06, + "loss": 1.3536, + "step": 1898 + }, + { + "epoch": 2.3634100808960796, + "grad_norm": 3.425493001937866, + "learning_rate": 1.21309963099631e-06, + "loss": 1.321, + "step": 1899 + }, + { + "epoch": 2.3646546359676415, + "grad_norm": 3.415886402130127, + "learning_rate": 1.2107933579335794e-06, + "loss": 1.3477, + "step": 1900 + }, + { + "epoch": 2.3646546359676415, + "eval_loss": 1.619869589805603, + "eval_runtime": 54.2118, + "eval_samples_per_second": 18.446, + "eval_steps_per_second": 0.775, + "step": 1900 + }, + { + "epoch": 2.3658991910392033, + "grad_norm": 3.6500039100646973, + "learning_rate": 1.2084870848708489e-06, + "loss": 1.406, + "step": 1901 + }, + { + "epoch": 2.3671437461107656, + "grad_norm": 3.332648992538452, + "learning_rate": 1.2061808118081183e-06, + "loss": 1.3288, + "step": 1902 + }, + { + "epoch": 2.3683883011823275, + "grad_norm": 3.2978408336639404, + "learning_rate": 1.2038745387453875e-06, + "loss": 1.3087, + "step": 1903 + }, + { + "epoch": 2.3696328562538893, + "grad_norm": 3.459388494491577, + "learning_rate": 1.201568265682657e-06, + "loss": 1.3754, + "step": 1904 + }, + { + "epoch": 2.370877411325451, + "grad_norm": 3.4902889728546143, + "learning_rate": 1.1992619926199263e-06, + "loss": 1.3529, + "step": 1905 + }, + { + "epoch": 2.372121966397013, + "grad_norm": 3.393232822418213, + "learning_rate": 1.1969557195571958e-06, + "loss": 1.3517, + "step": 1906 + }, + { + "epoch": 2.373366521468575, + "grad_norm": 3.4551451206207275, + "learning_rate": 1.1946494464944652e-06, + "loss": 1.3372, + "step": 1907 + }, + { + "epoch": 2.3746110765401367, + "grad_norm": 3.3197243213653564, + "learning_rate": 1.1923431734317344e-06, + "loss": 1.3169, + "step": 1908 + }, + { + "epoch": 2.375855631611699, + "grad_norm": 3.3255650997161865, + "learning_rate": 1.1900369003690038e-06, + "loss": 1.3462, + "step": 1909 + }, + { + "epoch": 2.377100186683261, + "grad_norm": 3.4901344776153564, + "learning_rate": 1.1877306273062733e-06, + "loss": 1.3296, + "step": 1910 + }, + { + "epoch": 2.377100186683261, + "eval_loss": 1.6212565898895264, + "eval_runtime": 43.2416, + "eval_samples_per_second": 23.126, + "eval_steps_per_second": 0.971, + "step": 1910 + }, + { + "epoch": 2.3783447417548227, + "grad_norm": 3.489834785461426, + "learning_rate": 1.1854243542435427e-06, + "loss": 1.358, + "step": 1911 + }, + { + "epoch": 2.3795892968263845, + "grad_norm": 3.513260841369629, + "learning_rate": 1.1831180811808119e-06, + "loss": 1.3688, + "step": 1912 + }, + { + "epoch": 2.3808338518979464, + "grad_norm": 3.464785099029541, + "learning_rate": 1.1808118081180813e-06, + "loss": 1.3291, + "step": 1913 + }, + { + "epoch": 2.3820784069695082, + "grad_norm": 3.2861804962158203, + "learning_rate": 1.1785055350553507e-06, + "loss": 1.3056, + "step": 1914 + }, + { + "epoch": 2.3833229620410705, + "grad_norm": 3.435983419418335, + "learning_rate": 1.1761992619926202e-06, + "loss": 1.3551, + "step": 1915 + }, + { + "epoch": 2.3845675171126324, + "grad_norm": 3.330296516418457, + "learning_rate": 1.1738929889298894e-06, + "loss": 1.3152, + "step": 1916 + }, + { + "epoch": 2.385812072184194, + "grad_norm": 3.300424337387085, + "learning_rate": 1.1715867158671588e-06, + "loss": 1.3446, + "step": 1917 + }, + { + "epoch": 2.387056627255756, + "grad_norm": 3.2503697872161865, + "learning_rate": 1.1692804428044282e-06, + "loss": 1.3378, + "step": 1918 + }, + { + "epoch": 2.388301182327318, + "grad_norm": 3.330183982849121, + "learning_rate": 1.1669741697416974e-06, + "loss": 1.3308, + "step": 1919 + }, + { + "epoch": 2.3895457373988798, + "grad_norm": 3.4580652713775635, + "learning_rate": 1.1646678966789668e-06, + "loss": 1.3309, + "step": 1920 + }, + { + "epoch": 2.3895457373988798, + "eval_loss": 1.6166638135910034, + "eval_runtime": 44.9035, + "eval_samples_per_second": 22.27, + "eval_steps_per_second": 0.935, + "step": 1920 + }, + { + "epoch": 2.3907902924704416, + "grad_norm": 3.2584407329559326, + "learning_rate": 1.1623616236162363e-06, + "loss": 1.3084, + "step": 1921 + }, + { + "epoch": 2.392034847542004, + "grad_norm": 3.272989511489868, + "learning_rate": 1.1600553505535057e-06, + "loss": 1.3238, + "step": 1922 + }, + { + "epoch": 2.3932794026135658, + "grad_norm": 3.3732693195343018, + "learning_rate": 1.157749077490775e-06, + "loss": 1.3391, + "step": 1923 + }, + { + "epoch": 2.3945239576851276, + "grad_norm": 3.3817999362945557, + "learning_rate": 1.1554428044280443e-06, + "loss": 1.3568, + "step": 1924 + }, + { + "epoch": 2.3957685127566895, + "grad_norm": 3.285632610321045, + "learning_rate": 1.1531365313653137e-06, + "loss": 1.3158, + "step": 1925 + }, + { + "epoch": 2.3970130678282513, + "grad_norm": 3.267397880554199, + "learning_rate": 1.1508302583025832e-06, + "loss": 1.2795, + "step": 1926 + }, + { + "epoch": 2.398257622899813, + "grad_norm": 3.555927276611328, + "learning_rate": 1.1485239852398524e-06, + "loss": 1.3788, + "step": 1927 + }, + { + "epoch": 2.3995021779713754, + "grad_norm": 3.4175591468811035, + "learning_rate": 1.1462177121771218e-06, + "loss": 1.3019, + "step": 1928 + }, + { + "epoch": 2.4007467330429373, + "grad_norm": 3.4269983768463135, + "learning_rate": 1.1439114391143912e-06, + "loss": 1.3145, + "step": 1929 + }, + { + "epoch": 2.401991288114499, + "grad_norm": 3.3499817848205566, + "learning_rate": 1.1416051660516606e-06, + "loss": 1.3206, + "step": 1930 + }, + { + "epoch": 2.401991288114499, + "eval_loss": 1.6193190813064575, + "eval_runtime": 44.8302, + "eval_samples_per_second": 22.306, + "eval_steps_per_second": 0.937, + "step": 1930 + }, + { + "epoch": 2.403235843186061, + "grad_norm": 3.3621203899383545, + "learning_rate": 1.1392988929889299e-06, + "loss": 1.3276, + "step": 1931 + }, + { + "epoch": 2.404480398257623, + "grad_norm": 3.2818424701690674, + "learning_rate": 1.1369926199261993e-06, + "loss": 1.3234, + "step": 1932 + }, + { + "epoch": 2.4057249533291847, + "grad_norm": 3.2797811031341553, + "learning_rate": 1.1346863468634687e-06, + "loss": 1.3069, + "step": 1933 + }, + { + "epoch": 2.4069695084007465, + "grad_norm": 3.3689663410186768, + "learning_rate": 1.1323800738007381e-06, + "loss": 1.314, + "step": 1934 + }, + { + "epoch": 2.408214063472309, + "grad_norm": 3.4538166522979736, + "learning_rate": 1.1300738007380073e-06, + "loss": 1.3236, + "step": 1935 + }, + { + "epoch": 2.4094586185438707, + "grad_norm": 3.3834025859832764, + "learning_rate": 1.1277675276752768e-06, + "loss": 1.3587, + "step": 1936 + }, + { + "epoch": 2.4107031736154325, + "grad_norm": 3.3694424629211426, + "learning_rate": 1.1254612546125462e-06, + "loss": 1.3452, + "step": 1937 + }, + { + "epoch": 2.4119477286869944, + "grad_norm": 3.471663236618042, + "learning_rate": 1.1231549815498156e-06, + "loss": 1.3622, + "step": 1938 + }, + { + "epoch": 2.413192283758556, + "grad_norm": 3.505692958831787, + "learning_rate": 1.1208487084870848e-06, + "loss": 1.326, + "step": 1939 + }, + { + "epoch": 2.414436838830118, + "grad_norm": 3.4514808654785156, + "learning_rate": 1.1185424354243542e-06, + "loss": 1.3603, + "step": 1940 + }, + { + "epoch": 2.414436838830118, + "eval_loss": 1.6171061992645264, + "eval_runtime": 44.7206, + "eval_samples_per_second": 22.361, + "eval_steps_per_second": 0.939, + "step": 1940 + }, + { + "epoch": 2.4156813939016804, + "grad_norm": 3.5669631958007812, + "learning_rate": 1.1162361623616237e-06, + "loss": 1.3357, + "step": 1941 + }, + { + "epoch": 2.416925948973242, + "grad_norm": 3.396930694580078, + "learning_rate": 1.113929889298893e-06, + "loss": 1.337, + "step": 1942 + }, + { + "epoch": 2.418170504044804, + "grad_norm": 3.389120101928711, + "learning_rate": 1.1116236162361623e-06, + "loss": 1.319, + "step": 1943 + }, + { + "epoch": 2.419415059116366, + "grad_norm": 3.613391399383545, + "learning_rate": 1.1093173431734317e-06, + "loss": 1.371, + "step": 1944 + }, + { + "epoch": 2.4206596141879277, + "grad_norm": 3.441382646560669, + "learning_rate": 1.1070110701107011e-06, + "loss": 1.3522, + "step": 1945 + }, + { + "epoch": 2.4219041692594896, + "grad_norm": 3.3786375522613525, + "learning_rate": 1.1047047970479706e-06, + "loss": 1.3468, + "step": 1946 + }, + { + "epoch": 2.4231487243310514, + "grad_norm": 3.432734489440918, + "learning_rate": 1.10239852398524e-06, + "loss": 1.3571, + "step": 1947 + }, + { + "epoch": 2.4243932794026137, + "grad_norm": 3.621337413787842, + "learning_rate": 1.1000922509225092e-06, + "loss": 1.3591, + "step": 1948 + }, + { + "epoch": 2.4256378344741756, + "grad_norm": 3.4495420455932617, + "learning_rate": 1.0977859778597786e-06, + "loss": 1.375, + "step": 1949 + }, + { + "epoch": 2.4268823895457374, + "grad_norm": 3.4852328300476074, + "learning_rate": 1.095479704797048e-06, + "loss": 1.3391, + "step": 1950 + }, + { + "epoch": 2.4268823895457374, + "eval_loss": 1.6171005964279175, + "eval_runtime": 51.922, + "eval_samples_per_second": 19.26, + "eval_steps_per_second": 0.809, + "step": 1950 + }, + { + "epoch": 2.4281269446172993, + "grad_norm": 3.455693006515503, + "learning_rate": 1.0931734317343175e-06, + "loss": 1.3703, + "step": 1951 + }, + { + "epoch": 2.429371499688861, + "grad_norm": 3.2051644325256348, + "learning_rate": 1.0908671586715867e-06, + "loss": 1.2962, + "step": 1952 + }, + { + "epoch": 2.430616054760423, + "grad_norm": 3.384723424911499, + "learning_rate": 1.088560885608856e-06, + "loss": 1.3794, + "step": 1953 + }, + { + "epoch": 2.4318606098319853, + "grad_norm": 3.3716330528259277, + "learning_rate": 1.0862546125461255e-06, + "loss": 1.3045, + "step": 1954 + }, + { + "epoch": 2.433105164903547, + "grad_norm": 3.48067569732666, + "learning_rate": 1.083948339483395e-06, + "loss": 1.2728, + "step": 1955 + }, + { + "epoch": 2.434349719975109, + "grad_norm": 3.200029134750366, + "learning_rate": 1.0816420664206644e-06, + "loss": 1.353, + "step": 1956 + }, + { + "epoch": 2.435594275046671, + "grad_norm": 3.413900375366211, + "learning_rate": 1.0793357933579336e-06, + "loss": 1.3946, + "step": 1957 + }, + { + "epoch": 2.4368388301182327, + "grad_norm": 3.2179441452026367, + "learning_rate": 1.077029520295203e-06, + "loss": 1.3164, + "step": 1958 + }, + { + "epoch": 2.4380833851897945, + "grad_norm": 3.321248769760132, + "learning_rate": 1.0747232472324724e-06, + "loss": 1.2627, + "step": 1959 + }, + { + "epoch": 2.4393279402613564, + "grad_norm": 3.2880983352661133, + "learning_rate": 1.0724169741697419e-06, + "loss": 1.3106, + "step": 1960 + }, + { + "epoch": 2.4393279402613564, + "eval_loss": 1.6168919801712036, + "eval_runtime": 45.4429, + "eval_samples_per_second": 22.006, + "eval_steps_per_second": 0.924, + "step": 1960 + }, + { + "epoch": 2.4405724953329186, + "grad_norm": 3.445110559463501, + "learning_rate": 1.070110701107011e-06, + "loss": 1.4013, + "step": 1961 + }, + { + "epoch": 2.4418170504044805, + "grad_norm": 3.694018602371216, + "learning_rate": 1.0678044280442805e-06, + "loss": 1.3347, + "step": 1962 + }, + { + "epoch": 2.4430616054760423, + "grad_norm": 3.2435543537139893, + "learning_rate": 1.06549815498155e-06, + "loss": 1.3277, + "step": 1963 + }, + { + "epoch": 2.444306160547604, + "grad_norm": 3.271838903427124, + "learning_rate": 1.0631918819188193e-06, + "loss": 1.3149, + "step": 1964 + }, + { + "epoch": 2.445550715619166, + "grad_norm": 3.4159295558929443, + "learning_rate": 1.0608856088560888e-06, + "loss": 1.3658, + "step": 1965 + }, + { + "epoch": 2.446795270690728, + "grad_norm": 3.3939833641052246, + "learning_rate": 1.058579335793358e-06, + "loss": 1.3466, + "step": 1966 + }, + { + "epoch": 2.44803982576229, + "grad_norm": 3.384528636932373, + "learning_rate": 1.0562730627306274e-06, + "loss": 1.3756, + "step": 1967 + }, + { + "epoch": 2.449284380833852, + "grad_norm": 3.414613962173462, + "learning_rate": 1.0539667896678968e-06, + "loss": 1.3655, + "step": 1968 + }, + { + "epoch": 2.450528935905414, + "grad_norm": 3.403575897216797, + "learning_rate": 1.0516605166051662e-06, + "loss": 1.3402, + "step": 1969 + }, + { + "epoch": 2.4517734909769757, + "grad_norm": 3.4550907611846924, + "learning_rate": 1.0493542435424354e-06, + "loss": 1.3727, + "step": 1970 + }, + { + "epoch": 2.4517734909769757, + "eval_loss": 1.6175874471664429, + "eval_runtime": 46.6225, + "eval_samples_per_second": 21.449, + "eval_steps_per_second": 0.901, + "step": 1970 + }, + { + "epoch": 2.4530180460485376, + "grad_norm": 3.4611523151397705, + "learning_rate": 1.0470479704797049e-06, + "loss": 1.3432, + "step": 1971 + }, + { + "epoch": 2.4542626011200994, + "grad_norm": 3.4850761890411377, + "learning_rate": 1.0447416974169743e-06, + "loss": 1.3441, + "step": 1972 + }, + { + "epoch": 2.4555071561916613, + "grad_norm": 3.49817156791687, + "learning_rate": 1.0424354243542437e-06, + "loss": 1.3137, + "step": 1973 + }, + { + "epoch": 2.4567517112632236, + "grad_norm": 3.390441656112671, + "learning_rate": 1.0401291512915131e-06, + "loss": 1.2719, + "step": 1974 + }, + { + "epoch": 2.4579962663347854, + "grad_norm": 3.43735933303833, + "learning_rate": 1.0378228782287824e-06, + "loss": 1.3539, + "step": 1975 + }, + { + "epoch": 2.4592408214063473, + "grad_norm": 3.173717737197876, + "learning_rate": 1.0355166051660518e-06, + "loss": 1.2953, + "step": 1976 + }, + { + "epoch": 2.460485376477909, + "grad_norm": 3.313398599624634, + "learning_rate": 1.0332103321033212e-06, + "loss": 1.2967, + "step": 1977 + }, + { + "epoch": 2.461729931549471, + "grad_norm": 3.2390615940093994, + "learning_rate": 1.0309040590405906e-06, + "loss": 1.3208, + "step": 1978 + }, + { + "epoch": 2.462974486621033, + "grad_norm": 3.2934212684631348, + "learning_rate": 1.0285977859778598e-06, + "loss": 1.3129, + "step": 1979 + }, + { + "epoch": 2.464219041692595, + "grad_norm": 3.342038631439209, + "learning_rate": 1.0262915129151293e-06, + "loss": 1.3438, + "step": 1980 + }, + { + "epoch": 2.464219041692595, + "eval_loss": 1.616586685180664, + "eval_runtime": 44.6397, + "eval_samples_per_second": 22.402, + "eval_steps_per_second": 0.941, + "step": 1980 + }, + { + "epoch": 2.465463596764157, + "grad_norm": 3.3280296325683594, + "learning_rate": 1.0239852398523987e-06, + "loss": 1.3251, + "step": 1981 + }, + { + "epoch": 2.466708151835719, + "grad_norm": 3.2722764015197754, + "learning_rate": 1.021678966789668e-06, + "loss": 1.3422, + "step": 1982 + }, + { + "epoch": 2.4679527069072806, + "grad_norm": 3.3998475074768066, + "learning_rate": 1.0193726937269373e-06, + "loss": 1.3285, + "step": 1983 + }, + { + "epoch": 2.4691972619788425, + "grad_norm": 3.4187684059143066, + "learning_rate": 1.0170664206642067e-06, + "loss": 1.398, + "step": 1984 + }, + { + "epoch": 2.4704418170504043, + "grad_norm": 3.324049472808838, + "learning_rate": 1.0147601476014762e-06, + "loss": 1.3178, + "step": 1985 + }, + { + "epoch": 2.471686372121966, + "grad_norm": 3.333989143371582, + "learning_rate": 1.0124538745387456e-06, + "loss": 1.265, + "step": 1986 + }, + { + "epoch": 2.4729309271935285, + "grad_norm": 3.5451438426971436, + "learning_rate": 1.010147601476015e-06, + "loss": 1.3259, + "step": 1987 + }, + { + "epoch": 2.4741754822650903, + "grad_norm": 3.5002248287200928, + "learning_rate": 1.0078413284132842e-06, + "loss": 1.3452, + "step": 1988 + }, + { + "epoch": 2.475420037336652, + "grad_norm": 3.3579916954040527, + "learning_rate": 1.0055350553505536e-06, + "loss": 1.2945, + "step": 1989 + }, + { + "epoch": 2.476664592408214, + "grad_norm": 3.612515687942505, + "learning_rate": 1.003228782287823e-06, + "loss": 1.3483, + "step": 1990 + }, + { + "epoch": 2.476664592408214, + "eval_loss": 1.6154413223266602, + "eval_runtime": 45.8894, + "eval_samples_per_second": 21.792, + "eval_steps_per_second": 0.915, + "step": 1990 + }, + { + "epoch": 2.477909147479776, + "grad_norm": 3.379666805267334, + "learning_rate": 1.0009225092250923e-06, + "loss": 1.329, + "step": 1991 + }, + { + "epoch": 2.4791537025513377, + "grad_norm": 3.4102606773376465, + "learning_rate": 9.986162361623617e-07, + "loss": 1.3485, + "step": 1992 + }, + { + "epoch": 2.4803982576229, + "grad_norm": 3.489799737930298, + "learning_rate": 9.963099630996311e-07, + "loss": 1.3735, + "step": 1993 + }, + { + "epoch": 2.481642812694462, + "grad_norm": 3.348045825958252, + "learning_rate": 9.940036900369005e-07, + "loss": 1.3586, + "step": 1994 + }, + { + "epoch": 2.4828873677660237, + "grad_norm": 3.4739747047424316, + "learning_rate": 9.916974169741698e-07, + "loss": 1.3584, + "step": 1995 + }, + { + "epoch": 2.4841319228375855, + "grad_norm": 3.338620185852051, + "learning_rate": 9.893911439114392e-07, + "loss": 1.3364, + "step": 1996 + }, + { + "epoch": 2.4853764779091474, + "grad_norm": 3.3837015628814697, + "learning_rate": 9.870848708487086e-07, + "loss": 1.3663, + "step": 1997 + }, + { + "epoch": 2.4866210329807092, + "grad_norm": 3.451869010925293, + "learning_rate": 9.84778597785978e-07, + "loss": 1.349, + "step": 1998 + }, + { + "epoch": 2.487865588052271, + "grad_norm": 3.2855687141418457, + "learning_rate": 9.824723247232472e-07, + "loss": 1.3403, + "step": 1999 + }, + { + "epoch": 2.4891101431238334, + "grad_norm": 3.3178985118865967, + "learning_rate": 9.801660516605167e-07, + "loss": 1.3468, + "step": 2000 + }, + { + "epoch": 2.4891101431238334, + "eval_loss": 1.6164495944976807, + "eval_runtime": 45.108, + "eval_samples_per_second": 22.169, + "eval_steps_per_second": 0.931, + "step": 2000 + }, + { + "epoch": 2.4903546981953952, + "grad_norm": 3.420365333557129, + "learning_rate": 9.77859778597786e-07, + "loss": 1.3748, + "step": 2001 + }, + { + "epoch": 2.491599253266957, + "grad_norm": 3.394939422607422, + "learning_rate": 9.755535055350555e-07, + "loss": 1.3521, + "step": 2002 + }, + { + "epoch": 2.492843808338519, + "grad_norm": 3.3312697410583496, + "learning_rate": 9.732472324723247e-07, + "loss": 1.2752, + "step": 2003 + }, + { + "epoch": 2.4940883634100808, + "grad_norm": 3.345182418823242, + "learning_rate": 9.709409594095941e-07, + "loss": 1.3396, + "step": 2004 + }, + { + "epoch": 2.4953329184816426, + "grad_norm": 3.3405473232269287, + "learning_rate": 9.686346863468636e-07, + "loss": 1.3409, + "step": 2005 + }, + { + "epoch": 2.496577473553205, + "grad_norm": 3.3915798664093018, + "learning_rate": 9.66328413284133e-07, + "loss": 1.3044, + "step": 2006 + }, + { + "epoch": 2.4978220286247668, + "grad_norm": 3.439868927001953, + "learning_rate": 9.640221402214022e-07, + "loss": 1.35, + "step": 2007 + }, + { + "epoch": 2.4990665836963286, + "grad_norm": 3.291309118270874, + "learning_rate": 9.617158671586716e-07, + "loss": 1.3315, + "step": 2008 + }, + { + "epoch": 2.5003111387678905, + "grad_norm": 3.3656320571899414, + "learning_rate": 9.59409594095941e-07, + "loss": 1.3223, + "step": 2009 + }, + { + "epoch": 2.5015556938394523, + "grad_norm": 3.234790563583374, + "learning_rate": 9.571033210332105e-07, + "loss": 1.2996, + "step": 2010 + }, + { + "epoch": 2.5015556938394523, + "eval_loss": 1.6163933277130127, + "eval_runtime": 43.9451, + "eval_samples_per_second": 22.756, + "eval_steps_per_second": 0.956, + "step": 2010 + }, + { + "epoch": 2.502800248911014, + "grad_norm": 3.535348892211914, + "learning_rate": 9.547970479704797e-07, + "loss": 1.3793, + "step": 2011 + }, + { + "epoch": 2.504044803982576, + "grad_norm": 3.222407341003418, + "learning_rate": 9.524907749077492e-07, + "loss": 1.3345, + "step": 2012 + }, + { + "epoch": 2.5052893590541383, + "grad_norm": 3.373684883117676, + "learning_rate": 9.501845018450185e-07, + "loss": 1.3294, + "step": 2013 + }, + { + "epoch": 2.5065339141257, + "grad_norm": 3.3373374938964844, + "learning_rate": 9.478782287822879e-07, + "loss": 1.3337, + "step": 2014 + }, + { + "epoch": 2.507778469197262, + "grad_norm": 3.562657117843628, + "learning_rate": 9.455719557195573e-07, + "loss": 1.3335, + "step": 2015 + }, + { + "epoch": 2.509023024268824, + "grad_norm": 3.4051525592803955, + "learning_rate": 9.432656826568267e-07, + "loss": 1.3165, + "step": 2016 + }, + { + "epoch": 2.5102675793403857, + "grad_norm": 3.563354015350342, + "learning_rate": 9.40959409594096e-07, + "loss": 1.3749, + "step": 2017 + }, + { + "epoch": 2.511512134411948, + "grad_norm": 3.3118226528167725, + "learning_rate": 9.386531365313654e-07, + "loss": 1.3334, + "step": 2018 + }, + { + "epoch": 2.51275668948351, + "grad_norm": 3.393760919570923, + "learning_rate": 9.363468634686347e-07, + "loss": 1.3827, + "step": 2019 + }, + { + "epoch": 2.5140012445550717, + "grad_norm": 3.412921190261841, + "learning_rate": 9.340405904059042e-07, + "loss": 1.3591, + "step": 2020 + }, + { + "epoch": 2.5140012445550717, + "eval_loss": 1.6178748607635498, + "eval_runtime": 45.527, + "eval_samples_per_second": 21.965, + "eval_steps_per_second": 0.923, + "step": 2020 + }, + { + "epoch": 2.5152457996266335, + "grad_norm": 3.4539453983306885, + "learning_rate": 9.317343173431735e-07, + "loss": 1.3402, + "step": 2021 + }, + { + "epoch": 2.5164903546981954, + "grad_norm": 3.3637566566467285, + "learning_rate": 9.294280442804429e-07, + "loss": 1.3481, + "step": 2022 + }, + { + "epoch": 2.5177349097697572, + "grad_norm": 3.3450021743774414, + "learning_rate": 9.271217712177122e-07, + "loss": 1.3182, + "step": 2023 + }, + { + "epoch": 2.518979464841319, + "grad_norm": 3.393338918685913, + "learning_rate": 9.248154981549816e-07, + "loss": 1.3166, + "step": 2024 + }, + { + "epoch": 2.520224019912881, + "grad_norm": 3.3550498485565186, + "learning_rate": 9.22509225092251e-07, + "loss": 1.3194, + "step": 2025 + }, + { + "epoch": 2.521468574984443, + "grad_norm": 3.275648355484009, + "learning_rate": 9.202029520295204e-07, + "loss": 1.3548, + "step": 2026 + }, + { + "epoch": 2.522713130056005, + "grad_norm": 3.312330961227417, + "learning_rate": 9.178966789667898e-07, + "loss": 1.3163, + "step": 2027 + }, + { + "epoch": 2.523957685127567, + "grad_norm": 3.2279152870178223, + "learning_rate": 9.155904059040591e-07, + "loss": 1.3056, + "step": 2028 + }, + { + "epoch": 2.5252022401991288, + "grad_norm": 3.480659008026123, + "learning_rate": 9.132841328413284e-07, + "loss": 1.3606, + "step": 2029 + }, + { + "epoch": 2.5264467952706906, + "grad_norm": 3.5396623611450195, + "learning_rate": 9.109778597785979e-07, + "loss": 1.3762, + "step": 2030 + }, + { + "epoch": 2.5264467952706906, + "eval_loss": 1.619415044784546, + "eval_runtime": 46.552, + "eval_samples_per_second": 21.481, + "eval_steps_per_second": 0.902, + "step": 2030 + }, + { + "epoch": 2.527691350342253, + "grad_norm": 3.454069137573242, + "learning_rate": 9.086715867158673e-07, + "loss": 1.3446, + "step": 2031 + }, + { + "epoch": 2.5289359054138147, + "grad_norm": 3.436459541320801, + "learning_rate": 9.063653136531365e-07, + "loss": 1.3413, + "step": 2032 + }, + { + "epoch": 2.5301804604853766, + "grad_norm": 3.624943256378174, + "learning_rate": 9.040590405904059e-07, + "loss": 1.3622, + "step": 2033 + }, + { + "epoch": 2.5314250155569384, + "grad_norm": 3.3480710983276367, + "learning_rate": 9.017527675276753e-07, + "loss": 1.2926, + "step": 2034 + }, + { + "epoch": 2.5326695706285003, + "grad_norm": 3.4130468368530273, + "learning_rate": 8.994464944649448e-07, + "loss": 1.33, + "step": 2035 + }, + { + "epoch": 2.533914125700062, + "grad_norm": 3.4406704902648926, + "learning_rate": 8.971402214022142e-07, + "loss": 1.3599, + "step": 2036 + }, + { + "epoch": 2.535158680771624, + "grad_norm": 3.3342652320861816, + "learning_rate": 8.948339483394834e-07, + "loss": 1.3177, + "step": 2037 + }, + { + "epoch": 2.536403235843186, + "grad_norm": 3.417802095413208, + "learning_rate": 8.925276752767528e-07, + "loss": 1.3592, + "step": 2038 + }, + { + "epoch": 2.537647790914748, + "grad_norm": 3.539034366607666, + "learning_rate": 8.902214022140222e-07, + "loss": 1.3806, + "step": 2039 + }, + { + "epoch": 2.53889234598631, + "grad_norm": 3.2618255615234375, + "learning_rate": 8.879151291512917e-07, + "loss": 1.2727, + "step": 2040 + }, + { + "epoch": 2.53889234598631, + "eval_loss": 1.61531400680542, + "eval_runtime": 47.2592, + "eval_samples_per_second": 21.16, + "eval_steps_per_second": 0.889, + "step": 2040 + }, + { + "epoch": 2.540136901057872, + "grad_norm": 3.318431854248047, + "learning_rate": 8.856088560885609e-07, + "loss": 1.3278, + "step": 2041 + }, + { + "epoch": 2.5413814561294337, + "grad_norm": 3.2816710472106934, + "learning_rate": 8.833025830258303e-07, + "loss": 1.3247, + "step": 2042 + }, + { + "epoch": 2.5426260112009955, + "grad_norm": 3.5672521591186523, + "learning_rate": 8.809963099630997e-07, + "loss": 1.3768, + "step": 2043 + }, + { + "epoch": 2.543870566272558, + "grad_norm": 3.330008029937744, + "learning_rate": 8.786900369003691e-07, + "loss": 1.3331, + "step": 2044 + }, + { + "epoch": 2.5451151213441197, + "grad_norm": 3.709188461303711, + "learning_rate": 8.763837638376385e-07, + "loss": 1.413, + "step": 2045 + }, + { + "epoch": 2.5463596764156815, + "grad_norm": 3.4795358180999756, + "learning_rate": 8.740774907749078e-07, + "loss": 1.3068, + "step": 2046 + }, + { + "epoch": 2.5476042314872434, + "grad_norm": 3.4811947345733643, + "learning_rate": 8.717712177121772e-07, + "loss": 1.3213, + "step": 2047 + }, + { + "epoch": 2.548848786558805, + "grad_norm": 3.308023452758789, + "learning_rate": 8.694649446494466e-07, + "loss": 1.3287, + "step": 2048 + }, + { + "epoch": 2.550093341630367, + "grad_norm": 3.392299175262451, + "learning_rate": 8.671586715867159e-07, + "loss": 1.3232, + "step": 2049 + }, + { + "epoch": 2.551337896701929, + "grad_norm": 3.4251368045806885, + "learning_rate": 8.648523985239853e-07, + "loss": 1.3348, + "step": 2050 + }, + { + "epoch": 2.551337896701929, + "eval_loss": 1.6107450723648071, + "eval_runtime": 48.8497, + "eval_samples_per_second": 20.471, + "eval_steps_per_second": 0.86, + "step": 2050 + }, + { + "epoch": 2.5525824517734907, + "grad_norm": 3.386932849884033, + "learning_rate": 8.625461254612547e-07, + "loss": 1.3432, + "step": 2051 + }, + { + "epoch": 2.553827006845053, + "grad_norm": 3.4637904167175293, + "learning_rate": 8.602398523985241e-07, + "loss": 1.3677, + "step": 2052 + }, + { + "epoch": 2.555071561916615, + "grad_norm": 3.3859682083129883, + "learning_rate": 8.579335793357934e-07, + "loss": 1.3336, + "step": 2053 + }, + { + "epoch": 2.5563161169881767, + "grad_norm": 3.248939037322998, + "learning_rate": 8.556273062730628e-07, + "loss": 1.2979, + "step": 2054 + }, + { + "epoch": 2.5575606720597386, + "grad_norm": 3.315685749053955, + "learning_rate": 8.533210332103322e-07, + "loss": 1.3156, + "step": 2055 + }, + { + "epoch": 2.5588052271313004, + "grad_norm": 3.356520175933838, + "learning_rate": 8.510147601476016e-07, + "loss": 1.2986, + "step": 2056 + }, + { + "epoch": 2.5600497822028627, + "grad_norm": 3.4709551334381104, + "learning_rate": 8.487084870848709e-07, + "loss": 1.3833, + "step": 2057 + }, + { + "epoch": 2.5612943372744246, + "grad_norm": 3.326077461242676, + "learning_rate": 8.464022140221403e-07, + "loss": 1.3558, + "step": 2058 + }, + { + "epoch": 2.5625388923459864, + "grad_norm": 3.427700996398926, + "learning_rate": 8.440959409594096e-07, + "loss": 1.3364, + "step": 2059 + }, + { + "epoch": 2.5637834474175483, + "grad_norm": 3.313568592071533, + "learning_rate": 8.417896678966791e-07, + "loss": 1.3384, + "step": 2060 + }, + { + "epoch": 2.5637834474175483, + "eval_loss": 1.6162501573562622, + "eval_runtime": 52.8784, + "eval_samples_per_second": 18.911, + "eval_steps_per_second": 0.794, + "step": 2060 + }, + { + "epoch": 2.56502800248911, + "grad_norm": 3.400526762008667, + "learning_rate": 8.394833948339484e-07, + "loss": 1.3772, + "step": 2061 + }, + { + "epoch": 2.566272557560672, + "grad_norm": 3.547825574874878, + "learning_rate": 8.371771217712178e-07, + "loss": 1.3651, + "step": 2062 + }, + { + "epoch": 2.567517112632234, + "grad_norm": 3.6061952114105225, + "learning_rate": 8.348708487084871e-07, + "loss": 1.3562, + "step": 2063 + }, + { + "epoch": 2.5687616677037957, + "grad_norm": 3.557246685028076, + "learning_rate": 8.325645756457565e-07, + "loss": 1.3828, + "step": 2064 + }, + { + "epoch": 2.570006222775358, + "grad_norm": 3.4955503940582275, + "learning_rate": 8.302583025830259e-07, + "loss": 1.3291, + "step": 2065 + }, + { + "epoch": 2.57125077784692, + "grad_norm": 3.332042694091797, + "learning_rate": 8.279520295202953e-07, + "loss": 1.3533, + "step": 2066 + }, + { + "epoch": 2.5724953329184816, + "grad_norm": 3.5139050483703613, + "learning_rate": 8.256457564575647e-07, + "loss": 1.3567, + "step": 2067 + }, + { + "epoch": 2.5737398879900435, + "grad_norm": 3.475031852722168, + "learning_rate": 8.233394833948339e-07, + "loss": 1.3449, + "step": 2068 + }, + { + "epoch": 2.5749844430616053, + "grad_norm": 3.3193306922912598, + "learning_rate": 8.210332103321033e-07, + "loss": 1.3102, + "step": 2069 + }, + { + "epoch": 2.5762289981331676, + "grad_norm": 3.391435146331787, + "learning_rate": 8.187269372693728e-07, + "loss": 1.3525, + "step": 2070 + }, + { + "epoch": 2.5762289981331676, + "eval_loss": 1.6134655475616455, + "eval_runtime": 50.1685, + "eval_samples_per_second": 19.933, + "eval_steps_per_second": 0.837, + "step": 2070 + }, + { + "epoch": 2.5774735532047295, + "grad_norm": 3.3784847259521484, + "learning_rate": 8.164206642066422e-07, + "loss": 1.3897, + "step": 2071 + }, + { + "epoch": 2.5787181082762913, + "grad_norm": 3.4751358032226562, + "learning_rate": 8.141143911439114e-07, + "loss": 1.3488, + "step": 2072 + }, + { + "epoch": 2.579962663347853, + "grad_norm": 3.350978374481201, + "learning_rate": 8.118081180811808e-07, + "loss": 1.3458, + "step": 2073 + }, + { + "epoch": 2.581207218419415, + "grad_norm": 3.382488489151001, + "learning_rate": 8.095018450184502e-07, + "loss": 1.3188, + "step": 2074 + }, + { + "epoch": 2.582451773490977, + "grad_norm": 3.4435741901397705, + "learning_rate": 8.071955719557197e-07, + "loss": 1.3324, + "step": 2075 + }, + { + "epoch": 2.5836963285625387, + "grad_norm": 3.4512386322021484, + "learning_rate": 8.048892988929891e-07, + "loss": 1.3445, + "step": 2076 + }, + { + "epoch": 2.5849408836341006, + "grad_norm": 3.386406898498535, + "learning_rate": 8.025830258302583e-07, + "loss": 1.3333, + "step": 2077 + }, + { + "epoch": 2.586185438705663, + "grad_norm": 3.3547282218933105, + "learning_rate": 8.002767527675277e-07, + "loss": 1.3542, + "step": 2078 + }, + { + "epoch": 2.5874299937772247, + "grad_norm": 3.3398079872131348, + "learning_rate": 7.979704797047971e-07, + "loss": 1.2978, + "step": 2079 + }, + { + "epoch": 2.5886745488487866, + "grad_norm": 3.40075945854187, + "learning_rate": 7.956642066420666e-07, + "loss": 1.3235, + "step": 2080 + }, + { + "epoch": 2.5886745488487866, + "eval_loss": 1.611613392829895, + "eval_runtime": 48.4495, + "eval_samples_per_second": 20.64, + "eval_steps_per_second": 0.867, + "step": 2080 + }, + { + "epoch": 2.5899191039203484, + "grad_norm": 3.4547722339630127, + "learning_rate": 7.933579335793358e-07, + "loss": 1.3781, + "step": 2081 + }, + { + "epoch": 2.5911636589919103, + "grad_norm": 3.2891974449157715, + "learning_rate": 7.910516605166052e-07, + "loss": 1.3045, + "step": 2082 + }, + { + "epoch": 2.5924082140634725, + "grad_norm": 3.493992567062378, + "learning_rate": 7.887453874538746e-07, + "loss": 1.3531, + "step": 2083 + }, + { + "epoch": 2.5936527691350344, + "grad_norm": 3.363905906677246, + "learning_rate": 7.86439114391144e-07, + "loss": 1.3385, + "step": 2084 + }, + { + "epoch": 2.5948973242065962, + "grad_norm": 3.404083728790283, + "learning_rate": 7.841328413284134e-07, + "loss": 1.3756, + "step": 2085 + }, + { + "epoch": 2.596141879278158, + "grad_norm": 3.4895224571228027, + "learning_rate": 7.818265682656827e-07, + "loss": 1.35, + "step": 2086 + }, + { + "epoch": 2.59738643434972, + "grad_norm": 3.2790639400482178, + "learning_rate": 7.795202952029521e-07, + "loss": 1.3013, + "step": 2087 + }, + { + "epoch": 2.598630989421282, + "grad_norm": 3.298156499862671, + "learning_rate": 7.772140221402215e-07, + "loss": 1.3273, + "step": 2088 + }, + { + "epoch": 2.5998755444928436, + "grad_norm": 3.3549964427948, + "learning_rate": 7.749077490774908e-07, + "loss": 1.3549, + "step": 2089 + }, + { + "epoch": 2.6011200995644055, + "grad_norm": 3.380481243133545, + "learning_rate": 7.726014760147602e-07, + "loss": 1.3742, + "step": 2090 + }, + { + "epoch": 2.6011200995644055, + "eval_loss": 1.614044427871704, + "eval_runtime": 55.012, + "eval_samples_per_second": 18.178, + "eval_steps_per_second": 0.763, + "step": 2090 + }, + { + "epoch": 2.6023646546359678, + "grad_norm": 3.41013503074646, + "learning_rate": 7.702952029520296e-07, + "loss": 1.3019, + "step": 2091 + }, + { + "epoch": 2.6036092097075296, + "grad_norm": 3.360945701599121, + "learning_rate": 7.67988929889299e-07, + "loss": 1.3389, + "step": 2092 + }, + { + "epoch": 2.6048537647790915, + "grad_norm": 3.456594228744507, + "learning_rate": 7.656826568265683e-07, + "loss": 1.3591, + "step": 2093 + }, + { + "epoch": 2.6060983198506533, + "grad_norm": 3.368643283843994, + "learning_rate": 7.633763837638377e-07, + "loss": 1.3299, + "step": 2094 + }, + { + "epoch": 2.607342874922215, + "grad_norm": 3.3424596786499023, + "learning_rate": 7.610701107011071e-07, + "loss": 1.2477, + "step": 2095 + }, + { + "epoch": 2.6085874299937775, + "grad_norm": 3.443603754043579, + "learning_rate": 7.587638376383765e-07, + "loss": 1.3302, + "step": 2096 + }, + { + "epoch": 2.6098319850653393, + "grad_norm": 3.503922462463379, + "learning_rate": 7.564575645756458e-07, + "loss": 1.3527, + "step": 2097 + }, + { + "epoch": 2.611076540136901, + "grad_norm": 3.4641854763031006, + "learning_rate": 7.541512915129152e-07, + "loss": 1.3394, + "step": 2098 + }, + { + "epoch": 2.612321095208463, + "grad_norm": 3.4873971939086914, + "learning_rate": 7.518450184501845e-07, + "loss": 1.3197, + "step": 2099 + }, + { + "epoch": 2.613565650280025, + "grad_norm": 3.3636622428894043, + "learning_rate": 7.49538745387454e-07, + "loss": 1.296, + "step": 2100 + }, + { + "epoch": 2.613565650280025, + "eval_loss": 1.6113073825836182, + "eval_runtime": 52.1267, + "eval_samples_per_second": 19.184, + "eval_steps_per_second": 0.806, + "step": 2100 + } + ], + "logging_steps": 1, + "max_steps": 2409, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 100, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 6.546349759660032e+17, + "train_batch_size": 3, + "trial_name": null, + "trial_params": null +} diff --git a/babylm_hop_control_10M_seed0/runs/checkpoint-2100/zero_to_fp32.py b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/babylm_hop_control_10M_seed0/runs/checkpoint-2100/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/babylm_reverse_partial_10M_seed0/artifacts/.locks/models--meta-llama--Llama-3.2-3B/2d73a6863086ff9d491c28e49df9fb697cd92c2b.lock b/babylm_reverse_partial_10M_seed0/artifacts/.locks/models--meta-llama--Llama-3.2-3B/2d73a6863086ff9d491c28e49df9fb697cd92c2b.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/babylm_reverse_partial_10M_seed0/artifacts/.locks/models--meta-llama--Llama-3.2-3B/4719a04514ec2f060240711b7c33ab21187cac730ecaba3040b7a0fd95a9cefb.lock b/babylm_reverse_partial_10M_seed0/artifacts/.locks/models--meta-llama--Llama-3.2-3B/4719a04514ec2f060240711b7c33ab21187cac730ecaba3040b7a0fd95a9cefb.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/babylm_reverse_partial_10M_seed0/artifacts/.locks/models--meta-llama--Llama-3.2-3B/47d4a5aa69cdef91a53b77f5c5583647a578ca0e.lock b/babylm_reverse_partial_10M_seed0/artifacts/.locks/models--meta-llama--Llama-3.2-3B/47d4a5aa69cdef91a53b77f5c5583647a578ca0e.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/babylm_reverse_partial_10M_seed0/artifacts/.locks/models--meta-llama--Llama-3.2-3B/584d8d3e3f82f7964955174dfe5e3b1cf117a9d859f022cfdf7fcb884856e002.lock b/babylm_reverse_partial_10M_seed0/artifacts/.locks/models--meta-llama--Llama-3.2-3B/584d8d3e3f82f7964955174dfe5e3b1cf117a9d859f022cfdf7fcb884856e002.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/babylm_reverse_partial_10M_seed0/artifacts/.locks/models--meta-llama--Llama-3.2-3B/d3a1f0f5f401eeadca0c7a6786bd9e877fd42e58.lock b/babylm_reverse_partial_10M_seed0/artifacts/.locks/models--meta-llama--Llama-3.2-3B/d3a1f0f5f401eeadca0c7a6786bd9e877fd42e58.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/.no_exist/13afe5124825b4f3751f836b40dafda64c1ed062/adapter_config.json b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/.no_exist/13afe5124825b4f3751f836b40dafda64c1ed062/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/.no_exist/13afe5124825b4f3751f836b40dafda64c1ed062/model.safetensors b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/.no_exist/13afe5124825b4f3751f836b40dafda64c1ed062/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/blobs/2d73a6863086ff9d491c28e49df9fb697cd92c2b b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/blobs/2d73a6863086ff9d491c28e49df9fb697cd92c2b new file mode 100644 index 0000000000000000000000000000000000000000..2d73a6863086ff9d491c28e49df9fb697cd92c2b --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/blobs/2d73a6863086ff9d491c28e49df9fb697cd92c2b @@ -0,0 +1,9 @@ +{ + "_from_model_config": true, + "bos_token_id": 128000, + "eos_token_id": 128001, + "transformers_version": "4.45.0.dev0", + "do_sample": true, + "temperature": 0.6, + "top_p": 0.9 +} diff --git a/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/blobs/47d4a5aa69cdef91a53b77f5c5583647a578ca0e b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/blobs/47d4a5aa69cdef91a53b77f5c5583647a578ca0e new file mode 100644 index 0000000000000000000000000000000000000000..47d4a5aa69cdef91a53b77f5c5583647a578ca0e --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/blobs/47d4a5aa69cdef91a53b77f5c5583647a578ca0e @@ -0,0 +1,35 @@ +{ + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "eos_token_id": 128001, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 3072, + "initializer_range": 0.02, + "intermediate_size": 8192, + "max_position_embeddings": 131072, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 24, + "num_hidden_layers": 28, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": { + "factor": 32.0, + "high_freq_factor": 4.0, + "low_freq_factor": 1.0, + "original_max_position_embeddings": 8192, + "rope_type": "llama3" + }, + "rope_theta": 500000.0, + "tie_word_embeddings": true, + "torch_dtype": "bfloat16", + "transformers_version": "4.45.0.dev0", + "use_cache": true, + "vocab_size": 128256 +} diff --git a/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/blobs/d3a1f0f5f401eeadca0c7a6786bd9e877fd42e58 b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/blobs/d3a1f0f5f401eeadca0c7a6786bd9e877fd42e58 new file mode 100644 index 0000000000000000000000000000000000000000..d3a1f0f5f401eeadca0c7a6786bd9e877fd42e58 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/blobs/d3a1f0f5f401eeadca0c7a6786bd9e877fd42e58 @@ -0,0 +1,261 @@ +{ + "metadata": { + "total_size": 6425499648 + }, + "weight_map": { + "model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.norm.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/refs/main b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/refs/main new file mode 100644 index 0000000000000000000000000000000000000000..940c46e96db631910d2cc5186ab76cd83e32b124 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/refs/main @@ -0,0 +1 @@ +13afe5124825b4f3751f836b40dafda64c1ed062 \ No newline at end of file diff --git a/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/snapshots/13afe5124825b4f3751f836b40dafda64c1ed062/config.json b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/snapshots/13afe5124825b4f3751f836b40dafda64c1ed062/config.json new file mode 100644 index 0000000000000000000000000000000000000000..47d4a5aa69cdef91a53b77f5c5583647a578ca0e --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/snapshots/13afe5124825b4f3751f836b40dafda64c1ed062/config.json @@ -0,0 +1,35 @@ +{ + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "eos_token_id": 128001, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 3072, + "initializer_range": 0.02, + "intermediate_size": 8192, + "max_position_embeddings": 131072, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 24, + "num_hidden_layers": 28, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": { + "factor": 32.0, + "high_freq_factor": 4.0, + "low_freq_factor": 1.0, + "original_max_position_embeddings": 8192, + "rope_type": "llama3" + }, + "rope_theta": 500000.0, + "tie_word_embeddings": true, + "torch_dtype": "bfloat16", + "transformers_version": "4.45.0.dev0", + "use_cache": true, + "vocab_size": 128256 +} diff --git a/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/snapshots/13afe5124825b4f3751f836b40dafda64c1ed062/generation_config.json b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/snapshots/13afe5124825b4f3751f836b40dafda64c1ed062/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..2d73a6863086ff9d491c28e49df9fb697cd92c2b --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/snapshots/13afe5124825b4f3751f836b40dafda64c1ed062/generation_config.json @@ -0,0 +1,9 @@ +{ + "_from_model_config": true, + "bos_token_id": 128000, + "eos_token_id": 128001, + "transformers_version": "4.45.0.dev0", + "do_sample": true, + "temperature": 0.6, + "top_p": 0.9 +} diff --git a/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/snapshots/13afe5124825b4f3751f836b40dafda64c1ed062/model.safetensors.index.json b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/snapshots/13afe5124825b4f3751f836b40dafda64c1ed062/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..d3a1f0f5f401eeadca0c7a6786bd9e877fd42e58 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/artifacts/models--meta-llama--Llama-3.2-3B/snapshots/13afe5124825b4f3751f836b40dafda64c1ed062/model.safetensors.index.json @@ -0,0 +1,261 @@ +{ + "metadata": { + "total_size": 6425499648 + }, + "weight_map": { + "model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.norm.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fb0f9a53a507ffcdb60410deb71d60ef801bf350 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/config.json @@ -0,0 +1,36 @@ +{ + "_name_or_path": "meta-llama/Llama-3.2-3B", + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "eos_token_id": 128001, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 3072, + "initializer_range": 0.02, + "intermediate_size": 8192, + "max_position_embeddings": 131072, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 24, + "num_hidden_layers": 28, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": { + "factor": 32.0, + "high_freq_factor": 4.0, + "low_freq_factor": 1.0, + "original_max_position_embeddings": 8192, + "rope_type": "llama3" + }, + "rope_theta": 500000.0, + "tie_word_embeddings": true, + "torch_dtype": "float16", + "transformers_version": "4.45.1", + "use_cache": true, + "vocab_size": 128256 +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/generation_config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..9c389b841a9a29ddd904e02d1eb0e08dcce82ad9 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/generation_config.json @@ -0,0 +1,9 @@ +{ + "_from_model_config": true, + "bos_token_id": 128000, + "do_sample": true, + "eos_token_id": 128001, + "temperature": 0.6, + "top_p": 0.9, + "transformers_version": "4.45.1" +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/latest b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/latest new file mode 100644 index 0000000000000000000000000000000000000000..b22cbc7e055351e7aa89a5e0ea1fae3d6a0f0087 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/latest @@ -0,0 +1 @@ +global_step1800 \ No newline at end of file diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/model.safetensors.index.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..ed64de846d720b9a7859dc20575fea8e8ca51940 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/model.safetensors.index.json @@ -0,0 +1,262 @@ +{ + "metadata": { + "total_size": 7213504512 + }, + "weight_map": { + "lm_head.weight": "model-00002-of-00002.safetensors", + "model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.norm.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/special_tokens_map.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..6949c5975ee0e961ef61cf31010dce04df0a03f8 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/special_tokens_map.json @@ -0,0 +1,23 @@ +{ + "bos_token": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/tokenizer_config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..0b782d4602b361f7577abb52b710a8c0e088588d --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/tokenizer_config.json @@ -0,0 +1,2078 @@ +{ + "added_tokens_decoder": { + "128000": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128001": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128002": { + "content": "<|reserved_special_token_0|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128003": { + "content": "<|reserved_special_token_1|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128004": { + "content": "<|finetune_right_pad_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128005": { + "content": "<|reserved_special_token_2|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128006": { + "content": "<|start_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128007": { + "content": "<|end_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128008": { + "content": "<|eom_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128009": { + "content": "<|eot_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128010": { + "content": "<|python_tag|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128011": { + "content": "<|reserved_special_token_3|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128012": { + "content": "<|reserved_special_token_4|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128013": { + "content": "<|reserved_special_token_5|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128014": { + "content": "<|reserved_special_token_6|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128015": { + "content": "<|reserved_special_token_7|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128016": { + "content": "<|reserved_special_token_8|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128017": { + "content": "<|reserved_special_token_9|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128018": { + "content": "<|reserved_special_token_10|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128019": { + "content": "<|reserved_special_token_11|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128020": { + "content": "<|reserved_special_token_12|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128021": { + "content": "<|reserved_special_token_13|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128022": { + "content": "<|reserved_special_token_14|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128023": { + "content": "<|reserved_special_token_15|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128024": { + "content": "<|reserved_special_token_16|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128025": { + "content": "<|reserved_special_token_17|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128026": { + "content": "<|reserved_special_token_18|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128027": { + "content": "<|reserved_special_token_19|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128028": { + "content": "<|reserved_special_token_20|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128029": { + "content": "<|reserved_special_token_21|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128030": { + "content": "<|reserved_special_token_22|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128031": { + "content": "<|reserved_special_token_23|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128032": { + "content": "<|reserved_special_token_24|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128033": { + "content": "<|reserved_special_token_25|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128034": { + "content": "<|reserved_special_token_26|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128035": { + "content": "<|reserved_special_token_27|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128036": { + "content": "<|reserved_special_token_28|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128037": { + "content": "<|reserved_special_token_29|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128038": { + "content": "<|reserved_special_token_30|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128039": { + "content": "<|reserved_special_token_31|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128040": { + "content": "<|reserved_special_token_32|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128041": { + "content": "<|reserved_special_token_33|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128042": { + "content": "<|reserved_special_token_34|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128043": { + "content": "<|reserved_special_token_35|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128044": { + "content": "<|reserved_special_token_36|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128045": { + "content": "<|reserved_special_token_37|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128046": { + "content": "<|reserved_special_token_38|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128047": { + "content": "<|reserved_special_token_39|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128048": { + "content": "<|reserved_special_token_40|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128049": { + "content": "<|reserved_special_token_41|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128050": { + "content": "<|reserved_special_token_42|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128051": { + "content": "<|reserved_special_token_43|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128052": { + "content": "<|reserved_special_token_44|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128053": { + "content": "<|reserved_special_token_45|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128054": { + "content": "<|reserved_special_token_46|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128055": { + "content": "<|reserved_special_token_47|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128056": { + "content": "<|reserved_special_token_48|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128057": { + "content": "<|reserved_special_token_49|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128058": { + "content": "<|reserved_special_token_50|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128059": { + "content": "<|reserved_special_token_51|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128060": { + "content": "<|reserved_special_token_52|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128061": { + "content": "<|reserved_special_token_53|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128062": { + "content": "<|reserved_special_token_54|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128063": { + "content": "<|reserved_special_token_55|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128064": { + "content": "<|reserved_special_token_56|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128065": { + "content": "<|reserved_special_token_57|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128066": { + "content": "<|reserved_special_token_58|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128067": { + "content": "<|reserved_special_token_59|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128068": { + "content": "<|reserved_special_token_60|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128069": { + "content": "<|reserved_special_token_61|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128070": { + "content": "<|reserved_special_token_62|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128071": { + "content": "<|reserved_special_token_63|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128072": { + "content": "<|reserved_special_token_64|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128073": { + "content": "<|reserved_special_token_65|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128074": { + "content": "<|reserved_special_token_66|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128075": { + "content": "<|reserved_special_token_67|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128076": { + "content": "<|reserved_special_token_68|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128077": { + "content": "<|reserved_special_token_69|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128078": { + "content": "<|reserved_special_token_70|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128079": { + "content": "<|reserved_special_token_71|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128080": { + "content": "<|reserved_special_token_72|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128081": { + "content": "<|reserved_special_token_73|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128082": { + "content": "<|reserved_special_token_74|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128083": { + "content": "<|reserved_special_token_75|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128084": { + "content": "<|reserved_special_token_76|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128085": { + "content": "<|reserved_special_token_77|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128086": { + "content": "<|reserved_special_token_78|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128087": { + "content": "<|reserved_special_token_79|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128088": { + "content": "<|reserved_special_token_80|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128089": { + "content": "<|reserved_special_token_81|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128090": { + "content": "<|reserved_special_token_82|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128091": { + "content": "<|reserved_special_token_83|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128092": { + "content": "<|reserved_special_token_84|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128093": { + "content": "<|reserved_special_token_85|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128094": { + "content": "<|reserved_special_token_86|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128095": { + "content": "<|reserved_special_token_87|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128096": { + "content": "<|reserved_special_token_88|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128097": { + "content": "<|reserved_special_token_89|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128098": { + "content": "<|reserved_special_token_90|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128099": { + "content": "<|reserved_special_token_91|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128100": { + "content": "<|reserved_special_token_92|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128101": { + "content": "<|reserved_special_token_93|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128102": { + "content": "<|reserved_special_token_94|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128103": { + "content": "<|reserved_special_token_95|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128104": { + "content": "<|reserved_special_token_96|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128105": { + "content": "<|reserved_special_token_97|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128106": { + "content": "<|reserved_special_token_98|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128107": { + "content": "<|reserved_special_token_99|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128108": { + "content": "<|reserved_special_token_100|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128109": { + "content": "<|reserved_special_token_101|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128110": { + "content": "<|reserved_special_token_102|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128111": { + "content": "<|reserved_special_token_103|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128112": { + "content": "<|reserved_special_token_104|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128113": { + "content": "<|reserved_special_token_105|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128114": { + "content": "<|reserved_special_token_106|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128115": { + "content": "<|reserved_special_token_107|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128116": { + "content": "<|reserved_special_token_108|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128117": { + "content": "<|reserved_special_token_109|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128118": { + "content": "<|reserved_special_token_110|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128119": { + "content": "<|reserved_special_token_111|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128120": { + "content": "<|reserved_special_token_112|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128121": { + "content": "<|reserved_special_token_113|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128122": { + "content": "<|reserved_special_token_114|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128123": { + "content": "<|reserved_special_token_115|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128124": { + "content": "<|reserved_special_token_116|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128125": { + "content": "<|reserved_special_token_117|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128126": { + "content": "<|reserved_special_token_118|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128127": { + "content": "<|reserved_special_token_119|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128128": { + "content": "<|reserved_special_token_120|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128129": { + "content": "<|reserved_special_token_121|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128130": { + "content": "<|reserved_special_token_122|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128131": { + "content": "<|reserved_special_token_123|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128132": { + "content": "<|reserved_special_token_124|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128133": { + "content": "<|reserved_special_token_125|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128134": { + "content": "<|reserved_special_token_126|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128135": { + "content": "<|reserved_special_token_127|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128136": { + "content": "<|reserved_special_token_128|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128137": { + "content": "<|reserved_special_token_129|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128138": { + "content": "<|reserved_special_token_130|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128139": { + "content": "<|reserved_special_token_131|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128140": { + "content": "<|reserved_special_token_132|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128141": { + "content": "<|reserved_special_token_133|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128142": { + "content": "<|reserved_special_token_134|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128143": { + "content": "<|reserved_special_token_135|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128144": { + "content": "<|reserved_special_token_136|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128145": { + "content": "<|reserved_special_token_137|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128146": { + "content": "<|reserved_special_token_138|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128147": { + "content": "<|reserved_special_token_139|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128148": { + "content": "<|reserved_special_token_140|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128149": { + "content": "<|reserved_special_token_141|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128150": { + "content": "<|reserved_special_token_142|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128151": { + "content": "<|reserved_special_token_143|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128152": { + "content": "<|reserved_special_token_144|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128153": { + "content": "<|reserved_special_token_145|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128154": { + "content": "<|reserved_special_token_146|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128155": { + "content": "<|reserved_special_token_147|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128156": { + "content": "<|reserved_special_token_148|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128157": { + "content": "<|reserved_special_token_149|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128158": { + "content": "<|reserved_special_token_150|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128159": { + "content": "<|reserved_special_token_151|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128160": { + "content": "<|reserved_special_token_152|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128161": { + "content": "<|reserved_special_token_153|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128162": { + "content": "<|reserved_special_token_154|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128163": { + "content": "<|reserved_special_token_155|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128164": { + "content": "<|reserved_special_token_156|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128165": { + "content": "<|reserved_special_token_157|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128166": { + "content": "<|reserved_special_token_158|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128167": { + "content": "<|reserved_special_token_159|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128168": { + "content": "<|reserved_special_token_160|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128169": { + "content": "<|reserved_special_token_161|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128170": { + "content": "<|reserved_special_token_162|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128171": { + "content": "<|reserved_special_token_163|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128172": { + "content": "<|reserved_special_token_164|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128173": { + "content": "<|reserved_special_token_165|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128174": { + "content": "<|reserved_special_token_166|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128175": { + "content": "<|reserved_special_token_167|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128176": { + "content": "<|reserved_special_token_168|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128177": { + "content": "<|reserved_special_token_169|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128178": { + "content": "<|reserved_special_token_170|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128179": { + "content": "<|reserved_special_token_171|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128180": { + "content": "<|reserved_special_token_172|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128181": { + "content": "<|reserved_special_token_173|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128182": { + "content": "<|reserved_special_token_174|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128183": { + "content": "<|reserved_special_token_175|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128184": { + "content": "<|reserved_special_token_176|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128185": { + "content": "<|reserved_special_token_177|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128186": { + "content": "<|reserved_special_token_178|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128187": { + "content": "<|reserved_special_token_179|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128188": { + "content": "<|reserved_special_token_180|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128189": { + "content": "<|reserved_special_token_181|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128190": { + "content": "<|reserved_special_token_182|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128191": { + "content": "<|reserved_special_token_183|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128192": { + "content": "<|reserved_special_token_184|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128193": { + "content": "<|reserved_special_token_185|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128194": { + "content": "<|reserved_special_token_186|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128195": { + "content": "<|reserved_special_token_187|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128196": { + "content": "<|reserved_special_token_188|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128197": { + "content": "<|reserved_special_token_189|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128198": { + "content": "<|reserved_special_token_190|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128199": { + "content": "<|reserved_special_token_191|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128200": { + "content": "<|reserved_special_token_192|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128201": { + "content": "<|reserved_special_token_193|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128202": { + "content": "<|reserved_special_token_194|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128203": { + "content": "<|reserved_special_token_195|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128204": { + "content": "<|reserved_special_token_196|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128205": { + "content": "<|reserved_special_token_197|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128206": { + "content": "<|reserved_special_token_198|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128207": { + "content": "<|reserved_special_token_199|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128208": { + "content": "<|reserved_special_token_200|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128209": { + "content": "<|reserved_special_token_201|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128210": { + "content": "<|reserved_special_token_202|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128211": { + "content": "<|reserved_special_token_203|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128212": { + "content": "<|reserved_special_token_204|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128213": { + "content": "<|reserved_special_token_205|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128214": { + "content": "<|reserved_special_token_206|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128215": { + "content": "<|reserved_special_token_207|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128216": { + "content": "<|reserved_special_token_208|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128217": { + "content": "<|reserved_special_token_209|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128218": { + "content": "<|reserved_special_token_210|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128219": { + "content": "<|reserved_special_token_211|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128220": { + "content": "<|reserved_special_token_212|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128221": { + "content": "<|reserved_special_token_213|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128222": { + "content": "<|reserved_special_token_214|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128223": { + "content": "<|reserved_special_token_215|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128224": { + "content": "<|reserved_special_token_216|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128225": { + "content": "<|reserved_special_token_217|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128226": { + "content": "<|reserved_special_token_218|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128227": { + "content": "<|reserved_special_token_219|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128228": { + "content": "<|reserved_special_token_220|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128229": { + "content": "<|reserved_special_token_221|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128230": { + "content": "<|reserved_special_token_222|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128231": { + "content": "<|reserved_special_token_223|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128232": { + "content": "<|reserved_special_token_224|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128233": { + "content": "<|reserved_special_token_225|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128234": { + "content": "<|reserved_special_token_226|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128235": { + "content": "<|reserved_special_token_227|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128236": { + "content": "<|reserved_special_token_228|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128237": { + "content": "<|reserved_special_token_229|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128238": { + "content": "<|reserved_special_token_230|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128239": { + "content": "<|reserved_special_token_231|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128240": { + "content": "<|reserved_special_token_232|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128241": { + "content": "<|reserved_special_token_233|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128242": { + "content": "<|reserved_special_token_234|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128243": { + "content": "<|reserved_special_token_235|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128244": { + "content": "<|reserved_special_token_236|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128245": { + "content": "<|reserved_special_token_237|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128246": { + "content": "<|reserved_special_token_238|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128247": { + "content": "<|reserved_special_token_239|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128248": { + "content": "<|reserved_special_token_240|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128249": { + "content": "<|reserved_special_token_241|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128250": { + "content": "<|reserved_special_token_242|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128251": { + "content": "<|reserved_special_token_243|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128252": { + "content": "<|reserved_special_token_244|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128253": { + "content": "<|reserved_special_token_245|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128254": { + "content": "<|reserved_special_token_246|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128255": { + "content": "<|reserved_special_token_247|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128256": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128257": { + "content": "🅁", + "lstrip": true, + "normalized": true, + "rstrip": false, + "single_word": false, + "special": false + } + }, + "bos_token": "<|begin_of_text|>", + "clean_up_tokenization_spaces": true, + "eos_token": "<|end_of_text|>", + "model_input_names": [ + "input_ids", + "attention_mask" + ], + "model_max_length": 131072, + "pad_token": "[PAD]", + "tokenizer_class": "PreTrainedTokenizerFast" +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/trainer_state.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..1f93777cc5e61037eb225d3dd13171bf60bc9862 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/trainer_state.json @@ -0,0 +1,14073 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.8489984591679507, + "eval_steps": 10, + "global_step": 1800, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.001027221366204417, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7873, + "step": 1 + }, + { + "epoch": 0.002054442732408834, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7904, + "step": 2 + }, + { + "epoch": 0.0030816640986132513, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8348, + "step": 3 + }, + { + "epoch": 0.004108885464817668, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8357, + "step": 4 + }, + { + "epoch": 0.005136106831022085, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8153, + "step": 5 + }, + { + "epoch": 0.0061633281972265025, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.825, + "step": 6 + }, + { + "epoch": 0.007190549563430919, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8002, + "step": 7 + }, + { + "epoch": 0.008217770929635337, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8413, + "step": 8 + }, + { + "epoch": 0.009244992295839754, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8157, + "step": 9 + }, + { + "epoch": 0.01027221366204417, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8249, + "step": 10 + }, + { + "epoch": 0.01027221366204417, + "eval_loss": 2.8518412113189697, + "eval_runtime": 43.212, + "eval_samples_per_second": 23.142, + "eval_steps_per_second": 0.972, + "step": 10 + }, + { + "epoch": 0.011299435028248588, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7829, + "step": 11 + }, + { + "epoch": 0.012326656394453005, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7819, + "step": 12 + }, + { + "epoch": 0.01335387776065742, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8355, + "step": 13 + }, + { + "epoch": 0.014381099126861838, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8138, + "step": 14 + }, + { + "epoch": 0.015408320493066256, + "grad_norm": 5.162191390991211, + "learning_rate": 1.7123287671232876e-08, + "loss": 2.8088, + "step": 15 + }, + { + "epoch": 0.016435541859270673, + "grad_norm": 5.5954084396362305, + "learning_rate": 3.424657534246575e-08, + "loss": 2.7883, + "step": 16 + }, + { + "epoch": 0.01746276322547509, + "grad_norm": 5.542337417602539, + "learning_rate": 5.136986301369863e-08, + "loss": 2.7625, + "step": 17 + }, + { + "epoch": 0.01848998459167951, + "grad_norm": 5.300518989562988, + "learning_rate": 6.84931506849315e-08, + "loss": 2.84, + "step": 18 + }, + { + "epoch": 0.019517205957883924, + "grad_norm": 5.6646881103515625, + "learning_rate": 8.561643835616439e-08, + "loss": 2.761, + "step": 19 + }, + { + "epoch": 0.02054442732408834, + "grad_norm": 5.375557899475098, + "learning_rate": 1.0273972602739726e-07, + "loss": 2.8215, + "step": 20 + }, + { + "epoch": 0.02054442732408834, + "eval_loss": 2.8507144451141357, + "eval_runtime": 45.1133, + "eval_samples_per_second": 22.166, + "eval_steps_per_second": 0.931, + "step": 20 + }, + { + "epoch": 0.02157164869029276, + "grad_norm": 5.1731977462768555, + "learning_rate": 1.1986301369863014e-07, + "loss": 2.8348, + "step": 21 + }, + { + "epoch": 0.022598870056497175, + "grad_norm": 5.370945930480957, + "learning_rate": 1.36986301369863e-07, + "loss": 2.822, + "step": 22 + }, + { + "epoch": 0.02362609142270159, + "grad_norm": 5.550389289855957, + "learning_rate": 1.541095890410959e-07, + "loss": 2.7904, + "step": 23 + }, + { + "epoch": 0.02465331278890601, + "grad_norm": 5.339347839355469, + "learning_rate": 1.7123287671232878e-07, + "loss": 2.8066, + "step": 24 + }, + { + "epoch": 0.025680534155110426, + "grad_norm": 5.2319722175598145, + "learning_rate": 1.8835616438356165e-07, + "loss": 2.7976, + "step": 25 + }, + { + "epoch": 0.02670775552131484, + "grad_norm": 5.207149982452393, + "learning_rate": 2.0547945205479452e-07, + "loss": 2.8162, + "step": 26 + }, + { + "epoch": 0.02773497688751926, + "grad_norm": 4.400204181671143, + "learning_rate": 2.226027397260274e-07, + "loss": 2.7808, + "step": 27 + }, + { + "epoch": 0.028762198253723677, + "grad_norm": 4.21675968170166, + "learning_rate": 2.397260273972603e-07, + "loss": 2.8044, + "step": 28 + }, + { + "epoch": 0.029789419619928096, + "grad_norm": 4.406878471374512, + "learning_rate": 2.568493150684932e-07, + "loss": 2.8162, + "step": 29 + }, + { + "epoch": 0.030816640986132512, + "grad_norm": 4.024835586547852, + "learning_rate": 2.73972602739726e-07, + "loss": 2.7994, + "step": 30 + }, + { + "epoch": 0.030816640986132512, + "eval_loss": 2.8124029636383057, + "eval_runtime": 43.1772, + "eval_samples_per_second": 23.16, + "eval_steps_per_second": 0.973, + "step": 30 + }, + { + "epoch": 0.03184386235233693, + "grad_norm": 3.5378901958465576, + "learning_rate": 2.910958904109589e-07, + "loss": 2.7637, + "step": 31 + }, + { + "epoch": 0.03287108371854135, + "grad_norm": 3.6289916038513184, + "learning_rate": 3.082191780821918e-07, + "loss": 2.8017, + "step": 32 + }, + { + "epoch": 0.03389830508474576, + "grad_norm": 3.6789698600769043, + "learning_rate": 3.2534246575342466e-07, + "loss": 2.7723, + "step": 33 + }, + { + "epoch": 0.03492552645095018, + "grad_norm": 3.491389751434326, + "learning_rate": 3.4246575342465755e-07, + "loss": 2.7528, + "step": 34 + }, + { + "epoch": 0.035952747817154594, + "grad_norm": 3.383080005645752, + "learning_rate": 3.595890410958904e-07, + "loss": 2.7418, + "step": 35 + }, + { + "epoch": 0.03697996918335902, + "grad_norm": 3.4823594093322754, + "learning_rate": 3.767123287671233e-07, + "loss": 2.6662, + "step": 36 + }, + { + "epoch": 0.03800719054956343, + "grad_norm": 3.2146108150482178, + "learning_rate": 3.938356164383562e-07, + "loss": 2.7659, + "step": 37 + }, + { + "epoch": 0.03903441191576785, + "grad_norm": 2.9549412727355957, + "learning_rate": 4.1095890410958903e-07, + "loss": 2.6838, + "step": 38 + }, + { + "epoch": 0.040061633281972264, + "grad_norm": 3.0744080543518066, + "learning_rate": 4.2808219178082193e-07, + "loss": 2.6879, + "step": 39 + }, + { + "epoch": 0.04108885464817668, + "grad_norm": 3.19840669631958, + "learning_rate": 4.452054794520548e-07, + "loss": 2.7443, + "step": 40 + }, + { + "epoch": 0.04108885464817668, + "eval_loss": 2.7469444274902344, + "eval_runtime": 46.4676, + "eval_samples_per_second": 21.52, + "eval_steps_per_second": 0.904, + "step": 40 + }, + { + "epoch": 0.042116076014381096, + "grad_norm": 3.178406000137329, + "learning_rate": 4.6232876712328767e-07, + "loss": 2.7036, + "step": 41 + }, + { + "epoch": 0.04314329738058552, + "grad_norm": 3.2032673358917236, + "learning_rate": 4.794520547945206e-07, + "loss": 2.6465, + "step": 42 + }, + { + "epoch": 0.044170518746789934, + "grad_norm": 3.1525962352752686, + "learning_rate": 4.965753424657534e-07, + "loss": 2.6996, + "step": 43 + }, + { + "epoch": 0.04519774011299435, + "grad_norm": 2.907524585723877, + "learning_rate": 5.136986301369864e-07, + "loss": 2.6581, + "step": 44 + }, + { + "epoch": 0.046224961479198766, + "grad_norm": 2.5885250568389893, + "learning_rate": 5.308219178082192e-07, + "loss": 2.6773, + "step": 45 + }, + { + "epoch": 0.04725218284540318, + "grad_norm": 2.569068431854248, + "learning_rate": 5.47945205479452e-07, + "loss": 2.6635, + "step": 46 + }, + { + "epoch": 0.048279404211607604, + "grad_norm": 2.5313518047332764, + "learning_rate": 5.65068493150685e-07, + "loss": 2.706, + "step": 47 + }, + { + "epoch": 0.04930662557781202, + "grad_norm": 2.615442991256714, + "learning_rate": 5.821917808219178e-07, + "loss": 2.638, + "step": 48 + }, + { + "epoch": 0.050333846944016436, + "grad_norm": 2.7420806884765625, + "learning_rate": 5.993150684931507e-07, + "loss": 2.6398, + "step": 49 + }, + { + "epoch": 0.05136106831022085, + "grad_norm": 2.5246975421905518, + "learning_rate": 6.164383561643836e-07, + "loss": 2.5916, + "step": 50 + }, + { + "epoch": 0.05136106831022085, + "eval_loss": 2.6723692417144775, + "eval_runtime": 46.7206, + "eval_samples_per_second": 21.404, + "eval_steps_per_second": 0.899, + "step": 50 + }, + { + "epoch": 0.05238828967642527, + "grad_norm": 2.390143632888794, + "learning_rate": 6.335616438356165e-07, + "loss": 2.5773, + "step": 51 + }, + { + "epoch": 0.05341551104262968, + "grad_norm": 2.0792484283447266, + "learning_rate": 6.506849315068493e-07, + "loss": 2.6249, + "step": 52 + }, + { + "epoch": 0.054442732408834106, + "grad_norm": 1.9995019435882568, + "learning_rate": 6.678082191780823e-07, + "loss": 2.5425, + "step": 53 + }, + { + "epoch": 0.05546995377503852, + "grad_norm": 1.985452651977539, + "learning_rate": 6.849315068493151e-07, + "loss": 2.5986, + "step": 54 + }, + { + "epoch": 0.05649717514124294, + "grad_norm": 2.10184907913208, + "learning_rate": 7.020547945205481e-07, + "loss": 2.5713, + "step": 55 + }, + { + "epoch": 0.05752439650744735, + "grad_norm": 2.0898985862731934, + "learning_rate": 7.191780821917808e-07, + "loss": 2.535, + "step": 56 + }, + { + "epoch": 0.05855161787365177, + "grad_norm": 2.0239579677581787, + "learning_rate": 7.363013698630137e-07, + "loss": 2.5667, + "step": 57 + }, + { + "epoch": 0.05957883923985619, + "grad_norm": 2.207080125808716, + "learning_rate": 7.534246575342466e-07, + "loss": 2.4999, + "step": 58 + }, + { + "epoch": 0.06060606060606061, + "grad_norm": 1.8541918992996216, + "learning_rate": 7.705479452054795e-07, + "loss": 2.5582, + "step": 59 + }, + { + "epoch": 0.061633281972265024, + "grad_norm": 1.8530476093292236, + "learning_rate": 7.876712328767124e-07, + "loss": 2.5535, + "step": 60 + }, + { + "epoch": 0.061633281972265024, + "eval_loss": 2.607963800430298, + "eval_runtime": 45.9754, + "eval_samples_per_second": 21.751, + "eval_steps_per_second": 0.914, + "step": 60 + }, + { + "epoch": 0.06266050333846944, + "grad_norm": 1.7978090047836304, + "learning_rate": 8.047945205479453e-07, + "loss": 2.53, + "step": 61 + }, + { + "epoch": 0.06368772470467386, + "grad_norm": 1.8173762559890747, + "learning_rate": 8.219178082191781e-07, + "loss": 2.5144, + "step": 62 + }, + { + "epoch": 0.06471494607087827, + "grad_norm": 2.0847485065460205, + "learning_rate": 8.39041095890411e-07, + "loss": 2.5525, + "step": 63 + }, + { + "epoch": 0.0657421674370827, + "grad_norm": 1.8691741228103638, + "learning_rate": 8.561643835616439e-07, + "loss": 2.5451, + "step": 64 + }, + { + "epoch": 0.0667693888032871, + "grad_norm": 2.0242726802825928, + "learning_rate": 8.732876712328768e-07, + "loss": 2.4145, + "step": 65 + }, + { + "epoch": 0.06779661016949153, + "grad_norm": 1.8268908262252808, + "learning_rate": 8.904109589041097e-07, + "loss": 2.5024, + "step": 66 + }, + { + "epoch": 0.06882383153569595, + "grad_norm": 1.8703057765960693, + "learning_rate": 9.075342465753426e-07, + "loss": 2.461, + "step": 67 + }, + { + "epoch": 0.06985105290190036, + "grad_norm": 1.8747144937515259, + "learning_rate": 9.246575342465753e-07, + "loss": 2.5122, + "step": 68 + }, + { + "epoch": 0.07087827426810478, + "grad_norm": 1.744924783706665, + "learning_rate": 9.417808219178083e-07, + "loss": 2.4673, + "step": 69 + }, + { + "epoch": 0.07190549563430919, + "grad_norm": 1.7492460012435913, + "learning_rate": 9.589041095890411e-07, + "loss": 2.5598, + "step": 70 + }, + { + "epoch": 0.07190549563430919, + "eval_loss": 2.547417640686035, + "eval_runtime": 49.7488, + "eval_samples_per_second": 20.101, + "eval_steps_per_second": 0.844, + "step": 70 + }, + { + "epoch": 0.07293271700051361, + "grad_norm": 1.9218230247497559, + "learning_rate": 9.76027397260274e-07, + "loss": 2.4562, + "step": 71 + }, + { + "epoch": 0.07395993836671803, + "grad_norm": 2.0127363204956055, + "learning_rate": 9.931506849315068e-07, + "loss": 2.4141, + "step": 72 + }, + { + "epoch": 0.07498715973292244, + "grad_norm": 1.8700369596481323, + "learning_rate": 1.0102739726027399e-06, + "loss": 2.4696, + "step": 73 + }, + { + "epoch": 0.07601438109912687, + "grad_norm": 2.1706016063690186, + "learning_rate": 1.0273972602739727e-06, + "loss": 2.4404, + "step": 74 + }, + { + "epoch": 0.07704160246533127, + "grad_norm": 2.470412254333496, + "learning_rate": 1.0445205479452056e-06, + "loss": 2.4409, + "step": 75 + }, + { + "epoch": 0.0780688238315357, + "grad_norm": 1.906569242477417, + "learning_rate": 1.0616438356164384e-06, + "loss": 2.4313, + "step": 76 + }, + { + "epoch": 0.07909604519774012, + "grad_norm": 2.122936487197876, + "learning_rate": 1.0787671232876712e-06, + "loss": 2.4369, + "step": 77 + }, + { + "epoch": 0.08012326656394453, + "grad_norm": 2.079127311706543, + "learning_rate": 1.095890410958904e-06, + "loss": 2.449, + "step": 78 + }, + { + "epoch": 0.08115048793014895, + "grad_norm": 1.9173146486282349, + "learning_rate": 1.1130136986301371e-06, + "loss": 2.3782, + "step": 79 + }, + { + "epoch": 0.08217770929635336, + "grad_norm": 1.9856903553009033, + "learning_rate": 1.13013698630137e-06, + "loss": 2.4215, + "step": 80 + }, + { + "epoch": 0.08217770929635336, + "eval_loss": 2.479673385620117, + "eval_runtime": 46.2211, + "eval_samples_per_second": 21.635, + "eval_steps_per_second": 0.909, + "step": 80 + }, + { + "epoch": 0.08320493066255778, + "grad_norm": 2.3764865398406982, + "learning_rate": 1.1472602739726028e-06, + "loss": 2.3744, + "step": 81 + }, + { + "epoch": 0.08423215202876219, + "grad_norm": 1.980668544769287, + "learning_rate": 1.1643835616438357e-06, + "loss": 2.3856, + "step": 82 + }, + { + "epoch": 0.08525937339496661, + "grad_norm": 2.2589704990386963, + "learning_rate": 1.1815068493150685e-06, + "loss": 2.3853, + "step": 83 + }, + { + "epoch": 0.08628659476117104, + "grad_norm": 2.171780824661255, + "learning_rate": 1.1986301369863014e-06, + "loss": 2.3603, + "step": 84 + }, + { + "epoch": 0.08731381612737545, + "grad_norm": 2.25736403465271, + "learning_rate": 1.2157534246575344e-06, + "loss": 2.3189, + "step": 85 + }, + { + "epoch": 0.08834103749357987, + "grad_norm": 2.3643763065338135, + "learning_rate": 1.2328767123287673e-06, + "loss": 2.3795, + "step": 86 + }, + { + "epoch": 0.08936825885978428, + "grad_norm": 2.384516954421997, + "learning_rate": 1.25e-06, + "loss": 2.3257, + "step": 87 + }, + { + "epoch": 0.0903954802259887, + "grad_norm": 2.5072567462921143, + "learning_rate": 1.267123287671233e-06, + "loss": 2.3125, + "step": 88 + }, + { + "epoch": 0.09142270159219312, + "grad_norm": 2.2726807594299316, + "learning_rate": 1.284246575342466e-06, + "loss": 2.3297, + "step": 89 + }, + { + "epoch": 0.09244992295839753, + "grad_norm": 2.4817421436309814, + "learning_rate": 1.3013698630136986e-06, + "loss": 2.2878, + "step": 90 + }, + { + "epoch": 0.09244992295839753, + "eval_loss": 2.4065146446228027, + "eval_runtime": 47.9372, + "eval_samples_per_second": 20.861, + "eval_steps_per_second": 0.876, + "step": 90 + }, + { + "epoch": 0.09347714432460195, + "grad_norm": 2.2219603061676025, + "learning_rate": 1.3184931506849317e-06, + "loss": 2.2978, + "step": 91 + }, + { + "epoch": 0.09450436569080636, + "grad_norm": 2.8790745735168457, + "learning_rate": 1.3356164383561645e-06, + "loss": 2.313, + "step": 92 + }, + { + "epoch": 0.09553158705701079, + "grad_norm": 2.589200496673584, + "learning_rate": 1.3527397260273976e-06, + "loss": 2.2937, + "step": 93 + }, + { + "epoch": 0.09655880842321521, + "grad_norm": 2.760016679763794, + "learning_rate": 1.3698630136986302e-06, + "loss": 2.277, + "step": 94 + }, + { + "epoch": 0.09758602978941962, + "grad_norm": 2.2641494274139404, + "learning_rate": 1.386986301369863e-06, + "loss": 2.3336, + "step": 95 + }, + { + "epoch": 0.09861325115562404, + "grad_norm": 2.2786386013031006, + "learning_rate": 1.4041095890410961e-06, + "loss": 2.233, + "step": 96 + }, + { + "epoch": 0.09964047252182845, + "grad_norm": 2.5590052604675293, + "learning_rate": 1.421232876712329e-06, + "loss": 2.2851, + "step": 97 + }, + { + "epoch": 0.10066769388803287, + "grad_norm": 2.5609471797943115, + "learning_rate": 1.4383561643835616e-06, + "loss": 2.1848, + "step": 98 + }, + { + "epoch": 0.1016949152542373, + "grad_norm": 2.442521810531616, + "learning_rate": 1.4554794520547946e-06, + "loss": 2.2073, + "step": 99 + }, + { + "epoch": 0.1027221366204417, + "grad_norm": 2.3735859394073486, + "learning_rate": 1.4726027397260275e-06, + "loss": 2.2353, + "step": 100 + }, + { + "epoch": 0.1027221366204417, + "eval_loss": 2.3457911014556885, + "eval_runtime": 51.6469, + "eval_samples_per_second": 19.362, + "eval_steps_per_second": 0.813, + "step": 100 + }, + { + "epoch": 0.10374935798664613, + "grad_norm": 2.175354242324829, + "learning_rate": 1.4897260273972605e-06, + "loss": 2.2259, + "step": 101 + }, + { + "epoch": 0.10477657935285054, + "grad_norm": 2.217677116394043, + "learning_rate": 1.5068493150684932e-06, + "loss": 2.2082, + "step": 102 + }, + { + "epoch": 0.10580380071905496, + "grad_norm": 2.4618067741394043, + "learning_rate": 1.5239726027397262e-06, + "loss": 2.1649, + "step": 103 + }, + { + "epoch": 0.10683102208525937, + "grad_norm": 2.232060432434082, + "learning_rate": 1.541095890410959e-06, + "loss": 2.1962, + "step": 104 + }, + { + "epoch": 0.10785824345146379, + "grad_norm": 2.9140872955322266, + "learning_rate": 1.5582191780821921e-06, + "loss": 2.2446, + "step": 105 + }, + { + "epoch": 0.10888546481766821, + "grad_norm": 2.232158660888672, + "learning_rate": 1.5753424657534248e-06, + "loss": 2.2179, + "step": 106 + }, + { + "epoch": 0.10991268618387262, + "grad_norm": 2.804938793182373, + "learning_rate": 1.5924657534246576e-06, + "loss": 2.2449, + "step": 107 + }, + { + "epoch": 0.11093990755007704, + "grad_norm": 2.3690710067749023, + "learning_rate": 1.6095890410958907e-06, + "loss": 2.225, + "step": 108 + }, + { + "epoch": 0.11196712891628145, + "grad_norm": 2.7717416286468506, + "learning_rate": 1.6267123287671235e-06, + "loss": 2.0922, + "step": 109 + }, + { + "epoch": 0.11299435028248588, + "grad_norm": 3.053602933883667, + "learning_rate": 1.6438356164383561e-06, + "loss": 2.1702, + "step": 110 + }, + { + "epoch": 0.11299435028248588, + "eval_loss": 2.298161029815674, + "eval_runtime": 44.0776, + "eval_samples_per_second": 22.687, + "eval_steps_per_second": 0.953, + "step": 110 + }, + { + "epoch": 0.1140215716486903, + "grad_norm": 2.8673532009124756, + "learning_rate": 1.6609589041095892e-06, + "loss": 2.0938, + "step": 111 + }, + { + "epoch": 0.1150487930148947, + "grad_norm": 2.57185959815979, + "learning_rate": 1.678082191780822e-06, + "loss": 2.1892, + "step": 112 + }, + { + "epoch": 0.11607601438109913, + "grad_norm": 2.5278828144073486, + "learning_rate": 1.695205479452055e-06, + "loss": 2.1251, + "step": 113 + }, + { + "epoch": 0.11710323574730354, + "grad_norm": 2.8581748008728027, + "learning_rate": 1.7123287671232877e-06, + "loss": 2.1699, + "step": 114 + }, + { + "epoch": 0.11813045711350796, + "grad_norm": 2.924959897994995, + "learning_rate": 1.7294520547945206e-06, + "loss": 2.1005, + "step": 115 + }, + { + "epoch": 0.11915767847971238, + "grad_norm": 2.477034091949463, + "learning_rate": 1.7465753424657536e-06, + "loss": 2.2016, + "step": 116 + }, + { + "epoch": 0.12018489984591679, + "grad_norm": 2.8719305992126465, + "learning_rate": 1.7636986301369865e-06, + "loss": 2.1588, + "step": 117 + }, + { + "epoch": 0.12121212121212122, + "grad_norm": 2.4403953552246094, + "learning_rate": 1.7808219178082193e-06, + "loss": 2.1445, + "step": 118 + }, + { + "epoch": 0.12223934257832562, + "grad_norm": 2.7994632720947266, + "learning_rate": 1.7979452054794521e-06, + "loss": 2.1369, + "step": 119 + }, + { + "epoch": 0.12326656394453005, + "grad_norm": 2.163196563720703, + "learning_rate": 1.8150684931506852e-06, + "loss": 2.2195, + "step": 120 + }, + { + "epoch": 0.12326656394453005, + "eval_loss": 2.2648775577545166, + "eval_runtime": 49.47, + "eval_samples_per_second": 20.214, + "eval_steps_per_second": 0.849, + "step": 120 + }, + { + "epoch": 0.12429378531073447, + "grad_norm": 2.6743483543395996, + "learning_rate": 1.832191780821918e-06, + "loss": 2.1521, + "step": 121 + }, + { + "epoch": 0.12532100667693888, + "grad_norm": 2.54795503616333, + "learning_rate": 1.8493150684931507e-06, + "loss": 2.1158, + "step": 122 + }, + { + "epoch": 0.1263482280431433, + "grad_norm": 2.9093267917633057, + "learning_rate": 1.8664383561643837e-06, + "loss": 2.0707, + "step": 123 + }, + { + "epoch": 0.12737544940934772, + "grad_norm": 2.416940212249756, + "learning_rate": 1.8835616438356166e-06, + "loss": 2.0901, + "step": 124 + }, + { + "epoch": 0.12840267077555212, + "grad_norm": 2.8333663940429688, + "learning_rate": 1.9006849315068496e-06, + "loss": 2.0545, + "step": 125 + }, + { + "epoch": 0.12942989214175654, + "grad_norm": 2.483051061630249, + "learning_rate": 1.9178082191780823e-06, + "loss": 2.1274, + "step": 126 + }, + { + "epoch": 0.13045711350796096, + "grad_norm": 2.3486859798431396, + "learning_rate": 1.9349315068493153e-06, + "loss": 2.0478, + "step": 127 + }, + { + "epoch": 0.1314843348741654, + "grad_norm": 3.1512887477874756, + "learning_rate": 1.952054794520548e-06, + "loss": 2.1041, + "step": 128 + }, + { + "epoch": 0.1325115562403698, + "grad_norm": 2.902873992919922, + "learning_rate": 1.969178082191781e-06, + "loss": 2.1332, + "step": 129 + }, + { + "epoch": 0.1335387776065742, + "grad_norm": 2.779132127761841, + "learning_rate": 1.9863013698630136e-06, + "loss": 2.1297, + "step": 130 + }, + { + "epoch": 0.1335387776065742, + "eval_loss": 2.2391879558563232, + "eval_runtime": 55.3018, + "eval_samples_per_second": 18.083, + "eval_steps_per_second": 0.759, + "step": 130 + }, + { + "epoch": 0.13456599897277863, + "grad_norm": 2.703019618988037, + "learning_rate": 2.0034246575342467e-06, + "loss": 2.0506, + "step": 131 + }, + { + "epoch": 0.13559322033898305, + "grad_norm": 3.0187575817108154, + "learning_rate": 2.0205479452054797e-06, + "loss": 2.1291, + "step": 132 + }, + { + "epoch": 0.13662044170518747, + "grad_norm": 2.7629971504211426, + "learning_rate": 2.037671232876713e-06, + "loss": 2.1003, + "step": 133 + }, + { + "epoch": 0.1376476630713919, + "grad_norm": 3.175776720046997, + "learning_rate": 2.0547945205479454e-06, + "loss": 2.0429, + "step": 134 + }, + { + "epoch": 0.1386748844375963, + "grad_norm": 3.0342514514923096, + "learning_rate": 2.071917808219178e-06, + "loss": 2.0858, + "step": 135 + }, + { + "epoch": 0.1397021058038007, + "grad_norm": 3.0861928462982178, + "learning_rate": 2.089041095890411e-06, + "loss": 2.0964, + "step": 136 + }, + { + "epoch": 0.14072932717000514, + "grad_norm": 3.151945114135742, + "learning_rate": 2.106164383561644e-06, + "loss": 2.0194, + "step": 137 + }, + { + "epoch": 0.14175654853620956, + "grad_norm": 2.9069676399230957, + "learning_rate": 2.123287671232877e-06, + "loss": 2.1131, + "step": 138 + }, + { + "epoch": 0.14278376990241398, + "grad_norm": 3.2285282611846924, + "learning_rate": 2.14041095890411e-06, + "loss": 2.0643, + "step": 139 + }, + { + "epoch": 0.14381099126861838, + "grad_norm": 2.500481367111206, + "learning_rate": 2.1575342465753425e-06, + "loss": 2.0625, + "step": 140 + }, + { + "epoch": 0.14381099126861838, + "eval_loss": 2.211498975753784, + "eval_runtime": 55.1046, + "eval_samples_per_second": 18.147, + "eval_steps_per_second": 0.762, + "step": 140 + }, + { + "epoch": 0.1448382126348228, + "grad_norm": 4.004592418670654, + "learning_rate": 2.1746575342465755e-06, + "loss": 2.0644, + "step": 141 + }, + { + "epoch": 0.14586543400102722, + "grad_norm": 2.9258391857147217, + "learning_rate": 2.191780821917808e-06, + "loss": 2.0174, + "step": 142 + }, + { + "epoch": 0.14689265536723164, + "grad_norm": 3.0073509216308594, + "learning_rate": 2.2089041095890412e-06, + "loss": 2.1417, + "step": 143 + }, + { + "epoch": 0.14791987673343607, + "grad_norm": 3.040708065032959, + "learning_rate": 2.2260273972602743e-06, + "loss": 2.0357, + "step": 144 + }, + { + "epoch": 0.14894709809964046, + "grad_norm": 2.9062869548797607, + "learning_rate": 2.243150684931507e-06, + "loss": 2.0122, + "step": 145 + }, + { + "epoch": 0.14997431946584489, + "grad_norm": 3.28969669342041, + "learning_rate": 2.26027397260274e-06, + "loss": 2.0255, + "step": 146 + }, + { + "epoch": 0.1510015408320493, + "grad_norm": 3.608752727508545, + "learning_rate": 2.2773972602739726e-06, + "loss": 2.0184, + "step": 147 + }, + { + "epoch": 0.15202876219825373, + "grad_norm": 2.900341033935547, + "learning_rate": 2.2945205479452057e-06, + "loss": 2.0532, + "step": 148 + }, + { + "epoch": 0.15305598356445815, + "grad_norm": 2.881843328475952, + "learning_rate": 2.3116438356164387e-06, + "loss": 2.1336, + "step": 149 + }, + { + "epoch": 0.15408320493066255, + "grad_norm": 3.4043259620666504, + "learning_rate": 2.3287671232876713e-06, + "loss": 2.0536, + "step": 150 + }, + { + "epoch": 0.15408320493066255, + "eval_loss": 2.192880153656006, + "eval_runtime": 53.7838, + "eval_samples_per_second": 18.593, + "eval_steps_per_second": 0.781, + "step": 150 + }, + { + "epoch": 0.15511042629686697, + "grad_norm": 2.693107843399048, + "learning_rate": 2.3458904109589044e-06, + "loss": 1.9503, + "step": 151 + }, + { + "epoch": 0.1561376476630714, + "grad_norm": 2.885260581970215, + "learning_rate": 2.363013698630137e-06, + "loss": 2.0081, + "step": 152 + }, + { + "epoch": 0.15716486902927582, + "grad_norm": 3.4459924697875977, + "learning_rate": 2.38013698630137e-06, + "loss": 2.0649, + "step": 153 + }, + { + "epoch": 0.15819209039548024, + "grad_norm": 2.8447093963623047, + "learning_rate": 2.3972602739726027e-06, + "loss": 2.0476, + "step": 154 + }, + { + "epoch": 0.15921931176168463, + "grad_norm": 2.755176305770874, + "learning_rate": 2.4143835616438358e-06, + "loss": 2.0006, + "step": 155 + }, + { + "epoch": 0.16024653312788906, + "grad_norm": 2.8677561283111572, + "learning_rate": 2.431506849315069e-06, + "loss": 1.9663, + "step": 156 + }, + { + "epoch": 0.16127375449409348, + "grad_norm": 2.783351182937622, + "learning_rate": 2.4486301369863015e-06, + "loss": 2.0301, + "step": 157 + }, + { + "epoch": 0.1623009758602979, + "grad_norm": 2.7545998096466064, + "learning_rate": 2.4657534246575345e-06, + "loss": 2.0438, + "step": 158 + }, + { + "epoch": 0.1633281972265023, + "grad_norm": 2.8140594959259033, + "learning_rate": 2.482876712328767e-06, + "loss": 2.0055, + "step": 159 + }, + { + "epoch": 0.16435541859270672, + "grad_norm": 2.6072335243225098, + "learning_rate": 2.5e-06, + "loss": 2.0262, + "step": 160 + }, + { + "epoch": 0.16435541859270672, + "eval_loss": 2.179842710494995, + "eval_runtime": 52.6159, + "eval_samples_per_second": 19.006, + "eval_steps_per_second": 0.798, + "step": 160 + }, + { + "epoch": 0.16538263995891114, + "grad_norm": 2.830212354660034, + "learning_rate": 2.5171232876712333e-06, + "loss": 2.0154, + "step": 161 + }, + { + "epoch": 0.16640986132511557, + "grad_norm": 2.677140712738037, + "learning_rate": 2.534246575342466e-06, + "loss": 1.9933, + "step": 162 + }, + { + "epoch": 0.16743708269132, + "grad_norm": 2.837759017944336, + "learning_rate": 2.551369863013699e-06, + "loss": 1.9602, + "step": 163 + }, + { + "epoch": 0.16846430405752438, + "grad_norm": 3.9556589126586914, + "learning_rate": 2.568493150684932e-06, + "loss": 2.0121, + "step": 164 + }, + { + "epoch": 0.1694915254237288, + "grad_norm": 2.8715925216674805, + "learning_rate": 2.585616438356164e-06, + "loss": 2.0219, + "step": 165 + }, + { + "epoch": 0.17051874678993323, + "grad_norm": 3.5073623657226562, + "learning_rate": 2.6027397260273973e-06, + "loss": 2.0202, + "step": 166 + }, + { + "epoch": 0.17154596815613765, + "grad_norm": 2.751962900161743, + "learning_rate": 2.6198630136986303e-06, + "loss": 1.9528, + "step": 167 + }, + { + "epoch": 0.17257318952234207, + "grad_norm": 3.7753474712371826, + "learning_rate": 2.6369863013698634e-06, + "loss": 2.0208, + "step": 168 + }, + { + "epoch": 0.17360041088854647, + "grad_norm": 2.642289876937866, + "learning_rate": 2.654109589041096e-06, + "loss": 2.0386, + "step": 169 + }, + { + "epoch": 0.1746276322547509, + "grad_norm": 2.82173490524292, + "learning_rate": 2.671232876712329e-06, + "loss": 2.0355, + "step": 170 + }, + { + "epoch": 0.1746276322547509, + "eval_loss": 2.1699249744415283, + "eval_runtime": 50.0341, + "eval_samples_per_second": 19.986, + "eval_steps_per_second": 0.839, + "step": 170 + }, + { + "epoch": 0.17565485362095531, + "grad_norm": 3.342280149459839, + "learning_rate": 2.688356164383562e-06, + "loss": 1.9672, + "step": 171 + }, + { + "epoch": 0.17668207498715974, + "grad_norm": 2.6548173427581787, + "learning_rate": 2.705479452054795e-06, + "loss": 1.9669, + "step": 172 + }, + { + "epoch": 0.17770929635336416, + "grad_norm": 2.6548173427581787, + "learning_rate": 2.705479452054795e-06, + "loss": 1.961, + "step": 173 + }, + { + "epoch": 0.17873651771956856, + "grad_norm": 3.4306116104125977, + "learning_rate": 2.7226027397260274e-06, + "loss": 1.9173, + "step": 174 + }, + { + "epoch": 0.17976373908577298, + "grad_norm": 2.7991764545440674, + "learning_rate": 2.7397260273972604e-06, + "loss": 2.0339, + "step": 175 + }, + { + "epoch": 0.1807909604519774, + "grad_norm": 3.6040217876434326, + "learning_rate": 2.7568493150684935e-06, + "loss": 2.0218, + "step": 176 + }, + { + "epoch": 0.18181818181818182, + "grad_norm": 3.0253515243530273, + "learning_rate": 2.773972602739726e-06, + "loss": 1.9245, + "step": 177 + }, + { + "epoch": 0.18284540318438625, + "grad_norm": 3.405141830444336, + "learning_rate": 2.791095890410959e-06, + "loss": 2.0136, + "step": 178 + }, + { + "epoch": 0.18387262455059064, + "grad_norm": 2.7518203258514404, + "learning_rate": 2.8082191780821922e-06, + "loss": 1.8925, + "step": 179 + }, + { + "epoch": 0.18489984591679506, + "grad_norm": 3.1903367042541504, + "learning_rate": 2.8253424657534253e-06, + "loss": 2.0523, + "step": 180 + }, + { + "epoch": 0.18489984591679506, + "eval_loss": 2.158029794692993, + "eval_runtime": 48.6162, + "eval_samples_per_second": 20.569, + "eval_steps_per_second": 0.864, + "step": 180 + }, + { + "epoch": 0.1859270672829995, + "grad_norm": 3.0331413745880127, + "learning_rate": 2.842465753424658e-06, + "loss": 2.0167, + "step": 181 + }, + { + "epoch": 0.1869542886492039, + "grad_norm": 2.919433116912842, + "learning_rate": 2.8595890410958905e-06, + "loss": 2.0197, + "step": 182 + }, + { + "epoch": 0.18798151001540833, + "grad_norm": 2.9013049602508545, + "learning_rate": 2.876712328767123e-06, + "loss": 2.0301, + "step": 183 + }, + { + "epoch": 0.18900873138161273, + "grad_norm": 2.738800525665283, + "learning_rate": 2.8938356164383562e-06, + "loss": 1.894, + "step": 184 + }, + { + "epoch": 0.19003595274781715, + "grad_norm": 2.670806407928467, + "learning_rate": 2.9109589041095893e-06, + "loss": 1.9127, + "step": 185 + }, + { + "epoch": 0.19106317411402157, + "grad_norm": 3.0435004234313965, + "learning_rate": 2.9280821917808223e-06, + "loss": 1.8996, + "step": 186 + }, + { + "epoch": 0.192090395480226, + "grad_norm": 3.2139949798583984, + "learning_rate": 2.945205479452055e-06, + "loss": 1.9255, + "step": 187 + }, + { + "epoch": 0.19311761684643042, + "grad_norm": 3.252965211868286, + "learning_rate": 2.962328767123288e-06, + "loss": 1.9809, + "step": 188 + }, + { + "epoch": 0.1941448382126348, + "grad_norm": 2.9325191974639893, + "learning_rate": 2.979452054794521e-06, + "loss": 1.9691, + "step": 189 + }, + { + "epoch": 0.19517205957883924, + "grad_norm": 3.5879862308502197, + "learning_rate": 2.9965753424657533e-06, + "loss": 1.9205, + "step": 190 + }, + { + "epoch": 0.19517205957883924, + "eval_loss": 2.13791561126709, + "eval_runtime": 44.8545, + "eval_samples_per_second": 22.294, + "eval_steps_per_second": 0.936, + "step": 190 + }, + { + "epoch": 0.19619928094504366, + "grad_norm": 3.0025253295898438, + "learning_rate": 3.0136986301369864e-06, + "loss": 1.9717, + "step": 191 + }, + { + "epoch": 0.19722650231124808, + "grad_norm": 3.1785972118377686, + "learning_rate": 3.0308219178082194e-06, + "loss": 1.9535, + "step": 192 + }, + { + "epoch": 0.1982537236774525, + "grad_norm": 3.121065378189087, + "learning_rate": 3.0479452054794525e-06, + "loss": 1.9026, + "step": 193 + }, + { + "epoch": 0.1992809450436569, + "grad_norm": 2.838779926300049, + "learning_rate": 3.065068493150685e-06, + "loss": 1.941, + "step": 194 + }, + { + "epoch": 0.20030816640986132, + "grad_norm": 2.8916969299316406, + "learning_rate": 3.082191780821918e-06, + "loss": 1.8983, + "step": 195 + }, + { + "epoch": 0.20133538777606574, + "grad_norm": 2.9195168018341064, + "learning_rate": 3.099315068493151e-06, + "loss": 1.918, + "step": 196 + }, + { + "epoch": 0.20236260914227017, + "grad_norm": 3.235989809036255, + "learning_rate": 3.1164383561643843e-06, + "loss": 1.9111, + "step": 197 + }, + { + "epoch": 0.2033898305084746, + "grad_norm": 3.0316455364227295, + "learning_rate": 3.1335616438356165e-06, + "loss": 1.8839, + "step": 198 + }, + { + "epoch": 0.20441705187467898, + "grad_norm": 2.963610887527466, + "learning_rate": 3.1506849315068495e-06, + "loss": 2.0303, + "step": 199 + }, + { + "epoch": 0.2054442732408834, + "grad_norm": 2.944054365158081, + "learning_rate": 3.167808219178082e-06, + "loss": 1.8822, + "step": 200 + }, + { + "epoch": 0.2054442732408834, + "eval_loss": 2.1433050632476807, + "eval_runtime": 45.8205, + "eval_samples_per_second": 21.824, + "eval_steps_per_second": 0.917, + "step": 200 + }, + { + "epoch": 0.20647149460708783, + "grad_norm": 2.8648695945739746, + "learning_rate": 3.184931506849315e-06, + "loss": 1.8841, + "step": 201 + }, + { + "epoch": 0.20749871597329225, + "grad_norm": 3.2838797569274902, + "learning_rate": 3.2020547945205483e-06, + "loss": 1.8623, + "step": 202 + }, + { + "epoch": 0.20852593733949668, + "grad_norm": 2.7965173721313477, + "learning_rate": 3.2191780821917813e-06, + "loss": 1.8521, + "step": 203 + }, + { + "epoch": 0.20955315870570107, + "grad_norm": 3.3449349403381348, + "learning_rate": 3.236301369863014e-06, + "loss": 1.84, + "step": 204 + }, + { + "epoch": 0.2105803800719055, + "grad_norm": 3.4888951778411865, + "learning_rate": 3.253424657534247e-06, + "loss": 1.824, + "step": 205 + }, + { + "epoch": 0.21160760143810992, + "grad_norm": 2.991349220275879, + "learning_rate": 3.2705479452054796e-06, + "loss": 1.9101, + "step": 206 + }, + { + "epoch": 0.21263482280431434, + "grad_norm": 3.313509702682495, + "learning_rate": 3.2876712328767123e-06, + "loss": 1.8912, + "step": 207 + }, + { + "epoch": 0.21366204417051873, + "grad_norm": 2.8260746002197266, + "learning_rate": 3.3047945205479453e-06, + "loss": 1.8555, + "step": 208 + }, + { + "epoch": 0.21468926553672316, + "grad_norm": 2.7131688594818115, + "learning_rate": 3.3219178082191784e-06, + "loss": 1.8693, + "step": 209 + }, + { + "epoch": 0.21571648690292758, + "grad_norm": 2.5836338996887207, + "learning_rate": 3.3390410958904114e-06, + "loss": 1.8965, + "step": 210 + }, + { + "epoch": 0.21571648690292758, + "eval_loss": 2.1232542991638184, + "eval_runtime": 42.3352, + "eval_samples_per_second": 23.621, + "eval_steps_per_second": 0.992, + "step": 210 + }, + { + "epoch": 0.216743708269132, + "grad_norm": 2.8341355323791504, + "learning_rate": 3.356164383561644e-06, + "loss": 1.9131, + "step": 211 + }, + { + "epoch": 0.21777092963533642, + "grad_norm": 2.490853786468506, + "learning_rate": 3.373287671232877e-06, + "loss": 1.9626, + "step": 212 + }, + { + "epoch": 0.21879815100154082, + "grad_norm": 2.597956895828247, + "learning_rate": 3.39041095890411e-06, + "loss": 1.9203, + "step": 213 + }, + { + "epoch": 0.21982537236774524, + "grad_norm": 3.1104283332824707, + "learning_rate": 3.4075342465753424e-06, + "loss": 1.9442, + "step": 214 + }, + { + "epoch": 0.22085259373394966, + "grad_norm": 2.5698931217193604, + "learning_rate": 3.4246575342465754e-06, + "loss": 1.8385, + "step": 215 + }, + { + "epoch": 0.2218798151001541, + "grad_norm": 3.1637446880340576, + "learning_rate": 3.4417808219178085e-06, + "loss": 1.9005, + "step": 216 + }, + { + "epoch": 0.2229070364663585, + "grad_norm": 2.764070749282837, + "learning_rate": 3.458904109589041e-06, + "loss": 1.8886, + "step": 217 + }, + { + "epoch": 0.2239342578325629, + "grad_norm": 3.303781032562256, + "learning_rate": 3.476027397260274e-06, + "loss": 1.9085, + "step": 218 + }, + { + "epoch": 0.22496147919876733, + "grad_norm": 2.9543044567108154, + "learning_rate": 3.4931506849315072e-06, + "loss": 1.9014, + "step": 219 + }, + { + "epoch": 0.22598870056497175, + "grad_norm": 2.7277028560638428, + "learning_rate": 3.5102739726027403e-06, + "loss": 1.9017, + "step": 220 + }, + { + "epoch": 0.22598870056497175, + "eval_loss": 2.1024081707000732, + "eval_runtime": 45.4966, + "eval_samples_per_second": 21.98, + "eval_steps_per_second": 0.923, + "step": 220 + }, + { + "epoch": 0.22701592193117617, + "grad_norm": 2.864140033721924, + "learning_rate": 3.527397260273973e-06, + "loss": 1.8935, + "step": 221 + }, + { + "epoch": 0.2280431432973806, + "grad_norm": 3.4133172035217285, + "learning_rate": 3.5445205479452056e-06, + "loss": 1.874, + "step": 222 + }, + { + "epoch": 0.229070364663585, + "grad_norm": 3.33343243598938, + "learning_rate": 3.5616438356164386e-06, + "loss": 1.8789, + "step": 223 + }, + { + "epoch": 0.2300975860297894, + "grad_norm": 3.5947320461273193, + "learning_rate": 3.5787671232876712e-06, + "loss": 1.9529, + "step": 224 + }, + { + "epoch": 0.23112480739599384, + "grad_norm": 3.2194948196411133, + "learning_rate": 3.5958904109589043e-06, + "loss": 1.9205, + "step": 225 + }, + { + "epoch": 0.23215202876219826, + "grad_norm": 3.3137149810791016, + "learning_rate": 3.6130136986301373e-06, + "loss": 1.8584, + "step": 226 + }, + { + "epoch": 0.23317925012840268, + "grad_norm": 3.3358561992645264, + "learning_rate": 3.6301369863013704e-06, + "loss": 1.8817, + "step": 227 + }, + { + "epoch": 0.23420647149460708, + "grad_norm": 3.8612711429595947, + "learning_rate": 3.647260273972603e-06, + "loss": 1.9189, + "step": 228 + }, + { + "epoch": 0.2352336928608115, + "grad_norm": 3.0039336681365967, + "learning_rate": 3.664383561643836e-06, + "loss": 1.92, + "step": 229 + }, + { + "epoch": 0.23626091422701592, + "grad_norm": 3.8983118534088135, + "learning_rate": 3.6815068493150687e-06, + "loss": 1.8512, + "step": 230 + }, + { + "epoch": 0.23626091422701592, + "eval_loss": 2.1237711906433105, + "eval_runtime": 45.6126, + "eval_samples_per_second": 21.924, + "eval_steps_per_second": 0.921, + "step": 230 + }, + { + "epoch": 0.23728813559322035, + "grad_norm": 3.061300039291382, + "learning_rate": 3.6986301369863014e-06, + "loss": 1.8772, + "step": 231 + }, + { + "epoch": 0.23831535695942477, + "grad_norm": 3.3102450370788574, + "learning_rate": 3.7157534246575344e-06, + "loss": 1.8395, + "step": 232 + }, + { + "epoch": 0.23934257832562916, + "grad_norm": 2.9634461402893066, + "learning_rate": 3.7328767123287675e-06, + "loss": 1.9447, + "step": 233 + }, + { + "epoch": 0.24036979969183359, + "grad_norm": 3.0937695503234863, + "learning_rate": 3.7500000000000005e-06, + "loss": 1.8628, + "step": 234 + }, + { + "epoch": 0.241397021058038, + "grad_norm": 2.8129842281341553, + "learning_rate": 3.767123287671233e-06, + "loss": 1.8891, + "step": 235 + }, + { + "epoch": 0.24242424242424243, + "grad_norm": 2.9616780281066895, + "learning_rate": 3.784246575342466e-06, + "loss": 1.9112, + "step": 236 + }, + { + "epoch": 0.24345146379044685, + "grad_norm": 2.837721109390259, + "learning_rate": 3.8013698630136993e-06, + "loss": 1.8514, + "step": 237 + }, + { + "epoch": 0.24447868515665125, + "grad_norm": 3.137784004211426, + "learning_rate": 3.818493150684932e-06, + "loss": 1.8764, + "step": 238 + }, + { + "epoch": 0.24550590652285567, + "grad_norm": 2.9692037105560303, + "learning_rate": 3.8356164383561645e-06, + "loss": 1.8201, + "step": 239 + }, + { + "epoch": 0.2465331278890601, + "grad_norm": 2.6718621253967285, + "learning_rate": 3.852739726027397e-06, + "loss": 1.9309, + "step": 240 + }, + { + "epoch": 0.2465331278890601, + "eval_loss": 2.101331949234009, + "eval_runtime": 45.6346, + "eval_samples_per_second": 21.913, + "eval_steps_per_second": 0.92, + "step": 240 + }, + { + "epoch": 0.24756034925526452, + "grad_norm": 3.411193609237671, + "learning_rate": 3.869863013698631e-06, + "loss": 1.9171, + "step": 241 + }, + { + "epoch": 0.24858757062146894, + "grad_norm": 2.8848962783813477, + "learning_rate": 3.886986301369863e-06, + "loss": 1.8587, + "step": 242 + }, + { + "epoch": 0.24961479198767333, + "grad_norm": 2.756298065185547, + "learning_rate": 3.904109589041096e-06, + "loss": 1.8614, + "step": 243 + }, + { + "epoch": 0.25064201335387776, + "grad_norm": 2.8048956394195557, + "learning_rate": 3.921232876712329e-06, + "loss": 1.8225, + "step": 244 + }, + { + "epoch": 0.2516692347200822, + "grad_norm": 2.9872403144836426, + "learning_rate": 3.938356164383562e-06, + "loss": 1.8367, + "step": 245 + }, + { + "epoch": 0.2526964560862866, + "grad_norm": 2.710156202316284, + "learning_rate": 3.9554794520547955e-06, + "loss": 1.809, + "step": 246 + }, + { + "epoch": 0.253723677452491, + "grad_norm": 3.2177000045776367, + "learning_rate": 3.972602739726027e-06, + "loss": 1.8883, + "step": 247 + }, + { + "epoch": 0.25475089881869545, + "grad_norm": 3.160964250564575, + "learning_rate": 3.989726027397261e-06, + "loss": 1.8267, + "step": 248 + }, + { + "epoch": 0.25577812018489987, + "grad_norm": 2.853498935699463, + "learning_rate": 4.006849315068493e-06, + "loss": 1.8852, + "step": 249 + }, + { + "epoch": 0.25680534155110424, + "grad_norm": 2.571742296218872, + "learning_rate": 4.023972602739726e-06, + "loss": 1.8833, + "step": 250 + }, + { + "epoch": 0.25680534155110424, + "eval_loss": 2.0988597869873047, + "eval_runtime": 47.7045, + "eval_samples_per_second": 20.962, + "eval_steps_per_second": 0.88, + "step": 250 + }, + { + "epoch": 0.25783256291730866, + "grad_norm": 2.776742458343506, + "learning_rate": 4.0410958904109595e-06, + "loss": 1.8662, + "step": 251 + }, + { + "epoch": 0.2588597842835131, + "grad_norm": 2.8038573265075684, + "learning_rate": 4.058219178082192e-06, + "loss": 1.8528, + "step": 252 + }, + { + "epoch": 0.2598870056497175, + "grad_norm": 2.557623863220215, + "learning_rate": 4.075342465753426e-06, + "loss": 1.8765, + "step": 253 + }, + { + "epoch": 0.26091422701592193, + "grad_norm": 3.0477073192596436, + "learning_rate": 4.092465753424658e-06, + "loss": 1.8434, + "step": 254 + }, + { + "epoch": 0.26194144838212635, + "grad_norm": 2.6221649646759033, + "learning_rate": 4.109589041095891e-06, + "loss": 1.8348, + "step": 255 + }, + { + "epoch": 0.2629686697483308, + "grad_norm": 3.0849292278289795, + "learning_rate": 4.1267123287671235e-06, + "loss": 1.9057, + "step": 256 + }, + { + "epoch": 0.2639958911145352, + "grad_norm": 2.795184373855591, + "learning_rate": 4.143835616438356e-06, + "loss": 1.8528, + "step": 257 + }, + { + "epoch": 0.2650231124807396, + "grad_norm": 2.887193441390991, + "learning_rate": 4.16095890410959e-06, + "loss": 1.9176, + "step": 258 + }, + { + "epoch": 0.26605033384694404, + "grad_norm": 2.776801347732544, + "learning_rate": 4.178082191780822e-06, + "loss": 1.8911, + "step": 259 + }, + { + "epoch": 0.2670775552131484, + "grad_norm": 2.980426788330078, + "learning_rate": 4.195205479452055e-06, + "loss": 1.8288, + "step": 260 + }, + { + "epoch": 0.2670775552131484, + "eval_loss": 2.0636370182037354, + "eval_runtime": 49.3081, + "eval_samples_per_second": 20.281, + "eval_steps_per_second": 0.852, + "step": 260 + }, + { + "epoch": 0.26810477657935283, + "grad_norm": 2.7921340465545654, + "learning_rate": 4.212328767123288e-06, + "loss": 1.8405, + "step": 261 + }, + { + "epoch": 0.26913199794555726, + "grad_norm": 2.67714262008667, + "learning_rate": 4.229452054794521e-06, + "loss": 1.8342, + "step": 262 + }, + { + "epoch": 0.2701592193117617, + "grad_norm": 2.819868326187134, + "learning_rate": 4.246575342465754e-06, + "loss": 1.8466, + "step": 263 + }, + { + "epoch": 0.2711864406779661, + "grad_norm": 2.611250638961792, + "learning_rate": 4.263698630136986e-06, + "loss": 1.8314, + "step": 264 + }, + { + "epoch": 0.2722136620441705, + "grad_norm": 3.0406076908111572, + "learning_rate": 4.28082191780822e-06, + "loss": 1.7924, + "step": 265 + }, + { + "epoch": 0.27324088341037495, + "grad_norm": 3.212790012359619, + "learning_rate": 4.297945205479452e-06, + "loss": 1.9479, + "step": 266 + }, + { + "epoch": 0.27426810477657937, + "grad_norm": 2.4767045974731445, + "learning_rate": 4.315068493150685e-06, + "loss": 1.835, + "step": 267 + }, + { + "epoch": 0.2752953261427838, + "grad_norm": 3.191835641860962, + "learning_rate": 4.3321917808219185e-06, + "loss": 1.8261, + "step": 268 + }, + { + "epoch": 0.2763225475089882, + "grad_norm": 2.9970273971557617, + "learning_rate": 4.349315068493151e-06, + "loss": 1.8427, + "step": 269 + }, + { + "epoch": 0.2773497688751926, + "grad_norm": 2.888843297958374, + "learning_rate": 4.3664383561643846e-06, + "loss": 1.819, + "step": 270 + }, + { + "epoch": 0.2773497688751926, + "eval_loss": 2.054966926574707, + "eval_runtime": 45.9212, + "eval_samples_per_second": 21.776, + "eval_steps_per_second": 0.915, + "step": 270 + }, + { + "epoch": 0.278376990241397, + "grad_norm": 2.7771849632263184, + "learning_rate": 4.383561643835616e-06, + "loss": 1.8178, + "step": 271 + }, + { + "epoch": 0.2794042116076014, + "grad_norm": 2.9749014377593994, + "learning_rate": 4.40068493150685e-06, + "loss": 1.8982, + "step": 272 + }, + { + "epoch": 0.28043143297380585, + "grad_norm": 3.076021432876587, + "learning_rate": 4.4178082191780825e-06, + "loss": 1.8575, + "step": 273 + }, + { + "epoch": 0.2814586543400103, + "grad_norm": 3.2113568782806396, + "learning_rate": 4.434931506849315e-06, + "loss": 1.8241, + "step": 274 + }, + { + "epoch": 0.2824858757062147, + "grad_norm": 3.322063684463501, + "learning_rate": 4.4520547945205486e-06, + "loss": 1.7922, + "step": 275 + }, + { + "epoch": 0.2835130970724191, + "grad_norm": 3.2778918743133545, + "learning_rate": 4.469178082191781e-06, + "loss": 1.8412, + "step": 276 + }, + { + "epoch": 0.28454031843862354, + "grad_norm": 3.1802618503570557, + "learning_rate": 4.486301369863014e-06, + "loss": 1.8283, + "step": 277 + }, + { + "epoch": 0.28556753980482796, + "grad_norm": 2.9214258193969727, + "learning_rate": 4.503424657534247e-06, + "loss": 1.8817, + "step": 278 + }, + { + "epoch": 0.28659476117103233, + "grad_norm": 2.8809854984283447, + "learning_rate": 4.52054794520548e-06, + "loss": 1.7662, + "step": 279 + }, + { + "epoch": 0.28762198253723675, + "grad_norm": 2.925537586212158, + "learning_rate": 4.537671232876713e-06, + "loss": 1.8147, + "step": 280 + }, + { + "epoch": 0.28762198253723675, + "eval_loss": 2.054513454437256, + "eval_runtime": 51.5036, + "eval_samples_per_second": 19.416, + "eval_steps_per_second": 0.815, + "step": 280 + }, + { + "epoch": 0.2886492039034412, + "grad_norm": 2.9421887397766113, + "learning_rate": 4.554794520547945e-06, + "loss": 1.8581, + "step": 281 + }, + { + "epoch": 0.2896764252696456, + "grad_norm": 2.9733567237854004, + "learning_rate": 4.571917808219179e-06, + "loss": 1.8703, + "step": 282 + }, + { + "epoch": 0.29070364663585, + "grad_norm": 2.5352156162261963, + "learning_rate": 4.589041095890411e-06, + "loss": 1.8505, + "step": 283 + }, + { + "epoch": 0.29173086800205444, + "grad_norm": 2.773847818374634, + "learning_rate": 4.606164383561644e-06, + "loss": 1.8114, + "step": 284 + }, + { + "epoch": 0.29275808936825887, + "grad_norm": 2.709632635116577, + "learning_rate": 4.6232876712328774e-06, + "loss": 1.815, + "step": 285 + }, + { + "epoch": 0.2937853107344633, + "grad_norm": 3.1234357357025146, + "learning_rate": 4.64041095890411e-06, + "loss": 1.8391, + "step": 286 + }, + { + "epoch": 0.2948125321006677, + "grad_norm": 2.9376144409179688, + "learning_rate": 4.657534246575343e-06, + "loss": 1.7998, + "step": 287 + }, + { + "epoch": 0.29583975346687214, + "grad_norm": 3.0812032222747803, + "learning_rate": 4.674657534246575e-06, + "loss": 1.8462, + "step": 288 + }, + { + "epoch": 0.2968669748330765, + "grad_norm": 2.922722578048706, + "learning_rate": 4.691780821917809e-06, + "loss": 1.8263, + "step": 289 + }, + { + "epoch": 0.2978941961992809, + "grad_norm": 2.9484992027282715, + "learning_rate": 4.7089041095890414e-06, + "loss": 1.8754, + "step": 290 + }, + { + "epoch": 0.2978941961992809, + "eval_loss": 2.032902240753174, + "eval_runtime": 47.8711, + "eval_samples_per_second": 20.889, + "eval_steps_per_second": 0.877, + "step": 290 + }, + { + "epoch": 0.29892141756548535, + "grad_norm": 3.0068976879119873, + "learning_rate": 4.726027397260274e-06, + "loss": 1.8256, + "step": 291 + }, + { + "epoch": 0.29994863893168977, + "grad_norm": 2.6469006538391113, + "learning_rate": 4.7431506849315075e-06, + "loss": 1.8354, + "step": 292 + }, + { + "epoch": 0.3009758602978942, + "grad_norm": 3.06986927986145, + "learning_rate": 4.76027397260274e-06, + "loss": 1.8402, + "step": 293 + }, + { + "epoch": 0.3020030816640986, + "grad_norm": 2.8409981727600098, + "learning_rate": 4.777397260273973e-06, + "loss": 1.7915, + "step": 294 + }, + { + "epoch": 0.30303030303030304, + "grad_norm": 3.1027376651763916, + "learning_rate": 4.7945205479452054e-06, + "loss": 1.8153, + "step": 295 + }, + { + "epoch": 0.30405752439650746, + "grad_norm": 2.7221434116363525, + "learning_rate": 4.811643835616439e-06, + "loss": 1.8098, + "step": 296 + }, + { + "epoch": 0.3050847457627119, + "grad_norm": 3.0559370517730713, + "learning_rate": 4.8287671232876716e-06, + "loss": 1.7504, + "step": 297 + }, + { + "epoch": 0.3061119671289163, + "grad_norm": 2.9628236293792725, + "learning_rate": 4.845890410958904e-06, + "loss": 1.8019, + "step": 298 + }, + { + "epoch": 0.3071391884951207, + "grad_norm": 2.6248221397399902, + "learning_rate": 4.863013698630138e-06, + "loss": 1.8001, + "step": 299 + }, + { + "epoch": 0.3081664098613251, + "grad_norm": 3.0477702617645264, + "learning_rate": 4.88013698630137e-06, + "loss": 1.7915, + "step": 300 + }, + { + "epoch": 0.3081664098613251, + "eval_loss": 2.0522429943084717, + "eval_runtime": 46.796, + "eval_samples_per_second": 21.369, + "eval_steps_per_second": 0.898, + "step": 300 + }, + { + "epoch": 0.3091936312275295, + "grad_norm": 2.784707546234131, + "learning_rate": 4.897260273972603e-06, + "loss": 1.7757, + "step": 301 + }, + { + "epoch": 0.31022085259373394, + "grad_norm": 2.8283841609954834, + "learning_rate": 4.914383561643836e-06, + "loss": 1.8975, + "step": 302 + }, + { + "epoch": 0.31124807395993837, + "grad_norm": 2.879342555999756, + "learning_rate": 4.931506849315069e-06, + "loss": 1.8208, + "step": 303 + }, + { + "epoch": 0.3122752953261428, + "grad_norm": 3.2111175060272217, + "learning_rate": 4.948630136986302e-06, + "loss": 1.8631, + "step": 304 + }, + { + "epoch": 0.3133025166923472, + "grad_norm": 2.876168966293335, + "learning_rate": 4.965753424657534e-06, + "loss": 1.8863, + "step": 305 + }, + { + "epoch": 0.31432973805855163, + "grad_norm": 2.8958466053009033, + "learning_rate": 4.982876712328768e-06, + "loss": 1.7565, + "step": 306 + }, + { + "epoch": 0.31535695942475606, + "grad_norm": 2.6702723503112793, + "learning_rate": 5e-06, + "loss": 1.7844, + "step": 307 + }, + { + "epoch": 0.3163841807909605, + "grad_norm": 2.557966470718384, + "learning_rate": 4.998096688237534e-06, + "loss": 1.7472, + "step": 308 + }, + { + "epoch": 0.31741140215716485, + "grad_norm": 2.9257633686065674, + "learning_rate": 4.996193376475067e-06, + "loss": 1.853, + "step": 309 + }, + { + "epoch": 0.31843862352336927, + "grad_norm": 2.497093439102173, + "learning_rate": 4.9942900647126005e-06, + "loss": 1.7313, + "step": 310 + }, + { + "epoch": 0.31843862352336927, + "eval_loss": 2.0336520671844482, + "eval_runtime": 43.3773, + "eval_samples_per_second": 23.054, + "eval_steps_per_second": 0.968, + "step": 310 + }, + { + "epoch": 0.3194658448895737, + "grad_norm": 3.0313165187835693, + "learning_rate": 4.992386752950134e-06, + "loss": 1.7397, + "step": 311 + }, + { + "epoch": 0.3204930662557781, + "grad_norm": 2.629685163497925, + "learning_rate": 4.990483441187667e-06, + "loss": 1.7839, + "step": 312 + }, + { + "epoch": 0.32152028762198254, + "grad_norm": 2.6772103309631348, + "learning_rate": 4.9885801294252e-06, + "loss": 1.8038, + "step": 313 + }, + { + "epoch": 0.32254750898818696, + "grad_norm": 2.893958568572998, + "learning_rate": 4.986676817662733e-06, + "loss": 1.8072, + "step": 314 + }, + { + "epoch": 0.3235747303543914, + "grad_norm": 2.5317318439483643, + "learning_rate": 4.9847735059002665e-06, + "loss": 1.7632, + "step": 315 + }, + { + "epoch": 0.3246019517205958, + "grad_norm": 2.836867332458496, + "learning_rate": 4.9828701941378e-06, + "loss": 1.8499, + "step": 316 + }, + { + "epoch": 0.3256291730868002, + "grad_norm": 2.5564308166503906, + "learning_rate": 4.980966882375333e-06, + "loss": 1.7849, + "step": 317 + }, + { + "epoch": 0.3266563944530046, + "grad_norm": 2.4335806369781494, + "learning_rate": 4.979063570612867e-06, + "loss": 1.7425, + "step": 318 + }, + { + "epoch": 0.327683615819209, + "grad_norm": 2.7140371799468994, + "learning_rate": 4.9771602588504e-06, + "loss": 1.7173, + "step": 319 + }, + { + "epoch": 0.32871083718541344, + "grad_norm": 2.574345111846924, + "learning_rate": 4.975256947087933e-06, + "loss": 1.8307, + "step": 320 + }, + { + "epoch": 0.32871083718541344, + "eval_loss": 2.026960611343384, + "eval_runtime": 53.8982, + "eval_samples_per_second": 18.553, + "eval_steps_per_second": 0.779, + "step": 320 + }, + { + "epoch": 0.32973805855161786, + "grad_norm": 2.6877281665802, + "learning_rate": 4.973353635325467e-06, + "loss": 1.7216, + "step": 321 + }, + { + "epoch": 0.3307652799178223, + "grad_norm": 2.8152849674224854, + "learning_rate": 4.971450323563e-06, + "loss": 1.7831, + "step": 322 + }, + { + "epoch": 0.3317925012840267, + "grad_norm": 2.4208791255950928, + "learning_rate": 4.969547011800533e-06, + "loss": 1.7266, + "step": 323 + }, + { + "epoch": 0.33281972265023113, + "grad_norm": 2.765162229537964, + "learning_rate": 4.967643700038067e-06, + "loss": 1.7971, + "step": 324 + }, + { + "epoch": 0.33384694401643555, + "grad_norm": 2.5018413066864014, + "learning_rate": 4.9657403882756e-06, + "loss": 1.7842, + "step": 325 + }, + { + "epoch": 0.33487416538264, + "grad_norm": 2.901268482208252, + "learning_rate": 4.9638370765131335e-06, + "loss": 1.7522, + "step": 326 + }, + { + "epoch": 0.3359013867488444, + "grad_norm": 2.40067195892334, + "learning_rate": 4.961933764750667e-06, + "loss": 1.7371, + "step": 327 + }, + { + "epoch": 0.33692860811504877, + "grad_norm": 2.6963725090026855, + "learning_rate": 4.9600304529882e-06, + "loss": 1.8036, + "step": 328 + }, + { + "epoch": 0.3379558294812532, + "grad_norm": 2.483790397644043, + "learning_rate": 4.9581271412257335e-06, + "loss": 1.815, + "step": 329 + }, + { + "epoch": 0.3389830508474576, + "grad_norm": 2.7459559440612793, + "learning_rate": 4.956223829463266e-06, + "loss": 1.777, + "step": 330 + }, + { + "epoch": 0.3389830508474576, + "eval_loss": 2.046339988708496, + "eval_runtime": 53.909, + "eval_samples_per_second": 18.55, + "eval_steps_per_second": 0.779, + "step": 330 + }, + { + "epoch": 0.34001027221366203, + "grad_norm": 2.5724353790283203, + "learning_rate": 4.954320517700799e-06, + "loss": 1.7022, + "step": 331 + }, + { + "epoch": 0.34103749357986646, + "grad_norm": 2.510002613067627, + "learning_rate": 4.952417205938333e-06, + "loss": 1.802, + "step": 332 + }, + { + "epoch": 0.3420647149460709, + "grad_norm": 2.8722083568573, + "learning_rate": 4.950513894175866e-06, + "loss": 1.7772, + "step": 333 + }, + { + "epoch": 0.3430919363122753, + "grad_norm": 2.586747407913208, + "learning_rate": 4.9486105824133995e-06, + "loss": 1.7664, + "step": 334 + }, + { + "epoch": 0.3441191576784797, + "grad_norm": 2.8008205890655518, + "learning_rate": 4.946707270650933e-06, + "loss": 1.7892, + "step": 335 + }, + { + "epoch": 0.34514637904468415, + "grad_norm": 2.632664203643799, + "learning_rate": 4.944803958888466e-06, + "loss": 1.7967, + "step": 336 + }, + { + "epoch": 0.34617360041088857, + "grad_norm": 2.8898301124572754, + "learning_rate": 4.942900647126e-06, + "loss": 1.8126, + "step": 337 + }, + { + "epoch": 0.34720082177709294, + "grad_norm": 2.719019651412964, + "learning_rate": 4.940997335363533e-06, + "loss": 1.7544, + "step": 338 + }, + { + "epoch": 0.34822804314329736, + "grad_norm": 2.9995970726013184, + "learning_rate": 4.939094023601066e-06, + "loss": 1.7142, + "step": 339 + }, + { + "epoch": 0.3492552645095018, + "grad_norm": 2.9737157821655273, + "learning_rate": 4.9371907118386e-06, + "loss": 1.7659, + "step": 340 + }, + { + "epoch": 0.3492552645095018, + "eval_loss": 2.024946928024292, + "eval_runtime": 52.8069, + "eval_samples_per_second": 18.937, + "eval_steps_per_second": 0.795, + "step": 340 + }, + { + "epoch": 0.3502824858757062, + "grad_norm": 2.8651719093322754, + "learning_rate": 4.935287400076133e-06, + "loss": 1.7767, + "step": 341 + }, + { + "epoch": 0.35130970724191063, + "grad_norm": 2.6293981075286865, + "learning_rate": 4.933384088313666e-06, + "loss": 1.7815, + "step": 342 + }, + { + "epoch": 0.35233692860811505, + "grad_norm": 3.230978012084961, + "learning_rate": 4.9314807765512e-06, + "loss": 1.7423, + "step": 343 + }, + { + "epoch": 0.3533641499743195, + "grad_norm": 3.228300094604492, + "learning_rate": 4.929577464788733e-06, + "loss": 1.6845, + "step": 344 + }, + { + "epoch": 0.3543913713405239, + "grad_norm": 3.036733388900757, + "learning_rate": 4.9276741530262665e-06, + "loss": 1.7524, + "step": 345 + }, + { + "epoch": 0.3554185927067283, + "grad_norm": 2.778538227081299, + "learning_rate": 4.9257708412638e-06, + "loss": 1.7433, + "step": 346 + }, + { + "epoch": 0.35644581407293274, + "grad_norm": 3.0958566665649414, + "learning_rate": 4.923867529501332e-06, + "loss": 1.76, + "step": 347 + }, + { + "epoch": 0.3574730354391371, + "grad_norm": 3.1120026111602783, + "learning_rate": 4.921964217738866e-06, + "loss": 1.7399, + "step": 348 + }, + { + "epoch": 0.35850025680534153, + "grad_norm": 2.70768141746521, + "learning_rate": 4.920060905976399e-06, + "loss": 1.8169, + "step": 349 + }, + { + "epoch": 0.35952747817154596, + "grad_norm": 2.917478561401367, + "learning_rate": 4.9181575942139325e-06, + "loss": 1.7948, + "step": 350 + }, + { + "epoch": 0.35952747817154596, + "eval_loss": 2.0208380222320557, + "eval_runtime": 47.4243, + "eval_samples_per_second": 21.086, + "eval_steps_per_second": 0.886, + "step": 350 + }, + { + "epoch": 0.3605546995377504, + "grad_norm": 3.0564756393432617, + "learning_rate": 4.916254282451466e-06, + "loss": 1.7861, + "step": 351 + }, + { + "epoch": 0.3615819209039548, + "grad_norm": 2.709195137023926, + "learning_rate": 4.914350970688999e-06, + "loss": 1.8323, + "step": 352 + }, + { + "epoch": 0.3626091422701592, + "grad_norm": 2.8797669410705566, + "learning_rate": 4.9124476589265325e-06, + "loss": 1.722, + "step": 353 + }, + { + "epoch": 0.36363636363636365, + "grad_norm": 2.560702085494995, + "learning_rate": 4.910544347164066e-06, + "loss": 1.7289, + "step": 354 + }, + { + "epoch": 0.36466358500256807, + "grad_norm": 3.3781847953796387, + "learning_rate": 4.908641035401599e-06, + "loss": 1.8096, + "step": 355 + }, + { + "epoch": 0.3656908063687725, + "grad_norm": 2.582310676574707, + "learning_rate": 4.906737723639133e-06, + "loss": 1.8752, + "step": 356 + }, + { + "epoch": 0.3667180277349769, + "grad_norm": 3.506385564804077, + "learning_rate": 4.904834411876665e-06, + "loss": 1.7404, + "step": 357 + }, + { + "epoch": 0.3677452491011813, + "grad_norm": 2.4360411167144775, + "learning_rate": 4.902931100114199e-06, + "loss": 1.7812, + "step": 358 + }, + { + "epoch": 0.3687724704673857, + "grad_norm": 3.77639102935791, + "learning_rate": 4.901027788351733e-06, + "loss": 1.7103, + "step": 359 + }, + { + "epoch": 0.3697996918335901, + "grad_norm": 2.887723922729492, + "learning_rate": 4.899124476589266e-06, + "loss": 1.7485, + "step": 360 + }, + { + "epoch": 0.3697996918335901, + "eval_loss": 2.0298168659210205, + "eval_runtime": 46.0993, + "eval_samples_per_second": 21.692, + "eval_steps_per_second": 0.911, + "step": 360 + }, + { + "epoch": 0.37082691319979455, + "grad_norm": 3.5861997604370117, + "learning_rate": 4.8972211648267994e-06, + "loss": 1.7189, + "step": 361 + }, + { + "epoch": 0.371854134565999, + "grad_norm": 2.590808868408203, + "learning_rate": 4.895317853064333e-06, + "loss": 1.7683, + "step": 362 + }, + { + "epoch": 0.3728813559322034, + "grad_norm": 3.4451487064361572, + "learning_rate": 4.893414541301866e-06, + "loss": 1.7436, + "step": 363 + }, + { + "epoch": 0.3739085772984078, + "grad_norm": 2.406073808670044, + "learning_rate": 4.891511229539399e-06, + "loss": 1.7378, + "step": 364 + }, + { + "epoch": 0.37493579866461224, + "grad_norm": 2.7162997722625732, + "learning_rate": 4.889607917776932e-06, + "loss": 1.7558, + "step": 365 + }, + { + "epoch": 0.37596302003081666, + "grad_norm": 2.564753770828247, + "learning_rate": 4.887704606014465e-06, + "loss": 1.7559, + "step": 366 + }, + { + "epoch": 0.37699024139702103, + "grad_norm": 2.7196900844573975, + "learning_rate": 4.885801294251999e-06, + "loss": 1.7683, + "step": 367 + }, + { + "epoch": 0.37801746276322545, + "grad_norm": 2.9334428310394287, + "learning_rate": 4.883897982489532e-06, + "loss": 1.6932, + "step": 368 + }, + { + "epoch": 0.3790446841294299, + "grad_norm": 2.4899954795837402, + "learning_rate": 4.8819946707270655e-06, + "loss": 1.7943, + "step": 369 + }, + { + "epoch": 0.3800719054956343, + "grad_norm": 2.820594310760498, + "learning_rate": 4.880091358964599e-06, + "loss": 1.7729, + "step": 370 + }, + { + "epoch": 0.3800719054956343, + "eval_loss": 2.0204291343688965, + "eval_runtime": 44.9993, + "eval_samples_per_second": 22.223, + "eval_steps_per_second": 0.933, + "step": 370 + }, + { + "epoch": 0.3810991268618387, + "grad_norm": 2.4717929363250732, + "learning_rate": 4.878188047202132e-06, + "loss": 1.6813, + "step": 371 + }, + { + "epoch": 0.38212634822804314, + "grad_norm": 3.1815855503082275, + "learning_rate": 4.876284735439666e-06, + "loss": 1.8588, + "step": 372 + }, + { + "epoch": 0.38315356959424757, + "grad_norm": 2.7330422401428223, + "learning_rate": 4.874381423677199e-06, + "loss": 1.8245, + "step": 373 + }, + { + "epoch": 0.384180790960452, + "grad_norm": 3.129444122314453, + "learning_rate": 4.8724781119147315e-06, + "loss": 1.7494, + "step": 374 + }, + { + "epoch": 0.3852080123266564, + "grad_norm": 2.75551700592041, + "learning_rate": 4.870574800152265e-06, + "loss": 1.7761, + "step": 375 + }, + { + "epoch": 0.38623523369286084, + "grad_norm": 2.9313466548919678, + "learning_rate": 4.868671488389798e-06, + "loss": 1.7076, + "step": 376 + }, + { + "epoch": 0.3872624550590652, + "grad_norm": 2.7590556144714355, + "learning_rate": 4.8667681766273316e-06, + "loss": 1.7105, + "step": 377 + }, + { + "epoch": 0.3882896764252696, + "grad_norm": 2.8224680423736572, + "learning_rate": 4.864864864864866e-06, + "loss": 1.6853, + "step": 378 + }, + { + "epoch": 0.38931689779147405, + "grad_norm": 2.5148062705993652, + "learning_rate": 4.862961553102399e-06, + "loss": 1.7477, + "step": 379 + }, + { + "epoch": 0.39034411915767847, + "grad_norm": 3.144247055053711, + "learning_rate": 4.8610582413399325e-06, + "loss": 1.7726, + "step": 380 + }, + { + "epoch": 0.39034411915767847, + "eval_loss": 2.006993532180786, + "eval_runtime": 50.8909, + "eval_samples_per_second": 19.65, + "eval_steps_per_second": 0.825, + "step": 380 + }, + { + "epoch": 0.3913713405238829, + "grad_norm": 2.6818125247955322, + "learning_rate": 4.859154929577465e-06, + "loss": 1.7351, + "step": 381 + }, + { + "epoch": 0.3923985618900873, + "grad_norm": 2.8543903827667236, + "learning_rate": 4.857251617814998e-06, + "loss": 1.7499, + "step": 382 + }, + { + "epoch": 0.39342578325629174, + "grad_norm": 2.5483179092407227, + "learning_rate": 4.855348306052532e-06, + "loss": 1.7849, + "step": 383 + }, + { + "epoch": 0.39445300462249616, + "grad_norm": 2.7169013023376465, + "learning_rate": 4.853444994290065e-06, + "loss": 1.7664, + "step": 384 + }, + { + "epoch": 0.3954802259887006, + "grad_norm": 2.8304696083068848, + "learning_rate": 4.8515416825275985e-06, + "loss": 1.73, + "step": 385 + }, + { + "epoch": 0.396507447354905, + "grad_norm": 2.4217448234558105, + "learning_rate": 4.849638370765132e-06, + "loss": 1.7915, + "step": 386 + }, + { + "epoch": 0.3975346687211094, + "grad_norm": 2.523256301879883, + "learning_rate": 4.847735059002665e-06, + "loss": 1.7195, + "step": 387 + }, + { + "epoch": 0.3985618900873138, + "grad_norm": 2.813638925552368, + "learning_rate": 4.8458317472401985e-06, + "loss": 1.6766, + "step": 388 + }, + { + "epoch": 0.3995891114535182, + "grad_norm": 2.6701548099517822, + "learning_rate": 4.843928435477732e-06, + "loss": 1.7666, + "step": 389 + }, + { + "epoch": 0.40061633281972264, + "grad_norm": 2.5568342208862305, + "learning_rate": 4.842025123715265e-06, + "loss": 1.723, + "step": 390 + }, + { + "epoch": 0.40061633281972264, + "eval_loss": 2.0098180770874023, + "eval_runtime": 44.4503, + "eval_samples_per_second": 22.497, + "eval_steps_per_second": 0.945, + "step": 390 + }, + { + "epoch": 0.40164355418592707, + "grad_norm": 2.6287341117858887, + "learning_rate": 4.840121811952798e-06, + "loss": 1.759, + "step": 391 + }, + { + "epoch": 0.4026707755521315, + "grad_norm": 2.934105157852173, + "learning_rate": 4.838218500190331e-06, + "loss": 1.7501, + "step": 392 + }, + { + "epoch": 0.4036979969183359, + "grad_norm": 2.6563878059387207, + "learning_rate": 4.8363151884278645e-06, + "loss": 1.7624, + "step": 393 + }, + { + "epoch": 0.40472521828454033, + "grad_norm": 2.538278102874756, + "learning_rate": 4.834411876665398e-06, + "loss": 1.7457, + "step": 394 + }, + { + "epoch": 0.40575243965074476, + "grad_norm": 2.801630735397339, + "learning_rate": 4.832508564902931e-06, + "loss": 1.7549, + "step": 395 + }, + { + "epoch": 0.4067796610169492, + "grad_norm": 2.571937322616577, + "learning_rate": 4.830605253140465e-06, + "loss": 1.6954, + "step": 396 + }, + { + "epoch": 0.40780688238315355, + "grad_norm": 2.5652592182159424, + "learning_rate": 4.828701941377998e-06, + "loss": 1.7328, + "step": 397 + }, + { + "epoch": 0.40883410374935797, + "grad_norm": 2.6840121746063232, + "learning_rate": 4.826798629615531e-06, + "loss": 1.6934, + "step": 398 + }, + { + "epoch": 0.4098613251155624, + "grad_norm": 3.0103495121002197, + "learning_rate": 4.824895317853065e-06, + "loss": 1.7086, + "step": 399 + }, + { + "epoch": 0.4108885464817668, + "grad_norm": 2.7361223697662354, + "learning_rate": 4.822992006090598e-06, + "loss": 1.7301, + "step": 400 + }, + { + "epoch": 0.4108885464817668, + "eval_loss": 1.9975080490112305, + "eval_runtime": 44.5413, + "eval_samples_per_second": 22.451, + "eval_steps_per_second": 0.943, + "step": 400 + }, + { + "epoch": 0.41191576784797124, + "grad_norm": 2.6805901527404785, + "learning_rate": 4.821088694328131e-06, + "loss": 1.6896, + "step": 401 + }, + { + "epoch": 0.41294298921417566, + "grad_norm": 3.231706142425537, + "learning_rate": 4.819185382565665e-06, + "loss": 1.7134, + "step": 402 + }, + { + "epoch": 0.4139702105803801, + "grad_norm": 2.5561299324035645, + "learning_rate": 4.817282070803198e-06, + "loss": 1.764, + "step": 403 + }, + { + "epoch": 0.4149974319465845, + "grad_norm": 3.3125340938568115, + "learning_rate": 4.8153787590407315e-06, + "loss": 1.76, + "step": 404 + }, + { + "epoch": 0.41602465331278893, + "grad_norm": 2.750051498413086, + "learning_rate": 4.813475447278265e-06, + "loss": 1.6882, + "step": 405 + }, + { + "epoch": 0.41705187467899335, + "grad_norm": 4.006515979766846, + "learning_rate": 4.811572135515798e-06, + "loss": 1.7892, + "step": 406 + }, + { + "epoch": 0.4180790960451977, + "grad_norm": 2.5276904106140137, + "learning_rate": 4.809668823753332e-06, + "loss": 1.6995, + "step": 407 + }, + { + "epoch": 0.41910631741140214, + "grad_norm": 2.959444046020508, + "learning_rate": 4.807765511990864e-06, + "loss": 1.7058, + "step": 408 + }, + { + "epoch": 0.42013353877760656, + "grad_norm": 2.8472464084625244, + "learning_rate": 4.8058622002283975e-06, + "loss": 1.7258, + "step": 409 + }, + { + "epoch": 0.421160760143811, + "grad_norm": 3.0240273475646973, + "learning_rate": 4.803958888465931e-06, + "loss": 1.7789, + "step": 410 + }, + { + "epoch": 0.421160760143811, + "eval_loss": 1.9919871091842651, + "eval_runtime": 41.727, + "eval_samples_per_second": 23.965, + "eval_steps_per_second": 1.007, + "step": 410 + }, + { + "epoch": 0.4221879815100154, + "grad_norm": 2.6211893558502197, + "learning_rate": 4.802055576703464e-06, + "loss": 1.7315, + "step": 411 + }, + { + "epoch": 0.42321520287621983, + "grad_norm": 2.6072895526885986, + "learning_rate": 4.8001522649409976e-06, + "loss": 1.8176, + "step": 412 + }, + { + "epoch": 0.42424242424242425, + "grad_norm": 2.9197702407836914, + "learning_rate": 4.798248953178531e-06, + "loss": 1.6947, + "step": 413 + }, + { + "epoch": 0.4252696456086287, + "grad_norm": 2.4383556842803955, + "learning_rate": 4.796345641416064e-06, + "loss": 1.6728, + "step": 414 + }, + { + "epoch": 0.4262968669748331, + "grad_norm": 2.9951915740966797, + "learning_rate": 4.794442329653598e-06, + "loss": 1.7249, + "step": 415 + }, + { + "epoch": 0.42732408834103747, + "grad_norm": 2.6095519065856934, + "learning_rate": 4.792539017891131e-06, + "loss": 1.7735, + "step": 416 + }, + { + "epoch": 0.4283513097072419, + "grad_norm": 2.6429390907287598, + "learning_rate": 4.790635706128664e-06, + "loss": 1.704, + "step": 417 + }, + { + "epoch": 0.4293785310734463, + "grad_norm": 2.921812057495117, + "learning_rate": 4.788732394366197e-06, + "loss": 1.7355, + "step": 418 + }, + { + "epoch": 0.43040575243965074, + "grad_norm": 2.6050965785980225, + "learning_rate": 4.786829082603731e-06, + "loss": 1.6636, + "step": 419 + }, + { + "epoch": 0.43143297380585516, + "grad_norm": 2.672070264816284, + "learning_rate": 4.7849257708412645e-06, + "loss": 1.6627, + "step": 420 + }, + { + "epoch": 0.43143297380585516, + "eval_loss": 1.9990200996398926, + "eval_runtime": 45.0672, + "eval_samples_per_second": 22.189, + "eval_steps_per_second": 0.932, + "step": 420 + }, + { + "epoch": 0.4324601951720596, + "grad_norm": 2.6813488006591797, + "learning_rate": 4.783022459078798e-06, + "loss": 1.7453, + "step": 421 + }, + { + "epoch": 0.433487416538264, + "grad_norm": 2.6587321758270264, + "learning_rate": 4.781119147316331e-06, + "loss": 1.6795, + "step": 422 + }, + { + "epoch": 0.4345146379044684, + "grad_norm": 2.827758312225342, + "learning_rate": 4.7792158355538645e-06, + "loss": 1.7098, + "step": 423 + }, + { + "epoch": 0.43554185927067285, + "grad_norm": 2.8615901470184326, + "learning_rate": 4.777312523791398e-06, + "loss": 1.7171, + "step": 424 + }, + { + "epoch": 0.43656908063687727, + "grad_norm": 2.9509294033050537, + "learning_rate": 4.77540921202893e-06, + "loss": 1.7018, + "step": 425 + }, + { + "epoch": 0.43759630200308164, + "grad_norm": 2.7994418144226074, + "learning_rate": 4.773505900266464e-06, + "loss": 1.7328, + "step": 426 + }, + { + "epoch": 0.43862352336928606, + "grad_norm": 2.7476799488067627, + "learning_rate": 4.771602588503997e-06, + "loss": 1.7116, + "step": 427 + }, + { + "epoch": 0.4396507447354905, + "grad_norm": 2.7955048084259033, + "learning_rate": 4.7696992767415305e-06, + "loss": 1.7187, + "step": 428 + }, + { + "epoch": 0.4406779661016949, + "grad_norm": 2.4326071739196777, + "learning_rate": 4.767795964979064e-06, + "loss": 1.75, + "step": 429 + }, + { + "epoch": 0.44170518746789933, + "grad_norm": 2.7708163261413574, + "learning_rate": 4.765892653216597e-06, + "loss": 1.7562, + "step": 430 + }, + { + "epoch": 0.44170518746789933, + "eval_loss": 2.0262997150421143, + "eval_runtime": 46.1649, + "eval_samples_per_second": 21.661, + "eval_steps_per_second": 0.91, + "step": 430 + }, + { + "epoch": 0.44273240883410375, + "grad_norm": 2.872743606567383, + "learning_rate": 4.763989341454131e-06, + "loss": 1.7527, + "step": 431 + }, + { + "epoch": 0.4437596302003082, + "grad_norm": 2.6387035846710205, + "learning_rate": 4.762086029691664e-06, + "loss": 1.6742, + "step": 432 + }, + { + "epoch": 0.4447868515665126, + "grad_norm": 2.6522979736328125, + "learning_rate": 4.760182717929197e-06, + "loss": 1.6811, + "step": 433 + }, + { + "epoch": 0.445814072932717, + "grad_norm": 2.644155263900757, + "learning_rate": 4.758279406166731e-06, + "loss": 1.7083, + "step": 434 + }, + { + "epoch": 0.44684129429892144, + "grad_norm": 2.530224561691284, + "learning_rate": 4.756376094404263e-06, + "loss": 1.6318, + "step": 435 + }, + { + "epoch": 0.4478685156651258, + "grad_norm": 2.431049108505249, + "learning_rate": 4.7544727826417966e-06, + "loss": 1.7162, + "step": 436 + }, + { + "epoch": 0.44889573703133023, + "grad_norm": 2.8091282844543457, + "learning_rate": 4.75256947087933e-06, + "loss": 1.7717, + "step": 437 + }, + { + "epoch": 0.44992295839753466, + "grad_norm": 2.5520501136779785, + "learning_rate": 4.750666159116863e-06, + "loss": 1.7201, + "step": 438 + }, + { + "epoch": 0.4509501797637391, + "grad_norm": 2.596370220184326, + "learning_rate": 4.7487628473543975e-06, + "loss": 1.6882, + "step": 439 + }, + { + "epoch": 0.4519774011299435, + "grad_norm": 2.667034387588501, + "learning_rate": 4.746859535591931e-06, + "loss": 1.6499, + "step": 440 + }, + { + "epoch": 0.4519774011299435, + "eval_loss": 2.008521556854248, + "eval_runtime": 49.3626, + "eval_samples_per_second": 20.258, + "eval_steps_per_second": 0.851, + "step": 440 + }, + { + "epoch": 0.4530046224961479, + "grad_norm": 2.636491537094116, + "learning_rate": 4.744956223829464e-06, + "loss": 1.6987, + "step": 441 + }, + { + "epoch": 0.45403184386235235, + "grad_norm": 2.4653050899505615, + "learning_rate": 4.743052912066997e-06, + "loss": 1.6775, + "step": 442 + }, + { + "epoch": 0.45505906522855677, + "grad_norm": 2.3866467475891113, + "learning_rate": 4.74114960030453e-06, + "loss": 1.7234, + "step": 443 + }, + { + "epoch": 0.4560862865947612, + "grad_norm": 2.675262928009033, + "learning_rate": 4.7392462885420635e-06, + "loss": 1.7363, + "step": 444 + }, + { + "epoch": 0.4571135079609656, + "grad_norm": 2.6233296394348145, + "learning_rate": 4.737342976779597e-06, + "loss": 1.6236, + "step": 445 + }, + { + "epoch": 0.45814072932717, + "grad_norm": 2.5638058185577393, + "learning_rate": 4.73543966501713e-06, + "loss": 1.7097, + "step": 446 + }, + { + "epoch": 0.4591679506933744, + "grad_norm": 2.7206926345825195, + "learning_rate": 4.7335363532546636e-06, + "loss": 1.6909, + "step": 447 + }, + { + "epoch": 0.4601951720595788, + "grad_norm": 2.546241283416748, + "learning_rate": 4.731633041492197e-06, + "loss": 1.6951, + "step": 448 + }, + { + "epoch": 0.46122239342578325, + "grad_norm": 2.6549105644226074, + "learning_rate": 4.72972972972973e-06, + "loss": 1.7052, + "step": 449 + }, + { + "epoch": 0.4622496147919877, + "grad_norm": 2.6649065017700195, + "learning_rate": 4.727826417967264e-06, + "loss": 1.7406, + "step": 450 + }, + { + "epoch": 0.4622496147919877, + "eval_loss": 1.9886436462402344, + "eval_runtime": 47.48, + "eval_samples_per_second": 21.061, + "eval_steps_per_second": 0.885, + "step": 450 + }, + { + "epoch": 0.4632768361581921, + "grad_norm": 2.7138187885284424, + "learning_rate": 4.725923106204797e-06, + "loss": 1.7154, + "step": 451 + }, + { + "epoch": 0.4643040575243965, + "grad_norm": 2.6095476150512695, + "learning_rate": 4.72401979444233e-06, + "loss": 1.686, + "step": 452 + }, + { + "epoch": 0.46533127889060094, + "grad_norm": 2.866363048553467, + "learning_rate": 4.722116482679863e-06, + "loss": 1.721, + "step": 453 + }, + { + "epoch": 0.46635850025680536, + "grad_norm": 2.4947352409362793, + "learning_rate": 4.720213170917396e-06, + "loss": 1.708, + "step": 454 + }, + { + "epoch": 0.46738572162300973, + "grad_norm": 2.3888256549835205, + "learning_rate": 4.71830985915493e-06, + "loss": 1.7356, + "step": 455 + }, + { + "epoch": 0.46841294298921415, + "grad_norm": 2.7065343856811523, + "learning_rate": 4.716406547392463e-06, + "loss": 1.6943, + "step": 456 + }, + { + "epoch": 0.4694401643554186, + "grad_norm": 2.424285888671875, + "learning_rate": 4.714503235629996e-06, + "loss": 1.7385, + "step": 457 + }, + { + "epoch": 0.470467385721623, + "grad_norm": 2.4175925254821777, + "learning_rate": 4.71259992386753e-06, + "loss": 1.6654, + "step": 458 + }, + { + "epoch": 0.4714946070878274, + "grad_norm": 2.3203656673431396, + "learning_rate": 4.710696612105063e-06, + "loss": 1.6977, + "step": 459 + }, + { + "epoch": 0.47252182845403184, + "grad_norm": 2.6660172939300537, + "learning_rate": 4.708793300342596e-06, + "loss": 1.7055, + "step": 460 + }, + { + "epoch": 0.47252182845403184, + "eval_loss": 1.9864659309387207, + "eval_runtime": 47.6232, + "eval_samples_per_second": 20.998, + "eval_steps_per_second": 0.882, + "step": 460 + }, + { + "epoch": 0.47354904982023627, + "grad_norm": 2.515615701675415, + "learning_rate": 4.70688998858013e-06, + "loss": 1.689, + "step": 461 + }, + { + "epoch": 0.4745762711864407, + "grad_norm": 2.429264783859253, + "learning_rate": 4.704986676817663e-06, + "loss": 1.6796, + "step": 462 + }, + { + "epoch": 0.4756034925526451, + "grad_norm": 2.469543933868408, + "learning_rate": 4.7030833650551965e-06, + "loss": 1.7164, + "step": 463 + }, + { + "epoch": 0.47663071391884954, + "grad_norm": 2.5921683311462402, + "learning_rate": 4.70118005329273e-06, + "loss": 1.6746, + "step": 464 + }, + { + "epoch": 0.4776579352850539, + "grad_norm": 2.6612918376922607, + "learning_rate": 4.699276741530263e-06, + "loss": 1.7598, + "step": 465 + }, + { + "epoch": 0.4786851566512583, + "grad_norm": 2.4633305072784424, + "learning_rate": 4.697373429767797e-06, + "loss": 1.7241, + "step": 466 + }, + { + "epoch": 0.47971237801746275, + "grad_norm": 2.613023281097412, + "learning_rate": 4.69547011800533e-06, + "loss": 1.7088, + "step": 467 + }, + { + "epoch": 0.48073959938366717, + "grad_norm": 2.5329010486602783, + "learning_rate": 4.693566806242863e-06, + "loss": 1.7163, + "step": 468 + }, + { + "epoch": 0.4817668207498716, + "grad_norm": 2.7887041568756104, + "learning_rate": 4.691663494480397e-06, + "loss": 1.6969, + "step": 469 + }, + { + "epoch": 0.482794042116076, + "grad_norm": 2.72991943359375, + "learning_rate": 4.689760182717929e-06, + "loss": 1.7332, + "step": 470 + }, + { + "epoch": 0.482794042116076, + "eval_loss": 1.965472936630249, + "eval_runtime": 46.1059, + "eval_samples_per_second": 21.689, + "eval_steps_per_second": 0.911, + "step": 470 + }, + { + "epoch": 0.48382126348228044, + "grad_norm": 2.613785982131958, + "learning_rate": 4.6878568709554626e-06, + "loss": 1.6272, + "step": 471 + }, + { + "epoch": 0.48484848484848486, + "grad_norm": 2.605299472808838, + "learning_rate": 4.685953559192996e-06, + "loss": 1.6966, + "step": 472 + }, + { + "epoch": 0.4858757062146893, + "grad_norm": 2.672013998031616, + "learning_rate": 4.684050247430529e-06, + "loss": 1.7132, + "step": 473 + }, + { + "epoch": 0.4869029275808937, + "grad_norm": 2.576486825942993, + "learning_rate": 4.682146935668063e-06, + "loss": 1.6449, + "step": 474 + }, + { + "epoch": 0.4879301489470981, + "grad_norm": 2.626840114593506, + "learning_rate": 4.680243623905596e-06, + "loss": 1.6993, + "step": 475 + }, + { + "epoch": 0.4889573703133025, + "grad_norm": 2.974418878555298, + "learning_rate": 4.678340312143129e-06, + "loss": 1.7116, + "step": 476 + }, + { + "epoch": 0.4899845916795069, + "grad_norm": 2.541242837905884, + "learning_rate": 4.676437000380663e-06, + "loss": 1.6693, + "step": 477 + }, + { + "epoch": 0.49101181304571134, + "grad_norm": 2.5494985580444336, + "learning_rate": 4.674533688618196e-06, + "loss": 1.7819, + "step": 478 + }, + { + "epoch": 0.49203903441191577, + "grad_norm": 2.606137990951538, + "learning_rate": 4.6726303768557295e-06, + "loss": 1.7278, + "step": 479 + }, + { + "epoch": 0.4930662557781202, + "grad_norm": 2.5032176971435547, + "learning_rate": 4.670727065093263e-06, + "loss": 1.6318, + "step": 480 + }, + { + "epoch": 0.4930662557781202, + "eval_loss": 1.9851018190383911, + "eval_runtime": 46.0386, + "eval_samples_per_second": 21.721, + "eval_steps_per_second": 0.912, + "step": 480 + }, + { + "epoch": 0.4940934771443246, + "grad_norm": 2.703259229660034, + "learning_rate": 4.668823753330796e-06, + "loss": 1.6646, + "step": 481 + }, + { + "epoch": 0.49512069851052903, + "grad_norm": 2.681807041168213, + "learning_rate": 4.6669204415683295e-06, + "loss": 1.6759, + "step": 482 + }, + { + "epoch": 0.49614791987673346, + "grad_norm": 2.7280473709106445, + "learning_rate": 4.665017129805863e-06, + "loss": 1.7388, + "step": 483 + }, + { + "epoch": 0.4971751412429379, + "grad_norm": 2.420909881591797, + "learning_rate": 4.663113818043396e-06, + "loss": 1.729, + "step": 484 + }, + { + "epoch": 0.49820236260914225, + "grad_norm": 2.5764737129211426, + "learning_rate": 4.66121050628093e-06, + "loss": 1.6903, + "step": 485 + }, + { + "epoch": 0.49922958397534667, + "grad_norm": 2.389531135559082, + "learning_rate": 4.659307194518463e-06, + "loss": 1.7191, + "step": 486 + }, + { + "epoch": 0.5002568053415511, + "grad_norm": 2.5207972526550293, + "learning_rate": 4.6574038827559955e-06, + "loss": 1.6892, + "step": 487 + }, + { + "epoch": 0.5012840267077555, + "grad_norm": 2.5662360191345215, + "learning_rate": 4.655500570993529e-06, + "loss": 1.6978, + "step": 488 + }, + { + "epoch": 0.50231124807396, + "grad_norm": 2.50262713432312, + "learning_rate": 4.653597259231062e-06, + "loss": 1.7379, + "step": 489 + }, + { + "epoch": 0.5033384694401644, + "grad_norm": 2.56643009185791, + "learning_rate": 4.651693947468596e-06, + "loss": 1.6233, + "step": 490 + }, + { + "epoch": 0.5033384694401644, + "eval_loss": 1.9815036058425903, + "eval_runtime": 48.0257, + "eval_samples_per_second": 20.822, + "eval_steps_per_second": 0.875, + "step": 490 + }, + { + "epoch": 0.5043656908063687, + "grad_norm": 2.504117965698242, + "learning_rate": 4.649790635706129e-06, + "loss": 1.7664, + "step": 491 + }, + { + "epoch": 0.5053929121725732, + "grad_norm": 2.906843423843384, + "learning_rate": 4.647887323943662e-06, + "loss": 1.6731, + "step": 492 + }, + { + "epoch": 0.5064201335387776, + "grad_norm": 2.586747169494629, + "learning_rate": 4.645984012181196e-06, + "loss": 1.68, + "step": 493 + }, + { + "epoch": 0.507447354904982, + "grad_norm": 2.3300511837005615, + "learning_rate": 4.644080700418729e-06, + "loss": 1.5998, + "step": 494 + }, + { + "epoch": 0.5084745762711864, + "grad_norm": 2.8711049556732178, + "learning_rate": 4.642177388656262e-06, + "loss": 1.7198, + "step": 495 + }, + { + "epoch": 0.5095017976373909, + "grad_norm": 2.6463162899017334, + "learning_rate": 4.640274076893796e-06, + "loss": 1.6814, + "step": 496 + }, + { + "epoch": 0.5105290190035953, + "grad_norm": 2.650481700897217, + "learning_rate": 4.638370765131328e-06, + "loss": 1.7353, + "step": 497 + }, + { + "epoch": 0.5115562403697997, + "grad_norm": 2.4464986324310303, + "learning_rate": 4.636467453368862e-06, + "loss": 1.6579, + "step": 498 + }, + { + "epoch": 0.5125834617360041, + "grad_norm": 2.438021659851074, + "learning_rate": 4.634564141606396e-06, + "loss": 1.6485, + "step": 499 + }, + { + "epoch": 0.5136106831022085, + "grad_norm": 2.5354347229003906, + "learning_rate": 4.632660829843929e-06, + "loss": 1.7123, + "step": 500 + }, + { + "epoch": 0.5136106831022085, + "eval_loss": 1.9738385677337646, + "eval_runtime": 52.5306, + "eval_samples_per_second": 19.037, + "eval_steps_per_second": 0.8, + "step": 500 + }, + { + "epoch": 0.514637904468413, + "grad_norm": 2.3702070713043213, + "learning_rate": 4.630757518081463e-06, + "loss": 1.6423, + "step": 501 + }, + { + "epoch": 0.5156651258346173, + "grad_norm": 2.5336198806762695, + "learning_rate": 4.628854206318996e-06, + "loss": 1.771, + "step": 502 + }, + { + "epoch": 0.5166923472008218, + "grad_norm": 2.572751998901367, + "learning_rate": 4.626950894556529e-06, + "loss": 1.677, + "step": 503 + }, + { + "epoch": 0.5177195685670262, + "grad_norm": 2.3865134716033936, + "learning_rate": 4.625047582794062e-06, + "loss": 1.7111, + "step": 504 + }, + { + "epoch": 0.5187467899332306, + "grad_norm": 2.5152785778045654, + "learning_rate": 4.623144271031595e-06, + "loss": 1.6343, + "step": 505 + }, + { + "epoch": 0.519774011299435, + "grad_norm": 2.5121307373046875, + "learning_rate": 4.6212409592691286e-06, + "loss": 1.6357, + "step": 506 + }, + { + "epoch": 0.5208012326656395, + "grad_norm": 2.506908655166626, + "learning_rate": 4.619337647506662e-06, + "loss": 1.7295, + "step": 507 + }, + { + "epoch": 0.5218284540318439, + "grad_norm": 2.636082172393799, + "learning_rate": 4.617434335744195e-06, + "loss": 1.6916, + "step": 508 + }, + { + "epoch": 0.5228556753980482, + "grad_norm": 2.5987415313720703, + "learning_rate": 4.615531023981729e-06, + "loss": 1.659, + "step": 509 + }, + { + "epoch": 0.5238828967642527, + "grad_norm": 2.642770290374756, + "learning_rate": 4.613627712219262e-06, + "loss": 1.6789, + "step": 510 + }, + { + "epoch": 0.5238828967642527, + "eval_loss": 1.9659523963928223, + "eval_runtime": 42.9453, + "eval_samples_per_second": 23.285, + "eval_steps_per_second": 0.978, + "step": 510 + }, + { + "epoch": 0.5249101181304571, + "grad_norm": 2.438389301300049, + "learning_rate": 4.611724400456795e-06, + "loss": 1.68, + "step": 511 + }, + { + "epoch": 0.5259373394966615, + "grad_norm": 2.536546468734741, + "learning_rate": 4.609821088694329e-06, + "loss": 1.6847, + "step": 512 + }, + { + "epoch": 0.5269645608628659, + "grad_norm": 2.4056830406188965, + "learning_rate": 4.607917776931862e-06, + "loss": 1.6577, + "step": 513 + }, + { + "epoch": 0.5279917822290704, + "grad_norm": 2.81195330619812, + "learning_rate": 4.606014465169395e-06, + "loss": 1.7149, + "step": 514 + }, + { + "epoch": 0.5290190035952748, + "grad_norm": 2.684948205947876, + "learning_rate": 4.604111153406928e-06, + "loss": 1.717, + "step": 515 + }, + { + "epoch": 0.5300462249614792, + "grad_norm": 2.583092212677002, + "learning_rate": 4.602207841644461e-06, + "loss": 1.6238, + "step": 516 + }, + { + "epoch": 0.5310734463276836, + "grad_norm": 3.022528886795044, + "learning_rate": 4.600304529881995e-06, + "loss": 1.7262, + "step": 517 + }, + { + "epoch": 0.5321006676938881, + "grad_norm": 2.6941866874694824, + "learning_rate": 4.598401218119528e-06, + "loss": 1.7055, + "step": 518 + }, + { + "epoch": 0.5331278890600925, + "grad_norm": 2.5207839012145996, + "learning_rate": 4.5964979063570614e-06, + "loss": 1.7117, + "step": 519 + }, + { + "epoch": 0.5341551104262968, + "grad_norm": 2.8056678771972656, + "learning_rate": 4.594594594594596e-06, + "loss": 1.657, + "step": 520 + }, + { + "epoch": 0.5341551104262968, + "eval_loss": 1.9712116718292236, + "eval_runtime": 49.57, + "eval_samples_per_second": 20.173, + "eval_steps_per_second": 0.847, + "step": 520 + }, + { + "epoch": 0.5351823317925013, + "grad_norm": 2.6119930744171143, + "learning_rate": 4.592691282832128e-06, + "loss": 1.6897, + "step": 521 + }, + { + "epoch": 0.5362095531587057, + "grad_norm": 2.7165610790252686, + "learning_rate": 4.5907879710696615e-06, + "loss": 1.7038, + "step": 522 + }, + { + "epoch": 0.5372367745249101, + "grad_norm": 2.3613736629486084, + "learning_rate": 4.588884659307195e-06, + "loss": 1.6222, + "step": 523 + }, + { + "epoch": 0.5382639958911145, + "grad_norm": 2.647672176361084, + "learning_rate": 4.586981347544728e-06, + "loss": 1.702, + "step": 524 + }, + { + "epoch": 0.539291217257319, + "grad_norm": 2.704634428024292, + "learning_rate": 4.585078035782262e-06, + "loss": 1.6566, + "step": 525 + }, + { + "epoch": 0.5403184386235234, + "grad_norm": 2.6634597778320312, + "learning_rate": 4.583174724019795e-06, + "loss": 1.6903, + "step": 526 + }, + { + "epoch": 0.5413456599897278, + "grad_norm": 2.5745162963867188, + "learning_rate": 4.581271412257328e-06, + "loss": 1.6375, + "step": 527 + }, + { + "epoch": 0.5423728813559322, + "grad_norm": 2.5577778816223145, + "learning_rate": 4.579368100494862e-06, + "loss": 1.614, + "step": 528 + }, + { + "epoch": 0.5434001027221366, + "grad_norm": 2.7684454917907715, + "learning_rate": 4.577464788732395e-06, + "loss": 1.7168, + "step": 529 + }, + { + "epoch": 0.544427324088341, + "grad_norm": 2.582108736038208, + "learning_rate": 4.575561476969928e-06, + "loss": 1.6897, + "step": 530 + }, + { + "epoch": 0.544427324088341, + "eval_loss": 1.9821091890335083, + "eval_runtime": 50.9409, + "eval_samples_per_second": 19.631, + "eval_steps_per_second": 0.824, + "step": 530 + }, + { + "epoch": 0.5454545454545454, + "grad_norm": 2.516486406326294, + "learning_rate": 4.573658165207461e-06, + "loss": 1.7273, + "step": 531 + }, + { + "epoch": 0.5464817668207499, + "grad_norm": 2.786806344985962, + "learning_rate": 4.571754853444994e-06, + "loss": 1.6898, + "step": 532 + }, + { + "epoch": 0.5475089881869543, + "grad_norm": 2.3137319087982178, + "learning_rate": 4.569851541682528e-06, + "loss": 1.676, + "step": 533 + }, + { + "epoch": 0.5485362095531587, + "grad_norm": 2.7268903255462646, + "learning_rate": 4.567948229920061e-06, + "loss": 1.6483, + "step": 534 + }, + { + "epoch": 0.5495634309193631, + "grad_norm": 2.4385902881622314, + "learning_rate": 4.566044918157594e-06, + "loss": 1.6172, + "step": 535 + }, + { + "epoch": 0.5505906522855676, + "grad_norm": 2.420576333999634, + "learning_rate": 4.564141606395128e-06, + "loss": 1.6569, + "step": 536 + }, + { + "epoch": 0.551617873651772, + "grad_norm": 2.592290163040161, + "learning_rate": 4.562238294632661e-06, + "loss": 1.7085, + "step": 537 + }, + { + "epoch": 0.5526450950179764, + "grad_norm": 2.485285520553589, + "learning_rate": 4.5603349828701945e-06, + "loss": 1.6699, + "step": 538 + }, + { + "epoch": 0.5536723163841808, + "grad_norm": 2.844710111618042, + "learning_rate": 4.558431671107728e-06, + "loss": 1.6873, + "step": 539 + }, + { + "epoch": 0.5546995377503852, + "grad_norm": 2.5279181003570557, + "learning_rate": 4.556528359345261e-06, + "loss": 1.7285, + "step": 540 + }, + { + "epoch": 0.5546995377503852, + "eval_loss": 1.9825981855392456, + "eval_runtime": 52.461, + "eval_samples_per_second": 19.062, + "eval_steps_per_second": 0.801, + "step": 540 + }, + { + "epoch": 0.5557267591165896, + "grad_norm": 2.673997163772583, + "learning_rate": 4.5546250475827946e-06, + "loss": 1.6658, + "step": 541 + }, + { + "epoch": 0.556753980482794, + "grad_norm": 2.6023194789886475, + "learning_rate": 4.552721735820328e-06, + "loss": 1.6262, + "step": 542 + }, + { + "epoch": 0.5577812018489985, + "grad_norm": 2.6327662467956543, + "learning_rate": 4.550818424057861e-06, + "loss": 1.6574, + "step": 543 + }, + { + "epoch": 0.5588084232152029, + "grad_norm": 2.5532140731811523, + "learning_rate": 4.548915112295395e-06, + "loss": 1.6607, + "step": 544 + }, + { + "epoch": 0.5598356445814073, + "grad_norm": 2.523966073989868, + "learning_rate": 4.547011800532928e-06, + "loss": 1.6752, + "step": 545 + }, + { + "epoch": 0.5608628659476117, + "grad_norm": 2.6338107585906982, + "learning_rate": 4.545108488770461e-06, + "loss": 1.6258, + "step": 546 + }, + { + "epoch": 0.5618900873138162, + "grad_norm": 2.500096559524536, + "learning_rate": 4.543205177007995e-06, + "loss": 1.7189, + "step": 547 + }, + { + "epoch": 0.5629173086800205, + "grad_norm": 2.644413471221924, + "learning_rate": 4.541301865245527e-06, + "loss": 1.6594, + "step": 548 + }, + { + "epoch": 0.5639445300462249, + "grad_norm": 2.451965093612671, + "learning_rate": 4.539398553483061e-06, + "loss": 1.6984, + "step": 549 + }, + { + "epoch": 0.5649717514124294, + "grad_norm": 2.7554423809051514, + "learning_rate": 4.537495241720594e-06, + "loss": 1.6626, + "step": 550 + }, + { + "epoch": 0.5649717514124294, + "eval_loss": 1.9897041320800781, + "eval_runtime": 45.9349, + "eval_samples_per_second": 21.77, + "eval_steps_per_second": 0.914, + "step": 550 + }, + { + "epoch": 0.5659989727786338, + "grad_norm": 2.5962674617767334, + "learning_rate": 4.535591929958127e-06, + "loss": 1.7015, + "step": 551 + }, + { + "epoch": 0.5670261941448382, + "grad_norm": 2.7917397022247314, + "learning_rate": 4.533688618195661e-06, + "loss": 1.6811, + "step": 552 + }, + { + "epoch": 0.5680534155110426, + "grad_norm": 2.4205498695373535, + "learning_rate": 4.531785306433194e-06, + "loss": 1.6659, + "step": 553 + }, + { + "epoch": 0.5690806368772471, + "grad_norm": 2.74898624420166, + "learning_rate": 4.529881994670727e-06, + "loss": 1.6521, + "step": 554 + }, + { + "epoch": 0.5701078582434514, + "grad_norm": 2.859898328781128, + "learning_rate": 4.527978682908261e-06, + "loss": 1.6445, + "step": 555 + }, + { + "epoch": 0.5711350796096559, + "grad_norm": 2.5526154041290283, + "learning_rate": 4.526075371145794e-06, + "loss": 1.6679, + "step": 556 + }, + { + "epoch": 0.5721623009758603, + "grad_norm": 2.96175217628479, + "learning_rate": 4.5241720593833275e-06, + "loss": 1.6784, + "step": 557 + }, + { + "epoch": 0.5731895223420647, + "grad_norm": 2.5037007331848145, + "learning_rate": 4.52226874762086e-06, + "loss": 1.6689, + "step": 558 + }, + { + "epoch": 0.5742167437082691, + "grad_norm": 2.6513259410858154, + "learning_rate": 4.520365435858393e-06, + "loss": 1.7038, + "step": 559 + }, + { + "epoch": 0.5752439650744735, + "grad_norm": 2.762770175933838, + "learning_rate": 4.518462124095928e-06, + "loss": 1.728, + "step": 560 + }, + { + "epoch": 0.5752439650744735, + "eval_loss": 1.972934603691101, + "eval_runtime": 44.2572, + "eval_samples_per_second": 22.595, + "eval_steps_per_second": 0.949, + "step": 560 + }, + { + "epoch": 0.576271186440678, + "grad_norm": 2.637406349182129, + "learning_rate": 4.516558812333461e-06, + "loss": 1.682, + "step": 561 + }, + { + "epoch": 0.5772984078068824, + "grad_norm": 2.545583724975586, + "learning_rate": 4.514655500570994e-06, + "loss": 1.6669, + "step": 562 + }, + { + "epoch": 0.5783256291730868, + "grad_norm": 2.5655033588409424, + "learning_rate": 4.512752188808528e-06, + "loss": 1.7279, + "step": 563 + }, + { + "epoch": 0.5793528505392912, + "grad_norm": 2.6856470108032227, + "learning_rate": 4.510848877046061e-06, + "loss": 1.7331, + "step": 564 + }, + { + "epoch": 0.5803800719054957, + "grad_norm": 2.605252981185913, + "learning_rate": 4.5089455652835936e-06, + "loss": 1.6904, + "step": 565 + }, + { + "epoch": 0.5814072932717, + "grad_norm": 2.610267400741577, + "learning_rate": 4.507042253521127e-06, + "loss": 1.6607, + "step": 566 + }, + { + "epoch": 0.5824345146379045, + "grad_norm": 2.6955432891845703, + "learning_rate": 4.50513894175866e-06, + "loss": 1.6166, + "step": 567 + }, + { + "epoch": 0.5834617360041089, + "grad_norm": 2.5661842823028564, + "learning_rate": 4.503235629996194e-06, + "loss": 1.6955, + "step": 568 + }, + { + "epoch": 0.5844889573703133, + "grad_norm": 2.6393115520477295, + "learning_rate": 4.501332318233727e-06, + "loss": 1.6933, + "step": 569 + }, + { + "epoch": 0.5855161787365177, + "grad_norm": 2.612645387649536, + "learning_rate": 4.49942900647126e-06, + "loss": 1.6585, + "step": 570 + }, + { + "epoch": 0.5855161787365177, + "eval_loss": 1.9599064588546753, + "eval_runtime": 46.0676, + "eval_samples_per_second": 21.707, + "eval_steps_per_second": 0.912, + "step": 570 + }, + { + "epoch": 0.5865434001027221, + "grad_norm": 2.4182891845703125, + "learning_rate": 4.497525694708794e-06, + "loss": 1.707, + "step": 571 + }, + { + "epoch": 0.5875706214689266, + "grad_norm": 2.7214412689208984, + "learning_rate": 4.495622382946327e-06, + "loss": 1.6966, + "step": 572 + }, + { + "epoch": 0.588597842835131, + "grad_norm": 2.5138018131256104, + "learning_rate": 4.4937190711838605e-06, + "loss": 1.6373, + "step": 573 + }, + { + "epoch": 0.5896250642013354, + "grad_norm": 2.5814292430877686, + "learning_rate": 4.491815759421394e-06, + "loss": 1.7313, + "step": 574 + }, + { + "epoch": 0.5906522855675398, + "grad_norm": 2.6236367225646973, + "learning_rate": 4.489912447658926e-06, + "loss": 1.6647, + "step": 575 + }, + { + "epoch": 0.5916795069337443, + "grad_norm": 2.468557357788086, + "learning_rate": 4.48800913589646e-06, + "loss": 1.6412, + "step": 576 + }, + { + "epoch": 0.5927067282999486, + "grad_norm": 2.454141855239868, + "learning_rate": 4.486105824133993e-06, + "loss": 1.6762, + "step": 577 + }, + { + "epoch": 0.593733949666153, + "grad_norm": 3.0009915828704834, + "learning_rate": 4.4842025123715264e-06, + "loss": 1.6847, + "step": 578 + }, + { + "epoch": 0.5947611710323575, + "grad_norm": 2.574394702911377, + "learning_rate": 4.48229920060906e-06, + "loss": 1.6421, + "step": 579 + }, + { + "epoch": 0.5957883923985619, + "grad_norm": 2.6978182792663574, + "learning_rate": 4.480395888846594e-06, + "loss": 1.6198, + "step": 580 + }, + { + "epoch": 0.5957883923985619, + "eval_loss": 1.9570144414901733, + "eval_runtime": 44.7661, + "eval_samples_per_second": 22.338, + "eval_steps_per_second": 0.938, + "step": 580 + }, + { + "epoch": 0.5968156137647663, + "grad_norm": 2.652461051940918, + "learning_rate": 4.478492577084127e-06, + "loss": 1.7334, + "step": 581 + }, + { + "epoch": 0.5978428351309707, + "grad_norm": 2.6292548179626465, + "learning_rate": 4.47658926532166e-06, + "loss": 1.6242, + "step": 582 + }, + { + "epoch": 0.5988700564971752, + "grad_norm": 2.474919080734253, + "learning_rate": 4.474685953559193e-06, + "loss": 1.6732, + "step": 583 + }, + { + "epoch": 0.5998972778633795, + "grad_norm": 2.7282516956329346, + "learning_rate": 4.472782641796727e-06, + "loss": 1.802, + "step": 584 + }, + { + "epoch": 0.600924499229584, + "grad_norm": 2.770307779312134, + "learning_rate": 4.47087933003426e-06, + "loss": 1.6156, + "step": 585 + }, + { + "epoch": 0.6019517205957884, + "grad_norm": 2.570681095123291, + "learning_rate": 4.468976018271793e-06, + "loss": 1.6309, + "step": 586 + }, + { + "epoch": 0.6029789419619929, + "grad_norm": 2.4075732231140137, + "learning_rate": 4.467072706509327e-06, + "loss": 1.6436, + "step": 587 + }, + { + "epoch": 0.6040061633281972, + "grad_norm": 2.4570937156677246, + "learning_rate": 4.46516939474686e-06, + "loss": 1.6737, + "step": 588 + }, + { + "epoch": 0.6050333846944016, + "grad_norm": 2.407274007797241, + "learning_rate": 4.463266082984393e-06, + "loss": 1.6179, + "step": 589 + }, + { + "epoch": 0.6060606060606061, + "grad_norm": 2.745486259460449, + "learning_rate": 4.461362771221927e-06, + "loss": 1.7458, + "step": 590 + }, + { + "epoch": 0.6060606060606061, + "eval_loss": 1.9700438976287842, + "eval_runtime": 45.3018, + "eval_samples_per_second": 22.074, + "eval_steps_per_second": 0.927, + "step": 590 + }, + { + "epoch": 0.6070878274268104, + "grad_norm": 2.4258809089660645, + "learning_rate": 4.45945945945946e-06, + "loss": 1.6372, + "step": 591 + }, + { + "epoch": 0.6081150487930149, + "grad_norm": 2.683662176132202, + "learning_rate": 4.457556147696993e-06, + "loss": 1.6649, + "step": 592 + }, + { + "epoch": 0.6091422701592193, + "grad_norm": 2.879727602005005, + "learning_rate": 4.455652835934526e-06, + "loss": 1.7376, + "step": 593 + }, + { + "epoch": 0.6101694915254238, + "grad_norm": 2.5620689392089844, + "learning_rate": 4.453749524172059e-06, + "loss": 1.7203, + "step": 594 + }, + { + "epoch": 0.6111967128916281, + "grad_norm": 2.662031888961792, + "learning_rate": 4.451846212409593e-06, + "loss": 1.6835, + "step": 595 + }, + { + "epoch": 0.6122239342578326, + "grad_norm": 2.6269922256469727, + "learning_rate": 4.449942900647126e-06, + "loss": 1.6726, + "step": 596 + }, + { + "epoch": 0.613251155624037, + "grad_norm": 2.515852212905884, + "learning_rate": 4.4480395888846595e-06, + "loss": 1.6045, + "step": 597 + }, + { + "epoch": 0.6142783769902413, + "grad_norm": 3.006873607635498, + "learning_rate": 4.446136277122193e-06, + "loss": 1.6827, + "step": 598 + }, + { + "epoch": 0.6153055983564458, + "grad_norm": 2.2904858589172363, + "learning_rate": 4.444232965359726e-06, + "loss": 1.6123, + "step": 599 + }, + { + "epoch": 0.6163328197226502, + "grad_norm": 2.7345471382141113, + "learning_rate": 4.4423296535972596e-06, + "loss": 1.6697, + "step": 600 + }, + { + "epoch": 0.6163328197226502, + "eval_loss": 1.9713997840881348, + "eval_runtime": 45.6715, + "eval_samples_per_second": 21.895, + "eval_steps_per_second": 0.92, + "step": 600 + }, + { + "epoch": 0.6173600410888547, + "grad_norm": 2.955853223800659, + "learning_rate": 4.440426341834793e-06, + "loss": 1.6195, + "step": 601 + }, + { + "epoch": 0.618387262455059, + "grad_norm": 2.4079766273498535, + "learning_rate": 4.438523030072326e-06, + "loss": 1.6589, + "step": 602 + }, + { + "epoch": 0.6194144838212635, + "grad_norm": 2.8158490657806396, + "learning_rate": 4.43661971830986e-06, + "loss": 1.6895, + "step": 603 + }, + { + "epoch": 0.6204417051874679, + "grad_norm": 2.491471529006958, + "learning_rate": 4.434716406547393e-06, + "loss": 1.6351, + "step": 604 + }, + { + "epoch": 0.6214689265536724, + "grad_norm": 2.4093260765075684, + "learning_rate": 4.432813094784926e-06, + "loss": 1.6402, + "step": 605 + }, + { + "epoch": 0.6224961479198767, + "grad_norm": 2.751018762588501, + "learning_rate": 4.43090978302246e-06, + "loss": 1.6962, + "step": 606 + }, + { + "epoch": 0.6235233692860811, + "grad_norm": 2.4410409927368164, + "learning_rate": 4.429006471259993e-06, + "loss": 1.6549, + "step": 607 + }, + { + "epoch": 0.6245505906522856, + "grad_norm": 2.360455274581909, + "learning_rate": 4.4271031594975265e-06, + "loss": 1.6449, + "step": 608 + }, + { + "epoch": 0.6255778120184899, + "grad_norm": 2.5058023929595947, + "learning_rate": 4.425199847735059e-06, + "loss": 1.617, + "step": 609 + }, + { + "epoch": 0.6266050333846944, + "grad_norm": 2.3429927825927734, + "learning_rate": 4.423296535972592e-06, + "loss": 1.6262, + "step": 610 + }, + { + "epoch": 0.6266050333846944, + "eval_loss": 1.9597365856170654, + "eval_runtime": 42.2034, + "eval_samples_per_second": 23.695, + "eval_steps_per_second": 0.995, + "step": 610 + }, + { + "epoch": 0.6276322547508988, + "grad_norm": 2.401761054992676, + "learning_rate": 4.421393224210126e-06, + "loss": 1.6497, + "step": 611 + }, + { + "epoch": 0.6286594761171033, + "grad_norm": 2.433439254760742, + "learning_rate": 4.419489912447659e-06, + "loss": 1.6117, + "step": 612 + }, + { + "epoch": 0.6296866974833076, + "grad_norm": 2.5147452354431152, + "learning_rate": 4.4175866006851924e-06, + "loss": 1.7041, + "step": 613 + }, + { + "epoch": 0.6307139188495121, + "grad_norm": 2.4174447059631348, + "learning_rate": 4.415683288922726e-06, + "loss": 1.6652, + "step": 614 + }, + { + "epoch": 0.6317411402157165, + "grad_norm": 2.424800157546997, + "learning_rate": 4.413779977160259e-06, + "loss": 1.6407, + "step": 615 + }, + { + "epoch": 0.632768361581921, + "grad_norm": 2.6760315895080566, + "learning_rate": 4.4118766653977925e-06, + "loss": 1.6566, + "step": 616 + }, + { + "epoch": 0.6337955829481253, + "grad_norm": 2.694810628890991, + "learning_rate": 4.409973353635326e-06, + "loss": 1.7125, + "step": 617 + }, + { + "epoch": 0.6348228043143297, + "grad_norm": 2.475126266479492, + "learning_rate": 4.408070041872859e-06, + "loss": 1.6663, + "step": 618 + }, + { + "epoch": 0.6358500256805342, + "grad_norm": 2.734738349914551, + "learning_rate": 4.406166730110393e-06, + "loss": 1.6115, + "step": 619 + }, + { + "epoch": 0.6368772470467385, + "grad_norm": 2.4914753437042236, + "learning_rate": 4.404263418347925e-06, + "loss": 1.6096, + "step": 620 + }, + { + "epoch": 0.6368772470467385, + "eval_loss": 1.952081322669983, + "eval_runtime": 45.4451, + "eval_samples_per_second": 22.005, + "eval_steps_per_second": 0.924, + "step": 620 + }, + { + "epoch": 0.637904468412943, + "grad_norm": 2.3727800846099854, + "learning_rate": 4.402360106585459e-06, + "loss": 1.6275, + "step": 621 + }, + { + "epoch": 0.6389316897791474, + "grad_norm": 2.576402187347412, + "learning_rate": 4.400456794822993e-06, + "loss": 1.6252, + "step": 622 + }, + { + "epoch": 0.6399589111453519, + "grad_norm": 2.560185432434082, + "learning_rate": 4.398553483060526e-06, + "loss": 1.6522, + "step": 623 + }, + { + "epoch": 0.6409861325115562, + "grad_norm": 2.462785005569458, + "learning_rate": 4.396650171298059e-06, + "loss": 1.6124, + "step": 624 + }, + { + "epoch": 0.6420133538777607, + "grad_norm": 2.3942930698394775, + "learning_rate": 4.394746859535593e-06, + "loss": 1.6231, + "step": 625 + }, + { + "epoch": 0.6430405752439651, + "grad_norm": 2.4870522022247314, + "learning_rate": 4.392843547773125e-06, + "loss": 1.6086, + "step": 626 + }, + { + "epoch": 0.6440677966101694, + "grad_norm": 2.5702786445617676, + "learning_rate": 4.390940236010659e-06, + "loss": 1.6228, + "step": 627 + }, + { + "epoch": 0.6450950179763739, + "grad_norm": 2.743983268737793, + "learning_rate": 4.389036924248192e-06, + "loss": 1.6394, + "step": 628 + }, + { + "epoch": 0.6461222393425783, + "grad_norm": 2.59942364692688, + "learning_rate": 4.387133612485725e-06, + "loss": 1.6721, + "step": 629 + }, + { + "epoch": 0.6471494607087828, + "grad_norm": 2.4473204612731934, + "learning_rate": 4.385230300723259e-06, + "loss": 1.6407, + "step": 630 + }, + { + "epoch": 0.6471494607087828, + "eval_loss": 1.9426974058151245, + "eval_runtime": 45.2318, + "eval_samples_per_second": 22.108, + "eval_steps_per_second": 0.929, + "step": 630 + }, + { + "epoch": 0.6481766820749871, + "grad_norm": 2.6036875247955322, + "learning_rate": 4.383326988960792e-06, + "loss": 1.6307, + "step": 631 + }, + { + "epoch": 0.6492039034411916, + "grad_norm": 2.691011428833008, + "learning_rate": 4.3814236771983255e-06, + "loss": 1.6059, + "step": 632 + }, + { + "epoch": 0.650231124807396, + "grad_norm": 2.585758924484253, + "learning_rate": 4.379520365435859e-06, + "loss": 1.6781, + "step": 633 + }, + { + "epoch": 0.6512583461736005, + "grad_norm": 2.9522950649261475, + "learning_rate": 4.377617053673392e-06, + "loss": 1.6559, + "step": 634 + }, + { + "epoch": 0.6522855675398048, + "grad_norm": 2.5368709564208984, + "learning_rate": 4.3757137419109256e-06, + "loss": 1.6445, + "step": 635 + }, + { + "epoch": 0.6533127889060092, + "grad_norm": 2.5002150535583496, + "learning_rate": 4.373810430148459e-06, + "loss": 1.7184, + "step": 636 + }, + { + "epoch": 0.6543400102722137, + "grad_norm": 2.7623229026794434, + "learning_rate": 4.3719071183859914e-06, + "loss": 1.6195, + "step": 637 + }, + { + "epoch": 0.655367231638418, + "grad_norm": 2.6188151836395264, + "learning_rate": 4.370003806623525e-06, + "loss": 1.6686, + "step": 638 + }, + { + "epoch": 0.6563944530046225, + "grad_norm": 2.685776948928833, + "learning_rate": 4.368100494861058e-06, + "loss": 1.6134, + "step": 639 + }, + { + "epoch": 0.6574216743708269, + "grad_norm": 2.5703907012939453, + "learning_rate": 4.3661971830985915e-06, + "loss": 1.665, + "step": 640 + }, + { + "epoch": 0.6574216743708269, + "eval_loss": 1.9504806995391846, + "eval_runtime": 47.1947, + "eval_samples_per_second": 21.189, + "eval_steps_per_second": 0.89, + "step": 640 + }, + { + "epoch": 0.6584488957370314, + "grad_norm": 2.511617660522461, + "learning_rate": 4.364293871336126e-06, + "loss": 1.6201, + "step": 641 + }, + { + "epoch": 0.6594761171032357, + "grad_norm": 2.549302339553833, + "learning_rate": 4.362390559573659e-06, + "loss": 1.647, + "step": 642 + }, + { + "epoch": 0.6605033384694402, + "grad_norm": 2.348747730255127, + "learning_rate": 4.3604872478111925e-06, + "loss": 1.5993, + "step": 643 + }, + { + "epoch": 0.6615305598356446, + "grad_norm": 2.7393345832824707, + "learning_rate": 4.358583936048725e-06, + "loss": 1.543, + "step": 644 + }, + { + "epoch": 0.662557781201849, + "grad_norm": 2.4242193698883057, + "learning_rate": 4.356680624286258e-06, + "loss": 1.6726, + "step": 645 + }, + { + "epoch": 0.6635850025680534, + "grad_norm": 2.6446714401245117, + "learning_rate": 4.354777312523792e-06, + "loss": 1.6312, + "step": 646 + }, + { + "epoch": 0.6646122239342578, + "grad_norm": 2.7407124042510986, + "learning_rate": 4.352874000761325e-06, + "loss": 1.6868, + "step": 647 + }, + { + "epoch": 0.6656394453004623, + "grad_norm": 2.3706371784210205, + "learning_rate": 4.3509706889988584e-06, + "loss": 1.6301, + "step": 648 + }, + { + "epoch": 0.6666666666666666, + "grad_norm": 2.5197198390960693, + "learning_rate": 4.349067377236392e-06, + "loss": 1.702, + "step": 649 + }, + { + "epoch": 0.6676938880328711, + "grad_norm": 2.4792377948760986, + "learning_rate": 4.347164065473925e-06, + "loss": 1.6891, + "step": 650 + }, + { + "epoch": 0.6676938880328711, + "eval_loss": 1.9589039087295532, + "eval_runtime": 47.1512, + "eval_samples_per_second": 21.208, + "eval_steps_per_second": 0.891, + "step": 650 + }, + { + "epoch": 0.6687211093990755, + "grad_norm": 2.4003701210021973, + "learning_rate": 4.3452607537114585e-06, + "loss": 1.7015, + "step": 651 + }, + { + "epoch": 0.66974833076528, + "grad_norm": 2.5363786220550537, + "learning_rate": 4.343357441948992e-06, + "loss": 1.6133, + "step": 652 + }, + { + "epoch": 0.6707755521314843, + "grad_norm": 2.5183844566345215, + "learning_rate": 4.341454130186525e-06, + "loss": 1.6574, + "step": 653 + }, + { + "epoch": 0.6718027734976888, + "grad_norm": 2.4668304920196533, + "learning_rate": 4.339550818424058e-06, + "loss": 1.6253, + "step": 654 + }, + { + "epoch": 0.6728299948638932, + "grad_norm": 2.4829814434051514, + "learning_rate": 4.337647506661591e-06, + "loss": 1.6099, + "step": 655 + }, + { + "epoch": 0.6738572162300975, + "grad_norm": 2.3683176040649414, + "learning_rate": 4.3357441948991245e-06, + "loss": 1.6385, + "step": 656 + }, + { + "epoch": 0.674884437596302, + "grad_norm": 2.4906842708587646, + "learning_rate": 4.333840883136658e-06, + "loss": 1.6755, + "step": 657 + }, + { + "epoch": 0.6759116589625064, + "grad_norm": 2.558776617050171, + "learning_rate": 4.331937571374191e-06, + "loss": 1.6152, + "step": 658 + }, + { + "epoch": 0.6769388803287109, + "grad_norm": 2.4081523418426514, + "learning_rate": 4.3300342596117246e-06, + "loss": 1.6224, + "step": 659 + }, + { + "epoch": 0.6779661016949152, + "grad_norm": 2.488370180130005, + "learning_rate": 4.328130947849258e-06, + "loss": 1.6298, + "step": 660 + }, + { + "epoch": 0.6779661016949152, + "eval_loss": 1.9484621286392212, + "eval_runtime": 52.549, + "eval_samples_per_second": 19.03, + "eval_steps_per_second": 0.799, + "step": 660 + }, + { + "epoch": 0.6789933230611197, + "grad_norm": 2.495748281478882, + "learning_rate": 4.326227636086791e-06, + "loss": 1.5852, + "step": 661 + }, + { + "epoch": 0.6800205444273241, + "grad_norm": 2.525031566619873, + "learning_rate": 4.324324324324325e-06, + "loss": 1.6348, + "step": 662 + }, + { + "epoch": 0.6810477657935285, + "grad_norm": 2.427615165710449, + "learning_rate": 4.322421012561858e-06, + "loss": 1.6352, + "step": 663 + }, + { + "epoch": 0.6820749871597329, + "grad_norm": 2.5084927082061768, + "learning_rate": 4.320517700799391e-06, + "loss": 1.6334, + "step": 664 + }, + { + "epoch": 0.6831022085259374, + "grad_norm": 2.6018898487091064, + "learning_rate": 4.318614389036925e-06, + "loss": 1.7212, + "step": 665 + }, + { + "epoch": 0.6841294298921418, + "grad_norm": 2.44184947013855, + "learning_rate": 4.316711077274458e-06, + "loss": 1.6802, + "step": 666 + }, + { + "epoch": 0.6851566512583461, + "grad_norm": 2.542123317718506, + "learning_rate": 4.3148077655119915e-06, + "loss": 1.5912, + "step": 667 + }, + { + "epoch": 0.6861838726245506, + "grad_norm": 2.5208137035369873, + "learning_rate": 4.312904453749525e-06, + "loss": 1.6965, + "step": 668 + }, + { + "epoch": 0.687211093990755, + "grad_norm": 2.391904592514038, + "learning_rate": 4.311001141987058e-06, + "loss": 1.6388, + "step": 669 + }, + { + "epoch": 0.6882383153569595, + "grad_norm": 2.629565477371216, + "learning_rate": 4.3090978302245916e-06, + "loss": 1.6475, + "step": 670 + }, + { + "epoch": 0.6882383153569595, + "eval_loss": 1.9572818279266357, + "eval_runtime": 48.5354, + "eval_samples_per_second": 20.604, + "eval_steps_per_second": 0.865, + "step": 670 + }, + { + "epoch": 0.6892655367231638, + "grad_norm": 2.3132376670837402, + "learning_rate": 4.307194518462124e-06, + "loss": 1.6399, + "step": 671 + }, + { + "epoch": 0.6902927580893683, + "grad_norm": 2.5003411769866943, + "learning_rate": 4.3052912066996574e-06, + "loss": 1.7136, + "step": 672 + }, + { + "epoch": 0.6913199794555727, + "grad_norm": 2.545541286468506, + "learning_rate": 4.303387894937191e-06, + "loss": 1.6825, + "step": 673 + }, + { + "epoch": 0.6923472008217771, + "grad_norm": 2.342423439025879, + "learning_rate": 4.301484583174724e-06, + "loss": 1.6088, + "step": 674 + }, + { + "epoch": 0.6933744221879815, + "grad_norm": 2.5516135692596436, + "learning_rate": 4.2995812714122575e-06, + "loss": 1.5934, + "step": 675 + }, + { + "epoch": 0.6944016435541859, + "grad_norm": 2.438675880432129, + "learning_rate": 4.297677959649791e-06, + "loss": 1.6236, + "step": 676 + }, + { + "epoch": 0.6954288649203904, + "grad_norm": 2.767305612564087, + "learning_rate": 4.295774647887324e-06, + "loss": 1.6514, + "step": 677 + }, + { + "epoch": 0.6964560862865947, + "grad_norm": 2.591994047164917, + "learning_rate": 4.293871336124858e-06, + "loss": 1.6643, + "step": 678 + }, + { + "epoch": 0.6974833076527992, + "grad_norm": 2.4734244346618652, + "learning_rate": 4.291968024362391e-06, + "loss": 1.6362, + "step": 679 + }, + { + "epoch": 0.6985105290190036, + "grad_norm": 2.6181890964508057, + "learning_rate": 4.290064712599924e-06, + "loss": 1.6666, + "step": 680 + }, + { + "epoch": 0.6985105290190036, + "eval_loss": 1.9585388898849487, + "eval_runtime": 46.8094, + "eval_samples_per_second": 21.363, + "eval_steps_per_second": 0.897, + "step": 680 + }, + { + "epoch": 0.699537750385208, + "grad_norm": 2.465571403503418, + "learning_rate": 4.288161400837458e-06, + "loss": 1.6588, + "step": 681 + }, + { + "epoch": 0.7005649717514124, + "grad_norm": 2.664262294769287, + "learning_rate": 4.286258089074991e-06, + "loss": 1.5862, + "step": 682 + }, + { + "epoch": 0.7015921931176169, + "grad_norm": 2.677140474319458, + "learning_rate": 4.284354777312524e-06, + "loss": 1.6549, + "step": 683 + }, + { + "epoch": 0.7026194144838213, + "grad_norm": 2.559051275253296, + "learning_rate": 4.282451465550058e-06, + "loss": 1.609, + "step": 684 + }, + { + "epoch": 0.7036466358500256, + "grad_norm": 2.428424119949341, + "learning_rate": 4.280548153787591e-06, + "loss": 1.6243, + "step": 685 + }, + { + "epoch": 0.7046738572162301, + "grad_norm": 2.54136061668396, + "learning_rate": 4.2786448420251245e-06, + "loss": 1.6301, + "step": 686 + }, + { + "epoch": 0.7057010785824345, + "grad_norm": 2.377699136734009, + "learning_rate": 4.276741530262658e-06, + "loss": 1.5903, + "step": 687 + }, + { + "epoch": 0.706728299948639, + "grad_norm": 2.5548322200775146, + "learning_rate": 4.27483821850019e-06, + "loss": 1.6161, + "step": 688 + }, + { + "epoch": 0.7077555213148433, + "grad_norm": 2.5993409156799316, + "learning_rate": 4.272934906737724e-06, + "loss": 1.6882, + "step": 689 + }, + { + "epoch": 0.7087827426810478, + "grad_norm": 2.4249117374420166, + "learning_rate": 4.271031594975257e-06, + "loss": 1.6723, + "step": 690 + }, + { + "epoch": 0.7087827426810478, + "eval_loss": 1.9526821374893188, + "eval_runtime": 48.2168, + "eval_samples_per_second": 20.74, + "eval_steps_per_second": 0.871, + "step": 690 + }, + { + "epoch": 0.7098099640472522, + "grad_norm": 2.6965081691741943, + "learning_rate": 4.2691282832127905e-06, + "loss": 1.6636, + "step": 691 + }, + { + "epoch": 0.7108371854134566, + "grad_norm": 2.6205356121063232, + "learning_rate": 4.267224971450324e-06, + "loss": 1.6231, + "step": 692 + }, + { + "epoch": 0.711864406779661, + "grad_norm": 2.5924129486083984, + "learning_rate": 4.265321659687857e-06, + "loss": 1.5634, + "step": 693 + }, + { + "epoch": 0.7128916281458655, + "grad_norm": 2.9710943698883057, + "learning_rate": 4.2634183479253906e-06, + "loss": 1.6134, + "step": 694 + }, + { + "epoch": 0.7139188495120699, + "grad_norm": 2.6252782344818115, + "learning_rate": 4.261515036162924e-06, + "loss": 1.5997, + "step": 695 + }, + { + "epoch": 0.7149460708782742, + "grad_norm": 2.6178832054138184, + "learning_rate": 4.259611724400457e-06, + "loss": 1.6255, + "step": 696 + }, + { + "epoch": 0.7159732922444787, + "grad_norm": 2.82548189163208, + "learning_rate": 4.257708412637991e-06, + "loss": 1.6167, + "step": 697 + }, + { + "epoch": 0.7170005136106831, + "grad_norm": 2.6430351734161377, + "learning_rate": 4.255805100875523e-06, + "loss": 1.5585, + "step": 698 + }, + { + "epoch": 0.7180277349768875, + "grad_norm": 2.85046124458313, + "learning_rate": 4.2539017891130565e-06, + "loss": 1.5873, + "step": 699 + }, + { + "epoch": 0.7190549563430919, + "grad_norm": 2.5643832683563232, + "learning_rate": 4.25199847735059e-06, + "loss": 1.6705, + "step": 700 + }, + { + "epoch": 0.7190549563430919, + "eval_loss": 1.945617437362671, + "eval_runtime": 54.2778, + "eval_samples_per_second": 18.424, + "eval_steps_per_second": 0.774, + "step": 700 + }, + { + "epoch": 0.7200821777092964, + "grad_norm": 2.7856431007385254, + "learning_rate": 4.250095165588123e-06, + "loss": 1.5833, + "step": 701 + }, + { + "epoch": 0.7211093990755008, + "grad_norm": 2.9881107807159424, + "learning_rate": 4.2481918538256575e-06, + "loss": 1.6438, + "step": 702 + }, + { + "epoch": 0.7221366204417052, + "grad_norm": 2.602938652038574, + "learning_rate": 4.246288542063191e-06, + "loss": 1.6135, + "step": 703 + }, + { + "epoch": 0.7231638418079096, + "grad_norm": 2.997450590133667, + "learning_rate": 4.244385230300724e-06, + "loss": 1.6419, + "step": 704 + }, + { + "epoch": 0.724191063174114, + "grad_norm": 2.5592496395111084, + "learning_rate": 4.242481918538257e-06, + "loss": 1.5886, + "step": 705 + }, + { + "epoch": 0.7252182845403184, + "grad_norm": 2.4073948860168457, + "learning_rate": 4.24057860677579e-06, + "loss": 1.6219, + "step": 706 + }, + { + "epoch": 0.7262455059065228, + "grad_norm": 2.545997381210327, + "learning_rate": 4.2386752950133234e-06, + "loss": 1.6684, + "step": 707 + }, + { + "epoch": 0.7272727272727273, + "grad_norm": 2.468264579772949, + "learning_rate": 4.236771983250857e-06, + "loss": 1.6155, + "step": 708 + }, + { + "epoch": 0.7282999486389317, + "grad_norm": 2.6635043621063232, + "learning_rate": 4.23486867148839e-06, + "loss": 1.5956, + "step": 709 + }, + { + "epoch": 0.7293271700051361, + "grad_norm": 2.8751156330108643, + "learning_rate": 4.2329653597259235e-06, + "loss": 1.6211, + "step": 710 + }, + { + "epoch": 0.7293271700051361, + "eval_loss": 1.9319628477096558, + "eval_runtime": 43.9701, + "eval_samples_per_second": 22.743, + "eval_steps_per_second": 0.955, + "step": 710 + }, + { + "epoch": 0.7303543913713405, + "grad_norm": 2.5999035835266113, + "learning_rate": 4.231062047963457e-06, + "loss": 1.6442, + "step": 711 + }, + { + "epoch": 0.731381612737545, + "grad_norm": 2.611095666885376, + "learning_rate": 4.22915873620099e-06, + "loss": 1.6745, + "step": 712 + }, + { + "epoch": 0.7324088341037494, + "grad_norm": 2.7235333919525146, + "learning_rate": 4.227255424438524e-06, + "loss": 1.6353, + "step": 713 + }, + { + "epoch": 0.7334360554699538, + "grad_norm": 2.453639507293701, + "learning_rate": 4.225352112676057e-06, + "loss": 1.6657, + "step": 714 + }, + { + "epoch": 0.7344632768361582, + "grad_norm": 2.4031429290771484, + "learning_rate": 4.2234488009135895e-06, + "loss": 1.603, + "step": 715 + }, + { + "epoch": 0.7354904982023626, + "grad_norm": 2.631052017211914, + "learning_rate": 4.221545489151123e-06, + "loss": 1.5681, + "step": 716 + }, + { + "epoch": 0.736517719568567, + "grad_norm": 2.518758773803711, + "learning_rate": 4.219642177388656e-06, + "loss": 1.5941, + "step": 717 + }, + { + "epoch": 0.7375449409347714, + "grad_norm": 2.699491500854492, + "learning_rate": 4.2177388656261896e-06, + "loss": 1.6327, + "step": 718 + }, + { + "epoch": 0.7385721623009759, + "grad_norm": 2.749356985092163, + "learning_rate": 4.215835553863723e-06, + "loss": 1.5864, + "step": 719 + }, + { + "epoch": 0.7395993836671803, + "grad_norm": 2.5699031352996826, + "learning_rate": 4.213932242101256e-06, + "loss": 1.5444, + "step": 720 + }, + { + "epoch": 0.7395993836671803, + "eval_loss": 1.9521623849868774, + "eval_runtime": 46.5749, + "eval_samples_per_second": 21.471, + "eval_steps_per_second": 0.902, + "step": 720 + }, + { + "epoch": 0.7406266050333847, + "grad_norm": 3.197425127029419, + "learning_rate": 4.21202893033879e-06, + "loss": 1.6158, + "step": 721 + }, + { + "epoch": 0.7416538263995891, + "grad_norm": 2.619459390640259, + "learning_rate": 4.210125618576323e-06, + "loss": 1.6328, + "step": 722 + }, + { + "epoch": 0.7426810477657936, + "grad_norm": 2.792086601257324, + "learning_rate": 4.208222306813856e-06, + "loss": 1.6333, + "step": 723 + }, + { + "epoch": 0.743708269131998, + "grad_norm": 2.7023141384124756, + "learning_rate": 4.20631899505139e-06, + "loss": 1.6787, + "step": 724 + }, + { + "epoch": 0.7447354904982023, + "grad_norm": 2.6319735050201416, + "learning_rate": 4.204415683288923e-06, + "loss": 1.54, + "step": 725 + }, + { + "epoch": 0.7457627118644068, + "grad_norm": 2.7032439708709717, + "learning_rate": 4.2025123715264565e-06, + "loss": 1.641, + "step": 726 + }, + { + "epoch": 0.7467899332306112, + "grad_norm": 2.9133267402648926, + "learning_rate": 4.20060905976399e-06, + "loss": 1.5901, + "step": 727 + }, + { + "epoch": 0.7478171545968156, + "grad_norm": 2.492403745651245, + "learning_rate": 4.198705748001523e-06, + "loss": 1.5516, + "step": 728 + }, + { + "epoch": 0.74884437596302, + "grad_norm": 2.604520082473755, + "learning_rate": 4.1968024362390566e-06, + "loss": 1.6574, + "step": 729 + }, + { + "epoch": 0.7498715973292245, + "grad_norm": 2.521969795227051, + "learning_rate": 4.19489912447659e-06, + "loss": 1.6099, + "step": 730 + }, + { + "epoch": 0.7498715973292245, + "eval_loss": 1.9449840784072876, + "eval_runtime": 44.9334, + "eval_samples_per_second": 22.255, + "eval_steps_per_second": 0.935, + "step": 730 + }, + { + "epoch": 0.7508988186954288, + "grad_norm": 2.4645657539367676, + "learning_rate": 4.192995812714123e-06, + "loss": 1.6752, + "step": 731 + }, + { + "epoch": 0.7519260400616333, + "grad_norm": 2.7852981090545654, + "learning_rate": 4.191092500951656e-06, + "loss": 1.6176, + "step": 732 + }, + { + "epoch": 0.7529532614278377, + "grad_norm": 2.2817575931549072, + "learning_rate": 4.189189189189189e-06, + "loss": 1.5934, + "step": 733 + }, + { + "epoch": 0.7539804827940421, + "grad_norm": 2.6218173503875732, + "learning_rate": 4.1872858774267225e-06, + "loss": 1.5902, + "step": 734 + }, + { + "epoch": 0.7550077041602465, + "grad_norm": 2.6671013832092285, + "learning_rate": 4.185382565664256e-06, + "loss": 1.657, + "step": 735 + }, + { + "epoch": 0.7560349255264509, + "grad_norm": 2.669757843017578, + "learning_rate": 4.183479253901789e-06, + "loss": 1.5884, + "step": 736 + }, + { + "epoch": 0.7570621468926554, + "grad_norm": 2.7210471630096436, + "learning_rate": 4.181575942139323e-06, + "loss": 1.5558, + "step": 737 + }, + { + "epoch": 0.7580893682588598, + "grad_norm": 2.414276123046875, + "learning_rate": 4.179672630376856e-06, + "loss": 1.6124, + "step": 738 + }, + { + "epoch": 0.7591165896250642, + "grad_norm": 2.4643094539642334, + "learning_rate": 4.177769318614389e-06, + "loss": 1.5904, + "step": 739 + }, + { + "epoch": 0.7601438109912686, + "grad_norm": 2.5447018146514893, + "learning_rate": 4.175866006851923e-06, + "loss": 1.6705, + "step": 740 + }, + { + "epoch": 0.7601438109912686, + "eval_loss": 1.9437954425811768, + "eval_runtime": 49.2702, + "eval_samples_per_second": 20.296, + "eval_steps_per_second": 0.852, + "step": 740 + }, + { + "epoch": 0.7611710323574731, + "grad_norm": 2.41048002243042, + "learning_rate": 4.173962695089456e-06, + "loss": 1.6398, + "step": 741 + }, + { + "epoch": 0.7621982537236774, + "grad_norm": 2.3949906826019287, + "learning_rate": 4.1720593833269894e-06, + "loss": 1.6198, + "step": 742 + }, + { + "epoch": 0.7632254750898819, + "grad_norm": 2.36966609954834, + "learning_rate": 4.170156071564523e-06, + "loss": 1.6275, + "step": 743 + }, + { + "epoch": 0.7642526964560863, + "grad_norm": 2.623347043991089, + "learning_rate": 4.168252759802056e-06, + "loss": 1.7059, + "step": 744 + }, + { + "epoch": 0.7652799178222907, + "grad_norm": 2.384788751602173, + "learning_rate": 4.1663494480395895e-06, + "loss": 1.6015, + "step": 745 + }, + { + "epoch": 0.7663071391884951, + "grad_norm": 2.586009979248047, + "learning_rate": 4.164446136277123e-06, + "loss": 1.6445, + "step": 746 + }, + { + "epoch": 0.7673343605546995, + "grad_norm": 2.396684169769287, + "learning_rate": 4.162542824514656e-06, + "loss": 1.6539, + "step": 747 + }, + { + "epoch": 0.768361581920904, + "grad_norm": 2.416243553161621, + "learning_rate": 4.16063951275219e-06, + "loss": 1.6142, + "step": 748 + }, + { + "epoch": 0.7693888032871083, + "grad_norm": 2.480912208557129, + "learning_rate": 4.158736200989722e-06, + "loss": 1.5768, + "step": 749 + }, + { + "epoch": 0.7704160246533128, + "grad_norm": 2.5516929626464844, + "learning_rate": 4.1568328892272555e-06, + "loss": 1.6074, + "step": 750 + }, + { + "epoch": 0.7704160246533128, + "eval_loss": 1.9409116506576538, + "eval_runtime": 45.3465, + "eval_samples_per_second": 22.052, + "eval_steps_per_second": 0.926, + "step": 750 + }, + { + "epoch": 0.7714432460195172, + "grad_norm": 2.4248714447021484, + "learning_rate": 4.154929577464789e-06, + "loss": 1.6308, + "step": 751 + }, + { + "epoch": 0.7724704673857217, + "grad_norm": 2.520810604095459, + "learning_rate": 4.153026265702322e-06, + "loss": 1.6227, + "step": 752 + }, + { + "epoch": 0.773497688751926, + "grad_norm": 2.5938422679901123, + "learning_rate": 4.1511229539398556e-06, + "loss": 1.5601, + "step": 753 + }, + { + "epoch": 0.7745249101181304, + "grad_norm": 2.4538512229919434, + "learning_rate": 4.149219642177389e-06, + "loss": 1.6006, + "step": 754 + }, + { + "epoch": 0.7755521314843349, + "grad_norm": 2.4936258792877197, + "learning_rate": 4.147316330414922e-06, + "loss": 1.6281, + "step": 755 + }, + { + "epoch": 0.7765793528505393, + "grad_norm": 2.5774738788604736, + "learning_rate": 4.145413018652456e-06, + "loss": 1.5328, + "step": 756 + }, + { + "epoch": 0.7776065742167437, + "grad_norm": 2.3616905212402344, + "learning_rate": 4.143509706889989e-06, + "loss": 1.629, + "step": 757 + }, + { + "epoch": 0.7786337955829481, + "grad_norm": 2.4983086585998535, + "learning_rate": 4.141606395127522e-06, + "loss": 1.6298, + "step": 758 + }, + { + "epoch": 0.7796610169491526, + "grad_norm": 2.4313876628875732, + "learning_rate": 4.139703083365056e-06, + "loss": 1.6113, + "step": 759 + }, + { + "epoch": 0.7806882383153569, + "grad_norm": 2.6290934085845947, + "learning_rate": 4.137799771602588e-06, + "loss": 1.6632, + "step": 760 + }, + { + "epoch": 0.7806882383153569, + "eval_loss": 1.950036644935608, + "eval_runtime": 47.4523, + "eval_samples_per_second": 21.074, + "eval_steps_per_second": 0.885, + "step": 760 + }, + { + "epoch": 0.7817154596815614, + "grad_norm": 2.3637750148773193, + "learning_rate": 4.135896459840122e-06, + "loss": 1.6268, + "step": 761 + }, + { + "epoch": 0.7827426810477658, + "grad_norm": 2.5537843704223633, + "learning_rate": 4.133993148077656e-06, + "loss": 1.5736, + "step": 762 + }, + { + "epoch": 0.7837699024139703, + "grad_norm": 2.4945273399353027, + "learning_rate": 4.132089836315189e-06, + "loss": 1.612, + "step": 763 + }, + { + "epoch": 0.7847971237801746, + "grad_norm": 2.506983757019043, + "learning_rate": 4.1301865245527226e-06, + "loss": 1.6112, + "step": 764 + }, + { + "epoch": 0.785824345146379, + "grad_norm": 2.599721908569336, + "learning_rate": 4.128283212790256e-06, + "loss": 1.5698, + "step": 765 + }, + { + "epoch": 0.7868515665125835, + "grad_norm": 2.4561266899108887, + "learning_rate": 4.1263799010277884e-06, + "loss": 1.5354, + "step": 766 + }, + { + "epoch": 0.7878787878787878, + "grad_norm": 2.445420026779175, + "learning_rate": 4.124476589265322e-06, + "loss": 1.6555, + "step": 767 + }, + { + "epoch": 0.7889060092449923, + "grad_norm": 2.576533079147339, + "learning_rate": 4.122573277502855e-06, + "loss": 1.5915, + "step": 768 + }, + { + "epoch": 0.7899332306111967, + "grad_norm": 2.503164291381836, + "learning_rate": 4.1206699657403885e-06, + "loss": 1.5893, + "step": 769 + }, + { + "epoch": 0.7909604519774012, + "grad_norm": 2.9463603496551514, + "learning_rate": 4.118766653977922e-06, + "loss": 1.6178, + "step": 770 + }, + { + "epoch": 0.7909604519774012, + "eval_loss": 1.9529738426208496, + "eval_runtime": 45.0922, + "eval_samples_per_second": 22.177, + "eval_steps_per_second": 0.931, + "step": 770 + }, + { + "epoch": 0.7919876733436055, + "grad_norm": 2.47148060798645, + "learning_rate": 4.116863342215455e-06, + "loss": 1.5991, + "step": 771 + }, + { + "epoch": 0.79301489470981, + "grad_norm": 2.515717029571533, + "learning_rate": 4.114960030452989e-06, + "loss": 1.5725, + "step": 772 + }, + { + "epoch": 0.7940421160760144, + "grad_norm": 2.5231993198394775, + "learning_rate": 4.113056718690522e-06, + "loss": 1.6145, + "step": 773 + }, + { + "epoch": 0.7950693374422187, + "grad_norm": 2.479978084564209, + "learning_rate": 4.111153406928055e-06, + "loss": 1.5537, + "step": 774 + }, + { + "epoch": 0.7960965588084232, + "grad_norm": 2.5341250896453857, + "learning_rate": 4.109250095165589e-06, + "loss": 1.6419, + "step": 775 + }, + { + "epoch": 0.7971237801746276, + "grad_norm": 2.6277401447296143, + "learning_rate": 4.107346783403122e-06, + "loss": 1.6505, + "step": 776 + }, + { + "epoch": 0.7981510015408321, + "grad_norm": 2.4688351154327393, + "learning_rate": 4.105443471640655e-06, + "loss": 1.5896, + "step": 777 + }, + { + "epoch": 0.7991782229070364, + "grad_norm": 2.4134581089019775, + "learning_rate": 4.103540159878188e-06, + "loss": 1.5788, + "step": 778 + }, + { + "epoch": 0.8002054442732409, + "grad_norm": 2.5090012550354004, + "learning_rate": 4.101636848115721e-06, + "loss": 1.6162, + "step": 779 + }, + { + "epoch": 0.8012326656394453, + "grad_norm": 2.5123302936553955, + "learning_rate": 4.099733536353255e-06, + "loss": 1.5868, + "step": 780 + }, + { + "epoch": 0.8012326656394453, + "eval_loss": 1.9545625448226929, + "eval_runtime": 45.9316, + "eval_samples_per_second": 21.772, + "eval_steps_per_second": 0.914, + "step": 780 + }, + { + "epoch": 0.8022598870056498, + "grad_norm": 2.787181854248047, + "learning_rate": 4.097830224590788e-06, + "loss": 1.612, + "step": 781 + }, + { + "epoch": 0.8032871083718541, + "grad_norm": 2.414426326751709, + "learning_rate": 4.095926912828322e-06, + "loss": 1.6908, + "step": 782 + }, + { + "epoch": 0.8043143297380585, + "grad_norm": 2.452697992324829, + "learning_rate": 4.094023601065855e-06, + "loss": 1.6575, + "step": 783 + }, + { + "epoch": 0.805341551104263, + "grad_norm": 2.4377169609069824, + "learning_rate": 4.092120289303388e-06, + "loss": 1.6016, + "step": 784 + }, + { + "epoch": 0.8063687724704673, + "grad_norm": 2.369154214859009, + "learning_rate": 4.0902169775409215e-06, + "loss": 1.6133, + "step": 785 + }, + { + "epoch": 0.8073959938366718, + "grad_norm": 2.261869430541992, + "learning_rate": 4.088313665778455e-06, + "loss": 1.5555, + "step": 786 + }, + { + "epoch": 0.8084232152028762, + "grad_norm": 2.7058632373809814, + "learning_rate": 4.086410354015988e-06, + "loss": 1.5957, + "step": 787 + }, + { + "epoch": 0.8094504365690807, + "grad_norm": 2.3907508850097656, + "learning_rate": 4.0845070422535216e-06, + "loss": 1.5676, + "step": 788 + }, + { + "epoch": 0.810477657935285, + "grad_norm": 2.510284900665283, + "learning_rate": 4.082603730491055e-06, + "loss": 1.5872, + "step": 789 + }, + { + "epoch": 0.8115048793014895, + "grad_norm": 2.590810775756836, + "learning_rate": 4.080700418728588e-06, + "loss": 1.6315, + "step": 790 + }, + { + "epoch": 0.8115048793014895, + "eval_loss": 1.9573780298233032, + "eval_runtime": 44.5677, + "eval_samples_per_second": 22.438, + "eval_steps_per_second": 0.942, + "step": 790 + }, + { + "epoch": 0.8125321006676939, + "grad_norm": 2.466134786605835, + "learning_rate": 4.078797106966122e-06, + "loss": 1.6018, + "step": 791 + }, + { + "epoch": 0.8135593220338984, + "grad_norm": 2.474653482437134, + "learning_rate": 4.076893795203655e-06, + "loss": 1.5101, + "step": 792 + }, + { + "epoch": 0.8145865434001027, + "grad_norm": 2.4463517665863037, + "learning_rate": 4.074990483441188e-06, + "loss": 1.5686, + "step": 793 + }, + { + "epoch": 0.8156137647663071, + "grad_norm": 2.5148985385894775, + "learning_rate": 4.073087171678721e-06, + "loss": 1.6544, + "step": 794 + }, + { + "epoch": 0.8166409861325116, + "grad_norm": 2.6183927059173584, + "learning_rate": 4.071183859916254e-06, + "loss": 1.6277, + "step": 795 + }, + { + "epoch": 0.8176682074987159, + "grad_norm": 2.438427686691284, + "learning_rate": 4.069280548153788e-06, + "loss": 1.6002, + "step": 796 + }, + { + "epoch": 0.8186954288649204, + "grad_norm": 2.662855863571167, + "learning_rate": 4.067377236391321e-06, + "loss": 1.5594, + "step": 797 + }, + { + "epoch": 0.8197226502311248, + "grad_norm": 2.64789080619812, + "learning_rate": 4.065473924628854e-06, + "loss": 1.6007, + "step": 798 + }, + { + "epoch": 0.8207498715973293, + "grad_norm": 2.5152997970581055, + "learning_rate": 4.063570612866388e-06, + "loss": 1.6366, + "step": 799 + }, + { + "epoch": 0.8217770929635336, + "grad_norm": 2.468672037124634, + "learning_rate": 4.061667301103921e-06, + "loss": 1.6313, + "step": 800 + }, + { + "epoch": 0.8217770929635336, + "eval_loss": 1.95821213722229, + "eval_runtime": 45.9362, + "eval_samples_per_second": 21.769, + "eval_steps_per_second": 0.914, + "step": 800 + }, + { + "epoch": 0.8228043143297381, + "grad_norm": 2.440307855606079, + "learning_rate": 4.0597639893414544e-06, + "loss": 1.5993, + "step": 801 + }, + { + "epoch": 0.8238315356959425, + "grad_norm": 2.6519880294799805, + "learning_rate": 4.057860677578988e-06, + "loss": 1.6106, + "step": 802 + }, + { + "epoch": 0.8248587570621468, + "grad_norm": 2.4224767684936523, + "learning_rate": 4.055957365816521e-06, + "loss": 1.6247, + "step": 803 + }, + { + "epoch": 0.8258859784283513, + "grad_norm": 2.346858263015747, + "learning_rate": 4.0540540540540545e-06, + "loss": 1.5699, + "step": 804 + }, + { + "epoch": 0.8269131997945557, + "grad_norm": 2.577711582183838, + "learning_rate": 4.052150742291588e-06, + "loss": 1.5665, + "step": 805 + }, + { + "epoch": 0.8279404211607602, + "grad_norm": 2.604501724243164, + "learning_rate": 4.050247430529121e-06, + "loss": 1.561, + "step": 806 + }, + { + "epoch": 0.8289676425269645, + "grad_norm": 2.534820795059204, + "learning_rate": 4.048344118766655e-06, + "loss": 1.6391, + "step": 807 + }, + { + "epoch": 0.829994863893169, + "grad_norm": 2.496765613555908, + "learning_rate": 4.046440807004188e-06, + "loss": 1.6338, + "step": 808 + }, + { + "epoch": 0.8310220852593734, + "grad_norm": 2.587185859680176, + "learning_rate": 4.044537495241721e-06, + "loss": 1.6009, + "step": 809 + }, + { + "epoch": 0.8320493066255779, + "grad_norm": 2.7336575984954834, + "learning_rate": 4.042634183479255e-06, + "loss": 1.5911, + "step": 810 + }, + { + "epoch": 0.8320493066255779, + "eval_loss": 1.951606273651123, + "eval_runtime": 42.4435, + "eval_samples_per_second": 23.561, + "eval_steps_per_second": 0.99, + "step": 810 + }, + { + "epoch": 0.8330765279917822, + "grad_norm": 2.5524845123291016, + "learning_rate": 4.040730871716787e-06, + "loss": 1.6406, + "step": 811 + }, + { + "epoch": 0.8341037493579867, + "grad_norm": 2.4861226081848145, + "learning_rate": 4.038827559954321e-06, + "loss": 1.6092, + "step": 812 + }, + { + "epoch": 0.8351309707241911, + "grad_norm": 2.668187141418457, + "learning_rate": 4.036924248191854e-06, + "loss": 1.6529, + "step": 813 + }, + { + "epoch": 0.8361581920903954, + "grad_norm": 2.5616471767425537, + "learning_rate": 4.035020936429387e-06, + "loss": 1.6253, + "step": 814 + }, + { + "epoch": 0.8371854134565999, + "grad_norm": 2.4261975288391113, + "learning_rate": 4.033117624666921e-06, + "loss": 1.5955, + "step": 815 + }, + { + "epoch": 0.8382126348228043, + "grad_norm": 2.4947080612182617, + "learning_rate": 4.031214312904454e-06, + "loss": 1.5686, + "step": 816 + }, + { + "epoch": 0.8392398561890088, + "grad_norm": 2.747781276702881, + "learning_rate": 4.029311001141987e-06, + "loss": 1.5642, + "step": 817 + }, + { + "epoch": 0.8402670775552131, + "grad_norm": 2.4762001037597656, + "learning_rate": 4.027407689379521e-06, + "loss": 1.5301, + "step": 818 + }, + { + "epoch": 0.8412942989214176, + "grad_norm": 2.6253550052642822, + "learning_rate": 4.025504377617054e-06, + "loss": 1.5885, + "step": 819 + }, + { + "epoch": 0.842321520287622, + "grad_norm": 2.6974875926971436, + "learning_rate": 4.0236010658545875e-06, + "loss": 1.5656, + "step": 820 + }, + { + "epoch": 0.842321520287622, + "eval_loss": 1.9420732259750366, + "eval_runtime": 46.9525, + "eval_samples_per_second": 21.298, + "eval_steps_per_second": 0.895, + "step": 820 + }, + { + "epoch": 0.8433487416538265, + "grad_norm": 2.638587474822998, + "learning_rate": 4.02169775409212e-06, + "loss": 1.6081, + "step": 821 + }, + { + "epoch": 0.8443759630200308, + "grad_norm": 2.963974714279175, + "learning_rate": 4.019794442329653e-06, + "loss": 1.6851, + "step": 822 + }, + { + "epoch": 0.8454031843862352, + "grad_norm": 2.4657533168792725, + "learning_rate": 4.0178911305671876e-06, + "loss": 1.5396, + "step": 823 + }, + { + "epoch": 0.8464304057524397, + "grad_norm": 2.5373876094818115, + "learning_rate": 4.015987818804721e-06, + "loss": 1.6186, + "step": 824 + }, + { + "epoch": 0.847457627118644, + "grad_norm": 2.5873284339904785, + "learning_rate": 4.014084507042254e-06, + "loss": 1.5618, + "step": 825 + }, + { + "epoch": 0.8484848484848485, + "grad_norm": 2.4308598041534424, + "learning_rate": 4.012181195279788e-06, + "loss": 1.5399, + "step": 826 + }, + { + "epoch": 0.8495120698510529, + "grad_norm": 2.5274946689605713, + "learning_rate": 4.010277883517321e-06, + "loss": 1.6375, + "step": 827 + }, + { + "epoch": 0.8505392912172574, + "grad_norm": 2.4370181560516357, + "learning_rate": 4.0083745717548535e-06, + "loss": 1.6313, + "step": 828 + }, + { + "epoch": 0.8515665125834617, + "grad_norm": 2.4524898529052734, + "learning_rate": 4.006471259992387e-06, + "loss": 1.5865, + "step": 829 + }, + { + "epoch": 0.8525937339496662, + "grad_norm": 2.5587961673736572, + "learning_rate": 4.00456794822992e-06, + "loss": 1.608, + "step": 830 + }, + { + "epoch": 0.8525937339496662, + "eval_loss": 1.94539475440979, + "eval_runtime": 49.4543, + "eval_samples_per_second": 20.221, + "eval_steps_per_second": 0.849, + "step": 830 + }, + { + "epoch": 0.8536209553158706, + "grad_norm": 2.512253761291504, + "learning_rate": 4.002664636467454e-06, + "loss": 1.6607, + "step": 831 + }, + { + "epoch": 0.8546481766820749, + "grad_norm": 2.438451051712036, + "learning_rate": 4.000761324704987e-06, + "loss": 1.6153, + "step": 832 + }, + { + "epoch": 0.8556753980482794, + "grad_norm": 2.5550496578216553, + "learning_rate": 3.99885801294252e-06, + "loss": 1.5489, + "step": 833 + }, + { + "epoch": 0.8567026194144838, + "grad_norm": 2.559732437133789, + "learning_rate": 3.996954701180054e-06, + "loss": 1.6045, + "step": 834 + }, + { + "epoch": 0.8577298407806883, + "grad_norm": 2.317692995071411, + "learning_rate": 3.995051389417587e-06, + "loss": 1.5316, + "step": 835 + }, + { + "epoch": 0.8587570621468926, + "grad_norm": 2.3742363452911377, + "learning_rate": 3.9931480776551204e-06, + "loss": 1.553, + "step": 836 + }, + { + "epoch": 0.8597842835130971, + "grad_norm": 2.5645551681518555, + "learning_rate": 3.991244765892654e-06, + "loss": 1.6558, + "step": 837 + }, + { + "epoch": 0.8608115048793015, + "grad_norm": 2.494830846786499, + "learning_rate": 3.989341454130186e-06, + "loss": 1.5713, + "step": 838 + }, + { + "epoch": 0.861838726245506, + "grad_norm": 2.416020393371582, + "learning_rate": 3.98743814236772e-06, + "loss": 1.5383, + "step": 839 + }, + { + "epoch": 0.8628659476117103, + "grad_norm": 2.6063015460968018, + "learning_rate": 3.985534830605253e-06, + "loss": 1.5831, + "step": 840 + }, + { + "epoch": 0.8628659476117103, + "eval_loss": 1.9441509246826172, + "eval_runtime": 46.0083, + "eval_samples_per_second": 21.735, + "eval_steps_per_second": 0.913, + "step": 840 + }, + { + "epoch": 0.8638931689779148, + "grad_norm": 2.5771663188934326, + "learning_rate": 3.983631518842786e-06, + "loss": 1.6077, + "step": 841 + }, + { + "epoch": 0.8649203903441192, + "grad_norm": 2.573119878768921, + "learning_rate": 3.98172820708032e-06, + "loss": 1.6301, + "step": 842 + }, + { + "epoch": 0.8659476117103235, + "grad_norm": 2.41396164894104, + "learning_rate": 3.979824895317854e-06, + "loss": 1.5758, + "step": 843 + }, + { + "epoch": 0.866974833076528, + "grad_norm": 2.532883882522583, + "learning_rate": 3.977921583555387e-06, + "loss": 1.5525, + "step": 844 + }, + { + "epoch": 0.8680020544427324, + "grad_norm": 2.669311285018921, + "learning_rate": 3.97601827179292e-06, + "loss": 1.6183, + "step": 845 + }, + { + "epoch": 0.8690292758089369, + "grad_norm": 2.366861343383789, + "learning_rate": 3.974114960030453e-06, + "loss": 1.5551, + "step": 846 + }, + { + "epoch": 0.8700564971751412, + "grad_norm": 2.3896048069000244, + "learning_rate": 3.9722116482679866e-06, + "loss": 1.5894, + "step": 847 + }, + { + "epoch": 0.8710837185413457, + "grad_norm": 2.6509673595428467, + "learning_rate": 3.97030833650552e-06, + "loss": 1.5813, + "step": 848 + }, + { + "epoch": 0.8721109399075501, + "grad_norm": 2.531726121902466, + "learning_rate": 3.968405024743053e-06, + "loss": 1.5899, + "step": 849 + }, + { + "epoch": 0.8731381612737545, + "grad_norm": 2.4117019176483154, + "learning_rate": 3.966501712980587e-06, + "loss": 1.5117, + "step": 850 + }, + { + "epoch": 0.8731381612737545, + "eval_loss": 1.9224250316619873, + "eval_runtime": 47.7854, + "eval_samples_per_second": 20.927, + "eval_steps_per_second": 0.879, + "step": 850 + }, + { + "epoch": 0.8741653826399589, + "grad_norm": 2.3789749145507812, + "learning_rate": 3.96459840121812e-06, + "loss": 1.5708, + "step": 851 + }, + { + "epoch": 0.8751926040061633, + "grad_norm": 2.386664867401123, + "learning_rate": 3.962695089455653e-06, + "loss": 1.5734, + "step": 852 + }, + { + "epoch": 0.8762198253723678, + "grad_norm": 2.4788997173309326, + "learning_rate": 3.960791777693187e-06, + "loss": 1.6315, + "step": 853 + }, + { + "epoch": 0.8772470467385721, + "grad_norm": 2.630183219909668, + "learning_rate": 3.95888846593072e-06, + "loss": 1.614, + "step": 854 + }, + { + "epoch": 0.8782742681047766, + "grad_norm": 2.3957736492156982, + "learning_rate": 3.956985154168253e-06, + "loss": 1.61, + "step": 855 + }, + { + "epoch": 0.879301489470981, + "grad_norm": 2.652036428451538, + "learning_rate": 3.955081842405786e-06, + "loss": 1.6463, + "step": 856 + }, + { + "epoch": 0.8803287108371854, + "grad_norm": 2.579867124557495, + "learning_rate": 3.953178530643319e-06, + "loss": 1.6624, + "step": 857 + }, + { + "epoch": 0.8813559322033898, + "grad_norm": 2.5118343830108643, + "learning_rate": 3.951275218880853e-06, + "loss": 1.6816, + "step": 858 + }, + { + "epoch": 0.8823831535695943, + "grad_norm": 2.5963029861450195, + "learning_rate": 3.949371907118386e-06, + "loss": 1.5296, + "step": 859 + }, + { + "epoch": 0.8834103749357987, + "grad_norm": 2.571241855621338, + "learning_rate": 3.9474685953559194e-06, + "loss": 1.6362, + "step": 860 + }, + { + "epoch": 0.8834103749357987, + "eval_loss": 1.931492567062378, + "eval_runtime": 52.2417, + "eval_samples_per_second": 19.142, + "eval_steps_per_second": 0.804, + "step": 860 + }, + { + "epoch": 0.884437596302003, + "grad_norm": 2.518036365509033, + "learning_rate": 3.945565283593453e-06, + "loss": 1.5865, + "step": 861 + }, + { + "epoch": 0.8854648176682075, + "grad_norm": 3.14587664604187, + "learning_rate": 3.943661971830986e-06, + "loss": 1.586, + "step": 862 + }, + { + "epoch": 0.8864920390344119, + "grad_norm": 2.686350107192993, + "learning_rate": 3.9417586600685195e-06, + "loss": 1.6385, + "step": 863 + }, + { + "epoch": 0.8875192604006163, + "grad_norm": 3.0324997901916504, + "learning_rate": 3.939855348306053e-06, + "loss": 1.6199, + "step": 864 + }, + { + "epoch": 0.8885464817668207, + "grad_norm": 2.6129438877105713, + "learning_rate": 3.937952036543586e-06, + "loss": 1.6049, + "step": 865 + }, + { + "epoch": 0.8895737031330252, + "grad_norm": 2.795729160308838, + "learning_rate": 3.93604872478112e-06, + "loss": 1.6027, + "step": 866 + }, + { + "epoch": 0.8906009244992296, + "grad_norm": 3.2352302074432373, + "learning_rate": 3.934145413018653e-06, + "loss": 1.6063, + "step": 867 + }, + { + "epoch": 0.891628145865434, + "grad_norm": 2.5050623416900635, + "learning_rate": 3.932242101256186e-06, + "loss": 1.5622, + "step": 868 + }, + { + "epoch": 0.8926553672316384, + "grad_norm": 2.8998777866363525, + "learning_rate": 3.93033878949372e-06, + "loss": 1.5729, + "step": 869 + }, + { + "epoch": 0.8936825885978429, + "grad_norm": 2.60239577293396, + "learning_rate": 3.928435477731253e-06, + "loss": 1.5377, + "step": 870 + }, + { + "epoch": 0.8936825885978429, + "eval_loss": 1.9311563968658447, + "eval_runtime": 46.8098, + "eval_samples_per_second": 21.363, + "eval_steps_per_second": 0.897, + "step": 870 + }, + { + "epoch": 0.8947098099640473, + "grad_norm": 2.5061111450195312, + "learning_rate": 3.9265321659687864e-06, + "loss": 1.6216, + "step": 871 + }, + { + "epoch": 0.8957370313302516, + "grad_norm": 2.7476489543914795, + "learning_rate": 3.924628854206319e-06, + "loss": 1.6357, + "step": 872 + }, + { + "epoch": 0.8967642526964561, + "grad_norm": 2.605536937713623, + "learning_rate": 3.922725542443852e-06, + "loss": 1.5983, + "step": 873 + }, + { + "epoch": 0.8977914740626605, + "grad_norm": 2.6024532318115234, + "learning_rate": 3.920822230681386e-06, + "loss": 1.5571, + "step": 874 + }, + { + "epoch": 0.8988186954288649, + "grad_norm": 2.7357826232910156, + "learning_rate": 3.918918918918919e-06, + "loss": 1.5802, + "step": 875 + }, + { + "epoch": 0.8998459167950693, + "grad_norm": 2.6406939029693604, + "learning_rate": 3.917015607156452e-06, + "loss": 1.6046, + "step": 876 + }, + { + "epoch": 0.9008731381612738, + "grad_norm": 2.4611780643463135, + "learning_rate": 3.915112295393986e-06, + "loss": 1.5803, + "step": 877 + }, + { + "epoch": 0.9019003595274782, + "grad_norm": 3.0608181953430176, + "learning_rate": 3.913208983631519e-06, + "loss": 1.6655, + "step": 878 + }, + { + "epoch": 0.9029275808936826, + "grad_norm": 2.482142448425293, + "learning_rate": 3.9113056718690525e-06, + "loss": 1.6444, + "step": 879 + }, + { + "epoch": 0.903954802259887, + "grad_norm": 2.4936728477478027, + "learning_rate": 3.909402360106586e-06, + "loss": 1.5596, + "step": 880 + }, + { + "epoch": 0.903954802259887, + "eval_loss": 1.933276653289795, + "eval_runtime": 48.0836, + "eval_samples_per_second": 20.797, + "eval_steps_per_second": 0.873, + "step": 880 + }, + { + "epoch": 0.9049820236260914, + "grad_norm": 2.7802248001098633, + "learning_rate": 3.907499048344119e-06, + "loss": 1.6133, + "step": 881 + }, + { + "epoch": 0.9060092449922958, + "grad_norm": 2.486586093902588, + "learning_rate": 3.905595736581652e-06, + "loss": 1.623, + "step": 882 + }, + { + "epoch": 0.9070364663585002, + "grad_norm": 2.477555751800537, + "learning_rate": 3.903692424819186e-06, + "loss": 1.6219, + "step": 883 + }, + { + "epoch": 0.9080636877247047, + "grad_norm": 2.5033347606658936, + "learning_rate": 3.901789113056719e-06, + "loss": 1.5612, + "step": 884 + }, + { + "epoch": 0.9090909090909091, + "grad_norm": 2.410464286804199, + "learning_rate": 3.899885801294253e-06, + "loss": 1.6243, + "step": 885 + }, + { + "epoch": 0.9101181304571135, + "grad_norm": 2.451402425765991, + "learning_rate": 3.897982489531786e-06, + "loss": 1.5961, + "step": 886 + }, + { + "epoch": 0.9111453518233179, + "grad_norm": 2.439730167388916, + "learning_rate": 3.896079177769319e-06, + "loss": 1.6471, + "step": 887 + }, + { + "epoch": 0.9121725731895224, + "grad_norm": 2.527719259262085, + "learning_rate": 3.894175866006853e-06, + "loss": 1.6262, + "step": 888 + }, + { + "epoch": 0.9131997945557268, + "grad_norm": 2.612992763519287, + "learning_rate": 3.892272554244385e-06, + "loss": 1.6533, + "step": 889 + }, + { + "epoch": 0.9142270159219312, + "grad_norm": 2.509989023208618, + "learning_rate": 3.890369242481919e-06, + "loss": 1.5596, + "step": 890 + }, + { + "epoch": 0.9142270159219312, + "eval_loss": 1.9224061965942383, + "eval_runtime": 45.5462, + "eval_samples_per_second": 21.956, + "eval_steps_per_second": 0.922, + "step": 890 + }, + { + "epoch": 0.9152542372881356, + "grad_norm": 2.3849823474884033, + "learning_rate": 3.888465930719452e-06, + "loss": 1.6336, + "step": 891 + }, + { + "epoch": 0.91628145865434, + "grad_norm": 2.4989709854125977, + "learning_rate": 3.886562618956985e-06, + "loss": 1.5733, + "step": 892 + }, + { + "epoch": 0.9173086800205444, + "grad_norm": 2.4675447940826416, + "learning_rate": 3.884659307194519e-06, + "loss": 1.5818, + "step": 893 + }, + { + "epoch": 0.9183359013867488, + "grad_norm": 2.3105645179748535, + "learning_rate": 3.882755995432052e-06, + "loss": 1.5251, + "step": 894 + }, + { + "epoch": 0.9193631227529533, + "grad_norm": 2.5410287380218506, + "learning_rate": 3.8808526836695854e-06, + "loss": 1.6089, + "step": 895 + }, + { + "epoch": 0.9203903441191577, + "grad_norm": 2.617091655731201, + "learning_rate": 3.878949371907119e-06, + "loss": 1.5631, + "step": 896 + }, + { + "epoch": 0.9214175654853621, + "grad_norm": 2.4979820251464844, + "learning_rate": 3.877046060144652e-06, + "loss": 1.5577, + "step": 897 + }, + { + "epoch": 0.9224447868515665, + "grad_norm": 2.4857399463653564, + "learning_rate": 3.8751427483821855e-06, + "loss": 1.56, + "step": 898 + }, + { + "epoch": 0.923472008217771, + "grad_norm": 2.456709146499634, + "learning_rate": 3.873239436619718e-06, + "loss": 1.594, + "step": 899 + }, + { + "epoch": 0.9244992295839753, + "grad_norm": 2.355168342590332, + "learning_rate": 3.871336124857251e-06, + "loss": 1.5371, + "step": 900 + }, + { + "epoch": 0.9244992295839753, + "eval_loss": 1.9357926845550537, + "eval_runtime": 44.4044, + "eval_samples_per_second": 22.52, + "eval_steps_per_second": 0.946, + "step": 900 + }, + { + "epoch": 0.9255264509501797, + "grad_norm": 2.5965569019317627, + "learning_rate": 3.869432813094785e-06, + "loss": 1.5966, + "step": 901 + }, + { + "epoch": 0.9265536723163842, + "grad_norm": 2.443662405014038, + "learning_rate": 3.867529501332318e-06, + "loss": 1.6053, + "step": 902 + }, + { + "epoch": 0.9275808936825886, + "grad_norm": 2.3888511657714844, + "learning_rate": 3.8656261895698515e-06, + "loss": 1.5921, + "step": 903 + }, + { + "epoch": 0.928608115048793, + "grad_norm": 2.4648637771606445, + "learning_rate": 3.863722877807386e-06, + "loss": 1.5879, + "step": 904 + }, + { + "epoch": 0.9296353364149974, + "grad_norm": 2.4636101722717285, + "learning_rate": 3.861819566044919e-06, + "loss": 1.6286, + "step": 905 + }, + { + "epoch": 0.9306625577812019, + "grad_norm": 2.722627878189087, + "learning_rate": 3.859916254282452e-06, + "loss": 1.6666, + "step": 906 + }, + { + "epoch": 0.9316897791474062, + "grad_norm": 2.550255298614502, + "learning_rate": 3.858012942519985e-06, + "loss": 1.5711, + "step": 907 + }, + { + "epoch": 0.9327170005136107, + "grad_norm": 2.510209798812866, + "learning_rate": 3.856109630757518e-06, + "loss": 1.5987, + "step": 908 + }, + { + "epoch": 0.9337442218798151, + "grad_norm": 2.6028876304626465, + "learning_rate": 3.854206318995052e-06, + "loss": 1.5682, + "step": 909 + }, + { + "epoch": 0.9347714432460195, + "grad_norm": 2.487908124923706, + "learning_rate": 3.852303007232585e-06, + "loss": 1.608, + "step": 910 + }, + { + "epoch": 0.9347714432460195, + "eval_loss": 1.93544602394104, + "eval_runtime": 41.5201, + "eval_samples_per_second": 24.085, + "eval_steps_per_second": 1.012, + "step": 910 + }, + { + "epoch": 0.9357986646122239, + "grad_norm": 2.4569835662841797, + "learning_rate": 3.850399695470118e-06, + "loss": 1.5492, + "step": 911 + }, + { + "epoch": 0.9368258859784283, + "grad_norm": 2.572781562805176, + "learning_rate": 3.848496383707652e-06, + "loss": 1.5633, + "step": 912 + }, + { + "epoch": 0.9378531073446328, + "grad_norm": 2.6213486194610596, + "learning_rate": 3.846593071945185e-06, + "loss": 1.5488, + "step": 913 + }, + { + "epoch": 0.9388803287108372, + "grad_norm": 2.6079821586608887, + "learning_rate": 3.8446897601827185e-06, + "loss": 1.5311, + "step": 914 + }, + { + "epoch": 0.9399075500770416, + "grad_norm": 2.6774539947509766, + "learning_rate": 3.842786448420252e-06, + "loss": 1.6564, + "step": 915 + }, + { + "epoch": 0.940934771443246, + "grad_norm": 2.5028746128082275, + "learning_rate": 3.840883136657785e-06, + "loss": 1.6551, + "step": 916 + }, + { + "epoch": 0.9419619928094505, + "grad_norm": 2.5254387855529785, + "learning_rate": 3.838979824895318e-06, + "loss": 1.6492, + "step": 917 + }, + { + "epoch": 0.9429892141756548, + "grad_norm": 2.5451207160949707, + "learning_rate": 3.837076513132851e-06, + "loss": 1.5306, + "step": 918 + }, + { + "epoch": 0.9440164355418593, + "grad_norm": 2.526526689529419, + "learning_rate": 3.8351732013703844e-06, + "loss": 1.6131, + "step": 919 + }, + { + "epoch": 0.9450436569080637, + "grad_norm": 2.454895257949829, + "learning_rate": 3.833269889607918e-06, + "loss": 1.5592, + "step": 920 + }, + { + "epoch": 0.9450436569080637, + "eval_loss": 1.9236185550689697, + "eval_runtime": 44.149, + "eval_samples_per_second": 22.651, + "eval_steps_per_second": 0.951, + "step": 920 + }, + { + "epoch": 0.9460708782742681, + "grad_norm": 2.6773581504821777, + "learning_rate": 3.831366577845451e-06, + "loss": 1.5498, + "step": 921 + }, + { + "epoch": 0.9470980996404725, + "grad_norm": 2.63554310798645, + "learning_rate": 3.8294632660829845e-06, + "loss": 1.5413, + "step": 922 + }, + { + "epoch": 0.9481253210066769, + "grad_norm": 2.7118968963623047, + "learning_rate": 3.827559954320518e-06, + "loss": 1.572, + "step": 923 + }, + { + "epoch": 0.9491525423728814, + "grad_norm": 2.4681396484375, + "learning_rate": 3.825656642558051e-06, + "loss": 1.6003, + "step": 924 + }, + { + "epoch": 0.9501797637390857, + "grad_norm": 2.387608051300049, + "learning_rate": 3.823753330795585e-06, + "loss": 1.6057, + "step": 925 + }, + { + "epoch": 0.9512069851052902, + "grad_norm": 2.426020383834839, + "learning_rate": 3.821850019033118e-06, + "loss": 1.5976, + "step": 926 + }, + { + "epoch": 0.9522342064714946, + "grad_norm": 2.446044445037842, + "learning_rate": 3.819946707270651e-06, + "loss": 1.5363, + "step": 927 + }, + { + "epoch": 0.9532614278376991, + "grad_norm": 2.4415299892425537, + "learning_rate": 3.818043395508185e-06, + "loss": 1.5685, + "step": 928 + }, + { + "epoch": 0.9542886492039034, + "grad_norm": 2.4006199836730957, + "learning_rate": 3.816140083745718e-06, + "loss": 1.5823, + "step": 929 + }, + { + "epoch": 0.9553158705701078, + "grad_norm": 2.6479294300079346, + "learning_rate": 3.8142367719832514e-06, + "loss": 1.6658, + "step": 930 + }, + { + "epoch": 0.9553158705701078, + "eval_loss": 1.9434653520584106, + "eval_runtime": 45.5063, + "eval_samples_per_second": 21.975, + "eval_steps_per_second": 0.923, + "step": 930 + }, + { + "epoch": 0.9563430919363123, + "grad_norm": 2.530367851257324, + "learning_rate": 3.8123334602207844e-06, + "loss": 1.544, + "step": 931 + }, + { + "epoch": 0.9573703133025167, + "grad_norm": 2.641665458679199, + "learning_rate": 3.8104301484583177e-06, + "loss": 1.5658, + "step": 932 + }, + { + "epoch": 0.9583975346687211, + "grad_norm": 2.3992276191711426, + "learning_rate": 3.808526836695851e-06, + "loss": 1.5524, + "step": 933 + }, + { + "epoch": 0.9594247560349255, + "grad_norm": 2.640503168106079, + "learning_rate": 3.8066235249333845e-06, + "loss": 1.6127, + "step": 934 + }, + { + "epoch": 0.96045197740113, + "grad_norm": 2.513735771179199, + "learning_rate": 3.804720213170918e-06, + "loss": 1.5901, + "step": 935 + }, + { + "epoch": 0.9614791987673343, + "grad_norm": 2.74481463432312, + "learning_rate": 3.8028169014084508e-06, + "loss": 1.5792, + "step": 936 + }, + { + "epoch": 0.9625064201335388, + "grad_norm": 2.5238802433013916, + "learning_rate": 3.800913589645984e-06, + "loss": 1.602, + "step": 937 + }, + { + "epoch": 0.9635336414997432, + "grad_norm": 2.5111966133117676, + "learning_rate": 3.7990102778835175e-06, + "loss": 1.607, + "step": 938 + }, + { + "epoch": 0.9645608628659477, + "grad_norm": 2.658449649810791, + "learning_rate": 3.797106966121051e-06, + "loss": 1.593, + "step": 939 + }, + { + "epoch": 0.965588084232152, + "grad_norm": 2.6884148120880127, + "learning_rate": 3.7952036543585842e-06, + "loss": 1.6576, + "step": 940 + }, + { + "epoch": 0.965588084232152, + "eval_loss": 1.9191738367080688, + "eval_runtime": 45.1928, + "eval_samples_per_second": 22.127, + "eval_steps_per_second": 0.929, + "step": 940 + }, + { + "epoch": 0.9666153055983564, + "grad_norm": 2.4238054752349854, + "learning_rate": 3.793300342596117e-06, + "loss": 1.6079, + "step": 941 + }, + { + "epoch": 0.9676425269645609, + "grad_norm": 2.505571126937866, + "learning_rate": 3.7913970308336505e-06, + "loss": 1.6299, + "step": 942 + }, + { + "epoch": 0.9686697483307652, + "grad_norm": 2.4143226146698, + "learning_rate": 3.789493719071184e-06, + "loss": 1.6158, + "step": 943 + }, + { + "epoch": 0.9696969696969697, + "grad_norm": 2.4664456844329834, + "learning_rate": 3.7875904073087177e-06, + "loss": 1.5672, + "step": 944 + }, + { + "epoch": 0.9707241910631741, + "grad_norm": 2.5686466693878174, + "learning_rate": 3.785687095546251e-06, + "loss": 1.5338, + "step": 945 + }, + { + "epoch": 0.9717514124293786, + "grad_norm": 2.4666805267333984, + "learning_rate": 3.7837837837837844e-06, + "loss": 1.5773, + "step": 946 + }, + { + "epoch": 0.9727786337955829, + "grad_norm": 2.6416635513305664, + "learning_rate": 3.7818804720213178e-06, + "loss": 1.5469, + "step": 947 + }, + { + "epoch": 0.9738058551617874, + "grad_norm": 2.721205711364746, + "learning_rate": 3.7799771602588507e-06, + "loss": 1.5278, + "step": 948 + }, + { + "epoch": 0.9748330765279918, + "grad_norm": 2.4834787845611572, + "learning_rate": 3.778073848496384e-06, + "loss": 1.52, + "step": 949 + }, + { + "epoch": 0.9758602978941961, + "grad_norm": 2.42047381401062, + "learning_rate": 3.7761705367339174e-06, + "loss": 1.5814, + "step": 950 + }, + { + "epoch": 0.9758602978941961, + "eval_loss": 1.9303041696548462, + "eval_runtime": 47.3843, + "eval_samples_per_second": 21.104, + "eval_steps_per_second": 0.886, + "step": 950 + }, + { + "epoch": 0.9768875192604006, + "grad_norm": 2.70369815826416, + "learning_rate": 3.7742672249714508e-06, + "loss": 1.6143, + "step": 951 + }, + { + "epoch": 0.977914740626605, + "grad_norm": 2.393588066101074, + "learning_rate": 3.772363913208984e-06, + "loss": 1.6023, + "step": 952 + }, + { + "epoch": 0.9789419619928095, + "grad_norm": 2.361304998397827, + "learning_rate": 3.770460601446517e-06, + "loss": 1.5679, + "step": 953 + }, + { + "epoch": 0.9799691833590138, + "grad_norm": 2.4044971466064453, + "learning_rate": 3.7685572896840504e-06, + "loss": 1.6007, + "step": 954 + }, + { + "epoch": 0.9809964047252183, + "grad_norm": 2.498539447784424, + "learning_rate": 3.766653977921584e-06, + "loss": 1.5429, + "step": 955 + }, + { + "epoch": 0.9820236260914227, + "grad_norm": 2.457162857055664, + "learning_rate": 3.764750666159117e-06, + "loss": 1.4861, + "step": 956 + }, + { + "epoch": 0.9830508474576272, + "grad_norm": 2.52756404876709, + "learning_rate": 3.7628473543966505e-06, + "loss": 1.5848, + "step": 957 + }, + { + "epoch": 0.9840780688238315, + "grad_norm": 2.503998279571533, + "learning_rate": 3.7609440426341835e-06, + "loss": 1.5435, + "step": 958 + }, + { + "epoch": 0.9851052901900359, + "grad_norm": 2.461087226867676, + "learning_rate": 3.759040730871717e-06, + "loss": 1.6344, + "step": 959 + }, + { + "epoch": 0.9861325115562404, + "grad_norm": 2.4149672985076904, + "learning_rate": 3.75713741910925e-06, + "loss": 1.5273, + "step": 960 + }, + { + "epoch": 0.9861325115562404, + "eval_loss": 1.928680658340454, + "eval_runtime": 48.7892, + "eval_samples_per_second": 20.496, + "eval_steps_per_second": 0.861, + "step": 960 + }, + { + "epoch": 0.9871597329224447, + "grad_norm": 2.5741734504699707, + "learning_rate": 3.7552341073467836e-06, + "loss": 1.5244, + "step": 961 + }, + { + "epoch": 0.9881869542886492, + "grad_norm": 2.5517220497131348, + "learning_rate": 3.753330795584317e-06, + "loss": 1.5695, + "step": 962 + }, + { + "epoch": 0.9892141756548536, + "grad_norm": 2.5910379886627197, + "learning_rate": 3.75142748382185e-06, + "loss": 1.6322, + "step": 963 + }, + { + "epoch": 0.9902413970210581, + "grad_norm": 2.4780521392822266, + "learning_rate": 3.749524172059384e-06, + "loss": 1.5292, + "step": 964 + }, + { + "epoch": 0.9912686183872624, + "grad_norm": 2.6141762733459473, + "learning_rate": 3.747620860296917e-06, + "loss": 1.5809, + "step": 965 + }, + { + "epoch": 0.9922958397534669, + "grad_norm": 2.467606544494629, + "learning_rate": 3.7457175485344504e-06, + "loss": 1.6123, + "step": 966 + }, + { + "epoch": 0.9933230611196713, + "grad_norm": 2.6112184524536133, + "learning_rate": 3.7438142367719837e-06, + "loss": 1.6226, + "step": 967 + }, + { + "epoch": 0.9943502824858758, + "grad_norm": 2.352513313293457, + "learning_rate": 3.741910925009517e-06, + "loss": 1.5928, + "step": 968 + }, + { + "epoch": 0.9953775038520801, + "grad_norm": 2.4357361793518066, + "learning_rate": 3.7400076132470505e-06, + "loss": 1.5484, + "step": 969 + }, + { + "epoch": 0.9964047252182845, + "grad_norm": 2.5168254375457764, + "learning_rate": 3.7381043014845834e-06, + "loss": 1.5722, + "step": 970 + }, + { + "epoch": 0.9964047252182845, + "eval_loss": 1.929540753364563, + "eval_runtime": 49.7232, + "eval_samples_per_second": 20.111, + "eval_steps_per_second": 0.845, + "step": 970 + }, + { + "epoch": 0.997431946584489, + "grad_norm": 2.475173234939575, + "learning_rate": 3.7362009897221168e-06, + "loss": 1.5952, + "step": 971 + }, + { + "epoch": 0.9984591679506933, + "grad_norm": 2.5278515815734863, + "learning_rate": 3.73429767795965e-06, + "loss": 1.5769, + "step": 972 + }, + { + "epoch": 0.9994863893168978, + "grad_norm": 2.4988675117492676, + "learning_rate": 3.7323943661971835e-06, + "loss": 1.5912, + "step": 973 + }, + { + "epoch": 1.0005136106831023, + "grad_norm": 2.406353712081909, + "learning_rate": 3.730491054434717e-06, + "loss": 1.5022, + "step": 974 + }, + { + "epoch": 1.0015408320493067, + "grad_norm": 2.611318349838257, + "learning_rate": 3.72858774267225e-06, + "loss": 1.4633, + "step": 975 + }, + { + "epoch": 1.002568053415511, + "grad_norm": 2.504613161087036, + "learning_rate": 3.726684430909783e-06, + "loss": 1.5405, + "step": 976 + }, + { + "epoch": 1.0035952747817154, + "grad_norm": 2.6135828495025635, + "learning_rate": 3.7247811191473165e-06, + "loss": 1.5756, + "step": 977 + }, + { + "epoch": 1.00462249614792, + "grad_norm": 2.692592144012451, + "learning_rate": 3.72287780738485e-06, + "loss": 1.5777, + "step": 978 + }, + { + "epoch": 1.0056497175141244, + "grad_norm": 2.660323143005371, + "learning_rate": 3.7209744956223832e-06, + "loss": 1.5488, + "step": 979 + }, + { + "epoch": 1.0066769388803287, + "grad_norm": 2.647883415222168, + "learning_rate": 3.719071183859916e-06, + "loss": 1.5619, + "step": 980 + }, + { + "epoch": 1.0066769388803287, + "eval_loss": 1.9224871397018433, + "eval_runtime": 52.6149, + "eval_samples_per_second": 19.006, + "eval_steps_per_second": 0.798, + "step": 980 + }, + { + "epoch": 1.007704160246533, + "grad_norm": 2.6186633110046387, + "learning_rate": 3.7171678720974495e-06, + "loss": 1.492, + "step": 981 + }, + { + "epoch": 1.0087313816127375, + "grad_norm": 2.7562549114227295, + "learning_rate": 3.715264560334983e-06, + "loss": 1.5932, + "step": 982 + }, + { + "epoch": 1.009758602978942, + "grad_norm": 2.618856430053711, + "learning_rate": 3.7133612485725163e-06, + "loss": 1.5514, + "step": 983 + }, + { + "epoch": 1.0107858243451464, + "grad_norm": 2.522064685821533, + "learning_rate": 3.71145793681005e-06, + "loss": 1.5699, + "step": 984 + }, + { + "epoch": 1.0118130457113508, + "grad_norm": 2.627108335494995, + "learning_rate": 3.7095546250475834e-06, + "loss": 1.5033, + "step": 985 + }, + { + "epoch": 1.0128402670775551, + "grad_norm": 2.6393308639526367, + "learning_rate": 3.7076513132851168e-06, + "loss": 1.6064, + "step": 986 + }, + { + "epoch": 1.0138674884437597, + "grad_norm": 2.560675621032715, + "learning_rate": 3.7057480015226497e-06, + "loss": 1.5167, + "step": 987 + }, + { + "epoch": 1.014894709809964, + "grad_norm": 2.4677579402923584, + "learning_rate": 3.703844689760183e-06, + "loss": 1.4953, + "step": 988 + }, + { + "epoch": 1.0159219311761685, + "grad_norm": 2.9904487133026123, + "learning_rate": 3.7019413779977164e-06, + "loss": 1.5786, + "step": 989 + }, + { + "epoch": 1.0169491525423728, + "grad_norm": 2.7342119216918945, + "learning_rate": 3.70003806623525e-06, + "loss": 1.5459, + "step": 990 + }, + { + "epoch": 1.0169491525423728, + "eval_loss": 1.929221510887146, + "eval_runtime": 45.9704, + "eval_samples_per_second": 21.753, + "eval_steps_per_second": 0.914, + "step": 990 + }, + { + "epoch": 1.0179763739085772, + "grad_norm": 2.723241090774536, + "learning_rate": 3.698134754472783e-06, + "loss": 1.5317, + "step": 991 + }, + { + "epoch": 1.0190035952747818, + "grad_norm": 2.562356948852539, + "learning_rate": 3.696231442710316e-06, + "loss": 1.5957, + "step": 992 + }, + { + "epoch": 1.0200308166409862, + "grad_norm": 2.635737419128418, + "learning_rate": 3.6943281309478495e-06, + "loss": 1.5404, + "step": 993 + }, + { + "epoch": 1.0210580380071905, + "grad_norm": 2.588714361190796, + "learning_rate": 3.692424819185383e-06, + "loss": 1.5618, + "step": 994 + }, + { + "epoch": 1.022085259373395, + "grad_norm": 2.61643385887146, + "learning_rate": 3.690521507422916e-06, + "loss": 1.5625, + "step": 995 + }, + { + "epoch": 1.0231124807395995, + "grad_norm": 2.566077470779419, + "learning_rate": 3.6886181956604496e-06, + "loss": 1.5051, + "step": 996 + }, + { + "epoch": 1.0241397021058039, + "grad_norm": 2.5667169094085693, + "learning_rate": 3.686714883897983e-06, + "loss": 1.5142, + "step": 997 + }, + { + "epoch": 1.0251669234720082, + "grad_norm": 2.839078903198242, + "learning_rate": 3.684811572135516e-06, + "loss": 1.5833, + "step": 998 + }, + { + "epoch": 1.0261941448382126, + "grad_norm": 2.651643753051758, + "learning_rate": 3.6829082603730492e-06, + "loss": 1.5735, + "step": 999 + }, + { + "epoch": 1.027221366204417, + "grad_norm": 2.6310813426971436, + "learning_rate": 3.6810049486105826e-06, + "loss": 1.5412, + "step": 1000 + }, + { + "epoch": 1.027221366204417, + "eval_loss": 1.9219006299972534, + "eval_runtime": 53.0111, + "eval_samples_per_second": 18.864, + "eval_steps_per_second": 0.792, + "step": 1000 + }, + { + "epoch": 1.0282485875706215, + "grad_norm": 2.721156120300293, + "learning_rate": 3.679101636848116e-06, + "loss": 1.5686, + "step": 1001 + }, + { + "epoch": 1.029275808936826, + "grad_norm": 2.8999130725860596, + "learning_rate": 3.6771983250856493e-06, + "loss": 1.5699, + "step": 1002 + }, + { + "epoch": 1.0303030303030303, + "grad_norm": 2.8879904747009277, + "learning_rate": 3.6752950133231822e-06, + "loss": 1.5353, + "step": 1003 + }, + { + "epoch": 1.0313302516692346, + "grad_norm": 2.4753830432891846, + "learning_rate": 3.6733917015607156e-06, + "loss": 1.5684, + "step": 1004 + }, + { + "epoch": 1.0323574730354392, + "grad_norm": 2.6306774616241455, + "learning_rate": 3.6714883897982494e-06, + "loss": 1.5741, + "step": 1005 + }, + { + "epoch": 1.0333846944016436, + "grad_norm": 2.8131468296051025, + "learning_rate": 3.6695850780357828e-06, + "loss": 1.5127, + "step": 1006 + }, + { + "epoch": 1.034411915767848, + "grad_norm": 2.6741905212402344, + "learning_rate": 3.667681766273316e-06, + "loss": 1.5135, + "step": 1007 + }, + { + "epoch": 1.0354391371340523, + "grad_norm": 2.6110734939575195, + "learning_rate": 3.6657784545108495e-06, + "loss": 1.5287, + "step": 1008 + }, + { + "epoch": 1.0364663585002567, + "grad_norm": 2.6415274143218994, + "learning_rate": 3.6638751427483824e-06, + "loss": 1.5978, + "step": 1009 + }, + { + "epoch": 1.0374935798664613, + "grad_norm": 2.5623817443847656, + "learning_rate": 3.6619718309859158e-06, + "loss": 1.5586, + "step": 1010 + }, + { + "epoch": 1.0374935798664613, + "eval_loss": 1.915805459022522, + "eval_runtime": 43.3255, + "eval_samples_per_second": 23.081, + "eval_steps_per_second": 0.969, + "step": 1010 + }, + { + "epoch": 1.0385208012326657, + "grad_norm": 2.723954916000366, + "learning_rate": 3.660068519223449e-06, + "loss": 1.6013, + "step": 1011 + }, + { + "epoch": 1.03954802259887, + "grad_norm": 2.6564769744873047, + "learning_rate": 3.6581652074609825e-06, + "loss": 1.5653, + "step": 1012 + }, + { + "epoch": 1.0405752439650744, + "grad_norm": 2.6694817543029785, + "learning_rate": 3.656261895698516e-06, + "loss": 1.5297, + "step": 1013 + }, + { + "epoch": 1.041602465331279, + "grad_norm": 2.546647071838379, + "learning_rate": 3.6543585839360492e-06, + "loss": 1.5192, + "step": 1014 + }, + { + "epoch": 1.0426296866974833, + "grad_norm": 2.691000461578369, + "learning_rate": 3.652455272173582e-06, + "loss": 1.5284, + "step": 1015 + }, + { + "epoch": 1.0436569080636877, + "grad_norm": 2.627333641052246, + "learning_rate": 3.6505519604111155e-06, + "loss": 1.5239, + "step": 1016 + }, + { + "epoch": 1.044684129429892, + "grad_norm": 2.6125097274780273, + "learning_rate": 3.648648648648649e-06, + "loss": 1.5357, + "step": 1017 + }, + { + "epoch": 1.0457113507960965, + "grad_norm": 2.6235790252685547, + "learning_rate": 3.6467453368861823e-06, + "loss": 1.5831, + "step": 1018 + }, + { + "epoch": 1.046738572162301, + "grad_norm": 2.5727598667144775, + "learning_rate": 3.6448420251237156e-06, + "loss": 1.5264, + "step": 1019 + }, + { + "epoch": 1.0477657935285054, + "grad_norm": 2.71970534324646, + "learning_rate": 3.6429387133612486e-06, + "loss": 1.5593, + "step": 1020 + }, + { + "epoch": 1.0477657935285054, + "eval_loss": 1.9139868021011353, + "eval_runtime": 46.8291, + "eval_samples_per_second": 21.354, + "eval_steps_per_second": 0.897, + "step": 1020 + }, + { + "epoch": 1.0487930148947098, + "grad_norm": 2.6479101181030273, + "learning_rate": 3.641035401598782e-06, + "loss": 1.5875, + "step": 1021 + }, + { + "epoch": 1.0498202362609141, + "grad_norm": 2.6313018798828125, + "learning_rate": 3.6391320898363153e-06, + "loss": 1.5222, + "step": 1022 + }, + { + "epoch": 1.0508474576271187, + "grad_norm": 2.679009437561035, + "learning_rate": 3.6372287780738487e-06, + "loss": 1.5678, + "step": 1023 + }, + { + "epoch": 1.051874678993323, + "grad_norm": 2.8511667251586914, + "learning_rate": 3.635325466311382e-06, + "loss": 1.5401, + "step": 1024 + }, + { + "epoch": 1.0529019003595275, + "grad_norm": 2.717280864715576, + "learning_rate": 3.633422154548916e-06, + "loss": 1.5686, + "step": 1025 + }, + { + "epoch": 1.0539291217257318, + "grad_norm": 2.6042356491088867, + "learning_rate": 3.6315188427864487e-06, + "loss": 1.5945, + "step": 1026 + }, + { + "epoch": 1.0549563430919364, + "grad_norm": 2.601451873779297, + "learning_rate": 3.629615531023982e-06, + "loss": 1.556, + "step": 1027 + }, + { + "epoch": 1.0559835644581408, + "grad_norm": 2.5195374488830566, + "learning_rate": 3.6277122192615155e-06, + "loss": 1.5026, + "step": 1028 + }, + { + "epoch": 1.0570107858243452, + "grad_norm": 2.604405164718628, + "learning_rate": 3.625808907499049e-06, + "loss": 1.5629, + "step": 1029 + }, + { + "epoch": 1.0580380071905495, + "grad_norm": 2.6300103664398193, + "learning_rate": 3.623905595736582e-06, + "loss": 1.5253, + "step": 1030 + }, + { + "epoch": 1.0580380071905495, + "eval_loss": 1.9272329807281494, + "eval_runtime": 46.1105, + "eval_samples_per_second": 21.687, + "eval_steps_per_second": 0.911, + "step": 1030 + }, + { + "epoch": 1.059065228556754, + "grad_norm": 2.663590908050537, + "learning_rate": 3.6220022839741156e-06, + "loss": 1.5223, + "step": 1031 + }, + { + "epoch": 1.0600924499229585, + "grad_norm": 2.773226499557495, + "learning_rate": 3.6200989722116485e-06, + "loss": 1.5052, + "step": 1032 + }, + { + "epoch": 1.0611196712891628, + "grad_norm": 2.746915817260742, + "learning_rate": 3.618195660449182e-06, + "loss": 1.5371, + "step": 1033 + }, + { + "epoch": 1.0621468926553672, + "grad_norm": 3.0018815994262695, + "learning_rate": 3.6162923486867152e-06, + "loss": 1.5494, + "step": 1034 + }, + { + "epoch": 1.0631741140215716, + "grad_norm": 2.77616810798645, + "learning_rate": 3.6143890369242486e-06, + "loss": 1.4878, + "step": 1035 + }, + { + "epoch": 1.064201335387776, + "grad_norm": 2.638270854949951, + "learning_rate": 3.612485725161782e-06, + "loss": 1.5028, + "step": 1036 + }, + { + "epoch": 1.0652285567539805, + "grad_norm": 2.7400267124176025, + "learning_rate": 3.610582413399315e-06, + "loss": 1.5364, + "step": 1037 + }, + { + "epoch": 1.066255778120185, + "grad_norm": 2.603987455368042, + "learning_rate": 3.6086791016368482e-06, + "loss": 1.5259, + "step": 1038 + }, + { + "epoch": 1.0672829994863893, + "grad_norm": 2.5326781272888184, + "learning_rate": 3.6067757898743816e-06, + "loss": 1.5346, + "step": 1039 + }, + { + "epoch": 1.0683102208525936, + "grad_norm": 2.677980422973633, + "learning_rate": 3.604872478111915e-06, + "loss": 1.6011, + "step": 1040 + }, + { + "epoch": 1.0683102208525936, + "eval_loss": 1.9431331157684326, + "eval_runtime": 46.6449, + "eval_samples_per_second": 21.439, + "eval_steps_per_second": 0.9, + "step": 1040 + }, + { + "epoch": 1.0693374422187982, + "grad_norm": 2.7540171146392822, + "learning_rate": 3.6029691663494483e-06, + "loss": 1.5988, + "step": 1041 + }, + { + "epoch": 1.0703646635850026, + "grad_norm": 2.6210403442382812, + "learning_rate": 3.6010658545869813e-06, + "loss": 1.5373, + "step": 1042 + }, + { + "epoch": 1.071391884951207, + "grad_norm": 2.7578794956207275, + "learning_rate": 3.5991625428245146e-06, + "loss": 1.5707, + "step": 1043 + }, + { + "epoch": 1.0724191063174113, + "grad_norm": 2.787123203277588, + "learning_rate": 3.597259231062048e-06, + "loss": 1.5666, + "step": 1044 + }, + { + "epoch": 1.073446327683616, + "grad_norm": 2.5755369663238525, + "learning_rate": 3.5953559192995818e-06, + "loss": 1.5005, + "step": 1045 + }, + { + "epoch": 1.0744735490498203, + "grad_norm": 2.644008159637451, + "learning_rate": 3.593452607537115e-06, + "loss": 1.5408, + "step": 1046 + }, + { + "epoch": 1.0755007704160247, + "grad_norm": 2.5818886756896973, + "learning_rate": 3.5915492957746485e-06, + "loss": 1.6064, + "step": 1047 + }, + { + "epoch": 1.076527991782229, + "grad_norm": 2.6413795948028564, + "learning_rate": 3.589645984012182e-06, + "loss": 1.5481, + "step": 1048 + }, + { + "epoch": 1.0775552131484334, + "grad_norm": 2.8570823669433594, + "learning_rate": 3.587742672249715e-06, + "loss": 1.5713, + "step": 1049 + }, + { + "epoch": 1.078582434514638, + "grad_norm": 2.6338746547698975, + "learning_rate": 3.585839360487248e-06, + "loss": 1.5252, + "step": 1050 + }, + { + "epoch": 1.078582434514638, + "eval_loss": 1.9467660188674927, + "eval_runtime": 45.8203, + "eval_samples_per_second": 21.824, + "eval_steps_per_second": 0.917, + "step": 1050 + }, + { + "epoch": 1.0796096558808423, + "grad_norm": 2.7295382022857666, + "learning_rate": 3.5839360487247815e-06, + "loss": 1.515, + "step": 1051 + }, + { + "epoch": 1.0806368772470467, + "grad_norm": 2.7476232051849365, + "learning_rate": 3.582032736962315e-06, + "loss": 1.5301, + "step": 1052 + }, + { + "epoch": 1.081664098613251, + "grad_norm": 2.8701605796813965, + "learning_rate": 3.5801294251998483e-06, + "loss": 1.5971, + "step": 1053 + }, + { + "epoch": 1.0826913199794557, + "grad_norm": 2.638195276260376, + "learning_rate": 3.578226113437381e-06, + "loss": 1.4917, + "step": 1054 + }, + { + "epoch": 1.08371854134566, + "grad_norm": 2.587324857711792, + "learning_rate": 3.5763228016749146e-06, + "loss": 1.5349, + "step": 1055 + }, + { + "epoch": 1.0847457627118644, + "grad_norm": 2.645148754119873, + "learning_rate": 3.574419489912448e-06, + "loss": 1.6063, + "step": 1056 + }, + { + "epoch": 1.0857729840780688, + "grad_norm": 2.580125331878662, + "learning_rate": 3.5725161781499813e-06, + "loss": 1.5504, + "step": 1057 + }, + { + "epoch": 1.0868002054442731, + "grad_norm": 2.730672836303711, + "learning_rate": 3.5706128663875147e-06, + "loss": 1.5272, + "step": 1058 + }, + { + "epoch": 1.0878274268104777, + "grad_norm": 2.564277172088623, + "learning_rate": 3.5687095546250476e-06, + "loss": 1.5471, + "step": 1059 + }, + { + "epoch": 1.088854648176682, + "grad_norm": 2.603226661682129, + "learning_rate": 3.566806242862581e-06, + "loss": 1.5759, + "step": 1060 + }, + { + "epoch": 1.088854648176682, + "eval_loss": 1.9485995769500732, + "eval_runtime": 45.476, + "eval_samples_per_second": 21.99, + "eval_steps_per_second": 0.924, + "step": 1060 + }, + { + "epoch": 1.0898818695428865, + "grad_norm": 2.7261931896209717, + "learning_rate": 3.5649029311001143e-06, + "loss": 1.5455, + "step": 1061 + }, + { + "epoch": 1.0909090909090908, + "grad_norm": 2.5638797283172607, + "learning_rate": 3.5629996193376477e-06, + "loss": 1.5236, + "step": 1062 + }, + { + "epoch": 1.0919363122752954, + "grad_norm": 2.8362395763397217, + "learning_rate": 3.561096307575181e-06, + "loss": 1.5376, + "step": 1063 + }, + { + "epoch": 1.0929635336414998, + "grad_norm": 2.6815669536590576, + "learning_rate": 3.559192995812714e-06, + "loss": 1.5681, + "step": 1064 + }, + { + "epoch": 1.0939907550077042, + "grad_norm": 2.691917896270752, + "learning_rate": 3.557289684050248e-06, + "loss": 1.5868, + "step": 1065 + }, + { + "epoch": 1.0950179763739085, + "grad_norm": 2.8439927101135254, + "learning_rate": 3.555386372287781e-06, + "loss": 1.5763, + "step": 1066 + }, + { + "epoch": 1.0960451977401129, + "grad_norm": 2.6857359409332275, + "learning_rate": 3.5534830605253145e-06, + "loss": 1.5784, + "step": 1067 + }, + { + "epoch": 1.0970724191063175, + "grad_norm": 2.532336950302124, + "learning_rate": 3.551579748762848e-06, + "loss": 1.5409, + "step": 1068 + }, + { + "epoch": 1.0980996404725218, + "grad_norm": 2.6567013263702393, + "learning_rate": 3.5496764370003812e-06, + "loss": 1.5484, + "step": 1069 + }, + { + "epoch": 1.0991268618387262, + "grad_norm": 2.800318717956543, + "learning_rate": 3.5477731252379146e-06, + "loss": 1.5247, + "step": 1070 + }, + { + "epoch": 1.0991268618387262, + "eval_loss": 1.9268279075622559, + "eval_runtime": 45.4409, + "eval_samples_per_second": 22.007, + "eval_steps_per_second": 0.924, + "step": 1070 + }, + { + "epoch": 1.1001540832049306, + "grad_norm": 2.6051485538482666, + "learning_rate": 3.5458698134754475e-06, + "loss": 1.5002, + "step": 1071 + }, + { + "epoch": 1.1011813045711352, + "grad_norm": 2.590815544128418, + "learning_rate": 3.543966501712981e-06, + "loss": 1.4982, + "step": 1072 + }, + { + "epoch": 1.1022085259373395, + "grad_norm": 2.6754868030548096, + "learning_rate": 3.5420631899505142e-06, + "loss": 1.5303, + "step": 1073 + }, + { + "epoch": 1.103235747303544, + "grad_norm": 2.8132805824279785, + "learning_rate": 3.5401598781880476e-06, + "loss": 1.6188, + "step": 1074 + }, + { + "epoch": 1.1042629686697483, + "grad_norm": 2.6267812252044678, + "learning_rate": 3.538256566425581e-06, + "loss": 1.442, + "step": 1075 + }, + { + "epoch": 1.1052901900359529, + "grad_norm": 2.945460557937622, + "learning_rate": 3.536353254663114e-06, + "loss": 1.5787, + "step": 1076 + }, + { + "epoch": 1.1063174114021572, + "grad_norm": 2.831702709197998, + "learning_rate": 3.5344499429006473e-06, + "loss": 1.5396, + "step": 1077 + }, + { + "epoch": 1.1073446327683616, + "grad_norm": 2.614482879638672, + "learning_rate": 3.5325466311381806e-06, + "loss": 1.5101, + "step": 1078 + }, + { + "epoch": 1.108371854134566, + "grad_norm": 2.809013605117798, + "learning_rate": 3.530643319375714e-06, + "loss": 1.5678, + "step": 1079 + }, + { + "epoch": 1.1093990755007703, + "grad_norm": 2.6645243167877197, + "learning_rate": 3.5287400076132474e-06, + "loss": 1.4762, + "step": 1080 + }, + { + "epoch": 1.1093990755007703, + "eval_loss": 1.9386240243911743, + "eval_runtime": 43.5259, + "eval_samples_per_second": 22.975, + "eval_steps_per_second": 0.965, + "step": 1080 + }, + { + "epoch": 1.110426296866975, + "grad_norm": 2.5597710609436035, + "learning_rate": 3.5268366958507803e-06, + "loss": 1.5208, + "step": 1081 + }, + { + "epoch": 1.1114535182331793, + "grad_norm": 2.6656508445739746, + "learning_rate": 3.5249333840883137e-06, + "loss": 1.5548, + "step": 1082 + }, + { + "epoch": 1.1124807395993837, + "grad_norm": 2.651195526123047, + "learning_rate": 3.523030072325847e-06, + "loss": 1.546, + "step": 1083 + }, + { + "epoch": 1.113507960965588, + "grad_norm": 2.946263551712036, + "learning_rate": 3.5211267605633804e-06, + "loss": 1.5058, + "step": 1084 + }, + { + "epoch": 1.1145351823317924, + "grad_norm": 2.7462689876556396, + "learning_rate": 3.519223448800914e-06, + "loss": 1.5213, + "step": 1085 + }, + { + "epoch": 1.115562403697997, + "grad_norm": 2.803926944732666, + "learning_rate": 3.5173201370384475e-06, + "loss": 1.5266, + "step": 1086 + }, + { + "epoch": 1.1165896250642013, + "grad_norm": 2.6583237648010254, + "learning_rate": 3.515416825275981e-06, + "loss": 1.5362, + "step": 1087 + }, + { + "epoch": 1.1176168464304057, + "grad_norm": 2.659313201904297, + "learning_rate": 3.513513513513514e-06, + "loss": 1.5365, + "step": 1088 + }, + { + "epoch": 1.11864406779661, + "grad_norm": 2.7181215286254883, + "learning_rate": 3.511610201751047e-06, + "loss": 1.5584, + "step": 1089 + }, + { + "epoch": 1.1196712891628147, + "grad_norm": 2.543297052383423, + "learning_rate": 3.5097068899885806e-06, + "loss": 1.4599, + "step": 1090 + }, + { + "epoch": 1.1196712891628147, + "eval_loss": 1.9393529891967773, + "eval_runtime": 44.5005, + "eval_samples_per_second": 22.472, + "eval_steps_per_second": 0.944, + "step": 1090 + }, + { + "epoch": 1.120698510529019, + "grad_norm": 2.8569319248199463, + "learning_rate": 3.507803578226114e-06, + "loss": 1.5308, + "step": 1091 + }, + { + "epoch": 1.1217257318952234, + "grad_norm": 2.795915365219116, + "learning_rate": 3.5059002664636473e-06, + "loss": 1.6092, + "step": 1092 + }, + { + "epoch": 1.1227529532614278, + "grad_norm": 2.659125566482544, + "learning_rate": 3.5039969547011802e-06, + "loss": 1.4701, + "step": 1093 + }, + { + "epoch": 1.1237801746276324, + "grad_norm": 2.7170166969299316, + "learning_rate": 3.5020936429387136e-06, + "loss": 1.5232, + "step": 1094 + }, + { + "epoch": 1.1248073959938367, + "grad_norm": 2.6485073566436768, + "learning_rate": 3.500190331176247e-06, + "loss": 1.5274, + "step": 1095 + }, + { + "epoch": 1.125834617360041, + "grad_norm": 2.831904888153076, + "learning_rate": 3.4982870194137803e-06, + "loss": 1.5833, + "step": 1096 + }, + { + "epoch": 1.1268618387262455, + "grad_norm": 2.877516984939575, + "learning_rate": 3.4963837076513137e-06, + "loss": 1.5173, + "step": 1097 + }, + { + "epoch": 1.1278890600924498, + "grad_norm": 2.5502521991729736, + "learning_rate": 3.4944803958888466e-06, + "loss": 1.5119, + "step": 1098 + }, + { + "epoch": 1.1289162814586544, + "grad_norm": 2.530945301055908, + "learning_rate": 3.49257708412638e-06, + "loss": 1.4573, + "step": 1099 + }, + { + "epoch": 1.1299435028248588, + "grad_norm": 2.7266969680786133, + "learning_rate": 3.4906737723639133e-06, + "loss": 1.5812, + "step": 1100 + }, + { + "epoch": 1.1299435028248588, + "eval_loss": 1.9370596408843994, + "eval_runtime": 47.2327, + "eval_samples_per_second": 21.172, + "eval_steps_per_second": 0.889, + "step": 1100 + }, + { + "epoch": 1.1309707241910631, + "grad_norm": 2.8846628665924072, + "learning_rate": 3.4887704606014467e-06, + "loss": 1.5945, + "step": 1101 + }, + { + "epoch": 1.1319979455572675, + "grad_norm": 2.5253827571868896, + "learning_rate": 3.48686714883898e-06, + "loss": 1.5391, + "step": 1102 + }, + { + "epoch": 1.133025166923472, + "grad_norm": 2.6929149627685547, + "learning_rate": 3.484963837076513e-06, + "loss": 1.5767, + "step": 1103 + }, + { + "epoch": 1.1340523882896765, + "grad_norm": 2.7702295780181885, + "learning_rate": 3.4830605253140464e-06, + "loss": 1.5206, + "step": 1104 + }, + { + "epoch": 1.1350796096558808, + "grad_norm": 2.6476149559020996, + "learning_rate": 3.4811572135515797e-06, + "loss": 1.4735, + "step": 1105 + }, + { + "epoch": 1.1361068310220852, + "grad_norm": 2.5630619525909424, + "learning_rate": 3.4792539017891135e-06, + "loss": 1.5109, + "step": 1106 + }, + { + "epoch": 1.1371340523882898, + "grad_norm": 2.549581289291382, + "learning_rate": 3.477350590026647e-06, + "loss": 1.4919, + "step": 1107 + }, + { + "epoch": 1.1381612737544942, + "grad_norm": 2.8687074184417725, + "learning_rate": 3.4754472782641802e-06, + "loss": 1.5658, + "step": 1108 + }, + { + "epoch": 1.1391884951206985, + "grad_norm": 2.799149751663208, + "learning_rate": 3.4735439665017136e-06, + "loss": 1.5582, + "step": 1109 + }, + { + "epoch": 1.140215716486903, + "grad_norm": 2.8335328102111816, + "learning_rate": 3.4716406547392465e-06, + "loss": 1.5217, + "step": 1110 + }, + { + "epoch": 1.140215716486903, + "eval_loss": 1.9011008739471436, + "eval_runtime": 42.9222, + "eval_samples_per_second": 23.298, + "eval_steps_per_second": 0.979, + "step": 1110 + }, + { + "epoch": 1.1412429378531073, + "grad_norm": 2.7025723457336426, + "learning_rate": 3.46973734297678e-06, + "loss": 1.5233, + "step": 1111 + }, + { + "epoch": 1.1422701592193119, + "grad_norm": 2.6984663009643555, + "learning_rate": 3.4678340312143133e-06, + "loss": 1.5447, + "step": 1112 + }, + { + "epoch": 1.1432973805855162, + "grad_norm": 2.708031415939331, + "learning_rate": 3.4659307194518466e-06, + "loss": 1.5905, + "step": 1113 + }, + { + "epoch": 1.1443246019517206, + "grad_norm": 2.5180366039276123, + "learning_rate": 3.46402740768938e-06, + "loss": 1.5636, + "step": 1114 + }, + { + "epoch": 1.145351823317925, + "grad_norm": 2.540074348449707, + "learning_rate": 3.462124095926913e-06, + "loss": 1.5045, + "step": 1115 + }, + { + "epoch": 1.1463790446841293, + "grad_norm": 2.4670231342315674, + "learning_rate": 3.4602207841644463e-06, + "loss": 1.4969, + "step": 1116 + }, + { + "epoch": 1.147406266050334, + "grad_norm": 2.7803049087524414, + "learning_rate": 3.4583174724019797e-06, + "loss": 1.5471, + "step": 1117 + }, + { + "epoch": 1.1484334874165383, + "grad_norm": 2.58117413520813, + "learning_rate": 3.456414160639513e-06, + "loss": 1.592, + "step": 1118 + }, + { + "epoch": 1.1494607087827426, + "grad_norm": 2.5810937881469727, + "learning_rate": 3.4545108488770464e-06, + "loss": 1.5542, + "step": 1119 + }, + { + "epoch": 1.150487930148947, + "grad_norm": 2.757809638977051, + "learning_rate": 3.4526075371145793e-06, + "loss": 1.5403, + "step": 1120 + }, + { + "epoch": 1.150487930148947, + "eval_loss": 1.905651569366455, + "eval_runtime": 45.619, + "eval_samples_per_second": 21.921, + "eval_steps_per_second": 0.921, + "step": 1120 + }, + { + "epoch": 1.1515151515151516, + "grad_norm": 2.6740472316741943, + "learning_rate": 3.4507042253521127e-06, + "loss": 1.5422, + "step": 1121 + }, + { + "epoch": 1.152542372881356, + "grad_norm": 2.7041897773742676, + "learning_rate": 3.448800913589646e-06, + "loss": 1.6049, + "step": 1122 + }, + { + "epoch": 1.1535695942475603, + "grad_norm": 2.7759902477264404, + "learning_rate": 3.4468976018271794e-06, + "loss": 1.522, + "step": 1123 + }, + { + "epoch": 1.1545968156137647, + "grad_norm": 2.69626522064209, + "learning_rate": 3.4449942900647128e-06, + "loss": 1.6096, + "step": 1124 + }, + { + "epoch": 1.1556240369799693, + "grad_norm": 2.57637619972229, + "learning_rate": 3.443090978302246e-06, + "loss": 1.5702, + "step": 1125 + }, + { + "epoch": 1.1566512583461737, + "grad_norm": 2.6259238719940186, + "learning_rate": 3.44118766653978e-06, + "loss": 1.5561, + "step": 1126 + }, + { + "epoch": 1.157678479712378, + "grad_norm": 2.7071187496185303, + "learning_rate": 3.439284354777313e-06, + "loss": 1.5545, + "step": 1127 + }, + { + "epoch": 1.1587057010785824, + "grad_norm": 2.7928988933563232, + "learning_rate": 3.4373810430148462e-06, + "loss": 1.5551, + "step": 1128 + }, + { + "epoch": 1.1597329224447868, + "grad_norm": 2.703719139099121, + "learning_rate": 3.4354777312523796e-06, + "loss": 1.4947, + "step": 1129 + }, + { + "epoch": 1.1607601438109914, + "grad_norm": 2.6363744735717773, + "learning_rate": 3.433574419489913e-06, + "loss": 1.511, + "step": 1130 + }, + { + "epoch": 1.1607601438109914, + "eval_loss": 1.9003485441207886, + "eval_runtime": 46.8855, + "eval_samples_per_second": 21.329, + "eval_steps_per_second": 0.896, + "step": 1130 + }, + { + "epoch": 1.1617873651771957, + "grad_norm": 2.7165591716766357, + "learning_rate": 3.4316711077274463e-06, + "loss": 1.5529, + "step": 1131 + }, + { + "epoch": 1.1628145865434, + "grad_norm": 2.6615865230560303, + "learning_rate": 3.4297677959649792e-06, + "loss": 1.5213, + "step": 1132 + }, + { + "epoch": 1.1638418079096045, + "grad_norm": 2.6274473667144775, + "learning_rate": 3.4278644842025126e-06, + "loss": 1.5813, + "step": 1133 + }, + { + "epoch": 1.1648690292758088, + "grad_norm": 2.637418031692505, + "learning_rate": 3.425961172440046e-06, + "loss": 1.4963, + "step": 1134 + }, + { + "epoch": 1.1658962506420134, + "grad_norm": 2.601574182510376, + "learning_rate": 3.4240578606775793e-06, + "loss": 1.6146, + "step": 1135 + }, + { + "epoch": 1.1669234720082178, + "grad_norm": 2.599043369293213, + "learning_rate": 3.4221545489151127e-06, + "loss": 1.5251, + "step": 1136 + }, + { + "epoch": 1.1679506933744221, + "grad_norm": 2.8078091144561768, + "learning_rate": 3.4202512371526456e-06, + "loss": 1.5231, + "step": 1137 + }, + { + "epoch": 1.1689779147406265, + "grad_norm": 2.811053514480591, + "learning_rate": 3.418347925390179e-06, + "loss": 1.6289, + "step": 1138 + }, + { + "epoch": 1.170005136106831, + "grad_norm": 2.7277560234069824, + "learning_rate": 3.4164446136277124e-06, + "loss": 1.5733, + "step": 1139 + }, + { + "epoch": 1.1710323574730355, + "grad_norm": 2.587024211883545, + "learning_rate": 3.4145413018652457e-06, + "loss": 1.5332, + "step": 1140 + }, + { + "epoch": 1.1710323574730355, + "eval_loss": 1.8978687524795532, + "eval_runtime": 49.9935, + "eval_samples_per_second": 20.003, + "eval_steps_per_second": 0.84, + "step": 1140 + }, + { + "epoch": 1.1720595788392398, + "grad_norm": 2.536479949951172, + "learning_rate": 3.412637990102779e-06, + "loss": 1.5601, + "step": 1141 + }, + { + "epoch": 1.1730868002054442, + "grad_norm": 2.7315709590911865, + "learning_rate": 3.4107346783403125e-06, + "loss": 1.5783, + "step": 1142 + }, + { + "epoch": 1.1741140215716488, + "grad_norm": 2.507249593734741, + "learning_rate": 3.4088313665778454e-06, + "loss": 1.5, + "step": 1143 + }, + { + "epoch": 1.1751412429378532, + "grad_norm": 2.625411033630371, + "learning_rate": 3.4069280548153788e-06, + "loss": 1.5328, + "step": 1144 + }, + { + "epoch": 1.1761684643040575, + "grad_norm": 2.591282367706299, + "learning_rate": 3.405024743052912e-06, + "loss": 1.5119, + "step": 1145 + }, + { + "epoch": 1.177195685670262, + "grad_norm": 2.547508955001831, + "learning_rate": 3.403121431290446e-06, + "loss": 1.5361, + "step": 1146 + }, + { + "epoch": 1.1782229070364663, + "grad_norm": 2.797992706298828, + "learning_rate": 3.4012181195279793e-06, + "loss": 1.5221, + "step": 1147 + }, + { + "epoch": 1.1792501284026708, + "grad_norm": 2.905266761779785, + "learning_rate": 3.3993148077655126e-06, + "loss": 1.5209, + "step": 1148 + }, + { + "epoch": 1.1802773497688752, + "grad_norm": 2.9259016513824463, + "learning_rate": 3.3974114960030456e-06, + "loss": 1.5365, + "step": 1149 + }, + { + "epoch": 1.1813045711350796, + "grad_norm": 2.597102165222168, + "learning_rate": 3.395508184240579e-06, + "loss": 1.4663, + "step": 1150 + }, + { + "epoch": 1.1813045711350796, + "eval_loss": 1.8966383934020996, + "eval_runtime": 47.3579, + "eval_samples_per_second": 21.116, + "eval_steps_per_second": 0.887, + "step": 1150 + }, + { + "epoch": 1.182331792501284, + "grad_norm": 2.6672847270965576, + "learning_rate": 3.3936048724781123e-06, + "loss": 1.5115, + "step": 1151 + }, + { + "epoch": 1.1833590138674885, + "grad_norm": 2.870549201965332, + "learning_rate": 3.3917015607156457e-06, + "loss": 1.6097, + "step": 1152 + }, + { + "epoch": 1.184386235233693, + "grad_norm": 2.6559388637542725, + "learning_rate": 3.389798248953179e-06, + "loss": 1.4777, + "step": 1153 + }, + { + "epoch": 1.1854134565998973, + "grad_norm": 2.7968215942382812, + "learning_rate": 3.387894937190712e-06, + "loss": 1.509, + "step": 1154 + }, + { + "epoch": 1.1864406779661016, + "grad_norm": 2.689741611480713, + "learning_rate": 3.3859916254282453e-06, + "loss": 1.5424, + "step": 1155 + }, + { + "epoch": 1.1874678993323062, + "grad_norm": 2.872887134552002, + "learning_rate": 3.3840883136657787e-06, + "loss": 1.5404, + "step": 1156 + }, + { + "epoch": 1.1884951206985106, + "grad_norm": 2.6436610221862793, + "learning_rate": 3.382185001903312e-06, + "loss": 1.5889, + "step": 1157 + }, + { + "epoch": 1.189522342064715, + "grad_norm": 2.697906255722046, + "learning_rate": 3.3802816901408454e-06, + "loss": 1.5086, + "step": 1158 + }, + { + "epoch": 1.1905495634309193, + "grad_norm": 2.8897645473480225, + "learning_rate": 3.3783783783783788e-06, + "loss": 1.5441, + "step": 1159 + }, + { + "epoch": 1.1915767847971237, + "grad_norm": 2.722351312637329, + "learning_rate": 3.3764750666159117e-06, + "loss": 1.527, + "step": 1160 + }, + { + "epoch": 1.1915767847971237, + "eval_loss": 1.8928718566894531, + "eval_runtime": 48.9427, + "eval_samples_per_second": 20.432, + "eval_steps_per_second": 0.858, + "step": 1160 + }, + { + "epoch": 1.1926040061633283, + "grad_norm": 2.5705020427703857, + "learning_rate": 3.374571754853445e-06, + "loss": 1.5304, + "step": 1161 + }, + { + "epoch": 1.1936312275295327, + "grad_norm": 2.571810245513916, + "learning_rate": 3.3726684430909784e-06, + "loss": 1.5242, + "step": 1162 + }, + { + "epoch": 1.194658448895737, + "grad_norm": 2.552725076675415, + "learning_rate": 3.370765131328512e-06, + "loss": 1.4803, + "step": 1163 + }, + { + "epoch": 1.1956856702619414, + "grad_norm": 2.734027862548828, + "learning_rate": 3.368861819566045e-06, + "loss": 1.5041, + "step": 1164 + }, + { + "epoch": 1.1967128916281458, + "grad_norm": 2.594614028930664, + "learning_rate": 3.366958507803578e-06, + "loss": 1.5205, + "step": 1165 + }, + { + "epoch": 1.1977401129943503, + "grad_norm": 2.897482395172119, + "learning_rate": 3.365055196041112e-06, + "loss": 1.5372, + "step": 1166 + }, + { + "epoch": 1.1987673343605547, + "grad_norm": 2.6143221855163574, + "learning_rate": 3.3631518842786452e-06, + "loss": 1.4966, + "step": 1167 + }, + { + "epoch": 1.199794555726759, + "grad_norm": 2.845475912094116, + "learning_rate": 3.3612485725161786e-06, + "loss": 1.5759, + "step": 1168 + }, + { + "epoch": 1.2008217770929634, + "grad_norm": 2.667388916015625, + "learning_rate": 3.359345260753712e-06, + "loss": 1.5014, + "step": 1169 + }, + { + "epoch": 1.201848998459168, + "grad_norm": 2.890197277069092, + "learning_rate": 3.3574419489912453e-06, + "loss": 1.5939, + "step": 1170 + }, + { + "epoch": 1.201848998459168, + "eval_loss": 1.8978379964828491, + "eval_runtime": 48.9839, + "eval_samples_per_second": 20.415, + "eval_steps_per_second": 0.857, + "step": 1170 + }, + { + "epoch": 1.2028762198253724, + "grad_norm": 2.578139066696167, + "learning_rate": 3.3555386372287783e-06, + "loss": 1.5655, + "step": 1171 + }, + { + "epoch": 1.2039034411915768, + "grad_norm": 2.5672738552093506, + "learning_rate": 3.3536353254663116e-06, + "loss": 1.5998, + "step": 1172 + }, + { + "epoch": 1.2049306625577811, + "grad_norm": 2.7297089099884033, + "learning_rate": 3.351732013703845e-06, + "loss": 1.4938, + "step": 1173 + }, + { + "epoch": 1.2059578839239857, + "grad_norm": 2.7455339431762695, + "learning_rate": 3.3498287019413784e-06, + "loss": 1.5505, + "step": 1174 + }, + { + "epoch": 1.20698510529019, + "grad_norm": 2.725179433822632, + "learning_rate": 3.3479253901789117e-06, + "loss": 1.5524, + "step": 1175 + }, + { + "epoch": 1.2080123266563945, + "grad_norm": 2.6953988075256348, + "learning_rate": 3.346022078416445e-06, + "loss": 1.568, + "step": 1176 + }, + { + "epoch": 1.2090395480225988, + "grad_norm": 2.613600492477417, + "learning_rate": 3.344118766653978e-06, + "loss": 1.4608, + "step": 1177 + }, + { + "epoch": 1.2100667693888032, + "grad_norm": 2.7339162826538086, + "learning_rate": 3.3422154548915114e-06, + "loss": 1.5426, + "step": 1178 + }, + { + "epoch": 1.2110939907550078, + "grad_norm": 2.774726152420044, + "learning_rate": 3.3403121431290448e-06, + "loss": 1.5192, + "step": 1179 + }, + { + "epoch": 1.2121212121212122, + "grad_norm": 2.7361602783203125, + "learning_rate": 3.338408831366578e-06, + "loss": 1.5481, + "step": 1180 + }, + { + "epoch": 1.2121212121212122, + "eval_loss": 1.902246117591858, + "eval_runtime": 45.153, + "eval_samples_per_second": 22.147, + "eval_steps_per_second": 0.93, + "step": 1180 + }, + { + "epoch": 1.2131484334874165, + "grad_norm": 2.6936450004577637, + "learning_rate": 3.3365055196041115e-06, + "loss": 1.4821, + "step": 1181 + }, + { + "epoch": 1.214175654853621, + "grad_norm": 2.881898880004883, + "learning_rate": 3.3346022078416444e-06, + "loss": 1.5663, + "step": 1182 + }, + { + "epoch": 1.2152028762198253, + "grad_norm": 2.793069362640381, + "learning_rate": 3.3326988960791778e-06, + "loss": 1.5998, + "step": 1183 + }, + { + "epoch": 1.2162300975860298, + "grad_norm": 2.591367721557617, + "learning_rate": 3.330795584316711e-06, + "loss": 1.5199, + "step": 1184 + }, + { + "epoch": 1.2172573189522342, + "grad_norm": 2.8402175903320312, + "learning_rate": 3.3288922725542445e-06, + "loss": 1.451, + "step": 1185 + }, + { + "epoch": 1.2182845403184386, + "grad_norm": 2.737199068069458, + "learning_rate": 3.326988960791778e-06, + "loss": 1.5117, + "step": 1186 + }, + { + "epoch": 1.219311761684643, + "grad_norm": 2.915210723876953, + "learning_rate": 3.3250856490293117e-06, + "loss": 1.5388, + "step": 1187 + }, + { + "epoch": 1.2203389830508475, + "grad_norm": 3.091784715652466, + "learning_rate": 3.323182337266845e-06, + "loss": 1.5944, + "step": 1188 + }, + { + "epoch": 1.221366204417052, + "grad_norm": 2.770756483078003, + "learning_rate": 3.321279025504378e-06, + "loss": 1.5619, + "step": 1189 + }, + { + "epoch": 1.2223934257832563, + "grad_norm": 2.7989954948425293, + "learning_rate": 3.3193757137419113e-06, + "loss": 1.5581, + "step": 1190 + }, + { + "epoch": 1.2223934257832563, + "eval_loss": 1.907904863357544, + "eval_runtime": 46.8087, + "eval_samples_per_second": 21.364, + "eval_steps_per_second": 0.897, + "step": 1190 + }, + { + "epoch": 1.2234206471494606, + "grad_norm": 2.6740641593933105, + "learning_rate": 3.3174724019794447e-06, + "loss": 1.5556, + "step": 1191 + }, + { + "epoch": 1.2244478685156652, + "grad_norm": 2.760608196258545, + "learning_rate": 3.315569090216978e-06, + "loss": 1.5903, + "step": 1192 + }, + { + "epoch": 1.2254750898818696, + "grad_norm": 2.942370891571045, + "learning_rate": 3.3136657784545114e-06, + "loss": 1.5052, + "step": 1193 + }, + { + "epoch": 1.226502311248074, + "grad_norm": 2.8096179962158203, + "learning_rate": 3.3117624666920443e-06, + "loss": 1.5933, + "step": 1194 + }, + { + "epoch": 1.2275295326142783, + "grad_norm": 2.7357091903686523, + "learning_rate": 3.3098591549295777e-06, + "loss": 1.5074, + "step": 1195 + }, + { + "epoch": 1.2285567539804827, + "grad_norm": 2.728134870529175, + "learning_rate": 3.307955843167111e-06, + "loss": 1.48, + "step": 1196 + }, + { + "epoch": 1.2295839753466873, + "grad_norm": 2.921462059020996, + "learning_rate": 3.3060525314046444e-06, + "loss": 1.5602, + "step": 1197 + }, + { + "epoch": 1.2306111967128917, + "grad_norm": 2.7563631534576416, + "learning_rate": 3.304149219642178e-06, + "loss": 1.5309, + "step": 1198 + }, + { + "epoch": 1.231638418079096, + "grad_norm": 2.533001661300659, + "learning_rate": 3.3022459078797107e-06, + "loss": 1.5242, + "step": 1199 + }, + { + "epoch": 1.2326656394453004, + "grad_norm": 2.5509681701660156, + "learning_rate": 3.300342596117244e-06, + "loss": 1.5488, + "step": 1200 + }, + { + "epoch": 1.2326656394453004, + "eval_loss": 1.8983508348464966, + "eval_runtime": 44.8654, + "eval_samples_per_second": 22.289, + "eval_steps_per_second": 0.936, + "step": 1200 + }, + { + "epoch": 1.2336928608115048, + "grad_norm": 2.688228130340576, + "learning_rate": 3.2984392843547775e-06, + "loss": 1.5344, + "step": 1201 + }, + { + "epoch": 1.2347200821777093, + "grad_norm": 2.567621946334839, + "learning_rate": 3.296535972592311e-06, + "loss": 1.4926, + "step": 1202 + }, + { + "epoch": 1.2357473035439137, + "grad_norm": 2.586637258529663, + "learning_rate": 3.294632660829844e-06, + "loss": 1.4749, + "step": 1203 + }, + { + "epoch": 1.236774524910118, + "grad_norm": 2.697901725769043, + "learning_rate": 3.292729349067377e-06, + "loss": 1.5073, + "step": 1204 + }, + { + "epoch": 1.2378017462763227, + "grad_norm": 2.832059621810913, + "learning_rate": 3.2908260373049105e-06, + "loss": 1.5147, + "step": 1205 + }, + { + "epoch": 1.238828967642527, + "grad_norm": 2.7802395820617676, + "learning_rate": 3.288922725542444e-06, + "loss": 1.5031, + "step": 1206 + }, + { + "epoch": 1.2398561890087314, + "grad_norm": 2.8382885456085205, + "learning_rate": 3.2870194137799776e-06, + "loss": 1.5539, + "step": 1207 + }, + { + "epoch": 1.2408834103749358, + "grad_norm": 2.7725186347961426, + "learning_rate": 3.285116102017511e-06, + "loss": 1.5145, + "step": 1208 + }, + { + "epoch": 1.2419106317411401, + "grad_norm": 2.563866376876831, + "learning_rate": 3.2832127902550444e-06, + "loss": 1.4852, + "step": 1209 + }, + { + "epoch": 1.2429378531073447, + "grad_norm": 2.7052412033081055, + "learning_rate": 3.2813094784925777e-06, + "loss": 1.5615, + "step": 1210 + }, + { + "epoch": 1.2429378531073447, + "eval_loss": 1.9071296453475952, + "eval_runtime": 42.2147, + "eval_samples_per_second": 23.688, + "eval_steps_per_second": 0.995, + "step": 1210 + }, + { + "epoch": 1.243965074473549, + "grad_norm": 2.7794888019561768, + "learning_rate": 3.2794061667301107e-06, + "loss": 1.5445, + "step": 1211 + }, + { + "epoch": 1.2449922958397535, + "grad_norm": 2.6376712322235107, + "learning_rate": 3.277502854967644e-06, + "loss": 1.4728, + "step": 1212 + }, + { + "epoch": 1.2460195172059578, + "grad_norm": 2.5500779151916504, + "learning_rate": 3.2755995432051774e-06, + "loss": 1.5133, + "step": 1213 + }, + { + "epoch": 1.2470467385721622, + "grad_norm": 2.629063606262207, + "learning_rate": 3.2736962314427107e-06, + "loss": 1.5093, + "step": 1214 + }, + { + "epoch": 1.2480739599383668, + "grad_norm": 2.662497043609619, + "learning_rate": 3.271792919680244e-06, + "loss": 1.5237, + "step": 1215 + }, + { + "epoch": 1.2491011813045712, + "grad_norm": 2.5375289916992188, + "learning_rate": 3.269889607917777e-06, + "loss": 1.5211, + "step": 1216 + }, + { + "epoch": 1.2501284026707755, + "grad_norm": 2.8396782875061035, + "learning_rate": 3.2679862961553104e-06, + "loss": 1.5773, + "step": 1217 + }, + { + "epoch": 1.25115562403698, + "grad_norm": 2.8443665504455566, + "learning_rate": 3.2660829843928438e-06, + "loss": 1.4987, + "step": 1218 + }, + { + "epoch": 1.2521828454031843, + "grad_norm": 2.77935528755188, + "learning_rate": 3.264179672630377e-06, + "loss": 1.5344, + "step": 1219 + }, + { + "epoch": 1.2532100667693888, + "grad_norm": 3.0581092834472656, + "learning_rate": 3.2622763608679105e-06, + "loss": 1.5728, + "step": 1220 + }, + { + "epoch": 1.2532100667693888, + "eval_loss": 1.9198752641677856, + "eval_runtime": 45.1116, + "eval_samples_per_second": 22.167, + "eval_steps_per_second": 0.931, + "step": 1220 + }, + { + "epoch": 1.2542372881355932, + "grad_norm": 2.8741002082824707, + "learning_rate": 3.2603730491054434e-06, + "loss": 1.5444, + "step": 1221 + }, + { + "epoch": 1.2552645095017976, + "grad_norm": 2.6263391971588135, + "learning_rate": 3.258469737342977e-06, + "loss": 1.4997, + "step": 1222 + }, + { + "epoch": 1.2562917308680022, + "grad_norm": 2.949795961380005, + "learning_rate": 3.25656642558051e-06, + "loss": 1.5145, + "step": 1223 + }, + { + "epoch": 1.2573189522342065, + "grad_norm": 2.846130609512329, + "learning_rate": 3.2546631138180435e-06, + "loss": 1.4709, + "step": 1224 + }, + { + "epoch": 1.258346173600411, + "grad_norm": 2.6340930461883545, + "learning_rate": 3.252759802055577e-06, + "loss": 1.5087, + "step": 1225 + }, + { + "epoch": 1.2593733949666153, + "grad_norm": 2.9878976345062256, + "learning_rate": 3.25085649029311e-06, + "loss": 1.5142, + "step": 1226 + }, + { + "epoch": 1.2604006163328196, + "grad_norm": 2.9880733489990234, + "learning_rate": 3.248953178530644e-06, + "loss": 1.521, + "step": 1227 + }, + { + "epoch": 1.2614278376990242, + "grad_norm": 2.8507912158966064, + "learning_rate": 3.247049866768177e-06, + "loss": 1.5482, + "step": 1228 + }, + { + "epoch": 1.2624550590652286, + "grad_norm": 2.870678424835205, + "learning_rate": 3.2451465550057103e-06, + "loss": 1.4965, + "step": 1229 + }, + { + "epoch": 1.263482280431433, + "grad_norm": 2.644347906112671, + "learning_rate": 3.2432432432432437e-06, + "loss": 1.4614, + "step": 1230 + }, + { + "epoch": 1.263482280431433, + "eval_loss": 1.9100512266159058, + "eval_runtime": 44.6472, + "eval_samples_per_second": 22.398, + "eval_steps_per_second": 0.941, + "step": 1230 + }, + { + "epoch": 1.2645095017976373, + "grad_norm": 2.5852861404418945, + "learning_rate": 3.241339931480777e-06, + "loss": 1.5122, + "step": 1231 + }, + { + "epoch": 1.2655367231638417, + "grad_norm": 2.9572696685791016, + "learning_rate": 3.2394366197183104e-06, + "loss": 1.5529, + "step": 1232 + }, + { + "epoch": 1.2665639445300463, + "grad_norm": 2.7040176391601562, + "learning_rate": 3.2375333079558434e-06, + "loss": 1.5418, + "step": 1233 + }, + { + "epoch": 1.2675911658962506, + "grad_norm": 2.577706813812256, + "learning_rate": 3.2356299961933767e-06, + "loss": 1.5167, + "step": 1234 + }, + { + "epoch": 1.268618387262455, + "grad_norm": 2.6559629440307617, + "learning_rate": 3.23372668443091e-06, + "loss": 1.5278, + "step": 1235 + }, + { + "epoch": 1.2696456086286596, + "grad_norm": 2.673293113708496, + "learning_rate": 3.2318233726684435e-06, + "loss": 1.5393, + "step": 1236 + }, + { + "epoch": 1.270672829994864, + "grad_norm": 2.507797956466675, + "learning_rate": 3.229920060905977e-06, + "loss": 1.5573, + "step": 1237 + }, + { + "epoch": 1.2717000513610683, + "grad_norm": 2.8361523151397705, + "learning_rate": 3.2280167491435098e-06, + "loss": 1.4911, + "step": 1238 + }, + { + "epoch": 1.2727272727272727, + "grad_norm": 2.7824695110321045, + "learning_rate": 3.226113437381043e-06, + "loss": 1.4949, + "step": 1239 + }, + { + "epoch": 1.273754494093477, + "grad_norm": 2.4863975048065186, + "learning_rate": 3.2242101256185765e-06, + "loss": 1.5017, + "step": 1240 + }, + { + "epoch": 1.273754494093477, + "eval_loss": 1.896610140800476, + "eval_runtime": 46.9004, + "eval_samples_per_second": 21.322, + "eval_steps_per_second": 0.896, + "step": 1240 + }, + { + "epoch": 1.2747817154596817, + "grad_norm": 2.767557144165039, + "learning_rate": 3.22230681385611e-06, + "loss": 1.5137, + "step": 1241 + }, + { + "epoch": 1.275808936825886, + "grad_norm": 2.544274091720581, + "learning_rate": 3.220403502093643e-06, + "loss": 1.5759, + "step": 1242 + }, + { + "epoch": 1.2768361581920904, + "grad_norm": 2.6679677963256836, + "learning_rate": 3.218500190331176e-06, + "loss": 1.533, + "step": 1243 + }, + { + "epoch": 1.2778633795582948, + "grad_norm": 2.634310245513916, + "learning_rate": 3.2165968785687095e-06, + "loss": 1.5605, + "step": 1244 + }, + { + "epoch": 1.2788906009244991, + "grad_norm": 2.6534063816070557, + "learning_rate": 3.214693566806243e-06, + "loss": 1.5129, + "step": 1245 + }, + { + "epoch": 1.2799178222907037, + "grad_norm": 2.6762259006500244, + "learning_rate": 3.2127902550437762e-06, + "loss": 1.5435, + "step": 1246 + }, + { + "epoch": 1.280945043656908, + "grad_norm": 2.70534610748291, + "learning_rate": 3.21088694328131e-06, + "loss": 1.5215, + "step": 1247 + }, + { + "epoch": 1.2819722650231125, + "grad_norm": 2.929002046585083, + "learning_rate": 3.2089836315188434e-06, + "loss": 1.5237, + "step": 1248 + }, + { + "epoch": 1.2829994863893168, + "grad_norm": 2.6163816452026367, + "learning_rate": 3.2070803197563767e-06, + "loss": 1.5114, + "step": 1249 + }, + { + "epoch": 1.2840267077555212, + "grad_norm": 2.6484017372131348, + "learning_rate": 3.2051770079939097e-06, + "loss": 1.5054, + "step": 1250 + }, + { + "epoch": 1.2840267077555212, + "eval_loss": 1.9057985544204712, + "eval_runtime": 45.9877, + "eval_samples_per_second": 21.745, + "eval_steps_per_second": 0.913, + "step": 1250 + }, + { + "epoch": 1.2850539291217258, + "grad_norm": 2.694584608078003, + "learning_rate": 3.203273696231443e-06, + "loss": 1.5272, + "step": 1251 + }, + { + "epoch": 1.2860811504879301, + "grad_norm": 2.699054718017578, + "learning_rate": 3.2013703844689764e-06, + "loss": 1.5632, + "step": 1252 + }, + { + "epoch": 1.2871083718541345, + "grad_norm": 2.765612840652466, + "learning_rate": 3.1994670727065098e-06, + "loss": 1.4927, + "step": 1253 + }, + { + "epoch": 1.288135593220339, + "grad_norm": 2.7850136756896973, + "learning_rate": 3.197563760944043e-06, + "loss": 1.5412, + "step": 1254 + }, + { + "epoch": 1.2891628145865435, + "grad_norm": 2.6565496921539307, + "learning_rate": 3.195660449181576e-06, + "loss": 1.5606, + "step": 1255 + }, + { + "epoch": 1.2901900359527478, + "grad_norm": 2.627163887023926, + "learning_rate": 3.1937571374191094e-06, + "loss": 1.5612, + "step": 1256 + }, + { + "epoch": 1.2912172573189522, + "grad_norm": 2.5778250694274902, + "learning_rate": 3.191853825656643e-06, + "loss": 1.4988, + "step": 1257 + }, + { + "epoch": 1.2922444786851566, + "grad_norm": 2.7875125408172607, + "learning_rate": 3.189950513894176e-06, + "loss": 1.5237, + "step": 1258 + }, + { + "epoch": 1.2932717000513612, + "grad_norm": 2.645707607269287, + "learning_rate": 3.1880472021317095e-06, + "loss": 1.5157, + "step": 1259 + }, + { + "epoch": 1.2942989214175655, + "grad_norm": 2.6457326412200928, + "learning_rate": 3.1861438903692425e-06, + "loss": 1.5109, + "step": 1260 + }, + { + "epoch": 1.2942989214175655, + "eval_loss": 1.896759033203125, + "eval_runtime": 50.1276, + "eval_samples_per_second": 19.949, + "eval_steps_per_second": 0.838, + "step": 1260 + }, + { + "epoch": 1.29532614278377, + "grad_norm": 2.617196798324585, + "learning_rate": 3.184240578606776e-06, + "loss": 1.4646, + "step": 1261 + }, + { + "epoch": 1.2963533641499743, + "grad_norm": 2.618905782699585, + "learning_rate": 3.182337266844309e-06, + "loss": 1.5012, + "step": 1262 + }, + { + "epoch": 1.2973805855161786, + "grad_norm": 2.6222894191741943, + "learning_rate": 3.1804339550818426e-06, + "loss": 1.526, + "step": 1263 + }, + { + "epoch": 1.2984078068823832, + "grad_norm": 2.858421564102173, + "learning_rate": 3.178530643319376e-06, + "loss": 1.5621, + "step": 1264 + }, + { + "epoch": 1.2994350282485876, + "grad_norm": 2.579683303833008, + "learning_rate": 3.176627331556909e-06, + "loss": 1.5217, + "step": 1265 + }, + { + "epoch": 1.300462249614792, + "grad_norm": 2.721665859222412, + "learning_rate": 3.1747240197944422e-06, + "loss": 1.4732, + "step": 1266 + }, + { + "epoch": 1.3014894709809965, + "grad_norm": 2.6431221961975098, + "learning_rate": 3.172820708031976e-06, + "loss": 1.5219, + "step": 1267 + }, + { + "epoch": 1.3025166923472007, + "grad_norm": 2.706671714782715, + "learning_rate": 3.1709173962695094e-06, + "loss": 1.4829, + "step": 1268 + }, + { + "epoch": 1.3035439137134053, + "grad_norm": 2.81333589553833, + "learning_rate": 3.1690140845070427e-06, + "loss": 1.5669, + "step": 1269 + }, + { + "epoch": 1.3045711350796096, + "grad_norm": 2.942777633666992, + "learning_rate": 3.167110772744576e-06, + "loss": 1.5477, + "step": 1270 + }, + { + "epoch": 1.3045711350796096, + "eval_loss": 1.9051777124404907, + "eval_runtime": 48.2881, + "eval_samples_per_second": 20.709, + "eval_steps_per_second": 0.87, + "step": 1270 + }, + { + "epoch": 1.305598356445814, + "grad_norm": 2.721989393234253, + "learning_rate": 3.1652074609821095e-06, + "loss": 1.5269, + "step": 1271 + }, + { + "epoch": 1.3066255778120186, + "grad_norm": 2.8142035007476807, + "learning_rate": 3.1633041492196424e-06, + "loss": 1.5193, + "step": 1272 + }, + { + "epoch": 1.307652799178223, + "grad_norm": 2.796630382537842, + "learning_rate": 3.1614008374571758e-06, + "loss": 1.503, + "step": 1273 + }, + { + "epoch": 1.3086800205444273, + "grad_norm": 2.696720600128174, + "learning_rate": 3.159497525694709e-06, + "loss": 1.5301, + "step": 1274 + }, + { + "epoch": 1.3097072419106317, + "grad_norm": 2.6286635398864746, + "learning_rate": 3.1575942139322425e-06, + "loss": 1.5326, + "step": 1275 + }, + { + "epoch": 1.310734463276836, + "grad_norm": 2.662801504135132, + "learning_rate": 3.155690902169776e-06, + "loss": 1.5048, + "step": 1276 + }, + { + "epoch": 1.3117616846430407, + "grad_norm": 2.838228702545166, + "learning_rate": 3.1537875904073088e-06, + "loss": 1.5321, + "step": 1277 + }, + { + "epoch": 1.312788906009245, + "grad_norm": 2.7021734714508057, + "learning_rate": 3.151884278644842e-06, + "loss": 1.4917, + "step": 1278 + }, + { + "epoch": 1.3138161273754494, + "grad_norm": 2.696664571762085, + "learning_rate": 3.1499809668823755e-06, + "loss": 1.5231, + "step": 1279 + }, + { + "epoch": 1.3148433487416538, + "grad_norm": 2.720292091369629, + "learning_rate": 3.148077655119909e-06, + "loss": 1.497, + "step": 1280 + }, + { + "epoch": 1.3148433487416538, + "eval_loss": 1.8934897184371948, + "eval_runtime": 48.5119, + "eval_samples_per_second": 20.613, + "eval_steps_per_second": 0.866, + "step": 1280 + }, + { + "epoch": 1.3158705701078581, + "grad_norm": 2.6228737831115723, + "learning_rate": 3.1461743433574422e-06, + "loss": 1.5475, + "step": 1281 + }, + { + "epoch": 1.3168977914740627, + "grad_norm": 2.733198881149292, + "learning_rate": 3.144271031594975e-06, + "loss": 1.4909, + "step": 1282 + }, + { + "epoch": 1.317925012840267, + "grad_norm": 2.5863518714904785, + "learning_rate": 3.1423677198325085e-06, + "loss": 1.5446, + "step": 1283 + }, + { + "epoch": 1.3189522342064715, + "grad_norm": 2.775151491165161, + "learning_rate": 3.140464408070042e-06, + "loss": 1.5414, + "step": 1284 + }, + { + "epoch": 1.319979455572676, + "grad_norm": 2.7392020225524902, + "learning_rate": 3.1385610963075753e-06, + "loss": 1.5094, + "step": 1285 + }, + { + "epoch": 1.3210066769388804, + "grad_norm": 2.578563928604126, + "learning_rate": 3.1366577845451086e-06, + "loss": 1.4973, + "step": 1286 + }, + { + "epoch": 1.3220338983050848, + "grad_norm": 2.664562225341797, + "learning_rate": 3.134754472782642e-06, + "loss": 1.4874, + "step": 1287 + }, + { + "epoch": 1.3230611196712891, + "grad_norm": 2.713029146194458, + "learning_rate": 3.1328511610201758e-06, + "loss": 1.479, + "step": 1288 + }, + { + "epoch": 1.3240883410374935, + "grad_norm": 2.569258213043213, + "learning_rate": 3.1309478492577087e-06, + "loss": 1.4641, + "step": 1289 + }, + { + "epoch": 1.325115562403698, + "grad_norm": 2.4881503582000732, + "learning_rate": 3.129044537495242e-06, + "loss": 1.4634, + "step": 1290 + }, + { + "epoch": 1.325115562403698, + "eval_loss": 1.8945759534835815, + "eval_runtime": 48.2935, + "eval_samples_per_second": 20.707, + "eval_steps_per_second": 0.87, + "step": 1290 + }, + { + "epoch": 1.3261427837699025, + "grad_norm": 2.485280752182007, + "learning_rate": 3.1271412257327754e-06, + "loss": 1.507, + "step": 1291 + }, + { + "epoch": 1.3271700051361068, + "grad_norm": 2.7823352813720703, + "learning_rate": 3.125237913970309e-06, + "loss": 1.5414, + "step": 1292 + }, + { + "epoch": 1.3281972265023112, + "grad_norm": 2.516641139984131, + "learning_rate": 3.123334602207842e-06, + "loss": 1.4889, + "step": 1293 + }, + { + "epoch": 1.3292244478685156, + "grad_norm": 2.7155838012695312, + "learning_rate": 3.121431290445375e-06, + "loss": 1.4701, + "step": 1294 + }, + { + "epoch": 1.3302516692347202, + "grad_norm": 2.5513222217559814, + "learning_rate": 3.1195279786829085e-06, + "loss": 1.5514, + "step": 1295 + }, + { + "epoch": 1.3312788906009245, + "grad_norm": 2.7775397300720215, + "learning_rate": 3.117624666920442e-06, + "loss": 1.611, + "step": 1296 + }, + { + "epoch": 1.332306111967129, + "grad_norm": 2.693214178085327, + "learning_rate": 3.115721355157975e-06, + "loss": 1.497, + "step": 1297 + }, + { + "epoch": 1.3333333333333333, + "grad_norm": 2.6052982807159424, + "learning_rate": 3.1138180433955085e-06, + "loss": 1.4911, + "step": 1298 + }, + { + "epoch": 1.3343605546995376, + "grad_norm": 2.578404426574707, + "learning_rate": 3.1119147316330415e-06, + "loss": 1.4497, + "step": 1299 + }, + { + "epoch": 1.3353877760657422, + "grad_norm": 2.588397264480591, + "learning_rate": 3.110011419870575e-06, + "loss": 1.5339, + "step": 1300 + }, + { + "epoch": 1.3353877760657422, + "eval_loss": 1.8876255750656128, + "eval_runtime": 48.2305, + "eval_samples_per_second": 20.734, + "eval_steps_per_second": 0.871, + "step": 1300 + }, + { + "epoch": 1.3364149974319466, + "grad_norm": 2.835510730743408, + "learning_rate": 3.1081081081081082e-06, + "loss": 1.5649, + "step": 1301 + }, + { + "epoch": 1.337442218798151, + "grad_norm": 2.6646556854248047, + "learning_rate": 3.1062047963456416e-06, + "loss": 1.4969, + "step": 1302 + }, + { + "epoch": 1.3384694401643555, + "grad_norm": 2.6960718631744385, + "learning_rate": 3.104301484583175e-06, + "loss": 1.5332, + "step": 1303 + }, + { + "epoch": 1.33949666153056, + "grad_norm": 2.8116049766540527, + "learning_rate": 3.1023981728207083e-06, + "loss": 1.5766, + "step": 1304 + }, + { + "epoch": 1.3405238828967643, + "grad_norm": 2.7282750606536865, + "learning_rate": 3.1004948610582412e-06, + "loss": 1.551, + "step": 1305 + }, + { + "epoch": 1.3415511042629686, + "grad_norm": 2.6903767585754395, + "learning_rate": 3.0985915492957746e-06, + "loss": 1.5053, + "step": 1306 + }, + { + "epoch": 1.342578325629173, + "grad_norm": 2.525789976119995, + "learning_rate": 3.096688237533308e-06, + "loss": 1.4969, + "step": 1307 + }, + { + "epoch": 1.3436055469953776, + "grad_norm": 2.7578322887420654, + "learning_rate": 3.0947849257708418e-06, + "loss": 1.4701, + "step": 1308 + }, + { + "epoch": 1.344632768361582, + "grad_norm": 2.714743137359619, + "learning_rate": 3.092881614008375e-06, + "loss": 1.5628, + "step": 1309 + }, + { + "epoch": 1.3456599897277863, + "grad_norm": 2.5726380348205566, + "learning_rate": 3.0909783022459085e-06, + "loss": 1.5146, + "step": 1310 + }, + { + "epoch": 1.3456599897277863, + "eval_loss": 1.8887813091278076, + "eval_runtime": 44.978, + "eval_samples_per_second": 22.233, + "eval_steps_per_second": 0.934, + "step": 1310 + }, + { + "epoch": 1.3466872110939907, + "grad_norm": 2.82814621925354, + "learning_rate": 3.0890749904834414e-06, + "loss": 1.5684, + "step": 1311 + }, + { + "epoch": 1.347714432460195, + "grad_norm": 2.705440044403076, + "learning_rate": 3.0871716787209748e-06, + "loss": 1.5003, + "step": 1312 + }, + { + "epoch": 1.3487416538263997, + "grad_norm": 2.647636890411377, + "learning_rate": 3.085268366958508e-06, + "loss": 1.5528, + "step": 1313 + }, + { + "epoch": 1.349768875192604, + "grad_norm": 2.954592227935791, + "learning_rate": 3.0833650551960415e-06, + "loss": 1.5274, + "step": 1314 + }, + { + "epoch": 1.3507960965588084, + "grad_norm": 2.984116315841675, + "learning_rate": 3.081461743433575e-06, + "loss": 1.5411, + "step": 1315 + }, + { + "epoch": 1.3518233179250128, + "grad_norm": 2.5633351802825928, + "learning_rate": 3.079558431671108e-06, + "loss": 1.4853, + "step": 1316 + }, + { + "epoch": 1.3528505392912171, + "grad_norm": 2.8512840270996094, + "learning_rate": 3.077655119908641e-06, + "loss": 1.5117, + "step": 1317 + }, + { + "epoch": 1.3538777606574217, + "grad_norm": 2.709902286529541, + "learning_rate": 3.0757518081461745e-06, + "loss": 1.4716, + "step": 1318 + }, + { + "epoch": 1.354904982023626, + "grad_norm": 2.6622109413146973, + "learning_rate": 3.073848496383708e-06, + "loss": 1.5098, + "step": 1319 + }, + { + "epoch": 1.3559322033898304, + "grad_norm": 2.681844472885132, + "learning_rate": 3.0719451846212413e-06, + "loss": 1.5245, + "step": 1320 + }, + { + "epoch": 1.3559322033898304, + "eval_loss": 1.8935153484344482, + "eval_runtime": 51.161, + "eval_samples_per_second": 19.546, + "eval_steps_per_second": 0.821, + "step": 1320 + }, + { + "epoch": 1.356959424756035, + "grad_norm": 2.816667318344116, + "learning_rate": 3.0700418728587746e-06, + "loss": 1.484, + "step": 1321 + }, + { + "epoch": 1.3579866461222394, + "grad_norm": 2.748138427734375, + "learning_rate": 3.0681385610963076e-06, + "loss": 1.5235, + "step": 1322 + }, + { + "epoch": 1.3590138674884438, + "grad_norm": 2.746454954147339, + "learning_rate": 3.066235249333841e-06, + "loss": 1.4929, + "step": 1323 + }, + { + "epoch": 1.3600410888546481, + "grad_norm": 2.8291871547698975, + "learning_rate": 3.0643319375713743e-06, + "loss": 1.5381, + "step": 1324 + }, + { + "epoch": 1.3610683102208525, + "grad_norm": 2.8472836017608643, + "learning_rate": 3.0624286258089076e-06, + "loss": 1.5664, + "step": 1325 + }, + { + "epoch": 1.362095531587057, + "grad_norm": 2.7493903636932373, + "learning_rate": 3.060525314046441e-06, + "loss": 1.5088, + "step": 1326 + }, + { + "epoch": 1.3631227529532615, + "grad_norm": 2.6563873291015625, + "learning_rate": 3.058622002283974e-06, + "loss": 1.4899, + "step": 1327 + }, + { + "epoch": 1.3641499743194658, + "grad_norm": 2.8158602714538574, + "learning_rate": 3.0567186905215077e-06, + "loss": 1.546, + "step": 1328 + }, + { + "epoch": 1.3651771956856702, + "grad_norm": 2.722679615020752, + "learning_rate": 3.054815378759041e-06, + "loss": 1.5572, + "step": 1329 + }, + { + "epoch": 1.3662044170518746, + "grad_norm": 2.572859525680542, + "learning_rate": 3.0529120669965745e-06, + "loss": 1.5441, + "step": 1330 + }, + { + "epoch": 1.3662044170518746, + "eval_loss": 1.8944634199142456, + "eval_runtime": 49.7122, + "eval_samples_per_second": 20.116, + "eval_steps_per_second": 0.845, + "step": 1330 + }, + { + "epoch": 1.3672316384180792, + "grad_norm": 2.6468751430511475, + "learning_rate": 3.051008755234108e-06, + "loss": 1.5507, + "step": 1331 + }, + { + "epoch": 1.3682588597842835, + "grad_norm": 2.618972063064575, + "learning_rate": 3.049105443471641e-06, + "loss": 1.4799, + "step": 1332 + }, + { + "epoch": 1.3692860811504879, + "grad_norm": 2.808328628540039, + "learning_rate": 3.0472021317091745e-06, + "loss": 1.4917, + "step": 1333 + }, + { + "epoch": 1.3703133025166925, + "grad_norm": 2.7831826210021973, + "learning_rate": 3.0452988199467075e-06, + "loss": 1.6242, + "step": 1334 + }, + { + "epoch": 1.3713405238828968, + "grad_norm": 2.6808860301971436, + "learning_rate": 3.043395508184241e-06, + "loss": 1.5455, + "step": 1335 + }, + { + "epoch": 1.3723677452491012, + "grad_norm": 2.5987675189971924, + "learning_rate": 3.041492196421774e-06, + "loss": 1.5406, + "step": 1336 + }, + { + "epoch": 1.3733949666153056, + "grad_norm": 2.6476593017578125, + "learning_rate": 3.0395888846593076e-06, + "loss": 1.487, + "step": 1337 + }, + { + "epoch": 1.37442218798151, + "grad_norm": 2.724715232849121, + "learning_rate": 3.037685572896841e-06, + "loss": 1.5316, + "step": 1338 + }, + { + "epoch": 1.3754494093477145, + "grad_norm": 2.6245994567871094, + "learning_rate": 3.035782261134374e-06, + "loss": 1.5137, + "step": 1339 + }, + { + "epoch": 1.376476630713919, + "grad_norm": 2.676206588745117, + "learning_rate": 3.0338789493719072e-06, + "loss": 1.5321, + "step": 1340 + }, + { + "epoch": 1.376476630713919, + "eval_loss": 1.908184289932251, + "eval_runtime": 49.2092, + "eval_samples_per_second": 20.321, + "eval_steps_per_second": 0.853, + "step": 1340 + }, + { + "epoch": 1.3775038520801233, + "grad_norm": 2.9669554233551025, + "learning_rate": 3.0319756376094406e-06, + "loss": 1.5349, + "step": 1341 + }, + { + "epoch": 1.3785310734463276, + "grad_norm": 2.6755905151367188, + "learning_rate": 3.030072325846974e-06, + "loss": 1.536, + "step": 1342 + }, + { + "epoch": 1.379558294812532, + "grad_norm": 2.5073540210723877, + "learning_rate": 3.0281690140845073e-06, + "loss": 1.5167, + "step": 1343 + }, + { + "epoch": 1.3805855161787366, + "grad_norm": 2.520477533340454, + "learning_rate": 3.0262657023220403e-06, + "loss": 1.4827, + "step": 1344 + }, + { + "epoch": 1.381612737544941, + "grad_norm": 2.6202239990234375, + "learning_rate": 3.0243623905595736e-06, + "loss": 1.5271, + "step": 1345 + }, + { + "epoch": 1.3826399589111453, + "grad_norm": 2.667724609375, + "learning_rate": 3.022459078797107e-06, + "loss": 1.537, + "step": 1346 + }, + { + "epoch": 1.3836671802773497, + "grad_norm": 2.701108455657959, + "learning_rate": 3.0205557670346404e-06, + "loss": 1.53, + "step": 1347 + }, + { + "epoch": 1.384694401643554, + "grad_norm": 2.6228628158569336, + "learning_rate": 3.018652455272174e-06, + "loss": 1.5297, + "step": 1348 + }, + { + "epoch": 1.3857216230097587, + "grad_norm": 2.6589505672454834, + "learning_rate": 3.0167491435097075e-06, + "loss": 1.481, + "step": 1349 + }, + { + "epoch": 1.386748844375963, + "grad_norm": 2.805138349533081, + "learning_rate": 3.014845831747241e-06, + "loss": 1.6176, + "step": 1350 + }, + { + "epoch": 1.386748844375963, + "eval_loss": 1.9109174013137817, + "eval_runtime": 44.6864, + "eval_samples_per_second": 22.378, + "eval_steps_per_second": 0.94, + "step": 1350 + }, + { + "epoch": 1.3877760657421674, + "grad_norm": 2.6587769985198975, + "learning_rate": 3.012942519984774e-06, + "loss": 1.5188, + "step": 1351 + }, + { + "epoch": 1.388803287108372, + "grad_norm": 2.7695798873901367, + "learning_rate": 3.011039208222307e-06, + "loss": 1.4913, + "step": 1352 + }, + { + "epoch": 1.3898305084745763, + "grad_norm": 2.7030696868896484, + "learning_rate": 3.0091358964598405e-06, + "loss": 1.4976, + "step": 1353 + }, + { + "epoch": 1.3908577298407807, + "grad_norm": 2.63368558883667, + "learning_rate": 3.007232584697374e-06, + "loss": 1.4575, + "step": 1354 + }, + { + "epoch": 1.391884951206985, + "grad_norm": 2.6904921531677246, + "learning_rate": 3.0053292729349073e-06, + "loss": 1.5067, + "step": 1355 + }, + { + "epoch": 1.3929121725731894, + "grad_norm": 2.7085046768188477, + "learning_rate": 3.00342596117244e-06, + "loss": 1.5215, + "step": 1356 + }, + { + "epoch": 1.393939393939394, + "grad_norm": 2.744744062423706, + "learning_rate": 3.0015226494099736e-06, + "loss": 1.5213, + "step": 1357 + }, + { + "epoch": 1.3949666153055984, + "grad_norm": 2.8159263134002686, + "learning_rate": 2.999619337647507e-06, + "loss": 1.5091, + "step": 1358 + }, + { + "epoch": 1.3959938366718028, + "grad_norm": 2.7380170822143555, + "learning_rate": 2.9977160258850403e-06, + "loss": 1.4863, + "step": 1359 + }, + { + "epoch": 1.3970210580380071, + "grad_norm": 2.9817709922790527, + "learning_rate": 2.9958127141225736e-06, + "loss": 1.4977, + "step": 1360 + }, + { + "epoch": 1.3970210580380071, + "eval_loss": 1.9042565822601318, + "eval_runtime": 45.13, + "eval_samples_per_second": 22.158, + "eval_steps_per_second": 0.931, + "step": 1360 + }, + { + "epoch": 1.3980482794042115, + "grad_norm": 2.7616889476776123, + "learning_rate": 2.9939094023601066e-06, + "loss": 1.4817, + "step": 1361 + }, + { + "epoch": 1.399075500770416, + "grad_norm": 2.695251226425171, + "learning_rate": 2.99200609059764e-06, + "loss": 1.5512, + "step": 1362 + }, + { + "epoch": 1.4001027221366205, + "grad_norm": 2.645728349685669, + "learning_rate": 2.9901027788351733e-06, + "loss": 1.5232, + "step": 1363 + }, + { + "epoch": 1.4011299435028248, + "grad_norm": 2.5545313358306885, + "learning_rate": 2.9881994670727067e-06, + "loss": 1.4626, + "step": 1364 + }, + { + "epoch": 1.4021571648690292, + "grad_norm": 2.6554672718048096, + "learning_rate": 2.98629615531024e-06, + "loss": 1.4813, + "step": 1365 + }, + { + "epoch": 1.4031843862352336, + "grad_norm": 2.543102979660034, + "learning_rate": 2.984392843547773e-06, + "loss": 1.4954, + "step": 1366 + }, + { + "epoch": 1.4042116076014381, + "grad_norm": 2.5635626316070557, + "learning_rate": 2.9824895317853063e-06, + "loss": 1.4858, + "step": 1367 + }, + { + "epoch": 1.4052388289676425, + "grad_norm": 2.60005259513855, + "learning_rate": 2.98058622002284e-06, + "loss": 1.4991, + "step": 1368 + }, + { + "epoch": 1.4062660503338469, + "grad_norm": 2.664973020553589, + "learning_rate": 2.9786829082603735e-06, + "loss": 1.5083, + "step": 1369 + }, + { + "epoch": 1.4072932717000515, + "grad_norm": 2.642303466796875, + "learning_rate": 2.976779596497907e-06, + "loss": 1.5636, + "step": 1370 + }, + { + "epoch": 1.4072932717000515, + "eval_loss": 1.9043673276901245, + "eval_runtime": 45.8156, + "eval_samples_per_second": 21.827, + "eval_steps_per_second": 0.917, + "step": 1370 + }, + { + "epoch": 1.4083204930662558, + "grad_norm": 2.721719264984131, + "learning_rate": 2.97487628473544e-06, + "loss": 1.5627, + "step": 1371 + }, + { + "epoch": 1.4093477144324602, + "grad_norm": 2.665929079055786, + "learning_rate": 2.9729729729729736e-06, + "loss": 1.5684, + "step": 1372 + }, + { + "epoch": 1.4103749357986646, + "grad_norm": 2.672046661376953, + "learning_rate": 2.9710696612105065e-06, + "loss": 1.518, + "step": 1373 + }, + { + "epoch": 1.411402157164869, + "grad_norm": 2.675283432006836, + "learning_rate": 2.96916634944804e-06, + "loss": 1.5604, + "step": 1374 + }, + { + "epoch": 1.4124293785310735, + "grad_norm": 2.6286461353302, + "learning_rate": 2.9672630376855732e-06, + "loss": 1.5003, + "step": 1375 + }, + { + "epoch": 1.413456599897278, + "grad_norm": 2.6903879642486572, + "learning_rate": 2.9653597259231066e-06, + "loss": 1.5601, + "step": 1376 + }, + { + "epoch": 1.4144838212634823, + "grad_norm": 2.7762627601623535, + "learning_rate": 2.96345641416064e-06, + "loss": 1.5326, + "step": 1377 + }, + { + "epoch": 1.4155110426296866, + "grad_norm": 2.758028030395508, + "learning_rate": 2.961553102398173e-06, + "loss": 1.4837, + "step": 1378 + }, + { + "epoch": 1.416538263995891, + "grad_norm": 2.769880771636963, + "learning_rate": 2.9596497906357063e-06, + "loss": 1.4122, + "step": 1379 + }, + { + "epoch": 1.4175654853620956, + "grad_norm": 2.815645217895508, + "learning_rate": 2.9577464788732396e-06, + "loss": 1.555, + "step": 1380 + }, + { + "epoch": 1.4175654853620956, + "eval_loss": 1.9081141948699951, + "eval_runtime": 44.267, + "eval_samples_per_second": 22.59, + "eval_steps_per_second": 0.949, + "step": 1380 + }, + { + "epoch": 1.4185927067283, + "grad_norm": 2.8118088245391846, + "learning_rate": 2.955843167110773e-06, + "loss": 1.5708, + "step": 1381 + }, + { + "epoch": 1.4196199280945043, + "grad_norm": 2.620800495147705, + "learning_rate": 2.9539398553483064e-06, + "loss": 1.4901, + "step": 1382 + }, + { + "epoch": 1.420647149460709, + "grad_norm": 2.6867423057556152, + "learning_rate": 2.9520365435858393e-06, + "loss": 1.4964, + "step": 1383 + }, + { + "epoch": 1.4216743708269133, + "grad_norm": 2.569495916366577, + "learning_rate": 2.9501332318233727e-06, + "loss": 1.5103, + "step": 1384 + }, + { + "epoch": 1.4227015921931176, + "grad_norm": 2.6389119625091553, + "learning_rate": 2.948229920060906e-06, + "loss": 1.5946, + "step": 1385 + }, + { + "epoch": 1.423728813559322, + "grad_norm": 2.767986536026001, + "learning_rate": 2.9463266082984394e-06, + "loss": 1.5296, + "step": 1386 + }, + { + "epoch": 1.4247560349255264, + "grad_norm": 2.5122997760772705, + "learning_rate": 2.9444232965359727e-06, + "loss": 1.4941, + "step": 1387 + }, + { + "epoch": 1.425783256291731, + "grad_norm": 2.7231523990631104, + "learning_rate": 2.9425199847735057e-06, + "loss": 1.5184, + "step": 1388 + }, + { + "epoch": 1.4268104776579353, + "grad_norm": 2.7313649654388428, + "learning_rate": 2.94061667301104e-06, + "loss": 1.5368, + "step": 1389 + }, + { + "epoch": 1.4278376990241397, + "grad_norm": 2.656062602996826, + "learning_rate": 2.938713361248573e-06, + "loss": 1.4958, + "step": 1390 + }, + { + "epoch": 1.4278376990241397, + "eval_loss": 1.9070981740951538, + "eval_runtime": 43.6226, + "eval_samples_per_second": 22.924, + "eval_steps_per_second": 0.963, + "step": 1390 + }, + { + "epoch": 1.428864920390344, + "grad_norm": 2.6621580123901367, + "learning_rate": 2.936810049486106e-06, + "loss": 1.4315, + "step": 1391 + }, + { + "epoch": 1.4298921417565484, + "grad_norm": 2.7683629989624023, + "learning_rate": 2.9349067377236396e-06, + "loss": 1.5042, + "step": 1392 + }, + { + "epoch": 1.430919363122753, + "grad_norm": 2.695810317993164, + "learning_rate": 2.933003425961173e-06, + "loss": 1.5174, + "step": 1393 + }, + { + "epoch": 1.4319465844889574, + "grad_norm": 2.686978816986084, + "learning_rate": 2.9311001141987063e-06, + "loss": 1.4706, + "step": 1394 + }, + { + "epoch": 1.4329738058551618, + "grad_norm": 2.799234628677368, + "learning_rate": 2.9291968024362392e-06, + "loss": 1.4826, + "step": 1395 + }, + { + "epoch": 1.4340010272213661, + "grad_norm": 2.8211469650268555, + "learning_rate": 2.9272934906737726e-06, + "loss": 1.488, + "step": 1396 + }, + { + "epoch": 1.4350282485875705, + "grad_norm": 2.6225461959838867, + "learning_rate": 2.925390178911306e-06, + "loss": 1.4699, + "step": 1397 + }, + { + "epoch": 1.436055469953775, + "grad_norm": 2.6853697299957275, + "learning_rate": 2.9234868671488393e-06, + "loss": 1.4787, + "step": 1398 + }, + { + "epoch": 1.4370826913199795, + "grad_norm": 2.784473419189453, + "learning_rate": 2.9215835553863727e-06, + "loss": 1.5043, + "step": 1399 + }, + { + "epoch": 1.4381099126861838, + "grad_norm": 2.706911087036133, + "learning_rate": 2.9196802436239056e-06, + "loss": 1.512, + "step": 1400 + }, + { + "epoch": 1.4381099126861838, + "eval_loss": 1.9136111736297607, + "eval_runtime": 43.5675, + "eval_samples_per_second": 22.953, + "eval_steps_per_second": 0.964, + "step": 1400 + }, + { + "epoch": 1.4391371340523884, + "grad_norm": 2.6579840183258057, + "learning_rate": 2.917776931861439e-06, + "loss": 1.5073, + "step": 1401 + }, + { + "epoch": 1.4401643554185928, + "grad_norm": 2.7359421253204346, + "learning_rate": 2.9158736200989723e-06, + "loss": 1.4819, + "step": 1402 + }, + { + "epoch": 1.4411915767847971, + "grad_norm": 2.604591131210327, + "learning_rate": 2.9139703083365057e-06, + "loss": 1.4758, + "step": 1403 + }, + { + "epoch": 1.4422187981510015, + "grad_norm": 2.7267162799835205, + "learning_rate": 2.912066996574039e-06, + "loss": 1.4835, + "step": 1404 + }, + { + "epoch": 1.4432460195172059, + "grad_norm": 2.844331741333008, + "learning_rate": 2.910163684811572e-06, + "loss": 1.5096, + "step": 1405 + }, + { + "epoch": 1.4442732408834105, + "grad_norm": 2.6915996074676514, + "learning_rate": 2.9082603730491054e-06, + "loss": 1.4536, + "step": 1406 + }, + { + "epoch": 1.4453004622496148, + "grad_norm": 2.9222843647003174, + "learning_rate": 2.9063570612866387e-06, + "loss": 1.5769, + "step": 1407 + }, + { + "epoch": 1.4463276836158192, + "grad_norm": 2.667371988296509, + "learning_rate": 2.904453749524172e-06, + "loss": 1.533, + "step": 1408 + }, + { + "epoch": 1.4473549049820236, + "grad_norm": 2.643632411956787, + "learning_rate": 2.902550437761706e-06, + "loss": 1.4666, + "step": 1409 + }, + { + "epoch": 1.448382126348228, + "grad_norm": 2.8633203506469727, + "learning_rate": 2.9006471259992392e-06, + "loss": 1.4826, + "step": 1410 + }, + { + "epoch": 1.448382126348228, + "eval_loss": 1.9127975702285767, + "eval_runtime": 42.2168, + "eval_samples_per_second": 23.687, + "eval_steps_per_second": 0.995, + "step": 1410 + }, + { + "epoch": 1.4494093477144325, + "grad_norm": 2.7763702869415283, + "learning_rate": 2.8987438142367726e-06, + "loss": 1.4969, + "step": 1411 + }, + { + "epoch": 1.450436569080637, + "grad_norm": 2.598337411880493, + "learning_rate": 2.8968405024743055e-06, + "loss": 1.4993, + "step": 1412 + }, + { + "epoch": 1.4514637904468413, + "grad_norm": 2.717944622039795, + "learning_rate": 2.894937190711839e-06, + "loss": 1.5479, + "step": 1413 + }, + { + "epoch": 1.4524910118130456, + "grad_norm": 2.5288076400756836, + "learning_rate": 2.8930338789493723e-06, + "loss": 1.4564, + "step": 1414 + }, + { + "epoch": 1.45351823317925, + "grad_norm": 2.6042704582214355, + "learning_rate": 2.8911305671869056e-06, + "loss": 1.5267, + "step": 1415 + }, + { + "epoch": 1.4545454545454546, + "grad_norm": 2.5768539905548096, + "learning_rate": 2.889227255424439e-06, + "loss": 1.4958, + "step": 1416 + }, + { + "epoch": 1.455572675911659, + "grad_norm": 2.62872576713562, + "learning_rate": 2.887323943661972e-06, + "loss": 1.4407, + "step": 1417 + }, + { + "epoch": 1.4565998972778633, + "grad_norm": 2.8656342029571533, + "learning_rate": 2.8854206318995053e-06, + "loss": 1.4944, + "step": 1418 + }, + { + "epoch": 1.457627118644068, + "grad_norm": 2.7083373069763184, + "learning_rate": 2.8835173201370386e-06, + "loss": 1.474, + "step": 1419 + }, + { + "epoch": 1.4586543400102723, + "grad_norm": 2.712958812713623, + "learning_rate": 2.881614008374572e-06, + "loss": 1.4982, + "step": 1420 + }, + { + "epoch": 1.4586543400102723, + "eval_loss": 1.9154274463653564, + "eval_runtime": 45.8592, + "eval_samples_per_second": 21.806, + "eval_steps_per_second": 0.916, + "step": 1420 + }, + { + "epoch": 1.4596815613764766, + "grad_norm": 3.0084564685821533, + "learning_rate": 2.8797106966121054e-06, + "loss": 1.5505, + "step": 1421 + }, + { + "epoch": 1.460708782742681, + "grad_norm": 2.793546438217163, + "learning_rate": 2.8778073848496383e-06, + "loss": 1.5203, + "step": 1422 + }, + { + "epoch": 1.4617360041088854, + "grad_norm": 2.5944628715515137, + "learning_rate": 2.8759040730871717e-06, + "loss": 1.5016, + "step": 1423 + }, + { + "epoch": 1.46276322547509, + "grad_norm": 2.7558817863464355, + "learning_rate": 2.874000761324705e-06, + "loss": 1.5273, + "step": 1424 + }, + { + "epoch": 1.4637904468412943, + "grad_norm": 2.582310676574707, + "learning_rate": 2.8720974495622384e-06, + "loss": 1.4862, + "step": 1425 + }, + { + "epoch": 1.4648176682074987, + "grad_norm": 2.639944314956665, + "learning_rate": 2.8701941377997718e-06, + "loss": 1.5276, + "step": 1426 + }, + { + "epoch": 1.465844889573703, + "grad_norm": 2.613696575164795, + "learning_rate": 2.8682908260373047e-06, + "loss": 1.5061, + "step": 1427 + }, + { + "epoch": 1.4668721109399074, + "grad_norm": 2.7124717235565186, + "learning_rate": 2.866387514274838e-06, + "loss": 1.5517, + "step": 1428 + }, + { + "epoch": 1.467899332306112, + "grad_norm": 2.7124717235565186, + "learning_rate": 2.866387514274838e-06, + "loss": 1.5288, + "step": 1429 + }, + { + "epoch": 1.4689265536723164, + "grad_norm": 2.722407102584839, + "learning_rate": 2.864484202512372e-06, + "loss": 1.5256, + "step": 1430 + }, + { + "epoch": 1.4689265536723164, + "eval_loss": 1.900185227394104, + "eval_runtime": 47.7724, + "eval_samples_per_second": 20.933, + "eval_steps_per_second": 0.879, + "step": 1430 + }, + { + "epoch": 1.4699537750385208, + "grad_norm": 2.690464973449707, + "learning_rate": 2.8625808907499052e-06, + "loss": 1.4841, + "step": 1431 + }, + { + "epoch": 1.4709809964047253, + "grad_norm": 2.766082525253296, + "learning_rate": 2.8606775789874386e-06, + "loss": 1.5171, + "step": 1432 + }, + { + "epoch": 1.4720082177709297, + "grad_norm": 2.6420655250549316, + "learning_rate": 2.858774267224972e-06, + "loss": 1.4871, + "step": 1433 + }, + { + "epoch": 1.473035439137134, + "grad_norm": 2.687546730041504, + "learning_rate": 2.8568709554625053e-06, + "loss": 1.5116, + "step": 1434 + }, + { + "epoch": 1.4740626605033385, + "grad_norm": 2.6613612174987793, + "learning_rate": 2.8549676437000382e-06, + "loss": 1.5599, + "step": 1435 + }, + { + "epoch": 1.4750898818695428, + "grad_norm": 2.7249948978424072, + "learning_rate": 2.8530643319375716e-06, + "loss": 1.5342, + "step": 1436 + }, + { + "epoch": 1.4761171032357474, + "grad_norm": 2.768580913543701, + "learning_rate": 2.851161020175105e-06, + "loss": 1.5022, + "step": 1437 + }, + { + "epoch": 1.4771443246019518, + "grad_norm": 2.6677844524383545, + "learning_rate": 2.8492577084126383e-06, + "loss": 1.4712, + "step": 1438 + }, + { + "epoch": 1.4781715459681561, + "grad_norm": 2.693856716156006, + "learning_rate": 2.8473543966501717e-06, + "loss": 1.5163, + "step": 1439 + }, + { + "epoch": 1.4791987673343605, + "grad_norm": 2.667125701904297, + "learning_rate": 2.8454510848877046e-06, + "loss": 1.5601, + "step": 1440 + }, + { + "epoch": 1.4791987673343605, + "eval_loss": 1.8993381261825562, + "eval_runtime": 47.2966, + "eval_samples_per_second": 21.143, + "eval_steps_per_second": 0.888, + "step": 1440 + }, + { + "epoch": 1.4802259887005649, + "grad_norm": 2.6483078002929688, + "learning_rate": 2.843547773125238e-06, + "loss": 1.5042, + "step": 1441 + }, + { + "epoch": 1.4812532100667695, + "grad_norm": 2.6776845455169678, + "learning_rate": 2.8416444613627714e-06, + "loss": 1.5096, + "step": 1442 + }, + { + "epoch": 1.4822804314329738, + "grad_norm": 2.7866480350494385, + "learning_rate": 2.8397411496003047e-06, + "loss": 1.4684, + "step": 1443 + }, + { + "epoch": 1.4833076527991782, + "grad_norm": 2.844266176223755, + "learning_rate": 2.837837837837838e-06, + "loss": 1.499, + "step": 1444 + }, + { + "epoch": 1.4843348741653826, + "grad_norm": 2.5618107318878174, + "learning_rate": 2.835934526075371e-06, + "loss": 1.5212, + "step": 1445 + }, + { + "epoch": 1.485362095531587, + "grad_norm": 2.6264190673828125, + "learning_rate": 2.8340312143129044e-06, + "loss": 1.4763, + "step": 1446 + }, + { + "epoch": 1.4863893168977915, + "grad_norm": 2.68026065826416, + "learning_rate": 2.8321279025504377e-06, + "loss": 1.4572, + "step": 1447 + }, + { + "epoch": 1.487416538263996, + "grad_norm": 2.7439587116241455, + "learning_rate": 2.830224590787971e-06, + "loss": 1.5778, + "step": 1448 + }, + { + "epoch": 1.4884437596302003, + "grad_norm": 2.708627700805664, + "learning_rate": 2.8283212790255045e-06, + "loss": 1.5177, + "step": 1449 + }, + { + "epoch": 1.4894709809964048, + "grad_norm": 2.5931451320648193, + "learning_rate": 2.8264179672630383e-06, + "loss": 1.5499, + "step": 1450 + }, + { + "epoch": 1.4894709809964048, + "eval_loss": 1.8863205909729004, + "eval_runtime": 45.8354, + "eval_samples_per_second": 21.817, + "eval_steps_per_second": 0.916, + "step": 1450 + }, + { + "epoch": 1.4904982023626092, + "grad_norm": 2.6775002479553223, + "learning_rate": 2.8245146555005716e-06, + "loss": 1.4987, + "step": 1451 + }, + { + "epoch": 1.4915254237288136, + "grad_norm": 2.568394660949707, + "learning_rate": 2.8226113437381046e-06, + "loss": 1.449, + "step": 1452 + }, + { + "epoch": 1.492552645095018, + "grad_norm": 2.6354329586029053, + "learning_rate": 2.820708031975638e-06, + "loss": 1.5018, + "step": 1453 + }, + { + "epoch": 1.4935798664612223, + "grad_norm": 2.731614112854004, + "learning_rate": 2.8188047202131713e-06, + "loss": 1.5337, + "step": 1454 + }, + { + "epoch": 1.494607087827427, + "grad_norm": 2.662299633026123, + "learning_rate": 2.8169014084507046e-06, + "loss": 1.5449, + "step": 1455 + }, + { + "epoch": 1.4956343091936313, + "grad_norm": 2.6484053134918213, + "learning_rate": 2.814998096688238e-06, + "loss": 1.5048, + "step": 1456 + }, + { + "epoch": 1.4966615305598356, + "grad_norm": 2.5573649406433105, + "learning_rate": 2.813094784925771e-06, + "loss": 1.4417, + "step": 1457 + }, + { + "epoch": 1.49768875192604, + "grad_norm": 2.6440303325653076, + "learning_rate": 2.8111914731633043e-06, + "loss": 1.5545, + "step": 1458 + }, + { + "epoch": 1.4987159732922444, + "grad_norm": 2.6014726161956787, + "learning_rate": 2.8092881614008377e-06, + "loss": 1.4995, + "step": 1459 + }, + { + "epoch": 1.499743194658449, + "grad_norm": 2.8325798511505127, + "learning_rate": 2.807384849638371e-06, + "loss": 1.4762, + "step": 1460 + }, + { + "epoch": 1.499743194658449, + "eval_loss": 1.8842575550079346, + "eval_runtime": 46.058, + "eval_samples_per_second": 21.712, + "eval_steps_per_second": 0.912, + "step": 1460 + }, + { + "epoch": 1.5007704160246533, + "grad_norm": 2.629534959793091, + "learning_rate": 2.8054815378759044e-06, + "loss": 1.5235, + "step": 1461 + }, + { + "epoch": 1.5017976373908577, + "grad_norm": 2.723256826400757, + "learning_rate": 2.8035782261134378e-06, + "loss": 1.4962, + "step": 1462 + }, + { + "epoch": 1.5028248587570623, + "grad_norm": 2.7497754096984863, + "learning_rate": 2.8016749143509707e-06, + "loss": 1.4931, + "step": 1463 + }, + { + "epoch": 1.5038520801232664, + "grad_norm": 2.695408821105957, + "learning_rate": 2.799771602588504e-06, + "loss": 1.4921, + "step": 1464 + }, + { + "epoch": 1.504879301489471, + "grad_norm": 2.6144590377807617, + "learning_rate": 2.7978682908260374e-06, + "loss": 1.5191, + "step": 1465 + }, + { + "epoch": 1.5059065228556754, + "grad_norm": 2.7783288955688477, + "learning_rate": 2.795964979063571e-06, + "loss": 1.5416, + "step": 1466 + }, + { + "epoch": 1.5069337442218798, + "grad_norm": 2.7083182334899902, + "learning_rate": 2.794061667301104e-06, + "loss": 1.5126, + "step": 1467 + }, + { + "epoch": 1.5079609655880843, + "grad_norm": 2.7532167434692383, + "learning_rate": 2.792158355538637e-06, + "loss": 1.5244, + "step": 1468 + }, + { + "epoch": 1.5089881869542885, + "grad_norm": 2.6125259399414062, + "learning_rate": 2.7902550437761705e-06, + "loss": 1.5194, + "step": 1469 + }, + { + "epoch": 1.510015408320493, + "grad_norm": 2.585392713546753, + "learning_rate": 2.7883517320137042e-06, + "loss": 1.474, + "step": 1470 + }, + { + "epoch": 1.510015408320493, + "eval_loss": 1.8812134265899658, + "eval_runtime": 46.7392, + "eval_samples_per_second": 21.395, + "eval_steps_per_second": 0.899, + "step": 1470 + }, + { + "epoch": 1.5110426296866974, + "grad_norm": 2.586333751678467, + "learning_rate": 2.7864484202512376e-06, + "loss": 1.4666, + "step": 1471 + }, + { + "epoch": 1.5120698510529018, + "grad_norm": 2.7523810863494873, + "learning_rate": 2.784545108488771e-06, + "loss": 1.5113, + "step": 1472 + }, + { + "epoch": 1.5130970724191064, + "grad_norm": 2.688206434249878, + "learning_rate": 2.7826417967263043e-06, + "loss": 1.5049, + "step": 1473 + }, + { + "epoch": 1.5141242937853108, + "grad_norm": 2.642054319381714, + "learning_rate": 2.7807384849638373e-06, + "loss": 1.5534, + "step": 1474 + }, + { + "epoch": 1.5151515151515151, + "grad_norm": 2.694291591644287, + "learning_rate": 2.7788351732013706e-06, + "loss": 1.5052, + "step": 1475 + }, + { + "epoch": 1.5161787365177197, + "grad_norm": 2.6969082355499268, + "learning_rate": 2.776931861438904e-06, + "loss": 1.491, + "step": 1476 + }, + { + "epoch": 1.5172059578839239, + "grad_norm": 2.6937081813812256, + "learning_rate": 2.7750285496764374e-06, + "loss": 1.5081, + "step": 1477 + }, + { + "epoch": 1.5182331792501285, + "grad_norm": 2.6754133701324463, + "learning_rate": 2.7731252379139707e-06, + "loss": 1.5107, + "step": 1478 + }, + { + "epoch": 1.5192604006163328, + "grad_norm": 2.586491346359253, + "learning_rate": 2.771221926151504e-06, + "loss": 1.4486, + "step": 1479 + }, + { + "epoch": 1.5202876219825372, + "grad_norm": 2.82452130317688, + "learning_rate": 2.769318614389037e-06, + "loss": 1.5089, + "step": 1480 + }, + { + "epoch": 1.5202876219825372, + "eval_loss": 1.879558801651001, + "eval_runtime": 45.9914, + "eval_samples_per_second": 21.743, + "eval_steps_per_second": 0.913, + "step": 1480 + }, + { + "epoch": 1.5213148433487418, + "grad_norm": 2.69734263420105, + "learning_rate": 2.7674153026265704e-06, + "loss": 1.5069, + "step": 1481 + }, + { + "epoch": 1.522342064714946, + "grad_norm": 2.7553112506866455, + "learning_rate": 2.7655119908641037e-06, + "loss": 1.5211, + "step": 1482 + }, + { + "epoch": 1.5233692860811505, + "grad_norm": 2.659170389175415, + "learning_rate": 2.763608679101637e-06, + "loss": 1.4967, + "step": 1483 + }, + { + "epoch": 1.5243965074473549, + "grad_norm": 2.6266300678253174, + "learning_rate": 2.7617053673391705e-06, + "loss": 1.5376, + "step": 1484 + }, + { + "epoch": 1.5254237288135593, + "grad_norm": 2.6549487113952637, + "learning_rate": 2.7598020555767034e-06, + "loss": 1.546, + "step": 1485 + }, + { + "epoch": 1.5264509501797638, + "grad_norm": 2.651841402053833, + "learning_rate": 2.7578987438142368e-06, + "loss": 1.4625, + "step": 1486 + }, + { + "epoch": 1.5274781715459682, + "grad_norm": 2.783480644226074, + "learning_rate": 2.75599543205177e-06, + "loss": 1.5165, + "step": 1487 + }, + { + "epoch": 1.5285053929121726, + "grad_norm": 2.674964189529419, + "learning_rate": 2.7540921202893035e-06, + "loss": 1.539, + "step": 1488 + }, + { + "epoch": 1.529532614278377, + "grad_norm": 2.7219326496124268, + "learning_rate": 2.752188808526837e-06, + "loss": 1.5151, + "step": 1489 + }, + { + "epoch": 1.5305598356445813, + "grad_norm": 2.6149561405181885, + "learning_rate": 2.75028549676437e-06, + "loss": 1.498, + "step": 1490 + }, + { + "epoch": 1.5305598356445813, + "eval_loss": 1.8944125175476074, + "eval_runtime": 45.5669, + "eval_samples_per_second": 21.946, + "eval_steps_per_second": 0.922, + "step": 1490 + }, + { + "epoch": 1.531587057010786, + "grad_norm": 2.6966514587402344, + "learning_rate": 2.7483821850019036e-06, + "loss": 1.5026, + "step": 1491 + }, + { + "epoch": 1.5326142783769903, + "grad_norm": 2.7445616722106934, + "learning_rate": 2.746478873239437e-06, + "loss": 1.493, + "step": 1492 + }, + { + "epoch": 1.5336414997431946, + "grad_norm": 2.8391168117523193, + "learning_rate": 2.7445755614769703e-06, + "loss": 1.4918, + "step": 1493 + }, + { + "epoch": 1.5346687211093992, + "grad_norm": 2.79190731048584, + "learning_rate": 2.7426722497145037e-06, + "loss": 1.4654, + "step": 1494 + }, + { + "epoch": 1.5356959424756034, + "grad_norm": 2.7497293949127197, + "learning_rate": 2.740768937952037e-06, + "loss": 1.4702, + "step": 1495 + }, + { + "epoch": 1.536723163841808, + "grad_norm": 2.7948496341705322, + "learning_rate": 2.7388656261895704e-06, + "loss": 1.4879, + "step": 1496 + }, + { + "epoch": 1.5377503852080123, + "grad_norm": 2.9910364151000977, + "learning_rate": 2.7369623144271033e-06, + "loss": 1.5689, + "step": 1497 + }, + { + "epoch": 1.5387776065742167, + "grad_norm": 2.726814031600952, + "learning_rate": 2.7350590026646367e-06, + "loss": 1.492, + "step": 1498 + }, + { + "epoch": 1.5398048279404213, + "grad_norm": 2.839493751525879, + "learning_rate": 2.73315569090217e-06, + "loss": 1.4539, + "step": 1499 + }, + { + "epoch": 1.5408320493066254, + "grad_norm": 2.776416063308716, + "learning_rate": 2.7312523791397034e-06, + "loss": 1.5509, + "step": 1500 + }, + { + "epoch": 1.5408320493066254, + "eval_loss": 1.9009771347045898, + "eval_runtime": 51.232, + "eval_samples_per_second": 19.519, + "eval_steps_per_second": 0.82, + "step": 1500 + }, + { + "epoch": 1.54185927067283, + "grad_norm": 2.8426146507263184, + "learning_rate": 2.7293490673772368e-06, + "loss": 1.5696, + "step": 1501 + }, + { + "epoch": 1.5428864920390344, + "grad_norm": 2.8076331615448, + "learning_rate": 2.7274457556147697e-06, + "loss": 1.5216, + "step": 1502 + }, + { + "epoch": 1.5439137134052388, + "grad_norm": 2.800166368484497, + "learning_rate": 2.725542443852303e-06, + "loss": 1.5094, + "step": 1503 + }, + { + "epoch": 1.5449409347714433, + "grad_norm": 2.597820281982422, + "learning_rate": 2.7236391320898365e-06, + "loss": 1.5149, + "step": 1504 + }, + { + "epoch": 1.5459681561376477, + "grad_norm": 2.557790517807007, + "learning_rate": 2.72173582032737e-06, + "loss": 1.4817, + "step": 1505 + }, + { + "epoch": 1.546995377503852, + "grad_norm": 2.6866400241851807, + "learning_rate": 2.719832508564903e-06, + "loss": 1.5024, + "step": 1506 + }, + { + "epoch": 1.5480225988700564, + "grad_norm": 2.739360809326172, + "learning_rate": 2.717929196802436e-06, + "loss": 1.5151, + "step": 1507 + }, + { + "epoch": 1.5490498202362608, + "grad_norm": 2.694063186645508, + "learning_rate": 2.7160258850399695e-06, + "loss": 1.4815, + "step": 1508 + }, + { + "epoch": 1.5500770416024654, + "grad_norm": 2.4691152572631836, + "learning_rate": 2.714122573277503e-06, + "loss": 1.4989, + "step": 1509 + }, + { + "epoch": 1.5511042629686698, + "grad_norm": 2.594684362411499, + "learning_rate": 2.712219261515036e-06, + "loss": 1.499, + "step": 1510 + }, + { + "epoch": 1.5511042629686698, + "eval_loss": 1.891904592514038, + "eval_runtime": 44.1399, + "eval_samples_per_second": 22.655, + "eval_steps_per_second": 0.952, + "step": 1510 + }, + { + "epoch": 1.5521314843348741, + "grad_norm": 2.621588706970215, + "learning_rate": 2.71031594975257e-06, + "loss": 1.516, + "step": 1511 + }, + { + "epoch": 1.5531587057010787, + "grad_norm": 2.677280902862549, + "learning_rate": 2.7084126379901034e-06, + "loss": 1.5192, + "step": 1512 + }, + { + "epoch": 1.5541859270672829, + "grad_norm": 2.7408759593963623, + "learning_rate": 2.7065093262276367e-06, + "loss": 1.5377, + "step": 1513 + }, + { + "epoch": 1.5552131484334875, + "grad_norm": 2.622382402420044, + "learning_rate": 2.7046060144651697e-06, + "loss": 1.5437, + "step": 1514 + }, + { + "epoch": 1.5562403697996918, + "grad_norm": 2.8121471405029297, + "learning_rate": 2.702702702702703e-06, + "loss": 1.5698, + "step": 1515 + }, + { + "epoch": 1.5572675911658962, + "grad_norm": 2.5897836685180664, + "learning_rate": 2.7007993909402364e-06, + "loss": 1.4295, + "step": 1516 + }, + { + "epoch": 1.5582948125321008, + "grad_norm": 2.793713092803955, + "learning_rate": 2.6988960791777697e-06, + "loss": 1.5575, + "step": 1517 + }, + { + "epoch": 1.559322033898305, + "grad_norm": 2.767385244369507, + "learning_rate": 2.696992767415303e-06, + "loss": 1.5007, + "step": 1518 + }, + { + "epoch": 1.5603492552645095, + "grad_norm": 2.6531004905700684, + "learning_rate": 2.695089455652836e-06, + "loss": 1.4977, + "step": 1519 + }, + { + "epoch": 1.5613764766307139, + "grad_norm": 2.686450719833374, + "learning_rate": 2.6931861438903694e-06, + "loss": 1.479, + "step": 1520 + }, + { + "epoch": 1.5613764766307139, + "eval_loss": 1.88840651512146, + "eval_runtime": 44.5631, + "eval_samples_per_second": 22.44, + "eval_steps_per_second": 0.942, + "step": 1520 + }, + { + "epoch": 1.5624036979969183, + "grad_norm": 2.6664063930511475, + "learning_rate": 2.6912828321279028e-06, + "loss": 1.5267, + "step": 1521 + }, + { + "epoch": 1.5634309193631228, + "grad_norm": 2.676542282104492, + "learning_rate": 2.689379520365436e-06, + "loss": 1.5005, + "step": 1522 + }, + { + "epoch": 1.5644581407293272, + "grad_norm": 2.8331971168518066, + "learning_rate": 2.6874762086029695e-06, + "loss": 1.5308, + "step": 1523 + }, + { + "epoch": 1.5654853620955316, + "grad_norm": 2.6427881717681885, + "learning_rate": 2.6855728968405024e-06, + "loss": 1.4813, + "step": 1524 + }, + { + "epoch": 1.5665125834617362, + "grad_norm": 2.7058424949645996, + "learning_rate": 2.683669585078036e-06, + "loss": 1.5149, + "step": 1525 + }, + { + "epoch": 1.5675398048279403, + "grad_norm": 2.7069029808044434, + "learning_rate": 2.681766273315569e-06, + "loss": 1.4588, + "step": 1526 + }, + { + "epoch": 1.568567026194145, + "grad_norm": 2.7757389545440674, + "learning_rate": 2.6798629615531025e-06, + "loss": 1.4953, + "step": 1527 + }, + { + "epoch": 1.5695942475603493, + "grad_norm": 2.788466215133667, + "learning_rate": 2.677959649790636e-06, + "loss": 1.5028, + "step": 1528 + }, + { + "epoch": 1.5706214689265536, + "grad_norm": 2.659156322479248, + "learning_rate": 2.676056338028169e-06, + "loss": 1.5365, + "step": 1529 + }, + { + "epoch": 1.5716486902927582, + "grad_norm": 2.7179627418518066, + "learning_rate": 2.674153026265702e-06, + "loss": 1.5232, + "step": 1530 + }, + { + "epoch": 1.5716486902927582, + "eval_loss": 1.8908981084823608, + "eval_runtime": 44.1745, + "eval_samples_per_second": 22.637, + "eval_steps_per_second": 0.951, + "step": 1530 + }, + { + "epoch": 1.5726759116589624, + "grad_norm": 2.676546812057495, + "learning_rate": 2.672249714503236e-06, + "loss": 1.5247, + "step": 1531 + }, + { + "epoch": 1.573703133025167, + "grad_norm": 2.629784345626831, + "learning_rate": 2.6703464027407693e-06, + "loss": 1.488, + "step": 1532 + }, + { + "epoch": 1.5747303543913713, + "grad_norm": 2.7101712226867676, + "learning_rate": 2.6684430909783027e-06, + "loss": 1.5135, + "step": 1533 + }, + { + "epoch": 1.5757575757575757, + "grad_norm": 2.8025403022766113, + "learning_rate": 2.666539779215836e-06, + "loss": 1.5024, + "step": 1534 + }, + { + "epoch": 1.5767847971237803, + "grad_norm": 2.6695220470428467, + "learning_rate": 2.6646364674533694e-06, + "loss": 1.4844, + "step": 1535 + }, + { + "epoch": 1.5778120184899846, + "grad_norm": 2.610874652862549, + "learning_rate": 2.6627331556909024e-06, + "loss": 1.3949, + "step": 1536 + }, + { + "epoch": 1.578839239856189, + "grad_norm": 2.926985025405884, + "learning_rate": 2.6608298439284357e-06, + "loss": 1.4941, + "step": 1537 + }, + { + "epoch": 1.5798664612223934, + "grad_norm": 2.610355854034424, + "learning_rate": 2.658926532165969e-06, + "loss": 1.4895, + "step": 1538 + }, + { + "epoch": 1.5808936825885977, + "grad_norm": 2.6426877975463867, + "learning_rate": 2.6570232204035024e-06, + "loss": 1.4483, + "step": 1539 + }, + { + "epoch": 1.5819209039548023, + "grad_norm": 2.7410695552825928, + "learning_rate": 2.655119908641036e-06, + "loss": 1.4931, + "step": 1540 + }, + { + "epoch": 1.5819209039548023, + "eval_loss": 1.903024673461914, + "eval_runtime": 43.387, + "eval_samples_per_second": 23.048, + "eval_steps_per_second": 0.968, + "step": 1540 + }, + { + "epoch": 1.5829481253210067, + "grad_norm": 2.6797378063201904, + "learning_rate": 2.6532165968785687e-06, + "loss": 1.497, + "step": 1541 + }, + { + "epoch": 1.583975346687211, + "grad_norm": 2.697439670562744, + "learning_rate": 2.651313285116102e-06, + "loss": 1.4596, + "step": 1542 + }, + { + "epoch": 1.5850025680534157, + "grad_norm": 2.6819119453430176, + "learning_rate": 2.6494099733536355e-06, + "loss": 1.4712, + "step": 1543 + }, + { + "epoch": 1.5860297894196198, + "grad_norm": 2.9221737384796143, + "learning_rate": 2.647506661591169e-06, + "loss": 1.6449, + "step": 1544 + }, + { + "epoch": 1.5870570107858244, + "grad_norm": 2.757866859436035, + "learning_rate": 2.645603349828702e-06, + "loss": 1.5251, + "step": 1545 + }, + { + "epoch": 1.5880842321520288, + "grad_norm": 2.693976402282715, + "learning_rate": 2.643700038066235e-06, + "loss": 1.462, + "step": 1546 + }, + { + "epoch": 1.5891114535182331, + "grad_norm": 2.732808828353882, + "learning_rate": 2.6417967263037685e-06, + "loss": 1.5115, + "step": 1547 + }, + { + "epoch": 1.5901386748844377, + "grad_norm": 2.5835330486297607, + "learning_rate": 2.639893414541302e-06, + "loss": 1.533, + "step": 1548 + }, + { + "epoch": 1.5911658962506419, + "grad_norm": 2.6625828742980957, + "learning_rate": 2.6379901027788352e-06, + "loss": 1.555, + "step": 1549 + }, + { + "epoch": 1.5921931176168465, + "grad_norm": 2.5941221714019775, + "learning_rate": 2.6360867910163686e-06, + "loss": 1.4881, + "step": 1550 + }, + { + "epoch": 1.5921931176168465, + "eval_loss": 1.9022443294525146, + "eval_runtime": 46.3086, + "eval_samples_per_second": 21.594, + "eval_steps_per_second": 0.907, + "step": 1550 + }, + { + "epoch": 1.5932203389830508, + "grad_norm": 2.6347010135650635, + "learning_rate": 2.6341834792539024e-06, + "loss": 1.5118, + "step": 1551 + }, + { + "epoch": 1.5942475603492552, + "grad_norm": 2.6772608757019043, + "learning_rate": 2.6322801674914357e-06, + "loss": 1.5107, + "step": 1552 + }, + { + "epoch": 1.5952747817154598, + "grad_norm": 2.819559097290039, + "learning_rate": 2.6303768557289687e-06, + "loss": 1.5008, + "step": 1553 + }, + { + "epoch": 1.5963020030816641, + "grad_norm": 2.5580296516418457, + "learning_rate": 2.628473543966502e-06, + "loss": 1.4561, + "step": 1554 + }, + { + "epoch": 1.5973292244478685, + "grad_norm": 2.6931111812591553, + "learning_rate": 2.6265702322040354e-06, + "loss": 1.5222, + "step": 1555 + }, + { + "epoch": 1.5983564458140729, + "grad_norm": 2.6396803855895996, + "learning_rate": 2.6246669204415688e-06, + "loss": 1.4791, + "step": 1556 + }, + { + "epoch": 1.5993836671802772, + "grad_norm": 2.6619536876678467, + "learning_rate": 2.622763608679102e-06, + "loss": 1.471, + "step": 1557 + }, + { + "epoch": 1.6004108885464818, + "grad_norm": 2.578331232070923, + "learning_rate": 2.620860296916635e-06, + "loss": 1.4787, + "step": 1558 + }, + { + "epoch": 1.6014381099126862, + "grad_norm": 2.764533042907715, + "learning_rate": 2.6189569851541684e-06, + "loss": 1.4305, + "step": 1559 + }, + { + "epoch": 1.6024653312788906, + "grad_norm": 2.673753499984741, + "learning_rate": 2.617053673391702e-06, + "loss": 1.536, + "step": 1560 + }, + { + "epoch": 1.6024653312788906, + "eval_loss": 1.8979606628417969, + "eval_runtime": 46.5101, + "eval_samples_per_second": 21.501, + "eval_steps_per_second": 0.903, + "step": 1560 + }, + { + "epoch": 1.6034925526450952, + "grad_norm": 2.7593281269073486, + "learning_rate": 2.615150361629235e-06, + "loss": 1.5721, + "step": 1561 + }, + { + "epoch": 1.6045197740112993, + "grad_norm": 2.72963285446167, + "learning_rate": 2.6132470498667685e-06, + "loss": 1.5198, + "step": 1562 + }, + { + "epoch": 1.605546995377504, + "grad_norm": 2.6438205242156982, + "learning_rate": 2.6113437381043015e-06, + "loss": 1.4922, + "step": 1563 + }, + { + "epoch": 1.6065742167437083, + "grad_norm": 2.6523141860961914, + "learning_rate": 2.609440426341835e-06, + "loss": 1.4775, + "step": 1564 + }, + { + "epoch": 1.6076014381099126, + "grad_norm": 2.7506513595581055, + "learning_rate": 2.607537114579368e-06, + "loss": 1.524, + "step": 1565 + }, + { + "epoch": 1.6086286594761172, + "grad_norm": 2.589853286743164, + "learning_rate": 2.6056338028169015e-06, + "loss": 1.4946, + "step": 1566 + }, + { + "epoch": 1.6096558808423214, + "grad_norm": 2.6591460704803467, + "learning_rate": 2.603730491054435e-06, + "loss": 1.4686, + "step": 1567 + }, + { + "epoch": 1.610683102208526, + "grad_norm": 2.8163933753967285, + "learning_rate": 2.601827179291968e-06, + "loss": 1.5196, + "step": 1568 + }, + { + "epoch": 1.6117103235747303, + "grad_norm": 2.6938319206237793, + "learning_rate": 2.599923867529501e-06, + "loss": 1.5082, + "step": 1569 + }, + { + "epoch": 1.6127375449409347, + "grad_norm": 2.7489402294158936, + "learning_rate": 2.5980205557670346e-06, + "loss": 1.4978, + "step": 1570 + }, + { + "epoch": 1.6127375449409347, + "eval_loss": 1.897243618965149, + "eval_runtime": 45.6931, + "eval_samples_per_second": 21.885, + "eval_steps_per_second": 0.919, + "step": 1570 + }, + { + "epoch": 1.6137647663071393, + "grad_norm": 2.763204336166382, + "learning_rate": 2.5961172440045684e-06, + "loss": 1.5382, + "step": 1571 + }, + { + "epoch": 1.6147919876733436, + "grad_norm": 2.726813316345215, + "learning_rate": 2.5942139322421017e-06, + "loss": 1.5377, + "step": 1572 + }, + { + "epoch": 1.615819209039548, + "grad_norm": 2.6197266578674316, + "learning_rate": 2.592310620479635e-06, + "loss": 1.4887, + "step": 1573 + }, + { + "epoch": 1.6168464304057526, + "grad_norm": 2.739987373352051, + "learning_rate": 2.5904073087171684e-06, + "loss": 1.4546, + "step": 1574 + }, + { + "epoch": 1.6178736517719567, + "grad_norm": 2.6571943759918213, + "learning_rate": 2.5885039969547014e-06, + "loss": 1.5129, + "step": 1575 + }, + { + "epoch": 1.6189008731381613, + "grad_norm": 2.7591447830200195, + "learning_rate": 2.5866006851922347e-06, + "loss": 1.4627, + "step": 1576 + }, + { + "epoch": 1.6199280945043657, + "grad_norm": 2.6228175163269043, + "learning_rate": 2.584697373429768e-06, + "loss": 1.4712, + "step": 1577 + }, + { + "epoch": 1.62095531587057, + "grad_norm": 2.7786920070648193, + "learning_rate": 2.5827940616673015e-06, + "loss": 1.4995, + "step": 1578 + }, + { + "epoch": 1.6219825372367747, + "grad_norm": 2.782848596572876, + "learning_rate": 2.580890749904835e-06, + "loss": 1.5387, + "step": 1579 + }, + { + "epoch": 1.6230097586029788, + "grad_norm": 2.7152798175811768, + "learning_rate": 2.5789874381423678e-06, + "loss": 1.4813, + "step": 1580 + }, + { + "epoch": 1.6230097586029788, + "eval_loss": 1.8939763307571411, + "eval_runtime": 46.2339, + "eval_samples_per_second": 21.629, + "eval_steps_per_second": 0.908, + "step": 1580 + }, + { + "epoch": 1.6240369799691834, + "grad_norm": 2.6328132152557373, + "learning_rate": 2.577084126379901e-06, + "loss": 1.4801, + "step": 1581 + }, + { + "epoch": 1.6250642013353878, + "grad_norm": 2.6837892532348633, + "learning_rate": 2.5751808146174345e-06, + "loss": 1.453, + "step": 1582 + }, + { + "epoch": 1.6260914227015921, + "grad_norm": 2.6900155544281006, + "learning_rate": 2.573277502854968e-06, + "loss": 1.5044, + "step": 1583 + }, + { + "epoch": 1.6271186440677967, + "grad_norm": 2.6830708980560303, + "learning_rate": 2.5713741910925012e-06, + "loss": 1.4662, + "step": 1584 + }, + { + "epoch": 1.6281458654340009, + "grad_norm": 2.6451282501220703, + "learning_rate": 2.569470879330034e-06, + "loss": 1.4749, + "step": 1585 + }, + { + "epoch": 1.6291730868002054, + "grad_norm": 2.8980915546417236, + "learning_rate": 2.5675675675675675e-06, + "loss": 1.5729, + "step": 1586 + }, + { + "epoch": 1.6302003081664098, + "grad_norm": 2.715513229370117, + "learning_rate": 2.565664255805101e-06, + "loss": 1.4848, + "step": 1587 + }, + { + "epoch": 1.6312275295326142, + "grad_norm": 2.8286707401275635, + "learning_rate": 2.5637609440426343e-06, + "loss": 1.5164, + "step": 1588 + }, + { + "epoch": 1.6322547508988188, + "grad_norm": 2.7614853382110596, + "learning_rate": 2.5618576322801676e-06, + "loss": 1.4623, + "step": 1589 + }, + { + "epoch": 1.6332819722650231, + "grad_norm": 2.826911687850952, + "learning_rate": 2.5599543205177006e-06, + "loss": 1.5593, + "step": 1590 + }, + { + "epoch": 1.6332819722650231, + "eval_loss": 1.8977031707763672, + "eval_runtime": 46.7228, + "eval_samples_per_second": 21.403, + "eval_steps_per_second": 0.899, + "step": 1590 + }, + { + "epoch": 1.6343091936312275, + "grad_norm": 2.707435131072998, + "learning_rate": 2.558051008755234e-06, + "loss": 1.5186, + "step": 1591 + }, + { + "epoch": 1.635336414997432, + "grad_norm": 2.701657295227051, + "learning_rate": 2.5561476969927677e-06, + "loss": 1.5643, + "step": 1592 + }, + { + "epoch": 1.6363636363636362, + "grad_norm": 2.7254910469055176, + "learning_rate": 2.554244385230301e-06, + "loss": 1.4701, + "step": 1593 + }, + { + "epoch": 1.6373908577298408, + "grad_norm": 2.6841659545898438, + "learning_rate": 2.5523410734678344e-06, + "loss": 1.4949, + "step": 1594 + }, + { + "epoch": 1.6384180790960452, + "grad_norm": 2.746089220046997, + "learning_rate": 2.550437761705368e-06, + "loss": 1.435, + "step": 1595 + }, + { + "epoch": 1.6394453004622496, + "grad_norm": 2.6010215282440186, + "learning_rate": 2.548534449942901e-06, + "loss": 1.4594, + "step": 1596 + }, + { + "epoch": 1.6404725218284542, + "grad_norm": 2.9612057209014893, + "learning_rate": 2.546631138180434e-06, + "loss": 1.5084, + "step": 1597 + }, + { + "epoch": 1.6414997431946583, + "grad_norm": 3.105278253555298, + "learning_rate": 2.5447278264179675e-06, + "loss": 1.5041, + "step": 1598 + }, + { + "epoch": 1.642526964560863, + "grad_norm": 2.714082956314087, + "learning_rate": 2.542824514655501e-06, + "loss": 1.462, + "step": 1599 + }, + { + "epoch": 1.6435541859270673, + "grad_norm": 2.8359603881835938, + "learning_rate": 2.540921202893034e-06, + "loss": 1.5545, + "step": 1600 + }, + { + "epoch": 1.6435541859270673, + "eval_loss": 1.9013447761535645, + "eval_runtime": 47.8392, + "eval_samples_per_second": 20.903, + "eval_steps_per_second": 0.878, + "step": 1600 + }, + { + "epoch": 1.6445814072932716, + "grad_norm": 2.8133420944213867, + "learning_rate": 2.5390178911305675e-06, + "loss": 1.5362, + "step": 1601 + }, + { + "epoch": 1.6456086286594762, + "grad_norm": 2.7947328090667725, + "learning_rate": 2.5371145793681005e-06, + "loss": 1.5067, + "step": 1602 + }, + { + "epoch": 1.6466358500256806, + "grad_norm": 2.705209732055664, + "learning_rate": 2.535211267605634e-06, + "loss": 1.4921, + "step": 1603 + }, + { + "epoch": 1.647663071391885, + "grad_norm": 2.7078511714935303, + "learning_rate": 2.533307955843167e-06, + "loss": 1.4398, + "step": 1604 + }, + { + "epoch": 1.6486902927580893, + "grad_norm": 2.749514579772949, + "learning_rate": 2.5314046440807006e-06, + "loss": 1.4931, + "step": 1605 + }, + { + "epoch": 1.6497175141242937, + "grad_norm": 2.703160285949707, + "learning_rate": 2.529501332318234e-06, + "loss": 1.4693, + "step": 1606 + }, + { + "epoch": 1.6507447354904983, + "grad_norm": 2.8278794288635254, + "learning_rate": 2.5275980205557673e-06, + "loss": 1.5353, + "step": 1607 + }, + { + "epoch": 1.6517719568567026, + "grad_norm": 2.927949905395508, + "learning_rate": 2.5256947087933002e-06, + "loss": 1.5665, + "step": 1608 + }, + { + "epoch": 1.652799178222907, + "grad_norm": 2.714650869369507, + "learning_rate": 2.5237913970308336e-06, + "loss": 1.5503, + "step": 1609 + }, + { + "epoch": 1.6538263995891116, + "grad_norm": 2.7319588661193848, + "learning_rate": 2.521888085268367e-06, + "loss": 1.5018, + "step": 1610 + }, + { + "epoch": 1.6538263995891116, + "eval_loss": 1.8987922668457031, + "eval_runtime": 43.4436, + "eval_samples_per_second": 23.018, + "eval_steps_per_second": 0.967, + "step": 1610 + }, + { + "epoch": 1.6548536209553157, + "grad_norm": 2.746079921722412, + "learning_rate": 2.5199847735059003e-06, + "loss": 1.5447, + "step": 1611 + }, + { + "epoch": 1.6558808423215203, + "grad_norm": 2.7670583724975586, + "learning_rate": 2.518081461743434e-06, + "loss": 1.4723, + "step": 1612 + }, + { + "epoch": 1.6569080636877247, + "grad_norm": 2.680903196334839, + "learning_rate": 2.5161781499809675e-06, + "loss": 1.5116, + "step": 1613 + }, + { + "epoch": 1.657935285053929, + "grad_norm": 2.5936760902404785, + "learning_rate": 2.5142748382185004e-06, + "loss": 1.4607, + "step": 1614 + }, + { + "epoch": 1.6589625064201337, + "grad_norm": 2.742476224899292, + "learning_rate": 2.5123715264560338e-06, + "loss": 1.4579, + "step": 1615 + }, + { + "epoch": 1.6599897277863378, + "grad_norm": 2.9192259311676025, + "learning_rate": 2.510468214693567e-06, + "loss": 1.4815, + "step": 1616 + }, + { + "epoch": 1.6610169491525424, + "grad_norm": 2.7400426864624023, + "learning_rate": 2.5085649029311005e-06, + "loss": 1.4268, + "step": 1617 + }, + { + "epoch": 1.6620441705187468, + "grad_norm": 2.6352646350860596, + "learning_rate": 2.506661591168634e-06, + "loss": 1.3921, + "step": 1618 + }, + { + "epoch": 1.6630713918849511, + "grad_norm": 2.6632823944091797, + "learning_rate": 2.504758279406167e-06, + "loss": 1.4754, + "step": 1619 + }, + { + "epoch": 1.6640986132511557, + "grad_norm": 2.8280115127563477, + "learning_rate": 2.5028549676437e-06, + "loss": 1.5208, + "step": 1620 + }, + { + "epoch": 1.6640986132511557, + "eval_loss": 1.8962472677230835, + "eval_runtime": 45.5434, + "eval_samples_per_second": 21.957, + "eval_steps_per_second": 0.922, + "step": 1620 + }, + { + "epoch": 1.66512583461736, + "grad_norm": 2.84045672416687, + "learning_rate": 2.5009516558812335e-06, + "loss": 1.4669, + "step": 1621 + }, + { + "epoch": 1.6661530559835644, + "grad_norm": 2.564673900604248, + "learning_rate": 2.499048344118767e-06, + "loss": 1.5092, + "step": 1622 + }, + { + "epoch": 1.667180277349769, + "grad_norm": 2.622130870819092, + "learning_rate": 2.4971450323563002e-06, + "loss": 1.5606, + "step": 1623 + }, + { + "epoch": 1.6682074987159732, + "grad_norm": 2.5972182750701904, + "learning_rate": 2.4952417205938336e-06, + "loss": 1.4751, + "step": 1624 + }, + { + "epoch": 1.6692347200821778, + "grad_norm": 2.6662817001342773, + "learning_rate": 2.4933384088313666e-06, + "loss": 1.4789, + "step": 1625 + }, + { + "epoch": 1.6702619414483821, + "grad_norm": 2.6823441982269287, + "learning_rate": 2.4914350970689e-06, + "loss": 1.462, + "step": 1626 + }, + { + "epoch": 1.6712891628145865, + "grad_norm": 2.7047946453094482, + "learning_rate": 2.4895317853064337e-06, + "loss": 1.4809, + "step": 1627 + }, + { + "epoch": 1.672316384180791, + "grad_norm": 2.7356603145599365, + "learning_rate": 2.4876284735439666e-06, + "loss": 1.5316, + "step": 1628 + }, + { + "epoch": 1.6733436055469952, + "grad_norm": 2.8151392936706543, + "learning_rate": 2.4857251617815e-06, + "loss": 1.4969, + "step": 1629 + }, + { + "epoch": 1.6743708269131998, + "grad_norm": 2.8269588947296143, + "learning_rate": 2.4838218500190334e-06, + "loss": 1.5041, + "step": 1630 + }, + { + "epoch": 1.6743708269131998, + "eval_loss": 1.9033366441726685, + "eval_runtime": 47.514, + "eval_samples_per_second": 21.046, + "eval_steps_per_second": 0.884, + "step": 1630 + }, + { + "epoch": 1.6753980482794042, + "grad_norm": 2.7226245403289795, + "learning_rate": 2.4819185382565667e-06, + "loss": 1.4924, + "step": 1631 + }, + { + "epoch": 1.6764252696456086, + "grad_norm": 2.6129348278045654, + "learning_rate": 2.4800152264941e-06, + "loss": 1.4829, + "step": 1632 + }, + { + "epoch": 1.6774524910118132, + "grad_norm": 2.618772029876709, + "learning_rate": 2.478111914731633e-06, + "loss": 1.452, + "step": 1633 + }, + { + "epoch": 1.6784797123780173, + "grad_norm": 2.7626497745513916, + "learning_rate": 2.4762086029691664e-06, + "loss": 1.4875, + "step": 1634 + }, + { + "epoch": 1.6795069337442219, + "grad_norm": 3.0622305870056152, + "learning_rate": 2.4743052912066998e-06, + "loss": 1.6208, + "step": 1635 + }, + { + "epoch": 1.6805341551104263, + "grad_norm": 2.6085500717163086, + "learning_rate": 2.472401979444233e-06, + "loss": 1.457, + "step": 1636 + }, + { + "epoch": 1.6815613764766306, + "grad_norm": 2.7662956714630127, + "learning_rate": 2.4704986676817665e-06, + "loss": 1.5572, + "step": 1637 + }, + { + "epoch": 1.6825885978428352, + "grad_norm": 2.8282175064086914, + "learning_rate": 2.4685953559193e-06, + "loss": 1.4922, + "step": 1638 + }, + { + "epoch": 1.6836158192090396, + "grad_norm": 2.930837631225586, + "learning_rate": 2.466692044156833e-06, + "loss": 1.5296, + "step": 1639 + }, + { + "epoch": 1.684643040575244, + "grad_norm": 2.7347614765167236, + "learning_rate": 2.4647887323943666e-06, + "loss": 1.4739, + "step": 1640 + }, + { + "epoch": 1.684643040575244, + "eval_loss": 1.8892933130264282, + "eval_runtime": 48.2635, + "eval_samples_per_second": 20.72, + "eval_steps_per_second": 0.87, + "step": 1640 + }, + { + "epoch": 1.6856702619414485, + "grad_norm": 2.6786420345306396, + "learning_rate": 2.4628854206319e-06, + "loss": 1.4429, + "step": 1641 + }, + { + "epoch": 1.6866974833076527, + "grad_norm": 2.754531145095825, + "learning_rate": 2.460982108869433e-06, + "loss": 1.4375, + "step": 1642 + }, + { + "epoch": 1.6877247046738573, + "grad_norm": 2.8799307346343994, + "learning_rate": 2.4590787971069662e-06, + "loss": 1.5438, + "step": 1643 + }, + { + "epoch": 1.6887519260400616, + "grad_norm": 2.6701910495758057, + "learning_rate": 2.4571754853444996e-06, + "loss": 1.4691, + "step": 1644 + }, + { + "epoch": 1.689779147406266, + "grad_norm": 2.760960340499878, + "learning_rate": 2.455272173582033e-06, + "loss": 1.4607, + "step": 1645 + }, + { + "epoch": 1.6908063687724706, + "grad_norm": 2.7178521156311035, + "learning_rate": 2.4533688618195663e-06, + "loss": 1.466, + "step": 1646 + }, + { + "epoch": 1.6918335901386747, + "grad_norm": 2.7641139030456543, + "learning_rate": 2.4514655500570997e-06, + "loss": 1.4065, + "step": 1647 + }, + { + "epoch": 1.6928608115048793, + "grad_norm": 2.7858457565307617, + "learning_rate": 2.449562238294633e-06, + "loss": 1.4587, + "step": 1648 + }, + { + "epoch": 1.6938880328710837, + "grad_norm": 2.6738100051879883, + "learning_rate": 2.4476589265321664e-06, + "loss": 1.4205, + "step": 1649 + }, + { + "epoch": 1.694915254237288, + "grad_norm": 2.748340129852295, + "learning_rate": 2.4457556147696993e-06, + "loss": 1.5212, + "step": 1650 + }, + { + "epoch": 1.694915254237288, + "eval_loss": 1.9066362380981445, + "eval_runtime": 52.2773, + "eval_samples_per_second": 19.129, + "eval_steps_per_second": 0.803, + "step": 1650 + }, + { + "epoch": 1.6959424756034926, + "grad_norm": 2.7664692401885986, + "learning_rate": 2.4438523030072327e-06, + "loss": 1.5045, + "step": 1651 + }, + { + "epoch": 1.696969696969697, + "grad_norm": 2.7928621768951416, + "learning_rate": 2.441948991244766e-06, + "loss": 1.5134, + "step": 1652 + }, + { + "epoch": 1.6979969183359014, + "grad_norm": 2.7065975666046143, + "learning_rate": 2.4400456794822994e-06, + "loss": 1.5072, + "step": 1653 + }, + { + "epoch": 1.6990241397021058, + "grad_norm": 2.7586593627929688, + "learning_rate": 2.438142367719833e-06, + "loss": 1.5276, + "step": 1654 + }, + { + "epoch": 1.7000513610683101, + "grad_norm": 2.689526081085205, + "learning_rate": 2.4362390559573657e-06, + "loss": 1.4902, + "step": 1655 + }, + { + "epoch": 1.7010785824345147, + "grad_norm": 2.716808795928955, + "learning_rate": 2.434335744194899e-06, + "loss": 1.4772, + "step": 1656 + }, + { + "epoch": 1.702105803800719, + "grad_norm": 2.825169086456299, + "learning_rate": 2.432432432432433e-06, + "loss": 1.5185, + "step": 1657 + }, + { + "epoch": 1.7031330251669234, + "grad_norm": 2.727210283279419, + "learning_rate": 2.4305291206699662e-06, + "loss": 1.4353, + "step": 1658 + }, + { + "epoch": 1.704160246533128, + "grad_norm": 2.753382921218872, + "learning_rate": 2.428625808907499e-06, + "loss": 1.5161, + "step": 1659 + }, + { + "epoch": 1.7051874678993322, + "grad_norm": 2.7188405990600586, + "learning_rate": 2.4267224971450325e-06, + "loss": 1.5324, + "step": 1660 + }, + { + "epoch": 1.7051874678993322, + "eval_loss": 1.8945986032485962, + "eval_runtime": 49.8237, + "eval_samples_per_second": 20.071, + "eval_steps_per_second": 0.843, + "step": 1660 + }, + { + "epoch": 1.7062146892655368, + "grad_norm": 2.702894687652588, + "learning_rate": 2.424819185382566e-06, + "loss": 1.4915, + "step": 1661 + }, + { + "epoch": 1.7072419106317411, + "grad_norm": 2.6755716800689697, + "learning_rate": 2.4229158736200993e-06, + "loss": 1.4471, + "step": 1662 + }, + { + "epoch": 1.7082691319979455, + "grad_norm": 2.637878656387329, + "learning_rate": 2.4210125618576326e-06, + "loss": 1.4488, + "step": 1663 + }, + { + "epoch": 1.70929635336415, + "grad_norm": 2.6714439392089844, + "learning_rate": 2.4191092500951656e-06, + "loss": 1.4708, + "step": 1664 + }, + { + "epoch": 1.7103235747303542, + "grad_norm": 2.703479528427124, + "learning_rate": 2.417205938332699e-06, + "loss": 1.5039, + "step": 1665 + }, + { + "epoch": 1.7113507960965588, + "grad_norm": 2.7005186080932617, + "learning_rate": 2.4153026265702323e-06, + "loss": 1.4242, + "step": 1666 + }, + { + "epoch": 1.7123780174627632, + "grad_norm": 2.7262654304504395, + "learning_rate": 2.4133993148077657e-06, + "loss": 1.4649, + "step": 1667 + }, + { + "epoch": 1.7134052388289676, + "grad_norm": 2.912041425704956, + "learning_rate": 2.411496003045299e-06, + "loss": 1.5247, + "step": 1668 + }, + { + "epoch": 1.7144324601951721, + "grad_norm": 2.713007926940918, + "learning_rate": 2.4095926912828324e-06, + "loss": 1.4585, + "step": 1669 + }, + { + "epoch": 1.7154596815613765, + "grad_norm": 2.6106162071228027, + "learning_rate": 2.4076893795203658e-06, + "loss": 1.4932, + "step": 1670 + }, + { + "epoch": 1.7154596815613765, + "eval_loss": 1.8863319158554077, + "eval_runtime": 45.1847, + "eval_samples_per_second": 22.131, + "eval_steps_per_second": 0.93, + "step": 1670 + }, + { + "epoch": 1.7164869029275809, + "grad_norm": 2.9076762199401855, + "learning_rate": 2.405786067757899e-06, + "loss": 1.4725, + "step": 1671 + }, + { + "epoch": 1.7175141242937855, + "grad_norm": 2.9677484035491943, + "learning_rate": 2.403882755995432e-06, + "loss": 1.551, + "step": 1672 + }, + { + "epoch": 1.7185413456599896, + "grad_norm": 2.6974384784698486, + "learning_rate": 2.4019794442329654e-06, + "loss": 1.4897, + "step": 1673 + }, + { + "epoch": 1.7195685670261942, + "grad_norm": 2.796464443206787, + "learning_rate": 2.4000761324704988e-06, + "loss": 1.4573, + "step": 1674 + }, + { + "epoch": 1.7205957883923986, + "grad_norm": 2.653627872467041, + "learning_rate": 2.398172820708032e-06, + "loss": 1.5025, + "step": 1675 + }, + { + "epoch": 1.721623009758603, + "grad_norm": 2.716545343399048, + "learning_rate": 2.3962695089455655e-06, + "loss": 1.4687, + "step": 1676 + }, + { + "epoch": 1.7226502311248075, + "grad_norm": 2.737752676010132, + "learning_rate": 2.3943661971830984e-06, + "loss": 1.4515, + "step": 1677 + }, + { + "epoch": 1.7236774524910117, + "grad_norm": 2.7596747875213623, + "learning_rate": 2.3924628854206322e-06, + "loss": 1.4712, + "step": 1678 + }, + { + "epoch": 1.7247046738572163, + "grad_norm": 2.9202370643615723, + "learning_rate": 2.3905595736581656e-06, + "loss": 1.4818, + "step": 1679 + }, + { + "epoch": 1.7257318952234206, + "grad_norm": 2.7129971981048584, + "learning_rate": 2.388656261895699e-06, + "loss": 1.5269, + "step": 1680 + }, + { + "epoch": 1.7257318952234206, + "eval_loss": 1.8941866159439087, + "eval_runtime": 45.4463, + "eval_samples_per_second": 22.004, + "eval_steps_per_second": 0.924, + "step": 1680 + }, + { + "epoch": 1.726759116589625, + "grad_norm": 2.956148862838745, + "learning_rate": 2.386752950133232e-06, + "loss": 1.4977, + "step": 1681 + }, + { + "epoch": 1.7277863379558296, + "grad_norm": 2.8728253841400146, + "learning_rate": 2.3848496383707653e-06, + "loss": 1.4961, + "step": 1682 + }, + { + "epoch": 1.7288135593220337, + "grad_norm": 2.7839162349700928, + "learning_rate": 2.3829463266082986e-06, + "loss": 1.4757, + "step": 1683 + }, + { + "epoch": 1.7298407806882383, + "grad_norm": 2.7487802505493164, + "learning_rate": 2.381043014845832e-06, + "loss": 1.5048, + "step": 1684 + }, + { + "epoch": 1.7308680020544427, + "grad_norm": 2.80683970451355, + "learning_rate": 2.3791397030833653e-06, + "loss": 1.4916, + "step": 1685 + }, + { + "epoch": 1.731895223420647, + "grad_norm": 2.6910040378570557, + "learning_rate": 2.3772363913208983e-06, + "loss": 1.4711, + "step": 1686 + }, + { + "epoch": 1.7329224447868516, + "grad_norm": 2.594284772872925, + "learning_rate": 2.3753330795584316e-06, + "loss": 1.4558, + "step": 1687 + }, + { + "epoch": 1.733949666153056, + "grad_norm": 2.7420263290405273, + "learning_rate": 2.3734297677959654e-06, + "loss": 1.4769, + "step": 1688 + }, + { + "epoch": 1.7349768875192604, + "grad_norm": 2.80002760887146, + "learning_rate": 2.3715264560334984e-06, + "loss": 1.5289, + "step": 1689 + }, + { + "epoch": 1.736004108885465, + "grad_norm": 2.799618721008301, + "learning_rate": 2.3696231442710317e-06, + "loss": 1.5131, + "step": 1690 + }, + { + "epoch": 1.736004108885465, + "eval_loss": 1.8925220966339111, + "eval_runtime": 45.89, + "eval_samples_per_second": 21.791, + "eval_steps_per_second": 0.915, + "step": 1690 + }, + { + "epoch": 1.7370313302516691, + "grad_norm": 2.701648712158203, + "learning_rate": 2.367719832508565e-06, + "loss": 1.5076, + "step": 1691 + }, + { + "epoch": 1.7380585516178737, + "grad_norm": 2.6925787925720215, + "learning_rate": 2.3658165207460985e-06, + "loss": 1.5331, + "step": 1692 + }, + { + "epoch": 1.739085772984078, + "grad_norm": 2.8448283672332764, + "learning_rate": 2.363913208983632e-06, + "loss": 1.4933, + "step": 1693 + }, + { + "epoch": 1.7401129943502824, + "grad_norm": 2.8252811431884766, + "learning_rate": 2.362009897221165e-06, + "loss": 1.5205, + "step": 1694 + }, + { + "epoch": 1.741140215716487, + "grad_norm": 2.675318956375122, + "learning_rate": 2.360106585458698e-06, + "loss": 1.4958, + "step": 1695 + }, + { + "epoch": 1.7421674370826912, + "grad_norm": 2.8392488956451416, + "learning_rate": 2.3582032736962315e-06, + "loss": 1.4522, + "step": 1696 + }, + { + "epoch": 1.7431946584488958, + "grad_norm": 2.8221116065979004, + "learning_rate": 2.356299961933765e-06, + "loss": 1.4924, + "step": 1697 + }, + { + "epoch": 1.7442218798151001, + "grad_norm": 2.685828447341919, + "learning_rate": 2.354396650171298e-06, + "loss": 1.4668, + "step": 1698 + }, + { + "epoch": 1.7452491011813045, + "grad_norm": 2.622481107711792, + "learning_rate": 2.3524933384088316e-06, + "loss": 1.451, + "step": 1699 + }, + { + "epoch": 1.746276322547509, + "grad_norm": 2.79746413230896, + "learning_rate": 2.350590026646365e-06, + "loss": 1.5473, + "step": 1700 + }, + { + "epoch": 1.746276322547509, + "eval_loss": 1.898282766342163, + "eval_runtime": 45.0673, + "eval_samples_per_second": 22.189, + "eval_steps_per_second": 0.932, + "step": 1700 + }, + { + "epoch": 1.7473035439137135, + "grad_norm": 2.8120598793029785, + "learning_rate": 2.3486867148838983e-06, + "loss": 1.4692, + "step": 1701 + }, + { + "epoch": 1.7483307652799178, + "grad_norm": 2.7376883029937744, + "learning_rate": 2.3467834031214317e-06, + "loss": 1.4749, + "step": 1702 + }, + { + "epoch": 1.7493579866461222, + "grad_norm": 2.7579424381256104, + "learning_rate": 2.3448800913589646e-06, + "loss": 1.4726, + "step": 1703 + }, + { + "epoch": 1.7503852080123266, + "grad_norm": 2.648026466369629, + "learning_rate": 2.342976779596498e-06, + "loss": 1.4938, + "step": 1704 + }, + { + "epoch": 1.7514124293785311, + "grad_norm": 2.6477997303009033, + "learning_rate": 2.3410734678340313e-06, + "loss": 1.4813, + "step": 1705 + }, + { + "epoch": 1.7524396507447355, + "grad_norm": 2.742249011993408, + "learning_rate": 2.3391701560715647e-06, + "loss": 1.5257, + "step": 1706 + }, + { + "epoch": 1.7534668721109399, + "grad_norm": 2.6430649757385254, + "learning_rate": 2.337266844309098e-06, + "loss": 1.4781, + "step": 1707 + }, + { + "epoch": 1.7544940934771445, + "grad_norm": 2.6746671199798584, + "learning_rate": 2.3353635325466314e-06, + "loss": 1.4807, + "step": 1708 + }, + { + "epoch": 1.7555213148433486, + "grad_norm": 2.8837218284606934, + "learning_rate": 2.3334602207841648e-06, + "loss": 1.4932, + "step": 1709 + }, + { + "epoch": 1.7565485362095532, + "grad_norm": 2.703761577606201, + "learning_rate": 2.331556909021698e-06, + "loss": 1.4969, + "step": 1710 + }, + { + "epoch": 1.7565485362095532, + "eval_loss": 1.8966599702835083, + "eval_runtime": 42.6325, + "eval_samples_per_second": 23.456, + "eval_steps_per_second": 0.985, + "step": 1710 + }, + { + "epoch": 1.7575757575757576, + "grad_norm": 2.760502338409424, + "learning_rate": 2.3296535972592315e-06, + "loss": 1.5148, + "step": 1711 + }, + { + "epoch": 1.758602978941962, + "grad_norm": 2.696356773376465, + "learning_rate": 2.3277502854967644e-06, + "loss": 1.4604, + "step": 1712 + }, + { + "epoch": 1.7596302003081665, + "grad_norm": 2.7124855518341064, + "learning_rate": 2.325846973734298e-06, + "loss": 1.5178, + "step": 1713 + }, + { + "epoch": 1.7606574216743707, + "grad_norm": 2.624866008758545, + "learning_rate": 2.323943661971831e-06, + "loss": 1.4786, + "step": 1714 + }, + { + "epoch": 1.7616846430405753, + "grad_norm": 2.785651922225952, + "learning_rate": 2.3220403502093645e-06, + "loss": 1.5077, + "step": 1715 + }, + { + "epoch": 1.7627118644067796, + "grad_norm": 2.7688722610473633, + "learning_rate": 2.320137038446898e-06, + "loss": 1.5299, + "step": 1716 + }, + { + "epoch": 1.763739085772984, + "grad_norm": 2.72312593460083, + "learning_rate": 2.318233726684431e-06, + "loss": 1.5352, + "step": 1717 + }, + { + "epoch": 1.7647663071391886, + "grad_norm": 2.7877933979034424, + "learning_rate": 2.3163304149219646e-06, + "loss": 1.5148, + "step": 1718 + }, + { + "epoch": 1.765793528505393, + "grad_norm": 2.6651675701141357, + "learning_rate": 2.314427103159498e-06, + "loss": 1.4979, + "step": 1719 + }, + { + "epoch": 1.7668207498715973, + "grad_norm": 2.747264862060547, + "learning_rate": 2.312523791397031e-06, + "loss": 1.4785, + "step": 1720 + }, + { + "epoch": 1.7668207498715973, + "eval_loss": 1.9028254747390747, + "eval_runtime": 47.0728, + "eval_samples_per_second": 21.244, + "eval_steps_per_second": 0.892, + "step": 1720 + }, + { + "epoch": 1.767847971237802, + "grad_norm": 2.6331684589385986, + "learning_rate": 2.3106204796345643e-06, + "loss": 1.4479, + "step": 1721 + }, + { + "epoch": 1.768875192604006, + "grad_norm": 2.662079095840454, + "learning_rate": 2.3087171678720976e-06, + "loss": 1.4827, + "step": 1722 + }, + { + "epoch": 1.7699024139702106, + "grad_norm": 2.735499382019043, + "learning_rate": 2.306813856109631e-06, + "loss": 1.5324, + "step": 1723 + }, + { + "epoch": 1.770929635336415, + "grad_norm": 2.66086745262146, + "learning_rate": 2.3049105443471644e-06, + "loss": 1.5002, + "step": 1724 + }, + { + "epoch": 1.7719568567026194, + "grad_norm": 2.7453548908233643, + "learning_rate": 2.3030072325846973e-06, + "loss": 1.48, + "step": 1725 + }, + { + "epoch": 1.772984078068824, + "grad_norm": 2.701510429382324, + "learning_rate": 2.3011039208222307e-06, + "loss": 1.4642, + "step": 1726 + }, + { + "epoch": 1.774011299435028, + "grad_norm": 2.6908695697784424, + "learning_rate": 2.299200609059764e-06, + "loss": 1.4607, + "step": 1727 + }, + { + "epoch": 1.7750385208012327, + "grad_norm": 2.7793400287628174, + "learning_rate": 2.297297297297298e-06, + "loss": 1.4756, + "step": 1728 + }, + { + "epoch": 1.776065742167437, + "grad_norm": 2.6336896419525146, + "learning_rate": 2.2953939855348308e-06, + "loss": 1.4336, + "step": 1729 + }, + { + "epoch": 1.7770929635336414, + "grad_norm": 2.82281231880188, + "learning_rate": 2.293490673772364e-06, + "loss": 1.5273, + "step": 1730 + }, + { + "epoch": 1.7770929635336414, + "eval_loss": 1.9012898206710815, + "eval_runtime": 46.5443, + "eval_samples_per_second": 21.485, + "eval_steps_per_second": 0.902, + "step": 1730 + }, + { + "epoch": 1.778120184899846, + "grad_norm": 2.7541351318359375, + "learning_rate": 2.2915873620098975e-06, + "loss": 1.5071, + "step": 1731 + }, + { + "epoch": 1.7791474062660502, + "grad_norm": 2.771777629852295, + "learning_rate": 2.289684050247431e-06, + "loss": 1.503, + "step": 1732 + }, + { + "epoch": 1.7801746276322548, + "grad_norm": 2.5365493297576904, + "learning_rate": 2.287780738484964e-06, + "loss": 1.4635, + "step": 1733 + }, + { + "epoch": 1.7812018489984591, + "grad_norm": 2.691694498062134, + "learning_rate": 2.285877426722497e-06, + "loss": 1.5027, + "step": 1734 + }, + { + "epoch": 1.7822290703646635, + "grad_norm": 2.782968282699585, + "learning_rate": 2.2839741149600305e-06, + "loss": 1.5204, + "step": 1735 + }, + { + "epoch": 1.783256291730868, + "grad_norm": 2.6389904022216797, + "learning_rate": 2.282070803197564e-06, + "loss": 1.5135, + "step": 1736 + }, + { + "epoch": 1.7842835130970724, + "grad_norm": 2.8609848022460938, + "learning_rate": 2.2801674914350972e-06, + "loss": 1.5013, + "step": 1737 + }, + { + "epoch": 1.7853107344632768, + "grad_norm": 2.6814842224121094, + "learning_rate": 2.2782641796726306e-06, + "loss": 1.4874, + "step": 1738 + }, + { + "epoch": 1.7863379558294814, + "grad_norm": 2.7633700370788574, + "learning_rate": 2.276360867910164e-06, + "loss": 1.5129, + "step": 1739 + }, + { + "epoch": 1.7873651771956856, + "grad_norm": 2.637755870819092, + "learning_rate": 2.2744575561476973e-06, + "loss": 1.4486, + "step": 1740 + }, + { + "epoch": 1.7873651771956856, + "eval_loss": 1.8927199840545654, + "eval_runtime": 46.5881, + "eval_samples_per_second": 21.465, + "eval_steps_per_second": 0.902, + "step": 1740 + }, + { + "epoch": 1.7883923985618901, + "grad_norm": 2.815883159637451, + "learning_rate": 2.2725542443852307e-06, + "loss": 1.4815, + "step": 1741 + }, + { + "epoch": 1.7894196199280945, + "grad_norm": 2.758363723754883, + "learning_rate": 2.2706509326227636e-06, + "loss": 1.5235, + "step": 1742 + }, + { + "epoch": 1.7904468412942989, + "grad_norm": 2.6607120037078857, + "learning_rate": 2.268747620860297e-06, + "loss": 1.503, + "step": 1743 + }, + { + "epoch": 1.7914740626605035, + "grad_norm": 2.7479255199432373, + "learning_rate": 2.2668443090978303e-06, + "loss": 1.503, + "step": 1744 + }, + { + "epoch": 1.7925012840267076, + "grad_norm": 2.746718168258667, + "learning_rate": 2.2649409973353637e-06, + "loss": 1.5142, + "step": 1745 + }, + { + "epoch": 1.7935285053929122, + "grad_norm": 2.6848037242889404, + "learning_rate": 2.263037685572897e-06, + "loss": 1.4585, + "step": 1746 + }, + { + "epoch": 1.7945557267591166, + "grad_norm": 2.6677210330963135, + "learning_rate": 2.26113437381043e-06, + "loss": 1.4617, + "step": 1747 + }, + { + "epoch": 1.795582948125321, + "grad_norm": 2.866431474685669, + "learning_rate": 2.259231062047964e-06, + "loss": 1.4945, + "step": 1748 + }, + { + "epoch": 1.7966101694915255, + "grad_norm": 2.953123092651367, + "learning_rate": 2.257327750285497e-06, + "loss": 1.5225, + "step": 1749 + }, + { + "epoch": 1.7976373908577299, + "grad_norm": 2.6214499473571777, + "learning_rate": 2.2554244385230305e-06, + "loss": 1.4707, + "step": 1750 + }, + { + "epoch": 1.7976373908577299, + "eval_loss": 1.8950202465057373, + "eval_runtime": 44.7553, + "eval_samples_per_second": 22.344, + "eval_steps_per_second": 0.938, + "step": 1750 + }, + { + "epoch": 1.7986646122239343, + "grad_norm": 2.7714295387268066, + "learning_rate": 2.2535211267605635e-06, + "loss": 1.4788, + "step": 1751 + }, + { + "epoch": 1.7996918335901386, + "grad_norm": 2.840985059738159, + "learning_rate": 2.251617814998097e-06, + "loss": 1.5004, + "step": 1752 + }, + { + "epoch": 1.800719054956343, + "grad_norm": 2.6215717792510986, + "learning_rate": 2.24971450323563e-06, + "loss": 1.4063, + "step": 1753 + }, + { + "epoch": 1.8017462763225476, + "grad_norm": 2.7458207607269287, + "learning_rate": 2.2478111914731636e-06, + "loss": 1.5045, + "step": 1754 + }, + { + "epoch": 1.802773497688752, + "grad_norm": 2.5622992515563965, + "learning_rate": 2.245907879710697e-06, + "loss": 1.3671, + "step": 1755 + }, + { + "epoch": 1.8038007190549563, + "grad_norm": 3.09394907951355, + "learning_rate": 2.24400456794823e-06, + "loss": 1.5151, + "step": 1756 + }, + { + "epoch": 1.804827940421161, + "grad_norm": 2.740785837173462, + "learning_rate": 2.2421012561857632e-06, + "loss": 1.4635, + "step": 1757 + }, + { + "epoch": 1.805855161787365, + "grad_norm": 2.8012168407440186, + "learning_rate": 2.240197944423297e-06, + "loss": 1.4694, + "step": 1758 + }, + { + "epoch": 1.8068823831535696, + "grad_norm": 2.799586534500122, + "learning_rate": 2.23829463266083e-06, + "loss": 1.4627, + "step": 1759 + }, + { + "epoch": 1.807909604519774, + "grad_norm": 2.6345391273498535, + "learning_rate": 2.2363913208983633e-06, + "loss": 1.4498, + "step": 1760 + }, + { + "epoch": 1.807909604519774, + "eval_loss": 1.9042582511901855, + "eval_runtime": 47.4919, + "eval_samples_per_second": 21.056, + "eval_steps_per_second": 0.884, + "step": 1760 + }, + { + "epoch": 1.8089368258859784, + "grad_norm": 2.715463638305664, + "learning_rate": 2.2344880091358967e-06, + "loss": 1.493, + "step": 1761 + }, + { + "epoch": 1.809964047252183, + "grad_norm": 2.6666505336761475, + "learning_rate": 2.23258469737343e-06, + "loss": 1.4678, + "step": 1762 + }, + { + "epoch": 1.810991268618387, + "grad_norm": 2.6605517864227295, + "learning_rate": 2.2306813856109634e-06, + "loss": 1.5154, + "step": 1763 + }, + { + "epoch": 1.8120184899845917, + "grad_norm": 2.621842384338379, + "learning_rate": 2.2287780738484963e-06, + "loss": 1.4626, + "step": 1764 + }, + { + "epoch": 1.813045711350796, + "grad_norm": 2.7204320430755615, + "learning_rate": 2.2268747620860297e-06, + "loss": 1.5461, + "step": 1765 + }, + { + "epoch": 1.8140729327170004, + "grad_norm": 2.8174679279327393, + "learning_rate": 2.224971450323563e-06, + "loss": 1.4881, + "step": 1766 + }, + { + "epoch": 1.815100154083205, + "grad_norm": 2.707890272140503, + "learning_rate": 2.2230681385610964e-06, + "loss": 1.4123, + "step": 1767 + }, + { + "epoch": 1.8161273754494094, + "grad_norm": 2.753523111343384, + "learning_rate": 2.2211648267986298e-06, + "loss": 1.4783, + "step": 1768 + }, + { + "epoch": 1.8171545968156138, + "grad_norm": 2.830095052719116, + "learning_rate": 2.219261515036163e-06, + "loss": 1.4536, + "step": 1769 + }, + { + "epoch": 1.8181818181818183, + "grad_norm": 2.803318738937378, + "learning_rate": 2.2173582032736965e-06, + "loss": 1.4939, + "step": 1770 + }, + { + "epoch": 1.8181818181818183, + "eval_loss": 1.899889349937439, + "eval_runtime": 46.7673, + "eval_samples_per_second": 21.382, + "eval_steps_per_second": 0.898, + "step": 1770 + }, + { + "epoch": 1.8192090395480225, + "grad_norm": 2.8235929012298584, + "learning_rate": 2.21545489151123e-06, + "loss": 1.5176, + "step": 1771 + }, + { + "epoch": 1.820236260914227, + "grad_norm": 2.761497974395752, + "learning_rate": 2.2135515797487632e-06, + "loss": 1.4685, + "step": 1772 + }, + { + "epoch": 1.8212634822804314, + "grad_norm": 2.589054822921753, + "learning_rate": 2.211648267986296e-06, + "loss": 1.4673, + "step": 1773 + }, + { + "epoch": 1.8222907036466358, + "grad_norm": 2.763709306716919, + "learning_rate": 2.2097449562238295e-06, + "loss": 1.4727, + "step": 1774 + }, + { + "epoch": 1.8233179250128404, + "grad_norm": 2.7250869274139404, + "learning_rate": 2.207841644461363e-06, + "loss": 1.4729, + "step": 1775 + }, + { + "epoch": 1.8243451463790445, + "grad_norm": 2.7706706523895264, + "learning_rate": 2.2059383326988963e-06, + "loss": 1.5382, + "step": 1776 + }, + { + "epoch": 1.8253723677452491, + "grad_norm": 2.9034879207611084, + "learning_rate": 2.2040350209364296e-06, + "loss": 1.4936, + "step": 1777 + }, + { + "epoch": 1.8263995891114535, + "grad_norm": 2.625908851623535, + "learning_rate": 2.2021317091739626e-06, + "loss": 1.4601, + "step": 1778 + }, + { + "epoch": 1.8274268104776579, + "grad_norm": 2.703167676925659, + "learning_rate": 2.2002283974114963e-06, + "loss": 1.4876, + "step": 1779 + }, + { + "epoch": 1.8284540318438625, + "grad_norm": 2.8047842979431152, + "learning_rate": 2.1983250856490297e-06, + "loss": 1.428, + "step": 1780 + }, + { + "epoch": 1.8284540318438625, + "eval_loss": 1.8968960046768188, + "eval_runtime": 46.5826, + "eval_samples_per_second": 21.467, + "eval_steps_per_second": 0.902, + "step": 1780 + }, + { + "epoch": 1.8294812532100666, + "grad_norm": 2.8730416297912598, + "learning_rate": 2.1964217738865626e-06, + "loss": 1.4697, + "step": 1781 + }, + { + "epoch": 1.8305084745762712, + "grad_norm": 2.958203077316284, + "learning_rate": 2.194518462124096e-06, + "loss": 1.4955, + "step": 1782 + }, + { + "epoch": 1.8315356959424756, + "grad_norm": 2.8280398845672607, + "learning_rate": 2.1926151503616294e-06, + "loss": 1.5391, + "step": 1783 + }, + { + "epoch": 1.83256291730868, + "grad_norm": 2.776029348373413, + "learning_rate": 2.1907118385991627e-06, + "loss": 1.4945, + "step": 1784 + }, + { + "epoch": 1.8335901386748845, + "grad_norm": 2.7049412727355957, + "learning_rate": 2.188808526836696e-06, + "loss": 1.4481, + "step": 1785 + }, + { + "epoch": 1.8346173600410889, + "grad_norm": 2.71523118019104, + "learning_rate": 2.1869052150742295e-06, + "loss": 1.4756, + "step": 1786 + }, + { + "epoch": 1.8356445814072933, + "grad_norm": 2.7659823894500732, + "learning_rate": 2.1850019033117624e-06, + "loss": 1.4065, + "step": 1787 + }, + { + "epoch": 1.8366718027734978, + "grad_norm": 2.679553508758545, + "learning_rate": 2.1830985915492958e-06, + "loss": 1.4744, + "step": 1788 + }, + { + "epoch": 1.837699024139702, + "grad_norm": 2.6452908515930176, + "learning_rate": 2.1811952797868295e-06, + "loss": 1.4922, + "step": 1789 + }, + { + "epoch": 1.8387262455059066, + "grad_norm": 2.780548095703125, + "learning_rate": 2.1792919680243625e-06, + "loss": 1.458, + "step": 1790 + }, + { + "epoch": 1.8387262455059066, + "eval_loss": 1.8877090215682983, + "eval_runtime": 47.6844, + "eval_samples_per_second": 20.971, + "eval_steps_per_second": 0.881, + "step": 1790 + }, + { + "epoch": 1.839753466872111, + "grad_norm": 2.8762049674987793, + "learning_rate": 2.177388656261896e-06, + "loss": 1.489, + "step": 1791 + }, + { + "epoch": 1.8407806882383153, + "grad_norm": 3.0053279399871826, + "learning_rate": 2.1754853444994292e-06, + "loss": 1.5681, + "step": 1792 + }, + { + "epoch": 1.84180790960452, + "grad_norm": 2.737626791000366, + "learning_rate": 2.1735820327369626e-06, + "loss": 1.3764, + "step": 1793 + }, + { + "epoch": 1.842835130970724, + "grad_norm": 2.6998884677886963, + "learning_rate": 2.171678720974496e-06, + "loss": 1.5099, + "step": 1794 + }, + { + "epoch": 1.8438623523369286, + "grad_norm": 2.8110108375549316, + "learning_rate": 2.169775409212029e-06, + "loss": 1.5079, + "step": 1795 + }, + { + "epoch": 1.844889573703133, + "grad_norm": 2.7312750816345215, + "learning_rate": 2.1678720974495622e-06, + "loss": 1.5136, + "step": 1796 + }, + { + "epoch": 1.8459167950693374, + "grad_norm": 2.692768096923828, + "learning_rate": 2.1659687856870956e-06, + "loss": 1.5005, + "step": 1797 + }, + { + "epoch": 1.846944016435542, + "grad_norm": 2.673879623413086, + "learning_rate": 2.164065473924629e-06, + "loss": 1.5127, + "step": 1798 + }, + { + "epoch": 1.8479712378017463, + "grad_norm": 2.6218647956848145, + "learning_rate": 2.1621621621621623e-06, + "loss": 1.4529, + "step": 1799 + }, + { + "epoch": 1.8489984591679507, + "grad_norm": 2.7149012088775635, + "learning_rate": 2.1602588503996957e-06, + "loss": 1.5048, + "step": 1800 + }, + { + "epoch": 1.8489984591679507, + "eval_loss": 1.883215069770813, + "eval_runtime": 50.8526, + "eval_samples_per_second": 19.665, + "eval_steps_per_second": 0.826, + "step": 1800 + } + ], + "logging_steps": 1, + "max_steps": 2919, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 100, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 5.611156936851456e+17, + "train_batch_size": 3, + "trial_name": null, + "trial_params": null +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/zero_to_fp32.py b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-1800/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fb0f9a53a507ffcdb60410deb71d60ef801bf350 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/config.json @@ -0,0 +1,36 @@ +{ + "_name_or_path": "meta-llama/Llama-3.2-3B", + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "eos_token_id": 128001, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 3072, + "initializer_range": 0.02, + "intermediate_size": 8192, + "max_position_embeddings": 131072, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 24, + "num_hidden_layers": 28, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": { + "factor": 32.0, + "high_freq_factor": 4.0, + "low_freq_factor": 1.0, + "original_max_position_embeddings": 8192, + "rope_type": "llama3" + }, + "rope_theta": 500000.0, + "tie_word_embeddings": true, + "torch_dtype": "float16", + "transformers_version": "4.45.1", + "use_cache": true, + "vocab_size": 128256 +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/generation_config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..9c389b841a9a29ddd904e02d1eb0e08dcce82ad9 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/generation_config.json @@ -0,0 +1,9 @@ +{ + "_from_model_config": true, + "bos_token_id": 128000, + "do_sample": true, + "eos_token_id": 128001, + "temperature": 0.6, + "top_p": 0.9, + "transformers_version": "4.45.1" +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/latest b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/latest new file mode 100644 index 0000000000000000000000000000000000000000..753e24e10f3a2489150f458205cf759fd8b6081f --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/latest @@ -0,0 +1 @@ +global_step200 \ No newline at end of file diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/model.safetensors.index.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..ed64de846d720b9a7859dc20575fea8e8ca51940 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/model.safetensors.index.json @@ -0,0 +1,262 @@ +{ + "metadata": { + "total_size": 7213504512 + }, + "weight_map": { + "lm_head.weight": "model-00002-of-00002.safetensors", + "model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.norm.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/special_tokens_map.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..6949c5975ee0e961ef61cf31010dce04df0a03f8 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/special_tokens_map.json @@ -0,0 +1,23 @@ +{ + "bos_token": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/tokenizer_config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..0b782d4602b361f7577abb52b710a8c0e088588d --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/tokenizer_config.json @@ -0,0 +1,2078 @@ +{ + "added_tokens_decoder": { + "128000": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128001": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128002": { + "content": "<|reserved_special_token_0|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128003": { + "content": "<|reserved_special_token_1|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128004": { + "content": "<|finetune_right_pad_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128005": { + "content": "<|reserved_special_token_2|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128006": { + "content": "<|start_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128007": { + "content": "<|end_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128008": { + "content": "<|eom_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128009": { + "content": "<|eot_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128010": { + "content": "<|python_tag|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128011": { + "content": "<|reserved_special_token_3|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128012": { + "content": "<|reserved_special_token_4|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128013": { + "content": "<|reserved_special_token_5|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128014": { + "content": "<|reserved_special_token_6|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128015": { + "content": "<|reserved_special_token_7|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128016": { + "content": "<|reserved_special_token_8|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128017": { + "content": "<|reserved_special_token_9|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128018": { + "content": "<|reserved_special_token_10|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128019": { + "content": "<|reserved_special_token_11|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128020": { + "content": "<|reserved_special_token_12|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128021": { + "content": "<|reserved_special_token_13|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128022": { + "content": "<|reserved_special_token_14|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128023": { + "content": "<|reserved_special_token_15|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128024": { + "content": "<|reserved_special_token_16|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128025": { + "content": "<|reserved_special_token_17|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128026": { + "content": "<|reserved_special_token_18|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128027": { + "content": "<|reserved_special_token_19|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128028": { + "content": "<|reserved_special_token_20|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128029": { + "content": "<|reserved_special_token_21|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128030": { + "content": "<|reserved_special_token_22|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128031": { + "content": "<|reserved_special_token_23|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128032": { + "content": "<|reserved_special_token_24|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128033": { + "content": "<|reserved_special_token_25|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128034": { + "content": "<|reserved_special_token_26|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128035": { + "content": "<|reserved_special_token_27|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128036": { + "content": "<|reserved_special_token_28|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128037": { + "content": "<|reserved_special_token_29|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128038": { + "content": "<|reserved_special_token_30|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128039": { + "content": "<|reserved_special_token_31|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128040": { + "content": "<|reserved_special_token_32|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128041": { + "content": "<|reserved_special_token_33|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128042": { + "content": "<|reserved_special_token_34|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128043": { + "content": "<|reserved_special_token_35|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128044": { + "content": "<|reserved_special_token_36|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128045": { + "content": "<|reserved_special_token_37|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128046": { + "content": "<|reserved_special_token_38|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128047": { + "content": "<|reserved_special_token_39|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128048": { + "content": "<|reserved_special_token_40|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128049": { + "content": "<|reserved_special_token_41|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128050": { + "content": "<|reserved_special_token_42|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128051": { + "content": "<|reserved_special_token_43|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128052": { + "content": "<|reserved_special_token_44|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128053": { + "content": "<|reserved_special_token_45|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128054": { + "content": "<|reserved_special_token_46|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128055": { + "content": "<|reserved_special_token_47|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128056": { + "content": "<|reserved_special_token_48|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128057": { + "content": "<|reserved_special_token_49|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128058": { + "content": "<|reserved_special_token_50|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128059": { + "content": "<|reserved_special_token_51|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128060": { + "content": "<|reserved_special_token_52|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128061": { + "content": "<|reserved_special_token_53|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128062": { + "content": "<|reserved_special_token_54|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128063": { + "content": "<|reserved_special_token_55|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128064": { + "content": "<|reserved_special_token_56|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128065": { + "content": "<|reserved_special_token_57|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128066": { + "content": "<|reserved_special_token_58|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128067": { + "content": "<|reserved_special_token_59|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128068": { + "content": "<|reserved_special_token_60|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128069": { + "content": "<|reserved_special_token_61|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128070": { + "content": "<|reserved_special_token_62|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128071": { + "content": "<|reserved_special_token_63|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128072": { + "content": "<|reserved_special_token_64|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128073": { + "content": "<|reserved_special_token_65|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128074": { + "content": "<|reserved_special_token_66|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128075": { + "content": "<|reserved_special_token_67|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128076": { + "content": "<|reserved_special_token_68|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128077": { + "content": "<|reserved_special_token_69|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128078": { + "content": "<|reserved_special_token_70|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128079": { + "content": "<|reserved_special_token_71|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128080": { + "content": "<|reserved_special_token_72|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128081": { + "content": "<|reserved_special_token_73|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128082": { + "content": "<|reserved_special_token_74|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128083": { + "content": "<|reserved_special_token_75|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128084": { + "content": "<|reserved_special_token_76|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128085": { + "content": "<|reserved_special_token_77|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128086": { + "content": "<|reserved_special_token_78|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128087": { + "content": "<|reserved_special_token_79|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128088": { + "content": "<|reserved_special_token_80|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128089": { + "content": "<|reserved_special_token_81|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128090": { + "content": "<|reserved_special_token_82|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128091": { + "content": "<|reserved_special_token_83|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128092": { + "content": "<|reserved_special_token_84|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128093": { + "content": "<|reserved_special_token_85|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128094": { + "content": "<|reserved_special_token_86|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128095": { + "content": "<|reserved_special_token_87|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128096": { + "content": "<|reserved_special_token_88|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128097": { + "content": "<|reserved_special_token_89|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128098": { + "content": "<|reserved_special_token_90|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128099": { + "content": "<|reserved_special_token_91|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128100": { + "content": "<|reserved_special_token_92|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128101": { + "content": "<|reserved_special_token_93|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128102": { + "content": "<|reserved_special_token_94|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128103": { + "content": "<|reserved_special_token_95|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128104": { + "content": "<|reserved_special_token_96|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128105": { + "content": "<|reserved_special_token_97|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128106": { + "content": "<|reserved_special_token_98|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128107": { + "content": "<|reserved_special_token_99|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128108": { + "content": "<|reserved_special_token_100|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128109": { + "content": "<|reserved_special_token_101|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128110": { + "content": "<|reserved_special_token_102|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128111": { + "content": "<|reserved_special_token_103|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128112": { + "content": "<|reserved_special_token_104|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128113": { + "content": "<|reserved_special_token_105|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128114": { + "content": "<|reserved_special_token_106|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128115": { + "content": "<|reserved_special_token_107|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128116": { + "content": "<|reserved_special_token_108|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128117": { + "content": "<|reserved_special_token_109|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128118": { + "content": "<|reserved_special_token_110|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128119": { + "content": "<|reserved_special_token_111|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128120": { + "content": "<|reserved_special_token_112|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128121": { + "content": "<|reserved_special_token_113|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128122": { + "content": "<|reserved_special_token_114|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128123": { + "content": "<|reserved_special_token_115|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128124": { + "content": "<|reserved_special_token_116|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128125": { + "content": "<|reserved_special_token_117|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128126": { + "content": "<|reserved_special_token_118|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128127": { + "content": "<|reserved_special_token_119|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128128": { + "content": "<|reserved_special_token_120|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128129": { + "content": "<|reserved_special_token_121|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128130": { + "content": "<|reserved_special_token_122|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128131": { + "content": "<|reserved_special_token_123|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128132": { + "content": "<|reserved_special_token_124|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128133": { + "content": "<|reserved_special_token_125|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128134": { + "content": "<|reserved_special_token_126|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128135": { + "content": "<|reserved_special_token_127|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128136": { + "content": "<|reserved_special_token_128|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128137": { + "content": "<|reserved_special_token_129|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128138": { + "content": "<|reserved_special_token_130|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128139": { + "content": "<|reserved_special_token_131|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128140": { + "content": "<|reserved_special_token_132|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128141": { + "content": "<|reserved_special_token_133|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128142": { + "content": "<|reserved_special_token_134|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128143": { + "content": "<|reserved_special_token_135|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128144": { + "content": "<|reserved_special_token_136|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128145": { + "content": "<|reserved_special_token_137|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128146": { + "content": "<|reserved_special_token_138|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128147": { + "content": "<|reserved_special_token_139|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128148": { + "content": "<|reserved_special_token_140|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128149": { + "content": "<|reserved_special_token_141|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128150": { + "content": "<|reserved_special_token_142|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128151": { + "content": "<|reserved_special_token_143|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128152": { + "content": "<|reserved_special_token_144|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128153": { + "content": "<|reserved_special_token_145|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128154": { + "content": "<|reserved_special_token_146|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128155": { + "content": "<|reserved_special_token_147|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128156": { + "content": "<|reserved_special_token_148|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128157": { + "content": "<|reserved_special_token_149|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128158": { + "content": "<|reserved_special_token_150|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128159": { + "content": "<|reserved_special_token_151|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128160": { + "content": "<|reserved_special_token_152|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128161": { + "content": "<|reserved_special_token_153|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128162": { + "content": "<|reserved_special_token_154|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128163": { + "content": "<|reserved_special_token_155|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128164": { + "content": "<|reserved_special_token_156|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128165": { + "content": "<|reserved_special_token_157|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128166": { + "content": "<|reserved_special_token_158|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128167": { + "content": "<|reserved_special_token_159|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128168": { + "content": "<|reserved_special_token_160|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128169": { + "content": "<|reserved_special_token_161|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128170": { + "content": "<|reserved_special_token_162|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128171": { + "content": "<|reserved_special_token_163|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128172": { + "content": "<|reserved_special_token_164|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128173": { + "content": "<|reserved_special_token_165|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128174": { + "content": "<|reserved_special_token_166|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128175": { + "content": "<|reserved_special_token_167|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128176": { + "content": "<|reserved_special_token_168|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128177": { + "content": "<|reserved_special_token_169|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128178": { + "content": "<|reserved_special_token_170|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128179": { + "content": "<|reserved_special_token_171|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128180": { + "content": "<|reserved_special_token_172|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128181": { + "content": "<|reserved_special_token_173|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128182": { + "content": "<|reserved_special_token_174|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128183": { + "content": "<|reserved_special_token_175|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128184": { + "content": "<|reserved_special_token_176|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128185": { + "content": "<|reserved_special_token_177|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128186": { + "content": "<|reserved_special_token_178|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128187": { + "content": "<|reserved_special_token_179|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128188": { + "content": "<|reserved_special_token_180|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128189": { + "content": "<|reserved_special_token_181|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128190": { + "content": "<|reserved_special_token_182|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128191": { + "content": "<|reserved_special_token_183|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128192": { + "content": "<|reserved_special_token_184|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128193": { + "content": "<|reserved_special_token_185|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128194": { + "content": "<|reserved_special_token_186|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128195": { + "content": "<|reserved_special_token_187|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128196": { + "content": "<|reserved_special_token_188|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128197": { + "content": "<|reserved_special_token_189|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128198": { + "content": "<|reserved_special_token_190|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128199": { + "content": "<|reserved_special_token_191|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128200": { + "content": "<|reserved_special_token_192|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128201": { + "content": "<|reserved_special_token_193|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128202": { + "content": "<|reserved_special_token_194|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128203": { + "content": "<|reserved_special_token_195|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128204": { + "content": "<|reserved_special_token_196|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128205": { + "content": "<|reserved_special_token_197|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128206": { + "content": "<|reserved_special_token_198|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128207": { + "content": "<|reserved_special_token_199|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128208": { + "content": "<|reserved_special_token_200|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128209": { + "content": "<|reserved_special_token_201|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128210": { + "content": "<|reserved_special_token_202|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128211": { + "content": "<|reserved_special_token_203|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128212": { + "content": "<|reserved_special_token_204|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128213": { + "content": "<|reserved_special_token_205|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128214": { + "content": "<|reserved_special_token_206|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128215": { + "content": "<|reserved_special_token_207|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128216": { + "content": "<|reserved_special_token_208|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128217": { + "content": "<|reserved_special_token_209|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128218": { + "content": "<|reserved_special_token_210|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128219": { + "content": "<|reserved_special_token_211|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128220": { + "content": "<|reserved_special_token_212|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128221": { + "content": "<|reserved_special_token_213|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128222": { + "content": "<|reserved_special_token_214|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128223": { + "content": "<|reserved_special_token_215|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128224": { + "content": "<|reserved_special_token_216|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128225": { + "content": "<|reserved_special_token_217|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128226": { + "content": "<|reserved_special_token_218|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128227": { + "content": "<|reserved_special_token_219|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128228": { + "content": "<|reserved_special_token_220|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128229": { + "content": "<|reserved_special_token_221|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128230": { + "content": "<|reserved_special_token_222|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128231": { + "content": "<|reserved_special_token_223|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128232": { + "content": "<|reserved_special_token_224|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128233": { + "content": "<|reserved_special_token_225|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128234": { + "content": "<|reserved_special_token_226|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128235": { + "content": "<|reserved_special_token_227|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128236": { + "content": "<|reserved_special_token_228|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128237": { + "content": "<|reserved_special_token_229|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128238": { + "content": "<|reserved_special_token_230|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128239": { + "content": "<|reserved_special_token_231|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128240": { + "content": "<|reserved_special_token_232|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128241": { + "content": "<|reserved_special_token_233|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128242": { + "content": "<|reserved_special_token_234|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128243": { + "content": "<|reserved_special_token_235|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128244": { + "content": "<|reserved_special_token_236|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128245": { + "content": "<|reserved_special_token_237|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128246": { + "content": "<|reserved_special_token_238|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128247": { + "content": "<|reserved_special_token_239|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128248": { + "content": "<|reserved_special_token_240|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128249": { + "content": "<|reserved_special_token_241|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128250": { + "content": "<|reserved_special_token_242|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128251": { + "content": "<|reserved_special_token_243|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128252": { + "content": "<|reserved_special_token_244|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128253": { + "content": "<|reserved_special_token_245|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128254": { + "content": "<|reserved_special_token_246|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128255": { + "content": "<|reserved_special_token_247|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128256": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128257": { + "content": "🅁", + "lstrip": true, + "normalized": true, + "rstrip": false, + "single_word": false, + "special": false + } + }, + "bos_token": "<|begin_of_text|>", + "clean_up_tokenization_spaces": true, + "eos_token": "<|end_of_text|>", + "model_input_names": [ + "input_ids", + "attention_mask" + ], + "model_max_length": 131072, + "pad_token": "[PAD]", + "tokenizer_class": "PreTrainedTokenizerFast" +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/trainer_state.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..407fe9477aef49007cd327b52eb6c86661985b4f --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/trainer_state.json @@ -0,0 +1,1593 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.2054442732408834, + "eval_steps": 10, + "global_step": 200, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.001027221366204417, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7873, + "step": 1 + }, + { + "epoch": 0.002054442732408834, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7904, + "step": 2 + }, + { + "epoch": 0.0030816640986132513, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8348, + "step": 3 + }, + { + "epoch": 0.004108885464817668, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8357, + "step": 4 + }, + { + "epoch": 0.005136106831022085, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8153, + "step": 5 + }, + { + "epoch": 0.0061633281972265025, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.825, + "step": 6 + }, + { + "epoch": 0.007190549563430919, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8002, + "step": 7 + }, + { + "epoch": 0.008217770929635337, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8413, + "step": 8 + }, + { + "epoch": 0.009244992295839754, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8157, + "step": 9 + }, + { + "epoch": 0.01027221366204417, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8249, + "step": 10 + }, + { + "epoch": 0.01027221366204417, + "eval_loss": 2.8518412113189697, + "eval_runtime": 43.212, + "eval_samples_per_second": 23.142, + "eval_steps_per_second": 0.972, + "step": 10 + }, + { + "epoch": 0.011299435028248588, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7829, + "step": 11 + }, + { + "epoch": 0.012326656394453005, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7819, + "step": 12 + }, + { + "epoch": 0.01335387776065742, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8355, + "step": 13 + }, + { + "epoch": 0.014381099126861838, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8138, + "step": 14 + }, + { + "epoch": 0.015408320493066256, + "grad_norm": 5.162191390991211, + "learning_rate": 1.7123287671232876e-08, + "loss": 2.8088, + "step": 15 + }, + { + "epoch": 0.016435541859270673, + "grad_norm": 5.5954084396362305, + "learning_rate": 3.424657534246575e-08, + "loss": 2.7883, + "step": 16 + }, + { + "epoch": 0.01746276322547509, + "grad_norm": 5.542337417602539, + "learning_rate": 5.136986301369863e-08, + "loss": 2.7625, + "step": 17 + }, + { + "epoch": 0.01848998459167951, + "grad_norm": 5.300518989562988, + "learning_rate": 6.84931506849315e-08, + "loss": 2.84, + "step": 18 + }, + { + "epoch": 0.019517205957883924, + "grad_norm": 5.6646881103515625, + "learning_rate": 8.561643835616439e-08, + "loss": 2.761, + "step": 19 + }, + { + "epoch": 0.02054442732408834, + "grad_norm": 5.375557899475098, + "learning_rate": 1.0273972602739726e-07, + "loss": 2.8215, + "step": 20 + }, + { + "epoch": 0.02054442732408834, + "eval_loss": 2.8507144451141357, + "eval_runtime": 45.1133, + "eval_samples_per_second": 22.166, + "eval_steps_per_second": 0.931, + "step": 20 + }, + { + "epoch": 0.02157164869029276, + "grad_norm": 5.1731977462768555, + "learning_rate": 1.1986301369863014e-07, + "loss": 2.8348, + "step": 21 + }, + { + "epoch": 0.022598870056497175, + "grad_norm": 5.370945930480957, + "learning_rate": 1.36986301369863e-07, + "loss": 2.822, + "step": 22 + }, + { + "epoch": 0.02362609142270159, + "grad_norm": 5.550389289855957, + "learning_rate": 1.541095890410959e-07, + "loss": 2.7904, + "step": 23 + }, + { + "epoch": 0.02465331278890601, + "grad_norm": 5.339347839355469, + "learning_rate": 1.7123287671232878e-07, + "loss": 2.8066, + "step": 24 + }, + { + "epoch": 0.025680534155110426, + "grad_norm": 5.2319722175598145, + "learning_rate": 1.8835616438356165e-07, + "loss": 2.7976, + "step": 25 + }, + { + "epoch": 0.02670775552131484, + "grad_norm": 5.207149982452393, + "learning_rate": 2.0547945205479452e-07, + "loss": 2.8162, + "step": 26 + }, + { + "epoch": 0.02773497688751926, + "grad_norm": 4.400204181671143, + "learning_rate": 2.226027397260274e-07, + "loss": 2.7808, + "step": 27 + }, + { + "epoch": 0.028762198253723677, + "grad_norm": 4.21675968170166, + "learning_rate": 2.397260273972603e-07, + "loss": 2.8044, + "step": 28 + }, + { + "epoch": 0.029789419619928096, + "grad_norm": 4.406878471374512, + "learning_rate": 2.568493150684932e-07, + "loss": 2.8162, + "step": 29 + }, + { + "epoch": 0.030816640986132512, + "grad_norm": 4.024835586547852, + "learning_rate": 2.73972602739726e-07, + "loss": 2.7994, + "step": 30 + }, + { + "epoch": 0.030816640986132512, + "eval_loss": 2.8124029636383057, + "eval_runtime": 43.1772, + "eval_samples_per_second": 23.16, + "eval_steps_per_second": 0.973, + "step": 30 + }, + { + "epoch": 0.03184386235233693, + "grad_norm": 3.5378901958465576, + "learning_rate": 2.910958904109589e-07, + "loss": 2.7637, + "step": 31 + }, + { + "epoch": 0.03287108371854135, + "grad_norm": 3.6289916038513184, + "learning_rate": 3.082191780821918e-07, + "loss": 2.8017, + "step": 32 + }, + { + "epoch": 0.03389830508474576, + "grad_norm": 3.6789698600769043, + "learning_rate": 3.2534246575342466e-07, + "loss": 2.7723, + "step": 33 + }, + { + "epoch": 0.03492552645095018, + "grad_norm": 3.491389751434326, + "learning_rate": 3.4246575342465755e-07, + "loss": 2.7528, + "step": 34 + }, + { + "epoch": 0.035952747817154594, + "grad_norm": 3.383080005645752, + "learning_rate": 3.595890410958904e-07, + "loss": 2.7418, + "step": 35 + }, + { + "epoch": 0.03697996918335902, + "grad_norm": 3.4823594093322754, + "learning_rate": 3.767123287671233e-07, + "loss": 2.6662, + "step": 36 + }, + { + "epoch": 0.03800719054956343, + "grad_norm": 3.2146108150482178, + "learning_rate": 3.938356164383562e-07, + "loss": 2.7659, + "step": 37 + }, + { + "epoch": 0.03903441191576785, + "grad_norm": 2.9549412727355957, + "learning_rate": 4.1095890410958903e-07, + "loss": 2.6838, + "step": 38 + }, + { + "epoch": 0.040061633281972264, + "grad_norm": 3.0744080543518066, + "learning_rate": 4.2808219178082193e-07, + "loss": 2.6879, + "step": 39 + }, + { + "epoch": 0.04108885464817668, + "grad_norm": 3.19840669631958, + "learning_rate": 4.452054794520548e-07, + "loss": 2.7443, + "step": 40 + }, + { + "epoch": 0.04108885464817668, + "eval_loss": 2.7469444274902344, + "eval_runtime": 46.4676, + "eval_samples_per_second": 21.52, + "eval_steps_per_second": 0.904, + "step": 40 + }, + { + "epoch": 0.042116076014381096, + "grad_norm": 3.178406000137329, + "learning_rate": 4.6232876712328767e-07, + "loss": 2.7036, + "step": 41 + }, + { + "epoch": 0.04314329738058552, + "grad_norm": 3.2032673358917236, + "learning_rate": 4.794520547945206e-07, + "loss": 2.6465, + "step": 42 + }, + { + "epoch": 0.044170518746789934, + "grad_norm": 3.1525962352752686, + "learning_rate": 4.965753424657534e-07, + "loss": 2.6996, + "step": 43 + }, + { + "epoch": 0.04519774011299435, + "grad_norm": 2.907524585723877, + "learning_rate": 5.136986301369864e-07, + "loss": 2.6581, + "step": 44 + }, + { + "epoch": 0.046224961479198766, + "grad_norm": 2.5885250568389893, + "learning_rate": 5.308219178082192e-07, + "loss": 2.6773, + "step": 45 + }, + { + "epoch": 0.04725218284540318, + "grad_norm": 2.569068431854248, + "learning_rate": 5.47945205479452e-07, + "loss": 2.6635, + "step": 46 + }, + { + "epoch": 0.048279404211607604, + "grad_norm": 2.5313518047332764, + "learning_rate": 5.65068493150685e-07, + "loss": 2.706, + "step": 47 + }, + { + "epoch": 0.04930662557781202, + "grad_norm": 2.615442991256714, + "learning_rate": 5.821917808219178e-07, + "loss": 2.638, + "step": 48 + }, + { + "epoch": 0.050333846944016436, + "grad_norm": 2.7420806884765625, + "learning_rate": 5.993150684931507e-07, + "loss": 2.6398, + "step": 49 + }, + { + "epoch": 0.05136106831022085, + "grad_norm": 2.5246975421905518, + "learning_rate": 6.164383561643836e-07, + "loss": 2.5916, + "step": 50 + }, + { + "epoch": 0.05136106831022085, + "eval_loss": 2.6723692417144775, + "eval_runtime": 46.7206, + "eval_samples_per_second": 21.404, + "eval_steps_per_second": 0.899, + "step": 50 + }, + { + "epoch": 0.05238828967642527, + "grad_norm": 2.390143632888794, + "learning_rate": 6.335616438356165e-07, + "loss": 2.5773, + "step": 51 + }, + { + "epoch": 0.05341551104262968, + "grad_norm": 2.0792484283447266, + "learning_rate": 6.506849315068493e-07, + "loss": 2.6249, + "step": 52 + }, + { + "epoch": 0.054442732408834106, + "grad_norm": 1.9995019435882568, + "learning_rate": 6.678082191780823e-07, + "loss": 2.5425, + "step": 53 + }, + { + "epoch": 0.05546995377503852, + "grad_norm": 1.985452651977539, + "learning_rate": 6.849315068493151e-07, + "loss": 2.5986, + "step": 54 + }, + { + "epoch": 0.05649717514124294, + "grad_norm": 2.10184907913208, + "learning_rate": 7.020547945205481e-07, + "loss": 2.5713, + "step": 55 + }, + { + "epoch": 0.05752439650744735, + "grad_norm": 2.0898985862731934, + "learning_rate": 7.191780821917808e-07, + "loss": 2.535, + "step": 56 + }, + { + "epoch": 0.05855161787365177, + "grad_norm": 2.0239579677581787, + "learning_rate": 7.363013698630137e-07, + "loss": 2.5667, + "step": 57 + }, + { + "epoch": 0.05957883923985619, + "grad_norm": 2.207080125808716, + "learning_rate": 7.534246575342466e-07, + "loss": 2.4999, + "step": 58 + }, + { + "epoch": 0.06060606060606061, + "grad_norm": 1.8541918992996216, + "learning_rate": 7.705479452054795e-07, + "loss": 2.5582, + "step": 59 + }, + { + "epoch": 0.061633281972265024, + "grad_norm": 1.8530476093292236, + "learning_rate": 7.876712328767124e-07, + "loss": 2.5535, + "step": 60 + }, + { + "epoch": 0.061633281972265024, + "eval_loss": 2.607963800430298, + "eval_runtime": 45.9754, + "eval_samples_per_second": 21.751, + "eval_steps_per_second": 0.914, + "step": 60 + }, + { + "epoch": 0.06266050333846944, + "grad_norm": 1.7978090047836304, + "learning_rate": 8.047945205479453e-07, + "loss": 2.53, + "step": 61 + }, + { + "epoch": 0.06368772470467386, + "grad_norm": 1.8173762559890747, + "learning_rate": 8.219178082191781e-07, + "loss": 2.5144, + "step": 62 + }, + { + "epoch": 0.06471494607087827, + "grad_norm": 2.0847485065460205, + "learning_rate": 8.39041095890411e-07, + "loss": 2.5525, + "step": 63 + }, + { + "epoch": 0.0657421674370827, + "grad_norm": 1.8691741228103638, + "learning_rate": 8.561643835616439e-07, + "loss": 2.5451, + "step": 64 + }, + { + "epoch": 0.0667693888032871, + "grad_norm": 2.0242726802825928, + "learning_rate": 8.732876712328768e-07, + "loss": 2.4145, + "step": 65 + }, + { + "epoch": 0.06779661016949153, + "grad_norm": 1.8268908262252808, + "learning_rate": 8.904109589041097e-07, + "loss": 2.5024, + "step": 66 + }, + { + "epoch": 0.06882383153569595, + "grad_norm": 1.8703057765960693, + "learning_rate": 9.075342465753426e-07, + "loss": 2.461, + "step": 67 + }, + { + "epoch": 0.06985105290190036, + "grad_norm": 1.8747144937515259, + "learning_rate": 9.246575342465753e-07, + "loss": 2.5122, + "step": 68 + }, + { + "epoch": 0.07087827426810478, + "grad_norm": 1.744924783706665, + "learning_rate": 9.417808219178083e-07, + "loss": 2.4673, + "step": 69 + }, + { + "epoch": 0.07190549563430919, + "grad_norm": 1.7492460012435913, + "learning_rate": 9.589041095890411e-07, + "loss": 2.5598, + "step": 70 + }, + { + "epoch": 0.07190549563430919, + "eval_loss": 2.547417640686035, + "eval_runtime": 49.7488, + "eval_samples_per_second": 20.101, + "eval_steps_per_second": 0.844, + "step": 70 + }, + { + "epoch": 0.07293271700051361, + "grad_norm": 1.9218230247497559, + "learning_rate": 9.76027397260274e-07, + "loss": 2.4562, + "step": 71 + }, + { + "epoch": 0.07395993836671803, + "grad_norm": 2.0127363204956055, + "learning_rate": 9.931506849315068e-07, + "loss": 2.4141, + "step": 72 + }, + { + "epoch": 0.07498715973292244, + "grad_norm": 1.8700369596481323, + "learning_rate": 1.0102739726027399e-06, + "loss": 2.4696, + "step": 73 + }, + { + "epoch": 0.07601438109912687, + "grad_norm": 2.1706016063690186, + "learning_rate": 1.0273972602739727e-06, + "loss": 2.4404, + "step": 74 + }, + { + "epoch": 0.07704160246533127, + "grad_norm": 2.470412254333496, + "learning_rate": 1.0445205479452056e-06, + "loss": 2.4409, + "step": 75 + }, + { + "epoch": 0.0780688238315357, + "grad_norm": 1.906569242477417, + "learning_rate": 1.0616438356164384e-06, + "loss": 2.4313, + "step": 76 + }, + { + "epoch": 0.07909604519774012, + "grad_norm": 2.122936487197876, + "learning_rate": 1.0787671232876712e-06, + "loss": 2.4369, + "step": 77 + }, + { + "epoch": 0.08012326656394453, + "grad_norm": 2.079127311706543, + "learning_rate": 1.095890410958904e-06, + "loss": 2.449, + "step": 78 + }, + { + "epoch": 0.08115048793014895, + "grad_norm": 1.9173146486282349, + "learning_rate": 1.1130136986301371e-06, + "loss": 2.3782, + "step": 79 + }, + { + "epoch": 0.08217770929635336, + "grad_norm": 1.9856903553009033, + "learning_rate": 1.13013698630137e-06, + "loss": 2.4215, + "step": 80 + }, + { + "epoch": 0.08217770929635336, + "eval_loss": 2.479673385620117, + "eval_runtime": 46.2211, + "eval_samples_per_second": 21.635, + "eval_steps_per_second": 0.909, + "step": 80 + }, + { + "epoch": 0.08320493066255778, + "grad_norm": 2.3764865398406982, + "learning_rate": 1.1472602739726028e-06, + "loss": 2.3744, + "step": 81 + }, + { + "epoch": 0.08423215202876219, + "grad_norm": 1.980668544769287, + "learning_rate": 1.1643835616438357e-06, + "loss": 2.3856, + "step": 82 + }, + { + "epoch": 0.08525937339496661, + "grad_norm": 2.2589704990386963, + "learning_rate": 1.1815068493150685e-06, + "loss": 2.3853, + "step": 83 + }, + { + "epoch": 0.08628659476117104, + "grad_norm": 2.171780824661255, + "learning_rate": 1.1986301369863014e-06, + "loss": 2.3603, + "step": 84 + }, + { + "epoch": 0.08731381612737545, + "grad_norm": 2.25736403465271, + "learning_rate": 1.2157534246575344e-06, + "loss": 2.3189, + "step": 85 + }, + { + "epoch": 0.08834103749357987, + "grad_norm": 2.3643763065338135, + "learning_rate": 1.2328767123287673e-06, + "loss": 2.3795, + "step": 86 + }, + { + "epoch": 0.08936825885978428, + "grad_norm": 2.384516954421997, + "learning_rate": 1.25e-06, + "loss": 2.3257, + "step": 87 + }, + { + "epoch": 0.0903954802259887, + "grad_norm": 2.5072567462921143, + "learning_rate": 1.267123287671233e-06, + "loss": 2.3125, + "step": 88 + }, + { + "epoch": 0.09142270159219312, + "grad_norm": 2.2726807594299316, + "learning_rate": 1.284246575342466e-06, + "loss": 2.3297, + "step": 89 + }, + { + "epoch": 0.09244992295839753, + "grad_norm": 2.4817421436309814, + "learning_rate": 1.3013698630136986e-06, + "loss": 2.2878, + "step": 90 + }, + { + "epoch": 0.09244992295839753, + "eval_loss": 2.4065146446228027, + "eval_runtime": 47.9372, + "eval_samples_per_second": 20.861, + "eval_steps_per_second": 0.876, + "step": 90 + }, + { + "epoch": 0.09347714432460195, + "grad_norm": 2.2219603061676025, + "learning_rate": 1.3184931506849317e-06, + "loss": 2.2978, + "step": 91 + }, + { + "epoch": 0.09450436569080636, + "grad_norm": 2.8790745735168457, + "learning_rate": 1.3356164383561645e-06, + "loss": 2.313, + "step": 92 + }, + { + "epoch": 0.09553158705701079, + "grad_norm": 2.589200496673584, + "learning_rate": 1.3527397260273976e-06, + "loss": 2.2937, + "step": 93 + }, + { + "epoch": 0.09655880842321521, + "grad_norm": 2.760016679763794, + "learning_rate": 1.3698630136986302e-06, + "loss": 2.277, + "step": 94 + }, + { + "epoch": 0.09758602978941962, + "grad_norm": 2.2641494274139404, + "learning_rate": 1.386986301369863e-06, + "loss": 2.3336, + "step": 95 + }, + { + "epoch": 0.09861325115562404, + "grad_norm": 2.2786386013031006, + "learning_rate": 1.4041095890410961e-06, + "loss": 2.233, + "step": 96 + }, + { + "epoch": 0.09964047252182845, + "grad_norm": 2.5590052604675293, + "learning_rate": 1.421232876712329e-06, + "loss": 2.2851, + "step": 97 + }, + { + "epoch": 0.10066769388803287, + "grad_norm": 2.5609471797943115, + "learning_rate": 1.4383561643835616e-06, + "loss": 2.1848, + "step": 98 + }, + { + "epoch": 0.1016949152542373, + "grad_norm": 2.442521810531616, + "learning_rate": 1.4554794520547946e-06, + "loss": 2.2073, + "step": 99 + }, + { + "epoch": 0.1027221366204417, + "grad_norm": 2.3735859394073486, + "learning_rate": 1.4726027397260275e-06, + "loss": 2.2353, + "step": 100 + }, + { + "epoch": 0.1027221366204417, + "eval_loss": 2.3457911014556885, + "eval_runtime": 51.6469, + "eval_samples_per_second": 19.362, + "eval_steps_per_second": 0.813, + "step": 100 + }, + { + "epoch": 0.10374935798664613, + "grad_norm": 2.175354242324829, + "learning_rate": 1.4897260273972605e-06, + "loss": 2.2259, + "step": 101 + }, + { + "epoch": 0.10477657935285054, + "grad_norm": 2.217677116394043, + "learning_rate": 1.5068493150684932e-06, + "loss": 2.2082, + "step": 102 + }, + { + "epoch": 0.10580380071905496, + "grad_norm": 2.4618067741394043, + "learning_rate": 1.5239726027397262e-06, + "loss": 2.1649, + "step": 103 + }, + { + "epoch": 0.10683102208525937, + "grad_norm": 2.232060432434082, + "learning_rate": 1.541095890410959e-06, + "loss": 2.1962, + "step": 104 + }, + { + "epoch": 0.10785824345146379, + "grad_norm": 2.9140872955322266, + "learning_rate": 1.5582191780821921e-06, + "loss": 2.2446, + "step": 105 + }, + { + "epoch": 0.10888546481766821, + "grad_norm": 2.232158660888672, + "learning_rate": 1.5753424657534248e-06, + "loss": 2.2179, + "step": 106 + }, + { + "epoch": 0.10991268618387262, + "grad_norm": 2.804938793182373, + "learning_rate": 1.5924657534246576e-06, + "loss": 2.2449, + "step": 107 + }, + { + "epoch": 0.11093990755007704, + "grad_norm": 2.3690710067749023, + "learning_rate": 1.6095890410958907e-06, + "loss": 2.225, + "step": 108 + }, + { + "epoch": 0.11196712891628145, + "grad_norm": 2.7717416286468506, + "learning_rate": 1.6267123287671235e-06, + "loss": 2.0922, + "step": 109 + }, + { + "epoch": 0.11299435028248588, + "grad_norm": 3.053602933883667, + "learning_rate": 1.6438356164383561e-06, + "loss": 2.1702, + "step": 110 + }, + { + "epoch": 0.11299435028248588, + "eval_loss": 2.298161029815674, + "eval_runtime": 44.0776, + "eval_samples_per_second": 22.687, + "eval_steps_per_second": 0.953, + "step": 110 + }, + { + "epoch": 0.1140215716486903, + "grad_norm": 2.8673532009124756, + "learning_rate": 1.6609589041095892e-06, + "loss": 2.0938, + "step": 111 + }, + { + "epoch": 0.1150487930148947, + "grad_norm": 2.57185959815979, + "learning_rate": 1.678082191780822e-06, + "loss": 2.1892, + "step": 112 + }, + { + "epoch": 0.11607601438109913, + "grad_norm": 2.5278828144073486, + "learning_rate": 1.695205479452055e-06, + "loss": 2.1251, + "step": 113 + }, + { + "epoch": 0.11710323574730354, + "grad_norm": 2.8581748008728027, + "learning_rate": 1.7123287671232877e-06, + "loss": 2.1699, + "step": 114 + }, + { + "epoch": 0.11813045711350796, + "grad_norm": 2.924959897994995, + "learning_rate": 1.7294520547945206e-06, + "loss": 2.1005, + "step": 115 + }, + { + "epoch": 0.11915767847971238, + "grad_norm": 2.477034091949463, + "learning_rate": 1.7465753424657536e-06, + "loss": 2.2016, + "step": 116 + }, + { + "epoch": 0.12018489984591679, + "grad_norm": 2.8719305992126465, + "learning_rate": 1.7636986301369865e-06, + "loss": 2.1588, + "step": 117 + }, + { + "epoch": 0.12121212121212122, + "grad_norm": 2.4403953552246094, + "learning_rate": 1.7808219178082193e-06, + "loss": 2.1445, + "step": 118 + }, + { + "epoch": 0.12223934257832562, + "grad_norm": 2.7994632720947266, + "learning_rate": 1.7979452054794521e-06, + "loss": 2.1369, + "step": 119 + }, + { + "epoch": 0.12326656394453005, + "grad_norm": 2.163196563720703, + "learning_rate": 1.8150684931506852e-06, + "loss": 2.2195, + "step": 120 + }, + { + "epoch": 0.12326656394453005, + "eval_loss": 2.2648775577545166, + "eval_runtime": 49.47, + "eval_samples_per_second": 20.214, + "eval_steps_per_second": 0.849, + "step": 120 + }, + { + "epoch": 0.12429378531073447, + "grad_norm": 2.6743483543395996, + "learning_rate": 1.832191780821918e-06, + "loss": 2.1521, + "step": 121 + }, + { + "epoch": 0.12532100667693888, + "grad_norm": 2.54795503616333, + "learning_rate": 1.8493150684931507e-06, + "loss": 2.1158, + "step": 122 + }, + { + "epoch": 0.1263482280431433, + "grad_norm": 2.9093267917633057, + "learning_rate": 1.8664383561643837e-06, + "loss": 2.0707, + "step": 123 + }, + { + "epoch": 0.12737544940934772, + "grad_norm": 2.416940212249756, + "learning_rate": 1.8835616438356166e-06, + "loss": 2.0901, + "step": 124 + }, + { + "epoch": 0.12840267077555212, + "grad_norm": 2.8333663940429688, + "learning_rate": 1.9006849315068496e-06, + "loss": 2.0545, + "step": 125 + }, + { + "epoch": 0.12942989214175654, + "grad_norm": 2.483051061630249, + "learning_rate": 1.9178082191780823e-06, + "loss": 2.1274, + "step": 126 + }, + { + "epoch": 0.13045711350796096, + "grad_norm": 2.3486859798431396, + "learning_rate": 1.9349315068493153e-06, + "loss": 2.0478, + "step": 127 + }, + { + "epoch": 0.1314843348741654, + "grad_norm": 3.1512887477874756, + "learning_rate": 1.952054794520548e-06, + "loss": 2.1041, + "step": 128 + }, + { + "epoch": 0.1325115562403698, + "grad_norm": 2.902873992919922, + "learning_rate": 1.969178082191781e-06, + "loss": 2.1332, + "step": 129 + }, + { + "epoch": 0.1335387776065742, + "grad_norm": 2.779132127761841, + "learning_rate": 1.9863013698630136e-06, + "loss": 2.1297, + "step": 130 + }, + { + "epoch": 0.1335387776065742, + "eval_loss": 2.2391879558563232, + "eval_runtime": 55.3018, + "eval_samples_per_second": 18.083, + "eval_steps_per_second": 0.759, + "step": 130 + }, + { + "epoch": 0.13456599897277863, + "grad_norm": 2.703019618988037, + "learning_rate": 2.0034246575342467e-06, + "loss": 2.0506, + "step": 131 + }, + { + "epoch": 0.13559322033898305, + "grad_norm": 3.0187575817108154, + "learning_rate": 2.0205479452054797e-06, + "loss": 2.1291, + "step": 132 + }, + { + "epoch": 0.13662044170518747, + "grad_norm": 2.7629971504211426, + "learning_rate": 2.037671232876713e-06, + "loss": 2.1003, + "step": 133 + }, + { + "epoch": 0.1376476630713919, + "grad_norm": 3.175776720046997, + "learning_rate": 2.0547945205479454e-06, + "loss": 2.0429, + "step": 134 + }, + { + "epoch": 0.1386748844375963, + "grad_norm": 3.0342514514923096, + "learning_rate": 2.071917808219178e-06, + "loss": 2.0858, + "step": 135 + }, + { + "epoch": 0.1397021058038007, + "grad_norm": 3.0861928462982178, + "learning_rate": 2.089041095890411e-06, + "loss": 2.0964, + "step": 136 + }, + { + "epoch": 0.14072932717000514, + "grad_norm": 3.151945114135742, + "learning_rate": 2.106164383561644e-06, + "loss": 2.0194, + "step": 137 + }, + { + "epoch": 0.14175654853620956, + "grad_norm": 2.9069676399230957, + "learning_rate": 2.123287671232877e-06, + "loss": 2.1131, + "step": 138 + }, + { + "epoch": 0.14278376990241398, + "grad_norm": 3.2285282611846924, + "learning_rate": 2.14041095890411e-06, + "loss": 2.0643, + "step": 139 + }, + { + "epoch": 0.14381099126861838, + "grad_norm": 2.500481367111206, + "learning_rate": 2.1575342465753425e-06, + "loss": 2.0625, + "step": 140 + }, + { + "epoch": 0.14381099126861838, + "eval_loss": 2.211498975753784, + "eval_runtime": 55.1046, + "eval_samples_per_second": 18.147, + "eval_steps_per_second": 0.762, + "step": 140 + }, + { + "epoch": 0.1448382126348228, + "grad_norm": 4.004592418670654, + "learning_rate": 2.1746575342465755e-06, + "loss": 2.0644, + "step": 141 + }, + { + "epoch": 0.14586543400102722, + "grad_norm": 2.9258391857147217, + "learning_rate": 2.191780821917808e-06, + "loss": 2.0174, + "step": 142 + }, + { + "epoch": 0.14689265536723164, + "grad_norm": 3.0073509216308594, + "learning_rate": 2.2089041095890412e-06, + "loss": 2.1417, + "step": 143 + }, + { + "epoch": 0.14791987673343607, + "grad_norm": 3.040708065032959, + "learning_rate": 2.2260273972602743e-06, + "loss": 2.0357, + "step": 144 + }, + { + "epoch": 0.14894709809964046, + "grad_norm": 2.9062869548797607, + "learning_rate": 2.243150684931507e-06, + "loss": 2.0122, + "step": 145 + }, + { + "epoch": 0.14997431946584489, + "grad_norm": 3.28969669342041, + "learning_rate": 2.26027397260274e-06, + "loss": 2.0255, + "step": 146 + }, + { + "epoch": 0.1510015408320493, + "grad_norm": 3.608752727508545, + "learning_rate": 2.2773972602739726e-06, + "loss": 2.0184, + "step": 147 + }, + { + "epoch": 0.15202876219825373, + "grad_norm": 2.900341033935547, + "learning_rate": 2.2945205479452057e-06, + "loss": 2.0532, + "step": 148 + }, + { + "epoch": 0.15305598356445815, + "grad_norm": 2.881843328475952, + "learning_rate": 2.3116438356164387e-06, + "loss": 2.1336, + "step": 149 + }, + { + "epoch": 0.15408320493066255, + "grad_norm": 3.4043259620666504, + "learning_rate": 2.3287671232876713e-06, + "loss": 2.0536, + "step": 150 + }, + { + "epoch": 0.15408320493066255, + "eval_loss": 2.192880153656006, + "eval_runtime": 53.7838, + "eval_samples_per_second": 18.593, + "eval_steps_per_second": 0.781, + "step": 150 + }, + { + "epoch": 0.15511042629686697, + "grad_norm": 2.693107843399048, + "learning_rate": 2.3458904109589044e-06, + "loss": 1.9503, + "step": 151 + }, + { + "epoch": 0.1561376476630714, + "grad_norm": 2.885260581970215, + "learning_rate": 2.363013698630137e-06, + "loss": 2.0081, + "step": 152 + }, + { + "epoch": 0.15716486902927582, + "grad_norm": 3.4459924697875977, + "learning_rate": 2.38013698630137e-06, + "loss": 2.0649, + "step": 153 + }, + { + "epoch": 0.15819209039548024, + "grad_norm": 2.8447093963623047, + "learning_rate": 2.3972602739726027e-06, + "loss": 2.0476, + "step": 154 + }, + { + "epoch": 0.15921931176168463, + "grad_norm": 2.755176305770874, + "learning_rate": 2.4143835616438358e-06, + "loss": 2.0006, + "step": 155 + }, + { + "epoch": 0.16024653312788906, + "grad_norm": 2.8677561283111572, + "learning_rate": 2.431506849315069e-06, + "loss": 1.9663, + "step": 156 + }, + { + "epoch": 0.16127375449409348, + "grad_norm": 2.783351182937622, + "learning_rate": 2.4486301369863015e-06, + "loss": 2.0301, + "step": 157 + }, + { + "epoch": 0.1623009758602979, + "grad_norm": 2.7545998096466064, + "learning_rate": 2.4657534246575345e-06, + "loss": 2.0438, + "step": 158 + }, + { + "epoch": 0.1633281972265023, + "grad_norm": 2.8140594959259033, + "learning_rate": 2.482876712328767e-06, + "loss": 2.0055, + "step": 159 + }, + { + "epoch": 0.16435541859270672, + "grad_norm": 2.6072335243225098, + "learning_rate": 2.5e-06, + "loss": 2.0262, + "step": 160 + }, + { + "epoch": 0.16435541859270672, + "eval_loss": 2.179842710494995, + "eval_runtime": 52.6159, + "eval_samples_per_second": 19.006, + "eval_steps_per_second": 0.798, + "step": 160 + }, + { + "epoch": 0.16538263995891114, + "grad_norm": 2.830212354660034, + "learning_rate": 2.5171232876712333e-06, + "loss": 2.0154, + "step": 161 + }, + { + "epoch": 0.16640986132511557, + "grad_norm": 2.677140712738037, + "learning_rate": 2.534246575342466e-06, + "loss": 1.9933, + "step": 162 + }, + { + "epoch": 0.16743708269132, + "grad_norm": 2.837759017944336, + "learning_rate": 2.551369863013699e-06, + "loss": 1.9602, + "step": 163 + }, + { + "epoch": 0.16846430405752438, + "grad_norm": 3.9556589126586914, + "learning_rate": 2.568493150684932e-06, + "loss": 2.0121, + "step": 164 + }, + { + "epoch": 0.1694915254237288, + "grad_norm": 2.8715925216674805, + "learning_rate": 2.585616438356164e-06, + "loss": 2.0219, + "step": 165 + }, + { + "epoch": 0.17051874678993323, + "grad_norm": 3.5073623657226562, + "learning_rate": 2.6027397260273973e-06, + "loss": 2.0202, + "step": 166 + }, + { + "epoch": 0.17154596815613765, + "grad_norm": 2.751962900161743, + "learning_rate": 2.6198630136986303e-06, + "loss": 1.9528, + "step": 167 + }, + { + "epoch": 0.17257318952234207, + "grad_norm": 3.7753474712371826, + "learning_rate": 2.6369863013698634e-06, + "loss": 2.0208, + "step": 168 + }, + { + "epoch": 0.17360041088854647, + "grad_norm": 2.642289876937866, + "learning_rate": 2.654109589041096e-06, + "loss": 2.0386, + "step": 169 + }, + { + "epoch": 0.1746276322547509, + "grad_norm": 2.82173490524292, + "learning_rate": 2.671232876712329e-06, + "loss": 2.0355, + "step": 170 + }, + { + "epoch": 0.1746276322547509, + "eval_loss": 2.1699249744415283, + "eval_runtime": 50.0341, + "eval_samples_per_second": 19.986, + "eval_steps_per_second": 0.839, + "step": 170 + }, + { + "epoch": 0.17565485362095531, + "grad_norm": 3.342280149459839, + "learning_rate": 2.688356164383562e-06, + "loss": 1.9672, + "step": 171 + }, + { + "epoch": 0.17668207498715974, + "grad_norm": 2.6548173427581787, + "learning_rate": 2.705479452054795e-06, + "loss": 1.9669, + "step": 172 + }, + { + "epoch": 0.17770929635336416, + "grad_norm": 2.6548173427581787, + "learning_rate": 2.705479452054795e-06, + "loss": 1.961, + "step": 173 + }, + { + "epoch": 0.17873651771956856, + "grad_norm": 3.4306116104125977, + "learning_rate": 2.7226027397260274e-06, + "loss": 1.9173, + "step": 174 + }, + { + "epoch": 0.17976373908577298, + "grad_norm": 2.7991764545440674, + "learning_rate": 2.7397260273972604e-06, + "loss": 2.0339, + "step": 175 + }, + { + "epoch": 0.1807909604519774, + "grad_norm": 3.6040217876434326, + "learning_rate": 2.7568493150684935e-06, + "loss": 2.0218, + "step": 176 + }, + { + "epoch": 0.18181818181818182, + "grad_norm": 3.0253515243530273, + "learning_rate": 2.773972602739726e-06, + "loss": 1.9245, + "step": 177 + }, + { + "epoch": 0.18284540318438625, + "grad_norm": 3.405141830444336, + "learning_rate": 2.791095890410959e-06, + "loss": 2.0136, + "step": 178 + }, + { + "epoch": 0.18387262455059064, + "grad_norm": 2.7518203258514404, + "learning_rate": 2.8082191780821922e-06, + "loss": 1.8925, + "step": 179 + }, + { + "epoch": 0.18489984591679506, + "grad_norm": 3.1903367042541504, + "learning_rate": 2.8253424657534253e-06, + "loss": 2.0523, + "step": 180 + }, + { + "epoch": 0.18489984591679506, + "eval_loss": 2.158029794692993, + "eval_runtime": 48.6162, + "eval_samples_per_second": 20.569, + "eval_steps_per_second": 0.864, + "step": 180 + }, + { + "epoch": 0.1859270672829995, + "grad_norm": 3.0331413745880127, + "learning_rate": 2.842465753424658e-06, + "loss": 2.0167, + "step": 181 + }, + { + "epoch": 0.1869542886492039, + "grad_norm": 2.919433116912842, + "learning_rate": 2.8595890410958905e-06, + "loss": 2.0197, + "step": 182 + }, + { + "epoch": 0.18798151001540833, + "grad_norm": 2.9013049602508545, + "learning_rate": 2.876712328767123e-06, + "loss": 2.0301, + "step": 183 + }, + { + "epoch": 0.18900873138161273, + "grad_norm": 2.738800525665283, + "learning_rate": 2.8938356164383562e-06, + "loss": 1.894, + "step": 184 + }, + { + "epoch": 0.19003595274781715, + "grad_norm": 2.670806407928467, + "learning_rate": 2.9109589041095893e-06, + "loss": 1.9127, + "step": 185 + }, + { + "epoch": 0.19106317411402157, + "grad_norm": 3.0435004234313965, + "learning_rate": 2.9280821917808223e-06, + "loss": 1.8996, + "step": 186 + }, + { + "epoch": 0.192090395480226, + "grad_norm": 3.2139949798583984, + "learning_rate": 2.945205479452055e-06, + "loss": 1.9255, + "step": 187 + }, + { + "epoch": 0.19311761684643042, + "grad_norm": 3.252965211868286, + "learning_rate": 2.962328767123288e-06, + "loss": 1.9809, + "step": 188 + }, + { + "epoch": 0.1941448382126348, + "grad_norm": 2.9325191974639893, + "learning_rate": 2.979452054794521e-06, + "loss": 1.9691, + "step": 189 + }, + { + "epoch": 0.19517205957883924, + "grad_norm": 3.5879862308502197, + "learning_rate": 2.9965753424657533e-06, + "loss": 1.9205, + "step": 190 + }, + { + "epoch": 0.19517205957883924, + "eval_loss": 2.13791561126709, + "eval_runtime": 44.8545, + "eval_samples_per_second": 22.294, + "eval_steps_per_second": 0.936, + "step": 190 + }, + { + "epoch": 0.19619928094504366, + "grad_norm": 3.0025253295898438, + "learning_rate": 3.0136986301369864e-06, + "loss": 1.9717, + "step": 191 + }, + { + "epoch": 0.19722650231124808, + "grad_norm": 3.1785972118377686, + "learning_rate": 3.0308219178082194e-06, + "loss": 1.9535, + "step": 192 + }, + { + "epoch": 0.1982537236774525, + "grad_norm": 3.121065378189087, + "learning_rate": 3.0479452054794525e-06, + "loss": 1.9026, + "step": 193 + }, + { + "epoch": 0.1992809450436569, + "grad_norm": 2.838779926300049, + "learning_rate": 3.065068493150685e-06, + "loss": 1.941, + "step": 194 + }, + { + "epoch": 0.20030816640986132, + "grad_norm": 2.8916969299316406, + "learning_rate": 3.082191780821918e-06, + "loss": 1.8983, + "step": 195 + }, + { + "epoch": 0.20133538777606574, + "grad_norm": 2.9195168018341064, + "learning_rate": 3.099315068493151e-06, + "loss": 1.918, + "step": 196 + }, + { + "epoch": 0.20236260914227017, + "grad_norm": 3.235989809036255, + "learning_rate": 3.1164383561643843e-06, + "loss": 1.9111, + "step": 197 + }, + { + "epoch": 0.2033898305084746, + "grad_norm": 3.0316455364227295, + "learning_rate": 3.1335616438356165e-06, + "loss": 1.8839, + "step": 198 + }, + { + "epoch": 0.20441705187467898, + "grad_norm": 2.963610887527466, + "learning_rate": 3.1506849315068495e-06, + "loss": 2.0303, + "step": 199 + }, + { + "epoch": 0.2054442732408834, + "grad_norm": 2.944054365158081, + "learning_rate": 3.167808219178082e-06, + "loss": 1.8822, + "step": 200 + }, + { + "epoch": 0.2054442732408834, + "eval_loss": 2.1433050632476807, + "eval_runtime": 45.8205, + "eval_samples_per_second": 21.824, + "eval_steps_per_second": 0.917, + "step": 200 + } + ], + "logging_steps": 1, + "max_steps": 2919, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 100, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 6.23461881872384e+16, + "train_batch_size": 3, + "trial_name": null, + "trial_params": null +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/zero_to_fp32.py b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-200/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fb0f9a53a507ffcdb60410deb71d60ef801bf350 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/config.json @@ -0,0 +1,36 @@ +{ + "_name_or_path": "meta-llama/Llama-3.2-3B", + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "eos_token_id": 128001, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 3072, + "initializer_range": 0.02, + "intermediate_size": 8192, + "max_position_embeddings": 131072, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 24, + "num_hidden_layers": 28, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": { + "factor": 32.0, + "high_freq_factor": 4.0, + "low_freq_factor": 1.0, + "original_max_position_embeddings": 8192, + "rope_type": "llama3" + }, + "rope_theta": 500000.0, + "tie_word_embeddings": true, + "torch_dtype": "float16", + "transformers_version": "4.45.1", + "use_cache": true, + "vocab_size": 128256 +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/generation_config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..9c389b841a9a29ddd904e02d1eb0e08dcce82ad9 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/generation_config.json @@ -0,0 +1,9 @@ +{ + "_from_model_config": true, + "bos_token_id": 128000, + "do_sample": true, + "eos_token_id": 128001, + "temperature": 0.6, + "top_p": 0.9, + "transformers_version": "4.45.1" +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/latest b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/latest new file mode 100644 index 0000000000000000000000000000000000000000..361f8a1a19370f3a8801a82732f02b0e9e639e14 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/latest @@ -0,0 +1 @@ +global_step2400 \ No newline at end of file diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/model.safetensors.index.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..ed64de846d720b9a7859dc20575fea8e8ca51940 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/model.safetensors.index.json @@ -0,0 +1,262 @@ +{ + "metadata": { + "total_size": 7213504512 + }, + "weight_map": { + "lm_head.weight": "model-00002-of-00002.safetensors", + "model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.norm.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/special_tokens_map.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..6949c5975ee0e961ef61cf31010dce04df0a03f8 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/special_tokens_map.json @@ -0,0 +1,23 @@ +{ + "bos_token": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/tokenizer_config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..0b782d4602b361f7577abb52b710a8c0e088588d --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/tokenizer_config.json @@ -0,0 +1,2078 @@ +{ + "added_tokens_decoder": { + "128000": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128001": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128002": { + "content": "<|reserved_special_token_0|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128003": { + "content": "<|reserved_special_token_1|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128004": { + "content": "<|finetune_right_pad_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128005": { + "content": "<|reserved_special_token_2|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128006": { + "content": "<|start_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128007": { + "content": "<|end_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128008": { + "content": "<|eom_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128009": { + "content": "<|eot_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128010": { + "content": "<|python_tag|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128011": { + "content": "<|reserved_special_token_3|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128012": { + "content": "<|reserved_special_token_4|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128013": { + "content": "<|reserved_special_token_5|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128014": { + "content": "<|reserved_special_token_6|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128015": { + "content": "<|reserved_special_token_7|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128016": { + "content": "<|reserved_special_token_8|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128017": { + "content": "<|reserved_special_token_9|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128018": { + "content": "<|reserved_special_token_10|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128019": { + "content": "<|reserved_special_token_11|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128020": { + "content": "<|reserved_special_token_12|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128021": { + "content": "<|reserved_special_token_13|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128022": { + "content": "<|reserved_special_token_14|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128023": { + "content": "<|reserved_special_token_15|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128024": { + "content": "<|reserved_special_token_16|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128025": { + "content": "<|reserved_special_token_17|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128026": { + "content": "<|reserved_special_token_18|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128027": { + "content": "<|reserved_special_token_19|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128028": { + "content": "<|reserved_special_token_20|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128029": { + "content": "<|reserved_special_token_21|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128030": { + "content": "<|reserved_special_token_22|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128031": { + "content": "<|reserved_special_token_23|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128032": { + "content": "<|reserved_special_token_24|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128033": { + "content": "<|reserved_special_token_25|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128034": { + "content": "<|reserved_special_token_26|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128035": { + "content": "<|reserved_special_token_27|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128036": { + "content": "<|reserved_special_token_28|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128037": { + "content": "<|reserved_special_token_29|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128038": { + "content": "<|reserved_special_token_30|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128039": { + "content": "<|reserved_special_token_31|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128040": { + "content": "<|reserved_special_token_32|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128041": { + "content": "<|reserved_special_token_33|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128042": { + "content": "<|reserved_special_token_34|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128043": { + "content": "<|reserved_special_token_35|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128044": { + "content": "<|reserved_special_token_36|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128045": { + "content": "<|reserved_special_token_37|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128046": { + "content": "<|reserved_special_token_38|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128047": { + "content": "<|reserved_special_token_39|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128048": { + "content": "<|reserved_special_token_40|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128049": { + "content": "<|reserved_special_token_41|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128050": { + "content": "<|reserved_special_token_42|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128051": { + "content": "<|reserved_special_token_43|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128052": { + "content": "<|reserved_special_token_44|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128053": { + "content": "<|reserved_special_token_45|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128054": { + "content": "<|reserved_special_token_46|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128055": { + "content": "<|reserved_special_token_47|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128056": { + "content": "<|reserved_special_token_48|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128057": { + "content": "<|reserved_special_token_49|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128058": { + "content": "<|reserved_special_token_50|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128059": { + "content": "<|reserved_special_token_51|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128060": { + "content": "<|reserved_special_token_52|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128061": { + "content": "<|reserved_special_token_53|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128062": { + "content": "<|reserved_special_token_54|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128063": { + "content": "<|reserved_special_token_55|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128064": { + "content": "<|reserved_special_token_56|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128065": { + "content": "<|reserved_special_token_57|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128066": { + "content": "<|reserved_special_token_58|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128067": { + "content": "<|reserved_special_token_59|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128068": { + "content": "<|reserved_special_token_60|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128069": { + "content": "<|reserved_special_token_61|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128070": { + "content": "<|reserved_special_token_62|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128071": { + "content": "<|reserved_special_token_63|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128072": { + "content": "<|reserved_special_token_64|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128073": { + "content": "<|reserved_special_token_65|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128074": { + "content": "<|reserved_special_token_66|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128075": { + "content": "<|reserved_special_token_67|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128076": { + "content": "<|reserved_special_token_68|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128077": { + "content": "<|reserved_special_token_69|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128078": { + "content": "<|reserved_special_token_70|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128079": { + "content": "<|reserved_special_token_71|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128080": { + "content": "<|reserved_special_token_72|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128081": { + "content": "<|reserved_special_token_73|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128082": { + "content": "<|reserved_special_token_74|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128083": { + "content": "<|reserved_special_token_75|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128084": { + "content": "<|reserved_special_token_76|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128085": { + "content": "<|reserved_special_token_77|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128086": { + "content": "<|reserved_special_token_78|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128087": { + "content": "<|reserved_special_token_79|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128088": { + "content": "<|reserved_special_token_80|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128089": { + "content": "<|reserved_special_token_81|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128090": { + "content": "<|reserved_special_token_82|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128091": { + "content": "<|reserved_special_token_83|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128092": { + "content": "<|reserved_special_token_84|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128093": { + "content": "<|reserved_special_token_85|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128094": { + "content": "<|reserved_special_token_86|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128095": { + "content": "<|reserved_special_token_87|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128096": { + "content": "<|reserved_special_token_88|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128097": { + "content": "<|reserved_special_token_89|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128098": { + "content": "<|reserved_special_token_90|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128099": { + "content": "<|reserved_special_token_91|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128100": { + "content": "<|reserved_special_token_92|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128101": { + "content": "<|reserved_special_token_93|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128102": { + "content": "<|reserved_special_token_94|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128103": { + "content": "<|reserved_special_token_95|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128104": { + "content": "<|reserved_special_token_96|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128105": { + "content": "<|reserved_special_token_97|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128106": { + "content": "<|reserved_special_token_98|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128107": { + "content": "<|reserved_special_token_99|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128108": { + "content": "<|reserved_special_token_100|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128109": { + "content": "<|reserved_special_token_101|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128110": { + "content": "<|reserved_special_token_102|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128111": { + "content": "<|reserved_special_token_103|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128112": { + "content": "<|reserved_special_token_104|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128113": { + "content": "<|reserved_special_token_105|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128114": { + "content": "<|reserved_special_token_106|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128115": { + "content": "<|reserved_special_token_107|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128116": { + "content": "<|reserved_special_token_108|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128117": { + "content": "<|reserved_special_token_109|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128118": { + "content": "<|reserved_special_token_110|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128119": { + "content": "<|reserved_special_token_111|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128120": { + "content": "<|reserved_special_token_112|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128121": { + "content": "<|reserved_special_token_113|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128122": { + "content": "<|reserved_special_token_114|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128123": { + "content": "<|reserved_special_token_115|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128124": { + "content": "<|reserved_special_token_116|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128125": { + "content": "<|reserved_special_token_117|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128126": { + "content": "<|reserved_special_token_118|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128127": { + "content": "<|reserved_special_token_119|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128128": { + "content": "<|reserved_special_token_120|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128129": { + "content": "<|reserved_special_token_121|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128130": { + "content": "<|reserved_special_token_122|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128131": { + "content": "<|reserved_special_token_123|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128132": { + "content": "<|reserved_special_token_124|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128133": { + "content": "<|reserved_special_token_125|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128134": { + "content": "<|reserved_special_token_126|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128135": { + "content": "<|reserved_special_token_127|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128136": { + "content": "<|reserved_special_token_128|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128137": { + "content": "<|reserved_special_token_129|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128138": { + "content": "<|reserved_special_token_130|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128139": { + "content": "<|reserved_special_token_131|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128140": { + "content": "<|reserved_special_token_132|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128141": { + "content": "<|reserved_special_token_133|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128142": { + "content": "<|reserved_special_token_134|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128143": { + "content": "<|reserved_special_token_135|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128144": { + "content": "<|reserved_special_token_136|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128145": { + "content": "<|reserved_special_token_137|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128146": { + "content": "<|reserved_special_token_138|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128147": { + "content": "<|reserved_special_token_139|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128148": { + "content": "<|reserved_special_token_140|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128149": { + "content": "<|reserved_special_token_141|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128150": { + "content": "<|reserved_special_token_142|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128151": { + "content": "<|reserved_special_token_143|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128152": { + "content": "<|reserved_special_token_144|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128153": { + "content": "<|reserved_special_token_145|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128154": { + "content": "<|reserved_special_token_146|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128155": { + "content": "<|reserved_special_token_147|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128156": { + "content": "<|reserved_special_token_148|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128157": { + "content": "<|reserved_special_token_149|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128158": { + "content": "<|reserved_special_token_150|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128159": { + "content": "<|reserved_special_token_151|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128160": { + "content": "<|reserved_special_token_152|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128161": { + "content": "<|reserved_special_token_153|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128162": { + "content": "<|reserved_special_token_154|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128163": { + "content": "<|reserved_special_token_155|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128164": { + "content": "<|reserved_special_token_156|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128165": { + "content": "<|reserved_special_token_157|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128166": { + "content": "<|reserved_special_token_158|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128167": { + "content": "<|reserved_special_token_159|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128168": { + "content": "<|reserved_special_token_160|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128169": { + "content": "<|reserved_special_token_161|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128170": { + "content": "<|reserved_special_token_162|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128171": { + "content": "<|reserved_special_token_163|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128172": { + "content": "<|reserved_special_token_164|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128173": { + "content": "<|reserved_special_token_165|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128174": { + "content": "<|reserved_special_token_166|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128175": { + "content": "<|reserved_special_token_167|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128176": { + "content": "<|reserved_special_token_168|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128177": { + "content": "<|reserved_special_token_169|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128178": { + "content": "<|reserved_special_token_170|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128179": { + "content": "<|reserved_special_token_171|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128180": { + "content": "<|reserved_special_token_172|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128181": { + "content": "<|reserved_special_token_173|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128182": { + "content": "<|reserved_special_token_174|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128183": { + "content": "<|reserved_special_token_175|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128184": { + "content": "<|reserved_special_token_176|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128185": { + "content": "<|reserved_special_token_177|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128186": { + "content": "<|reserved_special_token_178|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128187": { + "content": "<|reserved_special_token_179|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128188": { + "content": "<|reserved_special_token_180|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128189": { + "content": "<|reserved_special_token_181|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128190": { + "content": "<|reserved_special_token_182|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128191": { + "content": "<|reserved_special_token_183|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128192": { + "content": "<|reserved_special_token_184|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128193": { + "content": "<|reserved_special_token_185|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128194": { + "content": "<|reserved_special_token_186|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128195": { + "content": "<|reserved_special_token_187|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128196": { + "content": "<|reserved_special_token_188|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128197": { + "content": "<|reserved_special_token_189|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128198": { + "content": "<|reserved_special_token_190|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128199": { + "content": "<|reserved_special_token_191|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128200": { + "content": "<|reserved_special_token_192|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128201": { + "content": "<|reserved_special_token_193|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128202": { + "content": "<|reserved_special_token_194|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128203": { + "content": "<|reserved_special_token_195|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128204": { + "content": "<|reserved_special_token_196|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128205": { + "content": "<|reserved_special_token_197|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128206": { + "content": "<|reserved_special_token_198|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128207": { + "content": "<|reserved_special_token_199|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128208": { + "content": "<|reserved_special_token_200|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128209": { + "content": "<|reserved_special_token_201|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128210": { + "content": "<|reserved_special_token_202|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128211": { + "content": "<|reserved_special_token_203|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128212": { + "content": "<|reserved_special_token_204|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128213": { + "content": "<|reserved_special_token_205|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128214": { + "content": "<|reserved_special_token_206|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128215": { + "content": "<|reserved_special_token_207|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128216": { + "content": "<|reserved_special_token_208|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128217": { + "content": "<|reserved_special_token_209|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128218": { + "content": "<|reserved_special_token_210|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128219": { + "content": "<|reserved_special_token_211|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128220": { + "content": "<|reserved_special_token_212|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128221": { + "content": "<|reserved_special_token_213|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128222": { + "content": "<|reserved_special_token_214|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128223": { + "content": "<|reserved_special_token_215|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128224": { + "content": "<|reserved_special_token_216|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128225": { + "content": "<|reserved_special_token_217|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128226": { + "content": "<|reserved_special_token_218|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128227": { + "content": "<|reserved_special_token_219|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128228": { + "content": "<|reserved_special_token_220|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128229": { + "content": "<|reserved_special_token_221|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128230": { + "content": "<|reserved_special_token_222|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128231": { + "content": "<|reserved_special_token_223|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128232": { + "content": "<|reserved_special_token_224|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128233": { + "content": "<|reserved_special_token_225|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128234": { + "content": "<|reserved_special_token_226|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128235": { + "content": "<|reserved_special_token_227|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128236": { + "content": "<|reserved_special_token_228|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128237": { + "content": "<|reserved_special_token_229|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128238": { + "content": "<|reserved_special_token_230|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128239": { + "content": "<|reserved_special_token_231|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128240": { + "content": "<|reserved_special_token_232|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128241": { + "content": "<|reserved_special_token_233|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128242": { + "content": "<|reserved_special_token_234|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128243": { + "content": "<|reserved_special_token_235|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128244": { + "content": "<|reserved_special_token_236|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128245": { + "content": "<|reserved_special_token_237|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128246": { + "content": "<|reserved_special_token_238|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128247": { + "content": "<|reserved_special_token_239|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128248": { + "content": "<|reserved_special_token_240|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128249": { + "content": "<|reserved_special_token_241|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128250": { + "content": "<|reserved_special_token_242|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128251": { + "content": "<|reserved_special_token_243|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128252": { + "content": "<|reserved_special_token_244|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128253": { + "content": "<|reserved_special_token_245|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128254": { + "content": "<|reserved_special_token_246|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128255": { + "content": "<|reserved_special_token_247|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128256": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128257": { + "content": "🅁", + "lstrip": true, + "normalized": true, + "rstrip": false, + "single_word": false, + "special": false + } + }, + "bos_token": "<|begin_of_text|>", + "clean_up_tokenization_spaces": true, + "eos_token": "<|end_of_text|>", + "model_input_names": [ + "input_ids", + "attention_mask" + ], + "model_max_length": 131072, + "pad_token": "[PAD]", + "tokenizer_class": "PreTrainedTokenizerFast" +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/trainer_state.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..41423b5ee4206da446b22bb16c19839bbd15aad1 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/trainer_state.json @@ -0,0 +1,18753 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 2.4653312788906008, + "eval_steps": 10, + "global_step": 2400, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.001027221366204417, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7873, + "step": 1 + }, + { + "epoch": 0.002054442732408834, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7904, + "step": 2 + }, + { + "epoch": 0.0030816640986132513, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8348, + "step": 3 + }, + { + "epoch": 0.004108885464817668, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8357, + "step": 4 + }, + { + "epoch": 0.005136106831022085, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8153, + "step": 5 + }, + { + "epoch": 0.0061633281972265025, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.825, + "step": 6 + }, + { + "epoch": 0.007190549563430919, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8002, + "step": 7 + }, + { + "epoch": 0.008217770929635337, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8413, + "step": 8 + }, + { + "epoch": 0.009244992295839754, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8157, + "step": 9 + }, + { + "epoch": 0.01027221366204417, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8249, + "step": 10 + }, + { + "epoch": 0.01027221366204417, + "eval_loss": 2.8518412113189697, + "eval_runtime": 43.212, + "eval_samples_per_second": 23.142, + "eval_steps_per_second": 0.972, + "step": 10 + }, + { + "epoch": 0.011299435028248588, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7829, + "step": 11 + }, + { + "epoch": 0.012326656394453005, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7819, + "step": 12 + }, + { + "epoch": 0.01335387776065742, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8355, + "step": 13 + }, + { + "epoch": 0.014381099126861838, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8138, + "step": 14 + }, + { + "epoch": 0.015408320493066256, + "grad_norm": 5.162191390991211, + "learning_rate": 1.7123287671232876e-08, + "loss": 2.8088, + "step": 15 + }, + { + "epoch": 0.016435541859270673, + "grad_norm": 5.5954084396362305, + "learning_rate": 3.424657534246575e-08, + "loss": 2.7883, + "step": 16 + }, + { + "epoch": 0.01746276322547509, + "grad_norm": 5.542337417602539, + "learning_rate": 5.136986301369863e-08, + "loss": 2.7625, + "step": 17 + }, + { + "epoch": 0.01848998459167951, + "grad_norm": 5.300518989562988, + "learning_rate": 6.84931506849315e-08, + "loss": 2.84, + "step": 18 + }, + { + "epoch": 0.019517205957883924, + "grad_norm": 5.6646881103515625, + "learning_rate": 8.561643835616439e-08, + "loss": 2.761, + "step": 19 + }, + { + "epoch": 0.02054442732408834, + "grad_norm": 5.375557899475098, + "learning_rate": 1.0273972602739726e-07, + "loss": 2.8215, + "step": 20 + }, + { + "epoch": 0.02054442732408834, + "eval_loss": 2.8507144451141357, + "eval_runtime": 45.1133, + "eval_samples_per_second": 22.166, + "eval_steps_per_second": 0.931, + "step": 20 + }, + { + "epoch": 0.02157164869029276, + "grad_norm": 5.1731977462768555, + "learning_rate": 1.1986301369863014e-07, + "loss": 2.8348, + "step": 21 + }, + { + "epoch": 0.022598870056497175, + "grad_norm": 5.370945930480957, + "learning_rate": 1.36986301369863e-07, + "loss": 2.822, + "step": 22 + }, + { + "epoch": 0.02362609142270159, + "grad_norm": 5.550389289855957, + "learning_rate": 1.541095890410959e-07, + "loss": 2.7904, + "step": 23 + }, + { + "epoch": 0.02465331278890601, + "grad_norm": 5.339347839355469, + "learning_rate": 1.7123287671232878e-07, + "loss": 2.8066, + "step": 24 + }, + { + "epoch": 0.025680534155110426, + "grad_norm": 5.2319722175598145, + "learning_rate": 1.8835616438356165e-07, + "loss": 2.7976, + "step": 25 + }, + { + "epoch": 0.02670775552131484, + "grad_norm": 5.207149982452393, + "learning_rate": 2.0547945205479452e-07, + "loss": 2.8162, + "step": 26 + }, + { + "epoch": 0.02773497688751926, + "grad_norm": 4.400204181671143, + "learning_rate": 2.226027397260274e-07, + "loss": 2.7808, + "step": 27 + }, + { + "epoch": 0.028762198253723677, + "grad_norm": 4.21675968170166, + "learning_rate": 2.397260273972603e-07, + "loss": 2.8044, + "step": 28 + }, + { + "epoch": 0.029789419619928096, + "grad_norm": 4.406878471374512, + "learning_rate": 2.568493150684932e-07, + "loss": 2.8162, + "step": 29 + }, + { + "epoch": 0.030816640986132512, + "grad_norm": 4.024835586547852, + "learning_rate": 2.73972602739726e-07, + "loss": 2.7994, + "step": 30 + }, + { + "epoch": 0.030816640986132512, + "eval_loss": 2.8124029636383057, + "eval_runtime": 43.1772, + "eval_samples_per_second": 23.16, + "eval_steps_per_second": 0.973, + "step": 30 + }, + { + "epoch": 0.03184386235233693, + "grad_norm": 3.5378901958465576, + "learning_rate": 2.910958904109589e-07, + "loss": 2.7637, + "step": 31 + }, + { + "epoch": 0.03287108371854135, + "grad_norm": 3.6289916038513184, + "learning_rate": 3.082191780821918e-07, + "loss": 2.8017, + "step": 32 + }, + { + "epoch": 0.03389830508474576, + "grad_norm": 3.6789698600769043, + "learning_rate": 3.2534246575342466e-07, + "loss": 2.7723, + "step": 33 + }, + { + "epoch": 0.03492552645095018, + "grad_norm": 3.491389751434326, + "learning_rate": 3.4246575342465755e-07, + "loss": 2.7528, + "step": 34 + }, + { + "epoch": 0.035952747817154594, + "grad_norm": 3.383080005645752, + "learning_rate": 3.595890410958904e-07, + "loss": 2.7418, + "step": 35 + }, + { + "epoch": 0.03697996918335902, + "grad_norm": 3.4823594093322754, + "learning_rate": 3.767123287671233e-07, + "loss": 2.6662, + "step": 36 + }, + { + "epoch": 0.03800719054956343, + "grad_norm": 3.2146108150482178, + "learning_rate": 3.938356164383562e-07, + "loss": 2.7659, + "step": 37 + }, + { + "epoch": 0.03903441191576785, + "grad_norm": 2.9549412727355957, + "learning_rate": 4.1095890410958903e-07, + "loss": 2.6838, + "step": 38 + }, + { + "epoch": 0.040061633281972264, + "grad_norm": 3.0744080543518066, + "learning_rate": 4.2808219178082193e-07, + "loss": 2.6879, + "step": 39 + }, + { + "epoch": 0.04108885464817668, + "grad_norm": 3.19840669631958, + "learning_rate": 4.452054794520548e-07, + "loss": 2.7443, + "step": 40 + }, + { + "epoch": 0.04108885464817668, + "eval_loss": 2.7469444274902344, + "eval_runtime": 46.4676, + "eval_samples_per_second": 21.52, + "eval_steps_per_second": 0.904, + "step": 40 + }, + { + "epoch": 0.042116076014381096, + "grad_norm": 3.178406000137329, + "learning_rate": 4.6232876712328767e-07, + "loss": 2.7036, + "step": 41 + }, + { + "epoch": 0.04314329738058552, + "grad_norm": 3.2032673358917236, + "learning_rate": 4.794520547945206e-07, + "loss": 2.6465, + "step": 42 + }, + { + "epoch": 0.044170518746789934, + "grad_norm": 3.1525962352752686, + "learning_rate": 4.965753424657534e-07, + "loss": 2.6996, + "step": 43 + }, + { + "epoch": 0.04519774011299435, + "grad_norm": 2.907524585723877, + "learning_rate": 5.136986301369864e-07, + "loss": 2.6581, + "step": 44 + }, + { + "epoch": 0.046224961479198766, + "grad_norm": 2.5885250568389893, + "learning_rate": 5.308219178082192e-07, + "loss": 2.6773, + "step": 45 + }, + { + "epoch": 0.04725218284540318, + "grad_norm": 2.569068431854248, + "learning_rate": 5.47945205479452e-07, + "loss": 2.6635, + "step": 46 + }, + { + "epoch": 0.048279404211607604, + "grad_norm": 2.5313518047332764, + "learning_rate": 5.65068493150685e-07, + "loss": 2.706, + "step": 47 + }, + { + "epoch": 0.04930662557781202, + "grad_norm": 2.615442991256714, + "learning_rate": 5.821917808219178e-07, + "loss": 2.638, + "step": 48 + }, + { + "epoch": 0.050333846944016436, + "grad_norm": 2.7420806884765625, + "learning_rate": 5.993150684931507e-07, + "loss": 2.6398, + "step": 49 + }, + { + "epoch": 0.05136106831022085, + "grad_norm": 2.5246975421905518, + "learning_rate": 6.164383561643836e-07, + "loss": 2.5916, + "step": 50 + }, + { + "epoch": 0.05136106831022085, + "eval_loss": 2.6723692417144775, + "eval_runtime": 46.7206, + "eval_samples_per_second": 21.404, + "eval_steps_per_second": 0.899, + "step": 50 + }, + { + "epoch": 0.05238828967642527, + "grad_norm": 2.390143632888794, + "learning_rate": 6.335616438356165e-07, + "loss": 2.5773, + "step": 51 + }, + { + "epoch": 0.05341551104262968, + "grad_norm": 2.0792484283447266, + "learning_rate": 6.506849315068493e-07, + "loss": 2.6249, + "step": 52 + }, + { + "epoch": 0.054442732408834106, + "grad_norm": 1.9995019435882568, + "learning_rate": 6.678082191780823e-07, + "loss": 2.5425, + "step": 53 + }, + { + "epoch": 0.05546995377503852, + "grad_norm": 1.985452651977539, + "learning_rate": 6.849315068493151e-07, + "loss": 2.5986, + "step": 54 + }, + { + "epoch": 0.05649717514124294, + "grad_norm": 2.10184907913208, + "learning_rate": 7.020547945205481e-07, + "loss": 2.5713, + "step": 55 + }, + { + "epoch": 0.05752439650744735, + "grad_norm": 2.0898985862731934, + "learning_rate": 7.191780821917808e-07, + "loss": 2.535, + "step": 56 + }, + { + "epoch": 0.05855161787365177, + "grad_norm": 2.0239579677581787, + "learning_rate": 7.363013698630137e-07, + "loss": 2.5667, + "step": 57 + }, + { + "epoch": 0.05957883923985619, + "grad_norm": 2.207080125808716, + "learning_rate": 7.534246575342466e-07, + "loss": 2.4999, + "step": 58 + }, + { + "epoch": 0.06060606060606061, + "grad_norm": 1.8541918992996216, + "learning_rate": 7.705479452054795e-07, + "loss": 2.5582, + "step": 59 + }, + { + "epoch": 0.061633281972265024, + "grad_norm": 1.8530476093292236, + "learning_rate": 7.876712328767124e-07, + "loss": 2.5535, + "step": 60 + }, + { + "epoch": 0.061633281972265024, + "eval_loss": 2.607963800430298, + "eval_runtime": 45.9754, + "eval_samples_per_second": 21.751, + "eval_steps_per_second": 0.914, + "step": 60 + }, + { + "epoch": 0.06266050333846944, + "grad_norm": 1.7978090047836304, + "learning_rate": 8.047945205479453e-07, + "loss": 2.53, + "step": 61 + }, + { + "epoch": 0.06368772470467386, + "grad_norm": 1.8173762559890747, + "learning_rate": 8.219178082191781e-07, + "loss": 2.5144, + "step": 62 + }, + { + "epoch": 0.06471494607087827, + "grad_norm": 2.0847485065460205, + "learning_rate": 8.39041095890411e-07, + "loss": 2.5525, + "step": 63 + }, + { + "epoch": 0.0657421674370827, + "grad_norm": 1.8691741228103638, + "learning_rate": 8.561643835616439e-07, + "loss": 2.5451, + "step": 64 + }, + { + "epoch": 0.0667693888032871, + "grad_norm": 2.0242726802825928, + "learning_rate": 8.732876712328768e-07, + "loss": 2.4145, + "step": 65 + }, + { + "epoch": 0.06779661016949153, + "grad_norm": 1.8268908262252808, + "learning_rate": 8.904109589041097e-07, + "loss": 2.5024, + "step": 66 + }, + { + "epoch": 0.06882383153569595, + "grad_norm": 1.8703057765960693, + "learning_rate": 9.075342465753426e-07, + "loss": 2.461, + "step": 67 + }, + { + "epoch": 0.06985105290190036, + "grad_norm": 1.8747144937515259, + "learning_rate": 9.246575342465753e-07, + "loss": 2.5122, + "step": 68 + }, + { + "epoch": 0.07087827426810478, + "grad_norm": 1.744924783706665, + "learning_rate": 9.417808219178083e-07, + "loss": 2.4673, + "step": 69 + }, + { + "epoch": 0.07190549563430919, + "grad_norm": 1.7492460012435913, + "learning_rate": 9.589041095890411e-07, + "loss": 2.5598, + "step": 70 + }, + { + "epoch": 0.07190549563430919, + "eval_loss": 2.547417640686035, + "eval_runtime": 49.7488, + "eval_samples_per_second": 20.101, + "eval_steps_per_second": 0.844, + "step": 70 + }, + { + "epoch": 0.07293271700051361, + "grad_norm": 1.9218230247497559, + "learning_rate": 9.76027397260274e-07, + "loss": 2.4562, + "step": 71 + }, + { + "epoch": 0.07395993836671803, + "grad_norm": 2.0127363204956055, + "learning_rate": 9.931506849315068e-07, + "loss": 2.4141, + "step": 72 + }, + { + "epoch": 0.07498715973292244, + "grad_norm": 1.8700369596481323, + "learning_rate": 1.0102739726027399e-06, + "loss": 2.4696, + "step": 73 + }, + { + "epoch": 0.07601438109912687, + "grad_norm": 2.1706016063690186, + "learning_rate": 1.0273972602739727e-06, + "loss": 2.4404, + "step": 74 + }, + { + "epoch": 0.07704160246533127, + "grad_norm": 2.470412254333496, + "learning_rate": 1.0445205479452056e-06, + "loss": 2.4409, + "step": 75 + }, + { + "epoch": 0.0780688238315357, + "grad_norm": 1.906569242477417, + "learning_rate": 1.0616438356164384e-06, + "loss": 2.4313, + "step": 76 + }, + { + "epoch": 0.07909604519774012, + "grad_norm": 2.122936487197876, + "learning_rate": 1.0787671232876712e-06, + "loss": 2.4369, + "step": 77 + }, + { + "epoch": 0.08012326656394453, + "grad_norm": 2.079127311706543, + "learning_rate": 1.095890410958904e-06, + "loss": 2.449, + "step": 78 + }, + { + "epoch": 0.08115048793014895, + "grad_norm": 1.9173146486282349, + "learning_rate": 1.1130136986301371e-06, + "loss": 2.3782, + "step": 79 + }, + { + "epoch": 0.08217770929635336, + "grad_norm": 1.9856903553009033, + "learning_rate": 1.13013698630137e-06, + "loss": 2.4215, + "step": 80 + }, + { + "epoch": 0.08217770929635336, + "eval_loss": 2.479673385620117, + "eval_runtime": 46.2211, + "eval_samples_per_second": 21.635, + "eval_steps_per_second": 0.909, + "step": 80 + }, + { + "epoch": 0.08320493066255778, + "grad_norm": 2.3764865398406982, + "learning_rate": 1.1472602739726028e-06, + "loss": 2.3744, + "step": 81 + }, + { + "epoch": 0.08423215202876219, + "grad_norm": 1.980668544769287, + "learning_rate": 1.1643835616438357e-06, + "loss": 2.3856, + "step": 82 + }, + { + "epoch": 0.08525937339496661, + "grad_norm": 2.2589704990386963, + "learning_rate": 1.1815068493150685e-06, + "loss": 2.3853, + "step": 83 + }, + { + "epoch": 0.08628659476117104, + "grad_norm": 2.171780824661255, + "learning_rate": 1.1986301369863014e-06, + "loss": 2.3603, + "step": 84 + }, + { + "epoch": 0.08731381612737545, + "grad_norm": 2.25736403465271, + "learning_rate": 1.2157534246575344e-06, + "loss": 2.3189, + "step": 85 + }, + { + "epoch": 0.08834103749357987, + "grad_norm": 2.3643763065338135, + "learning_rate": 1.2328767123287673e-06, + "loss": 2.3795, + "step": 86 + }, + { + "epoch": 0.08936825885978428, + "grad_norm": 2.384516954421997, + "learning_rate": 1.25e-06, + "loss": 2.3257, + "step": 87 + }, + { + "epoch": 0.0903954802259887, + "grad_norm": 2.5072567462921143, + "learning_rate": 1.267123287671233e-06, + "loss": 2.3125, + "step": 88 + }, + { + "epoch": 0.09142270159219312, + "grad_norm": 2.2726807594299316, + "learning_rate": 1.284246575342466e-06, + "loss": 2.3297, + "step": 89 + }, + { + "epoch": 0.09244992295839753, + "grad_norm": 2.4817421436309814, + "learning_rate": 1.3013698630136986e-06, + "loss": 2.2878, + "step": 90 + }, + { + "epoch": 0.09244992295839753, + "eval_loss": 2.4065146446228027, + "eval_runtime": 47.9372, + "eval_samples_per_second": 20.861, + "eval_steps_per_second": 0.876, + "step": 90 + }, + { + "epoch": 0.09347714432460195, + "grad_norm": 2.2219603061676025, + "learning_rate": 1.3184931506849317e-06, + "loss": 2.2978, + "step": 91 + }, + { + "epoch": 0.09450436569080636, + "grad_norm": 2.8790745735168457, + "learning_rate": 1.3356164383561645e-06, + "loss": 2.313, + "step": 92 + }, + { + "epoch": 0.09553158705701079, + "grad_norm": 2.589200496673584, + "learning_rate": 1.3527397260273976e-06, + "loss": 2.2937, + "step": 93 + }, + { + "epoch": 0.09655880842321521, + "grad_norm": 2.760016679763794, + "learning_rate": 1.3698630136986302e-06, + "loss": 2.277, + "step": 94 + }, + { + "epoch": 0.09758602978941962, + "grad_norm": 2.2641494274139404, + "learning_rate": 1.386986301369863e-06, + "loss": 2.3336, + "step": 95 + }, + { + "epoch": 0.09861325115562404, + "grad_norm": 2.2786386013031006, + "learning_rate": 1.4041095890410961e-06, + "loss": 2.233, + "step": 96 + }, + { + "epoch": 0.09964047252182845, + "grad_norm": 2.5590052604675293, + "learning_rate": 1.421232876712329e-06, + "loss": 2.2851, + "step": 97 + }, + { + "epoch": 0.10066769388803287, + "grad_norm": 2.5609471797943115, + "learning_rate": 1.4383561643835616e-06, + "loss": 2.1848, + "step": 98 + }, + { + "epoch": 0.1016949152542373, + "grad_norm": 2.442521810531616, + "learning_rate": 1.4554794520547946e-06, + "loss": 2.2073, + "step": 99 + }, + { + "epoch": 0.1027221366204417, + "grad_norm": 2.3735859394073486, + "learning_rate": 1.4726027397260275e-06, + "loss": 2.2353, + "step": 100 + }, + { + "epoch": 0.1027221366204417, + "eval_loss": 2.3457911014556885, + "eval_runtime": 51.6469, + "eval_samples_per_second": 19.362, + "eval_steps_per_second": 0.813, + "step": 100 + }, + { + "epoch": 0.10374935798664613, + "grad_norm": 2.175354242324829, + "learning_rate": 1.4897260273972605e-06, + "loss": 2.2259, + "step": 101 + }, + { + "epoch": 0.10477657935285054, + "grad_norm": 2.217677116394043, + "learning_rate": 1.5068493150684932e-06, + "loss": 2.2082, + "step": 102 + }, + { + "epoch": 0.10580380071905496, + "grad_norm": 2.4618067741394043, + "learning_rate": 1.5239726027397262e-06, + "loss": 2.1649, + "step": 103 + }, + { + "epoch": 0.10683102208525937, + "grad_norm": 2.232060432434082, + "learning_rate": 1.541095890410959e-06, + "loss": 2.1962, + "step": 104 + }, + { + "epoch": 0.10785824345146379, + "grad_norm": 2.9140872955322266, + "learning_rate": 1.5582191780821921e-06, + "loss": 2.2446, + "step": 105 + }, + { + "epoch": 0.10888546481766821, + "grad_norm": 2.232158660888672, + "learning_rate": 1.5753424657534248e-06, + "loss": 2.2179, + "step": 106 + }, + { + "epoch": 0.10991268618387262, + "grad_norm": 2.804938793182373, + "learning_rate": 1.5924657534246576e-06, + "loss": 2.2449, + "step": 107 + }, + { + "epoch": 0.11093990755007704, + "grad_norm": 2.3690710067749023, + "learning_rate": 1.6095890410958907e-06, + "loss": 2.225, + "step": 108 + }, + { + "epoch": 0.11196712891628145, + "grad_norm": 2.7717416286468506, + "learning_rate": 1.6267123287671235e-06, + "loss": 2.0922, + "step": 109 + }, + { + "epoch": 0.11299435028248588, + "grad_norm": 3.053602933883667, + "learning_rate": 1.6438356164383561e-06, + "loss": 2.1702, + "step": 110 + }, + { + "epoch": 0.11299435028248588, + "eval_loss": 2.298161029815674, + "eval_runtime": 44.0776, + "eval_samples_per_second": 22.687, + "eval_steps_per_second": 0.953, + "step": 110 + }, + { + "epoch": 0.1140215716486903, + "grad_norm": 2.8673532009124756, + "learning_rate": 1.6609589041095892e-06, + "loss": 2.0938, + "step": 111 + }, + { + "epoch": 0.1150487930148947, + "grad_norm": 2.57185959815979, + "learning_rate": 1.678082191780822e-06, + "loss": 2.1892, + "step": 112 + }, + { + "epoch": 0.11607601438109913, + "grad_norm": 2.5278828144073486, + "learning_rate": 1.695205479452055e-06, + "loss": 2.1251, + "step": 113 + }, + { + "epoch": 0.11710323574730354, + "grad_norm": 2.8581748008728027, + "learning_rate": 1.7123287671232877e-06, + "loss": 2.1699, + "step": 114 + }, + { + "epoch": 0.11813045711350796, + "grad_norm": 2.924959897994995, + "learning_rate": 1.7294520547945206e-06, + "loss": 2.1005, + "step": 115 + }, + { + "epoch": 0.11915767847971238, + "grad_norm": 2.477034091949463, + "learning_rate": 1.7465753424657536e-06, + "loss": 2.2016, + "step": 116 + }, + { + "epoch": 0.12018489984591679, + "grad_norm": 2.8719305992126465, + "learning_rate": 1.7636986301369865e-06, + "loss": 2.1588, + "step": 117 + }, + { + "epoch": 0.12121212121212122, + "grad_norm": 2.4403953552246094, + "learning_rate": 1.7808219178082193e-06, + "loss": 2.1445, + "step": 118 + }, + { + "epoch": 0.12223934257832562, + "grad_norm": 2.7994632720947266, + "learning_rate": 1.7979452054794521e-06, + "loss": 2.1369, + "step": 119 + }, + { + "epoch": 0.12326656394453005, + "grad_norm": 2.163196563720703, + "learning_rate": 1.8150684931506852e-06, + "loss": 2.2195, + "step": 120 + }, + { + "epoch": 0.12326656394453005, + "eval_loss": 2.2648775577545166, + "eval_runtime": 49.47, + "eval_samples_per_second": 20.214, + "eval_steps_per_second": 0.849, + "step": 120 + }, + { + "epoch": 0.12429378531073447, + "grad_norm": 2.6743483543395996, + "learning_rate": 1.832191780821918e-06, + "loss": 2.1521, + "step": 121 + }, + { + "epoch": 0.12532100667693888, + "grad_norm": 2.54795503616333, + "learning_rate": 1.8493150684931507e-06, + "loss": 2.1158, + "step": 122 + }, + { + "epoch": 0.1263482280431433, + "grad_norm": 2.9093267917633057, + "learning_rate": 1.8664383561643837e-06, + "loss": 2.0707, + "step": 123 + }, + { + "epoch": 0.12737544940934772, + "grad_norm": 2.416940212249756, + "learning_rate": 1.8835616438356166e-06, + "loss": 2.0901, + "step": 124 + }, + { + "epoch": 0.12840267077555212, + "grad_norm": 2.8333663940429688, + "learning_rate": 1.9006849315068496e-06, + "loss": 2.0545, + "step": 125 + }, + { + "epoch": 0.12942989214175654, + "grad_norm": 2.483051061630249, + "learning_rate": 1.9178082191780823e-06, + "loss": 2.1274, + "step": 126 + }, + { + "epoch": 0.13045711350796096, + "grad_norm": 2.3486859798431396, + "learning_rate": 1.9349315068493153e-06, + "loss": 2.0478, + "step": 127 + }, + { + "epoch": 0.1314843348741654, + "grad_norm": 3.1512887477874756, + "learning_rate": 1.952054794520548e-06, + "loss": 2.1041, + "step": 128 + }, + { + "epoch": 0.1325115562403698, + "grad_norm": 2.902873992919922, + "learning_rate": 1.969178082191781e-06, + "loss": 2.1332, + "step": 129 + }, + { + "epoch": 0.1335387776065742, + "grad_norm": 2.779132127761841, + "learning_rate": 1.9863013698630136e-06, + "loss": 2.1297, + "step": 130 + }, + { + "epoch": 0.1335387776065742, + "eval_loss": 2.2391879558563232, + "eval_runtime": 55.3018, + "eval_samples_per_second": 18.083, + "eval_steps_per_second": 0.759, + "step": 130 + }, + { + "epoch": 0.13456599897277863, + "grad_norm": 2.703019618988037, + "learning_rate": 2.0034246575342467e-06, + "loss": 2.0506, + "step": 131 + }, + { + "epoch": 0.13559322033898305, + "grad_norm": 3.0187575817108154, + "learning_rate": 2.0205479452054797e-06, + "loss": 2.1291, + "step": 132 + }, + { + "epoch": 0.13662044170518747, + "grad_norm": 2.7629971504211426, + "learning_rate": 2.037671232876713e-06, + "loss": 2.1003, + "step": 133 + }, + { + "epoch": 0.1376476630713919, + "grad_norm": 3.175776720046997, + "learning_rate": 2.0547945205479454e-06, + "loss": 2.0429, + "step": 134 + }, + { + "epoch": 0.1386748844375963, + "grad_norm": 3.0342514514923096, + "learning_rate": 2.071917808219178e-06, + "loss": 2.0858, + "step": 135 + }, + { + "epoch": 0.1397021058038007, + "grad_norm": 3.0861928462982178, + "learning_rate": 2.089041095890411e-06, + "loss": 2.0964, + "step": 136 + }, + { + "epoch": 0.14072932717000514, + "grad_norm": 3.151945114135742, + "learning_rate": 2.106164383561644e-06, + "loss": 2.0194, + "step": 137 + }, + { + "epoch": 0.14175654853620956, + "grad_norm": 2.9069676399230957, + "learning_rate": 2.123287671232877e-06, + "loss": 2.1131, + "step": 138 + }, + { + "epoch": 0.14278376990241398, + "grad_norm": 3.2285282611846924, + "learning_rate": 2.14041095890411e-06, + "loss": 2.0643, + "step": 139 + }, + { + "epoch": 0.14381099126861838, + "grad_norm": 2.500481367111206, + "learning_rate": 2.1575342465753425e-06, + "loss": 2.0625, + "step": 140 + }, + { + "epoch": 0.14381099126861838, + "eval_loss": 2.211498975753784, + "eval_runtime": 55.1046, + "eval_samples_per_second": 18.147, + "eval_steps_per_second": 0.762, + "step": 140 + }, + { + "epoch": 0.1448382126348228, + "grad_norm": 4.004592418670654, + "learning_rate": 2.1746575342465755e-06, + "loss": 2.0644, + "step": 141 + }, + { + "epoch": 0.14586543400102722, + "grad_norm": 2.9258391857147217, + "learning_rate": 2.191780821917808e-06, + "loss": 2.0174, + "step": 142 + }, + { + "epoch": 0.14689265536723164, + "grad_norm": 3.0073509216308594, + "learning_rate": 2.2089041095890412e-06, + "loss": 2.1417, + "step": 143 + }, + { + "epoch": 0.14791987673343607, + "grad_norm": 3.040708065032959, + "learning_rate": 2.2260273972602743e-06, + "loss": 2.0357, + "step": 144 + }, + { + "epoch": 0.14894709809964046, + "grad_norm": 2.9062869548797607, + "learning_rate": 2.243150684931507e-06, + "loss": 2.0122, + "step": 145 + }, + { + "epoch": 0.14997431946584489, + "grad_norm": 3.28969669342041, + "learning_rate": 2.26027397260274e-06, + "loss": 2.0255, + "step": 146 + }, + { + "epoch": 0.1510015408320493, + "grad_norm": 3.608752727508545, + "learning_rate": 2.2773972602739726e-06, + "loss": 2.0184, + "step": 147 + }, + { + "epoch": 0.15202876219825373, + "grad_norm": 2.900341033935547, + "learning_rate": 2.2945205479452057e-06, + "loss": 2.0532, + "step": 148 + }, + { + "epoch": 0.15305598356445815, + "grad_norm": 2.881843328475952, + "learning_rate": 2.3116438356164387e-06, + "loss": 2.1336, + "step": 149 + }, + { + "epoch": 0.15408320493066255, + "grad_norm": 3.4043259620666504, + "learning_rate": 2.3287671232876713e-06, + "loss": 2.0536, + "step": 150 + }, + { + "epoch": 0.15408320493066255, + "eval_loss": 2.192880153656006, + "eval_runtime": 53.7838, + "eval_samples_per_second": 18.593, + "eval_steps_per_second": 0.781, + "step": 150 + }, + { + "epoch": 0.15511042629686697, + "grad_norm": 2.693107843399048, + "learning_rate": 2.3458904109589044e-06, + "loss": 1.9503, + "step": 151 + }, + { + "epoch": 0.1561376476630714, + "grad_norm": 2.885260581970215, + "learning_rate": 2.363013698630137e-06, + "loss": 2.0081, + "step": 152 + }, + { + "epoch": 0.15716486902927582, + "grad_norm": 3.4459924697875977, + "learning_rate": 2.38013698630137e-06, + "loss": 2.0649, + "step": 153 + }, + { + "epoch": 0.15819209039548024, + "grad_norm": 2.8447093963623047, + "learning_rate": 2.3972602739726027e-06, + "loss": 2.0476, + "step": 154 + }, + { + "epoch": 0.15921931176168463, + "grad_norm": 2.755176305770874, + "learning_rate": 2.4143835616438358e-06, + "loss": 2.0006, + "step": 155 + }, + { + "epoch": 0.16024653312788906, + "grad_norm": 2.8677561283111572, + "learning_rate": 2.431506849315069e-06, + "loss": 1.9663, + "step": 156 + }, + { + "epoch": 0.16127375449409348, + "grad_norm": 2.783351182937622, + "learning_rate": 2.4486301369863015e-06, + "loss": 2.0301, + "step": 157 + }, + { + "epoch": 0.1623009758602979, + "grad_norm": 2.7545998096466064, + "learning_rate": 2.4657534246575345e-06, + "loss": 2.0438, + "step": 158 + }, + { + "epoch": 0.1633281972265023, + "grad_norm": 2.8140594959259033, + "learning_rate": 2.482876712328767e-06, + "loss": 2.0055, + "step": 159 + }, + { + "epoch": 0.16435541859270672, + "grad_norm": 2.6072335243225098, + "learning_rate": 2.5e-06, + "loss": 2.0262, + "step": 160 + }, + { + "epoch": 0.16435541859270672, + "eval_loss": 2.179842710494995, + "eval_runtime": 52.6159, + "eval_samples_per_second": 19.006, + "eval_steps_per_second": 0.798, + "step": 160 + }, + { + "epoch": 0.16538263995891114, + "grad_norm": 2.830212354660034, + "learning_rate": 2.5171232876712333e-06, + "loss": 2.0154, + "step": 161 + }, + { + "epoch": 0.16640986132511557, + "grad_norm": 2.677140712738037, + "learning_rate": 2.534246575342466e-06, + "loss": 1.9933, + "step": 162 + }, + { + "epoch": 0.16743708269132, + "grad_norm": 2.837759017944336, + "learning_rate": 2.551369863013699e-06, + "loss": 1.9602, + "step": 163 + }, + { + "epoch": 0.16846430405752438, + "grad_norm": 3.9556589126586914, + "learning_rate": 2.568493150684932e-06, + "loss": 2.0121, + "step": 164 + }, + { + "epoch": 0.1694915254237288, + "grad_norm": 2.8715925216674805, + "learning_rate": 2.585616438356164e-06, + "loss": 2.0219, + "step": 165 + }, + { + "epoch": 0.17051874678993323, + "grad_norm": 3.5073623657226562, + "learning_rate": 2.6027397260273973e-06, + "loss": 2.0202, + "step": 166 + }, + { + "epoch": 0.17154596815613765, + "grad_norm": 2.751962900161743, + "learning_rate": 2.6198630136986303e-06, + "loss": 1.9528, + "step": 167 + }, + { + "epoch": 0.17257318952234207, + "grad_norm": 3.7753474712371826, + "learning_rate": 2.6369863013698634e-06, + "loss": 2.0208, + "step": 168 + }, + { + "epoch": 0.17360041088854647, + "grad_norm": 2.642289876937866, + "learning_rate": 2.654109589041096e-06, + "loss": 2.0386, + "step": 169 + }, + { + "epoch": 0.1746276322547509, + "grad_norm": 2.82173490524292, + "learning_rate": 2.671232876712329e-06, + "loss": 2.0355, + "step": 170 + }, + { + "epoch": 0.1746276322547509, + "eval_loss": 2.1699249744415283, + "eval_runtime": 50.0341, + "eval_samples_per_second": 19.986, + "eval_steps_per_second": 0.839, + "step": 170 + }, + { + "epoch": 0.17565485362095531, + "grad_norm": 3.342280149459839, + "learning_rate": 2.688356164383562e-06, + "loss": 1.9672, + "step": 171 + }, + { + "epoch": 0.17668207498715974, + "grad_norm": 2.6548173427581787, + "learning_rate": 2.705479452054795e-06, + "loss": 1.9669, + "step": 172 + }, + { + "epoch": 0.17770929635336416, + "grad_norm": 2.6548173427581787, + "learning_rate": 2.705479452054795e-06, + "loss": 1.961, + "step": 173 + }, + { + "epoch": 0.17873651771956856, + "grad_norm": 3.4306116104125977, + "learning_rate": 2.7226027397260274e-06, + "loss": 1.9173, + "step": 174 + }, + { + "epoch": 0.17976373908577298, + "grad_norm": 2.7991764545440674, + "learning_rate": 2.7397260273972604e-06, + "loss": 2.0339, + "step": 175 + }, + { + "epoch": 0.1807909604519774, + "grad_norm": 3.6040217876434326, + "learning_rate": 2.7568493150684935e-06, + "loss": 2.0218, + "step": 176 + }, + { + "epoch": 0.18181818181818182, + "grad_norm": 3.0253515243530273, + "learning_rate": 2.773972602739726e-06, + "loss": 1.9245, + "step": 177 + }, + { + "epoch": 0.18284540318438625, + "grad_norm": 3.405141830444336, + "learning_rate": 2.791095890410959e-06, + "loss": 2.0136, + "step": 178 + }, + { + "epoch": 0.18387262455059064, + "grad_norm": 2.7518203258514404, + "learning_rate": 2.8082191780821922e-06, + "loss": 1.8925, + "step": 179 + }, + { + "epoch": 0.18489984591679506, + "grad_norm": 3.1903367042541504, + "learning_rate": 2.8253424657534253e-06, + "loss": 2.0523, + "step": 180 + }, + { + "epoch": 0.18489984591679506, + "eval_loss": 2.158029794692993, + "eval_runtime": 48.6162, + "eval_samples_per_second": 20.569, + "eval_steps_per_second": 0.864, + "step": 180 + }, + { + "epoch": 0.1859270672829995, + "grad_norm": 3.0331413745880127, + "learning_rate": 2.842465753424658e-06, + "loss": 2.0167, + "step": 181 + }, + { + "epoch": 0.1869542886492039, + "grad_norm": 2.919433116912842, + "learning_rate": 2.8595890410958905e-06, + "loss": 2.0197, + "step": 182 + }, + { + "epoch": 0.18798151001540833, + "grad_norm": 2.9013049602508545, + "learning_rate": 2.876712328767123e-06, + "loss": 2.0301, + "step": 183 + }, + { + "epoch": 0.18900873138161273, + "grad_norm": 2.738800525665283, + "learning_rate": 2.8938356164383562e-06, + "loss": 1.894, + "step": 184 + }, + { + "epoch": 0.19003595274781715, + "grad_norm": 2.670806407928467, + "learning_rate": 2.9109589041095893e-06, + "loss": 1.9127, + "step": 185 + }, + { + "epoch": 0.19106317411402157, + "grad_norm": 3.0435004234313965, + "learning_rate": 2.9280821917808223e-06, + "loss": 1.8996, + "step": 186 + }, + { + "epoch": 0.192090395480226, + "grad_norm": 3.2139949798583984, + "learning_rate": 2.945205479452055e-06, + "loss": 1.9255, + "step": 187 + }, + { + "epoch": 0.19311761684643042, + "grad_norm": 3.252965211868286, + "learning_rate": 2.962328767123288e-06, + "loss": 1.9809, + "step": 188 + }, + { + "epoch": 0.1941448382126348, + "grad_norm": 2.9325191974639893, + "learning_rate": 2.979452054794521e-06, + "loss": 1.9691, + "step": 189 + }, + { + "epoch": 0.19517205957883924, + "grad_norm": 3.5879862308502197, + "learning_rate": 2.9965753424657533e-06, + "loss": 1.9205, + "step": 190 + }, + { + "epoch": 0.19517205957883924, + "eval_loss": 2.13791561126709, + "eval_runtime": 44.8545, + "eval_samples_per_second": 22.294, + "eval_steps_per_second": 0.936, + "step": 190 + }, + { + "epoch": 0.19619928094504366, + "grad_norm": 3.0025253295898438, + "learning_rate": 3.0136986301369864e-06, + "loss": 1.9717, + "step": 191 + }, + { + "epoch": 0.19722650231124808, + "grad_norm": 3.1785972118377686, + "learning_rate": 3.0308219178082194e-06, + "loss": 1.9535, + "step": 192 + }, + { + "epoch": 0.1982537236774525, + "grad_norm": 3.121065378189087, + "learning_rate": 3.0479452054794525e-06, + "loss": 1.9026, + "step": 193 + }, + { + "epoch": 0.1992809450436569, + "grad_norm": 2.838779926300049, + "learning_rate": 3.065068493150685e-06, + "loss": 1.941, + "step": 194 + }, + { + "epoch": 0.20030816640986132, + "grad_norm": 2.8916969299316406, + "learning_rate": 3.082191780821918e-06, + "loss": 1.8983, + "step": 195 + }, + { + "epoch": 0.20133538777606574, + "grad_norm": 2.9195168018341064, + "learning_rate": 3.099315068493151e-06, + "loss": 1.918, + "step": 196 + }, + { + "epoch": 0.20236260914227017, + "grad_norm": 3.235989809036255, + "learning_rate": 3.1164383561643843e-06, + "loss": 1.9111, + "step": 197 + }, + { + "epoch": 0.2033898305084746, + "grad_norm": 3.0316455364227295, + "learning_rate": 3.1335616438356165e-06, + "loss": 1.8839, + "step": 198 + }, + { + "epoch": 0.20441705187467898, + "grad_norm": 2.963610887527466, + "learning_rate": 3.1506849315068495e-06, + "loss": 2.0303, + "step": 199 + }, + { + "epoch": 0.2054442732408834, + "grad_norm": 2.944054365158081, + "learning_rate": 3.167808219178082e-06, + "loss": 1.8822, + "step": 200 + }, + { + "epoch": 0.2054442732408834, + "eval_loss": 2.1433050632476807, + "eval_runtime": 45.8205, + "eval_samples_per_second": 21.824, + "eval_steps_per_second": 0.917, + "step": 200 + }, + { + "epoch": 0.20647149460708783, + "grad_norm": 2.8648695945739746, + "learning_rate": 3.184931506849315e-06, + "loss": 1.8841, + "step": 201 + }, + { + "epoch": 0.20749871597329225, + "grad_norm": 3.2838797569274902, + "learning_rate": 3.2020547945205483e-06, + "loss": 1.8623, + "step": 202 + }, + { + "epoch": 0.20852593733949668, + "grad_norm": 2.7965173721313477, + "learning_rate": 3.2191780821917813e-06, + "loss": 1.8521, + "step": 203 + }, + { + "epoch": 0.20955315870570107, + "grad_norm": 3.3449349403381348, + "learning_rate": 3.236301369863014e-06, + "loss": 1.84, + "step": 204 + }, + { + "epoch": 0.2105803800719055, + "grad_norm": 3.4888951778411865, + "learning_rate": 3.253424657534247e-06, + "loss": 1.824, + "step": 205 + }, + { + "epoch": 0.21160760143810992, + "grad_norm": 2.991349220275879, + "learning_rate": 3.2705479452054796e-06, + "loss": 1.9101, + "step": 206 + }, + { + "epoch": 0.21263482280431434, + "grad_norm": 3.313509702682495, + "learning_rate": 3.2876712328767123e-06, + "loss": 1.8912, + "step": 207 + }, + { + "epoch": 0.21366204417051873, + "grad_norm": 2.8260746002197266, + "learning_rate": 3.3047945205479453e-06, + "loss": 1.8555, + "step": 208 + }, + { + "epoch": 0.21468926553672316, + "grad_norm": 2.7131688594818115, + "learning_rate": 3.3219178082191784e-06, + "loss": 1.8693, + "step": 209 + }, + { + "epoch": 0.21571648690292758, + "grad_norm": 2.5836338996887207, + "learning_rate": 3.3390410958904114e-06, + "loss": 1.8965, + "step": 210 + }, + { + "epoch": 0.21571648690292758, + "eval_loss": 2.1232542991638184, + "eval_runtime": 42.3352, + "eval_samples_per_second": 23.621, + "eval_steps_per_second": 0.992, + "step": 210 + }, + { + "epoch": 0.216743708269132, + "grad_norm": 2.8341355323791504, + "learning_rate": 3.356164383561644e-06, + "loss": 1.9131, + "step": 211 + }, + { + "epoch": 0.21777092963533642, + "grad_norm": 2.490853786468506, + "learning_rate": 3.373287671232877e-06, + "loss": 1.9626, + "step": 212 + }, + { + "epoch": 0.21879815100154082, + "grad_norm": 2.597956895828247, + "learning_rate": 3.39041095890411e-06, + "loss": 1.9203, + "step": 213 + }, + { + "epoch": 0.21982537236774524, + "grad_norm": 3.1104283332824707, + "learning_rate": 3.4075342465753424e-06, + "loss": 1.9442, + "step": 214 + }, + { + "epoch": 0.22085259373394966, + "grad_norm": 2.5698931217193604, + "learning_rate": 3.4246575342465754e-06, + "loss": 1.8385, + "step": 215 + }, + { + "epoch": 0.2218798151001541, + "grad_norm": 3.1637446880340576, + "learning_rate": 3.4417808219178085e-06, + "loss": 1.9005, + "step": 216 + }, + { + "epoch": 0.2229070364663585, + "grad_norm": 2.764070749282837, + "learning_rate": 3.458904109589041e-06, + "loss": 1.8886, + "step": 217 + }, + { + "epoch": 0.2239342578325629, + "grad_norm": 3.303781032562256, + "learning_rate": 3.476027397260274e-06, + "loss": 1.9085, + "step": 218 + }, + { + "epoch": 0.22496147919876733, + "grad_norm": 2.9543044567108154, + "learning_rate": 3.4931506849315072e-06, + "loss": 1.9014, + "step": 219 + }, + { + "epoch": 0.22598870056497175, + "grad_norm": 2.7277028560638428, + "learning_rate": 3.5102739726027403e-06, + "loss": 1.9017, + "step": 220 + }, + { + "epoch": 0.22598870056497175, + "eval_loss": 2.1024081707000732, + "eval_runtime": 45.4966, + "eval_samples_per_second": 21.98, + "eval_steps_per_second": 0.923, + "step": 220 + }, + { + "epoch": 0.22701592193117617, + "grad_norm": 2.864140033721924, + "learning_rate": 3.527397260273973e-06, + "loss": 1.8935, + "step": 221 + }, + { + "epoch": 0.2280431432973806, + "grad_norm": 3.4133172035217285, + "learning_rate": 3.5445205479452056e-06, + "loss": 1.874, + "step": 222 + }, + { + "epoch": 0.229070364663585, + "grad_norm": 3.33343243598938, + "learning_rate": 3.5616438356164386e-06, + "loss": 1.8789, + "step": 223 + }, + { + "epoch": 0.2300975860297894, + "grad_norm": 3.5947320461273193, + "learning_rate": 3.5787671232876712e-06, + "loss": 1.9529, + "step": 224 + }, + { + "epoch": 0.23112480739599384, + "grad_norm": 3.2194948196411133, + "learning_rate": 3.5958904109589043e-06, + "loss": 1.9205, + "step": 225 + }, + { + "epoch": 0.23215202876219826, + "grad_norm": 3.3137149810791016, + "learning_rate": 3.6130136986301373e-06, + "loss": 1.8584, + "step": 226 + }, + { + "epoch": 0.23317925012840268, + "grad_norm": 3.3358561992645264, + "learning_rate": 3.6301369863013704e-06, + "loss": 1.8817, + "step": 227 + }, + { + "epoch": 0.23420647149460708, + "grad_norm": 3.8612711429595947, + "learning_rate": 3.647260273972603e-06, + "loss": 1.9189, + "step": 228 + }, + { + "epoch": 0.2352336928608115, + "grad_norm": 3.0039336681365967, + "learning_rate": 3.664383561643836e-06, + "loss": 1.92, + "step": 229 + }, + { + "epoch": 0.23626091422701592, + "grad_norm": 3.8983118534088135, + "learning_rate": 3.6815068493150687e-06, + "loss": 1.8512, + "step": 230 + }, + { + "epoch": 0.23626091422701592, + "eval_loss": 2.1237711906433105, + "eval_runtime": 45.6126, + "eval_samples_per_second": 21.924, + "eval_steps_per_second": 0.921, + "step": 230 + }, + { + "epoch": 0.23728813559322035, + "grad_norm": 3.061300039291382, + "learning_rate": 3.6986301369863014e-06, + "loss": 1.8772, + "step": 231 + }, + { + "epoch": 0.23831535695942477, + "grad_norm": 3.3102450370788574, + "learning_rate": 3.7157534246575344e-06, + "loss": 1.8395, + "step": 232 + }, + { + "epoch": 0.23934257832562916, + "grad_norm": 2.9634461402893066, + "learning_rate": 3.7328767123287675e-06, + "loss": 1.9447, + "step": 233 + }, + { + "epoch": 0.24036979969183359, + "grad_norm": 3.0937695503234863, + "learning_rate": 3.7500000000000005e-06, + "loss": 1.8628, + "step": 234 + }, + { + "epoch": 0.241397021058038, + "grad_norm": 2.8129842281341553, + "learning_rate": 3.767123287671233e-06, + "loss": 1.8891, + "step": 235 + }, + { + "epoch": 0.24242424242424243, + "grad_norm": 2.9616780281066895, + "learning_rate": 3.784246575342466e-06, + "loss": 1.9112, + "step": 236 + }, + { + "epoch": 0.24345146379044685, + "grad_norm": 2.837721109390259, + "learning_rate": 3.8013698630136993e-06, + "loss": 1.8514, + "step": 237 + }, + { + "epoch": 0.24447868515665125, + "grad_norm": 3.137784004211426, + "learning_rate": 3.818493150684932e-06, + "loss": 1.8764, + "step": 238 + }, + { + "epoch": 0.24550590652285567, + "grad_norm": 2.9692037105560303, + "learning_rate": 3.8356164383561645e-06, + "loss": 1.8201, + "step": 239 + }, + { + "epoch": 0.2465331278890601, + "grad_norm": 2.6718621253967285, + "learning_rate": 3.852739726027397e-06, + "loss": 1.9309, + "step": 240 + }, + { + "epoch": 0.2465331278890601, + "eval_loss": 2.101331949234009, + "eval_runtime": 45.6346, + "eval_samples_per_second": 21.913, + "eval_steps_per_second": 0.92, + "step": 240 + }, + { + "epoch": 0.24756034925526452, + "grad_norm": 3.411193609237671, + "learning_rate": 3.869863013698631e-06, + "loss": 1.9171, + "step": 241 + }, + { + "epoch": 0.24858757062146894, + "grad_norm": 2.8848962783813477, + "learning_rate": 3.886986301369863e-06, + "loss": 1.8587, + "step": 242 + }, + { + "epoch": 0.24961479198767333, + "grad_norm": 2.756298065185547, + "learning_rate": 3.904109589041096e-06, + "loss": 1.8614, + "step": 243 + }, + { + "epoch": 0.25064201335387776, + "grad_norm": 2.8048956394195557, + "learning_rate": 3.921232876712329e-06, + "loss": 1.8225, + "step": 244 + }, + { + "epoch": 0.2516692347200822, + "grad_norm": 2.9872403144836426, + "learning_rate": 3.938356164383562e-06, + "loss": 1.8367, + "step": 245 + }, + { + "epoch": 0.2526964560862866, + "grad_norm": 2.710156202316284, + "learning_rate": 3.9554794520547955e-06, + "loss": 1.809, + "step": 246 + }, + { + "epoch": 0.253723677452491, + "grad_norm": 3.2177000045776367, + "learning_rate": 3.972602739726027e-06, + "loss": 1.8883, + "step": 247 + }, + { + "epoch": 0.25475089881869545, + "grad_norm": 3.160964250564575, + "learning_rate": 3.989726027397261e-06, + "loss": 1.8267, + "step": 248 + }, + { + "epoch": 0.25577812018489987, + "grad_norm": 2.853498935699463, + "learning_rate": 4.006849315068493e-06, + "loss": 1.8852, + "step": 249 + }, + { + "epoch": 0.25680534155110424, + "grad_norm": 2.571742296218872, + "learning_rate": 4.023972602739726e-06, + "loss": 1.8833, + "step": 250 + }, + { + "epoch": 0.25680534155110424, + "eval_loss": 2.0988597869873047, + "eval_runtime": 47.7045, + "eval_samples_per_second": 20.962, + "eval_steps_per_second": 0.88, + "step": 250 + }, + { + "epoch": 0.25783256291730866, + "grad_norm": 2.776742458343506, + "learning_rate": 4.0410958904109595e-06, + "loss": 1.8662, + "step": 251 + }, + { + "epoch": 0.2588597842835131, + "grad_norm": 2.8038573265075684, + "learning_rate": 4.058219178082192e-06, + "loss": 1.8528, + "step": 252 + }, + { + "epoch": 0.2598870056497175, + "grad_norm": 2.557623863220215, + "learning_rate": 4.075342465753426e-06, + "loss": 1.8765, + "step": 253 + }, + { + "epoch": 0.26091422701592193, + "grad_norm": 3.0477073192596436, + "learning_rate": 4.092465753424658e-06, + "loss": 1.8434, + "step": 254 + }, + { + "epoch": 0.26194144838212635, + "grad_norm": 2.6221649646759033, + "learning_rate": 4.109589041095891e-06, + "loss": 1.8348, + "step": 255 + }, + { + "epoch": 0.2629686697483308, + "grad_norm": 3.0849292278289795, + "learning_rate": 4.1267123287671235e-06, + "loss": 1.9057, + "step": 256 + }, + { + "epoch": 0.2639958911145352, + "grad_norm": 2.795184373855591, + "learning_rate": 4.143835616438356e-06, + "loss": 1.8528, + "step": 257 + }, + { + "epoch": 0.2650231124807396, + "grad_norm": 2.887193441390991, + "learning_rate": 4.16095890410959e-06, + "loss": 1.9176, + "step": 258 + }, + { + "epoch": 0.26605033384694404, + "grad_norm": 2.776801347732544, + "learning_rate": 4.178082191780822e-06, + "loss": 1.8911, + "step": 259 + }, + { + "epoch": 0.2670775552131484, + "grad_norm": 2.980426788330078, + "learning_rate": 4.195205479452055e-06, + "loss": 1.8288, + "step": 260 + }, + { + "epoch": 0.2670775552131484, + "eval_loss": 2.0636370182037354, + "eval_runtime": 49.3081, + "eval_samples_per_second": 20.281, + "eval_steps_per_second": 0.852, + "step": 260 + }, + { + "epoch": 0.26810477657935283, + "grad_norm": 2.7921340465545654, + "learning_rate": 4.212328767123288e-06, + "loss": 1.8405, + "step": 261 + }, + { + "epoch": 0.26913199794555726, + "grad_norm": 2.67714262008667, + "learning_rate": 4.229452054794521e-06, + "loss": 1.8342, + "step": 262 + }, + { + "epoch": 0.2701592193117617, + "grad_norm": 2.819868326187134, + "learning_rate": 4.246575342465754e-06, + "loss": 1.8466, + "step": 263 + }, + { + "epoch": 0.2711864406779661, + "grad_norm": 2.611250638961792, + "learning_rate": 4.263698630136986e-06, + "loss": 1.8314, + "step": 264 + }, + { + "epoch": 0.2722136620441705, + "grad_norm": 3.0406076908111572, + "learning_rate": 4.28082191780822e-06, + "loss": 1.7924, + "step": 265 + }, + { + "epoch": 0.27324088341037495, + "grad_norm": 3.212790012359619, + "learning_rate": 4.297945205479452e-06, + "loss": 1.9479, + "step": 266 + }, + { + "epoch": 0.27426810477657937, + "grad_norm": 2.4767045974731445, + "learning_rate": 4.315068493150685e-06, + "loss": 1.835, + "step": 267 + }, + { + "epoch": 0.2752953261427838, + "grad_norm": 3.191835641860962, + "learning_rate": 4.3321917808219185e-06, + "loss": 1.8261, + "step": 268 + }, + { + "epoch": 0.2763225475089882, + "grad_norm": 2.9970273971557617, + "learning_rate": 4.349315068493151e-06, + "loss": 1.8427, + "step": 269 + }, + { + "epoch": 0.2773497688751926, + "grad_norm": 2.888843297958374, + "learning_rate": 4.3664383561643846e-06, + "loss": 1.819, + "step": 270 + }, + { + "epoch": 0.2773497688751926, + "eval_loss": 2.054966926574707, + "eval_runtime": 45.9212, + "eval_samples_per_second": 21.776, + "eval_steps_per_second": 0.915, + "step": 270 + }, + { + "epoch": 0.278376990241397, + "grad_norm": 2.7771849632263184, + "learning_rate": 4.383561643835616e-06, + "loss": 1.8178, + "step": 271 + }, + { + "epoch": 0.2794042116076014, + "grad_norm": 2.9749014377593994, + "learning_rate": 4.40068493150685e-06, + "loss": 1.8982, + "step": 272 + }, + { + "epoch": 0.28043143297380585, + "grad_norm": 3.076021432876587, + "learning_rate": 4.4178082191780825e-06, + "loss": 1.8575, + "step": 273 + }, + { + "epoch": 0.2814586543400103, + "grad_norm": 3.2113568782806396, + "learning_rate": 4.434931506849315e-06, + "loss": 1.8241, + "step": 274 + }, + { + "epoch": 0.2824858757062147, + "grad_norm": 3.322063684463501, + "learning_rate": 4.4520547945205486e-06, + "loss": 1.7922, + "step": 275 + }, + { + "epoch": 0.2835130970724191, + "grad_norm": 3.2778918743133545, + "learning_rate": 4.469178082191781e-06, + "loss": 1.8412, + "step": 276 + }, + { + "epoch": 0.28454031843862354, + "grad_norm": 3.1802618503570557, + "learning_rate": 4.486301369863014e-06, + "loss": 1.8283, + "step": 277 + }, + { + "epoch": 0.28556753980482796, + "grad_norm": 2.9214258193969727, + "learning_rate": 4.503424657534247e-06, + "loss": 1.8817, + "step": 278 + }, + { + "epoch": 0.28659476117103233, + "grad_norm": 2.8809854984283447, + "learning_rate": 4.52054794520548e-06, + "loss": 1.7662, + "step": 279 + }, + { + "epoch": 0.28762198253723675, + "grad_norm": 2.925537586212158, + "learning_rate": 4.537671232876713e-06, + "loss": 1.8147, + "step": 280 + }, + { + "epoch": 0.28762198253723675, + "eval_loss": 2.054513454437256, + "eval_runtime": 51.5036, + "eval_samples_per_second": 19.416, + "eval_steps_per_second": 0.815, + "step": 280 + }, + { + "epoch": 0.2886492039034412, + "grad_norm": 2.9421887397766113, + "learning_rate": 4.554794520547945e-06, + "loss": 1.8581, + "step": 281 + }, + { + "epoch": 0.2896764252696456, + "grad_norm": 2.9733567237854004, + "learning_rate": 4.571917808219179e-06, + "loss": 1.8703, + "step": 282 + }, + { + "epoch": 0.29070364663585, + "grad_norm": 2.5352156162261963, + "learning_rate": 4.589041095890411e-06, + "loss": 1.8505, + "step": 283 + }, + { + "epoch": 0.29173086800205444, + "grad_norm": 2.773847818374634, + "learning_rate": 4.606164383561644e-06, + "loss": 1.8114, + "step": 284 + }, + { + "epoch": 0.29275808936825887, + "grad_norm": 2.709632635116577, + "learning_rate": 4.6232876712328774e-06, + "loss": 1.815, + "step": 285 + }, + { + "epoch": 0.2937853107344633, + "grad_norm": 3.1234357357025146, + "learning_rate": 4.64041095890411e-06, + "loss": 1.8391, + "step": 286 + }, + { + "epoch": 0.2948125321006677, + "grad_norm": 2.9376144409179688, + "learning_rate": 4.657534246575343e-06, + "loss": 1.7998, + "step": 287 + }, + { + "epoch": 0.29583975346687214, + "grad_norm": 3.0812032222747803, + "learning_rate": 4.674657534246575e-06, + "loss": 1.8462, + "step": 288 + }, + { + "epoch": 0.2968669748330765, + "grad_norm": 2.922722578048706, + "learning_rate": 4.691780821917809e-06, + "loss": 1.8263, + "step": 289 + }, + { + "epoch": 0.2978941961992809, + "grad_norm": 2.9484992027282715, + "learning_rate": 4.7089041095890414e-06, + "loss": 1.8754, + "step": 290 + }, + { + "epoch": 0.2978941961992809, + "eval_loss": 2.032902240753174, + "eval_runtime": 47.8711, + "eval_samples_per_second": 20.889, + "eval_steps_per_second": 0.877, + "step": 290 + }, + { + "epoch": 0.29892141756548535, + "grad_norm": 3.0068976879119873, + "learning_rate": 4.726027397260274e-06, + "loss": 1.8256, + "step": 291 + }, + { + "epoch": 0.29994863893168977, + "grad_norm": 2.6469006538391113, + "learning_rate": 4.7431506849315075e-06, + "loss": 1.8354, + "step": 292 + }, + { + "epoch": 0.3009758602978942, + "grad_norm": 3.06986927986145, + "learning_rate": 4.76027397260274e-06, + "loss": 1.8402, + "step": 293 + }, + { + "epoch": 0.3020030816640986, + "grad_norm": 2.8409981727600098, + "learning_rate": 4.777397260273973e-06, + "loss": 1.7915, + "step": 294 + }, + { + "epoch": 0.30303030303030304, + "grad_norm": 3.1027376651763916, + "learning_rate": 4.7945205479452054e-06, + "loss": 1.8153, + "step": 295 + }, + { + "epoch": 0.30405752439650746, + "grad_norm": 2.7221434116363525, + "learning_rate": 4.811643835616439e-06, + "loss": 1.8098, + "step": 296 + }, + { + "epoch": 0.3050847457627119, + "grad_norm": 3.0559370517730713, + "learning_rate": 4.8287671232876716e-06, + "loss": 1.7504, + "step": 297 + }, + { + "epoch": 0.3061119671289163, + "grad_norm": 2.9628236293792725, + "learning_rate": 4.845890410958904e-06, + "loss": 1.8019, + "step": 298 + }, + { + "epoch": 0.3071391884951207, + "grad_norm": 2.6248221397399902, + "learning_rate": 4.863013698630138e-06, + "loss": 1.8001, + "step": 299 + }, + { + "epoch": 0.3081664098613251, + "grad_norm": 3.0477702617645264, + "learning_rate": 4.88013698630137e-06, + "loss": 1.7915, + "step": 300 + }, + { + "epoch": 0.3081664098613251, + "eval_loss": 2.0522429943084717, + "eval_runtime": 46.796, + "eval_samples_per_second": 21.369, + "eval_steps_per_second": 0.898, + "step": 300 + }, + { + "epoch": 0.3091936312275295, + "grad_norm": 2.784707546234131, + "learning_rate": 4.897260273972603e-06, + "loss": 1.7757, + "step": 301 + }, + { + "epoch": 0.31022085259373394, + "grad_norm": 2.8283841609954834, + "learning_rate": 4.914383561643836e-06, + "loss": 1.8975, + "step": 302 + }, + { + "epoch": 0.31124807395993837, + "grad_norm": 2.879342555999756, + "learning_rate": 4.931506849315069e-06, + "loss": 1.8208, + "step": 303 + }, + { + "epoch": 0.3122752953261428, + "grad_norm": 3.2111175060272217, + "learning_rate": 4.948630136986302e-06, + "loss": 1.8631, + "step": 304 + }, + { + "epoch": 0.3133025166923472, + "grad_norm": 2.876168966293335, + "learning_rate": 4.965753424657534e-06, + "loss": 1.8863, + "step": 305 + }, + { + "epoch": 0.31432973805855163, + "grad_norm": 2.8958466053009033, + "learning_rate": 4.982876712328768e-06, + "loss": 1.7565, + "step": 306 + }, + { + "epoch": 0.31535695942475606, + "grad_norm": 2.6702723503112793, + "learning_rate": 5e-06, + "loss": 1.7844, + "step": 307 + }, + { + "epoch": 0.3163841807909605, + "grad_norm": 2.557966470718384, + "learning_rate": 4.998096688237534e-06, + "loss": 1.7472, + "step": 308 + }, + { + "epoch": 0.31741140215716485, + "grad_norm": 2.9257633686065674, + "learning_rate": 4.996193376475067e-06, + "loss": 1.853, + "step": 309 + }, + { + "epoch": 0.31843862352336927, + "grad_norm": 2.497093439102173, + "learning_rate": 4.9942900647126005e-06, + "loss": 1.7313, + "step": 310 + }, + { + "epoch": 0.31843862352336927, + "eval_loss": 2.0336520671844482, + "eval_runtime": 43.3773, + "eval_samples_per_second": 23.054, + "eval_steps_per_second": 0.968, + "step": 310 + }, + { + "epoch": 0.3194658448895737, + "grad_norm": 3.0313165187835693, + "learning_rate": 4.992386752950134e-06, + "loss": 1.7397, + "step": 311 + }, + { + "epoch": 0.3204930662557781, + "grad_norm": 2.629685163497925, + "learning_rate": 4.990483441187667e-06, + "loss": 1.7839, + "step": 312 + }, + { + "epoch": 0.32152028762198254, + "grad_norm": 2.6772103309631348, + "learning_rate": 4.9885801294252e-06, + "loss": 1.8038, + "step": 313 + }, + { + "epoch": 0.32254750898818696, + "grad_norm": 2.893958568572998, + "learning_rate": 4.986676817662733e-06, + "loss": 1.8072, + "step": 314 + }, + { + "epoch": 0.3235747303543914, + "grad_norm": 2.5317318439483643, + "learning_rate": 4.9847735059002665e-06, + "loss": 1.7632, + "step": 315 + }, + { + "epoch": 0.3246019517205958, + "grad_norm": 2.836867332458496, + "learning_rate": 4.9828701941378e-06, + "loss": 1.8499, + "step": 316 + }, + { + "epoch": 0.3256291730868002, + "grad_norm": 2.5564308166503906, + "learning_rate": 4.980966882375333e-06, + "loss": 1.7849, + "step": 317 + }, + { + "epoch": 0.3266563944530046, + "grad_norm": 2.4335806369781494, + "learning_rate": 4.979063570612867e-06, + "loss": 1.7425, + "step": 318 + }, + { + "epoch": 0.327683615819209, + "grad_norm": 2.7140371799468994, + "learning_rate": 4.9771602588504e-06, + "loss": 1.7173, + "step": 319 + }, + { + "epoch": 0.32871083718541344, + "grad_norm": 2.574345111846924, + "learning_rate": 4.975256947087933e-06, + "loss": 1.8307, + "step": 320 + }, + { + "epoch": 0.32871083718541344, + "eval_loss": 2.026960611343384, + "eval_runtime": 53.8982, + "eval_samples_per_second": 18.553, + "eval_steps_per_second": 0.779, + "step": 320 + }, + { + "epoch": 0.32973805855161786, + "grad_norm": 2.6877281665802, + "learning_rate": 4.973353635325467e-06, + "loss": 1.7216, + "step": 321 + }, + { + "epoch": 0.3307652799178223, + "grad_norm": 2.8152849674224854, + "learning_rate": 4.971450323563e-06, + "loss": 1.7831, + "step": 322 + }, + { + "epoch": 0.3317925012840267, + "grad_norm": 2.4208791255950928, + "learning_rate": 4.969547011800533e-06, + "loss": 1.7266, + "step": 323 + }, + { + "epoch": 0.33281972265023113, + "grad_norm": 2.765162229537964, + "learning_rate": 4.967643700038067e-06, + "loss": 1.7971, + "step": 324 + }, + { + "epoch": 0.33384694401643555, + "grad_norm": 2.5018413066864014, + "learning_rate": 4.9657403882756e-06, + "loss": 1.7842, + "step": 325 + }, + { + "epoch": 0.33487416538264, + "grad_norm": 2.901268482208252, + "learning_rate": 4.9638370765131335e-06, + "loss": 1.7522, + "step": 326 + }, + { + "epoch": 0.3359013867488444, + "grad_norm": 2.40067195892334, + "learning_rate": 4.961933764750667e-06, + "loss": 1.7371, + "step": 327 + }, + { + "epoch": 0.33692860811504877, + "grad_norm": 2.6963725090026855, + "learning_rate": 4.9600304529882e-06, + "loss": 1.8036, + "step": 328 + }, + { + "epoch": 0.3379558294812532, + "grad_norm": 2.483790397644043, + "learning_rate": 4.9581271412257335e-06, + "loss": 1.815, + "step": 329 + }, + { + "epoch": 0.3389830508474576, + "grad_norm": 2.7459559440612793, + "learning_rate": 4.956223829463266e-06, + "loss": 1.777, + "step": 330 + }, + { + "epoch": 0.3389830508474576, + "eval_loss": 2.046339988708496, + "eval_runtime": 53.909, + "eval_samples_per_second": 18.55, + "eval_steps_per_second": 0.779, + "step": 330 + }, + { + "epoch": 0.34001027221366203, + "grad_norm": 2.5724353790283203, + "learning_rate": 4.954320517700799e-06, + "loss": 1.7022, + "step": 331 + }, + { + "epoch": 0.34103749357986646, + "grad_norm": 2.510002613067627, + "learning_rate": 4.952417205938333e-06, + "loss": 1.802, + "step": 332 + }, + { + "epoch": 0.3420647149460709, + "grad_norm": 2.8722083568573, + "learning_rate": 4.950513894175866e-06, + "loss": 1.7772, + "step": 333 + }, + { + "epoch": 0.3430919363122753, + "grad_norm": 2.586747407913208, + "learning_rate": 4.9486105824133995e-06, + "loss": 1.7664, + "step": 334 + }, + { + "epoch": 0.3441191576784797, + "grad_norm": 2.8008205890655518, + "learning_rate": 4.946707270650933e-06, + "loss": 1.7892, + "step": 335 + }, + { + "epoch": 0.34514637904468415, + "grad_norm": 2.632664203643799, + "learning_rate": 4.944803958888466e-06, + "loss": 1.7967, + "step": 336 + }, + { + "epoch": 0.34617360041088857, + "grad_norm": 2.8898301124572754, + "learning_rate": 4.942900647126e-06, + "loss": 1.8126, + "step": 337 + }, + { + "epoch": 0.34720082177709294, + "grad_norm": 2.719019651412964, + "learning_rate": 4.940997335363533e-06, + "loss": 1.7544, + "step": 338 + }, + { + "epoch": 0.34822804314329736, + "grad_norm": 2.9995970726013184, + "learning_rate": 4.939094023601066e-06, + "loss": 1.7142, + "step": 339 + }, + { + "epoch": 0.3492552645095018, + "grad_norm": 2.9737157821655273, + "learning_rate": 4.9371907118386e-06, + "loss": 1.7659, + "step": 340 + }, + { + "epoch": 0.3492552645095018, + "eval_loss": 2.024946928024292, + "eval_runtime": 52.8069, + "eval_samples_per_second": 18.937, + "eval_steps_per_second": 0.795, + "step": 340 + }, + { + "epoch": 0.3502824858757062, + "grad_norm": 2.8651719093322754, + "learning_rate": 4.935287400076133e-06, + "loss": 1.7767, + "step": 341 + }, + { + "epoch": 0.35130970724191063, + "grad_norm": 2.6293981075286865, + "learning_rate": 4.933384088313666e-06, + "loss": 1.7815, + "step": 342 + }, + { + "epoch": 0.35233692860811505, + "grad_norm": 3.230978012084961, + "learning_rate": 4.9314807765512e-06, + "loss": 1.7423, + "step": 343 + }, + { + "epoch": 0.3533641499743195, + "grad_norm": 3.228300094604492, + "learning_rate": 4.929577464788733e-06, + "loss": 1.6845, + "step": 344 + }, + { + "epoch": 0.3543913713405239, + "grad_norm": 3.036733388900757, + "learning_rate": 4.9276741530262665e-06, + "loss": 1.7524, + "step": 345 + }, + { + "epoch": 0.3554185927067283, + "grad_norm": 2.778538227081299, + "learning_rate": 4.9257708412638e-06, + "loss": 1.7433, + "step": 346 + }, + { + "epoch": 0.35644581407293274, + "grad_norm": 3.0958566665649414, + "learning_rate": 4.923867529501332e-06, + "loss": 1.76, + "step": 347 + }, + { + "epoch": 0.3574730354391371, + "grad_norm": 3.1120026111602783, + "learning_rate": 4.921964217738866e-06, + "loss": 1.7399, + "step": 348 + }, + { + "epoch": 0.35850025680534153, + "grad_norm": 2.70768141746521, + "learning_rate": 4.920060905976399e-06, + "loss": 1.8169, + "step": 349 + }, + { + "epoch": 0.35952747817154596, + "grad_norm": 2.917478561401367, + "learning_rate": 4.9181575942139325e-06, + "loss": 1.7948, + "step": 350 + }, + { + "epoch": 0.35952747817154596, + "eval_loss": 2.0208380222320557, + "eval_runtime": 47.4243, + "eval_samples_per_second": 21.086, + "eval_steps_per_second": 0.886, + "step": 350 + }, + { + "epoch": 0.3605546995377504, + "grad_norm": 3.0564756393432617, + "learning_rate": 4.916254282451466e-06, + "loss": 1.7861, + "step": 351 + }, + { + "epoch": 0.3615819209039548, + "grad_norm": 2.709195137023926, + "learning_rate": 4.914350970688999e-06, + "loss": 1.8323, + "step": 352 + }, + { + "epoch": 0.3626091422701592, + "grad_norm": 2.8797669410705566, + "learning_rate": 4.9124476589265325e-06, + "loss": 1.722, + "step": 353 + }, + { + "epoch": 0.36363636363636365, + "grad_norm": 2.560702085494995, + "learning_rate": 4.910544347164066e-06, + "loss": 1.7289, + "step": 354 + }, + { + "epoch": 0.36466358500256807, + "grad_norm": 3.3781847953796387, + "learning_rate": 4.908641035401599e-06, + "loss": 1.8096, + "step": 355 + }, + { + "epoch": 0.3656908063687725, + "grad_norm": 2.582310676574707, + "learning_rate": 4.906737723639133e-06, + "loss": 1.8752, + "step": 356 + }, + { + "epoch": 0.3667180277349769, + "grad_norm": 3.506385564804077, + "learning_rate": 4.904834411876665e-06, + "loss": 1.7404, + "step": 357 + }, + { + "epoch": 0.3677452491011813, + "grad_norm": 2.4360411167144775, + "learning_rate": 4.902931100114199e-06, + "loss": 1.7812, + "step": 358 + }, + { + "epoch": 0.3687724704673857, + "grad_norm": 3.77639102935791, + "learning_rate": 4.901027788351733e-06, + "loss": 1.7103, + "step": 359 + }, + { + "epoch": 0.3697996918335901, + "grad_norm": 2.887723922729492, + "learning_rate": 4.899124476589266e-06, + "loss": 1.7485, + "step": 360 + }, + { + "epoch": 0.3697996918335901, + "eval_loss": 2.0298168659210205, + "eval_runtime": 46.0993, + "eval_samples_per_second": 21.692, + "eval_steps_per_second": 0.911, + "step": 360 + }, + { + "epoch": 0.37082691319979455, + "grad_norm": 3.5861997604370117, + "learning_rate": 4.8972211648267994e-06, + "loss": 1.7189, + "step": 361 + }, + { + "epoch": 0.371854134565999, + "grad_norm": 2.590808868408203, + "learning_rate": 4.895317853064333e-06, + "loss": 1.7683, + "step": 362 + }, + { + "epoch": 0.3728813559322034, + "grad_norm": 3.4451487064361572, + "learning_rate": 4.893414541301866e-06, + "loss": 1.7436, + "step": 363 + }, + { + "epoch": 0.3739085772984078, + "grad_norm": 2.406073808670044, + "learning_rate": 4.891511229539399e-06, + "loss": 1.7378, + "step": 364 + }, + { + "epoch": 0.37493579866461224, + "grad_norm": 2.7162997722625732, + "learning_rate": 4.889607917776932e-06, + "loss": 1.7558, + "step": 365 + }, + { + "epoch": 0.37596302003081666, + "grad_norm": 2.564753770828247, + "learning_rate": 4.887704606014465e-06, + "loss": 1.7559, + "step": 366 + }, + { + "epoch": 0.37699024139702103, + "grad_norm": 2.7196900844573975, + "learning_rate": 4.885801294251999e-06, + "loss": 1.7683, + "step": 367 + }, + { + "epoch": 0.37801746276322545, + "grad_norm": 2.9334428310394287, + "learning_rate": 4.883897982489532e-06, + "loss": 1.6932, + "step": 368 + }, + { + "epoch": 0.3790446841294299, + "grad_norm": 2.4899954795837402, + "learning_rate": 4.8819946707270655e-06, + "loss": 1.7943, + "step": 369 + }, + { + "epoch": 0.3800719054956343, + "grad_norm": 2.820594310760498, + "learning_rate": 4.880091358964599e-06, + "loss": 1.7729, + "step": 370 + }, + { + "epoch": 0.3800719054956343, + "eval_loss": 2.0204291343688965, + "eval_runtime": 44.9993, + "eval_samples_per_second": 22.223, + "eval_steps_per_second": 0.933, + "step": 370 + }, + { + "epoch": 0.3810991268618387, + "grad_norm": 2.4717929363250732, + "learning_rate": 4.878188047202132e-06, + "loss": 1.6813, + "step": 371 + }, + { + "epoch": 0.38212634822804314, + "grad_norm": 3.1815855503082275, + "learning_rate": 4.876284735439666e-06, + "loss": 1.8588, + "step": 372 + }, + { + "epoch": 0.38315356959424757, + "grad_norm": 2.7330422401428223, + "learning_rate": 4.874381423677199e-06, + "loss": 1.8245, + "step": 373 + }, + { + "epoch": 0.384180790960452, + "grad_norm": 3.129444122314453, + "learning_rate": 4.8724781119147315e-06, + "loss": 1.7494, + "step": 374 + }, + { + "epoch": 0.3852080123266564, + "grad_norm": 2.75551700592041, + "learning_rate": 4.870574800152265e-06, + "loss": 1.7761, + "step": 375 + }, + { + "epoch": 0.38623523369286084, + "grad_norm": 2.9313466548919678, + "learning_rate": 4.868671488389798e-06, + "loss": 1.7076, + "step": 376 + }, + { + "epoch": 0.3872624550590652, + "grad_norm": 2.7590556144714355, + "learning_rate": 4.8667681766273316e-06, + "loss": 1.7105, + "step": 377 + }, + { + "epoch": 0.3882896764252696, + "grad_norm": 2.8224680423736572, + "learning_rate": 4.864864864864866e-06, + "loss": 1.6853, + "step": 378 + }, + { + "epoch": 0.38931689779147405, + "grad_norm": 2.5148062705993652, + "learning_rate": 4.862961553102399e-06, + "loss": 1.7477, + "step": 379 + }, + { + "epoch": 0.39034411915767847, + "grad_norm": 3.144247055053711, + "learning_rate": 4.8610582413399325e-06, + "loss": 1.7726, + "step": 380 + }, + { + "epoch": 0.39034411915767847, + "eval_loss": 2.006993532180786, + "eval_runtime": 50.8909, + "eval_samples_per_second": 19.65, + "eval_steps_per_second": 0.825, + "step": 380 + }, + { + "epoch": 0.3913713405238829, + "grad_norm": 2.6818125247955322, + "learning_rate": 4.859154929577465e-06, + "loss": 1.7351, + "step": 381 + }, + { + "epoch": 0.3923985618900873, + "grad_norm": 2.8543903827667236, + "learning_rate": 4.857251617814998e-06, + "loss": 1.7499, + "step": 382 + }, + { + "epoch": 0.39342578325629174, + "grad_norm": 2.5483179092407227, + "learning_rate": 4.855348306052532e-06, + "loss": 1.7849, + "step": 383 + }, + { + "epoch": 0.39445300462249616, + "grad_norm": 2.7169013023376465, + "learning_rate": 4.853444994290065e-06, + "loss": 1.7664, + "step": 384 + }, + { + "epoch": 0.3954802259887006, + "grad_norm": 2.8304696083068848, + "learning_rate": 4.8515416825275985e-06, + "loss": 1.73, + "step": 385 + }, + { + "epoch": 0.396507447354905, + "grad_norm": 2.4217448234558105, + "learning_rate": 4.849638370765132e-06, + "loss": 1.7915, + "step": 386 + }, + { + "epoch": 0.3975346687211094, + "grad_norm": 2.523256301879883, + "learning_rate": 4.847735059002665e-06, + "loss": 1.7195, + "step": 387 + }, + { + "epoch": 0.3985618900873138, + "grad_norm": 2.813638925552368, + "learning_rate": 4.8458317472401985e-06, + "loss": 1.6766, + "step": 388 + }, + { + "epoch": 0.3995891114535182, + "grad_norm": 2.6701548099517822, + "learning_rate": 4.843928435477732e-06, + "loss": 1.7666, + "step": 389 + }, + { + "epoch": 0.40061633281972264, + "grad_norm": 2.5568342208862305, + "learning_rate": 4.842025123715265e-06, + "loss": 1.723, + "step": 390 + }, + { + "epoch": 0.40061633281972264, + "eval_loss": 2.0098180770874023, + "eval_runtime": 44.4503, + "eval_samples_per_second": 22.497, + "eval_steps_per_second": 0.945, + "step": 390 + }, + { + "epoch": 0.40164355418592707, + "grad_norm": 2.6287341117858887, + "learning_rate": 4.840121811952798e-06, + "loss": 1.759, + "step": 391 + }, + { + "epoch": 0.4026707755521315, + "grad_norm": 2.934105157852173, + "learning_rate": 4.838218500190331e-06, + "loss": 1.7501, + "step": 392 + }, + { + "epoch": 0.4036979969183359, + "grad_norm": 2.6563878059387207, + "learning_rate": 4.8363151884278645e-06, + "loss": 1.7624, + "step": 393 + }, + { + "epoch": 0.40472521828454033, + "grad_norm": 2.538278102874756, + "learning_rate": 4.834411876665398e-06, + "loss": 1.7457, + "step": 394 + }, + { + "epoch": 0.40575243965074476, + "grad_norm": 2.801630735397339, + "learning_rate": 4.832508564902931e-06, + "loss": 1.7549, + "step": 395 + }, + { + "epoch": 0.4067796610169492, + "grad_norm": 2.571937322616577, + "learning_rate": 4.830605253140465e-06, + "loss": 1.6954, + "step": 396 + }, + { + "epoch": 0.40780688238315355, + "grad_norm": 2.5652592182159424, + "learning_rate": 4.828701941377998e-06, + "loss": 1.7328, + "step": 397 + }, + { + "epoch": 0.40883410374935797, + "grad_norm": 2.6840121746063232, + "learning_rate": 4.826798629615531e-06, + "loss": 1.6934, + "step": 398 + }, + { + "epoch": 0.4098613251155624, + "grad_norm": 3.0103495121002197, + "learning_rate": 4.824895317853065e-06, + "loss": 1.7086, + "step": 399 + }, + { + "epoch": 0.4108885464817668, + "grad_norm": 2.7361223697662354, + "learning_rate": 4.822992006090598e-06, + "loss": 1.7301, + "step": 400 + }, + { + "epoch": 0.4108885464817668, + "eval_loss": 1.9975080490112305, + "eval_runtime": 44.5413, + "eval_samples_per_second": 22.451, + "eval_steps_per_second": 0.943, + "step": 400 + }, + { + "epoch": 0.41191576784797124, + "grad_norm": 2.6805901527404785, + "learning_rate": 4.821088694328131e-06, + "loss": 1.6896, + "step": 401 + }, + { + "epoch": 0.41294298921417566, + "grad_norm": 3.231706142425537, + "learning_rate": 4.819185382565665e-06, + "loss": 1.7134, + "step": 402 + }, + { + "epoch": 0.4139702105803801, + "grad_norm": 2.5561299324035645, + "learning_rate": 4.817282070803198e-06, + "loss": 1.764, + "step": 403 + }, + { + "epoch": 0.4149974319465845, + "grad_norm": 3.3125340938568115, + "learning_rate": 4.8153787590407315e-06, + "loss": 1.76, + "step": 404 + }, + { + "epoch": 0.41602465331278893, + "grad_norm": 2.750051498413086, + "learning_rate": 4.813475447278265e-06, + "loss": 1.6882, + "step": 405 + }, + { + "epoch": 0.41705187467899335, + "grad_norm": 4.006515979766846, + "learning_rate": 4.811572135515798e-06, + "loss": 1.7892, + "step": 406 + }, + { + "epoch": 0.4180790960451977, + "grad_norm": 2.5276904106140137, + "learning_rate": 4.809668823753332e-06, + "loss": 1.6995, + "step": 407 + }, + { + "epoch": 0.41910631741140214, + "grad_norm": 2.959444046020508, + "learning_rate": 4.807765511990864e-06, + "loss": 1.7058, + "step": 408 + }, + { + "epoch": 0.42013353877760656, + "grad_norm": 2.8472464084625244, + "learning_rate": 4.8058622002283975e-06, + "loss": 1.7258, + "step": 409 + }, + { + "epoch": 0.421160760143811, + "grad_norm": 3.0240273475646973, + "learning_rate": 4.803958888465931e-06, + "loss": 1.7789, + "step": 410 + }, + { + "epoch": 0.421160760143811, + "eval_loss": 1.9919871091842651, + "eval_runtime": 41.727, + "eval_samples_per_second": 23.965, + "eval_steps_per_second": 1.007, + "step": 410 + }, + { + "epoch": 0.4221879815100154, + "grad_norm": 2.6211893558502197, + "learning_rate": 4.802055576703464e-06, + "loss": 1.7315, + "step": 411 + }, + { + "epoch": 0.42321520287621983, + "grad_norm": 2.6072895526885986, + "learning_rate": 4.8001522649409976e-06, + "loss": 1.8176, + "step": 412 + }, + { + "epoch": 0.42424242424242425, + "grad_norm": 2.9197702407836914, + "learning_rate": 4.798248953178531e-06, + "loss": 1.6947, + "step": 413 + }, + { + "epoch": 0.4252696456086287, + "grad_norm": 2.4383556842803955, + "learning_rate": 4.796345641416064e-06, + "loss": 1.6728, + "step": 414 + }, + { + "epoch": 0.4262968669748331, + "grad_norm": 2.9951915740966797, + "learning_rate": 4.794442329653598e-06, + "loss": 1.7249, + "step": 415 + }, + { + "epoch": 0.42732408834103747, + "grad_norm": 2.6095519065856934, + "learning_rate": 4.792539017891131e-06, + "loss": 1.7735, + "step": 416 + }, + { + "epoch": 0.4283513097072419, + "grad_norm": 2.6429390907287598, + "learning_rate": 4.790635706128664e-06, + "loss": 1.704, + "step": 417 + }, + { + "epoch": 0.4293785310734463, + "grad_norm": 2.921812057495117, + "learning_rate": 4.788732394366197e-06, + "loss": 1.7355, + "step": 418 + }, + { + "epoch": 0.43040575243965074, + "grad_norm": 2.6050965785980225, + "learning_rate": 4.786829082603731e-06, + "loss": 1.6636, + "step": 419 + }, + { + "epoch": 0.43143297380585516, + "grad_norm": 2.672070264816284, + "learning_rate": 4.7849257708412645e-06, + "loss": 1.6627, + "step": 420 + }, + { + "epoch": 0.43143297380585516, + "eval_loss": 1.9990200996398926, + "eval_runtime": 45.0672, + "eval_samples_per_second": 22.189, + "eval_steps_per_second": 0.932, + "step": 420 + }, + { + "epoch": 0.4324601951720596, + "grad_norm": 2.6813488006591797, + "learning_rate": 4.783022459078798e-06, + "loss": 1.7453, + "step": 421 + }, + { + "epoch": 0.433487416538264, + "grad_norm": 2.6587321758270264, + "learning_rate": 4.781119147316331e-06, + "loss": 1.6795, + "step": 422 + }, + { + "epoch": 0.4345146379044684, + "grad_norm": 2.827758312225342, + "learning_rate": 4.7792158355538645e-06, + "loss": 1.7098, + "step": 423 + }, + { + "epoch": 0.43554185927067285, + "grad_norm": 2.8615901470184326, + "learning_rate": 4.777312523791398e-06, + "loss": 1.7171, + "step": 424 + }, + { + "epoch": 0.43656908063687727, + "grad_norm": 2.9509294033050537, + "learning_rate": 4.77540921202893e-06, + "loss": 1.7018, + "step": 425 + }, + { + "epoch": 0.43759630200308164, + "grad_norm": 2.7994418144226074, + "learning_rate": 4.773505900266464e-06, + "loss": 1.7328, + "step": 426 + }, + { + "epoch": 0.43862352336928606, + "grad_norm": 2.7476799488067627, + "learning_rate": 4.771602588503997e-06, + "loss": 1.7116, + "step": 427 + }, + { + "epoch": 0.4396507447354905, + "grad_norm": 2.7955048084259033, + "learning_rate": 4.7696992767415305e-06, + "loss": 1.7187, + "step": 428 + }, + { + "epoch": 0.4406779661016949, + "grad_norm": 2.4326071739196777, + "learning_rate": 4.767795964979064e-06, + "loss": 1.75, + "step": 429 + }, + { + "epoch": 0.44170518746789933, + "grad_norm": 2.7708163261413574, + "learning_rate": 4.765892653216597e-06, + "loss": 1.7562, + "step": 430 + }, + { + "epoch": 0.44170518746789933, + "eval_loss": 2.0262997150421143, + "eval_runtime": 46.1649, + "eval_samples_per_second": 21.661, + "eval_steps_per_second": 0.91, + "step": 430 + }, + { + "epoch": 0.44273240883410375, + "grad_norm": 2.872743606567383, + "learning_rate": 4.763989341454131e-06, + "loss": 1.7527, + "step": 431 + }, + { + "epoch": 0.4437596302003082, + "grad_norm": 2.6387035846710205, + "learning_rate": 4.762086029691664e-06, + "loss": 1.6742, + "step": 432 + }, + { + "epoch": 0.4447868515665126, + "grad_norm": 2.6522979736328125, + "learning_rate": 4.760182717929197e-06, + "loss": 1.6811, + "step": 433 + }, + { + "epoch": 0.445814072932717, + "grad_norm": 2.644155263900757, + "learning_rate": 4.758279406166731e-06, + "loss": 1.7083, + "step": 434 + }, + { + "epoch": 0.44684129429892144, + "grad_norm": 2.530224561691284, + "learning_rate": 4.756376094404263e-06, + "loss": 1.6318, + "step": 435 + }, + { + "epoch": 0.4478685156651258, + "grad_norm": 2.431049108505249, + "learning_rate": 4.7544727826417966e-06, + "loss": 1.7162, + "step": 436 + }, + { + "epoch": 0.44889573703133023, + "grad_norm": 2.8091282844543457, + "learning_rate": 4.75256947087933e-06, + "loss": 1.7717, + "step": 437 + }, + { + "epoch": 0.44992295839753466, + "grad_norm": 2.5520501136779785, + "learning_rate": 4.750666159116863e-06, + "loss": 1.7201, + "step": 438 + }, + { + "epoch": 0.4509501797637391, + "grad_norm": 2.596370220184326, + "learning_rate": 4.7487628473543975e-06, + "loss": 1.6882, + "step": 439 + }, + { + "epoch": 0.4519774011299435, + "grad_norm": 2.667034387588501, + "learning_rate": 4.746859535591931e-06, + "loss": 1.6499, + "step": 440 + }, + { + "epoch": 0.4519774011299435, + "eval_loss": 2.008521556854248, + "eval_runtime": 49.3626, + "eval_samples_per_second": 20.258, + "eval_steps_per_second": 0.851, + "step": 440 + }, + { + "epoch": 0.4530046224961479, + "grad_norm": 2.636491537094116, + "learning_rate": 4.744956223829464e-06, + "loss": 1.6987, + "step": 441 + }, + { + "epoch": 0.45403184386235235, + "grad_norm": 2.4653050899505615, + "learning_rate": 4.743052912066997e-06, + "loss": 1.6775, + "step": 442 + }, + { + "epoch": 0.45505906522855677, + "grad_norm": 2.3866467475891113, + "learning_rate": 4.74114960030453e-06, + "loss": 1.7234, + "step": 443 + }, + { + "epoch": 0.4560862865947612, + "grad_norm": 2.675262928009033, + "learning_rate": 4.7392462885420635e-06, + "loss": 1.7363, + "step": 444 + }, + { + "epoch": 0.4571135079609656, + "grad_norm": 2.6233296394348145, + "learning_rate": 4.737342976779597e-06, + "loss": 1.6236, + "step": 445 + }, + { + "epoch": 0.45814072932717, + "grad_norm": 2.5638058185577393, + "learning_rate": 4.73543966501713e-06, + "loss": 1.7097, + "step": 446 + }, + { + "epoch": 0.4591679506933744, + "grad_norm": 2.7206926345825195, + "learning_rate": 4.7335363532546636e-06, + "loss": 1.6909, + "step": 447 + }, + { + "epoch": 0.4601951720595788, + "grad_norm": 2.546241283416748, + "learning_rate": 4.731633041492197e-06, + "loss": 1.6951, + "step": 448 + }, + { + "epoch": 0.46122239342578325, + "grad_norm": 2.6549105644226074, + "learning_rate": 4.72972972972973e-06, + "loss": 1.7052, + "step": 449 + }, + { + "epoch": 0.4622496147919877, + "grad_norm": 2.6649065017700195, + "learning_rate": 4.727826417967264e-06, + "loss": 1.7406, + "step": 450 + }, + { + "epoch": 0.4622496147919877, + "eval_loss": 1.9886436462402344, + "eval_runtime": 47.48, + "eval_samples_per_second": 21.061, + "eval_steps_per_second": 0.885, + "step": 450 + }, + { + "epoch": 0.4632768361581921, + "grad_norm": 2.7138187885284424, + "learning_rate": 4.725923106204797e-06, + "loss": 1.7154, + "step": 451 + }, + { + "epoch": 0.4643040575243965, + "grad_norm": 2.6095476150512695, + "learning_rate": 4.72401979444233e-06, + "loss": 1.686, + "step": 452 + }, + { + "epoch": 0.46533127889060094, + "grad_norm": 2.866363048553467, + "learning_rate": 4.722116482679863e-06, + "loss": 1.721, + "step": 453 + }, + { + "epoch": 0.46635850025680536, + "grad_norm": 2.4947352409362793, + "learning_rate": 4.720213170917396e-06, + "loss": 1.708, + "step": 454 + }, + { + "epoch": 0.46738572162300973, + "grad_norm": 2.3888256549835205, + "learning_rate": 4.71830985915493e-06, + "loss": 1.7356, + "step": 455 + }, + { + "epoch": 0.46841294298921415, + "grad_norm": 2.7065343856811523, + "learning_rate": 4.716406547392463e-06, + "loss": 1.6943, + "step": 456 + }, + { + "epoch": 0.4694401643554186, + "grad_norm": 2.424285888671875, + "learning_rate": 4.714503235629996e-06, + "loss": 1.7385, + "step": 457 + }, + { + "epoch": 0.470467385721623, + "grad_norm": 2.4175925254821777, + "learning_rate": 4.71259992386753e-06, + "loss": 1.6654, + "step": 458 + }, + { + "epoch": 0.4714946070878274, + "grad_norm": 2.3203656673431396, + "learning_rate": 4.710696612105063e-06, + "loss": 1.6977, + "step": 459 + }, + { + "epoch": 0.47252182845403184, + "grad_norm": 2.6660172939300537, + "learning_rate": 4.708793300342596e-06, + "loss": 1.7055, + "step": 460 + }, + { + "epoch": 0.47252182845403184, + "eval_loss": 1.9864659309387207, + "eval_runtime": 47.6232, + "eval_samples_per_second": 20.998, + "eval_steps_per_second": 0.882, + "step": 460 + }, + { + "epoch": 0.47354904982023627, + "grad_norm": 2.515615701675415, + "learning_rate": 4.70688998858013e-06, + "loss": 1.689, + "step": 461 + }, + { + "epoch": 0.4745762711864407, + "grad_norm": 2.429264783859253, + "learning_rate": 4.704986676817663e-06, + "loss": 1.6796, + "step": 462 + }, + { + "epoch": 0.4756034925526451, + "grad_norm": 2.469543933868408, + "learning_rate": 4.7030833650551965e-06, + "loss": 1.7164, + "step": 463 + }, + { + "epoch": 0.47663071391884954, + "grad_norm": 2.5921683311462402, + "learning_rate": 4.70118005329273e-06, + "loss": 1.6746, + "step": 464 + }, + { + "epoch": 0.4776579352850539, + "grad_norm": 2.6612918376922607, + "learning_rate": 4.699276741530263e-06, + "loss": 1.7598, + "step": 465 + }, + { + "epoch": 0.4786851566512583, + "grad_norm": 2.4633305072784424, + "learning_rate": 4.697373429767797e-06, + "loss": 1.7241, + "step": 466 + }, + { + "epoch": 0.47971237801746275, + "grad_norm": 2.613023281097412, + "learning_rate": 4.69547011800533e-06, + "loss": 1.7088, + "step": 467 + }, + { + "epoch": 0.48073959938366717, + "grad_norm": 2.5329010486602783, + "learning_rate": 4.693566806242863e-06, + "loss": 1.7163, + "step": 468 + }, + { + "epoch": 0.4817668207498716, + "grad_norm": 2.7887041568756104, + "learning_rate": 4.691663494480397e-06, + "loss": 1.6969, + "step": 469 + }, + { + "epoch": 0.482794042116076, + "grad_norm": 2.72991943359375, + "learning_rate": 4.689760182717929e-06, + "loss": 1.7332, + "step": 470 + }, + { + "epoch": 0.482794042116076, + "eval_loss": 1.965472936630249, + "eval_runtime": 46.1059, + "eval_samples_per_second": 21.689, + "eval_steps_per_second": 0.911, + "step": 470 + }, + { + "epoch": 0.48382126348228044, + "grad_norm": 2.613785982131958, + "learning_rate": 4.6878568709554626e-06, + "loss": 1.6272, + "step": 471 + }, + { + "epoch": 0.48484848484848486, + "grad_norm": 2.605299472808838, + "learning_rate": 4.685953559192996e-06, + "loss": 1.6966, + "step": 472 + }, + { + "epoch": 0.4858757062146893, + "grad_norm": 2.672013998031616, + "learning_rate": 4.684050247430529e-06, + "loss": 1.7132, + "step": 473 + }, + { + "epoch": 0.4869029275808937, + "grad_norm": 2.576486825942993, + "learning_rate": 4.682146935668063e-06, + "loss": 1.6449, + "step": 474 + }, + { + "epoch": 0.4879301489470981, + "grad_norm": 2.626840114593506, + "learning_rate": 4.680243623905596e-06, + "loss": 1.6993, + "step": 475 + }, + { + "epoch": 0.4889573703133025, + "grad_norm": 2.974418878555298, + "learning_rate": 4.678340312143129e-06, + "loss": 1.7116, + "step": 476 + }, + { + "epoch": 0.4899845916795069, + "grad_norm": 2.541242837905884, + "learning_rate": 4.676437000380663e-06, + "loss": 1.6693, + "step": 477 + }, + { + "epoch": 0.49101181304571134, + "grad_norm": 2.5494985580444336, + "learning_rate": 4.674533688618196e-06, + "loss": 1.7819, + "step": 478 + }, + { + "epoch": 0.49203903441191577, + "grad_norm": 2.606137990951538, + "learning_rate": 4.6726303768557295e-06, + "loss": 1.7278, + "step": 479 + }, + { + "epoch": 0.4930662557781202, + "grad_norm": 2.5032176971435547, + "learning_rate": 4.670727065093263e-06, + "loss": 1.6318, + "step": 480 + }, + { + "epoch": 0.4930662557781202, + "eval_loss": 1.9851018190383911, + "eval_runtime": 46.0386, + "eval_samples_per_second": 21.721, + "eval_steps_per_second": 0.912, + "step": 480 + }, + { + "epoch": 0.4940934771443246, + "grad_norm": 2.703259229660034, + "learning_rate": 4.668823753330796e-06, + "loss": 1.6646, + "step": 481 + }, + { + "epoch": 0.49512069851052903, + "grad_norm": 2.681807041168213, + "learning_rate": 4.6669204415683295e-06, + "loss": 1.6759, + "step": 482 + }, + { + "epoch": 0.49614791987673346, + "grad_norm": 2.7280473709106445, + "learning_rate": 4.665017129805863e-06, + "loss": 1.7388, + "step": 483 + }, + { + "epoch": 0.4971751412429379, + "grad_norm": 2.420909881591797, + "learning_rate": 4.663113818043396e-06, + "loss": 1.729, + "step": 484 + }, + { + "epoch": 0.49820236260914225, + "grad_norm": 2.5764737129211426, + "learning_rate": 4.66121050628093e-06, + "loss": 1.6903, + "step": 485 + }, + { + "epoch": 0.49922958397534667, + "grad_norm": 2.389531135559082, + "learning_rate": 4.659307194518463e-06, + "loss": 1.7191, + "step": 486 + }, + { + "epoch": 0.5002568053415511, + "grad_norm": 2.5207972526550293, + "learning_rate": 4.6574038827559955e-06, + "loss": 1.6892, + "step": 487 + }, + { + "epoch": 0.5012840267077555, + "grad_norm": 2.5662360191345215, + "learning_rate": 4.655500570993529e-06, + "loss": 1.6978, + "step": 488 + }, + { + "epoch": 0.50231124807396, + "grad_norm": 2.50262713432312, + "learning_rate": 4.653597259231062e-06, + "loss": 1.7379, + "step": 489 + }, + { + "epoch": 0.5033384694401644, + "grad_norm": 2.56643009185791, + "learning_rate": 4.651693947468596e-06, + "loss": 1.6233, + "step": 490 + }, + { + "epoch": 0.5033384694401644, + "eval_loss": 1.9815036058425903, + "eval_runtime": 48.0257, + "eval_samples_per_second": 20.822, + "eval_steps_per_second": 0.875, + "step": 490 + }, + { + "epoch": 0.5043656908063687, + "grad_norm": 2.504117965698242, + "learning_rate": 4.649790635706129e-06, + "loss": 1.7664, + "step": 491 + }, + { + "epoch": 0.5053929121725732, + "grad_norm": 2.906843423843384, + "learning_rate": 4.647887323943662e-06, + "loss": 1.6731, + "step": 492 + }, + { + "epoch": 0.5064201335387776, + "grad_norm": 2.586747169494629, + "learning_rate": 4.645984012181196e-06, + "loss": 1.68, + "step": 493 + }, + { + "epoch": 0.507447354904982, + "grad_norm": 2.3300511837005615, + "learning_rate": 4.644080700418729e-06, + "loss": 1.5998, + "step": 494 + }, + { + "epoch": 0.5084745762711864, + "grad_norm": 2.8711049556732178, + "learning_rate": 4.642177388656262e-06, + "loss": 1.7198, + "step": 495 + }, + { + "epoch": 0.5095017976373909, + "grad_norm": 2.6463162899017334, + "learning_rate": 4.640274076893796e-06, + "loss": 1.6814, + "step": 496 + }, + { + "epoch": 0.5105290190035953, + "grad_norm": 2.650481700897217, + "learning_rate": 4.638370765131328e-06, + "loss": 1.7353, + "step": 497 + }, + { + "epoch": 0.5115562403697997, + "grad_norm": 2.4464986324310303, + "learning_rate": 4.636467453368862e-06, + "loss": 1.6579, + "step": 498 + }, + { + "epoch": 0.5125834617360041, + "grad_norm": 2.438021659851074, + "learning_rate": 4.634564141606396e-06, + "loss": 1.6485, + "step": 499 + }, + { + "epoch": 0.5136106831022085, + "grad_norm": 2.5354347229003906, + "learning_rate": 4.632660829843929e-06, + "loss": 1.7123, + "step": 500 + }, + { + "epoch": 0.5136106831022085, + "eval_loss": 1.9738385677337646, + "eval_runtime": 52.5306, + "eval_samples_per_second": 19.037, + "eval_steps_per_second": 0.8, + "step": 500 + }, + { + "epoch": 0.514637904468413, + "grad_norm": 2.3702070713043213, + "learning_rate": 4.630757518081463e-06, + "loss": 1.6423, + "step": 501 + }, + { + "epoch": 0.5156651258346173, + "grad_norm": 2.5336198806762695, + "learning_rate": 4.628854206318996e-06, + "loss": 1.771, + "step": 502 + }, + { + "epoch": 0.5166923472008218, + "grad_norm": 2.572751998901367, + "learning_rate": 4.626950894556529e-06, + "loss": 1.677, + "step": 503 + }, + { + "epoch": 0.5177195685670262, + "grad_norm": 2.3865134716033936, + "learning_rate": 4.625047582794062e-06, + "loss": 1.7111, + "step": 504 + }, + { + "epoch": 0.5187467899332306, + "grad_norm": 2.5152785778045654, + "learning_rate": 4.623144271031595e-06, + "loss": 1.6343, + "step": 505 + }, + { + "epoch": 0.519774011299435, + "grad_norm": 2.5121307373046875, + "learning_rate": 4.6212409592691286e-06, + "loss": 1.6357, + "step": 506 + }, + { + "epoch": 0.5208012326656395, + "grad_norm": 2.506908655166626, + "learning_rate": 4.619337647506662e-06, + "loss": 1.7295, + "step": 507 + }, + { + "epoch": 0.5218284540318439, + "grad_norm": 2.636082172393799, + "learning_rate": 4.617434335744195e-06, + "loss": 1.6916, + "step": 508 + }, + { + "epoch": 0.5228556753980482, + "grad_norm": 2.5987415313720703, + "learning_rate": 4.615531023981729e-06, + "loss": 1.659, + "step": 509 + }, + { + "epoch": 0.5238828967642527, + "grad_norm": 2.642770290374756, + "learning_rate": 4.613627712219262e-06, + "loss": 1.6789, + "step": 510 + }, + { + "epoch": 0.5238828967642527, + "eval_loss": 1.9659523963928223, + "eval_runtime": 42.9453, + "eval_samples_per_second": 23.285, + "eval_steps_per_second": 0.978, + "step": 510 + }, + { + "epoch": 0.5249101181304571, + "grad_norm": 2.438389301300049, + "learning_rate": 4.611724400456795e-06, + "loss": 1.68, + "step": 511 + }, + { + "epoch": 0.5259373394966615, + "grad_norm": 2.536546468734741, + "learning_rate": 4.609821088694329e-06, + "loss": 1.6847, + "step": 512 + }, + { + "epoch": 0.5269645608628659, + "grad_norm": 2.4056830406188965, + "learning_rate": 4.607917776931862e-06, + "loss": 1.6577, + "step": 513 + }, + { + "epoch": 0.5279917822290704, + "grad_norm": 2.81195330619812, + "learning_rate": 4.606014465169395e-06, + "loss": 1.7149, + "step": 514 + }, + { + "epoch": 0.5290190035952748, + "grad_norm": 2.684948205947876, + "learning_rate": 4.604111153406928e-06, + "loss": 1.717, + "step": 515 + }, + { + "epoch": 0.5300462249614792, + "grad_norm": 2.583092212677002, + "learning_rate": 4.602207841644461e-06, + "loss": 1.6238, + "step": 516 + }, + { + "epoch": 0.5310734463276836, + "grad_norm": 3.022528886795044, + "learning_rate": 4.600304529881995e-06, + "loss": 1.7262, + "step": 517 + }, + { + "epoch": 0.5321006676938881, + "grad_norm": 2.6941866874694824, + "learning_rate": 4.598401218119528e-06, + "loss": 1.7055, + "step": 518 + }, + { + "epoch": 0.5331278890600925, + "grad_norm": 2.5207839012145996, + "learning_rate": 4.5964979063570614e-06, + "loss": 1.7117, + "step": 519 + }, + { + "epoch": 0.5341551104262968, + "grad_norm": 2.8056678771972656, + "learning_rate": 4.594594594594596e-06, + "loss": 1.657, + "step": 520 + }, + { + "epoch": 0.5341551104262968, + "eval_loss": 1.9712116718292236, + "eval_runtime": 49.57, + "eval_samples_per_second": 20.173, + "eval_steps_per_second": 0.847, + "step": 520 + }, + { + "epoch": 0.5351823317925013, + "grad_norm": 2.6119930744171143, + "learning_rate": 4.592691282832128e-06, + "loss": 1.6897, + "step": 521 + }, + { + "epoch": 0.5362095531587057, + "grad_norm": 2.7165610790252686, + "learning_rate": 4.5907879710696615e-06, + "loss": 1.7038, + "step": 522 + }, + { + "epoch": 0.5372367745249101, + "grad_norm": 2.3613736629486084, + "learning_rate": 4.588884659307195e-06, + "loss": 1.6222, + "step": 523 + }, + { + "epoch": 0.5382639958911145, + "grad_norm": 2.647672176361084, + "learning_rate": 4.586981347544728e-06, + "loss": 1.702, + "step": 524 + }, + { + "epoch": 0.539291217257319, + "grad_norm": 2.704634428024292, + "learning_rate": 4.585078035782262e-06, + "loss": 1.6566, + "step": 525 + }, + { + "epoch": 0.5403184386235234, + "grad_norm": 2.6634597778320312, + "learning_rate": 4.583174724019795e-06, + "loss": 1.6903, + "step": 526 + }, + { + "epoch": 0.5413456599897278, + "grad_norm": 2.5745162963867188, + "learning_rate": 4.581271412257328e-06, + "loss": 1.6375, + "step": 527 + }, + { + "epoch": 0.5423728813559322, + "grad_norm": 2.5577778816223145, + "learning_rate": 4.579368100494862e-06, + "loss": 1.614, + "step": 528 + }, + { + "epoch": 0.5434001027221366, + "grad_norm": 2.7684454917907715, + "learning_rate": 4.577464788732395e-06, + "loss": 1.7168, + "step": 529 + }, + { + "epoch": 0.544427324088341, + "grad_norm": 2.582108736038208, + "learning_rate": 4.575561476969928e-06, + "loss": 1.6897, + "step": 530 + }, + { + "epoch": 0.544427324088341, + "eval_loss": 1.9821091890335083, + "eval_runtime": 50.9409, + "eval_samples_per_second": 19.631, + "eval_steps_per_second": 0.824, + "step": 530 + }, + { + "epoch": 0.5454545454545454, + "grad_norm": 2.516486406326294, + "learning_rate": 4.573658165207461e-06, + "loss": 1.7273, + "step": 531 + }, + { + "epoch": 0.5464817668207499, + "grad_norm": 2.786806344985962, + "learning_rate": 4.571754853444994e-06, + "loss": 1.6898, + "step": 532 + }, + { + "epoch": 0.5475089881869543, + "grad_norm": 2.3137319087982178, + "learning_rate": 4.569851541682528e-06, + "loss": 1.676, + "step": 533 + }, + { + "epoch": 0.5485362095531587, + "grad_norm": 2.7268903255462646, + "learning_rate": 4.567948229920061e-06, + "loss": 1.6483, + "step": 534 + }, + { + "epoch": 0.5495634309193631, + "grad_norm": 2.4385902881622314, + "learning_rate": 4.566044918157594e-06, + "loss": 1.6172, + "step": 535 + }, + { + "epoch": 0.5505906522855676, + "grad_norm": 2.420576333999634, + "learning_rate": 4.564141606395128e-06, + "loss": 1.6569, + "step": 536 + }, + { + "epoch": 0.551617873651772, + "grad_norm": 2.592290163040161, + "learning_rate": 4.562238294632661e-06, + "loss": 1.7085, + "step": 537 + }, + { + "epoch": 0.5526450950179764, + "grad_norm": 2.485285520553589, + "learning_rate": 4.5603349828701945e-06, + "loss": 1.6699, + "step": 538 + }, + { + "epoch": 0.5536723163841808, + "grad_norm": 2.844710111618042, + "learning_rate": 4.558431671107728e-06, + "loss": 1.6873, + "step": 539 + }, + { + "epoch": 0.5546995377503852, + "grad_norm": 2.5279181003570557, + "learning_rate": 4.556528359345261e-06, + "loss": 1.7285, + "step": 540 + }, + { + "epoch": 0.5546995377503852, + "eval_loss": 1.9825981855392456, + "eval_runtime": 52.461, + "eval_samples_per_second": 19.062, + "eval_steps_per_second": 0.801, + "step": 540 + }, + { + "epoch": 0.5557267591165896, + "grad_norm": 2.673997163772583, + "learning_rate": 4.5546250475827946e-06, + "loss": 1.6658, + "step": 541 + }, + { + "epoch": 0.556753980482794, + "grad_norm": 2.6023194789886475, + "learning_rate": 4.552721735820328e-06, + "loss": 1.6262, + "step": 542 + }, + { + "epoch": 0.5577812018489985, + "grad_norm": 2.6327662467956543, + "learning_rate": 4.550818424057861e-06, + "loss": 1.6574, + "step": 543 + }, + { + "epoch": 0.5588084232152029, + "grad_norm": 2.5532140731811523, + "learning_rate": 4.548915112295395e-06, + "loss": 1.6607, + "step": 544 + }, + { + "epoch": 0.5598356445814073, + "grad_norm": 2.523966073989868, + "learning_rate": 4.547011800532928e-06, + "loss": 1.6752, + "step": 545 + }, + { + "epoch": 0.5608628659476117, + "grad_norm": 2.6338107585906982, + "learning_rate": 4.545108488770461e-06, + "loss": 1.6258, + "step": 546 + }, + { + "epoch": 0.5618900873138162, + "grad_norm": 2.500096559524536, + "learning_rate": 4.543205177007995e-06, + "loss": 1.7189, + "step": 547 + }, + { + "epoch": 0.5629173086800205, + "grad_norm": 2.644413471221924, + "learning_rate": 4.541301865245527e-06, + "loss": 1.6594, + "step": 548 + }, + { + "epoch": 0.5639445300462249, + "grad_norm": 2.451965093612671, + "learning_rate": 4.539398553483061e-06, + "loss": 1.6984, + "step": 549 + }, + { + "epoch": 0.5649717514124294, + "grad_norm": 2.7554423809051514, + "learning_rate": 4.537495241720594e-06, + "loss": 1.6626, + "step": 550 + }, + { + "epoch": 0.5649717514124294, + "eval_loss": 1.9897041320800781, + "eval_runtime": 45.9349, + "eval_samples_per_second": 21.77, + "eval_steps_per_second": 0.914, + "step": 550 + }, + { + "epoch": 0.5659989727786338, + "grad_norm": 2.5962674617767334, + "learning_rate": 4.535591929958127e-06, + "loss": 1.7015, + "step": 551 + }, + { + "epoch": 0.5670261941448382, + "grad_norm": 2.7917397022247314, + "learning_rate": 4.533688618195661e-06, + "loss": 1.6811, + "step": 552 + }, + { + "epoch": 0.5680534155110426, + "grad_norm": 2.4205498695373535, + "learning_rate": 4.531785306433194e-06, + "loss": 1.6659, + "step": 553 + }, + { + "epoch": 0.5690806368772471, + "grad_norm": 2.74898624420166, + "learning_rate": 4.529881994670727e-06, + "loss": 1.6521, + "step": 554 + }, + { + "epoch": 0.5701078582434514, + "grad_norm": 2.859898328781128, + "learning_rate": 4.527978682908261e-06, + "loss": 1.6445, + "step": 555 + }, + { + "epoch": 0.5711350796096559, + "grad_norm": 2.5526154041290283, + "learning_rate": 4.526075371145794e-06, + "loss": 1.6679, + "step": 556 + }, + { + "epoch": 0.5721623009758603, + "grad_norm": 2.96175217628479, + "learning_rate": 4.5241720593833275e-06, + "loss": 1.6784, + "step": 557 + }, + { + "epoch": 0.5731895223420647, + "grad_norm": 2.5037007331848145, + "learning_rate": 4.52226874762086e-06, + "loss": 1.6689, + "step": 558 + }, + { + "epoch": 0.5742167437082691, + "grad_norm": 2.6513259410858154, + "learning_rate": 4.520365435858393e-06, + "loss": 1.7038, + "step": 559 + }, + { + "epoch": 0.5752439650744735, + "grad_norm": 2.762770175933838, + "learning_rate": 4.518462124095928e-06, + "loss": 1.728, + "step": 560 + }, + { + "epoch": 0.5752439650744735, + "eval_loss": 1.972934603691101, + "eval_runtime": 44.2572, + "eval_samples_per_second": 22.595, + "eval_steps_per_second": 0.949, + "step": 560 + }, + { + "epoch": 0.576271186440678, + "grad_norm": 2.637406349182129, + "learning_rate": 4.516558812333461e-06, + "loss": 1.682, + "step": 561 + }, + { + "epoch": 0.5772984078068824, + "grad_norm": 2.545583724975586, + "learning_rate": 4.514655500570994e-06, + "loss": 1.6669, + "step": 562 + }, + { + "epoch": 0.5783256291730868, + "grad_norm": 2.5655033588409424, + "learning_rate": 4.512752188808528e-06, + "loss": 1.7279, + "step": 563 + }, + { + "epoch": 0.5793528505392912, + "grad_norm": 2.6856470108032227, + "learning_rate": 4.510848877046061e-06, + "loss": 1.7331, + "step": 564 + }, + { + "epoch": 0.5803800719054957, + "grad_norm": 2.605252981185913, + "learning_rate": 4.5089455652835936e-06, + "loss": 1.6904, + "step": 565 + }, + { + "epoch": 0.5814072932717, + "grad_norm": 2.610267400741577, + "learning_rate": 4.507042253521127e-06, + "loss": 1.6607, + "step": 566 + }, + { + "epoch": 0.5824345146379045, + "grad_norm": 2.6955432891845703, + "learning_rate": 4.50513894175866e-06, + "loss": 1.6166, + "step": 567 + }, + { + "epoch": 0.5834617360041089, + "grad_norm": 2.5661842823028564, + "learning_rate": 4.503235629996194e-06, + "loss": 1.6955, + "step": 568 + }, + { + "epoch": 0.5844889573703133, + "grad_norm": 2.6393115520477295, + "learning_rate": 4.501332318233727e-06, + "loss": 1.6933, + "step": 569 + }, + { + "epoch": 0.5855161787365177, + "grad_norm": 2.612645387649536, + "learning_rate": 4.49942900647126e-06, + "loss": 1.6585, + "step": 570 + }, + { + "epoch": 0.5855161787365177, + "eval_loss": 1.9599064588546753, + "eval_runtime": 46.0676, + "eval_samples_per_second": 21.707, + "eval_steps_per_second": 0.912, + "step": 570 + }, + { + "epoch": 0.5865434001027221, + "grad_norm": 2.4182891845703125, + "learning_rate": 4.497525694708794e-06, + "loss": 1.707, + "step": 571 + }, + { + "epoch": 0.5875706214689266, + "grad_norm": 2.7214412689208984, + "learning_rate": 4.495622382946327e-06, + "loss": 1.6966, + "step": 572 + }, + { + "epoch": 0.588597842835131, + "grad_norm": 2.5138018131256104, + "learning_rate": 4.4937190711838605e-06, + "loss": 1.6373, + "step": 573 + }, + { + "epoch": 0.5896250642013354, + "grad_norm": 2.5814292430877686, + "learning_rate": 4.491815759421394e-06, + "loss": 1.7313, + "step": 574 + }, + { + "epoch": 0.5906522855675398, + "grad_norm": 2.6236367225646973, + "learning_rate": 4.489912447658926e-06, + "loss": 1.6647, + "step": 575 + }, + { + "epoch": 0.5916795069337443, + "grad_norm": 2.468557357788086, + "learning_rate": 4.48800913589646e-06, + "loss": 1.6412, + "step": 576 + }, + { + "epoch": 0.5927067282999486, + "grad_norm": 2.454141855239868, + "learning_rate": 4.486105824133993e-06, + "loss": 1.6762, + "step": 577 + }, + { + "epoch": 0.593733949666153, + "grad_norm": 3.0009915828704834, + "learning_rate": 4.4842025123715264e-06, + "loss": 1.6847, + "step": 578 + }, + { + "epoch": 0.5947611710323575, + "grad_norm": 2.574394702911377, + "learning_rate": 4.48229920060906e-06, + "loss": 1.6421, + "step": 579 + }, + { + "epoch": 0.5957883923985619, + "grad_norm": 2.6978182792663574, + "learning_rate": 4.480395888846594e-06, + "loss": 1.6198, + "step": 580 + }, + { + "epoch": 0.5957883923985619, + "eval_loss": 1.9570144414901733, + "eval_runtime": 44.7661, + "eval_samples_per_second": 22.338, + "eval_steps_per_second": 0.938, + "step": 580 + }, + { + "epoch": 0.5968156137647663, + "grad_norm": 2.652461051940918, + "learning_rate": 4.478492577084127e-06, + "loss": 1.7334, + "step": 581 + }, + { + "epoch": 0.5978428351309707, + "grad_norm": 2.6292548179626465, + "learning_rate": 4.47658926532166e-06, + "loss": 1.6242, + "step": 582 + }, + { + "epoch": 0.5988700564971752, + "grad_norm": 2.474919080734253, + "learning_rate": 4.474685953559193e-06, + "loss": 1.6732, + "step": 583 + }, + { + "epoch": 0.5998972778633795, + "grad_norm": 2.7282516956329346, + "learning_rate": 4.472782641796727e-06, + "loss": 1.802, + "step": 584 + }, + { + "epoch": 0.600924499229584, + "grad_norm": 2.770307779312134, + "learning_rate": 4.47087933003426e-06, + "loss": 1.6156, + "step": 585 + }, + { + "epoch": 0.6019517205957884, + "grad_norm": 2.570681095123291, + "learning_rate": 4.468976018271793e-06, + "loss": 1.6309, + "step": 586 + }, + { + "epoch": 0.6029789419619929, + "grad_norm": 2.4075732231140137, + "learning_rate": 4.467072706509327e-06, + "loss": 1.6436, + "step": 587 + }, + { + "epoch": 0.6040061633281972, + "grad_norm": 2.4570937156677246, + "learning_rate": 4.46516939474686e-06, + "loss": 1.6737, + "step": 588 + }, + { + "epoch": 0.6050333846944016, + "grad_norm": 2.407274007797241, + "learning_rate": 4.463266082984393e-06, + "loss": 1.6179, + "step": 589 + }, + { + "epoch": 0.6060606060606061, + "grad_norm": 2.745486259460449, + "learning_rate": 4.461362771221927e-06, + "loss": 1.7458, + "step": 590 + }, + { + "epoch": 0.6060606060606061, + "eval_loss": 1.9700438976287842, + "eval_runtime": 45.3018, + "eval_samples_per_second": 22.074, + "eval_steps_per_second": 0.927, + "step": 590 + }, + { + "epoch": 0.6070878274268104, + "grad_norm": 2.4258809089660645, + "learning_rate": 4.45945945945946e-06, + "loss": 1.6372, + "step": 591 + }, + { + "epoch": 0.6081150487930149, + "grad_norm": 2.683662176132202, + "learning_rate": 4.457556147696993e-06, + "loss": 1.6649, + "step": 592 + }, + { + "epoch": 0.6091422701592193, + "grad_norm": 2.879727602005005, + "learning_rate": 4.455652835934526e-06, + "loss": 1.7376, + "step": 593 + }, + { + "epoch": 0.6101694915254238, + "grad_norm": 2.5620689392089844, + "learning_rate": 4.453749524172059e-06, + "loss": 1.7203, + "step": 594 + }, + { + "epoch": 0.6111967128916281, + "grad_norm": 2.662031888961792, + "learning_rate": 4.451846212409593e-06, + "loss": 1.6835, + "step": 595 + }, + { + "epoch": 0.6122239342578326, + "grad_norm": 2.6269922256469727, + "learning_rate": 4.449942900647126e-06, + "loss": 1.6726, + "step": 596 + }, + { + "epoch": 0.613251155624037, + "grad_norm": 2.515852212905884, + "learning_rate": 4.4480395888846595e-06, + "loss": 1.6045, + "step": 597 + }, + { + "epoch": 0.6142783769902413, + "grad_norm": 3.006873607635498, + "learning_rate": 4.446136277122193e-06, + "loss": 1.6827, + "step": 598 + }, + { + "epoch": 0.6153055983564458, + "grad_norm": 2.2904858589172363, + "learning_rate": 4.444232965359726e-06, + "loss": 1.6123, + "step": 599 + }, + { + "epoch": 0.6163328197226502, + "grad_norm": 2.7345471382141113, + "learning_rate": 4.4423296535972596e-06, + "loss": 1.6697, + "step": 600 + }, + { + "epoch": 0.6163328197226502, + "eval_loss": 1.9713997840881348, + "eval_runtime": 45.6715, + "eval_samples_per_second": 21.895, + "eval_steps_per_second": 0.92, + "step": 600 + }, + { + "epoch": 0.6173600410888547, + "grad_norm": 2.955853223800659, + "learning_rate": 4.440426341834793e-06, + "loss": 1.6195, + "step": 601 + }, + { + "epoch": 0.618387262455059, + "grad_norm": 2.4079766273498535, + "learning_rate": 4.438523030072326e-06, + "loss": 1.6589, + "step": 602 + }, + { + "epoch": 0.6194144838212635, + "grad_norm": 2.8158490657806396, + "learning_rate": 4.43661971830986e-06, + "loss": 1.6895, + "step": 603 + }, + { + "epoch": 0.6204417051874679, + "grad_norm": 2.491471529006958, + "learning_rate": 4.434716406547393e-06, + "loss": 1.6351, + "step": 604 + }, + { + "epoch": 0.6214689265536724, + "grad_norm": 2.4093260765075684, + "learning_rate": 4.432813094784926e-06, + "loss": 1.6402, + "step": 605 + }, + { + "epoch": 0.6224961479198767, + "grad_norm": 2.751018762588501, + "learning_rate": 4.43090978302246e-06, + "loss": 1.6962, + "step": 606 + }, + { + "epoch": 0.6235233692860811, + "grad_norm": 2.4410409927368164, + "learning_rate": 4.429006471259993e-06, + "loss": 1.6549, + "step": 607 + }, + { + "epoch": 0.6245505906522856, + "grad_norm": 2.360455274581909, + "learning_rate": 4.4271031594975265e-06, + "loss": 1.6449, + "step": 608 + }, + { + "epoch": 0.6255778120184899, + "grad_norm": 2.5058023929595947, + "learning_rate": 4.425199847735059e-06, + "loss": 1.617, + "step": 609 + }, + { + "epoch": 0.6266050333846944, + "grad_norm": 2.3429927825927734, + "learning_rate": 4.423296535972592e-06, + "loss": 1.6262, + "step": 610 + }, + { + "epoch": 0.6266050333846944, + "eval_loss": 1.9597365856170654, + "eval_runtime": 42.2034, + "eval_samples_per_second": 23.695, + "eval_steps_per_second": 0.995, + "step": 610 + }, + { + "epoch": 0.6276322547508988, + "grad_norm": 2.401761054992676, + "learning_rate": 4.421393224210126e-06, + "loss": 1.6497, + "step": 611 + }, + { + "epoch": 0.6286594761171033, + "grad_norm": 2.433439254760742, + "learning_rate": 4.419489912447659e-06, + "loss": 1.6117, + "step": 612 + }, + { + "epoch": 0.6296866974833076, + "grad_norm": 2.5147452354431152, + "learning_rate": 4.4175866006851924e-06, + "loss": 1.7041, + "step": 613 + }, + { + "epoch": 0.6307139188495121, + "grad_norm": 2.4174447059631348, + "learning_rate": 4.415683288922726e-06, + "loss": 1.6652, + "step": 614 + }, + { + "epoch": 0.6317411402157165, + "grad_norm": 2.424800157546997, + "learning_rate": 4.413779977160259e-06, + "loss": 1.6407, + "step": 615 + }, + { + "epoch": 0.632768361581921, + "grad_norm": 2.6760315895080566, + "learning_rate": 4.4118766653977925e-06, + "loss": 1.6566, + "step": 616 + }, + { + "epoch": 0.6337955829481253, + "grad_norm": 2.694810628890991, + "learning_rate": 4.409973353635326e-06, + "loss": 1.7125, + "step": 617 + }, + { + "epoch": 0.6348228043143297, + "grad_norm": 2.475126266479492, + "learning_rate": 4.408070041872859e-06, + "loss": 1.6663, + "step": 618 + }, + { + "epoch": 0.6358500256805342, + "grad_norm": 2.734738349914551, + "learning_rate": 4.406166730110393e-06, + "loss": 1.6115, + "step": 619 + }, + { + "epoch": 0.6368772470467385, + "grad_norm": 2.4914753437042236, + "learning_rate": 4.404263418347925e-06, + "loss": 1.6096, + "step": 620 + }, + { + "epoch": 0.6368772470467385, + "eval_loss": 1.952081322669983, + "eval_runtime": 45.4451, + "eval_samples_per_second": 22.005, + "eval_steps_per_second": 0.924, + "step": 620 + }, + { + "epoch": 0.637904468412943, + "grad_norm": 2.3727800846099854, + "learning_rate": 4.402360106585459e-06, + "loss": 1.6275, + "step": 621 + }, + { + "epoch": 0.6389316897791474, + "grad_norm": 2.576402187347412, + "learning_rate": 4.400456794822993e-06, + "loss": 1.6252, + "step": 622 + }, + { + "epoch": 0.6399589111453519, + "grad_norm": 2.560185432434082, + "learning_rate": 4.398553483060526e-06, + "loss": 1.6522, + "step": 623 + }, + { + "epoch": 0.6409861325115562, + "grad_norm": 2.462785005569458, + "learning_rate": 4.396650171298059e-06, + "loss": 1.6124, + "step": 624 + }, + { + "epoch": 0.6420133538777607, + "grad_norm": 2.3942930698394775, + "learning_rate": 4.394746859535593e-06, + "loss": 1.6231, + "step": 625 + }, + { + "epoch": 0.6430405752439651, + "grad_norm": 2.4870522022247314, + "learning_rate": 4.392843547773125e-06, + "loss": 1.6086, + "step": 626 + }, + { + "epoch": 0.6440677966101694, + "grad_norm": 2.5702786445617676, + "learning_rate": 4.390940236010659e-06, + "loss": 1.6228, + "step": 627 + }, + { + "epoch": 0.6450950179763739, + "grad_norm": 2.743983268737793, + "learning_rate": 4.389036924248192e-06, + "loss": 1.6394, + "step": 628 + }, + { + "epoch": 0.6461222393425783, + "grad_norm": 2.59942364692688, + "learning_rate": 4.387133612485725e-06, + "loss": 1.6721, + "step": 629 + }, + { + "epoch": 0.6471494607087828, + "grad_norm": 2.4473204612731934, + "learning_rate": 4.385230300723259e-06, + "loss": 1.6407, + "step": 630 + }, + { + "epoch": 0.6471494607087828, + "eval_loss": 1.9426974058151245, + "eval_runtime": 45.2318, + "eval_samples_per_second": 22.108, + "eval_steps_per_second": 0.929, + "step": 630 + }, + { + "epoch": 0.6481766820749871, + "grad_norm": 2.6036875247955322, + "learning_rate": 4.383326988960792e-06, + "loss": 1.6307, + "step": 631 + }, + { + "epoch": 0.6492039034411916, + "grad_norm": 2.691011428833008, + "learning_rate": 4.3814236771983255e-06, + "loss": 1.6059, + "step": 632 + }, + { + "epoch": 0.650231124807396, + "grad_norm": 2.585758924484253, + "learning_rate": 4.379520365435859e-06, + "loss": 1.6781, + "step": 633 + }, + { + "epoch": 0.6512583461736005, + "grad_norm": 2.9522950649261475, + "learning_rate": 4.377617053673392e-06, + "loss": 1.6559, + "step": 634 + }, + { + "epoch": 0.6522855675398048, + "grad_norm": 2.5368709564208984, + "learning_rate": 4.3757137419109256e-06, + "loss": 1.6445, + "step": 635 + }, + { + "epoch": 0.6533127889060092, + "grad_norm": 2.5002150535583496, + "learning_rate": 4.373810430148459e-06, + "loss": 1.7184, + "step": 636 + }, + { + "epoch": 0.6543400102722137, + "grad_norm": 2.7623229026794434, + "learning_rate": 4.3719071183859914e-06, + "loss": 1.6195, + "step": 637 + }, + { + "epoch": 0.655367231638418, + "grad_norm": 2.6188151836395264, + "learning_rate": 4.370003806623525e-06, + "loss": 1.6686, + "step": 638 + }, + { + "epoch": 0.6563944530046225, + "grad_norm": 2.685776948928833, + "learning_rate": 4.368100494861058e-06, + "loss": 1.6134, + "step": 639 + }, + { + "epoch": 0.6574216743708269, + "grad_norm": 2.5703907012939453, + "learning_rate": 4.3661971830985915e-06, + "loss": 1.665, + "step": 640 + }, + { + "epoch": 0.6574216743708269, + "eval_loss": 1.9504806995391846, + "eval_runtime": 47.1947, + "eval_samples_per_second": 21.189, + "eval_steps_per_second": 0.89, + "step": 640 + }, + { + "epoch": 0.6584488957370314, + "grad_norm": 2.511617660522461, + "learning_rate": 4.364293871336126e-06, + "loss": 1.6201, + "step": 641 + }, + { + "epoch": 0.6594761171032357, + "grad_norm": 2.549302339553833, + "learning_rate": 4.362390559573659e-06, + "loss": 1.647, + "step": 642 + }, + { + "epoch": 0.6605033384694402, + "grad_norm": 2.348747730255127, + "learning_rate": 4.3604872478111925e-06, + "loss": 1.5993, + "step": 643 + }, + { + "epoch": 0.6615305598356446, + "grad_norm": 2.7393345832824707, + "learning_rate": 4.358583936048725e-06, + "loss": 1.543, + "step": 644 + }, + { + "epoch": 0.662557781201849, + "grad_norm": 2.4242193698883057, + "learning_rate": 4.356680624286258e-06, + "loss": 1.6726, + "step": 645 + }, + { + "epoch": 0.6635850025680534, + "grad_norm": 2.6446714401245117, + "learning_rate": 4.354777312523792e-06, + "loss": 1.6312, + "step": 646 + }, + { + "epoch": 0.6646122239342578, + "grad_norm": 2.7407124042510986, + "learning_rate": 4.352874000761325e-06, + "loss": 1.6868, + "step": 647 + }, + { + "epoch": 0.6656394453004623, + "grad_norm": 2.3706371784210205, + "learning_rate": 4.3509706889988584e-06, + "loss": 1.6301, + "step": 648 + }, + { + "epoch": 0.6666666666666666, + "grad_norm": 2.5197198390960693, + "learning_rate": 4.349067377236392e-06, + "loss": 1.702, + "step": 649 + }, + { + "epoch": 0.6676938880328711, + "grad_norm": 2.4792377948760986, + "learning_rate": 4.347164065473925e-06, + "loss": 1.6891, + "step": 650 + }, + { + "epoch": 0.6676938880328711, + "eval_loss": 1.9589039087295532, + "eval_runtime": 47.1512, + "eval_samples_per_second": 21.208, + "eval_steps_per_second": 0.891, + "step": 650 + }, + { + "epoch": 0.6687211093990755, + "grad_norm": 2.4003701210021973, + "learning_rate": 4.3452607537114585e-06, + "loss": 1.7015, + "step": 651 + }, + { + "epoch": 0.66974833076528, + "grad_norm": 2.5363786220550537, + "learning_rate": 4.343357441948992e-06, + "loss": 1.6133, + "step": 652 + }, + { + "epoch": 0.6707755521314843, + "grad_norm": 2.5183844566345215, + "learning_rate": 4.341454130186525e-06, + "loss": 1.6574, + "step": 653 + }, + { + "epoch": 0.6718027734976888, + "grad_norm": 2.4668304920196533, + "learning_rate": 4.339550818424058e-06, + "loss": 1.6253, + "step": 654 + }, + { + "epoch": 0.6728299948638932, + "grad_norm": 2.4829814434051514, + "learning_rate": 4.337647506661591e-06, + "loss": 1.6099, + "step": 655 + }, + { + "epoch": 0.6738572162300975, + "grad_norm": 2.3683176040649414, + "learning_rate": 4.3357441948991245e-06, + "loss": 1.6385, + "step": 656 + }, + { + "epoch": 0.674884437596302, + "grad_norm": 2.4906842708587646, + "learning_rate": 4.333840883136658e-06, + "loss": 1.6755, + "step": 657 + }, + { + "epoch": 0.6759116589625064, + "grad_norm": 2.558776617050171, + "learning_rate": 4.331937571374191e-06, + "loss": 1.6152, + "step": 658 + }, + { + "epoch": 0.6769388803287109, + "grad_norm": 2.4081523418426514, + "learning_rate": 4.3300342596117246e-06, + "loss": 1.6224, + "step": 659 + }, + { + "epoch": 0.6779661016949152, + "grad_norm": 2.488370180130005, + "learning_rate": 4.328130947849258e-06, + "loss": 1.6298, + "step": 660 + }, + { + "epoch": 0.6779661016949152, + "eval_loss": 1.9484621286392212, + "eval_runtime": 52.549, + "eval_samples_per_second": 19.03, + "eval_steps_per_second": 0.799, + "step": 660 + }, + { + "epoch": 0.6789933230611197, + "grad_norm": 2.495748281478882, + "learning_rate": 4.326227636086791e-06, + "loss": 1.5852, + "step": 661 + }, + { + "epoch": 0.6800205444273241, + "grad_norm": 2.525031566619873, + "learning_rate": 4.324324324324325e-06, + "loss": 1.6348, + "step": 662 + }, + { + "epoch": 0.6810477657935285, + "grad_norm": 2.427615165710449, + "learning_rate": 4.322421012561858e-06, + "loss": 1.6352, + "step": 663 + }, + { + "epoch": 0.6820749871597329, + "grad_norm": 2.5084927082061768, + "learning_rate": 4.320517700799391e-06, + "loss": 1.6334, + "step": 664 + }, + { + "epoch": 0.6831022085259374, + "grad_norm": 2.6018898487091064, + "learning_rate": 4.318614389036925e-06, + "loss": 1.7212, + "step": 665 + }, + { + "epoch": 0.6841294298921418, + "grad_norm": 2.44184947013855, + "learning_rate": 4.316711077274458e-06, + "loss": 1.6802, + "step": 666 + }, + { + "epoch": 0.6851566512583461, + "grad_norm": 2.542123317718506, + "learning_rate": 4.3148077655119915e-06, + "loss": 1.5912, + "step": 667 + }, + { + "epoch": 0.6861838726245506, + "grad_norm": 2.5208137035369873, + "learning_rate": 4.312904453749525e-06, + "loss": 1.6965, + "step": 668 + }, + { + "epoch": 0.687211093990755, + "grad_norm": 2.391904592514038, + "learning_rate": 4.311001141987058e-06, + "loss": 1.6388, + "step": 669 + }, + { + "epoch": 0.6882383153569595, + "grad_norm": 2.629565477371216, + "learning_rate": 4.3090978302245916e-06, + "loss": 1.6475, + "step": 670 + }, + { + "epoch": 0.6882383153569595, + "eval_loss": 1.9572818279266357, + "eval_runtime": 48.5354, + "eval_samples_per_second": 20.604, + "eval_steps_per_second": 0.865, + "step": 670 + }, + { + "epoch": 0.6892655367231638, + "grad_norm": 2.3132376670837402, + "learning_rate": 4.307194518462124e-06, + "loss": 1.6399, + "step": 671 + }, + { + "epoch": 0.6902927580893683, + "grad_norm": 2.5003411769866943, + "learning_rate": 4.3052912066996574e-06, + "loss": 1.7136, + "step": 672 + }, + { + "epoch": 0.6913199794555727, + "grad_norm": 2.545541286468506, + "learning_rate": 4.303387894937191e-06, + "loss": 1.6825, + "step": 673 + }, + { + "epoch": 0.6923472008217771, + "grad_norm": 2.342423439025879, + "learning_rate": 4.301484583174724e-06, + "loss": 1.6088, + "step": 674 + }, + { + "epoch": 0.6933744221879815, + "grad_norm": 2.5516135692596436, + "learning_rate": 4.2995812714122575e-06, + "loss": 1.5934, + "step": 675 + }, + { + "epoch": 0.6944016435541859, + "grad_norm": 2.438675880432129, + "learning_rate": 4.297677959649791e-06, + "loss": 1.6236, + "step": 676 + }, + { + "epoch": 0.6954288649203904, + "grad_norm": 2.767305612564087, + "learning_rate": 4.295774647887324e-06, + "loss": 1.6514, + "step": 677 + }, + { + "epoch": 0.6964560862865947, + "grad_norm": 2.591994047164917, + "learning_rate": 4.293871336124858e-06, + "loss": 1.6643, + "step": 678 + }, + { + "epoch": 0.6974833076527992, + "grad_norm": 2.4734244346618652, + "learning_rate": 4.291968024362391e-06, + "loss": 1.6362, + "step": 679 + }, + { + "epoch": 0.6985105290190036, + "grad_norm": 2.6181890964508057, + "learning_rate": 4.290064712599924e-06, + "loss": 1.6666, + "step": 680 + }, + { + "epoch": 0.6985105290190036, + "eval_loss": 1.9585388898849487, + "eval_runtime": 46.8094, + "eval_samples_per_second": 21.363, + "eval_steps_per_second": 0.897, + "step": 680 + }, + { + "epoch": 0.699537750385208, + "grad_norm": 2.465571403503418, + "learning_rate": 4.288161400837458e-06, + "loss": 1.6588, + "step": 681 + }, + { + "epoch": 0.7005649717514124, + "grad_norm": 2.664262294769287, + "learning_rate": 4.286258089074991e-06, + "loss": 1.5862, + "step": 682 + }, + { + "epoch": 0.7015921931176169, + "grad_norm": 2.677140474319458, + "learning_rate": 4.284354777312524e-06, + "loss": 1.6549, + "step": 683 + }, + { + "epoch": 0.7026194144838213, + "grad_norm": 2.559051275253296, + "learning_rate": 4.282451465550058e-06, + "loss": 1.609, + "step": 684 + }, + { + "epoch": 0.7036466358500256, + "grad_norm": 2.428424119949341, + "learning_rate": 4.280548153787591e-06, + "loss": 1.6243, + "step": 685 + }, + { + "epoch": 0.7046738572162301, + "grad_norm": 2.54136061668396, + "learning_rate": 4.2786448420251245e-06, + "loss": 1.6301, + "step": 686 + }, + { + "epoch": 0.7057010785824345, + "grad_norm": 2.377699136734009, + "learning_rate": 4.276741530262658e-06, + "loss": 1.5903, + "step": 687 + }, + { + "epoch": 0.706728299948639, + "grad_norm": 2.5548322200775146, + "learning_rate": 4.27483821850019e-06, + "loss": 1.6161, + "step": 688 + }, + { + "epoch": 0.7077555213148433, + "grad_norm": 2.5993409156799316, + "learning_rate": 4.272934906737724e-06, + "loss": 1.6882, + "step": 689 + }, + { + "epoch": 0.7087827426810478, + "grad_norm": 2.4249117374420166, + "learning_rate": 4.271031594975257e-06, + "loss": 1.6723, + "step": 690 + }, + { + "epoch": 0.7087827426810478, + "eval_loss": 1.9526821374893188, + "eval_runtime": 48.2168, + "eval_samples_per_second": 20.74, + "eval_steps_per_second": 0.871, + "step": 690 + }, + { + "epoch": 0.7098099640472522, + "grad_norm": 2.6965081691741943, + "learning_rate": 4.2691282832127905e-06, + "loss": 1.6636, + "step": 691 + }, + { + "epoch": 0.7108371854134566, + "grad_norm": 2.6205356121063232, + "learning_rate": 4.267224971450324e-06, + "loss": 1.6231, + "step": 692 + }, + { + "epoch": 0.711864406779661, + "grad_norm": 2.5924129486083984, + "learning_rate": 4.265321659687857e-06, + "loss": 1.5634, + "step": 693 + }, + { + "epoch": 0.7128916281458655, + "grad_norm": 2.9710943698883057, + "learning_rate": 4.2634183479253906e-06, + "loss": 1.6134, + "step": 694 + }, + { + "epoch": 0.7139188495120699, + "grad_norm": 2.6252782344818115, + "learning_rate": 4.261515036162924e-06, + "loss": 1.5997, + "step": 695 + }, + { + "epoch": 0.7149460708782742, + "grad_norm": 2.6178832054138184, + "learning_rate": 4.259611724400457e-06, + "loss": 1.6255, + "step": 696 + }, + { + "epoch": 0.7159732922444787, + "grad_norm": 2.82548189163208, + "learning_rate": 4.257708412637991e-06, + "loss": 1.6167, + "step": 697 + }, + { + "epoch": 0.7170005136106831, + "grad_norm": 2.6430351734161377, + "learning_rate": 4.255805100875523e-06, + "loss": 1.5585, + "step": 698 + }, + { + "epoch": 0.7180277349768875, + "grad_norm": 2.85046124458313, + "learning_rate": 4.2539017891130565e-06, + "loss": 1.5873, + "step": 699 + }, + { + "epoch": 0.7190549563430919, + "grad_norm": 2.5643832683563232, + "learning_rate": 4.25199847735059e-06, + "loss": 1.6705, + "step": 700 + }, + { + "epoch": 0.7190549563430919, + "eval_loss": 1.945617437362671, + "eval_runtime": 54.2778, + "eval_samples_per_second": 18.424, + "eval_steps_per_second": 0.774, + "step": 700 + }, + { + "epoch": 0.7200821777092964, + "grad_norm": 2.7856431007385254, + "learning_rate": 4.250095165588123e-06, + "loss": 1.5833, + "step": 701 + }, + { + "epoch": 0.7211093990755008, + "grad_norm": 2.9881107807159424, + "learning_rate": 4.2481918538256575e-06, + "loss": 1.6438, + "step": 702 + }, + { + "epoch": 0.7221366204417052, + "grad_norm": 2.602938652038574, + "learning_rate": 4.246288542063191e-06, + "loss": 1.6135, + "step": 703 + }, + { + "epoch": 0.7231638418079096, + "grad_norm": 2.997450590133667, + "learning_rate": 4.244385230300724e-06, + "loss": 1.6419, + "step": 704 + }, + { + "epoch": 0.724191063174114, + "grad_norm": 2.5592496395111084, + "learning_rate": 4.242481918538257e-06, + "loss": 1.5886, + "step": 705 + }, + { + "epoch": 0.7252182845403184, + "grad_norm": 2.4073948860168457, + "learning_rate": 4.24057860677579e-06, + "loss": 1.6219, + "step": 706 + }, + { + "epoch": 0.7262455059065228, + "grad_norm": 2.545997381210327, + "learning_rate": 4.2386752950133234e-06, + "loss": 1.6684, + "step": 707 + }, + { + "epoch": 0.7272727272727273, + "grad_norm": 2.468264579772949, + "learning_rate": 4.236771983250857e-06, + "loss": 1.6155, + "step": 708 + }, + { + "epoch": 0.7282999486389317, + "grad_norm": 2.6635043621063232, + "learning_rate": 4.23486867148839e-06, + "loss": 1.5956, + "step": 709 + }, + { + "epoch": 0.7293271700051361, + "grad_norm": 2.8751156330108643, + "learning_rate": 4.2329653597259235e-06, + "loss": 1.6211, + "step": 710 + }, + { + "epoch": 0.7293271700051361, + "eval_loss": 1.9319628477096558, + "eval_runtime": 43.9701, + "eval_samples_per_second": 22.743, + "eval_steps_per_second": 0.955, + "step": 710 + }, + { + "epoch": 0.7303543913713405, + "grad_norm": 2.5999035835266113, + "learning_rate": 4.231062047963457e-06, + "loss": 1.6442, + "step": 711 + }, + { + "epoch": 0.731381612737545, + "grad_norm": 2.611095666885376, + "learning_rate": 4.22915873620099e-06, + "loss": 1.6745, + "step": 712 + }, + { + "epoch": 0.7324088341037494, + "grad_norm": 2.7235333919525146, + "learning_rate": 4.227255424438524e-06, + "loss": 1.6353, + "step": 713 + }, + { + "epoch": 0.7334360554699538, + "grad_norm": 2.453639507293701, + "learning_rate": 4.225352112676057e-06, + "loss": 1.6657, + "step": 714 + }, + { + "epoch": 0.7344632768361582, + "grad_norm": 2.4031429290771484, + "learning_rate": 4.2234488009135895e-06, + "loss": 1.603, + "step": 715 + }, + { + "epoch": 0.7354904982023626, + "grad_norm": 2.631052017211914, + "learning_rate": 4.221545489151123e-06, + "loss": 1.5681, + "step": 716 + }, + { + "epoch": 0.736517719568567, + "grad_norm": 2.518758773803711, + "learning_rate": 4.219642177388656e-06, + "loss": 1.5941, + "step": 717 + }, + { + "epoch": 0.7375449409347714, + "grad_norm": 2.699491500854492, + "learning_rate": 4.2177388656261896e-06, + "loss": 1.6327, + "step": 718 + }, + { + "epoch": 0.7385721623009759, + "grad_norm": 2.749356985092163, + "learning_rate": 4.215835553863723e-06, + "loss": 1.5864, + "step": 719 + }, + { + "epoch": 0.7395993836671803, + "grad_norm": 2.5699031352996826, + "learning_rate": 4.213932242101256e-06, + "loss": 1.5444, + "step": 720 + }, + { + "epoch": 0.7395993836671803, + "eval_loss": 1.9521623849868774, + "eval_runtime": 46.5749, + "eval_samples_per_second": 21.471, + "eval_steps_per_second": 0.902, + "step": 720 + }, + { + "epoch": 0.7406266050333847, + "grad_norm": 3.197425127029419, + "learning_rate": 4.21202893033879e-06, + "loss": 1.6158, + "step": 721 + }, + { + "epoch": 0.7416538263995891, + "grad_norm": 2.619459390640259, + "learning_rate": 4.210125618576323e-06, + "loss": 1.6328, + "step": 722 + }, + { + "epoch": 0.7426810477657936, + "grad_norm": 2.792086601257324, + "learning_rate": 4.208222306813856e-06, + "loss": 1.6333, + "step": 723 + }, + { + "epoch": 0.743708269131998, + "grad_norm": 2.7023141384124756, + "learning_rate": 4.20631899505139e-06, + "loss": 1.6787, + "step": 724 + }, + { + "epoch": 0.7447354904982023, + "grad_norm": 2.6319735050201416, + "learning_rate": 4.204415683288923e-06, + "loss": 1.54, + "step": 725 + }, + { + "epoch": 0.7457627118644068, + "grad_norm": 2.7032439708709717, + "learning_rate": 4.2025123715264565e-06, + "loss": 1.641, + "step": 726 + }, + { + "epoch": 0.7467899332306112, + "grad_norm": 2.9133267402648926, + "learning_rate": 4.20060905976399e-06, + "loss": 1.5901, + "step": 727 + }, + { + "epoch": 0.7478171545968156, + "grad_norm": 2.492403745651245, + "learning_rate": 4.198705748001523e-06, + "loss": 1.5516, + "step": 728 + }, + { + "epoch": 0.74884437596302, + "grad_norm": 2.604520082473755, + "learning_rate": 4.1968024362390566e-06, + "loss": 1.6574, + "step": 729 + }, + { + "epoch": 0.7498715973292245, + "grad_norm": 2.521969795227051, + "learning_rate": 4.19489912447659e-06, + "loss": 1.6099, + "step": 730 + }, + { + "epoch": 0.7498715973292245, + "eval_loss": 1.9449840784072876, + "eval_runtime": 44.9334, + "eval_samples_per_second": 22.255, + "eval_steps_per_second": 0.935, + "step": 730 + }, + { + "epoch": 0.7508988186954288, + "grad_norm": 2.4645657539367676, + "learning_rate": 4.192995812714123e-06, + "loss": 1.6752, + "step": 731 + }, + { + "epoch": 0.7519260400616333, + "grad_norm": 2.7852981090545654, + "learning_rate": 4.191092500951656e-06, + "loss": 1.6176, + "step": 732 + }, + { + "epoch": 0.7529532614278377, + "grad_norm": 2.2817575931549072, + "learning_rate": 4.189189189189189e-06, + "loss": 1.5934, + "step": 733 + }, + { + "epoch": 0.7539804827940421, + "grad_norm": 2.6218173503875732, + "learning_rate": 4.1872858774267225e-06, + "loss": 1.5902, + "step": 734 + }, + { + "epoch": 0.7550077041602465, + "grad_norm": 2.6671013832092285, + "learning_rate": 4.185382565664256e-06, + "loss": 1.657, + "step": 735 + }, + { + "epoch": 0.7560349255264509, + "grad_norm": 2.669757843017578, + "learning_rate": 4.183479253901789e-06, + "loss": 1.5884, + "step": 736 + }, + { + "epoch": 0.7570621468926554, + "grad_norm": 2.7210471630096436, + "learning_rate": 4.181575942139323e-06, + "loss": 1.5558, + "step": 737 + }, + { + "epoch": 0.7580893682588598, + "grad_norm": 2.414276123046875, + "learning_rate": 4.179672630376856e-06, + "loss": 1.6124, + "step": 738 + }, + { + "epoch": 0.7591165896250642, + "grad_norm": 2.4643094539642334, + "learning_rate": 4.177769318614389e-06, + "loss": 1.5904, + "step": 739 + }, + { + "epoch": 0.7601438109912686, + "grad_norm": 2.5447018146514893, + "learning_rate": 4.175866006851923e-06, + "loss": 1.6705, + "step": 740 + }, + { + "epoch": 0.7601438109912686, + "eval_loss": 1.9437954425811768, + "eval_runtime": 49.2702, + "eval_samples_per_second": 20.296, + "eval_steps_per_second": 0.852, + "step": 740 + }, + { + "epoch": 0.7611710323574731, + "grad_norm": 2.41048002243042, + "learning_rate": 4.173962695089456e-06, + "loss": 1.6398, + "step": 741 + }, + { + "epoch": 0.7621982537236774, + "grad_norm": 2.3949906826019287, + "learning_rate": 4.1720593833269894e-06, + "loss": 1.6198, + "step": 742 + }, + { + "epoch": 0.7632254750898819, + "grad_norm": 2.36966609954834, + "learning_rate": 4.170156071564523e-06, + "loss": 1.6275, + "step": 743 + }, + { + "epoch": 0.7642526964560863, + "grad_norm": 2.623347043991089, + "learning_rate": 4.168252759802056e-06, + "loss": 1.7059, + "step": 744 + }, + { + "epoch": 0.7652799178222907, + "grad_norm": 2.384788751602173, + "learning_rate": 4.1663494480395895e-06, + "loss": 1.6015, + "step": 745 + }, + { + "epoch": 0.7663071391884951, + "grad_norm": 2.586009979248047, + "learning_rate": 4.164446136277123e-06, + "loss": 1.6445, + "step": 746 + }, + { + "epoch": 0.7673343605546995, + "grad_norm": 2.396684169769287, + "learning_rate": 4.162542824514656e-06, + "loss": 1.6539, + "step": 747 + }, + { + "epoch": 0.768361581920904, + "grad_norm": 2.416243553161621, + "learning_rate": 4.16063951275219e-06, + "loss": 1.6142, + "step": 748 + }, + { + "epoch": 0.7693888032871083, + "grad_norm": 2.480912208557129, + "learning_rate": 4.158736200989722e-06, + "loss": 1.5768, + "step": 749 + }, + { + "epoch": 0.7704160246533128, + "grad_norm": 2.5516929626464844, + "learning_rate": 4.1568328892272555e-06, + "loss": 1.6074, + "step": 750 + }, + { + "epoch": 0.7704160246533128, + "eval_loss": 1.9409116506576538, + "eval_runtime": 45.3465, + "eval_samples_per_second": 22.052, + "eval_steps_per_second": 0.926, + "step": 750 + }, + { + "epoch": 0.7714432460195172, + "grad_norm": 2.4248714447021484, + "learning_rate": 4.154929577464789e-06, + "loss": 1.6308, + "step": 751 + }, + { + "epoch": 0.7724704673857217, + "grad_norm": 2.520810604095459, + "learning_rate": 4.153026265702322e-06, + "loss": 1.6227, + "step": 752 + }, + { + "epoch": 0.773497688751926, + "grad_norm": 2.5938422679901123, + "learning_rate": 4.1511229539398556e-06, + "loss": 1.5601, + "step": 753 + }, + { + "epoch": 0.7745249101181304, + "grad_norm": 2.4538512229919434, + "learning_rate": 4.149219642177389e-06, + "loss": 1.6006, + "step": 754 + }, + { + "epoch": 0.7755521314843349, + "grad_norm": 2.4936258792877197, + "learning_rate": 4.147316330414922e-06, + "loss": 1.6281, + "step": 755 + }, + { + "epoch": 0.7765793528505393, + "grad_norm": 2.5774738788604736, + "learning_rate": 4.145413018652456e-06, + "loss": 1.5328, + "step": 756 + }, + { + "epoch": 0.7776065742167437, + "grad_norm": 2.3616905212402344, + "learning_rate": 4.143509706889989e-06, + "loss": 1.629, + "step": 757 + }, + { + "epoch": 0.7786337955829481, + "grad_norm": 2.4983086585998535, + "learning_rate": 4.141606395127522e-06, + "loss": 1.6298, + "step": 758 + }, + { + "epoch": 0.7796610169491526, + "grad_norm": 2.4313876628875732, + "learning_rate": 4.139703083365056e-06, + "loss": 1.6113, + "step": 759 + }, + { + "epoch": 0.7806882383153569, + "grad_norm": 2.6290934085845947, + "learning_rate": 4.137799771602588e-06, + "loss": 1.6632, + "step": 760 + }, + { + "epoch": 0.7806882383153569, + "eval_loss": 1.950036644935608, + "eval_runtime": 47.4523, + "eval_samples_per_second": 21.074, + "eval_steps_per_second": 0.885, + "step": 760 + }, + { + "epoch": 0.7817154596815614, + "grad_norm": 2.3637750148773193, + "learning_rate": 4.135896459840122e-06, + "loss": 1.6268, + "step": 761 + }, + { + "epoch": 0.7827426810477658, + "grad_norm": 2.5537843704223633, + "learning_rate": 4.133993148077656e-06, + "loss": 1.5736, + "step": 762 + }, + { + "epoch": 0.7837699024139703, + "grad_norm": 2.4945273399353027, + "learning_rate": 4.132089836315189e-06, + "loss": 1.612, + "step": 763 + }, + { + "epoch": 0.7847971237801746, + "grad_norm": 2.506983757019043, + "learning_rate": 4.1301865245527226e-06, + "loss": 1.6112, + "step": 764 + }, + { + "epoch": 0.785824345146379, + "grad_norm": 2.599721908569336, + "learning_rate": 4.128283212790256e-06, + "loss": 1.5698, + "step": 765 + }, + { + "epoch": 0.7868515665125835, + "grad_norm": 2.4561266899108887, + "learning_rate": 4.1263799010277884e-06, + "loss": 1.5354, + "step": 766 + }, + { + "epoch": 0.7878787878787878, + "grad_norm": 2.445420026779175, + "learning_rate": 4.124476589265322e-06, + "loss": 1.6555, + "step": 767 + }, + { + "epoch": 0.7889060092449923, + "grad_norm": 2.576533079147339, + "learning_rate": 4.122573277502855e-06, + "loss": 1.5915, + "step": 768 + }, + { + "epoch": 0.7899332306111967, + "grad_norm": 2.503164291381836, + "learning_rate": 4.1206699657403885e-06, + "loss": 1.5893, + "step": 769 + }, + { + "epoch": 0.7909604519774012, + "grad_norm": 2.9463603496551514, + "learning_rate": 4.118766653977922e-06, + "loss": 1.6178, + "step": 770 + }, + { + "epoch": 0.7909604519774012, + "eval_loss": 1.9529738426208496, + "eval_runtime": 45.0922, + "eval_samples_per_second": 22.177, + "eval_steps_per_second": 0.931, + "step": 770 + }, + { + "epoch": 0.7919876733436055, + "grad_norm": 2.47148060798645, + "learning_rate": 4.116863342215455e-06, + "loss": 1.5991, + "step": 771 + }, + { + "epoch": 0.79301489470981, + "grad_norm": 2.515717029571533, + "learning_rate": 4.114960030452989e-06, + "loss": 1.5725, + "step": 772 + }, + { + "epoch": 0.7940421160760144, + "grad_norm": 2.5231993198394775, + "learning_rate": 4.113056718690522e-06, + "loss": 1.6145, + "step": 773 + }, + { + "epoch": 0.7950693374422187, + "grad_norm": 2.479978084564209, + "learning_rate": 4.111153406928055e-06, + "loss": 1.5537, + "step": 774 + }, + { + "epoch": 0.7960965588084232, + "grad_norm": 2.5341250896453857, + "learning_rate": 4.109250095165589e-06, + "loss": 1.6419, + "step": 775 + }, + { + "epoch": 0.7971237801746276, + "grad_norm": 2.6277401447296143, + "learning_rate": 4.107346783403122e-06, + "loss": 1.6505, + "step": 776 + }, + { + "epoch": 0.7981510015408321, + "grad_norm": 2.4688351154327393, + "learning_rate": 4.105443471640655e-06, + "loss": 1.5896, + "step": 777 + }, + { + "epoch": 0.7991782229070364, + "grad_norm": 2.4134581089019775, + "learning_rate": 4.103540159878188e-06, + "loss": 1.5788, + "step": 778 + }, + { + "epoch": 0.8002054442732409, + "grad_norm": 2.5090012550354004, + "learning_rate": 4.101636848115721e-06, + "loss": 1.6162, + "step": 779 + }, + { + "epoch": 0.8012326656394453, + "grad_norm": 2.5123302936553955, + "learning_rate": 4.099733536353255e-06, + "loss": 1.5868, + "step": 780 + }, + { + "epoch": 0.8012326656394453, + "eval_loss": 1.9545625448226929, + "eval_runtime": 45.9316, + "eval_samples_per_second": 21.772, + "eval_steps_per_second": 0.914, + "step": 780 + }, + { + "epoch": 0.8022598870056498, + "grad_norm": 2.787181854248047, + "learning_rate": 4.097830224590788e-06, + "loss": 1.612, + "step": 781 + }, + { + "epoch": 0.8032871083718541, + "grad_norm": 2.414426326751709, + "learning_rate": 4.095926912828322e-06, + "loss": 1.6908, + "step": 782 + }, + { + "epoch": 0.8043143297380585, + "grad_norm": 2.452697992324829, + "learning_rate": 4.094023601065855e-06, + "loss": 1.6575, + "step": 783 + }, + { + "epoch": 0.805341551104263, + "grad_norm": 2.4377169609069824, + "learning_rate": 4.092120289303388e-06, + "loss": 1.6016, + "step": 784 + }, + { + "epoch": 0.8063687724704673, + "grad_norm": 2.369154214859009, + "learning_rate": 4.0902169775409215e-06, + "loss": 1.6133, + "step": 785 + }, + { + "epoch": 0.8073959938366718, + "grad_norm": 2.261869430541992, + "learning_rate": 4.088313665778455e-06, + "loss": 1.5555, + "step": 786 + }, + { + "epoch": 0.8084232152028762, + "grad_norm": 2.7058632373809814, + "learning_rate": 4.086410354015988e-06, + "loss": 1.5957, + "step": 787 + }, + { + "epoch": 0.8094504365690807, + "grad_norm": 2.3907508850097656, + "learning_rate": 4.0845070422535216e-06, + "loss": 1.5676, + "step": 788 + }, + { + "epoch": 0.810477657935285, + "grad_norm": 2.510284900665283, + "learning_rate": 4.082603730491055e-06, + "loss": 1.5872, + "step": 789 + }, + { + "epoch": 0.8115048793014895, + "grad_norm": 2.590810775756836, + "learning_rate": 4.080700418728588e-06, + "loss": 1.6315, + "step": 790 + }, + { + "epoch": 0.8115048793014895, + "eval_loss": 1.9573780298233032, + "eval_runtime": 44.5677, + "eval_samples_per_second": 22.438, + "eval_steps_per_second": 0.942, + "step": 790 + }, + { + "epoch": 0.8125321006676939, + "grad_norm": 2.466134786605835, + "learning_rate": 4.078797106966122e-06, + "loss": 1.6018, + "step": 791 + }, + { + "epoch": 0.8135593220338984, + "grad_norm": 2.474653482437134, + "learning_rate": 4.076893795203655e-06, + "loss": 1.5101, + "step": 792 + }, + { + "epoch": 0.8145865434001027, + "grad_norm": 2.4463517665863037, + "learning_rate": 4.074990483441188e-06, + "loss": 1.5686, + "step": 793 + }, + { + "epoch": 0.8156137647663071, + "grad_norm": 2.5148985385894775, + "learning_rate": 4.073087171678721e-06, + "loss": 1.6544, + "step": 794 + }, + { + "epoch": 0.8166409861325116, + "grad_norm": 2.6183927059173584, + "learning_rate": 4.071183859916254e-06, + "loss": 1.6277, + "step": 795 + }, + { + "epoch": 0.8176682074987159, + "grad_norm": 2.438427686691284, + "learning_rate": 4.069280548153788e-06, + "loss": 1.6002, + "step": 796 + }, + { + "epoch": 0.8186954288649204, + "grad_norm": 2.662855863571167, + "learning_rate": 4.067377236391321e-06, + "loss": 1.5594, + "step": 797 + }, + { + "epoch": 0.8197226502311248, + "grad_norm": 2.64789080619812, + "learning_rate": 4.065473924628854e-06, + "loss": 1.6007, + "step": 798 + }, + { + "epoch": 0.8207498715973293, + "grad_norm": 2.5152997970581055, + "learning_rate": 4.063570612866388e-06, + "loss": 1.6366, + "step": 799 + }, + { + "epoch": 0.8217770929635336, + "grad_norm": 2.468672037124634, + "learning_rate": 4.061667301103921e-06, + "loss": 1.6313, + "step": 800 + }, + { + "epoch": 0.8217770929635336, + "eval_loss": 1.95821213722229, + "eval_runtime": 45.9362, + "eval_samples_per_second": 21.769, + "eval_steps_per_second": 0.914, + "step": 800 + }, + { + "epoch": 0.8228043143297381, + "grad_norm": 2.440307855606079, + "learning_rate": 4.0597639893414544e-06, + "loss": 1.5993, + "step": 801 + }, + { + "epoch": 0.8238315356959425, + "grad_norm": 2.6519880294799805, + "learning_rate": 4.057860677578988e-06, + "loss": 1.6106, + "step": 802 + }, + { + "epoch": 0.8248587570621468, + "grad_norm": 2.4224767684936523, + "learning_rate": 4.055957365816521e-06, + "loss": 1.6247, + "step": 803 + }, + { + "epoch": 0.8258859784283513, + "grad_norm": 2.346858263015747, + "learning_rate": 4.0540540540540545e-06, + "loss": 1.5699, + "step": 804 + }, + { + "epoch": 0.8269131997945557, + "grad_norm": 2.577711582183838, + "learning_rate": 4.052150742291588e-06, + "loss": 1.5665, + "step": 805 + }, + { + "epoch": 0.8279404211607602, + "grad_norm": 2.604501724243164, + "learning_rate": 4.050247430529121e-06, + "loss": 1.561, + "step": 806 + }, + { + "epoch": 0.8289676425269645, + "grad_norm": 2.534820795059204, + "learning_rate": 4.048344118766655e-06, + "loss": 1.6391, + "step": 807 + }, + { + "epoch": 0.829994863893169, + "grad_norm": 2.496765613555908, + "learning_rate": 4.046440807004188e-06, + "loss": 1.6338, + "step": 808 + }, + { + "epoch": 0.8310220852593734, + "grad_norm": 2.587185859680176, + "learning_rate": 4.044537495241721e-06, + "loss": 1.6009, + "step": 809 + }, + { + "epoch": 0.8320493066255779, + "grad_norm": 2.7336575984954834, + "learning_rate": 4.042634183479255e-06, + "loss": 1.5911, + "step": 810 + }, + { + "epoch": 0.8320493066255779, + "eval_loss": 1.951606273651123, + "eval_runtime": 42.4435, + "eval_samples_per_second": 23.561, + "eval_steps_per_second": 0.99, + "step": 810 + }, + { + "epoch": 0.8330765279917822, + "grad_norm": 2.5524845123291016, + "learning_rate": 4.040730871716787e-06, + "loss": 1.6406, + "step": 811 + }, + { + "epoch": 0.8341037493579867, + "grad_norm": 2.4861226081848145, + "learning_rate": 4.038827559954321e-06, + "loss": 1.6092, + "step": 812 + }, + { + "epoch": 0.8351309707241911, + "grad_norm": 2.668187141418457, + "learning_rate": 4.036924248191854e-06, + "loss": 1.6529, + "step": 813 + }, + { + "epoch": 0.8361581920903954, + "grad_norm": 2.5616471767425537, + "learning_rate": 4.035020936429387e-06, + "loss": 1.6253, + "step": 814 + }, + { + "epoch": 0.8371854134565999, + "grad_norm": 2.4261975288391113, + "learning_rate": 4.033117624666921e-06, + "loss": 1.5955, + "step": 815 + }, + { + "epoch": 0.8382126348228043, + "grad_norm": 2.4947080612182617, + "learning_rate": 4.031214312904454e-06, + "loss": 1.5686, + "step": 816 + }, + { + "epoch": 0.8392398561890088, + "grad_norm": 2.747781276702881, + "learning_rate": 4.029311001141987e-06, + "loss": 1.5642, + "step": 817 + }, + { + "epoch": 0.8402670775552131, + "grad_norm": 2.4762001037597656, + "learning_rate": 4.027407689379521e-06, + "loss": 1.5301, + "step": 818 + }, + { + "epoch": 0.8412942989214176, + "grad_norm": 2.6253550052642822, + "learning_rate": 4.025504377617054e-06, + "loss": 1.5885, + "step": 819 + }, + { + "epoch": 0.842321520287622, + "grad_norm": 2.6974875926971436, + "learning_rate": 4.0236010658545875e-06, + "loss": 1.5656, + "step": 820 + }, + { + "epoch": 0.842321520287622, + "eval_loss": 1.9420732259750366, + "eval_runtime": 46.9525, + "eval_samples_per_second": 21.298, + "eval_steps_per_second": 0.895, + "step": 820 + }, + { + "epoch": 0.8433487416538265, + "grad_norm": 2.638587474822998, + "learning_rate": 4.02169775409212e-06, + "loss": 1.6081, + "step": 821 + }, + { + "epoch": 0.8443759630200308, + "grad_norm": 2.963974714279175, + "learning_rate": 4.019794442329653e-06, + "loss": 1.6851, + "step": 822 + }, + { + "epoch": 0.8454031843862352, + "grad_norm": 2.4657533168792725, + "learning_rate": 4.0178911305671876e-06, + "loss": 1.5396, + "step": 823 + }, + { + "epoch": 0.8464304057524397, + "grad_norm": 2.5373876094818115, + "learning_rate": 4.015987818804721e-06, + "loss": 1.6186, + "step": 824 + }, + { + "epoch": 0.847457627118644, + "grad_norm": 2.5873284339904785, + "learning_rate": 4.014084507042254e-06, + "loss": 1.5618, + "step": 825 + }, + { + "epoch": 0.8484848484848485, + "grad_norm": 2.4308598041534424, + "learning_rate": 4.012181195279788e-06, + "loss": 1.5399, + "step": 826 + }, + { + "epoch": 0.8495120698510529, + "grad_norm": 2.5274946689605713, + "learning_rate": 4.010277883517321e-06, + "loss": 1.6375, + "step": 827 + }, + { + "epoch": 0.8505392912172574, + "grad_norm": 2.4370181560516357, + "learning_rate": 4.0083745717548535e-06, + "loss": 1.6313, + "step": 828 + }, + { + "epoch": 0.8515665125834617, + "grad_norm": 2.4524898529052734, + "learning_rate": 4.006471259992387e-06, + "loss": 1.5865, + "step": 829 + }, + { + "epoch": 0.8525937339496662, + "grad_norm": 2.5587961673736572, + "learning_rate": 4.00456794822992e-06, + "loss": 1.608, + "step": 830 + }, + { + "epoch": 0.8525937339496662, + "eval_loss": 1.94539475440979, + "eval_runtime": 49.4543, + "eval_samples_per_second": 20.221, + "eval_steps_per_second": 0.849, + "step": 830 + }, + { + "epoch": 0.8536209553158706, + "grad_norm": 2.512253761291504, + "learning_rate": 4.002664636467454e-06, + "loss": 1.6607, + "step": 831 + }, + { + "epoch": 0.8546481766820749, + "grad_norm": 2.438451051712036, + "learning_rate": 4.000761324704987e-06, + "loss": 1.6153, + "step": 832 + }, + { + "epoch": 0.8556753980482794, + "grad_norm": 2.5550496578216553, + "learning_rate": 3.99885801294252e-06, + "loss": 1.5489, + "step": 833 + }, + { + "epoch": 0.8567026194144838, + "grad_norm": 2.559732437133789, + "learning_rate": 3.996954701180054e-06, + "loss": 1.6045, + "step": 834 + }, + { + "epoch": 0.8577298407806883, + "grad_norm": 2.317692995071411, + "learning_rate": 3.995051389417587e-06, + "loss": 1.5316, + "step": 835 + }, + { + "epoch": 0.8587570621468926, + "grad_norm": 2.3742363452911377, + "learning_rate": 3.9931480776551204e-06, + "loss": 1.553, + "step": 836 + }, + { + "epoch": 0.8597842835130971, + "grad_norm": 2.5645551681518555, + "learning_rate": 3.991244765892654e-06, + "loss": 1.6558, + "step": 837 + }, + { + "epoch": 0.8608115048793015, + "grad_norm": 2.494830846786499, + "learning_rate": 3.989341454130186e-06, + "loss": 1.5713, + "step": 838 + }, + { + "epoch": 0.861838726245506, + "grad_norm": 2.416020393371582, + "learning_rate": 3.98743814236772e-06, + "loss": 1.5383, + "step": 839 + }, + { + "epoch": 0.8628659476117103, + "grad_norm": 2.6063015460968018, + "learning_rate": 3.985534830605253e-06, + "loss": 1.5831, + "step": 840 + }, + { + "epoch": 0.8628659476117103, + "eval_loss": 1.9441509246826172, + "eval_runtime": 46.0083, + "eval_samples_per_second": 21.735, + "eval_steps_per_second": 0.913, + "step": 840 + }, + { + "epoch": 0.8638931689779148, + "grad_norm": 2.5771663188934326, + "learning_rate": 3.983631518842786e-06, + "loss": 1.6077, + "step": 841 + }, + { + "epoch": 0.8649203903441192, + "grad_norm": 2.573119878768921, + "learning_rate": 3.98172820708032e-06, + "loss": 1.6301, + "step": 842 + }, + { + "epoch": 0.8659476117103235, + "grad_norm": 2.41396164894104, + "learning_rate": 3.979824895317854e-06, + "loss": 1.5758, + "step": 843 + }, + { + "epoch": 0.866974833076528, + "grad_norm": 2.532883882522583, + "learning_rate": 3.977921583555387e-06, + "loss": 1.5525, + "step": 844 + }, + { + "epoch": 0.8680020544427324, + "grad_norm": 2.669311285018921, + "learning_rate": 3.97601827179292e-06, + "loss": 1.6183, + "step": 845 + }, + { + "epoch": 0.8690292758089369, + "grad_norm": 2.366861343383789, + "learning_rate": 3.974114960030453e-06, + "loss": 1.5551, + "step": 846 + }, + { + "epoch": 0.8700564971751412, + "grad_norm": 2.3896048069000244, + "learning_rate": 3.9722116482679866e-06, + "loss": 1.5894, + "step": 847 + }, + { + "epoch": 0.8710837185413457, + "grad_norm": 2.6509673595428467, + "learning_rate": 3.97030833650552e-06, + "loss": 1.5813, + "step": 848 + }, + { + "epoch": 0.8721109399075501, + "grad_norm": 2.531726121902466, + "learning_rate": 3.968405024743053e-06, + "loss": 1.5899, + "step": 849 + }, + { + "epoch": 0.8731381612737545, + "grad_norm": 2.4117019176483154, + "learning_rate": 3.966501712980587e-06, + "loss": 1.5117, + "step": 850 + }, + { + "epoch": 0.8731381612737545, + "eval_loss": 1.9224250316619873, + "eval_runtime": 47.7854, + "eval_samples_per_second": 20.927, + "eval_steps_per_second": 0.879, + "step": 850 + }, + { + "epoch": 0.8741653826399589, + "grad_norm": 2.3789749145507812, + "learning_rate": 3.96459840121812e-06, + "loss": 1.5708, + "step": 851 + }, + { + "epoch": 0.8751926040061633, + "grad_norm": 2.386664867401123, + "learning_rate": 3.962695089455653e-06, + "loss": 1.5734, + "step": 852 + }, + { + "epoch": 0.8762198253723678, + "grad_norm": 2.4788997173309326, + "learning_rate": 3.960791777693187e-06, + "loss": 1.6315, + "step": 853 + }, + { + "epoch": 0.8772470467385721, + "grad_norm": 2.630183219909668, + "learning_rate": 3.95888846593072e-06, + "loss": 1.614, + "step": 854 + }, + { + "epoch": 0.8782742681047766, + "grad_norm": 2.3957736492156982, + "learning_rate": 3.956985154168253e-06, + "loss": 1.61, + "step": 855 + }, + { + "epoch": 0.879301489470981, + "grad_norm": 2.652036428451538, + "learning_rate": 3.955081842405786e-06, + "loss": 1.6463, + "step": 856 + }, + { + "epoch": 0.8803287108371854, + "grad_norm": 2.579867124557495, + "learning_rate": 3.953178530643319e-06, + "loss": 1.6624, + "step": 857 + }, + { + "epoch": 0.8813559322033898, + "grad_norm": 2.5118343830108643, + "learning_rate": 3.951275218880853e-06, + "loss": 1.6816, + "step": 858 + }, + { + "epoch": 0.8823831535695943, + "grad_norm": 2.5963029861450195, + "learning_rate": 3.949371907118386e-06, + "loss": 1.5296, + "step": 859 + }, + { + "epoch": 0.8834103749357987, + "grad_norm": 2.571241855621338, + "learning_rate": 3.9474685953559194e-06, + "loss": 1.6362, + "step": 860 + }, + { + "epoch": 0.8834103749357987, + "eval_loss": 1.931492567062378, + "eval_runtime": 52.2417, + "eval_samples_per_second": 19.142, + "eval_steps_per_second": 0.804, + "step": 860 + }, + { + "epoch": 0.884437596302003, + "grad_norm": 2.518036365509033, + "learning_rate": 3.945565283593453e-06, + "loss": 1.5865, + "step": 861 + }, + { + "epoch": 0.8854648176682075, + "grad_norm": 3.14587664604187, + "learning_rate": 3.943661971830986e-06, + "loss": 1.586, + "step": 862 + }, + { + "epoch": 0.8864920390344119, + "grad_norm": 2.686350107192993, + "learning_rate": 3.9417586600685195e-06, + "loss": 1.6385, + "step": 863 + }, + { + "epoch": 0.8875192604006163, + "grad_norm": 3.0324997901916504, + "learning_rate": 3.939855348306053e-06, + "loss": 1.6199, + "step": 864 + }, + { + "epoch": 0.8885464817668207, + "grad_norm": 2.6129438877105713, + "learning_rate": 3.937952036543586e-06, + "loss": 1.6049, + "step": 865 + }, + { + "epoch": 0.8895737031330252, + "grad_norm": 2.795729160308838, + "learning_rate": 3.93604872478112e-06, + "loss": 1.6027, + "step": 866 + }, + { + "epoch": 0.8906009244992296, + "grad_norm": 3.2352302074432373, + "learning_rate": 3.934145413018653e-06, + "loss": 1.6063, + "step": 867 + }, + { + "epoch": 0.891628145865434, + "grad_norm": 2.5050623416900635, + "learning_rate": 3.932242101256186e-06, + "loss": 1.5622, + "step": 868 + }, + { + "epoch": 0.8926553672316384, + "grad_norm": 2.8998777866363525, + "learning_rate": 3.93033878949372e-06, + "loss": 1.5729, + "step": 869 + }, + { + "epoch": 0.8936825885978429, + "grad_norm": 2.60239577293396, + "learning_rate": 3.928435477731253e-06, + "loss": 1.5377, + "step": 870 + }, + { + "epoch": 0.8936825885978429, + "eval_loss": 1.9311563968658447, + "eval_runtime": 46.8098, + "eval_samples_per_second": 21.363, + "eval_steps_per_second": 0.897, + "step": 870 + }, + { + "epoch": 0.8947098099640473, + "grad_norm": 2.5061111450195312, + "learning_rate": 3.9265321659687864e-06, + "loss": 1.6216, + "step": 871 + }, + { + "epoch": 0.8957370313302516, + "grad_norm": 2.7476489543914795, + "learning_rate": 3.924628854206319e-06, + "loss": 1.6357, + "step": 872 + }, + { + "epoch": 0.8967642526964561, + "grad_norm": 2.605536937713623, + "learning_rate": 3.922725542443852e-06, + "loss": 1.5983, + "step": 873 + }, + { + "epoch": 0.8977914740626605, + "grad_norm": 2.6024532318115234, + "learning_rate": 3.920822230681386e-06, + "loss": 1.5571, + "step": 874 + }, + { + "epoch": 0.8988186954288649, + "grad_norm": 2.7357826232910156, + "learning_rate": 3.918918918918919e-06, + "loss": 1.5802, + "step": 875 + }, + { + "epoch": 0.8998459167950693, + "grad_norm": 2.6406939029693604, + "learning_rate": 3.917015607156452e-06, + "loss": 1.6046, + "step": 876 + }, + { + "epoch": 0.9008731381612738, + "grad_norm": 2.4611780643463135, + "learning_rate": 3.915112295393986e-06, + "loss": 1.5803, + "step": 877 + }, + { + "epoch": 0.9019003595274782, + "grad_norm": 3.0608181953430176, + "learning_rate": 3.913208983631519e-06, + "loss": 1.6655, + "step": 878 + }, + { + "epoch": 0.9029275808936826, + "grad_norm": 2.482142448425293, + "learning_rate": 3.9113056718690525e-06, + "loss": 1.6444, + "step": 879 + }, + { + "epoch": 0.903954802259887, + "grad_norm": 2.4936728477478027, + "learning_rate": 3.909402360106586e-06, + "loss": 1.5596, + "step": 880 + }, + { + "epoch": 0.903954802259887, + "eval_loss": 1.933276653289795, + "eval_runtime": 48.0836, + "eval_samples_per_second": 20.797, + "eval_steps_per_second": 0.873, + "step": 880 + }, + { + "epoch": 0.9049820236260914, + "grad_norm": 2.7802248001098633, + "learning_rate": 3.907499048344119e-06, + "loss": 1.6133, + "step": 881 + }, + { + "epoch": 0.9060092449922958, + "grad_norm": 2.486586093902588, + "learning_rate": 3.905595736581652e-06, + "loss": 1.623, + "step": 882 + }, + { + "epoch": 0.9070364663585002, + "grad_norm": 2.477555751800537, + "learning_rate": 3.903692424819186e-06, + "loss": 1.6219, + "step": 883 + }, + { + "epoch": 0.9080636877247047, + "grad_norm": 2.5033347606658936, + "learning_rate": 3.901789113056719e-06, + "loss": 1.5612, + "step": 884 + }, + { + "epoch": 0.9090909090909091, + "grad_norm": 2.410464286804199, + "learning_rate": 3.899885801294253e-06, + "loss": 1.6243, + "step": 885 + }, + { + "epoch": 0.9101181304571135, + "grad_norm": 2.451402425765991, + "learning_rate": 3.897982489531786e-06, + "loss": 1.5961, + "step": 886 + }, + { + "epoch": 0.9111453518233179, + "grad_norm": 2.439730167388916, + "learning_rate": 3.896079177769319e-06, + "loss": 1.6471, + "step": 887 + }, + { + "epoch": 0.9121725731895224, + "grad_norm": 2.527719259262085, + "learning_rate": 3.894175866006853e-06, + "loss": 1.6262, + "step": 888 + }, + { + "epoch": 0.9131997945557268, + "grad_norm": 2.612992763519287, + "learning_rate": 3.892272554244385e-06, + "loss": 1.6533, + "step": 889 + }, + { + "epoch": 0.9142270159219312, + "grad_norm": 2.509989023208618, + "learning_rate": 3.890369242481919e-06, + "loss": 1.5596, + "step": 890 + }, + { + "epoch": 0.9142270159219312, + "eval_loss": 1.9224061965942383, + "eval_runtime": 45.5462, + "eval_samples_per_second": 21.956, + "eval_steps_per_second": 0.922, + "step": 890 + }, + { + "epoch": 0.9152542372881356, + "grad_norm": 2.3849823474884033, + "learning_rate": 3.888465930719452e-06, + "loss": 1.6336, + "step": 891 + }, + { + "epoch": 0.91628145865434, + "grad_norm": 2.4989709854125977, + "learning_rate": 3.886562618956985e-06, + "loss": 1.5733, + "step": 892 + }, + { + "epoch": 0.9173086800205444, + "grad_norm": 2.4675447940826416, + "learning_rate": 3.884659307194519e-06, + "loss": 1.5818, + "step": 893 + }, + { + "epoch": 0.9183359013867488, + "grad_norm": 2.3105645179748535, + "learning_rate": 3.882755995432052e-06, + "loss": 1.5251, + "step": 894 + }, + { + "epoch": 0.9193631227529533, + "grad_norm": 2.5410287380218506, + "learning_rate": 3.8808526836695854e-06, + "loss": 1.6089, + "step": 895 + }, + { + "epoch": 0.9203903441191577, + "grad_norm": 2.617091655731201, + "learning_rate": 3.878949371907119e-06, + "loss": 1.5631, + "step": 896 + }, + { + "epoch": 0.9214175654853621, + "grad_norm": 2.4979820251464844, + "learning_rate": 3.877046060144652e-06, + "loss": 1.5577, + "step": 897 + }, + { + "epoch": 0.9224447868515665, + "grad_norm": 2.4857399463653564, + "learning_rate": 3.8751427483821855e-06, + "loss": 1.56, + "step": 898 + }, + { + "epoch": 0.923472008217771, + "grad_norm": 2.456709146499634, + "learning_rate": 3.873239436619718e-06, + "loss": 1.594, + "step": 899 + }, + { + "epoch": 0.9244992295839753, + "grad_norm": 2.355168342590332, + "learning_rate": 3.871336124857251e-06, + "loss": 1.5371, + "step": 900 + }, + { + "epoch": 0.9244992295839753, + "eval_loss": 1.9357926845550537, + "eval_runtime": 44.4044, + "eval_samples_per_second": 22.52, + "eval_steps_per_second": 0.946, + "step": 900 + }, + { + "epoch": 0.9255264509501797, + "grad_norm": 2.5965569019317627, + "learning_rate": 3.869432813094785e-06, + "loss": 1.5966, + "step": 901 + }, + { + "epoch": 0.9265536723163842, + "grad_norm": 2.443662405014038, + "learning_rate": 3.867529501332318e-06, + "loss": 1.6053, + "step": 902 + }, + { + "epoch": 0.9275808936825886, + "grad_norm": 2.3888511657714844, + "learning_rate": 3.8656261895698515e-06, + "loss": 1.5921, + "step": 903 + }, + { + "epoch": 0.928608115048793, + "grad_norm": 2.4648637771606445, + "learning_rate": 3.863722877807386e-06, + "loss": 1.5879, + "step": 904 + }, + { + "epoch": 0.9296353364149974, + "grad_norm": 2.4636101722717285, + "learning_rate": 3.861819566044919e-06, + "loss": 1.6286, + "step": 905 + }, + { + "epoch": 0.9306625577812019, + "grad_norm": 2.722627878189087, + "learning_rate": 3.859916254282452e-06, + "loss": 1.6666, + "step": 906 + }, + { + "epoch": 0.9316897791474062, + "grad_norm": 2.550255298614502, + "learning_rate": 3.858012942519985e-06, + "loss": 1.5711, + "step": 907 + }, + { + "epoch": 0.9327170005136107, + "grad_norm": 2.510209798812866, + "learning_rate": 3.856109630757518e-06, + "loss": 1.5987, + "step": 908 + }, + { + "epoch": 0.9337442218798151, + "grad_norm": 2.6028876304626465, + "learning_rate": 3.854206318995052e-06, + "loss": 1.5682, + "step": 909 + }, + { + "epoch": 0.9347714432460195, + "grad_norm": 2.487908124923706, + "learning_rate": 3.852303007232585e-06, + "loss": 1.608, + "step": 910 + }, + { + "epoch": 0.9347714432460195, + "eval_loss": 1.93544602394104, + "eval_runtime": 41.5201, + "eval_samples_per_second": 24.085, + "eval_steps_per_second": 1.012, + "step": 910 + }, + { + "epoch": 0.9357986646122239, + "grad_norm": 2.4569835662841797, + "learning_rate": 3.850399695470118e-06, + "loss": 1.5492, + "step": 911 + }, + { + "epoch": 0.9368258859784283, + "grad_norm": 2.572781562805176, + "learning_rate": 3.848496383707652e-06, + "loss": 1.5633, + "step": 912 + }, + { + "epoch": 0.9378531073446328, + "grad_norm": 2.6213486194610596, + "learning_rate": 3.846593071945185e-06, + "loss": 1.5488, + "step": 913 + }, + { + "epoch": 0.9388803287108372, + "grad_norm": 2.6079821586608887, + "learning_rate": 3.8446897601827185e-06, + "loss": 1.5311, + "step": 914 + }, + { + "epoch": 0.9399075500770416, + "grad_norm": 2.6774539947509766, + "learning_rate": 3.842786448420252e-06, + "loss": 1.6564, + "step": 915 + }, + { + "epoch": 0.940934771443246, + "grad_norm": 2.5028746128082275, + "learning_rate": 3.840883136657785e-06, + "loss": 1.6551, + "step": 916 + }, + { + "epoch": 0.9419619928094505, + "grad_norm": 2.5254387855529785, + "learning_rate": 3.838979824895318e-06, + "loss": 1.6492, + "step": 917 + }, + { + "epoch": 0.9429892141756548, + "grad_norm": 2.5451207160949707, + "learning_rate": 3.837076513132851e-06, + "loss": 1.5306, + "step": 918 + }, + { + "epoch": 0.9440164355418593, + "grad_norm": 2.526526689529419, + "learning_rate": 3.8351732013703844e-06, + "loss": 1.6131, + "step": 919 + }, + { + "epoch": 0.9450436569080637, + "grad_norm": 2.454895257949829, + "learning_rate": 3.833269889607918e-06, + "loss": 1.5592, + "step": 920 + }, + { + "epoch": 0.9450436569080637, + "eval_loss": 1.9236185550689697, + "eval_runtime": 44.149, + "eval_samples_per_second": 22.651, + "eval_steps_per_second": 0.951, + "step": 920 + }, + { + "epoch": 0.9460708782742681, + "grad_norm": 2.6773581504821777, + "learning_rate": 3.831366577845451e-06, + "loss": 1.5498, + "step": 921 + }, + { + "epoch": 0.9470980996404725, + "grad_norm": 2.63554310798645, + "learning_rate": 3.8294632660829845e-06, + "loss": 1.5413, + "step": 922 + }, + { + "epoch": 0.9481253210066769, + "grad_norm": 2.7118968963623047, + "learning_rate": 3.827559954320518e-06, + "loss": 1.572, + "step": 923 + }, + { + "epoch": 0.9491525423728814, + "grad_norm": 2.4681396484375, + "learning_rate": 3.825656642558051e-06, + "loss": 1.6003, + "step": 924 + }, + { + "epoch": 0.9501797637390857, + "grad_norm": 2.387608051300049, + "learning_rate": 3.823753330795585e-06, + "loss": 1.6057, + "step": 925 + }, + { + "epoch": 0.9512069851052902, + "grad_norm": 2.426020383834839, + "learning_rate": 3.821850019033118e-06, + "loss": 1.5976, + "step": 926 + }, + { + "epoch": 0.9522342064714946, + "grad_norm": 2.446044445037842, + "learning_rate": 3.819946707270651e-06, + "loss": 1.5363, + "step": 927 + }, + { + "epoch": 0.9532614278376991, + "grad_norm": 2.4415299892425537, + "learning_rate": 3.818043395508185e-06, + "loss": 1.5685, + "step": 928 + }, + { + "epoch": 0.9542886492039034, + "grad_norm": 2.4006199836730957, + "learning_rate": 3.816140083745718e-06, + "loss": 1.5823, + "step": 929 + }, + { + "epoch": 0.9553158705701078, + "grad_norm": 2.6479294300079346, + "learning_rate": 3.8142367719832514e-06, + "loss": 1.6658, + "step": 930 + }, + { + "epoch": 0.9553158705701078, + "eval_loss": 1.9434653520584106, + "eval_runtime": 45.5063, + "eval_samples_per_second": 21.975, + "eval_steps_per_second": 0.923, + "step": 930 + }, + { + "epoch": 0.9563430919363123, + "grad_norm": 2.530367851257324, + "learning_rate": 3.8123334602207844e-06, + "loss": 1.544, + "step": 931 + }, + { + "epoch": 0.9573703133025167, + "grad_norm": 2.641665458679199, + "learning_rate": 3.8104301484583177e-06, + "loss": 1.5658, + "step": 932 + }, + { + "epoch": 0.9583975346687211, + "grad_norm": 2.3992276191711426, + "learning_rate": 3.808526836695851e-06, + "loss": 1.5524, + "step": 933 + }, + { + "epoch": 0.9594247560349255, + "grad_norm": 2.640503168106079, + "learning_rate": 3.8066235249333845e-06, + "loss": 1.6127, + "step": 934 + }, + { + "epoch": 0.96045197740113, + "grad_norm": 2.513735771179199, + "learning_rate": 3.804720213170918e-06, + "loss": 1.5901, + "step": 935 + }, + { + "epoch": 0.9614791987673343, + "grad_norm": 2.74481463432312, + "learning_rate": 3.8028169014084508e-06, + "loss": 1.5792, + "step": 936 + }, + { + "epoch": 0.9625064201335388, + "grad_norm": 2.5238802433013916, + "learning_rate": 3.800913589645984e-06, + "loss": 1.602, + "step": 937 + }, + { + "epoch": 0.9635336414997432, + "grad_norm": 2.5111966133117676, + "learning_rate": 3.7990102778835175e-06, + "loss": 1.607, + "step": 938 + }, + { + "epoch": 0.9645608628659477, + "grad_norm": 2.658449649810791, + "learning_rate": 3.797106966121051e-06, + "loss": 1.593, + "step": 939 + }, + { + "epoch": 0.965588084232152, + "grad_norm": 2.6884148120880127, + "learning_rate": 3.7952036543585842e-06, + "loss": 1.6576, + "step": 940 + }, + { + "epoch": 0.965588084232152, + "eval_loss": 1.9191738367080688, + "eval_runtime": 45.1928, + "eval_samples_per_second": 22.127, + "eval_steps_per_second": 0.929, + "step": 940 + }, + { + "epoch": 0.9666153055983564, + "grad_norm": 2.4238054752349854, + "learning_rate": 3.793300342596117e-06, + "loss": 1.6079, + "step": 941 + }, + { + "epoch": 0.9676425269645609, + "grad_norm": 2.505571126937866, + "learning_rate": 3.7913970308336505e-06, + "loss": 1.6299, + "step": 942 + }, + { + "epoch": 0.9686697483307652, + "grad_norm": 2.4143226146698, + "learning_rate": 3.789493719071184e-06, + "loss": 1.6158, + "step": 943 + }, + { + "epoch": 0.9696969696969697, + "grad_norm": 2.4664456844329834, + "learning_rate": 3.7875904073087177e-06, + "loss": 1.5672, + "step": 944 + }, + { + "epoch": 0.9707241910631741, + "grad_norm": 2.5686466693878174, + "learning_rate": 3.785687095546251e-06, + "loss": 1.5338, + "step": 945 + }, + { + "epoch": 0.9717514124293786, + "grad_norm": 2.4666805267333984, + "learning_rate": 3.7837837837837844e-06, + "loss": 1.5773, + "step": 946 + }, + { + "epoch": 0.9727786337955829, + "grad_norm": 2.6416635513305664, + "learning_rate": 3.7818804720213178e-06, + "loss": 1.5469, + "step": 947 + }, + { + "epoch": 0.9738058551617874, + "grad_norm": 2.721205711364746, + "learning_rate": 3.7799771602588507e-06, + "loss": 1.5278, + "step": 948 + }, + { + "epoch": 0.9748330765279918, + "grad_norm": 2.4834787845611572, + "learning_rate": 3.778073848496384e-06, + "loss": 1.52, + "step": 949 + }, + { + "epoch": 0.9758602978941961, + "grad_norm": 2.42047381401062, + "learning_rate": 3.7761705367339174e-06, + "loss": 1.5814, + "step": 950 + }, + { + "epoch": 0.9758602978941961, + "eval_loss": 1.9303041696548462, + "eval_runtime": 47.3843, + "eval_samples_per_second": 21.104, + "eval_steps_per_second": 0.886, + "step": 950 + }, + { + "epoch": 0.9768875192604006, + "grad_norm": 2.70369815826416, + "learning_rate": 3.7742672249714508e-06, + "loss": 1.6143, + "step": 951 + }, + { + "epoch": 0.977914740626605, + "grad_norm": 2.393588066101074, + "learning_rate": 3.772363913208984e-06, + "loss": 1.6023, + "step": 952 + }, + { + "epoch": 0.9789419619928095, + "grad_norm": 2.361304998397827, + "learning_rate": 3.770460601446517e-06, + "loss": 1.5679, + "step": 953 + }, + { + "epoch": 0.9799691833590138, + "grad_norm": 2.4044971466064453, + "learning_rate": 3.7685572896840504e-06, + "loss": 1.6007, + "step": 954 + }, + { + "epoch": 0.9809964047252183, + "grad_norm": 2.498539447784424, + "learning_rate": 3.766653977921584e-06, + "loss": 1.5429, + "step": 955 + }, + { + "epoch": 0.9820236260914227, + "grad_norm": 2.457162857055664, + "learning_rate": 3.764750666159117e-06, + "loss": 1.4861, + "step": 956 + }, + { + "epoch": 0.9830508474576272, + "grad_norm": 2.52756404876709, + "learning_rate": 3.7628473543966505e-06, + "loss": 1.5848, + "step": 957 + }, + { + "epoch": 0.9840780688238315, + "grad_norm": 2.503998279571533, + "learning_rate": 3.7609440426341835e-06, + "loss": 1.5435, + "step": 958 + }, + { + "epoch": 0.9851052901900359, + "grad_norm": 2.461087226867676, + "learning_rate": 3.759040730871717e-06, + "loss": 1.6344, + "step": 959 + }, + { + "epoch": 0.9861325115562404, + "grad_norm": 2.4149672985076904, + "learning_rate": 3.75713741910925e-06, + "loss": 1.5273, + "step": 960 + }, + { + "epoch": 0.9861325115562404, + "eval_loss": 1.928680658340454, + "eval_runtime": 48.7892, + "eval_samples_per_second": 20.496, + "eval_steps_per_second": 0.861, + "step": 960 + }, + { + "epoch": 0.9871597329224447, + "grad_norm": 2.5741734504699707, + "learning_rate": 3.7552341073467836e-06, + "loss": 1.5244, + "step": 961 + }, + { + "epoch": 0.9881869542886492, + "grad_norm": 2.5517220497131348, + "learning_rate": 3.753330795584317e-06, + "loss": 1.5695, + "step": 962 + }, + { + "epoch": 0.9892141756548536, + "grad_norm": 2.5910379886627197, + "learning_rate": 3.75142748382185e-06, + "loss": 1.6322, + "step": 963 + }, + { + "epoch": 0.9902413970210581, + "grad_norm": 2.4780521392822266, + "learning_rate": 3.749524172059384e-06, + "loss": 1.5292, + "step": 964 + }, + { + "epoch": 0.9912686183872624, + "grad_norm": 2.6141762733459473, + "learning_rate": 3.747620860296917e-06, + "loss": 1.5809, + "step": 965 + }, + { + "epoch": 0.9922958397534669, + "grad_norm": 2.467606544494629, + "learning_rate": 3.7457175485344504e-06, + "loss": 1.6123, + "step": 966 + }, + { + "epoch": 0.9933230611196713, + "grad_norm": 2.6112184524536133, + "learning_rate": 3.7438142367719837e-06, + "loss": 1.6226, + "step": 967 + }, + { + "epoch": 0.9943502824858758, + "grad_norm": 2.352513313293457, + "learning_rate": 3.741910925009517e-06, + "loss": 1.5928, + "step": 968 + }, + { + "epoch": 0.9953775038520801, + "grad_norm": 2.4357361793518066, + "learning_rate": 3.7400076132470505e-06, + "loss": 1.5484, + "step": 969 + }, + { + "epoch": 0.9964047252182845, + "grad_norm": 2.5168254375457764, + "learning_rate": 3.7381043014845834e-06, + "loss": 1.5722, + "step": 970 + }, + { + "epoch": 0.9964047252182845, + "eval_loss": 1.929540753364563, + "eval_runtime": 49.7232, + "eval_samples_per_second": 20.111, + "eval_steps_per_second": 0.845, + "step": 970 + }, + { + "epoch": 0.997431946584489, + "grad_norm": 2.475173234939575, + "learning_rate": 3.7362009897221168e-06, + "loss": 1.5952, + "step": 971 + }, + { + "epoch": 0.9984591679506933, + "grad_norm": 2.5278515815734863, + "learning_rate": 3.73429767795965e-06, + "loss": 1.5769, + "step": 972 + }, + { + "epoch": 0.9994863893168978, + "grad_norm": 2.4988675117492676, + "learning_rate": 3.7323943661971835e-06, + "loss": 1.5912, + "step": 973 + }, + { + "epoch": 1.0005136106831023, + "grad_norm": 2.406353712081909, + "learning_rate": 3.730491054434717e-06, + "loss": 1.5022, + "step": 974 + }, + { + "epoch": 1.0015408320493067, + "grad_norm": 2.611318349838257, + "learning_rate": 3.72858774267225e-06, + "loss": 1.4633, + "step": 975 + }, + { + "epoch": 1.002568053415511, + "grad_norm": 2.504613161087036, + "learning_rate": 3.726684430909783e-06, + "loss": 1.5405, + "step": 976 + }, + { + "epoch": 1.0035952747817154, + "grad_norm": 2.6135828495025635, + "learning_rate": 3.7247811191473165e-06, + "loss": 1.5756, + "step": 977 + }, + { + "epoch": 1.00462249614792, + "grad_norm": 2.692592144012451, + "learning_rate": 3.72287780738485e-06, + "loss": 1.5777, + "step": 978 + }, + { + "epoch": 1.0056497175141244, + "grad_norm": 2.660323143005371, + "learning_rate": 3.7209744956223832e-06, + "loss": 1.5488, + "step": 979 + }, + { + "epoch": 1.0066769388803287, + "grad_norm": 2.647883415222168, + "learning_rate": 3.719071183859916e-06, + "loss": 1.5619, + "step": 980 + }, + { + "epoch": 1.0066769388803287, + "eval_loss": 1.9224871397018433, + "eval_runtime": 52.6149, + "eval_samples_per_second": 19.006, + "eval_steps_per_second": 0.798, + "step": 980 + }, + { + "epoch": 1.007704160246533, + "grad_norm": 2.6186633110046387, + "learning_rate": 3.7171678720974495e-06, + "loss": 1.492, + "step": 981 + }, + { + "epoch": 1.0087313816127375, + "grad_norm": 2.7562549114227295, + "learning_rate": 3.715264560334983e-06, + "loss": 1.5932, + "step": 982 + }, + { + "epoch": 1.009758602978942, + "grad_norm": 2.618856430053711, + "learning_rate": 3.7133612485725163e-06, + "loss": 1.5514, + "step": 983 + }, + { + "epoch": 1.0107858243451464, + "grad_norm": 2.522064685821533, + "learning_rate": 3.71145793681005e-06, + "loss": 1.5699, + "step": 984 + }, + { + "epoch": 1.0118130457113508, + "grad_norm": 2.627108335494995, + "learning_rate": 3.7095546250475834e-06, + "loss": 1.5033, + "step": 985 + }, + { + "epoch": 1.0128402670775551, + "grad_norm": 2.6393308639526367, + "learning_rate": 3.7076513132851168e-06, + "loss": 1.6064, + "step": 986 + }, + { + "epoch": 1.0138674884437597, + "grad_norm": 2.560675621032715, + "learning_rate": 3.7057480015226497e-06, + "loss": 1.5167, + "step": 987 + }, + { + "epoch": 1.014894709809964, + "grad_norm": 2.4677579402923584, + "learning_rate": 3.703844689760183e-06, + "loss": 1.4953, + "step": 988 + }, + { + "epoch": 1.0159219311761685, + "grad_norm": 2.9904487133026123, + "learning_rate": 3.7019413779977164e-06, + "loss": 1.5786, + "step": 989 + }, + { + "epoch": 1.0169491525423728, + "grad_norm": 2.7342119216918945, + "learning_rate": 3.70003806623525e-06, + "loss": 1.5459, + "step": 990 + }, + { + "epoch": 1.0169491525423728, + "eval_loss": 1.929221510887146, + "eval_runtime": 45.9704, + "eval_samples_per_second": 21.753, + "eval_steps_per_second": 0.914, + "step": 990 + }, + { + "epoch": 1.0179763739085772, + "grad_norm": 2.723241090774536, + "learning_rate": 3.698134754472783e-06, + "loss": 1.5317, + "step": 991 + }, + { + "epoch": 1.0190035952747818, + "grad_norm": 2.562356948852539, + "learning_rate": 3.696231442710316e-06, + "loss": 1.5957, + "step": 992 + }, + { + "epoch": 1.0200308166409862, + "grad_norm": 2.635737419128418, + "learning_rate": 3.6943281309478495e-06, + "loss": 1.5404, + "step": 993 + }, + { + "epoch": 1.0210580380071905, + "grad_norm": 2.588714361190796, + "learning_rate": 3.692424819185383e-06, + "loss": 1.5618, + "step": 994 + }, + { + "epoch": 1.022085259373395, + "grad_norm": 2.61643385887146, + "learning_rate": 3.690521507422916e-06, + "loss": 1.5625, + "step": 995 + }, + { + "epoch": 1.0231124807395995, + "grad_norm": 2.566077470779419, + "learning_rate": 3.6886181956604496e-06, + "loss": 1.5051, + "step": 996 + }, + { + "epoch": 1.0241397021058039, + "grad_norm": 2.5667169094085693, + "learning_rate": 3.686714883897983e-06, + "loss": 1.5142, + "step": 997 + }, + { + "epoch": 1.0251669234720082, + "grad_norm": 2.839078903198242, + "learning_rate": 3.684811572135516e-06, + "loss": 1.5833, + "step": 998 + }, + { + "epoch": 1.0261941448382126, + "grad_norm": 2.651643753051758, + "learning_rate": 3.6829082603730492e-06, + "loss": 1.5735, + "step": 999 + }, + { + "epoch": 1.027221366204417, + "grad_norm": 2.6310813426971436, + "learning_rate": 3.6810049486105826e-06, + "loss": 1.5412, + "step": 1000 + }, + { + "epoch": 1.027221366204417, + "eval_loss": 1.9219006299972534, + "eval_runtime": 53.0111, + "eval_samples_per_second": 18.864, + "eval_steps_per_second": 0.792, + "step": 1000 + }, + { + "epoch": 1.0282485875706215, + "grad_norm": 2.721156120300293, + "learning_rate": 3.679101636848116e-06, + "loss": 1.5686, + "step": 1001 + }, + { + "epoch": 1.029275808936826, + "grad_norm": 2.8999130725860596, + "learning_rate": 3.6771983250856493e-06, + "loss": 1.5699, + "step": 1002 + }, + { + "epoch": 1.0303030303030303, + "grad_norm": 2.8879904747009277, + "learning_rate": 3.6752950133231822e-06, + "loss": 1.5353, + "step": 1003 + }, + { + "epoch": 1.0313302516692346, + "grad_norm": 2.4753830432891846, + "learning_rate": 3.6733917015607156e-06, + "loss": 1.5684, + "step": 1004 + }, + { + "epoch": 1.0323574730354392, + "grad_norm": 2.6306774616241455, + "learning_rate": 3.6714883897982494e-06, + "loss": 1.5741, + "step": 1005 + }, + { + "epoch": 1.0333846944016436, + "grad_norm": 2.8131468296051025, + "learning_rate": 3.6695850780357828e-06, + "loss": 1.5127, + "step": 1006 + }, + { + "epoch": 1.034411915767848, + "grad_norm": 2.6741905212402344, + "learning_rate": 3.667681766273316e-06, + "loss": 1.5135, + "step": 1007 + }, + { + "epoch": 1.0354391371340523, + "grad_norm": 2.6110734939575195, + "learning_rate": 3.6657784545108495e-06, + "loss": 1.5287, + "step": 1008 + }, + { + "epoch": 1.0364663585002567, + "grad_norm": 2.6415274143218994, + "learning_rate": 3.6638751427483824e-06, + "loss": 1.5978, + "step": 1009 + }, + { + "epoch": 1.0374935798664613, + "grad_norm": 2.5623817443847656, + "learning_rate": 3.6619718309859158e-06, + "loss": 1.5586, + "step": 1010 + }, + { + "epoch": 1.0374935798664613, + "eval_loss": 1.915805459022522, + "eval_runtime": 43.3255, + "eval_samples_per_second": 23.081, + "eval_steps_per_second": 0.969, + "step": 1010 + }, + { + "epoch": 1.0385208012326657, + "grad_norm": 2.723954916000366, + "learning_rate": 3.660068519223449e-06, + "loss": 1.6013, + "step": 1011 + }, + { + "epoch": 1.03954802259887, + "grad_norm": 2.6564769744873047, + "learning_rate": 3.6581652074609825e-06, + "loss": 1.5653, + "step": 1012 + }, + { + "epoch": 1.0405752439650744, + "grad_norm": 2.6694817543029785, + "learning_rate": 3.656261895698516e-06, + "loss": 1.5297, + "step": 1013 + }, + { + "epoch": 1.041602465331279, + "grad_norm": 2.546647071838379, + "learning_rate": 3.6543585839360492e-06, + "loss": 1.5192, + "step": 1014 + }, + { + "epoch": 1.0426296866974833, + "grad_norm": 2.691000461578369, + "learning_rate": 3.652455272173582e-06, + "loss": 1.5284, + "step": 1015 + }, + { + "epoch": 1.0436569080636877, + "grad_norm": 2.627333641052246, + "learning_rate": 3.6505519604111155e-06, + "loss": 1.5239, + "step": 1016 + }, + { + "epoch": 1.044684129429892, + "grad_norm": 2.6125097274780273, + "learning_rate": 3.648648648648649e-06, + "loss": 1.5357, + "step": 1017 + }, + { + "epoch": 1.0457113507960965, + "grad_norm": 2.6235790252685547, + "learning_rate": 3.6467453368861823e-06, + "loss": 1.5831, + "step": 1018 + }, + { + "epoch": 1.046738572162301, + "grad_norm": 2.5727598667144775, + "learning_rate": 3.6448420251237156e-06, + "loss": 1.5264, + "step": 1019 + }, + { + "epoch": 1.0477657935285054, + "grad_norm": 2.71970534324646, + "learning_rate": 3.6429387133612486e-06, + "loss": 1.5593, + "step": 1020 + }, + { + "epoch": 1.0477657935285054, + "eval_loss": 1.9139868021011353, + "eval_runtime": 46.8291, + "eval_samples_per_second": 21.354, + "eval_steps_per_second": 0.897, + "step": 1020 + }, + { + "epoch": 1.0487930148947098, + "grad_norm": 2.6479101181030273, + "learning_rate": 3.641035401598782e-06, + "loss": 1.5875, + "step": 1021 + }, + { + "epoch": 1.0498202362609141, + "grad_norm": 2.6313018798828125, + "learning_rate": 3.6391320898363153e-06, + "loss": 1.5222, + "step": 1022 + }, + { + "epoch": 1.0508474576271187, + "grad_norm": 2.679009437561035, + "learning_rate": 3.6372287780738487e-06, + "loss": 1.5678, + "step": 1023 + }, + { + "epoch": 1.051874678993323, + "grad_norm": 2.8511667251586914, + "learning_rate": 3.635325466311382e-06, + "loss": 1.5401, + "step": 1024 + }, + { + "epoch": 1.0529019003595275, + "grad_norm": 2.717280864715576, + "learning_rate": 3.633422154548916e-06, + "loss": 1.5686, + "step": 1025 + }, + { + "epoch": 1.0539291217257318, + "grad_norm": 2.6042356491088867, + "learning_rate": 3.6315188427864487e-06, + "loss": 1.5945, + "step": 1026 + }, + { + "epoch": 1.0549563430919364, + "grad_norm": 2.601451873779297, + "learning_rate": 3.629615531023982e-06, + "loss": 1.556, + "step": 1027 + }, + { + "epoch": 1.0559835644581408, + "grad_norm": 2.5195374488830566, + "learning_rate": 3.6277122192615155e-06, + "loss": 1.5026, + "step": 1028 + }, + { + "epoch": 1.0570107858243452, + "grad_norm": 2.604405164718628, + "learning_rate": 3.625808907499049e-06, + "loss": 1.5629, + "step": 1029 + }, + { + "epoch": 1.0580380071905495, + "grad_norm": 2.6300103664398193, + "learning_rate": 3.623905595736582e-06, + "loss": 1.5253, + "step": 1030 + }, + { + "epoch": 1.0580380071905495, + "eval_loss": 1.9272329807281494, + "eval_runtime": 46.1105, + "eval_samples_per_second": 21.687, + "eval_steps_per_second": 0.911, + "step": 1030 + }, + { + "epoch": 1.059065228556754, + "grad_norm": 2.663590908050537, + "learning_rate": 3.6220022839741156e-06, + "loss": 1.5223, + "step": 1031 + }, + { + "epoch": 1.0600924499229585, + "grad_norm": 2.773226499557495, + "learning_rate": 3.6200989722116485e-06, + "loss": 1.5052, + "step": 1032 + }, + { + "epoch": 1.0611196712891628, + "grad_norm": 2.746915817260742, + "learning_rate": 3.618195660449182e-06, + "loss": 1.5371, + "step": 1033 + }, + { + "epoch": 1.0621468926553672, + "grad_norm": 3.0018815994262695, + "learning_rate": 3.6162923486867152e-06, + "loss": 1.5494, + "step": 1034 + }, + { + "epoch": 1.0631741140215716, + "grad_norm": 2.77616810798645, + "learning_rate": 3.6143890369242486e-06, + "loss": 1.4878, + "step": 1035 + }, + { + "epoch": 1.064201335387776, + "grad_norm": 2.638270854949951, + "learning_rate": 3.612485725161782e-06, + "loss": 1.5028, + "step": 1036 + }, + { + "epoch": 1.0652285567539805, + "grad_norm": 2.7400267124176025, + "learning_rate": 3.610582413399315e-06, + "loss": 1.5364, + "step": 1037 + }, + { + "epoch": 1.066255778120185, + "grad_norm": 2.603987455368042, + "learning_rate": 3.6086791016368482e-06, + "loss": 1.5259, + "step": 1038 + }, + { + "epoch": 1.0672829994863893, + "grad_norm": 2.5326781272888184, + "learning_rate": 3.6067757898743816e-06, + "loss": 1.5346, + "step": 1039 + }, + { + "epoch": 1.0683102208525936, + "grad_norm": 2.677980422973633, + "learning_rate": 3.604872478111915e-06, + "loss": 1.6011, + "step": 1040 + }, + { + "epoch": 1.0683102208525936, + "eval_loss": 1.9431331157684326, + "eval_runtime": 46.6449, + "eval_samples_per_second": 21.439, + "eval_steps_per_second": 0.9, + "step": 1040 + }, + { + "epoch": 1.0693374422187982, + "grad_norm": 2.7540171146392822, + "learning_rate": 3.6029691663494483e-06, + "loss": 1.5988, + "step": 1041 + }, + { + "epoch": 1.0703646635850026, + "grad_norm": 2.6210403442382812, + "learning_rate": 3.6010658545869813e-06, + "loss": 1.5373, + "step": 1042 + }, + { + "epoch": 1.071391884951207, + "grad_norm": 2.7578794956207275, + "learning_rate": 3.5991625428245146e-06, + "loss": 1.5707, + "step": 1043 + }, + { + "epoch": 1.0724191063174113, + "grad_norm": 2.787123203277588, + "learning_rate": 3.597259231062048e-06, + "loss": 1.5666, + "step": 1044 + }, + { + "epoch": 1.073446327683616, + "grad_norm": 2.5755369663238525, + "learning_rate": 3.5953559192995818e-06, + "loss": 1.5005, + "step": 1045 + }, + { + "epoch": 1.0744735490498203, + "grad_norm": 2.644008159637451, + "learning_rate": 3.593452607537115e-06, + "loss": 1.5408, + "step": 1046 + }, + { + "epoch": 1.0755007704160247, + "grad_norm": 2.5818886756896973, + "learning_rate": 3.5915492957746485e-06, + "loss": 1.6064, + "step": 1047 + }, + { + "epoch": 1.076527991782229, + "grad_norm": 2.6413795948028564, + "learning_rate": 3.589645984012182e-06, + "loss": 1.5481, + "step": 1048 + }, + { + "epoch": 1.0775552131484334, + "grad_norm": 2.8570823669433594, + "learning_rate": 3.587742672249715e-06, + "loss": 1.5713, + "step": 1049 + }, + { + "epoch": 1.078582434514638, + "grad_norm": 2.6338746547698975, + "learning_rate": 3.585839360487248e-06, + "loss": 1.5252, + "step": 1050 + }, + { + "epoch": 1.078582434514638, + "eval_loss": 1.9467660188674927, + "eval_runtime": 45.8203, + "eval_samples_per_second": 21.824, + "eval_steps_per_second": 0.917, + "step": 1050 + }, + { + "epoch": 1.0796096558808423, + "grad_norm": 2.7295382022857666, + "learning_rate": 3.5839360487247815e-06, + "loss": 1.515, + "step": 1051 + }, + { + "epoch": 1.0806368772470467, + "grad_norm": 2.7476232051849365, + "learning_rate": 3.582032736962315e-06, + "loss": 1.5301, + "step": 1052 + }, + { + "epoch": 1.081664098613251, + "grad_norm": 2.8701605796813965, + "learning_rate": 3.5801294251998483e-06, + "loss": 1.5971, + "step": 1053 + }, + { + "epoch": 1.0826913199794557, + "grad_norm": 2.638195276260376, + "learning_rate": 3.578226113437381e-06, + "loss": 1.4917, + "step": 1054 + }, + { + "epoch": 1.08371854134566, + "grad_norm": 2.587324857711792, + "learning_rate": 3.5763228016749146e-06, + "loss": 1.5349, + "step": 1055 + }, + { + "epoch": 1.0847457627118644, + "grad_norm": 2.645148754119873, + "learning_rate": 3.574419489912448e-06, + "loss": 1.6063, + "step": 1056 + }, + { + "epoch": 1.0857729840780688, + "grad_norm": 2.580125331878662, + "learning_rate": 3.5725161781499813e-06, + "loss": 1.5504, + "step": 1057 + }, + { + "epoch": 1.0868002054442731, + "grad_norm": 2.730672836303711, + "learning_rate": 3.5706128663875147e-06, + "loss": 1.5272, + "step": 1058 + }, + { + "epoch": 1.0878274268104777, + "grad_norm": 2.564277172088623, + "learning_rate": 3.5687095546250476e-06, + "loss": 1.5471, + "step": 1059 + }, + { + "epoch": 1.088854648176682, + "grad_norm": 2.603226661682129, + "learning_rate": 3.566806242862581e-06, + "loss": 1.5759, + "step": 1060 + }, + { + "epoch": 1.088854648176682, + "eval_loss": 1.9485995769500732, + "eval_runtime": 45.476, + "eval_samples_per_second": 21.99, + "eval_steps_per_second": 0.924, + "step": 1060 + }, + { + "epoch": 1.0898818695428865, + "grad_norm": 2.7261931896209717, + "learning_rate": 3.5649029311001143e-06, + "loss": 1.5455, + "step": 1061 + }, + { + "epoch": 1.0909090909090908, + "grad_norm": 2.5638797283172607, + "learning_rate": 3.5629996193376477e-06, + "loss": 1.5236, + "step": 1062 + }, + { + "epoch": 1.0919363122752954, + "grad_norm": 2.8362395763397217, + "learning_rate": 3.561096307575181e-06, + "loss": 1.5376, + "step": 1063 + }, + { + "epoch": 1.0929635336414998, + "grad_norm": 2.6815669536590576, + "learning_rate": 3.559192995812714e-06, + "loss": 1.5681, + "step": 1064 + }, + { + "epoch": 1.0939907550077042, + "grad_norm": 2.691917896270752, + "learning_rate": 3.557289684050248e-06, + "loss": 1.5868, + "step": 1065 + }, + { + "epoch": 1.0950179763739085, + "grad_norm": 2.8439927101135254, + "learning_rate": 3.555386372287781e-06, + "loss": 1.5763, + "step": 1066 + }, + { + "epoch": 1.0960451977401129, + "grad_norm": 2.6857359409332275, + "learning_rate": 3.5534830605253145e-06, + "loss": 1.5784, + "step": 1067 + }, + { + "epoch": 1.0970724191063175, + "grad_norm": 2.532336950302124, + "learning_rate": 3.551579748762848e-06, + "loss": 1.5409, + "step": 1068 + }, + { + "epoch": 1.0980996404725218, + "grad_norm": 2.6567013263702393, + "learning_rate": 3.5496764370003812e-06, + "loss": 1.5484, + "step": 1069 + }, + { + "epoch": 1.0991268618387262, + "grad_norm": 2.800318717956543, + "learning_rate": 3.5477731252379146e-06, + "loss": 1.5247, + "step": 1070 + }, + { + "epoch": 1.0991268618387262, + "eval_loss": 1.9268279075622559, + "eval_runtime": 45.4409, + "eval_samples_per_second": 22.007, + "eval_steps_per_second": 0.924, + "step": 1070 + }, + { + "epoch": 1.1001540832049306, + "grad_norm": 2.6051485538482666, + "learning_rate": 3.5458698134754475e-06, + "loss": 1.5002, + "step": 1071 + }, + { + "epoch": 1.1011813045711352, + "grad_norm": 2.590815544128418, + "learning_rate": 3.543966501712981e-06, + "loss": 1.4982, + "step": 1072 + }, + { + "epoch": 1.1022085259373395, + "grad_norm": 2.6754868030548096, + "learning_rate": 3.5420631899505142e-06, + "loss": 1.5303, + "step": 1073 + }, + { + "epoch": 1.103235747303544, + "grad_norm": 2.8132805824279785, + "learning_rate": 3.5401598781880476e-06, + "loss": 1.6188, + "step": 1074 + }, + { + "epoch": 1.1042629686697483, + "grad_norm": 2.6267812252044678, + "learning_rate": 3.538256566425581e-06, + "loss": 1.442, + "step": 1075 + }, + { + "epoch": 1.1052901900359529, + "grad_norm": 2.945460557937622, + "learning_rate": 3.536353254663114e-06, + "loss": 1.5787, + "step": 1076 + }, + { + "epoch": 1.1063174114021572, + "grad_norm": 2.831702709197998, + "learning_rate": 3.5344499429006473e-06, + "loss": 1.5396, + "step": 1077 + }, + { + "epoch": 1.1073446327683616, + "grad_norm": 2.614482879638672, + "learning_rate": 3.5325466311381806e-06, + "loss": 1.5101, + "step": 1078 + }, + { + "epoch": 1.108371854134566, + "grad_norm": 2.809013605117798, + "learning_rate": 3.530643319375714e-06, + "loss": 1.5678, + "step": 1079 + }, + { + "epoch": 1.1093990755007703, + "grad_norm": 2.6645243167877197, + "learning_rate": 3.5287400076132474e-06, + "loss": 1.4762, + "step": 1080 + }, + { + "epoch": 1.1093990755007703, + "eval_loss": 1.9386240243911743, + "eval_runtime": 43.5259, + "eval_samples_per_second": 22.975, + "eval_steps_per_second": 0.965, + "step": 1080 + }, + { + "epoch": 1.110426296866975, + "grad_norm": 2.5597710609436035, + "learning_rate": 3.5268366958507803e-06, + "loss": 1.5208, + "step": 1081 + }, + { + "epoch": 1.1114535182331793, + "grad_norm": 2.6656508445739746, + "learning_rate": 3.5249333840883137e-06, + "loss": 1.5548, + "step": 1082 + }, + { + "epoch": 1.1124807395993837, + "grad_norm": 2.651195526123047, + "learning_rate": 3.523030072325847e-06, + "loss": 1.546, + "step": 1083 + }, + { + "epoch": 1.113507960965588, + "grad_norm": 2.946263551712036, + "learning_rate": 3.5211267605633804e-06, + "loss": 1.5058, + "step": 1084 + }, + { + "epoch": 1.1145351823317924, + "grad_norm": 2.7462689876556396, + "learning_rate": 3.519223448800914e-06, + "loss": 1.5213, + "step": 1085 + }, + { + "epoch": 1.115562403697997, + "grad_norm": 2.803926944732666, + "learning_rate": 3.5173201370384475e-06, + "loss": 1.5266, + "step": 1086 + }, + { + "epoch": 1.1165896250642013, + "grad_norm": 2.6583237648010254, + "learning_rate": 3.515416825275981e-06, + "loss": 1.5362, + "step": 1087 + }, + { + "epoch": 1.1176168464304057, + "grad_norm": 2.659313201904297, + "learning_rate": 3.513513513513514e-06, + "loss": 1.5365, + "step": 1088 + }, + { + "epoch": 1.11864406779661, + "grad_norm": 2.7181215286254883, + "learning_rate": 3.511610201751047e-06, + "loss": 1.5584, + "step": 1089 + }, + { + "epoch": 1.1196712891628147, + "grad_norm": 2.543297052383423, + "learning_rate": 3.5097068899885806e-06, + "loss": 1.4599, + "step": 1090 + }, + { + "epoch": 1.1196712891628147, + "eval_loss": 1.9393529891967773, + "eval_runtime": 44.5005, + "eval_samples_per_second": 22.472, + "eval_steps_per_second": 0.944, + "step": 1090 + }, + { + "epoch": 1.120698510529019, + "grad_norm": 2.8569319248199463, + "learning_rate": 3.507803578226114e-06, + "loss": 1.5308, + "step": 1091 + }, + { + "epoch": 1.1217257318952234, + "grad_norm": 2.795915365219116, + "learning_rate": 3.5059002664636473e-06, + "loss": 1.6092, + "step": 1092 + }, + { + "epoch": 1.1227529532614278, + "grad_norm": 2.659125566482544, + "learning_rate": 3.5039969547011802e-06, + "loss": 1.4701, + "step": 1093 + }, + { + "epoch": 1.1237801746276324, + "grad_norm": 2.7170166969299316, + "learning_rate": 3.5020936429387136e-06, + "loss": 1.5232, + "step": 1094 + }, + { + "epoch": 1.1248073959938367, + "grad_norm": 2.6485073566436768, + "learning_rate": 3.500190331176247e-06, + "loss": 1.5274, + "step": 1095 + }, + { + "epoch": 1.125834617360041, + "grad_norm": 2.831904888153076, + "learning_rate": 3.4982870194137803e-06, + "loss": 1.5833, + "step": 1096 + }, + { + "epoch": 1.1268618387262455, + "grad_norm": 2.877516984939575, + "learning_rate": 3.4963837076513137e-06, + "loss": 1.5173, + "step": 1097 + }, + { + "epoch": 1.1278890600924498, + "grad_norm": 2.5502521991729736, + "learning_rate": 3.4944803958888466e-06, + "loss": 1.5119, + "step": 1098 + }, + { + "epoch": 1.1289162814586544, + "grad_norm": 2.530945301055908, + "learning_rate": 3.49257708412638e-06, + "loss": 1.4573, + "step": 1099 + }, + { + "epoch": 1.1299435028248588, + "grad_norm": 2.7266969680786133, + "learning_rate": 3.4906737723639133e-06, + "loss": 1.5812, + "step": 1100 + }, + { + "epoch": 1.1299435028248588, + "eval_loss": 1.9370596408843994, + "eval_runtime": 47.2327, + "eval_samples_per_second": 21.172, + "eval_steps_per_second": 0.889, + "step": 1100 + }, + { + "epoch": 1.1309707241910631, + "grad_norm": 2.8846628665924072, + "learning_rate": 3.4887704606014467e-06, + "loss": 1.5945, + "step": 1101 + }, + { + "epoch": 1.1319979455572675, + "grad_norm": 2.5253827571868896, + "learning_rate": 3.48686714883898e-06, + "loss": 1.5391, + "step": 1102 + }, + { + "epoch": 1.133025166923472, + "grad_norm": 2.6929149627685547, + "learning_rate": 3.484963837076513e-06, + "loss": 1.5767, + "step": 1103 + }, + { + "epoch": 1.1340523882896765, + "grad_norm": 2.7702295780181885, + "learning_rate": 3.4830605253140464e-06, + "loss": 1.5206, + "step": 1104 + }, + { + "epoch": 1.1350796096558808, + "grad_norm": 2.6476149559020996, + "learning_rate": 3.4811572135515797e-06, + "loss": 1.4735, + "step": 1105 + }, + { + "epoch": 1.1361068310220852, + "grad_norm": 2.5630619525909424, + "learning_rate": 3.4792539017891135e-06, + "loss": 1.5109, + "step": 1106 + }, + { + "epoch": 1.1371340523882898, + "grad_norm": 2.549581289291382, + "learning_rate": 3.477350590026647e-06, + "loss": 1.4919, + "step": 1107 + }, + { + "epoch": 1.1381612737544942, + "grad_norm": 2.8687074184417725, + "learning_rate": 3.4754472782641802e-06, + "loss": 1.5658, + "step": 1108 + }, + { + "epoch": 1.1391884951206985, + "grad_norm": 2.799149751663208, + "learning_rate": 3.4735439665017136e-06, + "loss": 1.5582, + "step": 1109 + }, + { + "epoch": 1.140215716486903, + "grad_norm": 2.8335328102111816, + "learning_rate": 3.4716406547392465e-06, + "loss": 1.5217, + "step": 1110 + }, + { + "epoch": 1.140215716486903, + "eval_loss": 1.9011008739471436, + "eval_runtime": 42.9222, + "eval_samples_per_second": 23.298, + "eval_steps_per_second": 0.979, + "step": 1110 + }, + { + "epoch": 1.1412429378531073, + "grad_norm": 2.7025723457336426, + "learning_rate": 3.46973734297678e-06, + "loss": 1.5233, + "step": 1111 + }, + { + "epoch": 1.1422701592193119, + "grad_norm": 2.6984663009643555, + "learning_rate": 3.4678340312143133e-06, + "loss": 1.5447, + "step": 1112 + }, + { + "epoch": 1.1432973805855162, + "grad_norm": 2.708031415939331, + "learning_rate": 3.4659307194518466e-06, + "loss": 1.5905, + "step": 1113 + }, + { + "epoch": 1.1443246019517206, + "grad_norm": 2.5180366039276123, + "learning_rate": 3.46402740768938e-06, + "loss": 1.5636, + "step": 1114 + }, + { + "epoch": 1.145351823317925, + "grad_norm": 2.540074348449707, + "learning_rate": 3.462124095926913e-06, + "loss": 1.5045, + "step": 1115 + }, + { + "epoch": 1.1463790446841293, + "grad_norm": 2.4670231342315674, + "learning_rate": 3.4602207841644463e-06, + "loss": 1.4969, + "step": 1116 + }, + { + "epoch": 1.147406266050334, + "grad_norm": 2.7803049087524414, + "learning_rate": 3.4583174724019797e-06, + "loss": 1.5471, + "step": 1117 + }, + { + "epoch": 1.1484334874165383, + "grad_norm": 2.58117413520813, + "learning_rate": 3.456414160639513e-06, + "loss": 1.592, + "step": 1118 + }, + { + "epoch": 1.1494607087827426, + "grad_norm": 2.5810937881469727, + "learning_rate": 3.4545108488770464e-06, + "loss": 1.5542, + "step": 1119 + }, + { + "epoch": 1.150487930148947, + "grad_norm": 2.757809638977051, + "learning_rate": 3.4526075371145793e-06, + "loss": 1.5403, + "step": 1120 + }, + { + "epoch": 1.150487930148947, + "eval_loss": 1.905651569366455, + "eval_runtime": 45.619, + "eval_samples_per_second": 21.921, + "eval_steps_per_second": 0.921, + "step": 1120 + }, + { + "epoch": 1.1515151515151516, + "grad_norm": 2.6740472316741943, + "learning_rate": 3.4507042253521127e-06, + "loss": 1.5422, + "step": 1121 + }, + { + "epoch": 1.152542372881356, + "grad_norm": 2.7041897773742676, + "learning_rate": 3.448800913589646e-06, + "loss": 1.6049, + "step": 1122 + }, + { + "epoch": 1.1535695942475603, + "grad_norm": 2.7759902477264404, + "learning_rate": 3.4468976018271794e-06, + "loss": 1.522, + "step": 1123 + }, + { + "epoch": 1.1545968156137647, + "grad_norm": 2.69626522064209, + "learning_rate": 3.4449942900647128e-06, + "loss": 1.6096, + "step": 1124 + }, + { + "epoch": 1.1556240369799693, + "grad_norm": 2.57637619972229, + "learning_rate": 3.443090978302246e-06, + "loss": 1.5702, + "step": 1125 + }, + { + "epoch": 1.1566512583461737, + "grad_norm": 2.6259238719940186, + "learning_rate": 3.44118766653978e-06, + "loss": 1.5561, + "step": 1126 + }, + { + "epoch": 1.157678479712378, + "grad_norm": 2.7071187496185303, + "learning_rate": 3.439284354777313e-06, + "loss": 1.5545, + "step": 1127 + }, + { + "epoch": 1.1587057010785824, + "grad_norm": 2.7928988933563232, + "learning_rate": 3.4373810430148462e-06, + "loss": 1.5551, + "step": 1128 + }, + { + "epoch": 1.1597329224447868, + "grad_norm": 2.703719139099121, + "learning_rate": 3.4354777312523796e-06, + "loss": 1.4947, + "step": 1129 + }, + { + "epoch": 1.1607601438109914, + "grad_norm": 2.6363744735717773, + "learning_rate": 3.433574419489913e-06, + "loss": 1.511, + "step": 1130 + }, + { + "epoch": 1.1607601438109914, + "eval_loss": 1.9003485441207886, + "eval_runtime": 46.8855, + "eval_samples_per_second": 21.329, + "eval_steps_per_second": 0.896, + "step": 1130 + }, + { + "epoch": 1.1617873651771957, + "grad_norm": 2.7165591716766357, + "learning_rate": 3.4316711077274463e-06, + "loss": 1.5529, + "step": 1131 + }, + { + "epoch": 1.1628145865434, + "grad_norm": 2.6615865230560303, + "learning_rate": 3.4297677959649792e-06, + "loss": 1.5213, + "step": 1132 + }, + { + "epoch": 1.1638418079096045, + "grad_norm": 2.6274473667144775, + "learning_rate": 3.4278644842025126e-06, + "loss": 1.5813, + "step": 1133 + }, + { + "epoch": 1.1648690292758088, + "grad_norm": 2.637418031692505, + "learning_rate": 3.425961172440046e-06, + "loss": 1.4963, + "step": 1134 + }, + { + "epoch": 1.1658962506420134, + "grad_norm": 2.601574182510376, + "learning_rate": 3.4240578606775793e-06, + "loss": 1.6146, + "step": 1135 + }, + { + "epoch": 1.1669234720082178, + "grad_norm": 2.599043369293213, + "learning_rate": 3.4221545489151127e-06, + "loss": 1.5251, + "step": 1136 + }, + { + "epoch": 1.1679506933744221, + "grad_norm": 2.8078091144561768, + "learning_rate": 3.4202512371526456e-06, + "loss": 1.5231, + "step": 1137 + }, + { + "epoch": 1.1689779147406265, + "grad_norm": 2.811053514480591, + "learning_rate": 3.418347925390179e-06, + "loss": 1.6289, + "step": 1138 + }, + { + "epoch": 1.170005136106831, + "grad_norm": 2.7277560234069824, + "learning_rate": 3.4164446136277124e-06, + "loss": 1.5733, + "step": 1139 + }, + { + "epoch": 1.1710323574730355, + "grad_norm": 2.587024211883545, + "learning_rate": 3.4145413018652457e-06, + "loss": 1.5332, + "step": 1140 + }, + { + "epoch": 1.1710323574730355, + "eval_loss": 1.8978687524795532, + "eval_runtime": 49.9935, + "eval_samples_per_second": 20.003, + "eval_steps_per_second": 0.84, + "step": 1140 + }, + { + "epoch": 1.1720595788392398, + "grad_norm": 2.536479949951172, + "learning_rate": 3.412637990102779e-06, + "loss": 1.5601, + "step": 1141 + }, + { + "epoch": 1.1730868002054442, + "grad_norm": 2.7315709590911865, + "learning_rate": 3.4107346783403125e-06, + "loss": 1.5783, + "step": 1142 + }, + { + "epoch": 1.1741140215716488, + "grad_norm": 2.507249593734741, + "learning_rate": 3.4088313665778454e-06, + "loss": 1.5, + "step": 1143 + }, + { + "epoch": 1.1751412429378532, + "grad_norm": 2.625411033630371, + "learning_rate": 3.4069280548153788e-06, + "loss": 1.5328, + "step": 1144 + }, + { + "epoch": 1.1761684643040575, + "grad_norm": 2.591282367706299, + "learning_rate": 3.405024743052912e-06, + "loss": 1.5119, + "step": 1145 + }, + { + "epoch": 1.177195685670262, + "grad_norm": 2.547508955001831, + "learning_rate": 3.403121431290446e-06, + "loss": 1.5361, + "step": 1146 + }, + { + "epoch": 1.1782229070364663, + "grad_norm": 2.797992706298828, + "learning_rate": 3.4012181195279793e-06, + "loss": 1.5221, + "step": 1147 + }, + { + "epoch": 1.1792501284026708, + "grad_norm": 2.905266761779785, + "learning_rate": 3.3993148077655126e-06, + "loss": 1.5209, + "step": 1148 + }, + { + "epoch": 1.1802773497688752, + "grad_norm": 2.9259016513824463, + "learning_rate": 3.3974114960030456e-06, + "loss": 1.5365, + "step": 1149 + }, + { + "epoch": 1.1813045711350796, + "grad_norm": 2.597102165222168, + "learning_rate": 3.395508184240579e-06, + "loss": 1.4663, + "step": 1150 + }, + { + "epoch": 1.1813045711350796, + "eval_loss": 1.8966383934020996, + "eval_runtime": 47.3579, + "eval_samples_per_second": 21.116, + "eval_steps_per_second": 0.887, + "step": 1150 + }, + { + "epoch": 1.182331792501284, + "grad_norm": 2.6672847270965576, + "learning_rate": 3.3936048724781123e-06, + "loss": 1.5115, + "step": 1151 + }, + { + "epoch": 1.1833590138674885, + "grad_norm": 2.870549201965332, + "learning_rate": 3.3917015607156457e-06, + "loss": 1.6097, + "step": 1152 + }, + { + "epoch": 1.184386235233693, + "grad_norm": 2.6559388637542725, + "learning_rate": 3.389798248953179e-06, + "loss": 1.4777, + "step": 1153 + }, + { + "epoch": 1.1854134565998973, + "grad_norm": 2.7968215942382812, + "learning_rate": 3.387894937190712e-06, + "loss": 1.509, + "step": 1154 + }, + { + "epoch": 1.1864406779661016, + "grad_norm": 2.689741611480713, + "learning_rate": 3.3859916254282453e-06, + "loss": 1.5424, + "step": 1155 + }, + { + "epoch": 1.1874678993323062, + "grad_norm": 2.872887134552002, + "learning_rate": 3.3840883136657787e-06, + "loss": 1.5404, + "step": 1156 + }, + { + "epoch": 1.1884951206985106, + "grad_norm": 2.6436610221862793, + "learning_rate": 3.382185001903312e-06, + "loss": 1.5889, + "step": 1157 + }, + { + "epoch": 1.189522342064715, + "grad_norm": 2.697906255722046, + "learning_rate": 3.3802816901408454e-06, + "loss": 1.5086, + "step": 1158 + }, + { + "epoch": 1.1905495634309193, + "grad_norm": 2.8897645473480225, + "learning_rate": 3.3783783783783788e-06, + "loss": 1.5441, + "step": 1159 + }, + { + "epoch": 1.1915767847971237, + "grad_norm": 2.722351312637329, + "learning_rate": 3.3764750666159117e-06, + "loss": 1.527, + "step": 1160 + }, + { + "epoch": 1.1915767847971237, + "eval_loss": 1.8928718566894531, + "eval_runtime": 48.9427, + "eval_samples_per_second": 20.432, + "eval_steps_per_second": 0.858, + "step": 1160 + }, + { + "epoch": 1.1926040061633283, + "grad_norm": 2.5705020427703857, + "learning_rate": 3.374571754853445e-06, + "loss": 1.5304, + "step": 1161 + }, + { + "epoch": 1.1936312275295327, + "grad_norm": 2.571810245513916, + "learning_rate": 3.3726684430909784e-06, + "loss": 1.5242, + "step": 1162 + }, + { + "epoch": 1.194658448895737, + "grad_norm": 2.552725076675415, + "learning_rate": 3.370765131328512e-06, + "loss": 1.4803, + "step": 1163 + }, + { + "epoch": 1.1956856702619414, + "grad_norm": 2.734027862548828, + "learning_rate": 3.368861819566045e-06, + "loss": 1.5041, + "step": 1164 + }, + { + "epoch": 1.1967128916281458, + "grad_norm": 2.594614028930664, + "learning_rate": 3.366958507803578e-06, + "loss": 1.5205, + "step": 1165 + }, + { + "epoch": 1.1977401129943503, + "grad_norm": 2.897482395172119, + "learning_rate": 3.365055196041112e-06, + "loss": 1.5372, + "step": 1166 + }, + { + "epoch": 1.1987673343605547, + "grad_norm": 2.6143221855163574, + "learning_rate": 3.3631518842786452e-06, + "loss": 1.4966, + "step": 1167 + }, + { + "epoch": 1.199794555726759, + "grad_norm": 2.845475912094116, + "learning_rate": 3.3612485725161786e-06, + "loss": 1.5759, + "step": 1168 + }, + { + "epoch": 1.2008217770929634, + "grad_norm": 2.667388916015625, + "learning_rate": 3.359345260753712e-06, + "loss": 1.5014, + "step": 1169 + }, + { + "epoch": 1.201848998459168, + "grad_norm": 2.890197277069092, + "learning_rate": 3.3574419489912453e-06, + "loss": 1.5939, + "step": 1170 + }, + { + "epoch": 1.201848998459168, + "eval_loss": 1.8978379964828491, + "eval_runtime": 48.9839, + "eval_samples_per_second": 20.415, + "eval_steps_per_second": 0.857, + "step": 1170 + }, + { + "epoch": 1.2028762198253724, + "grad_norm": 2.578139066696167, + "learning_rate": 3.3555386372287783e-06, + "loss": 1.5655, + "step": 1171 + }, + { + "epoch": 1.2039034411915768, + "grad_norm": 2.5672738552093506, + "learning_rate": 3.3536353254663116e-06, + "loss": 1.5998, + "step": 1172 + }, + { + "epoch": 1.2049306625577811, + "grad_norm": 2.7297089099884033, + "learning_rate": 3.351732013703845e-06, + "loss": 1.4938, + "step": 1173 + }, + { + "epoch": 1.2059578839239857, + "grad_norm": 2.7455339431762695, + "learning_rate": 3.3498287019413784e-06, + "loss": 1.5505, + "step": 1174 + }, + { + "epoch": 1.20698510529019, + "grad_norm": 2.725179433822632, + "learning_rate": 3.3479253901789117e-06, + "loss": 1.5524, + "step": 1175 + }, + { + "epoch": 1.2080123266563945, + "grad_norm": 2.6953988075256348, + "learning_rate": 3.346022078416445e-06, + "loss": 1.568, + "step": 1176 + }, + { + "epoch": 1.2090395480225988, + "grad_norm": 2.613600492477417, + "learning_rate": 3.344118766653978e-06, + "loss": 1.4608, + "step": 1177 + }, + { + "epoch": 1.2100667693888032, + "grad_norm": 2.7339162826538086, + "learning_rate": 3.3422154548915114e-06, + "loss": 1.5426, + "step": 1178 + }, + { + "epoch": 1.2110939907550078, + "grad_norm": 2.774726152420044, + "learning_rate": 3.3403121431290448e-06, + "loss": 1.5192, + "step": 1179 + }, + { + "epoch": 1.2121212121212122, + "grad_norm": 2.7361602783203125, + "learning_rate": 3.338408831366578e-06, + "loss": 1.5481, + "step": 1180 + }, + { + "epoch": 1.2121212121212122, + "eval_loss": 1.902246117591858, + "eval_runtime": 45.153, + "eval_samples_per_second": 22.147, + "eval_steps_per_second": 0.93, + "step": 1180 + }, + { + "epoch": 1.2131484334874165, + "grad_norm": 2.6936450004577637, + "learning_rate": 3.3365055196041115e-06, + "loss": 1.4821, + "step": 1181 + }, + { + "epoch": 1.214175654853621, + "grad_norm": 2.881898880004883, + "learning_rate": 3.3346022078416444e-06, + "loss": 1.5663, + "step": 1182 + }, + { + "epoch": 1.2152028762198253, + "grad_norm": 2.793069362640381, + "learning_rate": 3.3326988960791778e-06, + "loss": 1.5998, + "step": 1183 + }, + { + "epoch": 1.2162300975860298, + "grad_norm": 2.591367721557617, + "learning_rate": 3.330795584316711e-06, + "loss": 1.5199, + "step": 1184 + }, + { + "epoch": 1.2172573189522342, + "grad_norm": 2.8402175903320312, + "learning_rate": 3.3288922725542445e-06, + "loss": 1.451, + "step": 1185 + }, + { + "epoch": 1.2182845403184386, + "grad_norm": 2.737199068069458, + "learning_rate": 3.326988960791778e-06, + "loss": 1.5117, + "step": 1186 + }, + { + "epoch": 1.219311761684643, + "grad_norm": 2.915210723876953, + "learning_rate": 3.3250856490293117e-06, + "loss": 1.5388, + "step": 1187 + }, + { + "epoch": 1.2203389830508475, + "grad_norm": 3.091784715652466, + "learning_rate": 3.323182337266845e-06, + "loss": 1.5944, + "step": 1188 + }, + { + "epoch": 1.221366204417052, + "grad_norm": 2.770756483078003, + "learning_rate": 3.321279025504378e-06, + "loss": 1.5619, + "step": 1189 + }, + { + "epoch": 1.2223934257832563, + "grad_norm": 2.7989954948425293, + "learning_rate": 3.3193757137419113e-06, + "loss": 1.5581, + "step": 1190 + }, + { + "epoch": 1.2223934257832563, + "eval_loss": 1.907904863357544, + "eval_runtime": 46.8087, + "eval_samples_per_second": 21.364, + "eval_steps_per_second": 0.897, + "step": 1190 + }, + { + "epoch": 1.2234206471494606, + "grad_norm": 2.6740641593933105, + "learning_rate": 3.3174724019794447e-06, + "loss": 1.5556, + "step": 1191 + }, + { + "epoch": 1.2244478685156652, + "grad_norm": 2.760608196258545, + "learning_rate": 3.315569090216978e-06, + "loss": 1.5903, + "step": 1192 + }, + { + "epoch": 1.2254750898818696, + "grad_norm": 2.942370891571045, + "learning_rate": 3.3136657784545114e-06, + "loss": 1.5052, + "step": 1193 + }, + { + "epoch": 1.226502311248074, + "grad_norm": 2.8096179962158203, + "learning_rate": 3.3117624666920443e-06, + "loss": 1.5933, + "step": 1194 + }, + { + "epoch": 1.2275295326142783, + "grad_norm": 2.7357091903686523, + "learning_rate": 3.3098591549295777e-06, + "loss": 1.5074, + "step": 1195 + }, + { + "epoch": 1.2285567539804827, + "grad_norm": 2.728134870529175, + "learning_rate": 3.307955843167111e-06, + "loss": 1.48, + "step": 1196 + }, + { + "epoch": 1.2295839753466873, + "grad_norm": 2.921462059020996, + "learning_rate": 3.3060525314046444e-06, + "loss": 1.5602, + "step": 1197 + }, + { + "epoch": 1.2306111967128917, + "grad_norm": 2.7563631534576416, + "learning_rate": 3.304149219642178e-06, + "loss": 1.5309, + "step": 1198 + }, + { + "epoch": 1.231638418079096, + "grad_norm": 2.533001661300659, + "learning_rate": 3.3022459078797107e-06, + "loss": 1.5242, + "step": 1199 + }, + { + "epoch": 1.2326656394453004, + "grad_norm": 2.5509681701660156, + "learning_rate": 3.300342596117244e-06, + "loss": 1.5488, + "step": 1200 + }, + { + "epoch": 1.2326656394453004, + "eval_loss": 1.8983508348464966, + "eval_runtime": 44.8654, + "eval_samples_per_second": 22.289, + "eval_steps_per_second": 0.936, + "step": 1200 + }, + { + "epoch": 1.2336928608115048, + "grad_norm": 2.688228130340576, + "learning_rate": 3.2984392843547775e-06, + "loss": 1.5344, + "step": 1201 + }, + { + "epoch": 1.2347200821777093, + "grad_norm": 2.567621946334839, + "learning_rate": 3.296535972592311e-06, + "loss": 1.4926, + "step": 1202 + }, + { + "epoch": 1.2357473035439137, + "grad_norm": 2.586637258529663, + "learning_rate": 3.294632660829844e-06, + "loss": 1.4749, + "step": 1203 + }, + { + "epoch": 1.236774524910118, + "grad_norm": 2.697901725769043, + "learning_rate": 3.292729349067377e-06, + "loss": 1.5073, + "step": 1204 + }, + { + "epoch": 1.2378017462763227, + "grad_norm": 2.832059621810913, + "learning_rate": 3.2908260373049105e-06, + "loss": 1.5147, + "step": 1205 + }, + { + "epoch": 1.238828967642527, + "grad_norm": 2.7802395820617676, + "learning_rate": 3.288922725542444e-06, + "loss": 1.5031, + "step": 1206 + }, + { + "epoch": 1.2398561890087314, + "grad_norm": 2.8382885456085205, + "learning_rate": 3.2870194137799776e-06, + "loss": 1.5539, + "step": 1207 + }, + { + "epoch": 1.2408834103749358, + "grad_norm": 2.7725186347961426, + "learning_rate": 3.285116102017511e-06, + "loss": 1.5145, + "step": 1208 + }, + { + "epoch": 1.2419106317411401, + "grad_norm": 2.563866376876831, + "learning_rate": 3.2832127902550444e-06, + "loss": 1.4852, + "step": 1209 + }, + { + "epoch": 1.2429378531073447, + "grad_norm": 2.7052412033081055, + "learning_rate": 3.2813094784925777e-06, + "loss": 1.5615, + "step": 1210 + }, + { + "epoch": 1.2429378531073447, + "eval_loss": 1.9071296453475952, + "eval_runtime": 42.2147, + "eval_samples_per_second": 23.688, + "eval_steps_per_second": 0.995, + "step": 1210 + }, + { + "epoch": 1.243965074473549, + "grad_norm": 2.7794888019561768, + "learning_rate": 3.2794061667301107e-06, + "loss": 1.5445, + "step": 1211 + }, + { + "epoch": 1.2449922958397535, + "grad_norm": 2.6376712322235107, + "learning_rate": 3.277502854967644e-06, + "loss": 1.4728, + "step": 1212 + }, + { + "epoch": 1.2460195172059578, + "grad_norm": 2.5500779151916504, + "learning_rate": 3.2755995432051774e-06, + "loss": 1.5133, + "step": 1213 + }, + { + "epoch": 1.2470467385721622, + "grad_norm": 2.629063606262207, + "learning_rate": 3.2736962314427107e-06, + "loss": 1.5093, + "step": 1214 + }, + { + "epoch": 1.2480739599383668, + "grad_norm": 2.662497043609619, + "learning_rate": 3.271792919680244e-06, + "loss": 1.5237, + "step": 1215 + }, + { + "epoch": 1.2491011813045712, + "grad_norm": 2.5375289916992188, + "learning_rate": 3.269889607917777e-06, + "loss": 1.5211, + "step": 1216 + }, + { + "epoch": 1.2501284026707755, + "grad_norm": 2.8396782875061035, + "learning_rate": 3.2679862961553104e-06, + "loss": 1.5773, + "step": 1217 + }, + { + "epoch": 1.25115562403698, + "grad_norm": 2.8443665504455566, + "learning_rate": 3.2660829843928438e-06, + "loss": 1.4987, + "step": 1218 + }, + { + "epoch": 1.2521828454031843, + "grad_norm": 2.77935528755188, + "learning_rate": 3.264179672630377e-06, + "loss": 1.5344, + "step": 1219 + }, + { + "epoch": 1.2532100667693888, + "grad_norm": 3.0581092834472656, + "learning_rate": 3.2622763608679105e-06, + "loss": 1.5728, + "step": 1220 + }, + { + "epoch": 1.2532100667693888, + "eval_loss": 1.9198752641677856, + "eval_runtime": 45.1116, + "eval_samples_per_second": 22.167, + "eval_steps_per_second": 0.931, + "step": 1220 + }, + { + "epoch": 1.2542372881355932, + "grad_norm": 2.8741002082824707, + "learning_rate": 3.2603730491054434e-06, + "loss": 1.5444, + "step": 1221 + }, + { + "epoch": 1.2552645095017976, + "grad_norm": 2.6263391971588135, + "learning_rate": 3.258469737342977e-06, + "loss": 1.4997, + "step": 1222 + }, + { + "epoch": 1.2562917308680022, + "grad_norm": 2.949795961380005, + "learning_rate": 3.25656642558051e-06, + "loss": 1.5145, + "step": 1223 + }, + { + "epoch": 1.2573189522342065, + "grad_norm": 2.846130609512329, + "learning_rate": 3.2546631138180435e-06, + "loss": 1.4709, + "step": 1224 + }, + { + "epoch": 1.258346173600411, + "grad_norm": 2.6340930461883545, + "learning_rate": 3.252759802055577e-06, + "loss": 1.5087, + "step": 1225 + }, + { + "epoch": 1.2593733949666153, + "grad_norm": 2.9878976345062256, + "learning_rate": 3.25085649029311e-06, + "loss": 1.5142, + "step": 1226 + }, + { + "epoch": 1.2604006163328196, + "grad_norm": 2.9880733489990234, + "learning_rate": 3.248953178530644e-06, + "loss": 1.521, + "step": 1227 + }, + { + "epoch": 1.2614278376990242, + "grad_norm": 2.8507912158966064, + "learning_rate": 3.247049866768177e-06, + "loss": 1.5482, + "step": 1228 + }, + { + "epoch": 1.2624550590652286, + "grad_norm": 2.870678424835205, + "learning_rate": 3.2451465550057103e-06, + "loss": 1.4965, + "step": 1229 + }, + { + "epoch": 1.263482280431433, + "grad_norm": 2.644347906112671, + "learning_rate": 3.2432432432432437e-06, + "loss": 1.4614, + "step": 1230 + }, + { + "epoch": 1.263482280431433, + "eval_loss": 1.9100512266159058, + "eval_runtime": 44.6472, + "eval_samples_per_second": 22.398, + "eval_steps_per_second": 0.941, + "step": 1230 + }, + { + "epoch": 1.2645095017976373, + "grad_norm": 2.5852861404418945, + "learning_rate": 3.241339931480777e-06, + "loss": 1.5122, + "step": 1231 + }, + { + "epoch": 1.2655367231638417, + "grad_norm": 2.9572696685791016, + "learning_rate": 3.2394366197183104e-06, + "loss": 1.5529, + "step": 1232 + }, + { + "epoch": 1.2665639445300463, + "grad_norm": 2.7040176391601562, + "learning_rate": 3.2375333079558434e-06, + "loss": 1.5418, + "step": 1233 + }, + { + "epoch": 1.2675911658962506, + "grad_norm": 2.577706813812256, + "learning_rate": 3.2356299961933767e-06, + "loss": 1.5167, + "step": 1234 + }, + { + "epoch": 1.268618387262455, + "grad_norm": 2.6559629440307617, + "learning_rate": 3.23372668443091e-06, + "loss": 1.5278, + "step": 1235 + }, + { + "epoch": 1.2696456086286596, + "grad_norm": 2.673293113708496, + "learning_rate": 3.2318233726684435e-06, + "loss": 1.5393, + "step": 1236 + }, + { + "epoch": 1.270672829994864, + "grad_norm": 2.507797956466675, + "learning_rate": 3.229920060905977e-06, + "loss": 1.5573, + "step": 1237 + }, + { + "epoch": 1.2717000513610683, + "grad_norm": 2.8361523151397705, + "learning_rate": 3.2280167491435098e-06, + "loss": 1.4911, + "step": 1238 + }, + { + "epoch": 1.2727272727272727, + "grad_norm": 2.7824695110321045, + "learning_rate": 3.226113437381043e-06, + "loss": 1.4949, + "step": 1239 + }, + { + "epoch": 1.273754494093477, + "grad_norm": 2.4863975048065186, + "learning_rate": 3.2242101256185765e-06, + "loss": 1.5017, + "step": 1240 + }, + { + "epoch": 1.273754494093477, + "eval_loss": 1.896610140800476, + "eval_runtime": 46.9004, + "eval_samples_per_second": 21.322, + "eval_steps_per_second": 0.896, + "step": 1240 + }, + { + "epoch": 1.2747817154596817, + "grad_norm": 2.767557144165039, + "learning_rate": 3.22230681385611e-06, + "loss": 1.5137, + "step": 1241 + }, + { + "epoch": 1.275808936825886, + "grad_norm": 2.544274091720581, + "learning_rate": 3.220403502093643e-06, + "loss": 1.5759, + "step": 1242 + }, + { + "epoch": 1.2768361581920904, + "grad_norm": 2.6679677963256836, + "learning_rate": 3.218500190331176e-06, + "loss": 1.533, + "step": 1243 + }, + { + "epoch": 1.2778633795582948, + "grad_norm": 2.634310245513916, + "learning_rate": 3.2165968785687095e-06, + "loss": 1.5605, + "step": 1244 + }, + { + "epoch": 1.2788906009244991, + "grad_norm": 2.6534063816070557, + "learning_rate": 3.214693566806243e-06, + "loss": 1.5129, + "step": 1245 + }, + { + "epoch": 1.2799178222907037, + "grad_norm": 2.6762259006500244, + "learning_rate": 3.2127902550437762e-06, + "loss": 1.5435, + "step": 1246 + }, + { + "epoch": 1.280945043656908, + "grad_norm": 2.70534610748291, + "learning_rate": 3.21088694328131e-06, + "loss": 1.5215, + "step": 1247 + }, + { + "epoch": 1.2819722650231125, + "grad_norm": 2.929002046585083, + "learning_rate": 3.2089836315188434e-06, + "loss": 1.5237, + "step": 1248 + }, + { + "epoch": 1.2829994863893168, + "grad_norm": 2.6163816452026367, + "learning_rate": 3.2070803197563767e-06, + "loss": 1.5114, + "step": 1249 + }, + { + "epoch": 1.2840267077555212, + "grad_norm": 2.6484017372131348, + "learning_rate": 3.2051770079939097e-06, + "loss": 1.5054, + "step": 1250 + }, + { + "epoch": 1.2840267077555212, + "eval_loss": 1.9057985544204712, + "eval_runtime": 45.9877, + "eval_samples_per_second": 21.745, + "eval_steps_per_second": 0.913, + "step": 1250 + }, + { + "epoch": 1.2850539291217258, + "grad_norm": 2.694584608078003, + "learning_rate": 3.203273696231443e-06, + "loss": 1.5272, + "step": 1251 + }, + { + "epoch": 1.2860811504879301, + "grad_norm": 2.699054718017578, + "learning_rate": 3.2013703844689764e-06, + "loss": 1.5632, + "step": 1252 + }, + { + "epoch": 1.2871083718541345, + "grad_norm": 2.765612840652466, + "learning_rate": 3.1994670727065098e-06, + "loss": 1.4927, + "step": 1253 + }, + { + "epoch": 1.288135593220339, + "grad_norm": 2.7850136756896973, + "learning_rate": 3.197563760944043e-06, + "loss": 1.5412, + "step": 1254 + }, + { + "epoch": 1.2891628145865435, + "grad_norm": 2.6565496921539307, + "learning_rate": 3.195660449181576e-06, + "loss": 1.5606, + "step": 1255 + }, + { + "epoch": 1.2901900359527478, + "grad_norm": 2.627163887023926, + "learning_rate": 3.1937571374191094e-06, + "loss": 1.5612, + "step": 1256 + }, + { + "epoch": 1.2912172573189522, + "grad_norm": 2.5778250694274902, + "learning_rate": 3.191853825656643e-06, + "loss": 1.4988, + "step": 1257 + }, + { + "epoch": 1.2922444786851566, + "grad_norm": 2.7875125408172607, + "learning_rate": 3.189950513894176e-06, + "loss": 1.5237, + "step": 1258 + }, + { + "epoch": 1.2932717000513612, + "grad_norm": 2.645707607269287, + "learning_rate": 3.1880472021317095e-06, + "loss": 1.5157, + "step": 1259 + }, + { + "epoch": 1.2942989214175655, + "grad_norm": 2.6457326412200928, + "learning_rate": 3.1861438903692425e-06, + "loss": 1.5109, + "step": 1260 + }, + { + "epoch": 1.2942989214175655, + "eval_loss": 1.896759033203125, + "eval_runtime": 50.1276, + "eval_samples_per_second": 19.949, + "eval_steps_per_second": 0.838, + "step": 1260 + }, + { + "epoch": 1.29532614278377, + "grad_norm": 2.617196798324585, + "learning_rate": 3.184240578606776e-06, + "loss": 1.4646, + "step": 1261 + }, + { + "epoch": 1.2963533641499743, + "grad_norm": 2.618905782699585, + "learning_rate": 3.182337266844309e-06, + "loss": 1.5012, + "step": 1262 + }, + { + "epoch": 1.2973805855161786, + "grad_norm": 2.6222894191741943, + "learning_rate": 3.1804339550818426e-06, + "loss": 1.526, + "step": 1263 + }, + { + "epoch": 1.2984078068823832, + "grad_norm": 2.858421564102173, + "learning_rate": 3.178530643319376e-06, + "loss": 1.5621, + "step": 1264 + }, + { + "epoch": 1.2994350282485876, + "grad_norm": 2.579683303833008, + "learning_rate": 3.176627331556909e-06, + "loss": 1.5217, + "step": 1265 + }, + { + "epoch": 1.300462249614792, + "grad_norm": 2.721665859222412, + "learning_rate": 3.1747240197944422e-06, + "loss": 1.4732, + "step": 1266 + }, + { + "epoch": 1.3014894709809965, + "grad_norm": 2.6431221961975098, + "learning_rate": 3.172820708031976e-06, + "loss": 1.5219, + "step": 1267 + }, + { + "epoch": 1.3025166923472007, + "grad_norm": 2.706671714782715, + "learning_rate": 3.1709173962695094e-06, + "loss": 1.4829, + "step": 1268 + }, + { + "epoch": 1.3035439137134053, + "grad_norm": 2.81333589553833, + "learning_rate": 3.1690140845070427e-06, + "loss": 1.5669, + "step": 1269 + }, + { + "epoch": 1.3045711350796096, + "grad_norm": 2.942777633666992, + "learning_rate": 3.167110772744576e-06, + "loss": 1.5477, + "step": 1270 + }, + { + "epoch": 1.3045711350796096, + "eval_loss": 1.9051777124404907, + "eval_runtime": 48.2881, + "eval_samples_per_second": 20.709, + "eval_steps_per_second": 0.87, + "step": 1270 + }, + { + "epoch": 1.305598356445814, + "grad_norm": 2.721989393234253, + "learning_rate": 3.1652074609821095e-06, + "loss": 1.5269, + "step": 1271 + }, + { + "epoch": 1.3066255778120186, + "grad_norm": 2.8142035007476807, + "learning_rate": 3.1633041492196424e-06, + "loss": 1.5193, + "step": 1272 + }, + { + "epoch": 1.307652799178223, + "grad_norm": 2.796630382537842, + "learning_rate": 3.1614008374571758e-06, + "loss": 1.503, + "step": 1273 + }, + { + "epoch": 1.3086800205444273, + "grad_norm": 2.696720600128174, + "learning_rate": 3.159497525694709e-06, + "loss": 1.5301, + "step": 1274 + }, + { + "epoch": 1.3097072419106317, + "grad_norm": 2.6286635398864746, + "learning_rate": 3.1575942139322425e-06, + "loss": 1.5326, + "step": 1275 + }, + { + "epoch": 1.310734463276836, + "grad_norm": 2.662801504135132, + "learning_rate": 3.155690902169776e-06, + "loss": 1.5048, + "step": 1276 + }, + { + "epoch": 1.3117616846430407, + "grad_norm": 2.838228702545166, + "learning_rate": 3.1537875904073088e-06, + "loss": 1.5321, + "step": 1277 + }, + { + "epoch": 1.312788906009245, + "grad_norm": 2.7021734714508057, + "learning_rate": 3.151884278644842e-06, + "loss": 1.4917, + "step": 1278 + }, + { + "epoch": 1.3138161273754494, + "grad_norm": 2.696664571762085, + "learning_rate": 3.1499809668823755e-06, + "loss": 1.5231, + "step": 1279 + }, + { + "epoch": 1.3148433487416538, + "grad_norm": 2.720292091369629, + "learning_rate": 3.148077655119909e-06, + "loss": 1.497, + "step": 1280 + }, + { + "epoch": 1.3148433487416538, + "eval_loss": 1.8934897184371948, + "eval_runtime": 48.5119, + "eval_samples_per_second": 20.613, + "eval_steps_per_second": 0.866, + "step": 1280 + }, + { + "epoch": 1.3158705701078581, + "grad_norm": 2.6228737831115723, + "learning_rate": 3.1461743433574422e-06, + "loss": 1.5475, + "step": 1281 + }, + { + "epoch": 1.3168977914740627, + "grad_norm": 2.733198881149292, + "learning_rate": 3.144271031594975e-06, + "loss": 1.4909, + "step": 1282 + }, + { + "epoch": 1.317925012840267, + "grad_norm": 2.5863518714904785, + "learning_rate": 3.1423677198325085e-06, + "loss": 1.5446, + "step": 1283 + }, + { + "epoch": 1.3189522342064715, + "grad_norm": 2.775151491165161, + "learning_rate": 3.140464408070042e-06, + "loss": 1.5414, + "step": 1284 + }, + { + "epoch": 1.319979455572676, + "grad_norm": 2.7392020225524902, + "learning_rate": 3.1385610963075753e-06, + "loss": 1.5094, + "step": 1285 + }, + { + "epoch": 1.3210066769388804, + "grad_norm": 2.578563928604126, + "learning_rate": 3.1366577845451086e-06, + "loss": 1.4973, + "step": 1286 + }, + { + "epoch": 1.3220338983050848, + "grad_norm": 2.664562225341797, + "learning_rate": 3.134754472782642e-06, + "loss": 1.4874, + "step": 1287 + }, + { + "epoch": 1.3230611196712891, + "grad_norm": 2.713029146194458, + "learning_rate": 3.1328511610201758e-06, + "loss": 1.479, + "step": 1288 + }, + { + "epoch": 1.3240883410374935, + "grad_norm": 2.569258213043213, + "learning_rate": 3.1309478492577087e-06, + "loss": 1.4641, + "step": 1289 + }, + { + "epoch": 1.325115562403698, + "grad_norm": 2.4881503582000732, + "learning_rate": 3.129044537495242e-06, + "loss": 1.4634, + "step": 1290 + }, + { + "epoch": 1.325115562403698, + "eval_loss": 1.8945759534835815, + "eval_runtime": 48.2935, + "eval_samples_per_second": 20.707, + "eval_steps_per_second": 0.87, + "step": 1290 + }, + { + "epoch": 1.3261427837699025, + "grad_norm": 2.485280752182007, + "learning_rate": 3.1271412257327754e-06, + "loss": 1.507, + "step": 1291 + }, + { + "epoch": 1.3271700051361068, + "grad_norm": 2.7823352813720703, + "learning_rate": 3.125237913970309e-06, + "loss": 1.5414, + "step": 1292 + }, + { + "epoch": 1.3281972265023112, + "grad_norm": 2.516641139984131, + "learning_rate": 3.123334602207842e-06, + "loss": 1.4889, + "step": 1293 + }, + { + "epoch": 1.3292244478685156, + "grad_norm": 2.7155838012695312, + "learning_rate": 3.121431290445375e-06, + "loss": 1.4701, + "step": 1294 + }, + { + "epoch": 1.3302516692347202, + "grad_norm": 2.5513222217559814, + "learning_rate": 3.1195279786829085e-06, + "loss": 1.5514, + "step": 1295 + }, + { + "epoch": 1.3312788906009245, + "grad_norm": 2.7775397300720215, + "learning_rate": 3.117624666920442e-06, + "loss": 1.611, + "step": 1296 + }, + { + "epoch": 1.332306111967129, + "grad_norm": 2.693214178085327, + "learning_rate": 3.115721355157975e-06, + "loss": 1.497, + "step": 1297 + }, + { + "epoch": 1.3333333333333333, + "grad_norm": 2.6052982807159424, + "learning_rate": 3.1138180433955085e-06, + "loss": 1.4911, + "step": 1298 + }, + { + "epoch": 1.3343605546995376, + "grad_norm": 2.578404426574707, + "learning_rate": 3.1119147316330415e-06, + "loss": 1.4497, + "step": 1299 + }, + { + "epoch": 1.3353877760657422, + "grad_norm": 2.588397264480591, + "learning_rate": 3.110011419870575e-06, + "loss": 1.5339, + "step": 1300 + }, + { + "epoch": 1.3353877760657422, + "eval_loss": 1.8876255750656128, + "eval_runtime": 48.2305, + "eval_samples_per_second": 20.734, + "eval_steps_per_second": 0.871, + "step": 1300 + }, + { + "epoch": 1.3364149974319466, + "grad_norm": 2.835510730743408, + "learning_rate": 3.1081081081081082e-06, + "loss": 1.5649, + "step": 1301 + }, + { + "epoch": 1.337442218798151, + "grad_norm": 2.6646556854248047, + "learning_rate": 3.1062047963456416e-06, + "loss": 1.4969, + "step": 1302 + }, + { + "epoch": 1.3384694401643555, + "grad_norm": 2.6960718631744385, + "learning_rate": 3.104301484583175e-06, + "loss": 1.5332, + "step": 1303 + }, + { + "epoch": 1.33949666153056, + "grad_norm": 2.8116049766540527, + "learning_rate": 3.1023981728207083e-06, + "loss": 1.5766, + "step": 1304 + }, + { + "epoch": 1.3405238828967643, + "grad_norm": 2.7282750606536865, + "learning_rate": 3.1004948610582412e-06, + "loss": 1.551, + "step": 1305 + }, + { + "epoch": 1.3415511042629686, + "grad_norm": 2.6903767585754395, + "learning_rate": 3.0985915492957746e-06, + "loss": 1.5053, + "step": 1306 + }, + { + "epoch": 1.342578325629173, + "grad_norm": 2.525789976119995, + "learning_rate": 3.096688237533308e-06, + "loss": 1.4969, + "step": 1307 + }, + { + "epoch": 1.3436055469953776, + "grad_norm": 2.7578322887420654, + "learning_rate": 3.0947849257708418e-06, + "loss": 1.4701, + "step": 1308 + }, + { + "epoch": 1.344632768361582, + "grad_norm": 2.714743137359619, + "learning_rate": 3.092881614008375e-06, + "loss": 1.5628, + "step": 1309 + }, + { + "epoch": 1.3456599897277863, + "grad_norm": 2.5726380348205566, + "learning_rate": 3.0909783022459085e-06, + "loss": 1.5146, + "step": 1310 + }, + { + "epoch": 1.3456599897277863, + "eval_loss": 1.8887813091278076, + "eval_runtime": 44.978, + "eval_samples_per_second": 22.233, + "eval_steps_per_second": 0.934, + "step": 1310 + }, + { + "epoch": 1.3466872110939907, + "grad_norm": 2.82814621925354, + "learning_rate": 3.0890749904834414e-06, + "loss": 1.5684, + "step": 1311 + }, + { + "epoch": 1.347714432460195, + "grad_norm": 2.705440044403076, + "learning_rate": 3.0871716787209748e-06, + "loss": 1.5003, + "step": 1312 + }, + { + "epoch": 1.3487416538263997, + "grad_norm": 2.647636890411377, + "learning_rate": 3.085268366958508e-06, + "loss": 1.5528, + "step": 1313 + }, + { + "epoch": 1.349768875192604, + "grad_norm": 2.954592227935791, + "learning_rate": 3.0833650551960415e-06, + "loss": 1.5274, + "step": 1314 + }, + { + "epoch": 1.3507960965588084, + "grad_norm": 2.984116315841675, + "learning_rate": 3.081461743433575e-06, + "loss": 1.5411, + "step": 1315 + }, + { + "epoch": 1.3518233179250128, + "grad_norm": 2.5633351802825928, + "learning_rate": 3.079558431671108e-06, + "loss": 1.4853, + "step": 1316 + }, + { + "epoch": 1.3528505392912171, + "grad_norm": 2.8512840270996094, + "learning_rate": 3.077655119908641e-06, + "loss": 1.5117, + "step": 1317 + }, + { + "epoch": 1.3538777606574217, + "grad_norm": 2.709902286529541, + "learning_rate": 3.0757518081461745e-06, + "loss": 1.4716, + "step": 1318 + }, + { + "epoch": 1.354904982023626, + "grad_norm": 2.6622109413146973, + "learning_rate": 3.073848496383708e-06, + "loss": 1.5098, + "step": 1319 + }, + { + "epoch": 1.3559322033898304, + "grad_norm": 2.681844472885132, + "learning_rate": 3.0719451846212413e-06, + "loss": 1.5245, + "step": 1320 + }, + { + "epoch": 1.3559322033898304, + "eval_loss": 1.8935153484344482, + "eval_runtime": 51.161, + "eval_samples_per_second": 19.546, + "eval_steps_per_second": 0.821, + "step": 1320 + }, + { + "epoch": 1.356959424756035, + "grad_norm": 2.816667318344116, + "learning_rate": 3.0700418728587746e-06, + "loss": 1.484, + "step": 1321 + }, + { + "epoch": 1.3579866461222394, + "grad_norm": 2.748138427734375, + "learning_rate": 3.0681385610963076e-06, + "loss": 1.5235, + "step": 1322 + }, + { + "epoch": 1.3590138674884438, + "grad_norm": 2.746454954147339, + "learning_rate": 3.066235249333841e-06, + "loss": 1.4929, + "step": 1323 + }, + { + "epoch": 1.3600410888546481, + "grad_norm": 2.8291871547698975, + "learning_rate": 3.0643319375713743e-06, + "loss": 1.5381, + "step": 1324 + }, + { + "epoch": 1.3610683102208525, + "grad_norm": 2.8472836017608643, + "learning_rate": 3.0624286258089076e-06, + "loss": 1.5664, + "step": 1325 + }, + { + "epoch": 1.362095531587057, + "grad_norm": 2.7493903636932373, + "learning_rate": 3.060525314046441e-06, + "loss": 1.5088, + "step": 1326 + }, + { + "epoch": 1.3631227529532615, + "grad_norm": 2.6563873291015625, + "learning_rate": 3.058622002283974e-06, + "loss": 1.4899, + "step": 1327 + }, + { + "epoch": 1.3641499743194658, + "grad_norm": 2.8158602714538574, + "learning_rate": 3.0567186905215077e-06, + "loss": 1.546, + "step": 1328 + }, + { + "epoch": 1.3651771956856702, + "grad_norm": 2.722679615020752, + "learning_rate": 3.054815378759041e-06, + "loss": 1.5572, + "step": 1329 + }, + { + "epoch": 1.3662044170518746, + "grad_norm": 2.572859525680542, + "learning_rate": 3.0529120669965745e-06, + "loss": 1.5441, + "step": 1330 + }, + { + "epoch": 1.3662044170518746, + "eval_loss": 1.8944634199142456, + "eval_runtime": 49.7122, + "eval_samples_per_second": 20.116, + "eval_steps_per_second": 0.845, + "step": 1330 + }, + { + "epoch": 1.3672316384180792, + "grad_norm": 2.6468751430511475, + "learning_rate": 3.051008755234108e-06, + "loss": 1.5507, + "step": 1331 + }, + { + "epoch": 1.3682588597842835, + "grad_norm": 2.618972063064575, + "learning_rate": 3.049105443471641e-06, + "loss": 1.4799, + "step": 1332 + }, + { + "epoch": 1.3692860811504879, + "grad_norm": 2.808328628540039, + "learning_rate": 3.0472021317091745e-06, + "loss": 1.4917, + "step": 1333 + }, + { + "epoch": 1.3703133025166925, + "grad_norm": 2.7831826210021973, + "learning_rate": 3.0452988199467075e-06, + "loss": 1.6242, + "step": 1334 + }, + { + "epoch": 1.3713405238828968, + "grad_norm": 2.6808860301971436, + "learning_rate": 3.043395508184241e-06, + "loss": 1.5455, + "step": 1335 + }, + { + "epoch": 1.3723677452491012, + "grad_norm": 2.5987675189971924, + "learning_rate": 3.041492196421774e-06, + "loss": 1.5406, + "step": 1336 + }, + { + "epoch": 1.3733949666153056, + "grad_norm": 2.6476593017578125, + "learning_rate": 3.0395888846593076e-06, + "loss": 1.487, + "step": 1337 + }, + { + "epoch": 1.37442218798151, + "grad_norm": 2.724715232849121, + "learning_rate": 3.037685572896841e-06, + "loss": 1.5316, + "step": 1338 + }, + { + "epoch": 1.3754494093477145, + "grad_norm": 2.6245994567871094, + "learning_rate": 3.035782261134374e-06, + "loss": 1.5137, + "step": 1339 + }, + { + "epoch": 1.376476630713919, + "grad_norm": 2.676206588745117, + "learning_rate": 3.0338789493719072e-06, + "loss": 1.5321, + "step": 1340 + }, + { + "epoch": 1.376476630713919, + "eval_loss": 1.908184289932251, + "eval_runtime": 49.2092, + "eval_samples_per_second": 20.321, + "eval_steps_per_second": 0.853, + "step": 1340 + }, + { + "epoch": 1.3775038520801233, + "grad_norm": 2.9669554233551025, + "learning_rate": 3.0319756376094406e-06, + "loss": 1.5349, + "step": 1341 + }, + { + "epoch": 1.3785310734463276, + "grad_norm": 2.6755905151367188, + "learning_rate": 3.030072325846974e-06, + "loss": 1.536, + "step": 1342 + }, + { + "epoch": 1.379558294812532, + "grad_norm": 2.5073540210723877, + "learning_rate": 3.0281690140845073e-06, + "loss": 1.5167, + "step": 1343 + }, + { + "epoch": 1.3805855161787366, + "grad_norm": 2.520477533340454, + "learning_rate": 3.0262657023220403e-06, + "loss": 1.4827, + "step": 1344 + }, + { + "epoch": 1.381612737544941, + "grad_norm": 2.6202239990234375, + "learning_rate": 3.0243623905595736e-06, + "loss": 1.5271, + "step": 1345 + }, + { + "epoch": 1.3826399589111453, + "grad_norm": 2.667724609375, + "learning_rate": 3.022459078797107e-06, + "loss": 1.537, + "step": 1346 + }, + { + "epoch": 1.3836671802773497, + "grad_norm": 2.701108455657959, + "learning_rate": 3.0205557670346404e-06, + "loss": 1.53, + "step": 1347 + }, + { + "epoch": 1.384694401643554, + "grad_norm": 2.6228628158569336, + "learning_rate": 3.018652455272174e-06, + "loss": 1.5297, + "step": 1348 + }, + { + "epoch": 1.3857216230097587, + "grad_norm": 2.6589505672454834, + "learning_rate": 3.0167491435097075e-06, + "loss": 1.481, + "step": 1349 + }, + { + "epoch": 1.386748844375963, + "grad_norm": 2.805138349533081, + "learning_rate": 3.014845831747241e-06, + "loss": 1.6176, + "step": 1350 + }, + { + "epoch": 1.386748844375963, + "eval_loss": 1.9109174013137817, + "eval_runtime": 44.6864, + "eval_samples_per_second": 22.378, + "eval_steps_per_second": 0.94, + "step": 1350 + }, + { + "epoch": 1.3877760657421674, + "grad_norm": 2.6587769985198975, + "learning_rate": 3.012942519984774e-06, + "loss": 1.5188, + "step": 1351 + }, + { + "epoch": 1.388803287108372, + "grad_norm": 2.7695798873901367, + "learning_rate": 3.011039208222307e-06, + "loss": 1.4913, + "step": 1352 + }, + { + "epoch": 1.3898305084745763, + "grad_norm": 2.7030696868896484, + "learning_rate": 3.0091358964598405e-06, + "loss": 1.4976, + "step": 1353 + }, + { + "epoch": 1.3908577298407807, + "grad_norm": 2.63368558883667, + "learning_rate": 3.007232584697374e-06, + "loss": 1.4575, + "step": 1354 + }, + { + "epoch": 1.391884951206985, + "grad_norm": 2.6904921531677246, + "learning_rate": 3.0053292729349073e-06, + "loss": 1.5067, + "step": 1355 + }, + { + "epoch": 1.3929121725731894, + "grad_norm": 2.7085046768188477, + "learning_rate": 3.00342596117244e-06, + "loss": 1.5215, + "step": 1356 + }, + { + "epoch": 1.393939393939394, + "grad_norm": 2.744744062423706, + "learning_rate": 3.0015226494099736e-06, + "loss": 1.5213, + "step": 1357 + }, + { + "epoch": 1.3949666153055984, + "grad_norm": 2.8159263134002686, + "learning_rate": 2.999619337647507e-06, + "loss": 1.5091, + "step": 1358 + }, + { + "epoch": 1.3959938366718028, + "grad_norm": 2.7380170822143555, + "learning_rate": 2.9977160258850403e-06, + "loss": 1.4863, + "step": 1359 + }, + { + "epoch": 1.3970210580380071, + "grad_norm": 2.9817709922790527, + "learning_rate": 2.9958127141225736e-06, + "loss": 1.4977, + "step": 1360 + }, + { + "epoch": 1.3970210580380071, + "eval_loss": 1.9042565822601318, + "eval_runtime": 45.13, + "eval_samples_per_second": 22.158, + "eval_steps_per_second": 0.931, + "step": 1360 + }, + { + "epoch": 1.3980482794042115, + "grad_norm": 2.7616889476776123, + "learning_rate": 2.9939094023601066e-06, + "loss": 1.4817, + "step": 1361 + }, + { + "epoch": 1.399075500770416, + "grad_norm": 2.695251226425171, + "learning_rate": 2.99200609059764e-06, + "loss": 1.5512, + "step": 1362 + }, + { + "epoch": 1.4001027221366205, + "grad_norm": 2.645728349685669, + "learning_rate": 2.9901027788351733e-06, + "loss": 1.5232, + "step": 1363 + }, + { + "epoch": 1.4011299435028248, + "grad_norm": 2.5545313358306885, + "learning_rate": 2.9881994670727067e-06, + "loss": 1.4626, + "step": 1364 + }, + { + "epoch": 1.4021571648690292, + "grad_norm": 2.6554672718048096, + "learning_rate": 2.98629615531024e-06, + "loss": 1.4813, + "step": 1365 + }, + { + "epoch": 1.4031843862352336, + "grad_norm": 2.543102979660034, + "learning_rate": 2.984392843547773e-06, + "loss": 1.4954, + "step": 1366 + }, + { + "epoch": 1.4042116076014381, + "grad_norm": 2.5635626316070557, + "learning_rate": 2.9824895317853063e-06, + "loss": 1.4858, + "step": 1367 + }, + { + "epoch": 1.4052388289676425, + "grad_norm": 2.60005259513855, + "learning_rate": 2.98058622002284e-06, + "loss": 1.4991, + "step": 1368 + }, + { + "epoch": 1.4062660503338469, + "grad_norm": 2.664973020553589, + "learning_rate": 2.9786829082603735e-06, + "loss": 1.5083, + "step": 1369 + }, + { + "epoch": 1.4072932717000515, + "grad_norm": 2.642303466796875, + "learning_rate": 2.976779596497907e-06, + "loss": 1.5636, + "step": 1370 + }, + { + "epoch": 1.4072932717000515, + "eval_loss": 1.9043673276901245, + "eval_runtime": 45.8156, + "eval_samples_per_second": 21.827, + "eval_steps_per_second": 0.917, + "step": 1370 + }, + { + "epoch": 1.4083204930662558, + "grad_norm": 2.721719264984131, + "learning_rate": 2.97487628473544e-06, + "loss": 1.5627, + "step": 1371 + }, + { + "epoch": 1.4093477144324602, + "grad_norm": 2.665929079055786, + "learning_rate": 2.9729729729729736e-06, + "loss": 1.5684, + "step": 1372 + }, + { + "epoch": 1.4103749357986646, + "grad_norm": 2.672046661376953, + "learning_rate": 2.9710696612105065e-06, + "loss": 1.518, + "step": 1373 + }, + { + "epoch": 1.411402157164869, + "grad_norm": 2.675283432006836, + "learning_rate": 2.96916634944804e-06, + "loss": 1.5604, + "step": 1374 + }, + { + "epoch": 1.4124293785310735, + "grad_norm": 2.6286461353302, + "learning_rate": 2.9672630376855732e-06, + "loss": 1.5003, + "step": 1375 + }, + { + "epoch": 1.413456599897278, + "grad_norm": 2.6903879642486572, + "learning_rate": 2.9653597259231066e-06, + "loss": 1.5601, + "step": 1376 + }, + { + "epoch": 1.4144838212634823, + "grad_norm": 2.7762627601623535, + "learning_rate": 2.96345641416064e-06, + "loss": 1.5326, + "step": 1377 + }, + { + "epoch": 1.4155110426296866, + "grad_norm": 2.758028030395508, + "learning_rate": 2.961553102398173e-06, + "loss": 1.4837, + "step": 1378 + }, + { + "epoch": 1.416538263995891, + "grad_norm": 2.769880771636963, + "learning_rate": 2.9596497906357063e-06, + "loss": 1.4122, + "step": 1379 + }, + { + "epoch": 1.4175654853620956, + "grad_norm": 2.815645217895508, + "learning_rate": 2.9577464788732396e-06, + "loss": 1.555, + "step": 1380 + }, + { + "epoch": 1.4175654853620956, + "eval_loss": 1.9081141948699951, + "eval_runtime": 44.267, + "eval_samples_per_second": 22.59, + "eval_steps_per_second": 0.949, + "step": 1380 + }, + { + "epoch": 1.4185927067283, + "grad_norm": 2.8118088245391846, + "learning_rate": 2.955843167110773e-06, + "loss": 1.5708, + "step": 1381 + }, + { + "epoch": 1.4196199280945043, + "grad_norm": 2.620800495147705, + "learning_rate": 2.9539398553483064e-06, + "loss": 1.4901, + "step": 1382 + }, + { + "epoch": 1.420647149460709, + "grad_norm": 2.6867423057556152, + "learning_rate": 2.9520365435858393e-06, + "loss": 1.4964, + "step": 1383 + }, + { + "epoch": 1.4216743708269133, + "grad_norm": 2.569495916366577, + "learning_rate": 2.9501332318233727e-06, + "loss": 1.5103, + "step": 1384 + }, + { + "epoch": 1.4227015921931176, + "grad_norm": 2.6389119625091553, + "learning_rate": 2.948229920060906e-06, + "loss": 1.5946, + "step": 1385 + }, + { + "epoch": 1.423728813559322, + "grad_norm": 2.767986536026001, + "learning_rate": 2.9463266082984394e-06, + "loss": 1.5296, + "step": 1386 + }, + { + "epoch": 1.4247560349255264, + "grad_norm": 2.5122997760772705, + "learning_rate": 2.9444232965359727e-06, + "loss": 1.4941, + "step": 1387 + }, + { + "epoch": 1.425783256291731, + "grad_norm": 2.7231523990631104, + "learning_rate": 2.9425199847735057e-06, + "loss": 1.5184, + "step": 1388 + }, + { + "epoch": 1.4268104776579353, + "grad_norm": 2.7313649654388428, + "learning_rate": 2.94061667301104e-06, + "loss": 1.5368, + "step": 1389 + }, + { + "epoch": 1.4278376990241397, + "grad_norm": 2.656062602996826, + "learning_rate": 2.938713361248573e-06, + "loss": 1.4958, + "step": 1390 + }, + { + "epoch": 1.4278376990241397, + "eval_loss": 1.9070981740951538, + "eval_runtime": 43.6226, + "eval_samples_per_second": 22.924, + "eval_steps_per_second": 0.963, + "step": 1390 + }, + { + "epoch": 1.428864920390344, + "grad_norm": 2.6621580123901367, + "learning_rate": 2.936810049486106e-06, + "loss": 1.4315, + "step": 1391 + }, + { + "epoch": 1.4298921417565484, + "grad_norm": 2.7683629989624023, + "learning_rate": 2.9349067377236396e-06, + "loss": 1.5042, + "step": 1392 + }, + { + "epoch": 1.430919363122753, + "grad_norm": 2.695810317993164, + "learning_rate": 2.933003425961173e-06, + "loss": 1.5174, + "step": 1393 + }, + { + "epoch": 1.4319465844889574, + "grad_norm": 2.686978816986084, + "learning_rate": 2.9311001141987063e-06, + "loss": 1.4706, + "step": 1394 + }, + { + "epoch": 1.4329738058551618, + "grad_norm": 2.799234628677368, + "learning_rate": 2.9291968024362392e-06, + "loss": 1.4826, + "step": 1395 + }, + { + "epoch": 1.4340010272213661, + "grad_norm": 2.8211469650268555, + "learning_rate": 2.9272934906737726e-06, + "loss": 1.488, + "step": 1396 + }, + { + "epoch": 1.4350282485875705, + "grad_norm": 2.6225461959838867, + "learning_rate": 2.925390178911306e-06, + "loss": 1.4699, + "step": 1397 + }, + { + "epoch": 1.436055469953775, + "grad_norm": 2.6853697299957275, + "learning_rate": 2.9234868671488393e-06, + "loss": 1.4787, + "step": 1398 + }, + { + "epoch": 1.4370826913199795, + "grad_norm": 2.784473419189453, + "learning_rate": 2.9215835553863727e-06, + "loss": 1.5043, + "step": 1399 + }, + { + "epoch": 1.4381099126861838, + "grad_norm": 2.706911087036133, + "learning_rate": 2.9196802436239056e-06, + "loss": 1.512, + "step": 1400 + }, + { + "epoch": 1.4381099126861838, + "eval_loss": 1.9136111736297607, + "eval_runtime": 43.5675, + "eval_samples_per_second": 22.953, + "eval_steps_per_second": 0.964, + "step": 1400 + }, + { + "epoch": 1.4391371340523884, + "grad_norm": 2.6579840183258057, + "learning_rate": 2.917776931861439e-06, + "loss": 1.5073, + "step": 1401 + }, + { + "epoch": 1.4401643554185928, + "grad_norm": 2.7359421253204346, + "learning_rate": 2.9158736200989723e-06, + "loss": 1.4819, + "step": 1402 + }, + { + "epoch": 1.4411915767847971, + "grad_norm": 2.604591131210327, + "learning_rate": 2.9139703083365057e-06, + "loss": 1.4758, + "step": 1403 + }, + { + "epoch": 1.4422187981510015, + "grad_norm": 2.7267162799835205, + "learning_rate": 2.912066996574039e-06, + "loss": 1.4835, + "step": 1404 + }, + { + "epoch": 1.4432460195172059, + "grad_norm": 2.844331741333008, + "learning_rate": 2.910163684811572e-06, + "loss": 1.5096, + "step": 1405 + }, + { + "epoch": 1.4442732408834105, + "grad_norm": 2.6915996074676514, + "learning_rate": 2.9082603730491054e-06, + "loss": 1.4536, + "step": 1406 + }, + { + "epoch": 1.4453004622496148, + "grad_norm": 2.9222843647003174, + "learning_rate": 2.9063570612866387e-06, + "loss": 1.5769, + "step": 1407 + }, + { + "epoch": 1.4463276836158192, + "grad_norm": 2.667371988296509, + "learning_rate": 2.904453749524172e-06, + "loss": 1.533, + "step": 1408 + }, + { + "epoch": 1.4473549049820236, + "grad_norm": 2.643632411956787, + "learning_rate": 2.902550437761706e-06, + "loss": 1.4666, + "step": 1409 + }, + { + "epoch": 1.448382126348228, + "grad_norm": 2.8633203506469727, + "learning_rate": 2.9006471259992392e-06, + "loss": 1.4826, + "step": 1410 + }, + { + "epoch": 1.448382126348228, + "eval_loss": 1.9127975702285767, + "eval_runtime": 42.2168, + "eval_samples_per_second": 23.687, + "eval_steps_per_second": 0.995, + "step": 1410 + }, + { + "epoch": 1.4494093477144325, + "grad_norm": 2.7763702869415283, + "learning_rate": 2.8987438142367726e-06, + "loss": 1.4969, + "step": 1411 + }, + { + "epoch": 1.450436569080637, + "grad_norm": 2.598337411880493, + "learning_rate": 2.8968405024743055e-06, + "loss": 1.4993, + "step": 1412 + }, + { + "epoch": 1.4514637904468413, + "grad_norm": 2.717944622039795, + "learning_rate": 2.894937190711839e-06, + "loss": 1.5479, + "step": 1413 + }, + { + "epoch": 1.4524910118130456, + "grad_norm": 2.5288076400756836, + "learning_rate": 2.8930338789493723e-06, + "loss": 1.4564, + "step": 1414 + }, + { + "epoch": 1.45351823317925, + "grad_norm": 2.6042704582214355, + "learning_rate": 2.8911305671869056e-06, + "loss": 1.5267, + "step": 1415 + }, + { + "epoch": 1.4545454545454546, + "grad_norm": 2.5768539905548096, + "learning_rate": 2.889227255424439e-06, + "loss": 1.4958, + "step": 1416 + }, + { + "epoch": 1.455572675911659, + "grad_norm": 2.62872576713562, + "learning_rate": 2.887323943661972e-06, + "loss": 1.4407, + "step": 1417 + }, + { + "epoch": 1.4565998972778633, + "grad_norm": 2.8656342029571533, + "learning_rate": 2.8854206318995053e-06, + "loss": 1.4944, + "step": 1418 + }, + { + "epoch": 1.457627118644068, + "grad_norm": 2.7083373069763184, + "learning_rate": 2.8835173201370386e-06, + "loss": 1.474, + "step": 1419 + }, + { + "epoch": 1.4586543400102723, + "grad_norm": 2.712958812713623, + "learning_rate": 2.881614008374572e-06, + "loss": 1.4982, + "step": 1420 + }, + { + "epoch": 1.4586543400102723, + "eval_loss": 1.9154274463653564, + "eval_runtime": 45.8592, + "eval_samples_per_second": 21.806, + "eval_steps_per_second": 0.916, + "step": 1420 + }, + { + "epoch": 1.4596815613764766, + "grad_norm": 3.0084564685821533, + "learning_rate": 2.8797106966121054e-06, + "loss": 1.5505, + "step": 1421 + }, + { + "epoch": 1.460708782742681, + "grad_norm": 2.793546438217163, + "learning_rate": 2.8778073848496383e-06, + "loss": 1.5203, + "step": 1422 + }, + { + "epoch": 1.4617360041088854, + "grad_norm": 2.5944628715515137, + "learning_rate": 2.8759040730871717e-06, + "loss": 1.5016, + "step": 1423 + }, + { + "epoch": 1.46276322547509, + "grad_norm": 2.7558817863464355, + "learning_rate": 2.874000761324705e-06, + "loss": 1.5273, + "step": 1424 + }, + { + "epoch": 1.4637904468412943, + "grad_norm": 2.582310676574707, + "learning_rate": 2.8720974495622384e-06, + "loss": 1.4862, + "step": 1425 + }, + { + "epoch": 1.4648176682074987, + "grad_norm": 2.639944314956665, + "learning_rate": 2.8701941377997718e-06, + "loss": 1.5276, + "step": 1426 + }, + { + "epoch": 1.465844889573703, + "grad_norm": 2.613696575164795, + "learning_rate": 2.8682908260373047e-06, + "loss": 1.5061, + "step": 1427 + }, + { + "epoch": 1.4668721109399074, + "grad_norm": 2.7124717235565186, + "learning_rate": 2.866387514274838e-06, + "loss": 1.5517, + "step": 1428 + }, + { + "epoch": 1.467899332306112, + "grad_norm": 2.7124717235565186, + "learning_rate": 2.866387514274838e-06, + "loss": 1.5288, + "step": 1429 + }, + { + "epoch": 1.4689265536723164, + "grad_norm": 2.722407102584839, + "learning_rate": 2.864484202512372e-06, + "loss": 1.5256, + "step": 1430 + }, + { + "epoch": 1.4689265536723164, + "eval_loss": 1.900185227394104, + "eval_runtime": 47.7724, + "eval_samples_per_second": 20.933, + "eval_steps_per_second": 0.879, + "step": 1430 + }, + { + "epoch": 1.4699537750385208, + "grad_norm": 2.690464973449707, + "learning_rate": 2.8625808907499052e-06, + "loss": 1.4841, + "step": 1431 + }, + { + "epoch": 1.4709809964047253, + "grad_norm": 2.766082525253296, + "learning_rate": 2.8606775789874386e-06, + "loss": 1.5171, + "step": 1432 + }, + { + "epoch": 1.4720082177709297, + "grad_norm": 2.6420655250549316, + "learning_rate": 2.858774267224972e-06, + "loss": 1.4871, + "step": 1433 + }, + { + "epoch": 1.473035439137134, + "grad_norm": 2.687546730041504, + "learning_rate": 2.8568709554625053e-06, + "loss": 1.5116, + "step": 1434 + }, + { + "epoch": 1.4740626605033385, + "grad_norm": 2.6613612174987793, + "learning_rate": 2.8549676437000382e-06, + "loss": 1.5599, + "step": 1435 + }, + { + "epoch": 1.4750898818695428, + "grad_norm": 2.7249948978424072, + "learning_rate": 2.8530643319375716e-06, + "loss": 1.5342, + "step": 1436 + }, + { + "epoch": 1.4761171032357474, + "grad_norm": 2.768580913543701, + "learning_rate": 2.851161020175105e-06, + "loss": 1.5022, + "step": 1437 + }, + { + "epoch": 1.4771443246019518, + "grad_norm": 2.6677844524383545, + "learning_rate": 2.8492577084126383e-06, + "loss": 1.4712, + "step": 1438 + }, + { + "epoch": 1.4781715459681561, + "grad_norm": 2.693856716156006, + "learning_rate": 2.8473543966501717e-06, + "loss": 1.5163, + "step": 1439 + }, + { + "epoch": 1.4791987673343605, + "grad_norm": 2.667125701904297, + "learning_rate": 2.8454510848877046e-06, + "loss": 1.5601, + "step": 1440 + }, + { + "epoch": 1.4791987673343605, + "eval_loss": 1.8993381261825562, + "eval_runtime": 47.2966, + "eval_samples_per_second": 21.143, + "eval_steps_per_second": 0.888, + "step": 1440 + }, + { + "epoch": 1.4802259887005649, + "grad_norm": 2.6483078002929688, + "learning_rate": 2.843547773125238e-06, + "loss": 1.5042, + "step": 1441 + }, + { + "epoch": 1.4812532100667695, + "grad_norm": 2.6776845455169678, + "learning_rate": 2.8416444613627714e-06, + "loss": 1.5096, + "step": 1442 + }, + { + "epoch": 1.4822804314329738, + "grad_norm": 2.7866480350494385, + "learning_rate": 2.8397411496003047e-06, + "loss": 1.4684, + "step": 1443 + }, + { + "epoch": 1.4833076527991782, + "grad_norm": 2.844266176223755, + "learning_rate": 2.837837837837838e-06, + "loss": 1.499, + "step": 1444 + }, + { + "epoch": 1.4843348741653826, + "grad_norm": 2.5618107318878174, + "learning_rate": 2.835934526075371e-06, + "loss": 1.5212, + "step": 1445 + }, + { + "epoch": 1.485362095531587, + "grad_norm": 2.6264190673828125, + "learning_rate": 2.8340312143129044e-06, + "loss": 1.4763, + "step": 1446 + }, + { + "epoch": 1.4863893168977915, + "grad_norm": 2.68026065826416, + "learning_rate": 2.8321279025504377e-06, + "loss": 1.4572, + "step": 1447 + }, + { + "epoch": 1.487416538263996, + "grad_norm": 2.7439587116241455, + "learning_rate": 2.830224590787971e-06, + "loss": 1.5778, + "step": 1448 + }, + { + "epoch": 1.4884437596302003, + "grad_norm": 2.708627700805664, + "learning_rate": 2.8283212790255045e-06, + "loss": 1.5177, + "step": 1449 + }, + { + "epoch": 1.4894709809964048, + "grad_norm": 2.5931451320648193, + "learning_rate": 2.8264179672630383e-06, + "loss": 1.5499, + "step": 1450 + }, + { + "epoch": 1.4894709809964048, + "eval_loss": 1.8863205909729004, + "eval_runtime": 45.8354, + "eval_samples_per_second": 21.817, + "eval_steps_per_second": 0.916, + "step": 1450 + }, + { + "epoch": 1.4904982023626092, + "grad_norm": 2.6775002479553223, + "learning_rate": 2.8245146555005716e-06, + "loss": 1.4987, + "step": 1451 + }, + { + "epoch": 1.4915254237288136, + "grad_norm": 2.568394660949707, + "learning_rate": 2.8226113437381046e-06, + "loss": 1.449, + "step": 1452 + }, + { + "epoch": 1.492552645095018, + "grad_norm": 2.6354329586029053, + "learning_rate": 2.820708031975638e-06, + "loss": 1.5018, + "step": 1453 + }, + { + "epoch": 1.4935798664612223, + "grad_norm": 2.731614112854004, + "learning_rate": 2.8188047202131713e-06, + "loss": 1.5337, + "step": 1454 + }, + { + "epoch": 1.494607087827427, + "grad_norm": 2.662299633026123, + "learning_rate": 2.8169014084507046e-06, + "loss": 1.5449, + "step": 1455 + }, + { + "epoch": 1.4956343091936313, + "grad_norm": 2.6484053134918213, + "learning_rate": 2.814998096688238e-06, + "loss": 1.5048, + "step": 1456 + }, + { + "epoch": 1.4966615305598356, + "grad_norm": 2.5573649406433105, + "learning_rate": 2.813094784925771e-06, + "loss": 1.4417, + "step": 1457 + }, + { + "epoch": 1.49768875192604, + "grad_norm": 2.6440303325653076, + "learning_rate": 2.8111914731633043e-06, + "loss": 1.5545, + "step": 1458 + }, + { + "epoch": 1.4987159732922444, + "grad_norm": 2.6014726161956787, + "learning_rate": 2.8092881614008377e-06, + "loss": 1.4995, + "step": 1459 + }, + { + "epoch": 1.499743194658449, + "grad_norm": 2.8325798511505127, + "learning_rate": 2.807384849638371e-06, + "loss": 1.4762, + "step": 1460 + }, + { + "epoch": 1.499743194658449, + "eval_loss": 1.8842575550079346, + "eval_runtime": 46.058, + "eval_samples_per_second": 21.712, + "eval_steps_per_second": 0.912, + "step": 1460 + }, + { + "epoch": 1.5007704160246533, + "grad_norm": 2.629534959793091, + "learning_rate": 2.8054815378759044e-06, + "loss": 1.5235, + "step": 1461 + }, + { + "epoch": 1.5017976373908577, + "grad_norm": 2.723256826400757, + "learning_rate": 2.8035782261134378e-06, + "loss": 1.4962, + "step": 1462 + }, + { + "epoch": 1.5028248587570623, + "grad_norm": 2.7497754096984863, + "learning_rate": 2.8016749143509707e-06, + "loss": 1.4931, + "step": 1463 + }, + { + "epoch": 1.5038520801232664, + "grad_norm": 2.695408821105957, + "learning_rate": 2.799771602588504e-06, + "loss": 1.4921, + "step": 1464 + }, + { + "epoch": 1.504879301489471, + "grad_norm": 2.6144590377807617, + "learning_rate": 2.7978682908260374e-06, + "loss": 1.5191, + "step": 1465 + }, + { + "epoch": 1.5059065228556754, + "grad_norm": 2.7783288955688477, + "learning_rate": 2.795964979063571e-06, + "loss": 1.5416, + "step": 1466 + }, + { + "epoch": 1.5069337442218798, + "grad_norm": 2.7083182334899902, + "learning_rate": 2.794061667301104e-06, + "loss": 1.5126, + "step": 1467 + }, + { + "epoch": 1.5079609655880843, + "grad_norm": 2.7532167434692383, + "learning_rate": 2.792158355538637e-06, + "loss": 1.5244, + "step": 1468 + }, + { + "epoch": 1.5089881869542885, + "grad_norm": 2.6125259399414062, + "learning_rate": 2.7902550437761705e-06, + "loss": 1.5194, + "step": 1469 + }, + { + "epoch": 1.510015408320493, + "grad_norm": 2.585392713546753, + "learning_rate": 2.7883517320137042e-06, + "loss": 1.474, + "step": 1470 + }, + { + "epoch": 1.510015408320493, + "eval_loss": 1.8812134265899658, + "eval_runtime": 46.7392, + "eval_samples_per_second": 21.395, + "eval_steps_per_second": 0.899, + "step": 1470 + }, + { + "epoch": 1.5110426296866974, + "grad_norm": 2.586333751678467, + "learning_rate": 2.7864484202512376e-06, + "loss": 1.4666, + "step": 1471 + }, + { + "epoch": 1.5120698510529018, + "grad_norm": 2.7523810863494873, + "learning_rate": 2.784545108488771e-06, + "loss": 1.5113, + "step": 1472 + }, + { + "epoch": 1.5130970724191064, + "grad_norm": 2.688206434249878, + "learning_rate": 2.7826417967263043e-06, + "loss": 1.5049, + "step": 1473 + }, + { + "epoch": 1.5141242937853108, + "grad_norm": 2.642054319381714, + "learning_rate": 2.7807384849638373e-06, + "loss": 1.5534, + "step": 1474 + }, + { + "epoch": 1.5151515151515151, + "grad_norm": 2.694291591644287, + "learning_rate": 2.7788351732013706e-06, + "loss": 1.5052, + "step": 1475 + }, + { + "epoch": 1.5161787365177197, + "grad_norm": 2.6969082355499268, + "learning_rate": 2.776931861438904e-06, + "loss": 1.491, + "step": 1476 + }, + { + "epoch": 1.5172059578839239, + "grad_norm": 2.6937081813812256, + "learning_rate": 2.7750285496764374e-06, + "loss": 1.5081, + "step": 1477 + }, + { + "epoch": 1.5182331792501285, + "grad_norm": 2.6754133701324463, + "learning_rate": 2.7731252379139707e-06, + "loss": 1.5107, + "step": 1478 + }, + { + "epoch": 1.5192604006163328, + "grad_norm": 2.586491346359253, + "learning_rate": 2.771221926151504e-06, + "loss": 1.4486, + "step": 1479 + }, + { + "epoch": 1.5202876219825372, + "grad_norm": 2.82452130317688, + "learning_rate": 2.769318614389037e-06, + "loss": 1.5089, + "step": 1480 + }, + { + "epoch": 1.5202876219825372, + "eval_loss": 1.879558801651001, + "eval_runtime": 45.9914, + "eval_samples_per_second": 21.743, + "eval_steps_per_second": 0.913, + "step": 1480 + }, + { + "epoch": 1.5213148433487418, + "grad_norm": 2.69734263420105, + "learning_rate": 2.7674153026265704e-06, + "loss": 1.5069, + "step": 1481 + }, + { + "epoch": 1.522342064714946, + "grad_norm": 2.7553112506866455, + "learning_rate": 2.7655119908641037e-06, + "loss": 1.5211, + "step": 1482 + }, + { + "epoch": 1.5233692860811505, + "grad_norm": 2.659170389175415, + "learning_rate": 2.763608679101637e-06, + "loss": 1.4967, + "step": 1483 + }, + { + "epoch": 1.5243965074473549, + "grad_norm": 2.6266300678253174, + "learning_rate": 2.7617053673391705e-06, + "loss": 1.5376, + "step": 1484 + }, + { + "epoch": 1.5254237288135593, + "grad_norm": 2.6549487113952637, + "learning_rate": 2.7598020555767034e-06, + "loss": 1.546, + "step": 1485 + }, + { + "epoch": 1.5264509501797638, + "grad_norm": 2.651841402053833, + "learning_rate": 2.7578987438142368e-06, + "loss": 1.4625, + "step": 1486 + }, + { + "epoch": 1.5274781715459682, + "grad_norm": 2.783480644226074, + "learning_rate": 2.75599543205177e-06, + "loss": 1.5165, + "step": 1487 + }, + { + "epoch": 1.5285053929121726, + "grad_norm": 2.674964189529419, + "learning_rate": 2.7540921202893035e-06, + "loss": 1.539, + "step": 1488 + }, + { + "epoch": 1.529532614278377, + "grad_norm": 2.7219326496124268, + "learning_rate": 2.752188808526837e-06, + "loss": 1.5151, + "step": 1489 + }, + { + "epoch": 1.5305598356445813, + "grad_norm": 2.6149561405181885, + "learning_rate": 2.75028549676437e-06, + "loss": 1.498, + "step": 1490 + }, + { + "epoch": 1.5305598356445813, + "eval_loss": 1.8944125175476074, + "eval_runtime": 45.5669, + "eval_samples_per_second": 21.946, + "eval_steps_per_second": 0.922, + "step": 1490 + }, + { + "epoch": 1.531587057010786, + "grad_norm": 2.6966514587402344, + "learning_rate": 2.7483821850019036e-06, + "loss": 1.5026, + "step": 1491 + }, + { + "epoch": 1.5326142783769903, + "grad_norm": 2.7445616722106934, + "learning_rate": 2.746478873239437e-06, + "loss": 1.493, + "step": 1492 + }, + { + "epoch": 1.5336414997431946, + "grad_norm": 2.8391168117523193, + "learning_rate": 2.7445755614769703e-06, + "loss": 1.4918, + "step": 1493 + }, + { + "epoch": 1.5346687211093992, + "grad_norm": 2.79190731048584, + "learning_rate": 2.7426722497145037e-06, + "loss": 1.4654, + "step": 1494 + }, + { + "epoch": 1.5356959424756034, + "grad_norm": 2.7497293949127197, + "learning_rate": 2.740768937952037e-06, + "loss": 1.4702, + "step": 1495 + }, + { + "epoch": 1.536723163841808, + "grad_norm": 2.7948496341705322, + "learning_rate": 2.7388656261895704e-06, + "loss": 1.4879, + "step": 1496 + }, + { + "epoch": 1.5377503852080123, + "grad_norm": 2.9910364151000977, + "learning_rate": 2.7369623144271033e-06, + "loss": 1.5689, + "step": 1497 + }, + { + "epoch": 1.5387776065742167, + "grad_norm": 2.726814031600952, + "learning_rate": 2.7350590026646367e-06, + "loss": 1.492, + "step": 1498 + }, + { + "epoch": 1.5398048279404213, + "grad_norm": 2.839493751525879, + "learning_rate": 2.73315569090217e-06, + "loss": 1.4539, + "step": 1499 + }, + { + "epoch": 1.5408320493066254, + "grad_norm": 2.776416063308716, + "learning_rate": 2.7312523791397034e-06, + "loss": 1.5509, + "step": 1500 + }, + { + "epoch": 1.5408320493066254, + "eval_loss": 1.9009771347045898, + "eval_runtime": 51.232, + "eval_samples_per_second": 19.519, + "eval_steps_per_second": 0.82, + "step": 1500 + }, + { + "epoch": 1.54185927067283, + "grad_norm": 2.8426146507263184, + "learning_rate": 2.7293490673772368e-06, + "loss": 1.5696, + "step": 1501 + }, + { + "epoch": 1.5428864920390344, + "grad_norm": 2.8076331615448, + "learning_rate": 2.7274457556147697e-06, + "loss": 1.5216, + "step": 1502 + }, + { + "epoch": 1.5439137134052388, + "grad_norm": 2.800166368484497, + "learning_rate": 2.725542443852303e-06, + "loss": 1.5094, + "step": 1503 + }, + { + "epoch": 1.5449409347714433, + "grad_norm": 2.597820281982422, + "learning_rate": 2.7236391320898365e-06, + "loss": 1.5149, + "step": 1504 + }, + { + "epoch": 1.5459681561376477, + "grad_norm": 2.557790517807007, + "learning_rate": 2.72173582032737e-06, + "loss": 1.4817, + "step": 1505 + }, + { + "epoch": 1.546995377503852, + "grad_norm": 2.6866400241851807, + "learning_rate": 2.719832508564903e-06, + "loss": 1.5024, + "step": 1506 + }, + { + "epoch": 1.5480225988700564, + "grad_norm": 2.739360809326172, + "learning_rate": 2.717929196802436e-06, + "loss": 1.5151, + "step": 1507 + }, + { + "epoch": 1.5490498202362608, + "grad_norm": 2.694063186645508, + "learning_rate": 2.7160258850399695e-06, + "loss": 1.4815, + "step": 1508 + }, + { + "epoch": 1.5500770416024654, + "grad_norm": 2.4691152572631836, + "learning_rate": 2.714122573277503e-06, + "loss": 1.4989, + "step": 1509 + }, + { + "epoch": 1.5511042629686698, + "grad_norm": 2.594684362411499, + "learning_rate": 2.712219261515036e-06, + "loss": 1.499, + "step": 1510 + }, + { + "epoch": 1.5511042629686698, + "eval_loss": 1.891904592514038, + "eval_runtime": 44.1399, + "eval_samples_per_second": 22.655, + "eval_steps_per_second": 0.952, + "step": 1510 + }, + { + "epoch": 1.5521314843348741, + "grad_norm": 2.621588706970215, + "learning_rate": 2.71031594975257e-06, + "loss": 1.516, + "step": 1511 + }, + { + "epoch": 1.5531587057010787, + "grad_norm": 2.677280902862549, + "learning_rate": 2.7084126379901034e-06, + "loss": 1.5192, + "step": 1512 + }, + { + "epoch": 1.5541859270672829, + "grad_norm": 2.7408759593963623, + "learning_rate": 2.7065093262276367e-06, + "loss": 1.5377, + "step": 1513 + }, + { + "epoch": 1.5552131484334875, + "grad_norm": 2.622382402420044, + "learning_rate": 2.7046060144651697e-06, + "loss": 1.5437, + "step": 1514 + }, + { + "epoch": 1.5562403697996918, + "grad_norm": 2.8121471405029297, + "learning_rate": 2.702702702702703e-06, + "loss": 1.5698, + "step": 1515 + }, + { + "epoch": 1.5572675911658962, + "grad_norm": 2.5897836685180664, + "learning_rate": 2.7007993909402364e-06, + "loss": 1.4295, + "step": 1516 + }, + { + "epoch": 1.5582948125321008, + "grad_norm": 2.793713092803955, + "learning_rate": 2.6988960791777697e-06, + "loss": 1.5575, + "step": 1517 + }, + { + "epoch": 1.559322033898305, + "grad_norm": 2.767385244369507, + "learning_rate": 2.696992767415303e-06, + "loss": 1.5007, + "step": 1518 + }, + { + "epoch": 1.5603492552645095, + "grad_norm": 2.6531004905700684, + "learning_rate": 2.695089455652836e-06, + "loss": 1.4977, + "step": 1519 + }, + { + "epoch": 1.5613764766307139, + "grad_norm": 2.686450719833374, + "learning_rate": 2.6931861438903694e-06, + "loss": 1.479, + "step": 1520 + }, + { + "epoch": 1.5613764766307139, + "eval_loss": 1.88840651512146, + "eval_runtime": 44.5631, + "eval_samples_per_second": 22.44, + "eval_steps_per_second": 0.942, + "step": 1520 + }, + { + "epoch": 1.5624036979969183, + "grad_norm": 2.6664063930511475, + "learning_rate": 2.6912828321279028e-06, + "loss": 1.5267, + "step": 1521 + }, + { + "epoch": 1.5634309193631228, + "grad_norm": 2.676542282104492, + "learning_rate": 2.689379520365436e-06, + "loss": 1.5005, + "step": 1522 + }, + { + "epoch": 1.5644581407293272, + "grad_norm": 2.8331971168518066, + "learning_rate": 2.6874762086029695e-06, + "loss": 1.5308, + "step": 1523 + }, + { + "epoch": 1.5654853620955316, + "grad_norm": 2.6427881717681885, + "learning_rate": 2.6855728968405024e-06, + "loss": 1.4813, + "step": 1524 + }, + { + "epoch": 1.5665125834617362, + "grad_norm": 2.7058424949645996, + "learning_rate": 2.683669585078036e-06, + "loss": 1.5149, + "step": 1525 + }, + { + "epoch": 1.5675398048279403, + "grad_norm": 2.7069029808044434, + "learning_rate": 2.681766273315569e-06, + "loss": 1.4588, + "step": 1526 + }, + { + "epoch": 1.568567026194145, + "grad_norm": 2.7757389545440674, + "learning_rate": 2.6798629615531025e-06, + "loss": 1.4953, + "step": 1527 + }, + { + "epoch": 1.5695942475603493, + "grad_norm": 2.788466215133667, + "learning_rate": 2.677959649790636e-06, + "loss": 1.5028, + "step": 1528 + }, + { + "epoch": 1.5706214689265536, + "grad_norm": 2.659156322479248, + "learning_rate": 2.676056338028169e-06, + "loss": 1.5365, + "step": 1529 + }, + { + "epoch": 1.5716486902927582, + "grad_norm": 2.7179627418518066, + "learning_rate": 2.674153026265702e-06, + "loss": 1.5232, + "step": 1530 + }, + { + "epoch": 1.5716486902927582, + "eval_loss": 1.8908981084823608, + "eval_runtime": 44.1745, + "eval_samples_per_second": 22.637, + "eval_steps_per_second": 0.951, + "step": 1530 + }, + { + "epoch": 1.5726759116589624, + "grad_norm": 2.676546812057495, + "learning_rate": 2.672249714503236e-06, + "loss": 1.5247, + "step": 1531 + }, + { + "epoch": 1.573703133025167, + "grad_norm": 2.629784345626831, + "learning_rate": 2.6703464027407693e-06, + "loss": 1.488, + "step": 1532 + }, + { + "epoch": 1.5747303543913713, + "grad_norm": 2.7101712226867676, + "learning_rate": 2.6684430909783027e-06, + "loss": 1.5135, + "step": 1533 + }, + { + "epoch": 1.5757575757575757, + "grad_norm": 2.8025403022766113, + "learning_rate": 2.666539779215836e-06, + "loss": 1.5024, + "step": 1534 + }, + { + "epoch": 1.5767847971237803, + "grad_norm": 2.6695220470428467, + "learning_rate": 2.6646364674533694e-06, + "loss": 1.4844, + "step": 1535 + }, + { + "epoch": 1.5778120184899846, + "grad_norm": 2.610874652862549, + "learning_rate": 2.6627331556909024e-06, + "loss": 1.3949, + "step": 1536 + }, + { + "epoch": 1.578839239856189, + "grad_norm": 2.926985025405884, + "learning_rate": 2.6608298439284357e-06, + "loss": 1.4941, + "step": 1537 + }, + { + "epoch": 1.5798664612223934, + "grad_norm": 2.610355854034424, + "learning_rate": 2.658926532165969e-06, + "loss": 1.4895, + "step": 1538 + }, + { + "epoch": 1.5808936825885977, + "grad_norm": 2.6426877975463867, + "learning_rate": 2.6570232204035024e-06, + "loss": 1.4483, + "step": 1539 + }, + { + "epoch": 1.5819209039548023, + "grad_norm": 2.7410695552825928, + "learning_rate": 2.655119908641036e-06, + "loss": 1.4931, + "step": 1540 + }, + { + "epoch": 1.5819209039548023, + "eval_loss": 1.903024673461914, + "eval_runtime": 43.387, + "eval_samples_per_second": 23.048, + "eval_steps_per_second": 0.968, + "step": 1540 + }, + { + "epoch": 1.5829481253210067, + "grad_norm": 2.6797378063201904, + "learning_rate": 2.6532165968785687e-06, + "loss": 1.497, + "step": 1541 + }, + { + "epoch": 1.583975346687211, + "grad_norm": 2.697439670562744, + "learning_rate": 2.651313285116102e-06, + "loss": 1.4596, + "step": 1542 + }, + { + "epoch": 1.5850025680534157, + "grad_norm": 2.6819119453430176, + "learning_rate": 2.6494099733536355e-06, + "loss": 1.4712, + "step": 1543 + }, + { + "epoch": 1.5860297894196198, + "grad_norm": 2.9221737384796143, + "learning_rate": 2.647506661591169e-06, + "loss": 1.6449, + "step": 1544 + }, + { + "epoch": 1.5870570107858244, + "grad_norm": 2.757866859436035, + "learning_rate": 2.645603349828702e-06, + "loss": 1.5251, + "step": 1545 + }, + { + "epoch": 1.5880842321520288, + "grad_norm": 2.693976402282715, + "learning_rate": 2.643700038066235e-06, + "loss": 1.462, + "step": 1546 + }, + { + "epoch": 1.5891114535182331, + "grad_norm": 2.732808828353882, + "learning_rate": 2.6417967263037685e-06, + "loss": 1.5115, + "step": 1547 + }, + { + "epoch": 1.5901386748844377, + "grad_norm": 2.5835330486297607, + "learning_rate": 2.639893414541302e-06, + "loss": 1.533, + "step": 1548 + }, + { + "epoch": 1.5911658962506419, + "grad_norm": 2.6625828742980957, + "learning_rate": 2.6379901027788352e-06, + "loss": 1.555, + "step": 1549 + }, + { + "epoch": 1.5921931176168465, + "grad_norm": 2.5941221714019775, + "learning_rate": 2.6360867910163686e-06, + "loss": 1.4881, + "step": 1550 + }, + { + "epoch": 1.5921931176168465, + "eval_loss": 1.9022443294525146, + "eval_runtime": 46.3086, + "eval_samples_per_second": 21.594, + "eval_steps_per_second": 0.907, + "step": 1550 + }, + { + "epoch": 1.5932203389830508, + "grad_norm": 2.6347010135650635, + "learning_rate": 2.6341834792539024e-06, + "loss": 1.5118, + "step": 1551 + }, + { + "epoch": 1.5942475603492552, + "grad_norm": 2.6772608757019043, + "learning_rate": 2.6322801674914357e-06, + "loss": 1.5107, + "step": 1552 + }, + { + "epoch": 1.5952747817154598, + "grad_norm": 2.819559097290039, + "learning_rate": 2.6303768557289687e-06, + "loss": 1.5008, + "step": 1553 + }, + { + "epoch": 1.5963020030816641, + "grad_norm": 2.5580296516418457, + "learning_rate": 2.628473543966502e-06, + "loss": 1.4561, + "step": 1554 + }, + { + "epoch": 1.5973292244478685, + "grad_norm": 2.6931111812591553, + "learning_rate": 2.6265702322040354e-06, + "loss": 1.5222, + "step": 1555 + }, + { + "epoch": 1.5983564458140729, + "grad_norm": 2.6396803855895996, + "learning_rate": 2.6246669204415688e-06, + "loss": 1.4791, + "step": 1556 + }, + { + "epoch": 1.5993836671802772, + "grad_norm": 2.6619536876678467, + "learning_rate": 2.622763608679102e-06, + "loss": 1.471, + "step": 1557 + }, + { + "epoch": 1.6004108885464818, + "grad_norm": 2.578331232070923, + "learning_rate": 2.620860296916635e-06, + "loss": 1.4787, + "step": 1558 + }, + { + "epoch": 1.6014381099126862, + "grad_norm": 2.764533042907715, + "learning_rate": 2.6189569851541684e-06, + "loss": 1.4305, + "step": 1559 + }, + { + "epoch": 1.6024653312788906, + "grad_norm": 2.673753499984741, + "learning_rate": 2.617053673391702e-06, + "loss": 1.536, + "step": 1560 + }, + { + "epoch": 1.6024653312788906, + "eval_loss": 1.8979606628417969, + "eval_runtime": 46.5101, + "eval_samples_per_second": 21.501, + "eval_steps_per_second": 0.903, + "step": 1560 + }, + { + "epoch": 1.6034925526450952, + "grad_norm": 2.7593281269073486, + "learning_rate": 2.615150361629235e-06, + "loss": 1.5721, + "step": 1561 + }, + { + "epoch": 1.6045197740112993, + "grad_norm": 2.72963285446167, + "learning_rate": 2.6132470498667685e-06, + "loss": 1.5198, + "step": 1562 + }, + { + "epoch": 1.605546995377504, + "grad_norm": 2.6438205242156982, + "learning_rate": 2.6113437381043015e-06, + "loss": 1.4922, + "step": 1563 + }, + { + "epoch": 1.6065742167437083, + "grad_norm": 2.6523141860961914, + "learning_rate": 2.609440426341835e-06, + "loss": 1.4775, + "step": 1564 + }, + { + "epoch": 1.6076014381099126, + "grad_norm": 2.7506513595581055, + "learning_rate": 2.607537114579368e-06, + "loss": 1.524, + "step": 1565 + }, + { + "epoch": 1.6086286594761172, + "grad_norm": 2.589853286743164, + "learning_rate": 2.6056338028169015e-06, + "loss": 1.4946, + "step": 1566 + }, + { + "epoch": 1.6096558808423214, + "grad_norm": 2.6591460704803467, + "learning_rate": 2.603730491054435e-06, + "loss": 1.4686, + "step": 1567 + }, + { + "epoch": 1.610683102208526, + "grad_norm": 2.8163933753967285, + "learning_rate": 2.601827179291968e-06, + "loss": 1.5196, + "step": 1568 + }, + { + "epoch": 1.6117103235747303, + "grad_norm": 2.6938319206237793, + "learning_rate": 2.599923867529501e-06, + "loss": 1.5082, + "step": 1569 + }, + { + "epoch": 1.6127375449409347, + "grad_norm": 2.7489402294158936, + "learning_rate": 2.5980205557670346e-06, + "loss": 1.4978, + "step": 1570 + }, + { + "epoch": 1.6127375449409347, + "eval_loss": 1.897243618965149, + "eval_runtime": 45.6931, + "eval_samples_per_second": 21.885, + "eval_steps_per_second": 0.919, + "step": 1570 + }, + { + "epoch": 1.6137647663071393, + "grad_norm": 2.763204336166382, + "learning_rate": 2.5961172440045684e-06, + "loss": 1.5382, + "step": 1571 + }, + { + "epoch": 1.6147919876733436, + "grad_norm": 2.726813316345215, + "learning_rate": 2.5942139322421017e-06, + "loss": 1.5377, + "step": 1572 + }, + { + "epoch": 1.615819209039548, + "grad_norm": 2.6197266578674316, + "learning_rate": 2.592310620479635e-06, + "loss": 1.4887, + "step": 1573 + }, + { + "epoch": 1.6168464304057526, + "grad_norm": 2.739987373352051, + "learning_rate": 2.5904073087171684e-06, + "loss": 1.4546, + "step": 1574 + }, + { + "epoch": 1.6178736517719567, + "grad_norm": 2.6571943759918213, + "learning_rate": 2.5885039969547014e-06, + "loss": 1.5129, + "step": 1575 + }, + { + "epoch": 1.6189008731381613, + "grad_norm": 2.7591447830200195, + "learning_rate": 2.5866006851922347e-06, + "loss": 1.4627, + "step": 1576 + }, + { + "epoch": 1.6199280945043657, + "grad_norm": 2.6228175163269043, + "learning_rate": 2.584697373429768e-06, + "loss": 1.4712, + "step": 1577 + }, + { + "epoch": 1.62095531587057, + "grad_norm": 2.7786920070648193, + "learning_rate": 2.5827940616673015e-06, + "loss": 1.4995, + "step": 1578 + }, + { + "epoch": 1.6219825372367747, + "grad_norm": 2.782848596572876, + "learning_rate": 2.580890749904835e-06, + "loss": 1.5387, + "step": 1579 + }, + { + "epoch": 1.6230097586029788, + "grad_norm": 2.7152798175811768, + "learning_rate": 2.5789874381423678e-06, + "loss": 1.4813, + "step": 1580 + }, + { + "epoch": 1.6230097586029788, + "eval_loss": 1.8939763307571411, + "eval_runtime": 46.2339, + "eval_samples_per_second": 21.629, + "eval_steps_per_second": 0.908, + "step": 1580 + }, + { + "epoch": 1.6240369799691834, + "grad_norm": 2.6328132152557373, + "learning_rate": 2.577084126379901e-06, + "loss": 1.4801, + "step": 1581 + }, + { + "epoch": 1.6250642013353878, + "grad_norm": 2.6837892532348633, + "learning_rate": 2.5751808146174345e-06, + "loss": 1.453, + "step": 1582 + }, + { + "epoch": 1.6260914227015921, + "grad_norm": 2.6900155544281006, + "learning_rate": 2.573277502854968e-06, + "loss": 1.5044, + "step": 1583 + }, + { + "epoch": 1.6271186440677967, + "grad_norm": 2.6830708980560303, + "learning_rate": 2.5713741910925012e-06, + "loss": 1.4662, + "step": 1584 + }, + { + "epoch": 1.6281458654340009, + "grad_norm": 2.6451282501220703, + "learning_rate": 2.569470879330034e-06, + "loss": 1.4749, + "step": 1585 + }, + { + "epoch": 1.6291730868002054, + "grad_norm": 2.8980915546417236, + "learning_rate": 2.5675675675675675e-06, + "loss": 1.5729, + "step": 1586 + }, + { + "epoch": 1.6302003081664098, + "grad_norm": 2.715513229370117, + "learning_rate": 2.565664255805101e-06, + "loss": 1.4848, + "step": 1587 + }, + { + "epoch": 1.6312275295326142, + "grad_norm": 2.8286707401275635, + "learning_rate": 2.5637609440426343e-06, + "loss": 1.5164, + "step": 1588 + }, + { + "epoch": 1.6322547508988188, + "grad_norm": 2.7614853382110596, + "learning_rate": 2.5618576322801676e-06, + "loss": 1.4623, + "step": 1589 + }, + { + "epoch": 1.6332819722650231, + "grad_norm": 2.826911687850952, + "learning_rate": 2.5599543205177006e-06, + "loss": 1.5593, + "step": 1590 + }, + { + "epoch": 1.6332819722650231, + "eval_loss": 1.8977031707763672, + "eval_runtime": 46.7228, + "eval_samples_per_second": 21.403, + "eval_steps_per_second": 0.899, + "step": 1590 + }, + { + "epoch": 1.6343091936312275, + "grad_norm": 2.707435131072998, + "learning_rate": 2.558051008755234e-06, + "loss": 1.5186, + "step": 1591 + }, + { + "epoch": 1.635336414997432, + "grad_norm": 2.701657295227051, + "learning_rate": 2.5561476969927677e-06, + "loss": 1.5643, + "step": 1592 + }, + { + "epoch": 1.6363636363636362, + "grad_norm": 2.7254910469055176, + "learning_rate": 2.554244385230301e-06, + "loss": 1.4701, + "step": 1593 + }, + { + "epoch": 1.6373908577298408, + "grad_norm": 2.6841659545898438, + "learning_rate": 2.5523410734678344e-06, + "loss": 1.4949, + "step": 1594 + }, + { + "epoch": 1.6384180790960452, + "grad_norm": 2.746089220046997, + "learning_rate": 2.550437761705368e-06, + "loss": 1.435, + "step": 1595 + }, + { + "epoch": 1.6394453004622496, + "grad_norm": 2.6010215282440186, + "learning_rate": 2.548534449942901e-06, + "loss": 1.4594, + "step": 1596 + }, + { + "epoch": 1.6404725218284542, + "grad_norm": 2.9612057209014893, + "learning_rate": 2.546631138180434e-06, + "loss": 1.5084, + "step": 1597 + }, + { + "epoch": 1.6414997431946583, + "grad_norm": 3.105278253555298, + "learning_rate": 2.5447278264179675e-06, + "loss": 1.5041, + "step": 1598 + }, + { + "epoch": 1.642526964560863, + "grad_norm": 2.714082956314087, + "learning_rate": 2.542824514655501e-06, + "loss": 1.462, + "step": 1599 + }, + { + "epoch": 1.6435541859270673, + "grad_norm": 2.8359603881835938, + "learning_rate": 2.540921202893034e-06, + "loss": 1.5545, + "step": 1600 + }, + { + "epoch": 1.6435541859270673, + "eval_loss": 1.9013447761535645, + "eval_runtime": 47.8392, + "eval_samples_per_second": 20.903, + "eval_steps_per_second": 0.878, + "step": 1600 + }, + { + "epoch": 1.6445814072932716, + "grad_norm": 2.8133420944213867, + "learning_rate": 2.5390178911305675e-06, + "loss": 1.5362, + "step": 1601 + }, + { + "epoch": 1.6456086286594762, + "grad_norm": 2.7947328090667725, + "learning_rate": 2.5371145793681005e-06, + "loss": 1.5067, + "step": 1602 + }, + { + "epoch": 1.6466358500256806, + "grad_norm": 2.705209732055664, + "learning_rate": 2.535211267605634e-06, + "loss": 1.4921, + "step": 1603 + }, + { + "epoch": 1.647663071391885, + "grad_norm": 2.7078511714935303, + "learning_rate": 2.533307955843167e-06, + "loss": 1.4398, + "step": 1604 + }, + { + "epoch": 1.6486902927580893, + "grad_norm": 2.749514579772949, + "learning_rate": 2.5314046440807006e-06, + "loss": 1.4931, + "step": 1605 + }, + { + "epoch": 1.6497175141242937, + "grad_norm": 2.703160285949707, + "learning_rate": 2.529501332318234e-06, + "loss": 1.4693, + "step": 1606 + }, + { + "epoch": 1.6507447354904983, + "grad_norm": 2.8278794288635254, + "learning_rate": 2.5275980205557673e-06, + "loss": 1.5353, + "step": 1607 + }, + { + "epoch": 1.6517719568567026, + "grad_norm": 2.927949905395508, + "learning_rate": 2.5256947087933002e-06, + "loss": 1.5665, + "step": 1608 + }, + { + "epoch": 1.652799178222907, + "grad_norm": 2.714650869369507, + "learning_rate": 2.5237913970308336e-06, + "loss": 1.5503, + "step": 1609 + }, + { + "epoch": 1.6538263995891116, + "grad_norm": 2.7319588661193848, + "learning_rate": 2.521888085268367e-06, + "loss": 1.5018, + "step": 1610 + }, + { + "epoch": 1.6538263995891116, + "eval_loss": 1.8987922668457031, + "eval_runtime": 43.4436, + "eval_samples_per_second": 23.018, + "eval_steps_per_second": 0.967, + "step": 1610 + }, + { + "epoch": 1.6548536209553157, + "grad_norm": 2.746079921722412, + "learning_rate": 2.5199847735059003e-06, + "loss": 1.5447, + "step": 1611 + }, + { + "epoch": 1.6558808423215203, + "grad_norm": 2.7670583724975586, + "learning_rate": 2.518081461743434e-06, + "loss": 1.4723, + "step": 1612 + }, + { + "epoch": 1.6569080636877247, + "grad_norm": 2.680903196334839, + "learning_rate": 2.5161781499809675e-06, + "loss": 1.5116, + "step": 1613 + }, + { + "epoch": 1.657935285053929, + "grad_norm": 2.5936760902404785, + "learning_rate": 2.5142748382185004e-06, + "loss": 1.4607, + "step": 1614 + }, + { + "epoch": 1.6589625064201337, + "grad_norm": 2.742476224899292, + "learning_rate": 2.5123715264560338e-06, + "loss": 1.4579, + "step": 1615 + }, + { + "epoch": 1.6599897277863378, + "grad_norm": 2.9192259311676025, + "learning_rate": 2.510468214693567e-06, + "loss": 1.4815, + "step": 1616 + }, + { + "epoch": 1.6610169491525424, + "grad_norm": 2.7400426864624023, + "learning_rate": 2.5085649029311005e-06, + "loss": 1.4268, + "step": 1617 + }, + { + "epoch": 1.6620441705187468, + "grad_norm": 2.6352646350860596, + "learning_rate": 2.506661591168634e-06, + "loss": 1.3921, + "step": 1618 + }, + { + "epoch": 1.6630713918849511, + "grad_norm": 2.6632823944091797, + "learning_rate": 2.504758279406167e-06, + "loss": 1.4754, + "step": 1619 + }, + { + "epoch": 1.6640986132511557, + "grad_norm": 2.8280115127563477, + "learning_rate": 2.5028549676437e-06, + "loss": 1.5208, + "step": 1620 + }, + { + "epoch": 1.6640986132511557, + "eval_loss": 1.8962472677230835, + "eval_runtime": 45.5434, + "eval_samples_per_second": 21.957, + "eval_steps_per_second": 0.922, + "step": 1620 + }, + { + "epoch": 1.66512583461736, + "grad_norm": 2.84045672416687, + "learning_rate": 2.5009516558812335e-06, + "loss": 1.4669, + "step": 1621 + }, + { + "epoch": 1.6661530559835644, + "grad_norm": 2.564673900604248, + "learning_rate": 2.499048344118767e-06, + "loss": 1.5092, + "step": 1622 + }, + { + "epoch": 1.667180277349769, + "grad_norm": 2.622130870819092, + "learning_rate": 2.4971450323563002e-06, + "loss": 1.5606, + "step": 1623 + }, + { + "epoch": 1.6682074987159732, + "grad_norm": 2.5972182750701904, + "learning_rate": 2.4952417205938336e-06, + "loss": 1.4751, + "step": 1624 + }, + { + "epoch": 1.6692347200821778, + "grad_norm": 2.6662817001342773, + "learning_rate": 2.4933384088313666e-06, + "loss": 1.4789, + "step": 1625 + }, + { + "epoch": 1.6702619414483821, + "grad_norm": 2.6823441982269287, + "learning_rate": 2.4914350970689e-06, + "loss": 1.462, + "step": 1626 + }, + { + "epoch": 1.6712891628145865, + "grad_norm": 2.7047946453094482, + "learning_rate": 2.4895317853064337e-06, + "loss": 1.4809, + "step": 1627 + }, + { + "epoch": 1.672316384180791, + "grad_norm": 2.7356603145599365, + "learning_rate": 2.4876284735439666e-06, + "loss": 1.5316, + "step": 1628 + }, + { + "epoch": 1.6733436055469952, + "grad_norm": 2.8151392936706543, + "learning_rate": 2.4857251617815e-06, + "loss": 1.4969, + "step": 1629 + }, + { + "epoch": 1.6743708269131998, + "grad_norm": 2.8269588947296143, + "learning_rate": 2.4838218500190334e-06, + "loss": 1.5041, + "step": 1630 + }, + { + "epoch": 1.6743708269131998, + "eval_loss": 1.9033366441726685, + "eval_runtime": 47.514, + "eval_samples_per_second": 21.046, + "eval_steps_per_second": 0.884, + "step": 1630 + }, + { + "epoch": 1.6753980482794042, + "grad_norm": 2.7226245403289795, + "learning_rate": 2.4819185382565667e-06, + "loss": 1.4924, + "step": 1631 + }, + { + "epoch": 1.6764252696456086, + "grad_norm": 2.6129348278045654, + "learning_rate": 2.4800152264941e-06, + "loss": 1.4829, + "step": 1632 + }, + { + "epoch": 1.6774524910118132, + "grad_norm": 2.618772029876709, + "learning_rate": 2.478111914731633e-06, + "loss": 1.452, + "step": 1633 + }, + { + "epoch": 1.6784797123780173, + "grad_norm": 2.7626497745513916, + "learning_rate": 2.4762086029691664e-06, + "loss": 1.4875, + "step": 1634 + }, + { + "epoch": 1.6795069337442219, + "grad_norm": 3.0622305870056152, + "learning_rate": 2.4743052912066998e-06, + "loss": 1.6208, + "step": 1635 + }, + { + "epoch": 1.6805341551104263, + "grad_norm": 2.6085500717163086, + "learning_rate": 2.472401979444233e-06, + "loss": 1.457, + "step": 1636 + }, + { + "epoch": 1.6815613764766306, + "grad_norm": 2.7662956714630127, + "learning_rate": 2.4704986676817665e-06, + "loss": 1.5572, + "step": 1637 + }, + { + "epoch": 1.6825885978428352, + "grad_norm": 2.8282175064086914, + "learning_rate": 2.4685953559193e-06, + "loss": 1.4922, + "step": 1638 + }, + { + "epoch": 1.6836158192090396, + "grad_norm": 2.930837631225586, + "learning_rate": 2.466692044156833e-06, + "loss": 1.5296, + "step": 1639 + }, + { + "epoch": 1.684643040575244, + "grad_norm": 2.7347614765167236, + "learning_rate": 2.4647887323943666e-06, + "loss": 1.4739, + "step": 1640 + }, + { + "epoch": 1.684643040575244, + "eval_loss": 1.8892933130264282, + "eval_runtime": 48.2635, + "eval_samples_per_second": 20.72, + "eval_steps_per_second": 0.87, + "step": 1640 + }, + { + "epoch": 1.6856702619414485, + "grad_norm": 2.6786420345306396, + "learning_rate": 2.4628854206319e-06, + "loss": 1.4429, + "step": 1641 + }, + { + "epoch": 1.6866974833076527, + "grad_norm": 2.754531145095825, + "learning_rate": 2.460982108869433e-06, + "loss": 1.4375, + "step": 1642 + }, + { + "epoch": 1.6877247046738573, + "grad_norm": 2.8799307346343994, + "learning_rate": 2.4590787971069662e-06, + "loss": 1.5438, + "step": 1643 + }, + { + "epoch": 1.6887519260400616, + "grad_norm": 2.6701910495758057, + "learning_rate": 2.4571754853444996e-06, + "loss": 1.4691, + "step": 1644 + }, + { + "epoch": 1.689779147406266, + "grad_norm": 2.760960340499878, + "learning_rate": 2.455272173582033e-06, + "loss": 1.4607, + "step": 1645 + }, + { + "epoch": 1.6908063687724706, + "grad_norm": 2.7178521156311035, + "learning_rate": 2.4533688618195663e-06, + "loss": 1.466, + "step": 1646 + }, + { + "epoch": 1.6918335901386747, + "grad_norm": 2.7641139030456543, + "learning_rate": 2.4514655500570997e-06, + "loss": 1.4065, + "step": 1647 + }, + { + "epoch": 1.6928608115048793, + "grad_norm": 2.7858457565307617, + "learning_rate": 2.449562238294633e-06, + "loss": 1.4587, + "step": 1648 + }, + { + "epoch": 1.6938880328710837, + "grad_norm": 2.6738100051879883, + "learning_rate": 2.4476589265321664e-06, + "loss": 1.4205, + "step": 1649 + }, + { + "epoch": 1.694915254237288, + "grad_norm": 2.748340129852295, + "learning_rate": 2.4457556147696993e-06, + "loss": 1.5212, + "step": 1650 + }, + { + "epoch": 1.694915254237288, + "eval_loss": 1.9066362380981445, + "eval_runtime": 52.2773, + "eval_samples_per_second": 19.129, + "eval_steps_per_second": 0.803, + "step": 1650 + }, + { + "epoch": 1.6959424756034926, + "grad_norm": 2.7664692401885986, + "learning_rate": 2.4438523030072327e-06, + "loss": 1.5045, + "step": 1651 + }, + { + "epoch": 1.696969696969697, + "grad_norm": 2.7928621768951416, + "learning_rate": 2.441948991244766e-06, + "loss": 1.5134, + "step": 1652 + }, + { + "epoch": 1.6979969183359014, + "grad_norm": 2.7065975666046143, + "learning_rate": 2.4400456794822994e-06, + "loss": 1.5072, + "step": 1653 + }, + { + "epoch": 1.6990241397021058, + "grad_norm": 2.7586593627929688, + "learning_rate": 2.438142367719833e-06, + "loss": 1.5276, + "step": 1654 + }, + { + "epoch": 1.7000513610683101, + "grad_norm": 2.689526081085205, + "learning_rate": 2.4362390559573657e-06, + "loss": 1.4902, + "step": 1655 + }, + { + "epoch": 1.7010785824345147, + "grad_norm": 2.716808795928955, + "learning_rate": 2.434335744194899e-06, + "loss": 1.4772, + "step": 1656 + }, + { + "epoch": 1.702105803800719, + "grad_norm": 2.825169086456299, + "learning_rate": 2.432432432432433e-06, + "loss": 1.5185, + "step": 1657 + }, + { + "epoch": 1.7031330251669234, + "grad_norm": 2.727210283279419, + "learning_rate": 2.4305291206699662e-06, + "loss": 1.4353, + "step": 1658 + }, + { + "epoch": 1.704160246533128, + "grad_norm": 2.753382921218872, + "learning_rate": 2.428625808907499e-06, + "loss": 1.5161, + "step": 1659 + }, + { + "epoch": 1.7051874678993322, + "grad_norm": 2.7188405990600586, + "learning_rate": 2.4267224971450325e-06, + "loss": 1.5324, + "step": 1660 + }, + { + "epoch": 1.7051874678993322, + "eval_loss": 1.8945986032485962, + "eval_runtime": 49.8237, + "eval_samples_per_second": 20.071, + "eval_steps_per_second": 0.843, + "step": 1660 + }, + { + "epoch": 1.7062146892655368, + "grad_norm": 2.702894687652588, + "learning_rate": 2.424819185382566e-06, + "loss": 1.4915, + "step": 1661 + }, + { + "epoch": 1.7072419106317411, + "grad_norm": 2.6755716800689697, + "learning_rate": 2.4229158736200993e-06, + "loss": 1.4471, + "step": 1662 + }, + { + "epoch": 1.7082691319979455, + "grad_norm": 2.637878656387329, + "learning_rate": 2.4210125618576326e-06, + "loss": 1.4488, + "step": 1663 + }, + { + "epoch": 1.70929635336415, + "grad_norm": 2.6714439392089844, + "learning_rate": 2.4191092500951656e-06, + "loss": 1.4708, + "step": 1664 + }, + { + "epoch": 1.7103235747303542, + "grad_norm": 2.703479528427124, + "learning_rate": 2.417205938332699e-06, + "loss": 1.5039, + "step": 1665 + }, + { + "epoch": 1.7113507960965588, + "grad_norm": 2.7005186080932617, + "learning_rate": 2.4153026265702323e-06, + "loss": 1.4242, + "step": 1666 + }, + { + "epoch": 1.7123780174627632, + "grad_norm": 2.7262654304504395, + "learning_rate": 2.4133993148077657e-06, + "loss": 1.4649, + "step": 1667 + }, + { + "epoch": 1.7134052388289676, + "grad_norm": 2.912041425704956, + "learning_rate": 2.411496003045299e-06, + "loss": 1.5247, + "step": 1668 + }, + { + "epoch": 1.7144324601951721, + "grad_norm": 2.713007926940918, + "learning_rate": 2.4095926912828324e-06, + "loss": 1.4585, + "step": 1669 + }, + { + "epoch": 1.7154596815613765, + "grad_norm": 2.6106162071228027, + "learning_rate": 2.4076893795203658e-06, + "loss": 1.4932, + "step": 1670 + }, + { + "epoch": 1.7154596815613765, + "eval_loss": 1.8863319158554077, + "eval_runtime": 45.1847, + "eval_samples_per_second": 22.131, + "eval_steps_per_second": 0.93, + "step": 1670 + }, + { + "epoch": 1.7164869029275809, + "grad_norm": 2.9076762199401855, + "learning_rate": 2.405786067757899e-06, + "loss": 1.4725, + "step": 1671 + }, + { + "epoch": 1.7175141242937855, + "grad_norm": 2.9677484035491943, + "learning_rate": 2.403882755995432e-06, + "loss": 1.551, + "step": 1672 + }, + { + "epoch": 1.7185413456599896, + "grad_norm": 2.6974384784698486, + "learning_rate": 2.4019794442329654e-06, + "loss": 1.4897, + "step": 1673 + }, + { + "epoch": 1.7195685670261942, + "grad_norm": 2.796464443206787, + "learning_rate": 2.4000761324704988e-06, + "loss": 1.4573, + "step": 1674 + }, + { + "epoch": 1.7205957883923986, + "grad_norm": 2.653627872467041, + "learning_rate": 2.398172820708032e-06, + "loss": 1.5025, + "step": 1675 + }, + { + "epoch": 1.721623009758603, + "grad_norm": 2.716545343399048, + "learning_rate": 2.3962695089455655e-06, + "loss": 1.4687, + "step": 1676 + }, + { + "epoch": 1.7226502311248075, + "grad_norm": 2.737752676010132, + "learning_rate": 2.3943661971830984e-06, + "loss": 1.4515, + "step": 1677 + }, + { + "epoch": 1.7236774524910117, + "grad_norm": 2.7596747875213623, + "learning_rate": 2.3924628854206322e-06, + "loss": 1.4712, + "step": 1678 + }, + { + "epoch": 1.7247046738572163, + "grad_norm": 2.9202370643615723, + "learning_rate": 2.3905595736581656e-06, + "loss": 1.4818, + "step": 1679 + }, + { + "epoch": 1.7257318952234206, + "grad_norm": 2.7129971981048584, + "learning_rate": 2.388656261895699e-06, + "loss": 1.5269, + "step": 1680 + }, + { + "epoch": 1.7257318952234206, + "eval_loss": 1.8941866159439087, + "eval_runtime": 45.4463, + "eval_samples_per_second": 22.004, + "eval_steps_per_second": 0.924, + "step": 1680 + }, + { + "epoch": 1.726759116589625, + "grad_norm": 2.956148862838745, + "learning_rate": 2.386752950133232e-06, + "loss": 1.4977, + "step": 1681 + }, + { + "epoch": 1.7277863379558296, + "grad_norm": 2.8728253841400146, + "learning_rate": 2.3848496383707653e-06, + "loss": 1.4961, + "step": 1682 + }, + { + "epoch": 1.7288135593220337, + "grad_norm": 2.7839162349700928, + "learning_rate": 2.3829463266082986e-06, + "loss": 1.4757, + "step": 1683 + }, + { + "epoch": 1.7298407806882383, + "grad_norm": 2.7487802505493164, + "learning_rate": 2.381043014845832e-06, + "loss": 1.5048, + "step": 1684 + }, + { + "epoch": 1.7308680020544427, + "grad_norm": 2.80683970451355, + "learning_rate": 2.3791397030833653e-06, + "loss": 1.4916, + "step": 1685 + }, + { + "epoch": 1.731895223420647, + "grad_norm": 2.6910040378570557, + "learning_rate": 2.3772363913208983e-06, + "loss": 1.4711, + "step": 1686 + }, + { + "epoch": 1.7329224447868516, + "grad_norm": 2.594284772872925, + "learning_rate": 2.3753330795584316e-06, + "loss": 1.4558, + "step": 1687 + }, + { + "epoch": 1.733949666153056, + "grad_norm": 2.7420263290405273, + "learning_rate": 2.3734297677959654e-06, + "loss": 1.4769, + "step": 1688 + }, + { + "epoch": 1.7349768875192604, + "grad_norm": 2.80002760887146, + "learning_rate": 2.3715264560334984e-06, + "loss": 1.5289, + "step": 1689 + }, + { + "epoch": 1.736004108885465, + "grad_norm": 2.799618721008301, + "learning_rate": 2.3696231442710317e-06, + "loss": 1.5131, + "step": 1690 + }, + { + "epoch": 1.736004108885465, + "eval_loss": 1.8925220966339111, + "eval_runtime": 45.89, + "eval_samples_per_second": 21.791, + "eval_steps_per_second": 0.915, + "step": 1690 + }, + { + "epoch": 1.7370313302516691, + "grad_norm": 2.701648712158203, + "learning_rate": 2.367719832508565e-06, + "loss": 1.5076, + "step": 1691 + }, + { + "epoch": 1.7380585516178737, + "grad_norm": 2.6925787925720215, + "learning_rate": 2.3658165207460985e-06, + "loss": 1.5331, + "step": 1692 + }, + { + "epoch": 1.739085772984078, + "grad_norm": 2.8448283672332764, + "learning_rate": 2.363913208983632e-06, + "loss": 1.4933, + "step": 1693 + }, + { + "epoch": 1.7401129943502824, + "grad_norm": 2.8252811431884766, + "learning_rate": 2.362009897221165e-06, + "loss": 1.5205, + "step": 1694 + }, + { + "epoch": 1.741140215716487, + "grad_norm": 2.675318956375122, + "learning_rate": 2.360106585458698e-06, + "loss": 1.4958, + "step": 1695 + }, + { + "epoch": 1.7421674370826912, + "grad_norm": 2.8392488956451416, + "learning_rate": 2.3582032736962315e-06, + "loss": 1.4522, + "step": 1696 + }, + { + "epoch": 1.7431946584488958, + "grad_norm": 2.8221116065979004, + "learning_rate": 2.356299961933765e-06, + "loss": 1.4924, + "step": 1697 + }, + { + "epoch": 1.7442218798151001, + "grad_norm": 2.685828447341919, + "learning_rate": 2.354396650171298e-06, + "loss": 1.4668, + "step": 1698 + }, + { + "epoch": 1.7452491011813045, + "grad_norm": 2.622481107711792, + "learning_rate": 2.3524933384088316e-06, + "loss": 1.451, + "step": 1699 + }, + { + "epoch": 1.746276322547509, + "grad_norm": 2.79746413230896, + "learning_rate": 2.350590026646365e-06, + "loss": 1.5473, + "step": 1700 + }, + { + "epoch": 1.746276322547509, + "eval_loss": 1.898282766342163, + "eval_runtime": 45.0673, + "eval_samples_per_second": 22.189, + "eval_steps_per_second": 0.932, + "step": 1700 + }, + { + "epoch": 1.7473035439137135, + "grad_norm": 2.8120598793029785, + "learning_rate": 2.3486867148838983e-06, + "loss": 1.4692, + "step": 1701 + }, + { + "epoch": 1.7483307652799178, + "grad_norm": 2.7376883029937744, + "learning_rate": 2.3467834031214317e-06, + "loss": 1.4749, + "step": 1702 + }, + { + "epoch": 1.7493579866461222, + "grad_norm": 2.7579424381256104, + "learning_rate": 2.3448800913589646e-06, + "loss": 1.4726, + "step": 1703 + }, + { + "epoch": 1.7503852080123266, + "grad_norm": 2.648026466369629, + "learning_rate": 2.342976779596498e-06, + "loss": 1.4938, + "step": 1704 + }, + { + "epoch": 1.7514124293785311, + "grad_norm": 2.6477997303009033, + "learning_rate": 2.3410734678340313e-06, + "loss": 1.4813, + "step": 1705 + }, + { + "epoch": 1.7524396507447355, + "grad_norm": 2.742249011993408, + "learning_rate": 2.3391701560715647e-06, + "loss": 1.5257, + "step": 1706 + }, + { + "epoch": 1.7534668721109399, + "grad_norm": 2.6430649757385254, + "learning_rate": 2.337266844309098e-06, + "loss": 1.4781, + "step": 1707 + }, + { + "epoch": 1.7544940934771445, + "grad_norm": 2.6746671199798584, + "learning_rate": 2.3353635325466314e-06, + "loss": 1.4807, + "step": 1708 + }, + { + "epoch": 1.7555213148433486, + "grad_norm": 2.8837218284606934, + "learning_rate": 2.3334602207841648e-06, + "loss": 1.4932, + "step": 1709 + }, + { + "epoch": 1.7565485362095532, + "grad_norm": 2.703761577606201, + "learning_rate": 2.331556909021698e-06, + "loss": 1.4969, + "step": 1710 + }, + { + "epoch": 1.7565485362095532, + "eval_loss": 1.8966599702835083, + "eval_runtime": 42.6325, + "eval_samples_per_second": 23.456, + "eval_steps_per_second": 0.985, + "step": 1710 + }, + { + "epoch": 1.7575757575757576, + "grad_norm": 2.760502338409424, + "learning_rate": 2.3296535972592315e-06, + "loss": 1.5148, + "step": 1711 + }, + { + "epoch": 1.758602978941962, + "grad_norm": 2.696356773376465, + "learning_rate": 2.3277502854967644e-06, + "loss": 1.4604, + "step": 1712 + }, + { + "epoch": 1.7596302003081665, + "grad_norm": 2.7124855518341064, + "learning_rate": 2.325846973734298e-06, + "loss": 1.5178, + "step": 1713 + }, + { + "epoch": 1.7606574216743707, + "grad_norm": 2.624866008758545, + "learning_rate": 2.323943661971831e-06, + "loss": 1.4786, + "step": 1714 + }, + { + "epoch": 1.7616846430405753, + "grad_norm": 2.785651922225952, + "learning_rate": 2.3220403502093645e-06, + "loss": 1.5077, + "step": 1715 + }, + { + "epoch": 1.7627118644067796, + "grad_norm": 2.7688722610473633, + "learning_rate": 2.320137038446898e-06, + "loss": 1.5299, + "step": 1716 + }, + { + "epoch": 1.763739085772984, + "grad_norm": 2.72312593460083, + "learning_rate": 2.318233726684431e-06, + "loss": 1.5352, + "step": 1717 + }, + { + "epoch": 1.7647663071391886, + "grad_norm": 2.7877933979034424, + "learning_rate": 2.3163304149219646e-06, + "loss": 1.5148, + "step": 1718 + }, + { + "epoch": 1.765793528505393, + "grad_norm": 2.6651675701141357, + "learning_rate": 2.314427103159498e-06, + "loss": 1.4979, + "step": 1719 + }, + { + "epoch": 1.7668207498715973, + "grad_norm": 2.747264862060547, + "learning_rate": 2.312523791397031e-06, + "loss": 1.4785, + "step": 1720 + }, + { + "epoch": 1.7668207498715973, + "eval_loss": 1.9028254747390747, + "eval_runtime": 47.0728, + "eval_samples_per_second": 21.244, + "eval_steps_per_second": 0.892, + "step": 1720 + }, + { + "epoch": 1.767847971237802, + "grad_norm": 2.6331684589385986, + "learning_rate": 2.3106204796345643e-06, + "loss": 1.4479, + "step": 1721 + }, + { + "epoch": 1.768875192604006, + "grad_norm": 2.662079095840454, + "learning_rate": 2.3087171678720976e-06, + "loss": 1.4827, + "step": 1722 + }, + { + "epoch": 1.7699024139702106, + "grad_norm": 2.735499382019043, + "learning_rate": 2.306813856109631e-06, + "loss": 1.5324, + "step": 1723 + }, + { + "epoch": 1.770929635336415, + "grad_norm": 2.66086745262146, + "learning_rate": 2.3049105443471644e-06, + "loss": 1.5002, + "step": 1724 + }, + { + "epoch": 1.7719568567026194, + "grad_norm": 2.7453548908233643, + "learning_rate": 2.3030072325846973e-06, + "loss": 1.48, + "step": 1725 + }, + { + "epoch": 1.772984078068824, + "grad_norm": 2.701510429382324, + "learning_rate": 2.3011039208222307e-06, + "loss": 1.4642, + "step": 1726 + }, + { + "epoch": 1.774011299435028, + "grad_norm": 2.6908695697784424, + "learning_rate": 2.299200609059764e-06, + "loss": 1.4607, + "step": 1727 + }, + { + "epoch": 1.7750385208012327, + "grad_norm": 2.7793400287628174, + "learning_rate": 2.297297297297298e-06, + "loss": 1.4756, + "step": 1728 + }, + { + "epoch": 1.776065742167437, + "grad_norm": 2.6336896419525146, + "learning_rate": 2.2953939855348308e-06, + "loss": 1.4336, + "step": 1729 + }, + { + "epoch": 1.7770929635336414, + "grad_norm": 2.82281231880188, + "learning_rate": 2.293490673772364e-06, + "loss": 1.5273, + "step": 1730 + }, + { + "epoch": 1.7770929635336414, + "eval_loss": 1.9012898206710815, + "eval_runtime": 46.5443, + "eval_samples_per_second": 21.485, + "eval_steps_per_second": 0.902, + "step": 1730 + }, + { + "epoch": 1.778120184899846, + "grad_norm": 2.7541351318359375, + "learning_rate": 2.2915873620098975e-06, + "loss": 1.5071, + "step": 1731 + }, + { + "epoch": 1.7791474062660502, + "grad_norm": 2.771777629852295, + "learning_rate": 2.289684050247431e-06, + "loss": 1.503, + "step": 1732 + }, + { + "epoch": 1.7801746276322548, + "grad_norm": 2.5365493297576904, + "learning_rate": 2.287780738484964e-06, + "loss": 1.4635, + "step": 1733 + }, + { + "epoch": 1.7812018489984591, + "grad_norm": 2.691694498062134, + "learning_rate": 2.285877426722497e-06, + "loss": 1.5027, + "step": 1734 + }, + { + "epoch": 1.7822290703646635, + "grad_norm": 2.782968282699585, + "learning_rate": 2.2839741149600305e-06, + "loss": 1.5204, + "step": 1735 + }, + { + "epoch": 1.783256291730868, + "grad_norm": 2.6389904022216797, + "learning_rate": 2.282070803197564e-06, + "loss": 1.5135, + "step": 1736 + }, + { + "epoch": 1.7842835130970724, + "grad_norm": 2.8609848022460938, + "learning_rate": 2.2801674914350972e-06, + "loss": 1.5013, + "step": 1737 + }, + { + "epoch": 1.7853107344632768, + "grad_norm": 2.6814842224121094, + "learning_rate": 2.2782641796726306e-06, + "loss": 1.4874, + "step": 1738 + }, + { + "epoch": 1.7863379558294814, + "grad_norm": 2.7633700370788574, + "learning_rate": 2.276360867910164e-06, + "loss": 1.5129, + "step": 1739 + }, + { + "epoch": 1.7873651771956856, + "grad_norm": 2.637755870819092, + "learning_rate": 2.2744575561476973e-06, + "loss": 1.4486, + "step": 1740 + }, + { + "epoch": 1.7873651771956856, + "eval_loss": 1.8927199840545654, + "eval_runtime": 46.5881, + "eval_samples_per_second": 21.465, + "eval_steps_per_second": 0.902, + "step": 1740 + }, + { + "epoch": 1.7883923985618901, + "grad_norm": 2.815883159637451, + "learning_rate": 2.2725542443852307e-06, + "loss": 1.4815, + "step": 1741 + }, + { + "epoch": 1.7894196199280945, + "grad_norm": 2.758363723754883, + "learning_rate": 2.2706509326227636e-06, + "loss": 1.5235, + "step": 1742 + }, + { + "epoch": 1.7904468412942989, + "grad_norm": 2.6607120037078857, + "learning_rate": 2.268747620860297e-06, + "loss": 1.503, + "step": 1743 + }, + { + "epoch": 1.7914740626605035, + "grad_norm": 2.7479255199432373, + "learning_rate": 2.2668443090978303e-06, + "loss": 1.503, + "step": 1744 + }, + { + "epoch": 1.7925012840267076, + "grad_norm": 2.746718168258667, + "learning_rate": 2.2649409973353637e-06, + "loss": 1.5142, + "step": 1745 + }, + { + "epoch": 1.7935285053929122, + "grad_norm": 2.6848037242889404, + "learning_rate": 2.263037685572897e-06, + "loss": 1.4585, + "step": 1746 + }, + { + "epoch": 1.7945557267591166, + "grad_norm": 2.6677210330963135, + "learning_rate": 2.26113437381043e-06, + "loss": 1.4617, + "step": 1747 + }, + { + "epoch": 1.795582948125321, + "grad_norm": 2.866431474685669, + "learning_rate": 2.259231062047964e-06, + "loss": 1.4945, + "step": 1748 + }, + { + "epoch": 1.7966101694915255, + "grad_norm": 2.953123092651367, + "learning_rate": 2.257327750285497e-06, + "loss": 1.5225, + "step": 1749 + }, + { + "epoch": 1.7976373908577299, + "grad_norm": 2.6214499473571777, + "learning_rate": 2.2554244385230305e-06, + "loss": 1.4707, + "step": 1750 + }, + { + "epoch": 1.7976373908577299, + "eval_loss": 1.8950202465057373, + "eval_runtime": 44.7553, + "eval_samples_per_second": 22.344, + "eval_steps_per_second": 0.938, + "step": 1750 + }, + { + "epoch": 1.7986646122239343, + "grad_norm": 2.7714295387268066, + "learning_rate": 2.2535211267605635e-06, + "loss": 1.4788, + "step": 1751 + }, + { + "epoch": 1.7996918335901386, + "grad_norm": 2.840985059738159, + "learning_rate": 2.251617814998097e-06, + "loss": 1.5004, + "step": 1752 + }, + { + "epoch": 1.800719054956343, + "grad_norm": 2.6215717792510986, + "learning_rate": 2.24971450323563e-06, + "loss": 1.4063, + "step": 1753 + }, + { + "epoch": 1.8017462763225476, + "grad_norm": 2.7458207607269287, + "learning_rate": 2.2478111914731636e-06, + "loss": 1.5045, + "step": 1754 + }, + { + "epoch": 1.802773497688752, + "grad_norm": 2.5622992515563965, + "learning_rate": 2.245907879710697e-06, + "loss": 1.3671, + "step": 1755 + }, + { + "epoch": 1.8038007190549563, + "grad_norm": 3.09394907951355, + "learning_rate": 2.24400456794823e-06, + "loss": 1.5151, + "step": 1756 + }, + { + "epoch": 1.804827940421161, + "grad_norm": 2.740785837173462, + "learning_rate": 2.2421012561857632e-06, + "loss": 1.4635, + "step": 1757 + }, + { + "epoch": 1.805855161787365, + "grad_norm": 2.8012168407440186, + "learning_rate": 2.240197944423297e-06, + "loss": 1.4694, + "step": 1758 + }, + { + "epoch": 1.8068823831535696, + "grad_norm": 2.799586534500122, + "learning_rate": 2.23829463266083e-06, + "loss": 1.4627, + "step": 1759 + }, + { + "epoch": 1.807909604519774, + "grad_norm": 2.6345391273498535, + "learning_rate": 2.2363913208983633e-06, + "loss": 1.4498, + "step": 1760 + }, + { + "epoch": 1.807909604519774, + "eval_loss": 1.9042582511901855, + "eval_runtime": 47.4919, + "eval_samples_per_second": 21.056, + "eval_steps_per_second": 0.884, + "step": 1760 + }, + { + "epoch": 1.8089368258859784, + "grad_norm": 2.715463638305664, + "learning_rate": 2.2344880091358967e-06, + "loss": 1.493, + "step": 1761 + }, + { + "epoch": 1.809964047252183, + "grad_norm": 2.6666505336761475, + "learning_rate": 2.23258469737343e-06, + "loss": 1.4678, + "step": 1762 + }, + { + "epoch": 1.810991268618387, + "grad_norm": 2.6605517864227295, + "learning_rate": 2.2306813856109634e-06, + "loss": 1.5154, + "step": 1763 + }, + { + "epoch": 1.8120184899845917, + "grad_norm": 2.621842384338379, + "learning_rate": 2.2287780738484963e-06, + "loss": 1.4626, + "step": 1764 + }, + { + "epoch": 1.813045711350796, + "grad_norm": 2.7204320430755615, + "learning_rate": 2.2268747620860297e-06, + "loss": 1.5461, + "step": 1765 + }, + { + "epoch": 1.8140729327170004, + "grad_norm": 2.8174679279327393, + "learning_rate": 2.224971450323563e-06, + "loss": 1.4881, + "step": 1766 + }, + { + "epoch": 1.815100154083205, + "grad_norm": 2.707890272140503, + "learning_rate": 2.2230681385610964e-06, + "loss": 1.4123, + "step": 1767 + }, + { + "epoch": 1.8161273754494094, + "grad_norm": 2.753523111343384, + "learning_rate": 2.2211648267986298e-06, + "loss": 1.4783, + "step": 1768 + }, + { + "epoch": 1.8171545968156138, + "grad_norm": 2.830095052719116, + "learning_rate": 2.219261515036163e-06, + "loss": 1.4536, + "step": 1769 + }, + { + "epoch": 1.8181818181818183, + "grad_norm": 2.803318738937378, + "learning_rate": 2.2173582032736965e-06, + "loss": 1.4939, + "step": 1770 + }, + { + "epoch": 1.8181818181818183, + "eval_loss": 1.899889349937439, + "eval_runtime": 46.7673, + "eval_samples_per_second": 21.382, + "eval_steps_per_second": 0.898, + "step": 1770 + }, + { + "epoch": 1.8192090395480225, + "grad_norm": 2.8235929012298584, + "learning_rate": 2.21545489151123e-06, + "loss": 1.5176, + "step": 1771 + }, + { + "epoch": 1.820236260914227, + "grad_norm": 2.761497974395752, + "learning_rate": 2.2135515797487632e-06, + "loss": 1.4685, + "step": 1772 + }, + { + "epoch": 1.8212634822804314, + "grad_norm": 2.589054822921753, + "learning_rate": 2.211648267986296e-06, + "loss": 1.4673, + "step": 1773 + }, + { + "epoch": 1.8222907036466358, + "grad_norm": 2.763709306716919, + "learning_rate": 2.2097449562238295e-06, + "loss": 1.4727, + "step": 1774 + }, + { + "epoch": 1.8233179250128404, + "grad_norm": 2.7250869274139404, + "learning_rate": 2.207841644461363e-06, + "loss": 1.4729, + "step": 1775 + }, + { + "epoch": 1.8243451463790445, + "grad_norm": 2.7706706523895264, + "learning_rate": 2.2059383326988963e-06, + "loss": 1.5382, + "step": 1776 + }, + { + "epoch": 1.8253723677452491, + "grad_norm": 2.9034879207611084, + "learning_rate": 2.2040350209364296e-06, + "loss": 1.4936, + "step": 1777 + }, + { + "epoch": 1.8263995891114535, + "grad_norm": 2.625908851623535, + "learning_rate": 2.2021317091739626e-06, + "loss": 1.4601, + "step": 1778 + }, + { + "epoch": 1.8274268104776579, + "grad_norm": 2.703167676925659, + "learning_rate": 2.2002283974114963e-06, + "loss": 1.4876, + "step": 1779 + }, + { + "epoch": 1.8284540318438625, + "grad_norm": 2.8047842979431152, + "learning_rate": 2.1983250856490297e-06, + "loss": 1.428, + "step": 1780 + }, + { + "epoch": 1.8284540318438625, + "eval_loss": 1.8968960046768188, + "eval_runtime": 46.5826, + "eval_samples_per_second": 21.467, + "eval_steps_per_second": 0.902, + "step": 1780 + }, + { + "epoch": 1.8294812532100666, + "grad_norm": 2.8730416297912598, + "learning_rate": 2.1964217738865626e-06, + "loss": 1.4697, + "step": 1781 + }, + { + "epoch": 1.8305084745762712, + "grad_norm": 2.958203077316284, + "learning_rate": 2.194518462124096e-06, + "loss": 1.4955, + "step": 1782 + }, + { + "epoch": 1.8315356959424756, + "grad_norm": 2.8280398845672607, + "learning_rate": 2.1926151503616294e-06, + "loss": 1.5391, + "step": 1783 + }, + { + "epoch": 1.83256291730868, + "grad_norm": 2.776029348373413, + "learning_rate": 2.1907118385991627e-06, + "loss": 1.4945, + "step": 1784 + }, + { + "epoch": 1.8335901386748845, + "grad_norm": 2.7049412727355957, + "learning_rate": 2.188808526836696e-06, + "loss": 1.4481, + "step": 1785 + }, + { + "epoch": 1.8346173600410889, + "grad_norm": 2.71523118019104, + "learning_rate": 2.1869052150742295e-06, + "loss": 1.4756, + "step": 1786 + }, + { + "epoch": 1.8356445814072933, + "grad_norm": 2.7659823894500732, + "learning_rate": 2.1850019033117624e-06, + "loss": 1.4065, + "step": 1787 + }, + { + "epoch": 1.8366718027734978, + "grad_norm": 2.679553508758545, + "learning_rate": 2.1830985915492958e-06, + "loss": 1.4744, + "step": 1788 + }, + { + "epoch": 1.837699024139702, + "grad_norm": 2.6452908515930176, + "learning_rate": 2.1811952797868295e-06, + "loss": 1.4922, + "step": 1789 + }, + { + "epoch": 1.8387262455059066, + "grad_norm": 2.780548095703125, + "learning_rate": 2.1792919680243625e-06, + "loss": 1.458, + "step": 1790 + }, + { + "epoch": 1.8387262455059066, + "eval_loss": 1.8877090215682983, + "eval_runtime": 47.6844, + "eval_samples_per_second": 20.971, + "eval_steps_per_second": 0.881, + "step": 1790 + }, + { + "epoch": 1.839753466872111, + "grad_norm": 2.8762049674987793, + "learning_rate": 2.177388656261896e-06, + "loss": 1.489, + "step": 1791 + }, + { + "epoch": 1.8407806882383153, + "grad_norm": 3.0053279399871826, + "learning_rate": 2.1754853444994292e-06, + "loss": 1.5681, + "step": 1792 + }, + { + "epoch": 1.84180790960452, + "grad_norm": 2.737626791000366, + "learning_rate": 2.1735820327369626e-06, + "loss": 1.3764, + "step": 1793 + }, + { + "epoch": 1.842835130970724, + "grad_norm": 2.6998884677886963, + "learning_rate": 2.171678720974496e-06, + "loss": 1.5099, + "step": 1794 + }, + { + "epoch": 1.8438623523369286, + "grad_norm": 2.8110108375549316, + "learning_rate": 2.169775409212029e-06, + "loss": 1.5079, + "step": 1795 + }, + { + "epoch": 1.844889573703133, + "grad_norm": 2.7312750816345215, + "learning_rate": 2.1678720974495622e-06, + "loss": 1.5136, + "step": 1796 + }, + { + "epoch": 1.8459167950693374, + "grad_norm": 2.692768096923828, + "learning_rate": 2.1659687856870956e-06, + "loss": 1.5005, + "step": 1797 + }, + { + "epoch": 1.846944016435542, + "grad_norm": 2.673879623413086, + "learning_rate": 2.164065473924629e-06, + "loss": 1.5127, + "step": 1798 + }, + { + "epoch": 1.8479712378017463, + "grad_norm": 2.6218647956848145, + "learning_rate": 2.1621621621621623e-06, + "loss": 1.4529, + "step": 1799 + }, + { + "epoch": 1.8489984591679507, + "grad_norm": 2.7149012088775635, + "learning_rate": 2.1602588503996957e-06, + "loss": 1.5048, + "step": 1800 + }, + { + "epoch": 1.8489984591679507, + "eval_loss": 1.883215069770813, + "eval_runtime": 50.8526, + "eval_samples_per_second": 19.665, + "eval_steps_per_second": 0.826, + "step": 1800 + }, + { + "epoch": 1.850025680534155, + "grad_norm": 2.60640025138855, + "learning_rate": 2.158355538637229e-06, + "loss": 1.4827, + "step": 1801 + }, + { + "epoch": 1.8510529019003594, + "grad_norm": 2.7535810470581055, + "learning_rate": 2.1564522268747624e-06, + "loss": 1.4709, + "step": 1802 + }, + { + "epoch": 1.852080123266564, + "grad_norm": 2.8784072399139404, + "learning_rate": 2.1545489151122958e-06, + "loss": 1.548, + "step": 1803 + }, + { + "epoch": 1.8531073446327684, + "grad_norm": 2.7007954120635986, + "learning_rate": 2.1526456033498287e-06, + "loss": 1.4926, + "step": 1804 + }, + { + "epoch": 1.8541345659989728, + "grad_norm": 2.699535369873047, + "learning_rate": 2.150742291587362e-06, + "loss": 1.4993, + "step": 1805 + }, + { + "epoch": 1.8551617873651773, + "grad_norm": 2.777305841445923, + "learning_rate": 2.1488389798248954e-06, + "loss": 1.4822, + "step": 1806 + }, + { + "epoch": 1.8561890087313815, + "grad_norm": 2.6989753246307373, + "learning_rate": 2.146935668062429e-06, + "loss": 1.515, + "step": 1807 + }, + { + "epoch": 1.857216230097586, + "grad_norm": 2.7000770568847656, + "learning_rate": 2.145032356299962e-06, + "loss": 1.4278, + "step": 1808 + }, + { + "epoch": 1.8582434514637904, + "grad_norm": 2.706331729888916, + "learning_rate": 2.1431290445374955e-06, + "loss": 1.4846, + "step": 1809 + }, + { + "epoch": 1.8592706728299948, + "grad_norm": 2.683302402496338, + "learning_rate": 2.141225732775029e-06, + "loss": 1.4501, + "step": 1810 + }, + { + "epoch": 1.8592706728299948, + "eval_loss": 1.8919310569763184, + "eval_runtime": 43.2772, + "eval_samples_per_second": 23.107, + "eval_steps_per_second": 0.97, + "step": 1810 + }, + { + "epoch": 1.8602978941961994, + "grad_norm": 2.9528095722198486, + "learning_rate": 2.1393224210125623e-06, + "loss": 1.4895, + "step": 1811 + }, + { + "epoch": 1.8613251155624035, + "grad_norm": 2.6797986030578613, + "learning_rate": 2.137419109250095e-06, + "loss": 1.4815, + "step": 1812 + }, + { + "epoch": 1.8623523369286081, + "grad_norm": 2.753924608230591, + "learning_rate": 2.1355157974876286e-06, + "loss": 1.5224, + "step": 1813 + }, + { + "epoch": 1.8633795582948125, + "grad_norm": 2.8037896156311035, + "learning_rate": 2.133612485725162e-06, + "loss": 1.5203, + "step": 1814 + }, + { + "epoch": 1.8644067796610169, + "grad_norm": 2.7934775352478027, + "learning_rate": 2.1317091739626953e-06, + "loss": 1.4307, + "step": 1815 + }, + { + "epoch": 1.8654340010272215, + "grad_norm": 2.919184684753418, + "learning_rate": 2.1298058622002286e-06, + "loss": 1.4654, + "step": 1816 + }, + { + "epoch": 1.8664612223934258, + "grad_norm": 2.671267509460449, + "learning_rate": 2.1279025504377616e-06, + "loss": 1.4776, + "step": 1817 + }, + { + "epoch": 1.8674884437596302, + "grad_norm": 2.6870152950286865, + "learning_rate": 2.125999238675295e-06, + "loss": 1.4759, + "step": 1818 + }, + { + "epoch": 1.8685156651258348, + "grad_norm": 2.8859241008758545, + "learning_rate": 2.1240959269128287e-06, + "loss": 1.515, + "step": 1819 + }, + { + "epoch": 1.869542886492039, + "grad_norm": 2.773226261138916, + "learning_rate": 2.122192615150362e-06, + "loss": 1.4743, + "step": 1820 + }, + { + "epoch": 1.869542886492039, + "eval_loss": 1.8901108503341675, + "eval_runtime": 45.5952, + "eval_samples_per_second": 21.932, + "eval_steps_per_second": 0.921, + "step": 1820 + }, + { + "epoch": 1.8705701078582435, + "grad_norm": 2.591169595718384, + "learning_rate": 2.120289303387895e-06, + "loss": 1.4245, + "step": 1821 + }, + { + "epoch": 1.8715973292244479, + "grad_norm": 2.977644205093384, + "learning_rate": 2.1183859916254284e-06, + "loss": 1.4755, + "step": 1822 + }, + { + "epoch": 1.8726245505906522, + "grad_norm": 2.7917141914367676, + "learning_rate": 2.1164826798629618e-06, + "loss": 1.5061, + "step": 1823 + }, + { + "epoch": 1.8736517719568568, + "grad_norm": 2.6995670795440674, + "learning_rate": 2.114579368100495e-06, + "loss": 1.4627, + "step": 1824 + }, + { + "epoch": 1.874678993323061, + "grad_norm": 2.7223551273345947, + "learning_rate": 2.1126760563380285e-06, + "loss": 1.4495, + "step": 1825 + }, + { + "epoch": 1.8757062146892656, + "grad_norm": 2.749028205871582, + "learning_rate": 2.1107727445755614e-06, + "loss": 1.4262, + "step": 1826 + }, + { + "epoch": 1.87673343605547, + "grad_norm": 3.0993876457214355, + "learning_rate": 2.1088694328130948e-06, + "loss": 1.5554, + "step": 1827 + }, + { + "epoch": 1.8777606574216743, + "grad_norm": 2.6830661296844482, + "learning_rate": 2.106966121050628e-06, + "loss": 1.4404, + "step": 1828 + }, + { + "epoch": 1.878787878787879, + "grad_norm": 2.7380309104919434, + "learning_rate": 2.1050628092881615e-06, + "loss": 1.5113, + "step": 1829 + }, + { + "epoch": 1.879815100154083, + "grad_norm": 2.873230457305908, + "learning_rate": 2.103159497525695e-06, + "loss": 1.545, + "step": 1830 + }, + { + "epoch": 1.879815100154083, + "eval_loss": 1.8950482606887817, + "eval_runtime": 46.8358, + "eval_samples_per_second": 21.351, + "eval_steps_per_second": 0.897, + "step": 1830 + }, + { + "epoch": 1.8808423215202876, + "grad_norm": 2.805757522583008, + "learning_rate": 2.1012561857632282e-06, + "loss": 1.4593, + "step": 1831 + }, + { + "epoch": 1.881869542886492, + "grad_norm": 2.7594594955444336, + "learning_rate": 2.0993528740007616e-06, + "loss": 1.5167, + "step": 1832 + }, + { + "epoch": 1.8828967642526964, + "grad_norm": 2.810119152069092, + "learning_rate": 2.097449562238295e-06, + "loss": 1.5199, + "step": 1833 + }, + { + "epoch": 1.883923985618901, + "grad_norm": 2.6915268898010254, + "learning_rate": 2.095546250475828e-06, + "loss": 1.47, + "step": 1834 + }, + { + "epoch": 1.8849512069851053, + "grad_norm": 2.966466188430786, + "learning_rate": 2.0936429387133613e-06, + "loss": 1.4815, + "step": 1835 + }, + { + "epoch": 1.8859784283513097, + "grad_norm": 2.719665050506592, + "learning_rate": 2.0917396269508946e-06, + "loss": 1.537, + "step": 1836 + }, + { + "epoch": 1.8870056497175143, + "grad_norm": 2.7843384742736816, + "learning_rate": 2.089836315188428e-06, + "loss": 1.5102, + "step": 1837 + }, + { + "epoch": 1.8880328710837184, + "grad_norm": 2.7335705757141113, + "learning_rate": 2.0879330034259614e-06, + "loss": 1.5199, + "step": 1838 + }, + { + "epoch": 1.889060092449923, + "grad_norm": 2.9173038005828857, + "learning_rate": 2.0860296916634947e-06, + "loss": 1.4605, + "step": 1839 + }, + { + "epoch": 1.8900873138161274, + "grad_norm": 2.7791881561279297, + "learning_rate": 2.084126379901028e-06, + "loss": 1.5482, + "step": 1840 + }, + { + "epoch": 1.8900873138161274, + "eval_loss": 1.8914867639541626, + "eval_runtime": 49.5764, + "eval_samples_per_second": 20.171, + "eval_steps_per_second": 0.847, + "step": 1840 + }, + { + "epoch": 1.8911145351823317, + "grad_norm": 2.700892686843872, + "learning_rate": 2.0822230681385614e-06, + "loss": 1.4509, + "step": 1841 + }, + { + "epoch": 1.8921417565485363, + "grad_norm": 2.9125330448150635, + "learning_rate": 2.080319756376095e-06, + "loss": 1.5081, + "step": 1842 + }, + { + "epoch": 1.8931689779147405, + "grad_norm": 2.762075424194336, + "learning_rate": 2.0784164446136277e-06, + "loss": 1.4989, + "step": 1843 + }, + { + "epoch": 1.894196199280945, + "grad_norm": 2.782738447189331, + "learning_rate": 2.076513132851161e-06, + "loss": 1.4643, + "step": 1844 + }, + { + "epoch": 1.8952234206471494, + "grad_norm": 2.7540040016174316, + "learning_rate": 2.0746098210886945e-06, + "loss": 1.5029, + "step": 1845 + }, + { + "epoch": 1.8962506420133538, + "grad_norm": 2.6205410957336426, + "learning_rate": 2.072706509326228e-06, + "loss": 1.4674, + "step": 1846 + }, + { + "epoch": 1.8972778633795584, + "grad_norm": 2.6946957111358643, + "learning_rate": 2.070803197563761e-06, + "loss": 1.4747, + "step": 1847 + }, + { + "epoch": 1.8983050847457628, + "grad_norm": 2.9114229679107666, + "learning_rate": 2.068899885801294e-06, + "loss": 1.4809, + "step": 1848 + }, + { + "epoch": 1.8993323061119671, + "grad_norm": 2.823211431503296, + "learning_rate": 2.066996574038828e-06, + "loss": 1.4887, + "step": 1849 + }, + { + "epoch": 1.9003595274781715, + "grad_norm": 2.779381513595581, + "learning_rate": 2.0650932622763613e-06, + "loss": 1.5217, + "step": 1850 + }, + { + "epoch": 1.9003595274781715, + "eval_loss": 1.8863105773925781, + "eval_runtime": 46.2064, + "eval_samples_per_second": 21.642, + "eval_steps_per_second": 0.909, + "step": 1850 + }, + { + "epoch": 1.9013867488443759, + "grad_norm": 2.7235658168792725, + "learning_rate": 2.0631899505138942e-06, + "loss": 1.4221, + "step": 1851 + }, + { + "epoch": 1.9024139702105805, + "grad_norm": 2.6994757652282715, + "learning_rate": 2.0612866387514276e-06, + "loss": 1.461, + "step": 1852 + }, + { + "epoch": 1.9034411915767848, + "grad_norm": 2.7634339332580566, + "learning_rate": 2.059383326988961e-06, + "loss": 1.4975, + "step": 1853 + }, + { + "epoch": 1.9044684129429892, + "grad_norm": 2.78421688079834, + "learning_rate": 2.0574800152264943e-06, + "loss": 1.4757, + "step": 1854 + }, + { + "epoch": 1.9054956343091938, + "grad_norm": 2.7357447147369385, + "learning_rate": 2.0555767034640277e-06, + "loss": 1.4524, + "step": 1855 + }, + { + "epoch": 1.906522855675398, + "grad_norm": 2.712550163269043, + "learning_rate": 2.053673391701561e-06, + "loss": 1.4232, + "step": 1856 + }, + { + "epoch": 1.9075500770416025, + "grad_norm": 2.7226524353027344, + "learning_rate": 2.051770079939094e-06, + "loss": 1.4846, + "step": 1857 + }, + { + "epoch": 1.9085772984078069, + "grad_norm": 2.762127637863159, + "learning_rate": 2.0498667681766273e-06, + "loss": 1.447, + "step": 1858 + }, + { + "epoch": 1.9096045197740112, + "grad_norm": 2.7561259269714355, + "learning_rate": 2.047963456414161e-06, + "loss": 1.4512, + "step": 1859 + }, + { + "epoch": 1.9106317411402158, + "grad_norm": 2.7032148838043213, + "learning_rate": 2.046060144651694e-06, + "loss": 1.4783, + "step": 1860 + }, + { + "epoch": 1.9106317411402158, + "eval_loss": 1.8838176727294922, + "eval_runtime": 46.0974, + "eval_samples_per_second": 21.693, + "eval_steps_per_second": 0.911, + "step": 1860 + }, + { + "epoch": 1.91165896250642, + "grad_norm": 2.8010241985321045, + "learning_rate": 2.0441568328892274e-06, + "loss": 1.4898, + "step": 1861 + }, + { + "epoch": 1.9126861838726246, + "grad_norm": 2.7780871391296387, + "learning_rate": 2.0422535211267608e-06, + "loss": 1.4848, + "step": 1862 + }, + { + "epoch": 1.913713405238829, + "grad_norm": 2.712059259414673, + "learning_rate": 2.040350209364294e-06, + "loss": 1.459, + "step": 1863 + }, + { + "epoch": 1.9147406266050333, + "grad_norm": 2.642489194869995, + "learning_rate": 2.0384468976018275e-06, + "loss": 1.4534, + "step": 1864 + }, + { + "epoch": 1.915767847971238, + "grad_norm": 2.645132064819336, + "learning_rate": 2.0365435858393604e-06, + "loss": 1.4573, + "step": 1865 + }, + { + "epoch": 1.9167950693374423, + "grad_norm": 2.747443914413452, + "learning_rate": 2.034640274076894e-06, + "loss": 1.4505, + "step": 1866 + }, + { + "epoch": 1.9178222907036466, + "grad_norm": 2.660188674926758, + "learning_rate": 2.032736962314427e-06, + "loss": 1.5146, + "step": 1867 + }, + { + "epoch": 1.9188495120698512, + "grad_norm": 2.7249484062194824, + "learning_rate": 2.0308336505519605e-06, + "loss": 1.4553, + "step": 1868 + }, + { + "epoch": 1.9198767334360554, + "grad_norm": 2.670896530151367, + "learning_rate": 2.028930338789494e-06, + "loss": 1.5204, + "step": 1869 + }, + { + "epoch": 1.92090395480226, + "grad_norm": 2.858780860900879, + "learning_rate": 2.0270270270270273e-06, + "loss": 1.446, + "step": 1870 + }, + { + "epoch": 1.92090395480226, + "eval_loss": 1.8811051845550537, + "eval_runtime": 45.7097, + "eval_samples_per_second": 21.877, + "eval_steps_per_second": 0.919, + "step": 1870 + }, + { + "epoch": 1.9219311761684643, + "grad_norm": 2.744133949279785, + "learning_rate": 2.0251237152645606e-06, + "loss": 1.4666, + "step": 1871 + }, + { + "epoch": 1.9229583975346687, + "grad_norm": 2.747067928314209, + "learning_rate": 2.023220403502094e-06, + "loss": 1.4466, + "step": 1872 + }, + { + "epoch": 1.9239856189008733, + "grad_norm": 2.7786505222320557, + "learning_rate": 2.0213170917396273e-06, + "loss": 1.4233, + "step": 1873 + }, + { + "epoch": 1.9250128402670774, + "grad_norm": 2.793522357940674, + "learning_rate": 2.0194137799771603e-06, + "loss": 1.4683, + "step": 1874 + }, + { + "epoch": 1.926040061633282, + "grad_norm": 2.7715535163879395, + "learning_rate": 2.0175104682146937e-06, + "loss": 1.4696, + "step": 1875 + }, + { + "epoch": 1.9270672829994864, + "grad_norm": 2.7226290702819824, + "learning_rate": 2.015607156452227e-06, + "loss": 1.412, + "step": 1876 + }, + { + "epoch": 1.9280945043656907, + "grad_norm": 2.722703456878662, + "learning_rate": 2.0137038446897604e-06, + "loss": 1.4503, + "step": 1877 + }, + { + "epoch": 1.9291217257318953, + "grad_norm": 2.717355251312256, + "learning_rate": 2.0118005329272937e-06, + "loss": 1.4423, + "step": 1878 + }, + { + "epoch": 1.9301489470980995, + "grad_norm": 2.719848394393921, + "learning_rate": 2.0098972211648267e-06, + "loss": 1.4609, + "step": 1879 + }, + { + "epoch": 1.931176168464304, + "grad_norm": 2.693844795227051, + "learning_rate": 2.0079939094023605e-06, + "loss": 1.4606, + "step": 1880 + }, + { + "epoch": 1.931176168464304, + "eval_loss": 1.8892173767089844, + "eval_runtime": 45.2703, + "eval_samples_per_second": 22.09, + "eval_steps_per_second": 0.928, + "step": 1880 + }, + { + "epoch": 1.9322033898305084, + "grad_norm": 2.649186372756958, + "learning_rate": 2.006090597639894e-06, + "loss": 1.5056, + "step": 1881 + }, + { + "epoch": 1.9332306111967128, + "grad_norm": 2.738987922668457, + "learning_rate": 2.0041872858774268e-06, + "loss": 1.5321, + "step": 1882 + }, + { + "epoch": 1.9342578325629174, + "grad_norm": 2.6583259105682373, + "learning_rate": 2.00228397411496e-06, + "loss": 1.5004, + "step": 1883 + }, + { + "epoch": 1.9352850539291218, + "grad_norm": 2.625006675720215, + "learning_rate": 2.0003806623524935e-06, + "loss": 1.4897, + "step": 1884 + }, + { + "epoch": 1.9363122752953261, + "grad_norm": 2.6956043243408203, + "learning_rate": 1.998477350590027e-06, + "loss": 1.4049, + "step": 1885 + }, + { + "epoch": 1.9373394966615307, + "grad_norm": 2.668865203857422, + "learning_rate": 1.9965740388275602e-06, + "loss": 1.4999, + "step": 1886 + }, + { + "epoch": 1.9383667180277349, + "grad_norm": 2.623006820678711, + "learning_rate": 1.994670727065093e-06, + "loss": 1.4742, + "step": 1887 + }, + { + "epoch": 1.9393939393939394, + "grad_norm": 2.7890007495880127, + "learning_rate": 1.9927674153026265e-06, + "loss": 1.4332, + "step": 1888 + }, + { + "epoch": 1.9404211607601438, + "grad_norm": 2.7803823947906494, + "learning_rate": 1.99086410354016e-06, + "loss": 1.5016, + "step": 1889 + }, + { + "epoch": 1.9414483821263482, + "grad_norm": 2.801680088043213, + "learning_rate": 1.9889607917776937e-06, + "loss": 1.4577, + "step": 1890 + }, + { + "epoch": 1.9414483821263482, + "eval_loss": 1.8877300024032593, + "eval_runtime": 43.6778, + "eval_samples_per_second": 22.895, + "eval_steps_per_second": 0.962, + "step": 1890 + }, + { + "epoch": 1.9424756034925528, + "grad_norm": 2.954679489135742, + "learning_rate": 1.9870574800152266e-06, + "loss": 1.4811, + "step": 1891 + }, + { + "epoch": 1.943502824858757, + "grad_norm": 2.6189258098602295, + "learning_rate": 1.98515416825276e-06, + "loss": 1.4955, + "step": 1892 + }, + { + "epoch": 1.9445300462249615, + "grad_norm": 2.7194509506225586, + "learning_rate": 1.9832508564902933e-06, + "loss": 1.4833, + "step": 1893 + }, + { + "epoch": 1.9455572675911659, + "grad_norm": 2.5767593383789062, + "learning_rate": 1.9813475447278267e-06, + "loss": 1.3973, + "step": 1894 + }, + { + "epoch": 1.9465844889573702, + "grad_norm": 2.7513067722320557, + "learning_rate": 1.97944423296536e-06, + "loss": 1.4814, + "step": 1895 + }, + { + "epoch": 1.9476117103235748, + "grad_norm": 2.712155818939209, + "learning_rate": 1.977540921202893e-06, + "loss": 1.4686, + "step": 1896 + }, + { + "epoch": 1.9486389316897792, + "grad_norm": 2.582233190536499, + "learning_rate": 1.9756376094404264e-06, + "loss": 1.4418, + "step": 1897 + }, + { + "epoch": 1.9496661530559836, + "grad_norm": 2.6065144538879395, + "learning_rate": 1.9737342976779597e-06, + "loss": 1.4198, + "step": 1898 + }, + { + "epoch": 1.950693374422188, + "grad_norm": 2.7891361713409424, + "learning_rate": 1.971830985915493e-06, + "loss": 1.5054, + "step": 1899 + }, + { + "epoch": 1.9517205957883923, + "grad_norm": 2.7199501991271973, + "learning_rate": 1.9699276741530264e-06, + "loss": 1.462, + "step": 1900 + }, + { + "epoch": 1.9517205957883923, + "eval_loss": 1.8931901454925537, + "eval_runtime": 44.0781, + "eval_samples_per_second": 22.687, + "eval_steps_per_second": 0.953, + "step": 1900 + }, + { + "epoch": 1.9527478171545969, + "grad_norm": 2.742461681365967, + "learning_rate": 1.96802436239056e-06, + "loss": 1.4716, + "step": 1901 + }, + { + "epoch": 1.9537750385208013, + "grad_norm": 2.689281702041626, + "learning_rate": 1.966121050628093e-06, + "loss": 1.4613, + "step": 1902 + }, + { + "epoch": 1.9548022598870056, + "grad_norm": 2.722919225692749, + "learning_rate": 1.9642177388656265e-06, + "loss": 1.4901, + "step": 1903 + }, + { + "epoch": 1.9558294812532102, + "grad_norm": 2.773724317550659, + "learning_rate": 1.9623144271031595e-06, + "loss": 1.5116, + "step": 1904 + }, + { + "epoch": 1.9568567026194144, + "grad_norm": 2.7967145442962646, + "learning_rate": 1.960411115340693e-06, + "loss": 1.5249, + "step": 1905 + }, + { + "epoch": 1.957883923985619, + "grad_norm": 2.7624928951263428, + "learning_rate": 1.958507803578226e-06, + "loss": 1.4751, + "step": 1906 + }, + { + "epoch": 1.9589111453518233, + "grad_norm": 2.787565231323242, + "learning_rate": 1.9566044918157596e-06, + "loss": 1.4878, + "step": 1907 + }, + { + "epoch": 1.9599383667180277, + "grad_norm": 2.6808853149414062, + "learning_rate": 1.954701180053293e-06, + "loss": 1.4963, + "step": 1908 + }, + { + "epoch": 1.9609655880842323, + "grad_norm": 2.7684340476989746, + "learning_rate": 1.952797868290826e-06, + "loss": 1.4528, + "step": 1909 + }, + { + "epoch": 1.9619928094504364, + "grad_norm": 2.7576661109924316, + "learning_rate": 1.9508945565283596e-06, + "loss": 1.5019, + "step": 1910 + }, + { + "epoch": 1.9619928094504364, + "eval_loss": 1.8857277631759644, + "eval_runtime": 43.0174, + "eval_samples_per_second": 23.246, + "eval_steps_per_second": 0.976, + "step": 1910 + }, + { + "epoch": 1.963020030816641, + "grad_norm": 2.6631407737731934, + "learning_rate": 1.948991244765893e-06, + "loss": 1.4638, + "step": 1911 + }, + { + "epoch": 1.9640472521828454, + "grad_norm": 2.720525026321411, + "learning_rate": 1.9470879330034264e-06, + "loss": 1.4951, + "step": 1912 + }, + { + "epoch": 1.9650744735490497, + "grad_norm": 2.780932903289795, + "learning_rate": 1.9451846212409593e-06, + "loss": 1.4351, + "step": 1913 + }, + { + "epoch": 1.9661016949152543, + "grad_norm": 2.8517262935638428, + "learning_rate": 1.9432813094784927e-06, + "loss": 1.4941, + "step": 1914 + }, + { + "epoch": 1.9671289162814587, + "grad_norm": 2.7812676429748535, + "learning_rate": 1.941377997716026e-06, + "loss": 1.4383, + "step": 1915 + }, + { + "epoch": 1.968156137647663, + "grad_norm": 2.815864086151123, + "learning_rate": 1.9394746859535594e-06, + "loss": 1.4562, + "step": 1916 + }, + { + "epoch": 1.9691833590138677, + "grad_norm": 2.841853618621826, + "learning_rate": 1.9375713741910928e-06, + "loss": 1.4289, + "step": 1917 + }, + { + "epoch": 1.9702105803800718, + "grad_norm": 2.751080274581909, + "learning_rate": 1.9356680624286257e-06, + "loss": 1.4567, + "step": 1918 + }, + { + "epoch": 1.9712378017462764, + "grad_norm": 2.7139065265655518, + "learning_rate": 1.933764750666159e-06, + "loss": 1.4158, + "step": 1919 + }, + { + "epoch": 1.9722650231124808, + "grad_norm": 2.6788330078125, + "learning_rate": 1.931861438903693e-06, + "loss": 1.4535, + "step": 1920 + }, + { + "epoch": 1.9722650231124808, + "eval_loss": 1.8828307390213013, + "eval_runtime": 44.8729, + "eval_samples_per_second": 22.285, + "eval_steps_per_second": 0.936, + "step": 1920 + }, + { + "epoch": 1.9732922444786851, + "grad_norm": 2.695631742477417, + "learning_rate": 1.929958127141226e-06, + "loss": 1.4348, + "step": 1921 + }, + { + "epoch": 1.9743194658448897, + "grad_norm": 2.6957788467407227, + "learning_rate": 1.928054815378759e-06, + "loss": 1.4627, + "step": 1922 + }, + { + "epoch": 1.9753466872110939, + "grad_norm": 2.8832242488861084, + "learning_rate": 1.9261515036162925e-06, + "loss": 1.5429, + "step": 1923 + }, + { + "epoch": 1.9763739085772984, + "grad_norm": 2.744378089904785, + "learning_rate": 1.924248191853826e-06, + "loss": 1.4973, + "step": 1924 + }, + { + "epoch": 1.9774011299435028, + "grad_norm": 2.695059299468994, + "learning_rate": 1.9223448800913592e-06, + "loss": 1.484, + "step": 1925 + }, + { + "epoch": 1.9784283513097072, + "grad_norm": 2.719637870788574, + "learning_rate": 1.9204415683288926e-06, + "loss": 1.4536, + "step": 1926 + }, + { + "epoch": 1.9794555726759118, + "grad_norm": 2.7248711585998535, + "learning_rate": 1.9185382565664255e-06, + "loss": 1.502, + "step": 1927 + }, + { + "epoch": 1.980482794042116, + "grad_norm": 2.6778104305267334, + "learning_rate": 1.916634944803959e-06, + "loss": 1.5065, + "step": 1928 + }, + { + "epoch": 1.9815100154083205, + "grad_norm": 2.859102964401245, + "learning_rate": 1.9147316330414923e-06, + "loss": 1.4791, + "step": 1929 + }, + { + "epoch": 1.9825372367745249, + "grad_norm": 2.708515167236328, + "learning_rate": 1.9128283212790256e-06, + "loss": 1.4693, + "step": 1930 + }, + { + "epoch": 1.9825372367745249, + "eval_loss": 1.8875067234039307, + "eval_runtime": 44.2114, + "eval_samples_per_second": 22.619, + "eval_steps_per_second": 0.95, + "step": 1930 + }, + { + "epoch": 1.9835644581407292, + "grad_norm": 2.6972970962524414, + "learning_rate": 1.910925009516559e-06, + "loss": 1.5149, + "step": 1931 + }, + { + "epoch": 1.9845916795069338, + "grad_norm": 2.800899028778076, + "learning_rate": 1.9090216977540924e-06, + "loss": 1.4785, + "step": 1932 + }, + { + "epoch": 1.9856189008731382, + "grad_norm": 2.8238046169281006, + "learning_rate": 1.9071183859916257e-06, + "loss": 1.5203, + "step": 1933 + }, + { + "epoch": 1.9866461222393426, + "grad_norm": 2.6949214935302734, + "learning_rate": 1.9052150742291589e-06, + "loss": 1.4919, + "step": 1934 + }, + { + "epoch": 1.9876733436055471, + "grad_norm": 2.7611637115478516, + "learning_rate": 1.9033117624666922e-06, + "loss": 1.4889, + "step": 1935 + }, + { + "epoch": 1.9887005649717513, + "grad_norm": 2.704080581665039, + "learning_rate": 1.9014084507042254e-06, + "loss": 1.4598, + "step": 1936 + }, + { + "epoch": 1.9897277863379559, + "grad_norm": 2.8081612586975098, + "learning_rate": 1.8995051389417587e-06, + "loss": 1.4324, + "step": 1937 + }, + { + "epoch": 1.9907550077041603, + "grad_norm": 2.597720146179199, + "learning_rate": 1.8976018271792921e-06, + "loss": 1.4196, + "step": 1938 + }, + { + "epoch": 1.9917822290703646, + "grad_norm": 2.716373920440674, + "learning_rate": 1.8956985154168253e-06, + "loss": 1.4795, + "step": 1939 + }, + { + "epoch": 1.9928094504365692, + "grad_norm": 2.7661051750183105, + "learning_rate": 1.8937952036543588e-06, + "loss": 1.457, + "step": 1940 + }, + { + "epoch": 1.9928094504365692, + "eval_loss": 1.8831615447998047, + "eval_runtime": 46.9632, + "eval_samples_per_second": 21.293, + "eval_steps_per_second": 0.894, + "step": 1940 + }, + { + "epoch": 1.9938366718027734, + "grad_norm": 2.66249942779541, + "learning_rate": 1.8918918918918922e-06, + "loss": 1.4213, + "step": 1941 + }, + { + "epoch": 1.994863893168978, + "grad_norm": 2.7329354286193848, + "learning_rate": 1.8899885801294253e-06, + "loss": 1.4707, + "step": 1942 + }, + { + "epoch": 1.9958911145351823, + "grad_norm": 2.844203472137451, + "learning_rate": 1.8880852683669587e-06, + "loss": 1.4847, + "step": 1943 + }, + { + "epoch": 1.9969183359013867, + "grad_norm": 2.9034390449523926, + "learning_rate": 1.886181956604492e-06, + "loss": 1.4605, + "step": 1944 + }, + { + "epoch": 1.9979455572675913, + "grad_norm": 2.8142337799072266, + "learning_rate": 1.8842786448420252e-06, + "loss": 1.4529, + "step": 1945 + }, + { + "epoch": 1.9989727786337956, + "grad_norm": 2.8397836685180664, + "learning_rate": 1.8823753330795586e-06, + "loss": 1.4863, + "step": 1946 + }, + { + "epoch": 2.0, + "grad_norm": 2.79947829246521, + "learning_rate": 1.8804720213170917e-06, + "loss": 1.3827, + "step": 1947 + }, + { + "epoch": 2.0010272213662046, + "grad_norm": 2.900805711746216, + "learning_rate": 1.878568709554625e-06, + "loss": 1.4183, + "step": 1948 + }, + { + "epoch": 2.0020544427324087, + "grad_norm": 2.8362643718719482, + "learning_rate": 1.8766653977921585e-06, + "loss": 1.468, + "step": 1949 + }, + { + "epoch": 2.0030816640986133, + "grad_norm": 2.774122476577759, + "learning_rate": 1.874762086029692e-06, + "loss": 1.4352, + "step": 1950 + }, + { + "epoch": 2.0030816640986133, + "eval_loss": 1.8892052173614502, + "eval_runtime": 49.6699, + "eval_samples_per_second": 20.133, + "eval_steps_per_second": 0.846, + "step": 1950 + }, + { + "epoch": 2.0041088854648175, + "grad_norm": 3.0558278560638428, + "learning_rate": 1.8728587742672252e-06, + "loss": 1.4593, + "step": 1951 + }, + { + "epoch": 2.005136106831022, + "grad_norm": 3.044110059738159, + "learning_rate": 1.8709554625047585e-06, + "loss": 1.4679, + "step": 1952 + }, + { + "epoch": 2.0061633281972266, + "grad_norm": 3.015784740447998, + "learning_rate": 1.8690521507422917e-06, + "loss": 1.4053, + "step": 1953 + }, + { + "epoch": 2.007190549563431, + "grad_norm": 3.1507861614227295, + "learning_rate": 1.867148838979825e-06, + "loss": 1.4289, + "step": 1954 + }, + { + "epoch": 2.0082177709296354, + "grad_norm": 2.9931421279907227, + "learning_rate": 1.8652455272173584e-06, + "loss": 1.4559, + "step": 1955 + }, + { + "epoch": 2.00924499229584, + "grad_norm": 3.0338571071624756, + "learning_rate": 1.8633422154548916e-06, + "loss": 1.473, + "step": 1956 + }, + { + "epoch": 2.010272213662044, + "grad_norm": 3.1007537841796875, + "learning_rate": 1.861438903692425e-06, + "loss": 1.3715, + "step": 1957 + }, + { + "epoch": 2.0112994350282487, + "grad_norm": 3.1187386512756348, + "learning_rate": 1.859535591929958e-06, + "loss": 1.4007, + "step": 1958 + }, + { + "epoch": 2.012326656394453, + "grad_norm": 3.045088529586792, + "learning_rate": 1.8576322801674915e-06, + "loss": 1.4866, + "step": 1959 + }, + { + "epoch": 2.0133538777606574, + "grad_norm": 2.9711339473724365, + "learning_rate": 1.855728968405025e-06, + "loss": 1.4193, + "step": 1960 + }, + { + "epoch": 2.0133538777606574, + "eval_loss": 1.8801918029785156, + "eval_runtime": 51.4737, + "eval_samples_per_second": 19.427, + "eval_steps_per_second": 0.816, + "step": 1960 + }, + { + "epoch": 2.014381099126862, + "grad_norm": 3.008835554122925, + "learning_rate": 1.8538256566425584e-06, + "loss": 1.4409, + "step": 1961 + }, + { + "epoch": 2.015408320493066, + "grad_norm": 3.013974189758301, + "learning_rate": 1.8519223448800915e-06, + "loss": 1.4708, + "step": 1962 + }, + { + "epoch": 2.0164355418592708, + "grad_norm": 3.00911545753479, + "learning_rate": 1.850019033117625e-06, + "loss": 1.4169, + "step": 1963 + }, + { + "epoch": 2.017462763225475, + "grad_norm": 2.9539992809295654, + "learning_rate": 1.848115721355158e-06, + "loss": 1.4172, + "step": 1964 + }, + { + "epoch": 2.0184899845916795, + "grad_norm": 3.2330169677734375, + "learning_rate": 1.8462124095926914e-06, + "loss": 1.4372, + "step": 1965 + }, + { + "epoch": 2.019517205957884, + "grad_norm": 3.1095242500305176, + "learning_rate": 1.8443090978302248e-06, + "loss": 1.4376, + "step": 1966 + }, + { + "epoch": 2.0205444273240882, + "grad_norm": 3.032968044281006, + "learning_rate": 1.842405786067758e-06, + "loss": 1.4784, + "step": 1967 + }, + { + "epoch": 2.021571648690293, + "grad_norm": 3.0439071655273438, + "learning_rate": 1.8405024743052913e-06, + "loss": 1.418, + "step": 1968 + }, + { + "epoch": 2.022598870056497, + "grad_norm": 3.0753591060638428, + "learning_rate": 1.8385991625428247e-06, + "loss": 1.4687, + "step": 1969 + }, + { + "epoch": 2.0236260914227016, + "grad_norm": 3.0760700702667236, + "learning_rate": 1.8366958507803578e-06, + "loss": 1.397, + "step": 1970 + }, + { + "epoch": 2.0236260914227016, + "eval_loss": 1.885207176208496, + "eval_runtime": 51.5554, + "eval_samples_per_second": 19.397, + "eval_steps_per_second": 0.815, + "step": 1970 + }, + { + "epoch": 2.024653312788906, + "grad_norm": 3.0004823207855225, + "learning_rate": 1.8347925390178914e-06, + "loss": 1.4413, + "step": 1971 + }, + { + "epoch": 2.0256805341551103, + "grad_norm": 2.861964702606201, + "learning_rate": 1.8328892272554247e-06, + "loss": 1.3885, + "step": 1972 + }, + { + "epoch": 2.026707755521315, + "grad_norm": 3.034388542175293, + "learning_rate": 1.8309859154929579e-06, + "loss": 1.4579, + "step": 1973 + }, + { + "epoch": 2.0277349768875195, + "grad_norm": 3.1062700748443604, + "learning_rate": 1.8290826037304913e-06, + "loss": 1.4396, + "step": 1974 + }, + { + "epoch": 2.0287621982537236, + "grad_norm": 2.986480712890625, + "learning_rate": 1.8271792919680246e-06, + "loss": 1.4048, + "step": 1975 + }, + { + "epoch": 2.029789419619928, + "grad_norm": 3.0232272148132324, + "learning_rate": 1.8252759802055578e-06, + "loss": 1.4107, + "step": 1976 + }, + { + "epoch": 2.0308166409861323, + "grad_norm": 3.131545305252075, + "learning_rate": 1.8233726684430911e-06, + "loss": 1.4532, + "step": 1977 + }, + { + "epoch": 2.031843862352337, + "grad_norm": 3.1322755813598633, + "learning_rate": 1.8214693566806243e-06, + "loss": 1.3931, + "step": 1978 + }, + { + "epoch": 2.0328710837185415, + "grad_norm": 3.06524395942688, + "learning_rate": 1.8195660449181576e-06, + "loss": 1.4203, + "step": 1979 + }, + { + "epoch": 2.0338983050847457, + "grad_norm": 3.1509804725646973, + "learning_rate": 1.817662733155691e-06, + "loss": 1.4451, + "step": 1980 + }, + { + "epoch": 2.0338983050847457, + "eval_loss": 1.8945039510726929, + "eval_runtime": 47.4904, + "eval_samples_per_second": 21.057, + "eval_steps_per_second": 0.884, + "step": 1980 + }, + { + "epoch": 2.0349255264509503, + "grad_norm": 3.01600980758667, + "learning_rate": 1.8157594213932244e-06, + "loss": 1.4225, + "step": 1981 + }, + { + "epoch": 2.0359527478171544, + "grad_norm": 3.092637777328491, + "learning_rate": 1.8138561096307577e-06, + "loss": 1.4495, + "step": 1982 + }, + { + "epoch": 2.036979969183359, + "grad_norm": 2.8800899982452393, + "learning_rate": 1.811952797868291e-06, + "loss": 1.3792, + "step": 1983 + }, + { + "epoch": 2.0380071905495636, + "grad_norm": 2.91048002243042, + "learning_rate": 1.8100494861058242e-06, + "loss": 1.427, + "step": 1984 + }, + { + "epoch": 2.0390344119157677, + "grad_norm": 2.9576427936553955, + "learning_rate": 1.8081461743433576e-06, + "loss": 1.4328, + "step": 1985 + }, + { + "epoch": 2.0400616332819723, + "grad_norm": 2.950599193572998, + "learning_rate": 1.806242862580891e-06, + "loss": 1.3877, + "step": 1986 + }, + { + "epoch": 2.0410888546481765, + "grad_norm": 3.0447452068328857, + "learning_rate": 1.8043395508184241e-06, + "loss": 1.4442, + "step": 1987 + }, + { + "epoch": 2.042116076014381, + "grad_norm": 3.058427095413208, + "learning_rate": 1.8024362390559575e-06, + "loss": 1.438, + "step": 1988 + }, + { + "epoch": 2.0431432973805856, + "grad_norm": 3.0351500511169434, + "learning_rate": 1.8005329272934906e-06, + "loss": 1.4007, + "step": 1989 + }, + { + "epoch": 2.04417051874679, + "grad_norm": 3.2143735885620117, + "learning_rate": 1.798629615531024e-06, + "loss": 1.4968, + "step": 1990 + }, + { + "epoch": 2.04417051874679, + "eval_loss": 1.895965337753296, + "eval_runtime": 47.6538, + "eval_samples_per_second": 20.985, + "eval_steps_per_second": 0.881, + "step": 1990 + }, + { + "epoch": 2.0451977401129944, + "grad_norm": 3.207671642303467, + "learning_rate": 1.7967263037685576e-06, + "loss": 1.419, + "step": 1991 + }, + { + "epoch": 2.046224961479199, + "grad_norm": 3.161807060241699, + "learning_rate": 1.794822992006091e-06, + "loss": 1.4193, + "step": 1992 + }, + { + "epoch": 2.047252182845403, + "grad_norm": 3.1695258617401123, + "learning_rate": 1.792919680243624e-06, + "loss": 1.4708, + "step": 1993 + }, + { + "epoch": 2.0482794042116077, + "grad_norm": 3.241506814956665, + "learning_rate": 1.7910163684811574e-06, + "loss": 1.3984, + "step": 1994 + }, + { + "epoch": 2.049306625577812, + "grad_norm": 3.014082193374634, + "learning_rate": 1.7891130567186906e-06, + "loss": 1.4208, + "step": 1995 + }, + { + "epoch": 2.0503338469440164, + "grad_norm": 3.0239710807800293, + "learning_rate": 1.787209744956224e-06, + "loss": 1.4036, + "step": 1996 + }, + { + "epoch": 2.051361068310221, + "grad_norm": 3.0406906604766846, + "learning_rate": 1.7853064331937573e-06, + "loss": 1.4266, + "step": 1997 + }, + { + "epoch": 2.052388289676425, + "grad_norm": 3.000096559524536, + "learning_rate": 1.7834031214312905e-06, + "loss": 1.4619, + "step": 1998 + }, + { + "epoch": 2.0534155110426298, + "grad_norm": 3.042457103729248, + "learning_rate": 1.7814998096688238e-06, + "loss": 1.3968, + "step": 1999 + }, + { + "epoch": 2.054442732408834, + "grad_norm": 3.0115537643432617, + "learning_rate": 1.779596497906357e-06, + "loss": 1.4404, + "step": 2000 + }, + { + "epoch": 2.054442732408834, + "eval_loss": 1.8855949640274048, + "eval_runtime": 49.9663, + "eval_samples_per_second": 20.013, + "eval_steps_per_second": 0.841, + "step": 2000 + }, + { + "epoch": 2.0554699537750385, + "grad_norm": 3.127013921737671, + "learning_rate": 1.7776931861438906e-06, + "loss": 1.4841, + "step": 2001 + }, + { + "epoch": 2.056497175141243, + "grad_norm": 3.080503463745117, + "learning_rate": 1.775789874381424e-06, + "loss": 1.477, + "step": 2002 + }, + { + "epoch": 2.0575243965074472, + "grad_norm": 3.193058967590332, + "learning_rate": 1.7738865626189573e-06, + "loss": 1.4539, + "step": 2003 + }, + { + "epoch": 2.058551617873652, + "grad_norm": 3.063589334487915, + "learning_rate": 1.7719832508564904e-06, + "loss": 1.3891, + "step": 2004 + }, + { + "epoch": 2.0595788392398564, + "grad_norm": 3.1374058723449707, + "learning_rate": 1.7700799390940238e-06, + "loss": 1.4131, + "step": 2005 + }, + { + "epoch": 2.0606060606060606, + "grad_norm": 3.029297351837158, + "learning_rate": 1.768176627331557e-06, + "loss": 1.4519, + "step": 2006 + }, + { + "epoch": 2.061633281972265, + "grad_norm": 3.0728025436401367, + "learning_rate": 1.7662733155690903e-06, + "loss": 1.434, + "step": 2007 + }, + { + "epoch": 2.0626605033384693, + "grad_norm": 3.0600240230560303, + "learning_rate": 1.7643700038066237e-06, + "loss": 1.464, + "step": 2008 + }, + { + "epoch": 2.063687724704674, + "grad_norm": 2.991814136505127, + "learning_rate": 1.7624666920441568e-06, + "loss": 1.3938, + "step": 2009 + }, + { + "epoch": 2.0647149460708785, + "grad_norm": 2.9615304470062256, + "learning_rate": 1.7605633802816902e-06, + "loss": 1.3883, + "step": 2010 + }, + { + "epoch": 2.0647149460708785, + "eval_loss": 1.8773977756500244, + "eval_runtime": 43.5833, + "eval_samples_per_second": 22.945, + "eval_steps_per_second": 0.964, + "step": 2010 + }, + { + "epoch": 2.0657421674370826, + "grad_norm": 3.108255624771118, + "learning_rate": 1.7586600685192238e-06, + "loss": 1.4956, + "step": 2011 + }, + { + "epoch": 2.066769388803287, + "grad_norm": 3.156050205230713, + "learning_rate": 1.756756756756757e-06, + "loss": 1.4922, + "step": 2012 + }, + { + "epoch": 2.0677966101694913, + "grad_norm": 3.24027943611145, + "learning_rate": 1.7548534449942903e-06, + "loss": 1.4504, + "step": 2013 + }, + { + "epoch": 2.068823831535696, + "grad_norm": 3.1180083751678467, + "learning_rate": 1.7529501332318236e-06, + "loss": 1.4183, + "step": 2014 + }, + { + "epoch": 2.0698510529019005, + "grad_norm": 3.178832769393921, + "learning_rate": 1.7510468214693568e-06, + "loss": 1.4178, + "step": 2015 + }, + { + "epoch": 2.0708782742681047, + "grad_norm": 3.058708906173706, + "learning_rate": 1.7491435097068902e-06, + "loss": 1.404, + "step": 2016 + }, + { + "epoch": 2.0719054956343093, + "grad_norm": 3.176765203475952, + "learning_rate": 1.7472401979444233e-06, + "loss": 1.4439, + "step": 2017 + }, + { + "epoch": 2.0729327170005134, + "grad_norm": 3.0423498153686523, + "learning_rate": 1.7453368861819567e-06, + "loss": 1.3661, + "step": 2018 + }, + { + "epoch": 2.073959938366718, + "grad_norm": 3.056716203689575, + "learning_rate": 1.74343357441949e-06, + "loss": 1.4196, + "step": 2019 + }, + { + "epoch": 2.0749871597329226, + "grad_norm": 3.0644123554229736, + "learning_rate": 1.7415302626570232e-06, + "loss": 1.4825, + "step": 2020 + }, + { + "epoch": 2.0749871597329226, + "eval_loss": 1.89279043674469, + "eval_runtime": 46.9776, + "eval_samples_per_second": 21.287, + "eval_steps_per_second": 0.894, + "step": 2020 + }, + { + "epoch": 2.0760143810991267, + "grad_norm": 2.9771924018859863, + "learning_rate": 1.7396269508945568e-06, + "loss": 1.4163, + "step": 2021 + }, + { + "epoch": 2.0770416024653313, + "grad_norm": 2.9193663597106934, + "learning_rate": 1.7377236391320901e-06, + "loss": 1.386, + "step": 2022 + }, + { + "epoch": 2.078068823831536, + "grad_norm": 3.194058418273926, + "learning_rate": 1.7358203273696233e-06, + "loss": 1.4696, + "step": 2023 + }, + { + "epoch": 2.07909604519774, + "grad_norm": 3.1295535564422607, + "learning_rate": 1.7339170156071566e-06, + "loss": 1.4252, + "step": 2024 + }, + { + "epoch": 2.0801232665639446, + "grad_norm": 2.948204755783081, + "learning_rate": 1.73201370384469e-06, + "loss": 1.416, + "step": 2025 + }, + { + "epoch": 2.081150487930149, + "grad_norm": 3.064087390899658, + "learning_rate": 1.7301103920822231e-06, + "loss": 1.431, + "step": 2026 + }, + { + "epoch": 2.0821777092963534, + "grad_norm": 3.230952739715576, + "learning_rate": 1.7282070803197565e-06, + "loss": 1.4454, + "step": 2027 + }, + { + "epoch": 2.083204930662558, + "grad_norm": 3.1373074054718018, + "learning_rate": 1.7263037685572897e-06, + "loss": 1.4218, + "step": 2028 + }, + { + "epoch": 2.084232152028762, + "grad_norm": 3.2426247596740723, + "learning_rate": 1.724400456794823e-06, + "loss": 1.4481, + "step": 2029 + }, + { + "epoch": 2.0852593733949667, + "grad_norm": 2.8744704723358154, + "learning_rate": 1.7224971450323564e-06, + "loss": 1.406, + "step": 2030 + }, + { + "epoch": 2.0852593733949667, + "eval_loss": 1.8881611824035645, + "eval_runtime": 49.2991, + "eval_samples_per_second": 20.284, + "eval_steps_per_second": 0.852, + "step": 2030 + }, + { + "epoch": 2.086286594761171, + "grad_norm": 3.0375564098358154, + "learning_rate": 1.72059383326989e-06, + "loss": 1.4187, + "step": 2031 + }, + { + "epoch": 2.0873138161273754, + "grad_norm": 3.069279909133911, + "learning_rate": 1.7186905215074231e-06, + "loss": 1.4847, + "step": 2032 + }, + { + "epoch": 2.08834103749358, + "grad_norm": 3.0621039867401123, + "learning_rate": 1.7167872097449565e-06, + "loss": 1.4048, + "step": 2033 + }, + { + "epoch": 2.089368258859784, + "grad_norm": 3.137110948562622, + "learning_rate": 1.7148838979824896e-06, + "loss": 1.4272, + "step": 2034 + }, + { + "epoch": 2.0903954802259888, + "grad_norm": 3.1717305183410645, + "learning_rate": 1.712980586220023e-06, + "loss": 1.424, + "step": 2035 + }, + { + "epoch": 2.091422701592193, + "grad_norm": 3.0675606727600098, + "learning_rate": 1.7110772744575563e-06, + "loss": 1.4076, + "step": 2036 + }, + { + "epoch": 2.0924499229583975, + "grad_norm": 3.0585849285125732, + "learning_rate": 1.7091739626950895e-06, + "loss": 1.4205, + "step": 2037 + }, + { + "epoch": 2.093477144324602, + "grad_norm": 3.1450932025909424, + "learning_rate": 1.7072706509326229e-06, + "loss": 1.4557, + "step": 2038 + }, + { + "epoch": 2.0945043656908062, + "grad_norm": 3.0912528038024902, + "learning_rate": 1.7053673391701562e-06, + "loss": 1.4335, + "step": 2039 + }, + { + "epoch": 2.095531587057011, + "grad_norm": 2.9372971057891846, + "learning_rate": 1.7034640274076894e-06, + "loss": 1.3959, + "step": 2040 + }, + { + "epoch": 2.095531587057011, + "eval_loss": 1.8857965469360352, + "eval_runtime": 50.3859, + "eval_samples_per_second": 19.847, + "eval_steps_per_second": 0.834, + "step": 2040 + }, + { + "epoch": 2.0965588084232154, + "grad_norm": 3.127795457839966, + "learning_rate": 1.701560715645223e-06, + "loss": 1.4741, + "step": 2041 + }, + { + "epoch": 2.0975860297894195, + "grad_norm": 3.1425771713256836, + "learning_rate": 1.6996574038827563e-06, + "loss": 1.3868, + "step": 2042 + }, + { + "epoch": 2.098613251155624, + "grad_norm": 3.158665180206299, + "learning_rate": 1.6977540921202895e-06, + "loss": 1.434, + "step": 2043 + }, + { + "epoch": 2.0996404725218283, + "grad_norm": 3.060809850692749, + "learning_rate": 1.6958507803578228e-06, + "loss": 1.4093, + "step": 2044 + }, + { + "epoch": 2.100667693888033, + "grad_norm": 3.138719320297241, + "learning_rate": 1.693947468595356e-06, + "loss": 1.4494, + "step": 2045 + }, + { + "epoch": 2.1016949152542375, + "grad_norm": 3.235567092895508, + "learning_rate": 1.6920441568328893e-06, + "loss": 1.4497, + "step": 2046 + }, + { + "epoch": 2.1027221366204416, + "grad_norm": 3.1233010292053223, + "learning_rate": 1.6901408450704227e-06, + "loss": 1.3903, + "step": 2047 + }, + { + "epoch": 2.103749357986646, + "grad_norm": 3.095599412918091, + "learning_rate": 1.6882375333079559e-06, + "loss": 1.416, + "step": 2048 + }, + { + "epoch": 2.1047765793528503, + "grad_norm": 3.266810178756714, + "learning_rate": 1.6863342215454892e-06, + "loss": 1.4676, + "step": 2049 + }, + { + "epoch": 2.105803800719055, + "grad_norm": 3.2977075576782227, + "learning_rate": 1.6844309097830226e-06, + "loss": 1.4641, + "step": 2050 + }, + { + "epoch": 2.105803800719055, + "eval_loss": 1.8875017166137695, + "eval_runtime": 44.2952, + "eval_samples_per_second": 22.576, + "eval_steps_per_second": 0.948, + "step": 2050 + }, + { + "epoch": 2.1068310220852595, + "grad_norm": 3.06746244430542, + "learning_rate": 1.682527598020556e-06, + "loss": 1.4443, + "step": 2051 + }, + { + "epoch": 2.1078582434514637, + "grad_norm": 3.0481631755828857, + "learning_rate": 1.6806242862580893e-06, + "loss": 1.4728, + "step": 2052 + }, + { + "epoch": 2.1088854648176683, + "grad_norm": 3.090182065963745, + "learning_rate": 1.6787209744956227e-06, + "loss": 1.3925, + "step": 2053 + }, + { + "epoch": 2.109912686183873, + "grad_norm": 3.0113272666931152, + "learning_rate": 1.6768176627331558e-06, + "loss": 1.3924, + "step": 2054 + }, + { + "epoch": 2.110939907550077, + "grad_norm": 3.0159542560577393, + "learning_rate": 1.6749143509706892e-06, + "loss": 1.4297, + "step": 2055 + }, + { + "epoch": 2.1119671289162816, + "grad_norm": 3.0569703578948975, + "learning_rate": 1.6730110392082225e-06, + "loss": 1.425, + "step": 2056 + }, + { + "epoch": 2.1129943502824857, + "grad_norm": 3.1391549110412598, + "learning_rate": 1.6711077274457557e-06, + "loss": 1.4357, + "step": 2057 + }, + { + "epoch": 2.1140215716486903, + "grad_norm": 3.0112249851226807, + "learning_rate": 1.669204415683289e-06, + "loss": 1.3948, + "step": 2058 + }, + { + "epoch": 2.115048793014895, + "grad_norm": 3.143165111541748, + "learning_rate": 1.6673011039208222e-06, + "loss": 1.4502, + "step": 2059 + }, + { + "epoch": 2.116076014381099, + "grad_norm": 3.030191659927368, + "learning_rate": 1.6653977921583556e-06, + "loss": 1.43, + "step": 2060 + }, + { + "epoch": 2.116076014381099, + "eval_loss": 1.885430932044983, + "eval_runtime": 44.6532, + "eval_samples_per_second": 22.395, + "eval_steps_per_second": 0.941, + "step": 2060 + }, + { + "epoch": 2.1171032357473036, + "grad_norm": 3.1698837280273438, + "learning_rate": 1.663494480395889e-06, + "loss": 1.4461, + "step": 2061 + }, + { + "epoch": 2.118130457113508, + "grad_norm": 3.0927395820617676, + "learning_rate": 1.6615911686334225e-06, + "loss": 1.3957, + "step": 2062 + }, + { + "epoch": 2.1191576784797124, + "grad_norm": 3.049832582473755, + "learning_rate": 1.6596878568709557e-06, + "loss": 1.4269, + "step": 2063 + }, + { + "epoch": 2.120184899845917, + "grad_norm": 3.031977653503418, + "learning_rate": 1.657784545108489e-06, + "loss": 1.4141, + "step": 2064 + }, + { + "epoch": 2.121212121212121, + "grad_norm": 3.0946176052093506, + "learning_rate": 1.6558812333460222e-06, + "loss": 1.4677, + "step": 2065 + }, + { + "epoch": 2.1222393425783257, + "grad_norm": 3.1852216720581055, + "learning_rate": 1.6539779215835555e-06, + "loss": 1.4476, + "step": 2066 + }, + { + "epoch": 2.12326656394453, + "grad_norm": 3.1202027797698975, + "learning_rate": 1.652074609821089e-06, + "loss": 1.4107, + "step": 2067 + }, + { + "epoch": 2.1242937853107344, + "grad_norm": 2.9965174198150635, + "learning_rate": 1.650171298058622e-06, + "loss": 1.4118, + "step": 2068 + }, + { + "epoch": 2.125321006676939, + "grad_norm": 3.142627239227295, + "learning_rate": 1.6482679862961554e-06, + "loss": 1.4148, + "step": 2069 + }, + { + "epoch": 2.126348228043143, + "grad_norm": 3.114840030670166, + "learning_rate": 1.6463646745336886e-06, + "loss": 1.4104, + "step": 2070 + }, + { + "epoch": 2.126348228043143, + "eval_loss": 1.8847553730010986, + "eval_runtime": 45.4766, + "eval_samples_per_second": 21.989, + "eval_steps_per_second": 0.924, + "step": 2070 + }, + { + "epoch": 2.1273754494093478, + "grad_norm": 3.0997679233551025, + "learning_rate": 1.644461362771222e-06, + "loss": 1.3903, + "step": 2071 + }, + { + "epoch": 2.128402670775552, + "grad_norm": 3.053497791290283, + "learning_rate": 1.6425580510087555e-06, + "loss": 1.4362, + "step": 2072 + }, + { + "epoch": 2.1294298921417565, + "grad_norm": 3.1247191429138184, + "learning_rate": 1.6406547392462889e-06, + "loss": 1.395, + "step": 2073 + }, + { + "epoch": 2.130457113507961, + "grad_norm": 3.0684449672698975, + "learning_rate": 1.638751427483822e-06, + "loss": 1.4, + "step": 2074 + }, + { + "epoch": 2.131484334874165, + "grad_norm": 3.0759975910186768, + "learning_rate": 1.6368481157213554e-06, + "loss": 1.4301, + "step": 2075 + }, + { + "epoch": 2.13251155624037, + "grad_norm": 3.1344354152679443, + "learning_rate": 1.6349448039588885e-06, + "loss": 1.4165, + "step": 2076 + }, + { + "epoch": 2.1335387776065744, + "grad_norm": 3.1342315673828125, + "learning_rate": 1.6330414921964219e-06, + "loss": 1.4811, + "step": 2077 + }, + { + "epoch": 2.1345659989727785, + "grad_norm": 2.9999516010284424, + "learning_rate": 1.6311381804339553e-06, + "loss": 1.3974, + "step": 2078 + }, + { + "epoch": 2.135593220338983, + "grad_norm": 3.10968279838562, + "learning_rate": 1.6292348686714884e-06, + "loss": 1.4308, + "step": 2079 + }, + { + "epoch": 2.1366204417051873, + "grad_norm": 3.123739004135132, + "learning_rate": 1.6273315569090218e-06, + "loss": 1.4116, + "step": 2080 + }, + { + "epoch": 2.1366204417051873, + "eval_loss": 1.8859853744506836, + "eval_runtime": 43.5954, + "eval_samples_per_second": 22.938, + "eval_steps_per_second": 0.963, + "step": 2080 + }, + { + "epoch": 2.137647663071392, + "grad_norm": 3.1350209712982178, + "learning_rate": 1.625428245146555e-06, + "loss": 1.4914, + "step": 2081 + }, + { + "epoch": 2.1386748844375965, + "grad_norm": 3.1349897384643555, + "learning_rate": 1.6235249333840885e-06, + "loss": 1.4681, + "step": 2082 + }, + { + "epoch": 2.1397021058038006, + "grad_norm": 3.1517136096954346, + "learning_rate": 1.6216216216216219e-06, + "loss": 1.4287, + "step": 2083 + }, + { + "epoch": 2.140729327170005, + "grad_norm": 3.109863758087158, + "learning_rate": 1.6197183098591552e-06, + "loss": 1.4557, + "step": 2084 + }, + { + "epoch": 2.1417565485362093, + "grad_norm": 3.017601728439331, + "learning_rate": 1.6178149980966884e-06, + "loss": 1.435, + "step": 2085 + }, + { + "epoch": 2.142783769902414, + "grad_norm": 3.0578653812408447, + "learning_rate": 1.6159116863342217e-06, + "loss": 1.4078, + "step": 2086 + }, + { + "epoch": 2.1438109912686185, + "grad_norm": 3.1092753410339355, + "learning_rate": 1.6140083745717549e-06, + "loss": 1.4809, + "step": 2087 + }, + { + "epoch": 2.1448382126348227, + "grad_norm": 3.012550115585327, + "learning_rate": 1.6121050628092882e-06, + "loss": 1.3883, + "step": 2088 + }, + { + "epoch": 2.1458654340010272, + "grad_norm": 3.239058256149292, + "learning_rate": 1.6102017510468216e-06, + "loss": 1.4733, + "step": 2089 + }, + { + "epoch": 2.146892655367232, + "grad_norm": 3.0761263370513916, + "learning_rate": 1.6082984392843548e-06, + "loss": 1.377, + "step": 2090 + }, + { + "epoch": 2.146892655367232, + "eval_loss": 1.8865137100219727, + "eval_runtime": 43.4996, + "eval_samples_per_second": 22.989, + "eval_steps_per_second": 0.966, + "step": 2090 + }, + { + "epoch": 2.147919876733436, + "grad_norm": 3.125427007675171, + "learning_rate": 1.6063951275218881e-06, + "loss": 1.4381, + "step": 2091 + }, + { + "epoch": 2.1489470980996406, + "grad_norm": 3.1415202617645264, + "learning_rate": 1.6044918157594217e-06, + "loss": 1.4823, + "step": 2092 + }, + { + "epoch": 2.1499743194658447, + "grad_norm": 3.061314821243286, + "learning_rate": 1.6025885039969548e-06, + "loss": 1.4177, + "step": 2093 + }, + { + "epoch": 2.1510015408320493, + "grad_norm": 3.0278573036193848, + "learning_rate": 1.6006851922344882e-06, + "loss": 1.4283, + "step": 2094 + }, + { + "epoch": 2.152028762198254, + "grad_norm": 3.0583488941192627, + "learning_rate": 1.5987818804720216e-06, + "loss": 1.4453, + "step": 2095 + }, + { + "epoch": 2.153055983564458, + "grad_norm": 3.149545907974243, + "learning_rate": 1.5968785687095547e-06, + "loss": 1.4481, + "step": 2096 + }, + { + "epoch": 2.1540832049306626, + "grad_norm": 3.031721591949463, + "learning_rate": 1.594975256947088e-06, + "loss": 1.3923, + "step": 2097 + }, + { + "epoch": 2.1551104262968668, + "grad_norm": 3.0269832611083984, + "learning_rate": 1.5930719451846212e-06, + "loss": 1.4489, + "step": 2098 + }, + { + "epoch": 2.1561376476630714, + "grad_norm": 3.0105485916137695, + "learning_rate": 1.5911686334221546e-06, + "loss": 1.3967, + "step": 2099 + }, + { + "epoch": 2.157164869029276, + "grad_norm": 3.170759916305542, + "learning_rate": 1.589265321659688e-06, + "loss": 1.4467, + "step": 2100 + }, + { + "epoch": 2.157164869029276, + "eval_loss": 1.887832760810852, + "eval_runtime": 44.2197, + "eval_samples_per_second": 22.614, + "eval_steps_per_second": 0.95, + "step": 2100 + }, + { + "epoch": 2.15819209039548, + "grad_norm": 2.9956490993499756, + "learning_rate": 1.5873620098972211e-06, + "loss": 1.3601, + "step": 2101 + }, + { + "epoch": 2.1592193117616847, + "grad_norm": 3.118528366088867, + "learning_rate": 1.5854586981347547e-06, + "loss": 1.4053, + "step": 2102 + }, + { + "epoch": 2.1602465331278893, + "grad_norm": 3.3036599159240723, + "learning_rate": 1.583555386372288e-06, + "loss": 1.4523, + "step": 2103 + }, + { + "epoch": 2.1612737544940934, + "grad_norm": 3.2302451133728027, + "learning_rate": 1.5816520746098212e-06, + "loss": 1.4646, + "step": 2104 + }, + { + "epoch": 2.162300975860298, + "grad_norm": 3.1311371326446533, + "learning_rate": 1.5797487628473546e-06, + "loss": 1.4233, + "step": 2105 + }, + { + "epoch": 2.163328197226502, + "grad_norm": 3.024657726287842, + "learning_rate": 1.577845451084888e-06, + "loss": 1.3791, + "step": 2106 + }, + { + "epoch": 2.1643554185927067, + "grad_norm": 3.1913206577301025, + "learning_rate": 1.575942139322421e-06, + "loss": 1.4284, + "step": 2107 + }, + { + "epoch": 2.1653826399589113, + "grad_norm": 3.010667085647583, + "learning_rate": 1.5740388275599544e-06, + "loss": 1.352, + "step": 2108 + }, + { + "epoch": 2.1664098613251155, + "grad_norm": 3.0839853286743164, + "learning_rate": 1.5721355157974876e-06, + "loss": 1.4342, + "step": 2109 + }, + { + "epoch": 2.16743708269132, + "grad_norm": 3.115177631378174, + "learning_rate": 1.570232204035021e-06, + "loss": 1.4188, + "step": 2110 + }, + { + "epoch": 2.16743708269132, + "eval_loss": 1.8869224786758423, + "eval_runtime": 42.2937, + "eval_samples_per_second": 23.644, + "eval_steps_per_second": 0.993, + "step": 2110 + }, + { + "epoch": 2.168464304057524, + "grad_norm": 3.1874091625213623, + "learning_rate": 1.5683288922725543e-06, + "loss": 1.4252, + "step": 2111 + }, + { + "epoch": 2.169491525423729, + "grad_norm": 3.149386405944824, + "learning_rate": 1.5664255805100879e-06, + "loss": 1.4252, + "step": 2112 + }, + { + "epoch": 2.1705187467899334, + "grad_norm": 3.1509039402008057, + "learning_rate": 1.564522268747621e-06, + "loss": 1.4135, + "step": 2113 + }, + { + "epoch": 2.1715459681561375, + "grad_norm": 3.1496448516845703, + "learning_rate": 1.5626189569851544e-06, + "loss": 1.4384, + "step": 2114 + }, + { + "epoch": 2.172573189522342, + "grad_norm": 2.9863228797912598, + "learning_rate": 1.5607156452226875e-06, + "loss": 1.3569, + "step": 2115 + }, + { + "epoch": 2.1736004108885463, + "grad_norm": 3.1577210426330566, + "learning_rate": 1.558812333460221e-06, + "loss": 1.4263, + "step": 2116 + }, + { + "epoch": 2.174627632254751, + "grad_norm": 3.079028844833374, + "learning_rate": 1.5569090216977543e-06, + "loss": 1.4363, + "step": 2117 + }, + { + "epoch": 2.1756548536209555, + "grad_norm": 3.0446267127990723, + "learning_rate": 1.5550057099352874e-06, + "loss": 1.4287, + "step": 2118 + }, + { + "epoch": 2.1766820749871596, + "grad_norm": 3.0535521507263184, + "learning_rate": 1.5531023981728208e-06, + "loss": 1.4457, + "step": 2119 + }, + { + "epoch": 2.177709296353364, + "grad_norm": 2.997999429702759, + "learning_rate": 1.5511990864103542e-06, + "loss": 1.4013, + "step": 2120 + }, + { + "epoch": 2.177709296353364, + "eval_loss": 1.8781853914260864, + "eval_runtime": 46.565, + "eval_samples_per_second": 21.475, + "eval_steps_per_second": 0.902, + "step": 2120 + }, + { + "epoch": 2.1787365177195683, + "grad_norm": 3.021111249923706, + "learning_rate": 1.5492957746478873e-06, + "loss": 1.4337, + "step": 2121 + }, + { + "epoch": 2.179763739085773, + "grad_norm": 3.147423267364502, + "learning_rate": 1.5473924628854209e-06, + "loss": 1.4566, + "step": 2122 + }, + { + "epoch": 2.1807909604519775, + "grad_norm": 3.2015597820281982, + "learning_rate": 1.5454891511229542e-06, + "loss": 1.4183, + "step": 2123 + }, + { + "epoch": 2.1818181818181817, + "grad_norm": 3.027254819869995, + "learning_rate": 1.5435858393604874e-06, + "loss": 1.4211, + "step": 2124 + }, + { + "epoch": 2.1828454031843862, + "grad_norm": 3.140397787094116, + "learning_rate": 1.5416825275980208e-06, + "loss": 1.4545, + "step": 2125 + }, + { + "epoch": 2.183872624550591, + "grad_norm": 3.087702751159668, + "learning_rate": 1.539779215835554e-06, + "loss": 1.4038, + "step": 2126 + }, + { + "epoch": 2.184899845916795, + "grad_norm": 3.1732728481292725, + "learning_rate": 1.5378759040730873e-06, + "loss": 1.4156, + "step": 2127 + }, + { + "epoch": 2.1859270672829996, + "grad_norm": 3.181870460510254, + "learning_rate": 1.5359725923106206e-06, + "loss": 1.468, + "step": 2128 + }, + { + "epoch": 2.1869542886492037, + "grad_norm": 3.111034393310547, + "learning_rate": 1.5340692805481538e-06, + "loss": 1.4402, + "step": 2129 + }, + { + "epoch": 2.1879815100154083, + "grad_norm": 3.07132887840271, + "learning_rate": 1.5321659687856871e-06, + "loss": 1.4072, + "step": 2130 + }, + { + "epoch": 2.1879815100154083, + "eval_loss": 1.8848137855529785, + "eval_runtime": 48.9628, + "eval_samples_per_second": 20.424, + "eval_steps_per_second": 0.858, + "step": 2130 + }, + { + "epoch": 2.189008731381613, + "grad_norm": 3.078758716583252, + "learning_rate": 1.5302626570232205e-06, + "loss": 1.4022, + "step": 2131 + }, + { + "epoch": 2.190035952747817, + "grad_norm": 3.104029655456543, + "learning_rate": 1.5283593452607539e-06, + "loss": 1.3619, + "step": 2132 + }, + { + "epoch": 2.1910631741140216, + "grad_norm": 3.167560577392578, + "learning_rate": 1.5264560334982872e-06, + "loss": 1.4784, + "step": 2133 + }, + { + "epoch": 2.1920903954802258, + "grad_norm": 3.055535078048706, + "learning_rate": 1.5245527217358206e-06, + "loss": 1.4053, + "step": 2134 + }, + { + "epoch": 2.1931176168464304, + "grad_norm": 3.1337225437164307, + "learning_rate": 1.5226494099733537e-06, + "loss": 1.4459, + "step": 2135 + }, + { + "epoch": 2.194144838212635, + "grad_norm": 3.0870962142944336, + "learning_rate": 1.520746098210887e-06, + "loss": 1.3966, + "step": 2136 + }, + { + "epoch": 2.195172059578839, + "grad_norm": 3.0952706336975098, + "learning_rate": 1.5188427864484205e-06, + "loss": 1.4625, + "step": 2137 + }, + { + "epoch": 2.1961992809450437, + "grad_norm": 3.064378023147583, + "learning_rate": 1.5169394746859536e-06, + "loss": 1.3913, + "step": 2138 + }, + { + "epoch": 2.1972265023112483, + "grad_norm": 3.181436538696289, + "learning_rate": 1.515036162923487e-06, + "loss": 1.4114, + "step": 2139 + }, + { + "epoch": 2.1982537236774524, + "grad_norm": 3.0721826553344727, + "learning_rate": 1.5131328511610201e-06, + "loss": 1.4337, + "step": 2140 + }, + { + "epoch": 2.1982537236774524, + "eval_loss": 1.8966083526611328, + "eval_runtime": 46.5515, + "eval_samples_per_second": 21.482, + "eval_steps_per_second": 0.902, + "step": 2140 + }, + { + "epoch": 2.199280945043657, + "grad_norm": 3.3798177242279053, + "learning_rate": 1.5112295393985535e-06, + "loss": 1.465, + "step": 2141 + }, + { + "epoch": 2.200308166409861, + "grad_norm": 3.2605955600738525, + "learning_rate": 1.509326227636087e-06, + "loss": 1.4434, + "step": 2142 + }, + { + "epoch": 2.2013353877760657, + "grad_norm": 3.14622163772583, + "learning_rate": 1.5074229158736204e-06, + "loss": 1.3743, + "step": 2143 + }, + { + "epoch": 2.2023626091422703, + "grad_norm": 3.0676040649414062, + "learning_rate": 1.5055196041111536e-06, + "loss": 1.4221, + "step": 2144 + }, + { + "epoch": 2.2033898305084745, + "grad_norm": 3.2191338539123535, + "learning_rate": 1.503616292348687e-06, + "loss": 1.4318, + "step": 2145 + }, + { + "epoch": 2.204417051874679, + "grad_norm": 3.293288230895996, + "learning_rate": 1.50171298058622e-06, + "loss": 1.4646, + "step": 2146 + }, + { + "epoch": 2.205444273240883, + "grad_norm": 3.0085601806640625, + "learning_rate": 1.4998096688237535e-06, + "loss": 1.3608, + "step": 2147 + }, + { + "epoch": 2.206471494607088, + "grad_norm": 3.1057655811309814, + "learning_rate": 1.4979063570612868e-06, + "loss": 1.4342, + "step": 2148 + }, + { + "epoch": 2.2074987159732924, + "grad_norm": 3.1611974239349365, + "learning_rate": 1.49600304529882e-06, + "loss": 1.4303, + "step": 2149 + }, + { + "epoch": 2.2085259373394965, + "grad_norm": 3.0769052505493164, + "learning_rate": 1.4940997335363533e-06, + "loss": 1.4049, + "step": 2150 + }, + { + "epoch": 2.2085259373394965, + "eval_loss": 1.8821698427200317, + "eval_runtime": 47.8527, + "eval_samples_per_second": 20.897, + "eval_steps_per_second": 0.878, + "step": 2150 + }, + { + "epoch": 2.209553158705701, + "grad_norm": 3.3353095054626465, + "learning_rate": 1.4921964217738865e-06, + "loss": 1.4316, + "step": 2151 + }, + { + "epoch": 2.2105803800719057, + "grad_norm": 3.109008550643921, + "learning_rate": 1.49029311001142e-06, + "loss": 1.4088, + "step": 2152 + }, + { + "epoch": 2.21160760143811, + "grad_norm": 3.315708875656128, + "learning_rate": 1.4883897982489534e-06, + "loss": 1.4372, + "step": 2153 + }, + { + "epoch": 2.2126348228043144, + "grad_norm": 3.41266131401062, + "learning_rate": 1.4864864864864868e-06, + "loss": 1.4516, + "step": 2154 + }, + { + "epoch": 2.2136620441705186, + "grad_norm": 3.206641435623169, + "learning_rate": 1.48458317472402e-06, + "loss": 1.4198, + "step": 2155 + }, + { + "epoch": 2.214689265536723, + "grad_norm": 3.3165881633758545, + "learning_rate": 1.4826798629615533e-06, + "loss": 1.4229, + "step": 2156 + }, + { + "epoch": 2.2157164869029278, + "grad_norm": 3.113398313522339, + "learning_rate": 1.4807765511990864e-06, + "loss": 1.3785, + "step": 2157 + }, + { + "epoch": 2.216743708269132, + "grad_norm": 3.155613422393799, + "learning_rate": 1.4788732394366198e-06, + "loss": 1.3985, + "step": 2158 + }, + { + "epoch": 2.2177709296353365, + "grad_norm": 3.1063179969787598, + "learning_rate": 1.4769699276741532e-06, + "loss": 1.4282, + "step": 2159 + }, + { + "epoch": 2.2187981510015407, + "grad_norm": 3.1550281047821045, + "learning_rate": 1.4750666159116863e-06, + "loss": 1.4086, + "step": 2160 + }, + { + "epoch": 2.2187981510015407, + "eval_loss": 1.8825618028640747, + "eval_runtime": 45.5113, + "eval_samples_per_second": 21.973, + "eval_steps_per_second": 0.923, + "step": 2160 + }, + { + "epoch": 2.2198253723677452, + "grad_norm": 3.0101521015167236, + "learning_rate": 1.4731633041492197e-06, + "loss": 1.3705, + "step": 2161 + }, + { + "epoch": 2.22085259373395, + "grad_norm": 3.0926272869110107, + "learning_rate": 1.4712599923867528e-06, + "loss": 1.443, + "step": 2162 + }, + { + "epoch": 2.221879815100154, + "grad_norm": 3.1270558834075928, + "learning_rate": 1.4693566806242864e-06, + "loss": 1.4641, + "step": 2163 + }, + { + "epoch": 2.2229070364663586, + "grad_norm": 2.9442856311798096, + "learning_rate": 1.4674533688618198e-06, + "loss": 1.3873, + "step": 2164 + }, + { + "epoch": 2.2239342578325627, + "grad_norm": 3.0549123287200928, + "learning_rate": 1.4655500570993531e-06, + "loss": 1.4236, + "step": 2165 + }, + { + "epoch": 2.2249614791987673, + "grad_norm": 3.1002092361450195, + "learning_rate": 1.4636467453368863e-06, + "loss": 1.4174, + "step": 2166 + }, + { + "epoch": 2.225988700564972, + "grad_norm": 3.118391990661621, + "learning_rate": 1.4617434335744197e-06, + "loss": 1.4783, + "step": 2167 + }, + { + "epoch": 2.227015921931176, + "grad_norm": 3.1835744380950928, + "learning_rate": 1.4598401218119528e-06, + "loss": 1.4542, + "step": 2168 + }, + { + "epoch": 2.2280431432973806, + "grad_norm": 3.1558971405029297, + "learning_rate": 1.4579368100494862e-06, + "loss": 1.4443, + "step": 2169 + }, + { + "epoch": 2.2290703646635848, + "grad_norm": 3.0535943508148193, + "learning_rate": 1.4560334982870195e-06, + "loss": 1.4209, + "step": 2170 + }, + { + "epoch": 2.2290703646635848, + "eval_loss": 1.882952094078064, + "eval_runtime": 47.6589, + "eval_samples_per_second": 20.982, + "eval_steps_per_second": 0.881, + "step": 2170 + }, + { + "epoch": 2.2300975860297894, + "grad_norm": 3.110412836074829, + "learning_rate": 1.4541301865245527e-06, + "loss": 1.432, + "step": 2171 + }, + { + "epoch": 2.231124807395994, + "grad_norm": 3.0594677925109863, + "learning_rate": 1.452226874762086e-06, + "loss": 1.4466, + "step": 2172 + }, + { + "epoch": 2.232152028762198, + "grad_norm": 3.1013123989105225, + "learning_rate": 1.4503235629996196e-06, + "loss": 1.4038, + "step": 2173 + }, + { + "epoch": 2.2331792501284027, + "grad_norm": 3.327085018157959, + "learning_rate": 1.4484202512371528e-06, + "loss": 1.4712, + "step": 2174 + }, + { + "epoch": 2.2342064714946073, + "grad_norm": 3.1586358547210693, + "learning_rate": 1.4465169394746861e-06, + "loss": 1.3606, + "step": 2175 + }, + { + "epoch": 2.2352336928608114, + "grad_norm": 3.200604200363159, + "learning_rate": 1.4446136277122195e-06, + "loss": 1.4792, + "step": 2176 + }, + { + "epoch": 2.236260914227016, + "grad_norm": 3.0358104705810547, + "learning_rate": 1.4427103159497526e-06, + "loss": 1.3812, + "step": 2177 + }, + { + "epoch": 2.23728813559322, + "grad_norm": 3.103911876678467, + "learning_rate": 1.440807004187286e-06, + "loss": 1.4049, + "step": 2178 + }, + { + "epoch": 2.2383153569594247, + "grad_norm": 3.0799686908721924, + "learning_rate": 1.4389036924248192e-06, + "loss": 1.437, + "step": 2179 + }, + { + "epoch": 2.2393425783256293, + "grad_norm": 3.0964274406433105, + "learning_rate": 1.4370003806623525e-06, + "loss": 1.4501, + "step": 2180 + }, + { + "epoch": 2.2393425783256293, + "eval_loss": 1.8832465410232544, + "eval_runtime": 50.3011, + "eval_samples_per_second": 19.88, + "eval_steps_per_second": 0.835, + "step": 2180 + }, + { + "epoch": 2.2403697996918335, + "grad_norm": 3.1642727851867676, + "learning_rate": 1.4350970688998859e-06, + "loss": 1.4198, + "step": 2181 + }, + { + "epoch": 2.241397021058038, + "grad_norm": 3.0271341800689697, + "learning_rate": 1.433193757137419e-06, + "loss": 1.4648, + "step": 2182 + }, + { + "epoch": 2.242424242424242, + "grad_norm": 2.9605119228363037, + "learning_rate": 1.4312904453749526e-06, + "loss": 1.4079, + "step": 2183 + }, + { + "epoch": 2.243451463790447, + "grad_norm": 3.034020185470581, + "learning_rate": 1.429387133612486e-06, + "loss": 1.3797, + "step": 2184 + }, + { + "epoch": 2.2444786851566514, + "grad_norm": 3.0324604511260986, + "learning_rate": 1.4274838218500191e-06, + "loss": 1.4268, + "step": 2185 + }, + { + "epoch": 2.2455059065228555, + "grad_norm": 3.260662317276001, + "learning_rate": 1.4255805100875525e-06, + "loss": 1.4302, + "step": 2186 + }, + { + "epoch": 2.24653312788906, + "grad_norm": 3.2712326049804688, + "learning_rate": 1.4236771983250858e-06, + "loss": 1.4863, + "step": 2187 + }, + { + "epoch": 2.2475603492552647, + "grad_norm": 3.1835451126098633, + "learning_rate": 1.421773886562619e-06, + "loss": 1.4176, + "step": 2188 + }, + { + "epoch": 2.248587570621469, + "grad_norm": 3.0387771129608154, + "learning_rate": 1.4198705748001524e-06, + "loss": 1.3861, + "step": 2189 + }, + { + "epoch": 2.2496147919876734, + "grad_norm": 3.074028968811035, + "learning_rate": 1.4179672630376855e-06, + "loss": 1.4074, + "step": 2190 + }, + { + "epoch": 2.2496147919876734, + "eval_loss": 1.8878331184387207, + "eval_runtime": 46.7284, + "eval_samples_per_second": 21.4, + "eval_steps_per_second": 0.899, + "step": 2190 + }, + { + "epoch": 2.2506420133538776, + "grad_norm": 3.131228446960449, + "learning_rate": 1.4160639512752189e-06, + "loss": 1.3913, + "step": 2191 + }, + { + "epoch": 2.251669234720082, + "grad_norm": 3.049567699432373, + "learning_rate": 1.4141606395127522e-06, + "loss": 1.4114, + "step": 2192 + }, + { + "epoch": 2.2526964560862868, + "grad_norm": 3.0833427906036377, + "learning_rate": 1.4122573277502858e-06, + "loss": 1.3855, + "step": 2193 + }, + { + "epoch": 2.253723677452491, + "grad_norm": 3.0678162574768066, + "learning_rate": 1.410354015987819e-06, + "loss": 1.4452, + "step": 2194 + }, + { + "epoch": 2.2547508988186955, + "grad_norm": 3.1660776138305664, + "learning_rate": 1.4084507042253523e-06, + "loss": 1.4169, + "step": 2195 + }, + { + "epoch": 2.2557781201848996, + "grad_norm": 2.9469871520996094, + "learning_rate": 1.4065473924628855e-06, + "loss": 1.443, + "step": 2196 + }, + { + "epoch": 2.2568053415511042, + "grad_norm": 2.9388298988342285, + "learning_rate": 1.4046440807004188e-06, + "loss": 1.411, + "step": 2197 + }, + { + "epoch": 2.257832562917309, + "grad_norm": 3.080441951751709, + "learning_rate": 1.4027407689379522e-06, + "loss": 1.4163, + "step": 2198 + }, + { + "epoch": 2.258859784283513, + "grad_norm": 3.086993932723999, + "learning_rate": 1.4008374571754854e-06, + "loss": 1.3996, + "step": 2199 + }, + { + "epoch": 2.2598870056497176, + "grad_norm": 3.1033880710601807, + "learning_rate": 1.3989341454130187e-06, + "loss": 1.4271, + "step": 2200 + }, + { + "epoch": 2.2598870056497176, + "eval_loss": 1.8867006301879883, + "eval_runtime": 47.3101, + "eval_samples_per_second": 21.137, + "eval_steps_per_second": 0.888, + "step": 2200 + }, + { + "epoch": 2.260914227015922, + "grad_norm": 3.137422561645508, + "learning_rate": 1.397030833650552e-06, + "loss": 1.4244, + "step": 2201 + }, + { + "epoch": 2.2619414483821263, + "grad_norm": 3.1550681591033936, + "learning_rate": 1.3951275218880852e-06, + "loss": 1.4401, + "step": 2202 + }, + { + "epoch": 2.262968669748331, + "grad_norm": 3.341092824935913, + "learning_rate": 1.3932242101256188e-06, + "loss": 1.505, + "step": 2203 + }, + { + "epoch": 2.263995891114535, + "grad_norm": 3.0959739685058594, + "learning_rate": 1.3913208983631522e-06, + "loss": 1.3924, + "step": 2204 + }, + { + "epoch": 2.2650231124807396, + "grad_norm": 3.121640920639038, + "learning_rate": 1.3894175866006853e-06, + "loss": 1.4263, + "step": 2205 + }, + { + "epoch": 2.266050333846944, + "grad_norm": 3.0325067043304443, + "learning_rate": 1.3875142748382187e-06, + "loss": 1.3724, + "step": 2206 + }, + { + "epoch": 2.2670775552131484, + "grad_norm": 3.130715847015381, + "learning_rate": 1.385610963075752e-06, + "loss": 1.3632, + "step": 2207 + }, + { + "epoch": 2.268104776579353, + "grad_norm": 3.10500431060791, + "learning_rate": 1.3837076513132852e-06, + "loss": 1.4227, + "step": 2208 + }, + { + "epoch": 2.269131997945557, + "grad_norm": 3.247288465499878, + "learning_rate": 1.3818043395508186e-06, + "loss": 1.4617, + "step": 2209 + }, + { + "epoch": 2.2701592193117617, + "grad_norm": 3.2476606369018555, + "learning_rate": 1.3799010277883517e-06, + "loss": 1.4325, + "step": 2210 + }, + { + "epoch": 2.2701592193117617, + "eval_loss": 1.8897888660430908, + "eval_runtime": 42.9294, + "eval_samples_per_second": 23.294, + "eval_steps_per_second": 0.978, + "step": 2210 + }, + { + "epoch": 2.2711864406779663, + "grad_norm": 3.0068607330322266, + "learning_rate": 1.377997716025885e-06, + "loss": 1.4119, + "step": 2211 + }, + { + "epoch": 2.2722136620441704, + "grad_norm": 3.185081958770752, + "learning_rate": 1.3760944042634184e-06, + "loss": 1.4334, + "step": 2212 + }, + { + "epoch": 2.273240883410375, + "grad_norm": 3.091458559036255, + "learning_rate": 1.3741910925009518e-06, + "loss": 1.4496, + "step": 2213 + }, + { + "epoch": 2.2742681047765796, + "grad_norm": 3.07018780708313, + "learning_rate": 1.3722877807384852e-06, + "loss": 1.4325, + "step": 2214 + }, + { + "epoch": 2.2752953261427837, + "grad_norm": 3.116154909133911, + "learning_rate": 1.3703844689760185e-06, + "loss": 1.4392, + "step": 2215 + }, + { + "epoch": 2.2763225475089883, + "grad_norm": 3.1154232025146484, + "learning_rate": 1.3684811572135517e-06, + "loss": 1.4062, + "step": 2216 + }, + { + "epoch": 2.2773497688751925, + "grad_norm": 3.0303139686584473, + "learning_rate": 1.366577845451085e-06, + "loss": 1.3569, + "step": 2217 + }, + { + "epoch": 2.278376990241397, + "grad_norm": 3.2322487831115723, + "learning_rate": 1.3646745336886184e-06, + "loss": 1.4498, + "step": 2218 + }, + { + "epoch": 2.279404211607601, + "grad_norm": 3.241386651992798, + "learning_rate": 1.3627712219261515e-06, + "loss": 1.4373, + "step": 2219 + }, + { + "epoch": 2.280431432973806, + "grad_norm": 3.052802801132202, + "learning_rate": 1.360867910163685e-06, + "loss": 1.3918, + "step": 2220 + }, + { + "epoch": 2.280431432973806, + "eval_loss": 1.882577657699585, + "eval_runtime": 48.945, + "eval_samples_per_second": 20.431, + "eval_steps_per_second": 0.858, + "step": 2220 + }, + { + "epoch": 2.2814586543400104, + "grad_norm": 3.256002426147461, + "learning_rate": 1.358964598401218e-06, + "loss": 1.4227, + "step": 2221 + }, + { + "epoch": 2.2824858757062145, + "grad_norm": 3.185438394546509, + "learning_rate": 1.3570612866387514e-06, + "loss": 1.423, + "step": 2222 + }, + { + "epoch": 2.283513097072419, + "grad_norm": 3.1974103450775146, + "learning_rate": 1.355157974876285e-06, + "loss": 1.4046, + "step": 2223 + }, + { + "epoch": 2.2845403184386237, + "grad_norm": 3.2578117847442627, + "learning_rate": 1.3532546631138184e-06, + "loss": 1.4692, + "step": 2224 + }, + { + "epoch": 2.285567539804828, + "grad_norm": 3.166027784347534, + "learning_rate": 1.3513513513513515e-06, + "loss": 1.4298, + "step": 2225 + }, + { + "epoch": 2.2865947611710324, + "grad_norm": 3.2637693881988525, + "learning_rate": 1.3494480395888849e-06, + "loss": 1.4132, + "step": 2226 + }, + { + "epoch": 2.2876219825372366, + "grad_norm": 3.124319314956665, + "learning_rate": 1.347544727826418e-06, + "loss": 1.4097, + "step": 2227 + }, + { + "epoch": 2.288649203903441, + "grad_norm": 3.164501428604126, + "learning_rate": 1.3456414160639514e-06, + "loss": 1.3947, + "step": 2228 + }, + { + "epoch": 2.2896764252696458, + "grad_norm": 3.2155773639678955, + "learning_rate": 1.3437381043014847e-06, + "loss": 1.4347, + "step": 2229 + }, + { + "epoch": 2.29070364663585, + "grad_norm": 3.2893707752227783, + "learning_rate": 1.341834792539018e-06, + "loss": 1.4258, + "step": 2230 + }, + { + "epoch": 2.29070364663585, + "eval_loss": 1.879811406135559, + "eval_runtime": 47.4777, + "eval_samples_per_second": 21.063, + "eval_steps_per_second": 0.885, + "step": 2230 + }, + { + "epoch": 2.2917308680020545, + "grad_norm": 3.095522403717041, + "learning_rate": 1.3399314807765513e-06, + "loss": 1.4227, + "step": 2231 + }, + { + "epoch": 2.2927580893682586, + "grad_norm": 3.1545920372009277, + "learning_rate": 1.3380281690140844e-06, + "loss": 1.4344, + "step": 2232 + }, + { + "epoch": 2.2937853107344632, + "grad_norm": 3.2042078971862793, + "learning_rate": 1.336124857251618e-06, + "loss": 1.4322, + "step": 2233 + }, + { + "epoch": 2.294812532100668, + "grad_norm": 3.162391424179077, + "learning_rate": 1.3342215454891513e-06, + "loss": 1.3924, + "step": 2234 + }, + { + "epoch": 2.295839753466872, + "grad_norm": 3.0768051147460938, + "learning_rate": 1.3323182337266847e-06, + "loss": 1.4, + "step": 2235 + }, + { + "epoch": 2.2968669748330766, + "grad_norm": 3.0797791481018066, + "learning_rate": 1.3304149219642179e-06, + "loss": 1.4268, + "step": 2236 + }, + { + "epoch": 2.297894196199281, + "grad_norm": 3.1587228775024414, + "learning_rate": 1.3285116102017512e-06, + "loss": 1.4665, + "step": 2237 + }, + { + "epoch": 2.2989214175654853, + "grad_norm": 3.0737040042877197, + "learning_rate": 1.3266082984392844e-06, + "loss": 1.4149, + "step": 2238 + }, + { + "epoch": 2.29994863893169, + "grad_norm": 3.1760494709014893, + "learning_rate": 1.3247049866768177e-06, + "loss": 1.4254, + "step": 2239 + }, + { + "epoch": 2.300975860297894, + "grad_norm": 3.1770107746124268, + "learning_rate": 1.322801674914351e-06, + "loss": 1.3979, + "step": 2240 + }, + { + "epoch": 2.300975860297894, + "eval_loss": 1.8801367282867432, + "eval_runtime": 50.137, + "eval_samples_per_second": 19.945, + "eval_steps_per_second": 0.838, + "step": 2240 + }, + { + "epoch": 2.3020030816640986, + "grad_norm": 3.215686559677124, + "learning_rate": 1.3208983631518843e-06, + "loss": 1.3948, + "step": 2241 + }, + { + "epoch": 2.303030303030303, + "grad_norm": 3.1983041763305664, + "learning_rate": 1.3189950513894176e-06, + "loss": 1.4276, + "step": 2242 + }, + { + "epoch": 2.3040575243965074, + "grad_norm": 3.2098212242126465, + "learning_rate": 1.3170917396269512e-06, + "loss": 1.4144, + "step": 2243 + }, + { + "epoch": 2.305084745762712, + "grad_norm": 3.170607089996338, + "learning_rate": 1.3151884278644843e-06, + "loss": 1.429, + "step": 2244 + }, + { + "epoch": 2.306111967128916, + "grad_norm": 3.1504342555999756, + "learning_rate": 1.3132851161020177e-06, + "loss": 1.4446, + "step": 2245 + }, + { + "epoch": 2.3071391884951207, + "grad_norm": 3.012388229370117, + "learning_rate": 1.311381804339551e-06, + "loss": 1.3926, + "step": 2246 + }, + { + "epoch": 2.3081664098613253, + "grad_norm": 3.1543102264404297, + "learning_rate": 1.3094784925770842e-06, + "loss": 1.4796, + "step": 2247 + }, + { + "epoch": 2.3091936312275294, + "grad_norm": 2.9769937992095947, + "learning_rate": 1.3075751808146176e-06, + "loss": 1.3916, + "step": 2248 + }, + { + "epoch": 2.310220852593734, + "grad_norm": 3.0640859603881836, + "learning_rate": 1.3056718690521507e-06, + "loss": 1.4255, + "step": 2249 + }, + { + "epoch": 2.3112480739599386, + "grad_norm": 2.9977078437805176, + "learning_rate": 1.303768557289684e-06, + "loss": 1.3964, + "step": 2250 + }, + { + "epoch": 2.3112480739599386, + "eval_loss": 1.8834030628204346, + "eval_runtime": 47.1318, + "eval_samples_per_second": 21.217, + "eval_steps_per_second": 0.891, + "step": 2250 + }, + { + "epoch": 2.3122752953261427, + "grad_norm": 3.0376105308532715, + "learning_rate": 1.3018652455272175e-06, + "loss": 1.4033, + "step": 2251 + }, + { + "epoch": 2.3133025166923473, + "grad_norm": 3.154709815979004, + "learning_rate": 1.2999619337647506e-06, + "loss": 1.4132, + "step": 2252 + }, + { + "epoch": 2.3143297380585515, + "grad_norm": 3.1585581302642822, + "learning_rate": 1.2980586220022842e-06, + "loss": 1.4045, + "step": 2253 + }, + { + "epoch": 2.315356959424756, + "grad_norm": 3.2512333393096924, + "learning_rate": 1.2961553102398175e-06, + "loss": 1.4431, + "step": 2254 + }, + { + "epoch": 2.3163841807909606, + "grad_norm": 3.093665599822998, + "learning_rate": 1.2942519984773507e-06, + "loss": 1.4519, + "step": 2255 + }, + { + "epoch": 2.317411402157165, + "grad_norm": 3.1790826320648193, + "learning_rate": 1.292348686714884e-06, + "loss": 1.4384, + "step": 2256 + }, + { + "epoch": 2.3184386235233694, + "grad_norm": 3.0824191570281982, + "learning_rate": 1.2904453749524174e-06, + "loss": 1.4011, + "step": 2257 + }, + { + "epoch": 2.3194658448895735, + "grad_norm": 3.218667507171631, + "learning_rate": 1.2885420631899506e-06, + "loss": 1.4446, + "step": 2258 + }, + { + "epoch": 2.320493066255778, + "grad_norm": 3.0636513233184814, + "learning_rate": 1.286638751427484e-06, + "loss": 1.3601, + "step": 2259 + }, + { + "epoch": 2.3215202876219827, + "grad_norm": 3.137908935546875, + "learning_rate": 1.284735439665017e-06, + "loss": 1.4033, + "step": 2260 + }, + { + "epoch": 2.3215202876219827, + "eval_loss": 1.8799747228622437, + "eval_runtime": 49.6332, + "eval_samples_per_second": 20.148, + "eval_steps_per_second": 0.846, + "step": 2260 + }, + { + "epoch": 2.322547508988187, + "grad_norm": 3.1707377433776855, + "learning_rate": 1.2828321279025504e-06, + "loss": 1.4306, + "step": 2261 + }, + { + "epoch": 2.3235747303543914, + "grad_norm": 3.2448198795318604, + "learning_rate": 1.2809288161400838e-06, + "loss": 1.4392, + "step": 2262 + }, + { + "epoch": 2.324601951720596, + "grad_norm": 3.115103006362915, + "learning_rate": 1.279025504377617e-06, + "loss": 1.4179, + "step": 2263 + }, + { + "epoch": 2.3256291730868, + "grad_norm": 3.1800553798675537, + "learning_rate": 1.2771221926151505e-06, + "loss": 1.3979, + "step": 2264 + }, + { + "epoch": 2.3266563944530048, + "grad_norm": 3.2946650981903076, + "learning_rate": 1.275218880852684e-06, + "loss": 1.4626, + "step": 2265 + }, + { + "epoch": 2.327683615819209, + "grad_norm": 3.1193125247955322, + "learning_rate": 1.273315569090217e-06, + "loss": 1.406, + "step": 2266 + }, + { + "epoch": 2.3287108371854135, + "grad_norm": 3.2309799194335938, + "learning_rate": 1.2714122573277504e-06, + "loss": 1.4179, + "step": 2267 + }, + { + "epoch": 2.3297380585516176, + "grad_norm": 3.1344757080078125, + "learning_rate": 1.2695089455652838e-06, + "loss": 1.4717, + "step": 2268 + }, + { + "epoch": 2.3307652799178222, + "grad_norm": 3.0551161766052246, + "learning_rate": 1.267605633802817e-06, + "loss": 1.4706, + "step": 2269 + }, + { + "epoch": 2.331792501284027, + "grad_norm": 3.1680619716644287, + "learning_rate": 1.2657023220403503e-06, + "loss": 1.4447, + "step": 2270 + }, + { + "epoch": 2.331792501284027, + "eval_loss": 1.872067928314209, + "eval_runtime": 44.2239, + "eval_samples_per_second": 22.612, + "eval_steps_per_second": 0.95, + "step": 2270 + }, + { + "epoch": 2.332819722650231, + "grad_norm": 3.0938210487365723, + "learning_rate": 1.2637990102778836e-06, + "loss": 1.4318, + "step": 2271 + }, + { + "epoch": 2.3338469440164356, + "grad_norm": 3.1710591316223145, + "learning_rate": 1.2618956985154168e-06, + "loss": 1.4182, + "step": 2272 + }, + { + "epoch": 2.33487416538264, + "grad_norm": 3.192415714263916, + "learning_rate": 1.2599923867529502e-06, + "loss": 1.423, + "step": 2273 + }, + { + "epoch": 2.3359013867488443, + "grad_norm": 3.219047784805298, + "learning_rate": 1.2580890749904837e-06, + "loss": 1.4457, + "step": 2274 + }, + { + "epoch": 2.336928608115049, + "grad_norm": 3.0913424491882324, + "learning_rate": 1.2561857632280169e-06, + "loss": 1.4174, + "step": 2275 + }, + { + "epoch": 2.337955829481253, + "grad_norm": 3.2987632751464844, + "learning_rate": 1.2542824514655502e-06, + "loss": 1.4551, + "step": 2276 + }, + { + "epoch": 2.3389830508474576, + "grad_norm": 3.5121114253997803, + "learning_rate": 1.2523791397030834e-06, + "loss": 1.4793, + "step": 2277 + }, + { + "epoch": 2.340010272213662, + "grad_norm": 3.3527626991271973, + "learning_rate": 1.2504758279406168e-06, + "loss": 1.4445, + "step": 2278 + }, + { + "epoch": 2.3410374935798663, + "grad_norm": 3.1356029510498047, + "learning_rate": 1.2485725161781501e-06, + "loss": 1.4062, + "step": 2279 + }, + { + "epoch": 2.342064714946071, + "grad_norm": 3.326028823852539, + "learning_rate": 1.2466692044156833e-06, + "loss": 1.4499, + "step": 2280 + }, + { + "epoch": 2.342064714946071, + "eval_loss": 1.8791024684906006, + "eval_runtime": 44.2336, + "eval_samples_per_second": 22.607, + "eval_steps_per_second": 0.95, + "step": 2280 + }, + { + "epoch": 2.343091936312275, + "grad_norm": 3.4187798500061035, + "learning_rate": 1.2447658926532168e-06, + "loss": 1.5179, + "step": 2281 + }, + { + "epoch": 2.3441191576784797, + "grad_norm": 3.1069252490997314, + "learning_rate": 1.24286258089075e-06, + "loss": 1.3825, + "step": 2282 + }, + { + "epoch": 2.3451463790446843, + "grad_norm": 2.9901397228240967, + "learning_rate": 1.2409592691282834e-06, + "loss": 1.4218, + "step": 2283 + }, + { + "epoch": 2.3461736004108884, + "grad_norm": 3.057490587234497, + "learning_rate": 1.2390559573658165e-06, + "loss": 1.4043, + "step": 2284 + }, + { + "epoch": 2.347200821777093, + "grad_norm": 3.265415906906128, + "learning_rate": 1.2371526456033499e-06, + "loss": 1.4315, + "step": 2285 + }, + { + "epoch": 2.3482280431432976, + "grad_norm": 3.2394862174987793, + "learning_rate": 1.2352493338408832e-06, + "loss": 1.4267, + "step": 2286 + }, + { + "epoch": 2.3492552645095017, + "grad_norm": 3.0940380096435547, + "learning_rate": 1.2333460220784166e-06, + "loss": 1.3947, + "step": 2287 + }, + { + "epoch": 2.3502824858757063, + "grad_norm": 3.230414628982544, + "learning_rate": 1.23144271031595e-06, + "loss": 1.5144, + "step": 2288 + }, + { + "epoch": 2.3513097072419105, + "grad_norm": 3.3229947090148926, + "learning_rate": 1.2295393985534831e-06, + "loss": 1.4413, + "step": 2289 + }, + { + "epoch": 2.352336928608115, + "grad_norm": 3.337873935699463, + "learning_rate": 1.2276360867910165e-06, + "loss": 1.461, + "step": 2290 + }, + { + "epoch": 2.352336928608115, + "eval_loss": 1.881198525428772, + "eval_runtime": 45.0014, + "eval_samples_per_second": 22.222, + "eval_steps_per_second": 0.933, + "step": 2290 + }, + { + "epoch": 2.3533641499743196, + "grad_norm": 3.064849615097046, + "learning_rate": 1.2257327750285498e-06, + "loss": 1.3404, + "step": 2291 + }, + { + "epoch": 2.354391371340524, + "grad_norm": 3.2958126068115234, + "learning_rate": 1.2238294632660832e-06, + "loss": 1.472, + "step": 2292 + }, + { + "epoch": 2.3554185927067284, + "grad_norm": 3.1702499389648438, + "learning_rate": 1.2219261515036164e-06, + "loss": 1.3842, + "step": 2293 + }, + { + "epoch": 2.3564458140729325, + "grad_norm": 3.022472620010376, + "learning_rate": 1.2200228397411497e-06, + "loss": 1.3946, + "step": 2294 + }, + { + "epoch": 2.357473035439137, + "grad_norm": 3.2721829414367676, + "learning_rate": 1.2181195279786829e-06, + "loss": 1.4444, + "step": 2295 + }, + { + "epoch": 2.3585002568053417, + "grad_norm": 3.1014726161956787, + "learning_rate": 1.2162162162162164e-06, + "loss": 1.478, + "step": 2296 + }, + { + "epoch": 2.359527478171546, + "grad_norm": 3.1011557579040527, + "learning_rate": 1.2143129044537496e-06, + "loss": 1.4067, + "step": 2297 + }, + { + "epoch": 2.3605546995377504, + "grad_norm": 3.09366512298584, + "learning_rate": 1.212409592691283e-06, + "loss": 1.3998, + "step": 2298 + }, + { + "epoch": 2.361581920903955, + "grad_norm": 3.1009769439697266, + "learning_rate": 1.2105062809288163e-06, + "loss": 1.4072, + "step": 2299 + }, + { + "epoch": 2.362609142270159, + "grad_norm": 3.1026828289031982, + "learning_rate": 1.2086029691663495e-06, + "loss": 1.4213, + "step": 2300 + }, + { + "epoch": 2.362609142270159, + "eval_loss": 1.8786081075668335, + "eval_runtime": 44.8368, + "eval_samples_per_second": 22.303, + "eval_steps_per_second": 0.937, + "step": 2300 + }, + { + "epoch": 2.3636363636363638, + "grad_norm": 3.214540481567383, + "learning_rate": 1.2066996574038828e-06, + "loss": 1.4397, + "step": 2301 + }, + { + "epoch": 2.364663585002568, + "grad_norm": 3.0264482498168945, + "learning_rate": 1.2047963456414162e-06, + "loss": 1.3706, + "step": 2302 + }, + { + "epoch": 2.3656908063687725, + "grad_norm": 3.361640453338623, + "learning_rate": 1.2028930338789496e-06, + "loss": 1.4427, + "step": 2303 + }, + { + "epoch": 2.366718027734977, + "grad_norm": 3.1837921142578125, + "learning_rate": 1.2009897221164827e-06, + "loss": 1.4235, + "step": 2304 + }, + { + "epoch": 2.3677452491011812, + "grad_norm": 3.1406962871551514, + "learning_rate": 1.199086410354016e-06, + "loss": 1.3666, + "step": 2305 + }, + { + "epoch": 2.368772470467386, + "grad_norm": 3.189342975616455, + "learning_rate": 1.1971830985915492e-06, + "loss": 1.4383, + "step": 2306 + }, + { + "epoch": 2.36979969183359, + "grad_norm": 3.163757801055908, + "learning_rate": 1.1952797868290828e-06, + "loss": 1.4025, + "step": 2307 + }, + { + "epoch": 2.3708269131997946, + "grad_norm": 3.415684223175049, + "learning_rate": 1.193376475066616e-06, + "loss": 1.4026, + "step": 2308 + }, + { + "epoch": 2.371854134565999, + "grad_norm": 3.28863787651062, + "learning_rate": 1.1914731633041493e-06, + "loss": 1.4249, + "step": 2309 + }, + { + "epoch": 2.3728813559322033, + "grad_norm": 3.2626993656158447, + "learning_rate": 1.1895698515416827e-06, + "loss": 1.4409, + "step": 2310 + }, + { + "epoch": 2.3728813559322033, + "eval_loss": 1.8808053731918335, + "eval_runtime": 41.9147, + "eval_samples_per_second": 23.858, + "eval_steps_per_second": 1.002, + "step": 2310 + }, + { + "epoch": 2.373908577298408, + "grad_norm": 3.252225160598755, + "learning_rate": 1.1876665397792158e-06, + "loss": 1.3958, + "step": 2311 + }, + { + "epoch": 2.3749357986646125, + "grad_norm": 3.0922813415527344, + "learning_rate": 1.1857632280167492e-06, + "loss": 1.3902, + "step": 2312 + }, + { + "epoch": 2.3759630200308166, + "grad_norm": 3.242309808731079, + "learning_rate": 1.1838599162542825e-06, + "loss": 1.4393, + "step": 2313 + }, + { + "epoch": 2.376990241397021, + "grad_norm": 3.0868842601776123, + "learning_rate": 1.181956604491816e-06, + "loss": 1.3587, + "step": 2314 + }, + { + "epoch": 2.3780174627632253, + "grad_norm": 3.1234078407287598, + "learning_rate": 1.180053292729349e-06, + "loss": 1.4387, + "step": 2315 + }, + { + "epoch": 2.37904468412943, + "grad_norm": 3.1602776050567627, + "learning_rate": 1.1781499809668824e-06, + "loss": 1.4151, + "step": 2316 + }, + { + "epoch": 2.380071905495634, + "grad_norm": 3.359309673309326, + "learning_rate": 1.1762466692044158e-06, + "loss": 1.4918, + "step": 2317 + }, + { + "epoch": 2.3810991268618387, + "grad_norm": 3.204411268234253, + "learning_rate": 1.1743433574419491e-06, + "loss": 1.4199, + "step": 2318 + }, + { + "epoch": 2.3821263482280433, + "grad_norm": 3.1535773277282715, + "learning_rate": 1.1724400456794823e-06, + "loss": 1.4494, + "step": 2319 + }, + { + "epoch": 2.3831535695942474, + "grad_norm": 3.124729871749878, + "learning_rate": 1.1705367339170157e-06, + "loss": 1.3915, + "step": 2320 + }, + { + "epoch": 2.3831535695942474, + "eval_loss": 1.8865554332733154, + "eval_runtime": 43.2617, + "eval_samples_per_second": 23.115, + "eval_steps_per_second": 0.971, + "step": 2320 + }, + { + "epoch": 2.384180790960452, + "grad_norm": 3.1918749809265137, + "learning_rate": 1.168633422154549e-06, + "loss": 1.4486, + "step": 2321 + }, + { + "epoch": 2.3852080123266566, + "grad_norm": 3.362424850463867, + "learning_rate": 1.1667301103920824e-06, + "loss": 1.4313, + "step": 2322 + }, + { + "epoch": 2.3862352336928607, + "grad_norm": 3.0750956535339355, + "learning_rate": 1.1648267986296157e-06, + "loss": 1.3796, + "step": 2323 + }, + { + "epoch": 2.3872624550590653, + "grad_norm": 3.081636428833008, + "learning_rate": 1.162923486867149e-06, + "loss": 1.3569, + "step": 2324 + }, + { + "epoch": 2.3882896764252695, + "grad_norm": 3.1504268646240234, + "learning_rate": 1.1610201751046823e-06, + "loss": 1.4058, + "step": 2325 + }, + { + "epoch": 2.389316897791474, + "grad_norm": 3.1414480209350586, + "learning_rate": 1.1591168633422154e-06, + "loss": 1.4215, + "step": 2326 + }, + { + "epoch": 2.3903441191576786, + "grad_norm": 3.262014389038086, + "learning_rate": 1.157213551579749e-06, + "loss": 1.4678, + "step": 2327 + }, + { + "epoch": 2.391371340523883, + "grad_norm": 3.163388729095459, + "learning_rate": 1.1553102398172821e-06, + "loss": 1.4479, + "step": 2328 + }, + { + "epoch": 2.3923985618900874, + "grad_norm": 3.0833804607391357, + "learning_rate": 1.1534069280548155e-06, + "loss": 1.4164, + "step": 2329 + }, + { + "epoch": 2.3934257832562915, + "grad_norm": 3.127962589263916, + "learning_rate": 1.1515036162923487e-06, + "loss": 1.4022, + "step": 2330 + }, + { + "epoch": 2.3934257832562915, + "eval_loss": 1.8800996541976929, + "eval_runtime": 46.0509, + "eval_samples_per_second": 21.715, + "eval_steps_per_second": 0.912, + "step": 2330 + }, + { + "epoch": 2.394453004622496, + "grad_norm": 3.239804983139038, + "learning_rate": 1.149600304529882e-06, + "loss": 1.4315, + "step": 2331 + }, + { + "epoch": 2.3954802259887007, + "grad_norm": 3.0544466972351074, + "learning_rate": 1.1476969927674154e-06, + "loss": 1.3993, + "step": 2332 + }, + { + "epoch": 2.396507447354905, + "grad_norm": 3.2019660472869873, + "learning_rate": 1.1457936810049487e-06, + "loss": 1.4055, + "step": 2333 + }, + { + "epoch": 2.3975346687211094, + "grad_norm": 3.1437528133392334, + "learning_rate": 1.143890369242482e-06, + "loss": 1.4109, + "step": 2334 + }, + { + "epoch": 2.398561890087314, + "grad_norm": 3.274785041809082, + "learning_rate": 1.1419870574800153e-06, + "loss": 1.4603, + "step": 2335 + }, + { + "epoch": 2.399589111453518, + "grad_norm": 3.129502058029175, + "learning_rate": 1.1400837457175486e-06, + "loss": 1.3705, + "step": 2336 + }, + { + "epoch": 2.4006163328197228, + "grad_norm": 3.1008121967315674, + "learning_rate": 1.138180433955082e-06, + "loss": 1.4041, + "step": 2337 + }, + { + "epoch": 2.401643554185927, + "grad_norm": 3.138871431350708, + "learning_rate": 1.1362771221926153e-06, + "loss": 1.375, + "step": 2338 + }, + { + "epoch": 2.4026707755521315, + "grad_norm": 3.1868844032287598, + "learning_rate": 1.1343738104301485e-06, + "loss": 1.4419, + "step": 2339 + }, + { + "epoch": 2.403697996918336, + "grad_norm": 3.2045419216156006, + "learning_rate": 1.1324704986676819e-06, + "loss": 1.4233, + "step": 2340 + }, + { + "epoch": 2.403697996918336, + "eval_loss": 1.8847991228103638, + "eval_runtime": 45.2785, + "eval_samples_per_second": 22.086, + "eval_steps_per_second": 0.928, + "step": 2340 + }, + { + "epoch": 2.4047252182845402, + "grad_norm": 3.190446615219116, + "learning_rate": 1.130567186905215e-06, + "loss": 1.3998, + "step": 2341 + }, + { + "epoch": 2.405752439650745, + "grad_norm": 3.1729485988616943, + "learning_rate": 1.1286638751427486e-06, + "loss": 1.4609, + "step": 2342 + }, + { + "epoch": 2.406779661016949, + "grad_norm": 3.0363855361938477, + "learning_rate": 1.1267605633802817e-06, + "loss": 1.3987, + "step": 2343 + }, + { + "epoch": 2.4078068823831535, + "grad_norm": 3.087101459503174, + "learning_rate": 1.124857251617815e-06, + "loss": 1.4006, + "step": 2344 + }, + { + "epoch": 2.408834103749358, + "grad_norm": 3.1294353008270264, + "learning_rate": 1.1229539398553485e-06, + "loss": 1.4475, + "step": 2345 + }, + { + "epoch": 2.4098613251155623, + "grad_norm": 3.2285165786743164, + "learning_rate": 1.1210506280928816e-06, + "loss": 1.4778, + "step": 2346 + }, + { + "epoch": 2.410888546481767, + "grad_norm": 3.1207237243652344, + "learning_rate": 1.119147316330415e-06, + "loss": 1.4174, + "step": 2347 + }, + { + "epoch": 2.4119157678479715, + "grad_norm": 3.248936653137207, + "learning_rate": 1.1172440045679483e-06, + "loss": 1.4532, + "step": 2348 + }, + { + "epoch": 2.4129429892141756, + "grad_norm": 3.1401236057281494, + "learning_rate": 1.1153406928054817e-06, + "loss": 1.4282, + "step": 2349 + }, + { + "epoch": 2.41397021058038, + "grad_norm": 3.179011821746826, + "learning_rate": 1.1134373810430148e-06, + "loss": 1.4169, + "step": 2350 + }, + { + "epoch": 2.41397021058038, + "eval_loss": 1.8840751647949219, + "eval_runtime": 44.964, + "eval_samples_per_second": 22.24, + "eval_steps_per_second": 0.934, + "step": 2350 + }, + { + "epoch": 2.4149974319465843, + "grad_norm": 3.1402602195739746, + "learning_rate": 1.1115340692805482e-06, + "loss": 1.4533, + "step": 2351 + }, + { + "epoch": 2.416024653312789, + "grad_norm": 3.315294027328491, + "learning_rate": 1.1096307575180816e-06, + "loss": 1.4546, + "step": 2352 + }, + { + "epoch": 2.4170518746789935, + "grad_norm": 3.172640085220337, + "learning_rate": 1.107727445755615e-06, + "loss": 1.4203, + "step": 2353 + }, + { + "epoch": 2.4180790960451977, + "grad_norm": 3.1118452548980713, + "learning_rate": 1.105824133993148e-06, + "loss": 1.3931, + "step": 2354 + }, + { + "epoch": 2.4191063174114023, + "grad_norm": 3.1557371616363525, + "learning_rate": 1.1039208222306814e-06, + "loss": 1.3937, + "step": 2355 + }, + { + "epoch": 2.4201335387776064, + "grad_norm": 3.1625313758850098, + "learning_rate": 1.1020175104682148e-06, + "loss": 1.4184, + "step": 2356 + }, + { + "epoch": 2.421160760143811, + "grad_norm": 3.0680766105651855, + "learning_rate": 1.1001141987057482e-06, + "loss": 1.4219, + "step": 2357 + }, + { + "epoch": 2.4221879815100156, + "grad_norm": 2.9973316192626953, + "learning_rate": 1.0982108869432813e-06, + "loss": 1.4075, + "step": 2358 + }, + { + "epoch": 2.4232152028762197, + "grad_norm": 3.202812433242798, + "learning_rate": 1.0963075751808147e-06, + "loss": 1.4227, + "step": 2359 + }, + { + "epoch": 2.4242424242424243, + "grad_norm": 3.296248197555542, + "learning_rate": 1.094404263418348e-06, + "loss": 1.4486, + "step": 2360 + }, + { + "epoch": 2.4242424242424243, + "eval_loss": 1.8838396072387695, + "eval_runtime": 45.3465, + "eval_samples_per_second": 22.052, + "eval_steps_per_second": 0.926, + "step": 2360 + }, + { + "epoch": 2.425269645608629, + "grad_norm": 3.1101279258728027, + "learning_rate": 1.0925009516558812e-06, + "loss": 1.413, + "step": 2361 + }, + { + "epoch": 2.426296866974833, + "grad_norm": 3.070812463760376, + "learning_rate": 1.0905976398934148e-06, + "loss": 1.4198, + "step": 2362 + }, + { + "epoch": 2.4273240883410376, + "grad_norm": 3.0629656314849854, + "learning_rate": 1.088694328130948e-06, + "loss": 1.4121, + "step": 2363 + }, + { + "epoch": 2.428351309707242, + "grad_norm": 3.228043556213379, + "learning_rate": 1.0867910163684813e-06, + "loss": 1.4292, + "step": 2364 + }, + { + "epoch": 2.4293785310734464, + "grad_norm": 3.068103790283203, + "learning_rate": 1.0848877046060144e-06, + "loss": 1.3747, + "step": 2365 + }, + { + "epoch": 2.4304057524396505, + "grad_norm": 3.105422019958496, + "learning_rate": 1.0829843928435478e-06, + "loss": 1.4128, + "step": 2366 + }, + { + "epoch": 2.431432973805855, + "grad_norm": 3.2109289169311523, + "learning_rate": 1.0810810810810812e-06, + "loss": 1.4051, + "step": 2367 + }, + { + "epoch": 2.4324601951720597, + "grad_norm": 3.1759419441223145, + "learning_rate": 1.0791777693186145e-06, + "loss": 1.4164, + "step": 2368 + }, + { + "epoch": 2.433487416538264, + "grad_norm": 3.2143466472625732, + "learning_rate": 1.0772744575561479e-06, + "loss": 1.4363, + "step": 2369 + }, + { + "epoch": 2.4345146379044684, + "grad_norm": 3.155259132385254, + "learning_rate": 1.075371145793681e-06, + "loss": 1.4089, + "step": 2370 + }, + { + "epoch": 2.4345146379044684, + "eval_loss": 1.8794890642166138, + "eval_runtime": 48.175, + "eval_samples_per_second": 20.758, + "eval_steps_per_second": 0.872, + "step": 2370 + }, + { + "epoch": 2.435541859270673, + "grad_norm": 3.2817187309265137, + "learning_rate": 1.0734678340312144e-06, + "loss": 1.404, + "step": 2371 + }, + { + "epoch": 2.436569080636877, + "grad_norm": 3.2742044925689697, + "learning_rate": 1.0715645222687478e-06, + "loss": 1.4048, + "step": 2372 + }, + { + "epoch": 2.4375963020030817, + "grad_norm": 3.111154556274414, + "learning_rate": 1.0696612105062811e-06, + "loss": 1.3966, + "step": 2373 + }, + { + "epoch": 2.438623523369286, + "grad_norm": 3.083907127380371, + "learning_rate": 1.0677578987438143e-06, + "loss": 1.3931, + "step": 2374 + }, + { + "epoch": 2.4396507447354905, + "grad_norm": 3.164278030395508, + "learning_rate": 1.0658545869813476e-06, + "loss": 1.4373, + "step": 2375 + }, + { + "epoch": 2.440677966101695, + "grad_norm": 3.17753005027771, + "learning_rate": 1.0639512752188808e-06, + "loss": 1.3874, + "step": 2376 + }, + { + "epoch": 2.441705187467899, + "grad_norm": 3.0339651107788086, + "learning_rate": 1.0620479634564144e-06, + "loss": 1.401, + "step": 2377 + }, + { + "epoch": 2.442732408834104, + "grad_norm": 3.1768946647644043, + "learning_rate": 1.0601446516939475e-06, + "loss": 1.4083, + "step": 2378 + }, + { + "epoch": 2.443759630200308, + "grad_norm": 3.0505385398864746, + "learning_rate": 1.0582413399314809e-06, + "loss": 1.3733, + "step": 2379 + }, + { + "epoch": 2.4447868515665125, + "grad_norm": 3.064157485961914, + "learning_rate": 1.0563380281690142e-06, + "loss": 1.3782, + "step": 2380 + }, + { + "epoch": 2.4447868515665125, + "eval_loss": 1.879464864730835, + "eval_runtime": 50.9219, + "eval_samples_per_second": 19.638, + "eval_steps_per_second": 0.825, + "step": 2380 + }, + { + "epoch": 2.445814072932717, + "grad_norm": 3.1522152423858643, + "learning_rate": 1.0544347164065474e-06, + "loss": 1.3926, + "step": 2381 + }, + { + "epoch": 2.4468412942989213, + "grad_norm": 3.073568820953369, + "learning_rate": 1.0525314046440808e-06, + "loss": 1.399, + "step": 2382 + }, + { + "epoch": 2.447868515665126, + "grad_norm": 3.171452522277832, + "learning_rate": 1.0506280928816141e-06, + "loss": 1.4208, + "step": 2383 + }, + { + "epoch": 2.4488957370313305, + "grad_norm": 3.1839122772216797, + "learning_rate": 1.0487247811191475e-06, + "loss": 1.3923, + "step": 2384 + }, + { + "epoch": 2.4499229583975346, + "grad_norm": 3.0783913135528564, + "learning_rate": 1.0468214693566806e-06, + "loss": 1.3984, + "step": 2385 + }, + { + "epoch": 2.450950179763739, + "grad_norm": 3.203411817550659, + "learning_rate": 1.044918157594214e-06, + "loss": 1.4106, + "step": 2386 + }, + { + "epoch": 2.4519774011299433, + "grad_norm": 3.254432201385498, + "learning_rate": 1.0430148458317474e-06, + "loss": 1.4426, + "step": 2387 + }, + { + "epoch": 2.453004622496148, + "grad_norm": 3.2302334308624268, + "learning_rate": 1.0411115340692807e-06, + "loss": 1.3964, + "step": 2388 + }, + { + "epoch": 2.4540318438623525, + "grad_norm": 3.0067214965820312, + "learning_rate": 1.0392082223068139e-06, + "loss": 1.3749, + "step": 2389 + }, + { + "epoch": 2.4550590652285567, + "grad_norm": 3.2783524990081787, + "learning_rate": 1.0373049105443472e-06, + "loss": 1.4603, + "step": 2390 + }, + { + "epoch": 2.4550590652285567, + "eval_loss": 1.8760793209075928, + "eval_runtime": 47.0458, + "eval_samples_per_second": 21.256, + "eval_steps_per_second": 0.893, + "step": 2390 + }, + { + "epoch": 2.4560862865947612, + "grad_norm": 3.2079291343688965, + "learning_rate": 1.0354015987818806e-06, + "loss": 1.4176, + "step": 2391 + }, + { + "epoch": 2.4571135079609654, + "grad_norm": 3.0905849933624268, + "learning_rate": 1.033498287019414e-06, + "loss": 1.428, + "step": 2392 + }, + { + "epoch": 2.45814072932717, + "grad_norm": 3.034698963165283, + "learning_rate": 1.0315949752569471e-06, + "loss": 1.3884, + "step": 2393 + }, + { + "epoch": 2.4591679506933746, + "grad_norm": 3.2759525775909424, + "learning_rate": 1.0296916634944805e-06, + "loss": 1.4455, + "step": 2394 + }, + { + "epoch": 2.4601951720595787, + "grad_norm": 3.2399072647094727, + "learning_rate": 1.0277883517320138e-06, + "loss": 1.3684, + "step": 2395 + }, + { + "epoch": 2.4612223934257833, + "grad_norm": 3.173449993133545, + "learning_rate": 1.025885039969547e-06, + "loss": 1.4227, + "step": 2396 + }, + { + "epoch": 2.462249614791988, + "grad_norm": 3.244391679763794, + "learning_rate": 1.0239817282070806e-06, + "loss": 1.4525, + "step": 2397 + }, + { + "epoch": 2.463276836158192, + "grad_norm": 3.0395283699035645, + "learning_rate": 1.0220784164446137e-06, + "loss": 1.3501, + "step": 2398 + }, + { + "epoch": 2.4643040575243966, + "grad_norm": 3.0880112648010254, + "learning_rate": 1.020175104682147e-06, + "loss": 1.395, + "step": 2399 + }, + { + "epoch": 2.4653312788906008, + "grad_norm": 3.1978697776794434, + "learning_rate": 1.0182717929196802e-06, + "loss": 1.3984, + "step": 2400 + }, + { + "epoch": 2.4653312788906008, + "eval_loss": 1.8770432472229004, + "eval_runtime": 45.2381, + "eval_samples_per_second": 22.105, + "eval_steps_per_second": 0.928, + "step": 2400 + } + ], + "logging_steps": 1, + "max_steps": 2919, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 100, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 7.481542582468608e+17, + "train_batch_size": 3, + "trial_name": null, + "trial_params": null +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/zero_to_fp32.py b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2400/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2600/model.safetensors.index.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2600/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..ed64de846d720b9a7859dc20575fea8e8ca51940 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2600/model.safetensors.index.json @@ -0,0 +1,262 @@ +{ + "metadata": { + "total_size": 7213504512 + }, + "weight_map": { + "lm_head.weight": "model-00002-of-00002.safetensors", + "model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.norm.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2600/zero_to_fp32.py b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2600/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2600/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fb0f9a53a507ffcdb60410deb71d60ef801bf350 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/config.json @@ -0,0 +1,36 @@ +{ + "_name_or_path": "meta-llama/Llama-3.2-3B", + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "eos_token_id": 128001, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 3072, + "initializer_range": 0.02, + "intermediate_size": 8192, + "max_position_embeddings": 131072, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 24, + "num_hidden_layers": 28, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": { + "factor": 32.0, + "high_freq_factor": 4.0, + "low_freq_factor": 1.0, + "original_max_position_embeddings": 8192, + "rope_type": "llama3" + }, + "rope_theta": 500000.0, + "tie_word_embeddings": true, + "torch_dtype": "float16", + "transformers_version": "4.45.1", + "use_cache": true, + "vocab_size": 128256 +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/generation_config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..9c389b841a9a29ddd904e02d1eb0e08dcce82ad9 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/generation_config.json @@ -0,0 +1,9 @@ +{ + "_from_model_config": true, + "bos_token_id": 128000, + "do_sample": true, + "eos_token_id": 128001, + "temperature": 0.6, + "top_p": 0.9, + "transformers_version": "4.45.1" +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/latest b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/latest new file mode 100644 index 0000000000000000000000000000000000000000..ca3fd898b349457b6d1b62649e5bdfc274b799e1 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/latest @@ -0,0 +1 @@ +global_step2700 \ No newline at end of file diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/model.safetensors.index.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..ed64de846d720b9a7859dc20575fea8e8ca51940 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/model.safetensors.index.json @@ -0,0 +1,262 @@ +{ + "metadata": { + "total_size": 7213504512 + }, + "weight_map": { + "lm_head.weight": "model-00002-of-00002.safetensors", + "model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.norm.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/special_tokens_map.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..6949c5975ee0e961ef61cf31010dce04df0a03f8 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/special_tokens_map.json @@ -0,0 +1,23 @@ +{ + "bos_token": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/tokenizer_config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..0b782d4602b361f7577abb52b710a8c0e088588d --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/tokenizer_config.json @@ -0,0 +1,2078 @@ +{ + "added_tokens_decoder": { + "128000": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128001": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128002": { + "content": "<|reserved_special_token_0|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128003": { + "content": "<|reserved_special_token_1|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128004": { + "content": "<|finetune_right_pad_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128005": { + "content": "<|reserved_special_token_2|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128006": { + "content": "<|start_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128007": { + "content": "<|end_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128008": { + "content": "<|eom_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128009": { + "content": "<|eot_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128010": { + "content": "<|python_tag|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128011": { + "content": "<|reserved_special_token_3|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128012": { + "content": "<|reserved_special_token_4|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128013": { + "content": "<|reserved_special_token_5|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128014": { + "content": "<|reserved_special_token_6|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128015": { + "content": "<|reserved_special_token_7|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128016": { + "content": "<|reserved_special_token_8|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128017": { + "content": "<|reserved_special_token_9|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128018": { + "content": "<|reserved_special_token_10|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128019": { + "content": "<|reserved_special_token_11|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128020": { + "content": "<|reserved_special_token_12|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128021": { + "content": "<|reserved_special_token_13|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128022": { + "content": "<|reserved_special_token_14|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128023": { + "content": "<|reserved_special_token_15|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128024": { + "content": "<|reserved_special_token_16|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128025": { + "content": "<|reserved_special_token_17|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128026": { + "content": "<|reserved_special_token_18|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128027": { + "content": "<|reserved_special_token_19|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128028": { + "content": "<|reserved_special_token_20|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128029": { + "content": "<|reserved_special_token_21|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128030": { + "content": "<|reserved_special_token_22|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128031": { + "content": "<|reserved_special_token_23|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128032": { + "content": "<|reserved_special_token_24|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128033": { + "content": "<|reserved_special_token_25|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128034": { + "content": "<|reserved_special_token_26|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128035": { + "content": "<|reserved_special_token_27|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128036": { + "content": "<|reserved_special_token_28|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128037": { + "content": "<|reserved_special_token_29|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128038": { + "content": "<|reserved_special_token_30|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128039": { + "content": "<|reserved_special_token_31|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128040": { + "content": "<|reserved_special_token_32|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128041": { + "content": "<|reserved_special_token_33|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128042": { + "content": "<|reserved_special_token_34|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128043": { + "content": "<|reserved_special_token_35|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128044": { + "content": "<|reserved_special_token_36|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128045": { + "content": "<|reserved_special_token_37|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128046": { + "content": "<|reserved_special_token_38|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128047": { + "content": "<|reserved_special_token_39|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128048": { + "content": "<|reserved_special_token_40|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128049": { + "content": "<|reserved_special_token_41|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128050": { + "content": "<|reserved_special_token_42|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128051": { + "content": "<|reserved_special_token_43|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128052": { + "content": "<|reserved_special_token_44|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128053": { + "content": "<|reserved_special_token_45|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128054": { + "content": "<|reserved_special_token_46|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128055": { + "content": "<|reserved_special_token_47|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128056": { + "content": "<|reserved_special_token_48|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128057": { + "content": "<|reserved_special_token_49|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128058": { + "content": "<|reserved_special_token_50|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128059": { + "content": "<|reserved_special_token_51|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128060": { + "content": "<|reserved_special_token_52|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128061": { + "content": "<|reserved_special_token_53|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128062": { + "content": "<|reserved_special_token_54|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128063": { + "content": "<|reserved_special_token_55|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128064": { + "content": "<|reserved_special_token_56|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128065": { + "content": "<|reserved_special_token_57|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128066": { + "content": "<|reserved_special_token_58|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128067": { + "content": "<|reserved_special_token_59|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128068": { + "content": "<|reserved_special_token_60|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128069": { + "content": "<|reserved_special_token_61|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128070": { + "content": "<|reserved_special_token_62|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128071": { + "content": "<|reserved_special_token_63|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128072": { + "content": "<|reserved_special_token_64|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128073": { + "content": "<|reserved_special_token_65|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128074": { + "content": "<|reserved_special_token_66|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128075": { + "content": "<|reserved_special_token_67|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128076": { + "content": "<|reserved_special_token_68|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128077": { + "content": "<|reserved_special_token_69|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128078": { + "content": "<|reserved_special_token_70|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128079": { + "content": "<|reserved_special_token_71|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128080": { + "content": "<|reserved_special_token_72|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128081": { + "content": "<|reserved_special_token_73|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128082": { + "content": "<|reserved_special_token_74|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128083": { + "content": "<|reserved_special_token_75|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128084": { + "content": "<|reserved_special_token_76|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128085": { + "content": "<|reserved_special_token_77|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128086": { + "content": "<|reserved_special_token_78|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128087": { + "content": "<|reserved_special_token_79|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128088": { + "content": "<|reserved_special_token_80|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128089": { + "content": "<|reserved_special_token_81|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128090": { + "content": "<|reserved_special_token_82|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128091": { + "content": "<|reserved_special_token_83|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128092": { + "content": "<|reserved_special_token_84|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128093": { + "content": "<|reserved_special_token_85|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128094": { + "content": "<|reserved_special_token_86|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128095": { + "content": "<|reserved_special_token_87|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128096": { + "content": "<|reserved_special_token_88|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128097": { + "content": "<|reserved_special_token_89|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128098": { + "content": "<|reserved_special_token_90|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128099": { + "content": "<|reserved_special_token_91|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128100": { + "content": "<|reserved_special_token_92|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128101": { + "content": "<|reserved_special_token_93|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128102": { + "content": "<|reserved_special_token_94|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128103": { + "content": "<|reserved_special_token_95|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128104": { + "content": "<|reserved_special_token_96|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128105": { + "content": "<|reserved_special_token_97|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128106": { + "content": "<|reserved_special_token_98|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128107": { + "content": "<|reserved_special_token_99|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128108": { + "content": "<|reserved_special_token_100|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128109": { + "content": "<|reserved_special_token_101|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128110": { + "content": "<|reserved_special_token_102|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128111": { + "content": "<|reserved_special_token_103|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128112": { + "content": "<|reserved_special_token_104|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128113": { + "content": "<|reserved_special_token_105|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128114": { + "content": "<|reserved_special_token_106|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128115": { + "content": "<|reserved_special_token_107|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128116": { + "content": "<|reserved_special_token_108|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128117": { + "content": "<|reserved_special_token_109|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128118": { + "content": "<|reserved_special_token_110|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128119": { + "content": "<|reserved_special_token_111|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128120": { + "content": "<|reserved_special_token_112|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128121": { + "content": "<|reserved_special_token_113|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128122": { + "content": "<|reserved_special_token_114|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128123": { + "content": "<|reserved_special_token_115|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128124": { + "content": "<|reserved_special_token_116|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128125": { + "content": "<|reserved_special_token_117|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128126": { + "content": "<|reserved_special_token_118|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128127": { + "content": "<|reserved_special_token_119|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128128": { + "content": "<|reserved_special_token_120|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128129": { + "content": "<|reserved_special_token_121|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128130": { + "content": "<|reserved_special_token_122|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128131": { + "content": "<|reserved_special_token_123|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128132": { + "content": "<|reserved_special_token_124|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128133": { + "content": "<|reserved_special_token_125|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128134": { + "content": "<|reserved_special_token_126|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128135": { + "content": "<|reserved_special_token_127|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128136": { + "content": "<|reserved_special_token_128|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128137": { + "content": "<|reserved_special_token_129|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128138": { + "content": "<|reserved_special_token_130|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128139": { + "content": "<|reserved_special_token_131|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128140": { + "content": "<|reserved_special_token_132|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128141": { + "content": "<|reserved_special_token_133|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128142": { + "content": "<|reserved_special_token_134|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128143": { + "content": "<|reserved_special_token_135|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128144": { + "content": "<|reserved_special_token_136|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128145": { + "content": "<|reserved_special_token_137|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128146": { + "content": "<|reserved_special_token_138|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128147": { + "content": "<|reserved_special_token_139|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128148": { + "content": "<|reserved_special_token_140|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128149": { + "content": "<|reserved_special_token_141|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128150": { + "content": "<|reserved_special_token_142|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128151": { + "content": "<|reserved_special_token_143|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128152": { + "content": "<|reserved_special_token_144|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128153": { + "content": "<|reserved_special_token_145|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128154": { + "content": "<|reserved_special_token_146|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128155": { + "content": "<|reserved_special_token_147|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128156": { + "content": "<|reserved_special_token_148|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128157": { + "content": "<|reserved_special_token_149|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128158": { + "content": "<|reserved_special_token_150|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128159": { + "content": "<|reserved_special_token_151|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128160": { + "content": "<|reserved_special_token_152|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128161": { + "content": "<|reserved_special_token_153|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128162": { + "content": "<|reserved_special_token_154|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128163": { + "content": "<|reserved_special_token_155|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128164": { + "content": "<|reserved_special_token_156|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128165": { + "content": "<|reserved_special_token_157|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128166": { + "content": "<|reserved_special_token_158|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128167": { + "content": "<|reserved_special_token_159|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128168": { + "content": "<|reserved_special_token_160|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128169": { + "content": "<|reserved_special_token_161|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128170": { + "content": "<|reserved_special_token_162|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128171": { + "content": "<|reserved_special_token_163|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128172": { + "content": "<|reserved_special_token_164|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128173": { + "content": "<|reserved_special_token_165|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128174": { + "content": "<|reserved_special_token_166|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128175": { + "content": "<|reserved_special_token_167|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128176": { + "content": "<|reserved_special_token_168|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128177": { + "content": "<|reserved_special_token_169|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128178": { + "content": "<|reserved_special_token_170|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128179": { + "content": "<|reserved_special_token_171|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128180": { + "content": "<|reserved_special_token_172|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128181": { + "content": "<|reserved_special_token_173|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128182": { + "content": "<|reserved_special_token_174|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128183": { + "content": "<|reserved_special_token_175|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128184": { + "content": "<|reserved_special_token_176|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128185": { + "content": "<|reserved_special_token_177|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128186": { + "content": "<|reserved_special_token_178|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128187": { + "content": "<|reserved_special_token_179|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128188": { + "content": "<|reserved_special_token_180|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128189": { + "content": "<|reserved_special_token_181|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128190": { + "content": "<|reserved_special_token_182|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128191": { + "content": "<|reserved_special_token_183|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128192": { + "content": "<|reserved_special_token_184|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128193": { + "content": "<|reserved_special_token_185|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128194": { + "content": "<|reserved_special_token_186|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128195": { + "content": "<|reserved_special_token_187|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128196": { + "content": "<|reserved_special_token_188|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128197": { + "content": "<|reserved_special_token_189|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128198": { + "content": "<|reserved_special_token_190|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128199": { + "content": "<|reserved_special_token_191|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128200": { + "content": "<|reserved_special_token_192|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128201": { + "content": "<|reserved_special_token_193|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128202": { + "content": "<|reserved_special_token_194|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128203": { + "content": "<|reserved_special_token_195|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128204": { + "content": "<|reserved_special_token_196|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128205": { + "content": "<|reserved_special_token_197|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128206": { + "content": "<|reserved_special_token_198|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128207": { + "content": "<|reserved_special_token_199|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128208": { + "content": "<|reserved_special_token_200|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128209": { + "content": "<|reserved_special_token_201|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128210": { + "content": "<|reserved_special_token_202|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128211": { + "content": "<|reserved_special_token_203|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128212": { + "content": "<|reserved_special_token_204|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128213": { + "content": "<|reserved_special_token_205|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128214": { + "content": "<|reserved_special_token_206|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128215": { + "content": "<|reserved_special_token_207|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128216": { + "content": "<|reserved_special_token_208|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128217": { + "content": "<|reserved_special_token_209|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128218": { + "content": "<|reserved_special_token_210|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128219": { + "content": "<|reserved_special_token_211|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128220": { + "content": "<|reserved_special_token_212|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128221": { + "content": "<|reserved_special_token_213|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128222": { + "content": "<|reserved_special_token_214|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128223": { + "content": "<|reserved_special_token_215|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128224": { + "content": "<|reserved_special_token_216|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128225": { + "content": "<|reserved_special_token_217|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128226": { + "content": "<|reserved_special_token_218|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128227": { + "content": "<|reserved_special_token_219|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128228": { + "content": "<|reserved_special_token_220|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128229": { + "content": "<|reserved_special_token_221|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128230": { + "content": "<|reserved_special_token_222|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128231": { + "content": "<|reserved_special_token_223|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128232": { + "content": "<|reserved_special_token_224|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128233": { + "content": "<|reserved_special_token_225|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128234": { + "content": "<|reserved_special_token_226|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128235": { + "content": "<|reserved_special_token_227|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128236": { + "content": "<|reserved_special_token_228|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128237": { + "content": "<|reserved_special_token_229|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128238": { + "content": "<|reserved_special_token_230|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128239": { + "content": "<|reserved_special_token_231|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128240": { + "content": "<|reserved_special_token_232|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128241": { + "content": "<|reserved_special_token_233|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128242": { + "content": "<|reserved_special_token_234|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128243": { + "content": "<|reserved_special_token_235|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128244": { + "content": "<|reserved_special_token_236|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128245": { + "content": "<|reserved_special_token_237|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128246": { + "content": "<|reserved_special_token_238|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128247": { + "content": "<|reserved_special_token_239|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128248": { + "content": "<|reserved_special_token_240|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128249": { + "content": "<|reserved_special_token_241|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128250": { + "content": "<|reserved_special_token_242|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128251": { + "content": "<|reserved_special_token_243|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128252": { + "content": "<|reserved_special_token_244|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128253": { + "content": "<|reserved_special_token_245|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128254": { + "content": "<|reserved_special_token_246|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128255": { + "content": "<|reserved_special_token_247|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128256": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128257": { + "content": "🅁", + "lstrip": true, + "normalized": true, + "rstrip": false, + "single_word": false, + "special": false + } + }, + "bos_token": "<|begin_of_text|>", + "clean_up_tokenization_spaces": true, + "eos_token": "<|end_of_text|>", + "model_input_names": [ + "input_ids", + "attention_mask" + ], + "model_max_length": 131072, + "pad_token": "[PAD]", + "tokenizer_class": "PreTrainedTokenizerFast" +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/trainer_state.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..08a3d55baf94fa78327355a53721aa23cd070c65 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/trainer_state.json @@ -0,0 +1,21093 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 2.773497688751926, + "eval_steps": 10, + "global_step": 2700, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.001027221366204417, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7873, + "step": 1 + }, + { + "epoch": 0.002054442732408834, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7904, + "step": 2 + }, + { + "epoch": 0.0030816640986132513, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8348, + "step": 3 + }, + { + "epoch": 0.004108885464817668, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8357, + "step": 4 + }, + { + "epoch": 0.005136106831022085, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8153, + "step": 5 + }, + { + "epoch": 0.0061633281972265025, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.825, + "step": 6 + }, + { + "epoch": 0.007190549563430919, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8002, + "step": 7 + }, + { + "epoch": 0.008217770929635337, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8413, + "step": 8 + }, + { + "epoch": 0.009244992295839754, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8157, + "step": 9 + }, + { + "epoch": 0.01027221366204417, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8249, + "step": 10 + }, + { + "epoch": 0.01027221366204417, + "eval_loss": 2.8518412113189697, + "eval_runtime": 43.212, + "eval_samples_per_second": 23.142, + "eval_steps_per_second": 0.972, + "step": 10 + }, + { + "epoch": 0.011299435028248588, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7829, + "step": 11 + }, + { + "epoch": 0.012326656394453005, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7819, + "step": 12 + }, + { + "epoch": 0.01335387776065742, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8355, + "step": 13 + }, + { + "epoch": 0.014381099126861838, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8138, + "step": 14 + }, + { + "epoch": 0.015408320493066256, + "grad_norm": 5.162191390991211, + "learning_rate": 1.7123287671232876e-08, + "loss": 2.8088, + "step": 15 + }, + { + "epoch": 0.016435541859270673, + "grad_norm": 5.5954084396362305, + "learning_rate": 3.424657534246575e-08, + "loss": 2.7883, + "step": 16 + }, + { + "epoch": 0.01746276322547509, + "grad_norm": 5.542337417602539, + "learning_rate": 5.136986301369863e-08, + "loss": 2.7625, + "step": 17 + }, + { + "epoch": 0.01848998459167951, + "grad_norm": 5.300518989562988, + "learning_rate": 6.84931506849315e-08, + "loss": 2.84, + "step": 18 + }, + { + "epoch": 0.019517205957883924, + "grad_norm": 5.6646881103515625, + "learning_rate": 8.561643835616439e-08, + "loss": 2.761, + "step": 19 + }, + { + "epoch": 0.02054442732408834, + "grad_norm": 5.375557899475098, + "learning_rate": 1.0273972602739726e-07, + "loss": 2.8215, + "step": 20 + }, + { + "epoch": 0.02054442732408834, + "eval_loss": 2.8507144451141357, + "eval_runtime": 45.1133, + "eval_samples_per_second": 22.166, + "eval_steps_per_second": 0.931, + "step": 20 + }, + { + "epoch": 0.02157164869029276, + "grad_norm": 5.1731977462768555, + "learning_rate": 1.1986301369863014e-07, + "loss": 2.8348, + "step": 21 + }, + { + "epoch": 0.022598870056497175, + "grad_norm": 5.370945930480957, + "learning_rate": 1.36986301369863e-07, + "loss": 2.822, + "step": 22 + }, + { + "epoch": 0.02362609142270159, + "grad_norm": 5.550389289855957, + "learning_rate": 1.541095890410959e-07, + "loss": 2.7904, + "step": 23 + }, + { + "epoch": 0.02465331278890601, + "grad_norm": 5.339347839355469, + "learning_rate": 1.7123287671232878e-07, + "loss": 2.8066, + "step": 24 + }, + { + "epoch": 0.025680534155110426, + "grad_norm": 5.2319722175598145, + "learning_rate": 1.8835616438356165e-07, + "loss": 2.7976, + "step": 25 + }, + { + "epoch": 0.02670775552131484, + "grad_norm": 5.207149982452393, + "learning_rate": 2.0547945205479452e-07, + "loss": 2.8162, + "step": 26 + }, + { + "epoch": 0.02773497688751926, + "grad_norm": 4.400204181671143, + "learning_rate": 2.226027397260274e-07, + "loss": 2.7808, + "step": 27 + }, + { + "epoch": 0.028762198253723677, + "grad_norm": 4.21675968170166, + "learning_rate": 2.397260273972603e-07, + "loss": 2.8044, + "step": 28 + }, + { + "epoch": 0.029789419619928096, + "grad_norm": 4.406878471374512, + "learning_rate": 2.568493150684932e-07, + "loss": 2.8162, + "step": 29 + }, + { + "epoch": 0.030816640986132512, + "grad_norm": 4.024835586547852, + "learning_rate": 2.73972602739726e-07, + "loss": 2.7994, + "step": 30 + }, + { + "epoch": 0.030816640986132512, + "eval_loss": 2.8124029636383057, + "eval_runtime": 43.1772, + "eval_samples_per_second": 23.16, + "eval_steps_per_second": 0.973, + "step": 30 + }, + { + "epoch": 0.03184386235233693, + "grad_norm": 3.5378901958465576, + "learning_rate": 2.910958904109589e-07, + "loss": 2.7637, + "step": 31 + }, + { + "epoch": 0.03287108371854135, + "grad_norm": 3.6289916038513184, + "learning_rate": 3.082191780821918e-07, + "loss": 2.8017, + "step": 32 + }, + { + "epoch": 0.03389830508474576, + "grad_norm": 3.6789698600769043, + "learning_rate": 3.2534246575342466e-07, + "loss": 2.7723, + "step": 33 + }, + { + "epoch": 0.03492552645095018, + "grad_norm": 3.491389751434326, + "learning_rate": 3.4246575342465755e-07, + "loss": 2.7528, + "step": 34 + }, + { + "epoch": 0.035952747817154594, + "grad_norm": 3.383080005645752, + "learning_rate": 3.595890410958904e-07, + "loss": 2.7418, + "step": 35 + }, + { + "epoch": 0.03697996918335902, + "grad_norm": 3.4823594093322754, + "learning_rate": 3.767123287671233e-07, + "loss": 2.6662, + "step": 36 + }, + { + "epoch": 0.03800719054956343, + "grad_norm": 3.2146108150482178, + "learning_rate": 3.938356164383562e-07, + "loss": 2.7659, + "step": 37 + }, + { + "epoch": 0.03903441191576785, + "grad_norm": 2.9549412727355957, + "learning_rate": 4.1095890410958903e-07, + "loss": 2.6838, + "step": 38 + }, + { + "epoch": 0.040061633281972264, + "grad_norm": 3.0744080543518066, + "learning_rate": 4.2808219178082193e-07, + "loss": 2.6879, + "step": 39 + }, + { + "epoch": 0.04108885464817668, + "grad_norm": 3.19840669631958, + "learning_rate": 4.452054794520548e-07, + "loss": 2.7443, + "step": 40 + }, + { + "epoch": 0.04108885464817668, + "eval_loss": 2.7469444274902344, + "eval_runtime": 46.4676, + "eval_samples_per_second": 21.52, + "eval_steps_per_second": 0.904, + "step": 40 + }, + { + "epoch": 0.042116076014381096, + "grad_norm": 3.178406000137329, + "learning_rate": 4.6232876712328767e-07, + "loss": 2.7036, + "step": 41 + }, + { + "epoch": 0.04314329738058552, + "grad_norm": 3.2032673358917236, + "learning_rate": 4.794520547945206e-07, + "loss": 2.6465, + "step": 42 + }, + { + "epoch": 0.044170518746789934, + "grad_norm": 3.1525962352752686, + "learning_rate": 4.965753424657534e-07, + "loss": 2.6996, + "step": 43 + }, + { + "epoch": 0.04519774011299435, + "grad_norm": 2.907524585723877, + "learning_rate": 5.136986301369864e-07, + "loss": 2.6581, + "step": 44 + }, + { + "epoch": 0.046224961479198766, + "grad_norm": 2.5885250568389893, + "learning_rate": 5.308219178082192e-07, + "loss": 2.6773, + "step": 45 + }, + { + "epoch": 0.04725218284540318, + "grad_norm": 2.569068431854248, + "learning_rate": 5.47945205479452e-07, + "loss": 2.6635, + "step": 46 + }, + { + "epoch": 0.048279404211607604, + "grad_norm": 2.5313518047332764, + "learning_rate": 5.65068493150685e-07, + "loss": 2.706, + "step": 47 + }, + { + "epoch": 0.04930662557781202, + "grad_norm": 2.615442991256714, + "learning_rate": 5.821917808219178e-07, + "loss": 2.638, + "step": 48 + }, + { + "epoch": 0.050333846944016436, + "grad_norm": 2.7420806884765625, + "learning_rate": 5.993150684931507e-07, + "loss": 2.6398, + "step": 49 + }, + { + "epoch": 0.05136106831022085, + "grad_norm": 2.5246975421905518, + "learning_rate": 6.164383561643836e-07, + "loss": 2.5916, + "step": 50 + }, + { + "epoch": 0.05136106831022085, + "eval_loss": 2.6723692417144775, + "eval_runtime": 46.7206, + "eval_samples_per_second": 21.404, + "eval_steps_per_second": 0.899, + "step": 50 + }, + { + "epoch": 0.05238828967642527, + "grad_norm": 2.390143632888794, + "learning_rate": 6.335616438356165e-07, + "loss": 2.5773, + "step": 51 + }, + { + "epoch": 0.05341551104262968, + "grad_norm": 2.0792484283447266, + "learning_rate": 6.506849315068493e-07, + "loss": 2.6249, + "step": 52 + }, + { + "epoch": 0.054442732408834106, + "grad_norm": 1.9995019435882568, + "learning_rate": 6.678082191780823e-07, + "loss": 2.5425, + "step": 53 + }, + { + "epoch": 0.05546995377503852, + "grad_norm": 1.985452651977539, + "learning_rate": 6.849315068493151e-07, + "loss": 2.5986, + "step": 54 + }, + { + "epoch": 0.05649717514124294, + "grad_norm": 2.10184907913208, + "learning_rate": 7.020547945205481e-07, + "loss": 2.5713, + "step": 55 + }, + { + "epoch": 0.05752439650744735, + "grad_norm": 2.0898985862731934, + "learning_rate": 7.191780821917808e-07, + "loss": 2.535, + "step": 56 + }, + { + "epoch": 0.05855161787365177, + "grad_norm": 2.0239579677581787, + "learning_rate": 7.363013698630137e-07, + "loss": 2.5667, + "step": 57 + }, + { + "epoch": 0.05957883923985619, + "grad_norm": 2.207080125808716, + "learning_rate": 7.534246575342466e-07, + "loss": 2.4999, + "step": 58 + }, + { + "epoch": 0.06060606060606061, + "grad_norm": 1.8541918992996216, + "learning_rate": 7.705479452054795e-07, + "loss": 2.5582, + "step": 59 + }, + { + "epoch": 0.061633281972265024, + "grad_norm": 1.8530476093292236, + "learning_rate": 7.876712328767124e-07, + "loss": 2.5535, + "step": 60 + }, + { + "epoch": 0.061633281972265024, + "eval_loss": 2.607963800430298, + "eval_runtime": 45.9754, + "eval_samples_per_second": 21.751, + "eval_steps_per_second": 0.914, + "step": 60 + }, + { + "epoch": 0.06266050333846944, + "grad_norm": 1.7978090047836304, + "learning_rate": 8.047945205479453e-07, + "loss": 2.53, + "step": 61 + }, + { + "epoch": 0.06368772470467386, + "grad_norm": 1.8173762559890747, + "learning_rate": 8.219178082191781e-07, + "loss": 2.5144, + "step": 62 + }, + { + "epoch": 0.06471494607087827, + "grad_norm": 2.0847485065460205, + "learning_rate": 8.39041095890411e-07, + "loss": 2.5525, + "step": 63 + }, + { + "epoch": 0.0657421674370827, + "grad_norm": 1.8691741228103638, + "learning_rate": 8.561643835616439e-07, + "loss": 2.5451, + "step": 64 + }, + { + "epoch": 0.0667693888032871, + "grad_norm": 2.0242726802825928, + "learning_rate": 8.732876712328768e-07, + "loss": 2.4145, + "step": 65 + }, + { + "epoch": 0.06779661016949153, + "grad_norm": 1.8268908262252808, + "learning_rate": 8.904109589041097e-07, + "loss": 2.5024, + "step": 66 + }, + { + "epoch": 0.06882383153569595, + "grad_norm": 1.8703057765960693, + "learning_rate": 9.075342465753426e-07, + "loss": 2.461, + "step": 67 + }, + { + "epoch": 0.06985105290190036, + "grad_norm": 1.8747144937515259, + "learning_rate": 9.246575342465753e-07, + "loss": 2.5122, + "step": 68 + }, + { + "epoch": 0.07087827426810478, + "grad_norm": 1.744924783706665, + "learning_rate": 9.417808219178083e-07, + "loss": 2.4673, + "step": 69 + }, + { + "epoch": 0.07190549563430919, + "grad_norm": 1.7492460012435913, + "learning_rate": 9.589041095890411e-07, + "loss": 2.5598, + "step": 70 + }, + { + "epoch": 0.07190549563430919, + "eval_loss": 2.547417640686035, + "eval_runtime": 49.7488, + "eval_samples_per_second": 20.101, + "eval_steps_per_second": 0.844, + "step": 70 + }, + { + "epoch": 0.07293271700051361, + "grad_norm": 1.9218230247497559, + "learning_rate": 9.76027397260274e-07, + "loss": 2.4562, + "step": 71 + }, + { + "epoch": 0.07395993836671803, + "grad_norm": 2.0127363204956055, + "learning_rate": 9.931506849315068e-07, + "loss": 2.4141, + "step": 72 + }, + { + "epoch": 0.07498715973292244, + "grad_norm": 1.8700369596481323, + "learning_rate": 1.0102739726027399e-06, + "loss": 2.4696, + "step": 73 + }, + { + "epoch": 0.07601438109912687, + "grad_norm": 2.1706016063690186, + "learning_rate": 1.0273972602739727e-06, + "loss": 2.4404, + "step": 74 + }, + { + "epoch": 0.07704160246533127, + "grad_norm": 2.470412254333496, + "learning_rate": 1.0445205479452056e-06, + "loss": 2.4409, + "step": 75 + }, + { + "epoch": 0.0780688238315357, + "grad_norm": 1.906569242477417, + "learning_rate": 1.0616438356164384e-06, + "loss": 2.4313, + "step": 76 + }, + { + "epoch": 0.07909604519774012, + "grad_norm": 2.122936487197876, + "learning_rate": 1.0787671232876712e-06, + "loss": 2.4369, + "step": 77 + }, + { + "epoch": 0.08012326656394453, + "grad_norm": 2.079127311706543, + "learning_rate": 1.095890410958904e-06, + "loss": 2.449, + "step": 78 + }, + { + "epoch": 0.08115048793014895, + "grad_norm": 1.9173146486282349, + "learning_rate": 1.1130136986301371e-06, + "loss": 2.3782, + "step": 79 + }, + { + "epoch": 0.08217770929635336, + "grad_norm": 1.9856903553009033, + "learning_rate": 1.13013698630137e-06, + "loss": 2.4215, + "step": 80 + }, + { + "epoch": 0.08217770929635336, + "eval_loss": 2.479673385620117, + "eval_runtime": 46.2211, + "eval_samples_per_second": 21.635, + "eval_steps_per_second": 0.909, + "step": 80 + }, + { + "epoch": 0.08320493066255778, + "grad_norm": 2.3764865398406982, + "learning_rate": 1.1472602739726028e-06, + "loss": 2.3744, + "step": 81 + }, + { + "epoch": 0.08423215202876219, + "grad_norm": 1.980668544769287, + "learning_rate": 1.1643835616438357e-06, + "loss": 2.3856, + "step": 82 + }, + { + "epoch": 0.08525937339496661, + "grad_norm": 2.2589704990386963, + "learning_rate": 1.1815068493150685e-06, + "loss": 2.3853, + "step": 83 + }, + { + "epoch": 0.08628659476117104, + "grad_norm": 2.171780824661255, + "learning_rate": 1.1986301369863014e-06, + "loss": 2.3603, + "step": 84 + }, + { + "epoch": 0.08731381612737545, + "grad_norm": 2.25736403465271, + "learning_rate": 1.2157534246575344e-06, + "loss": 2.3189, + "step": 85 + }, + { + "epoch": 0.08834103749357987, + "grad_norm": 2.3643763065338135, + "learning_rate": 1.2328767123287673e-06, + "loss": 2.3795, + "step": 86 + }, + { + "epoch": 0.08936825885978428, + "grad_norm": 2.384516954421997, + "learning_rate": 1.25e-06, + "loss": 2.3257, + "step": 87 + }, + { + "epoch": 0.0903954802259887, + "grad_norm": 2.5072567462921143, + "learning_rate": 1.267123287671233e-06, + "loss": 2.3125, + "step": 88 + }, + { + "epoch": 0.09142270159219312, + "grad_norm": 2.2726807594299316, + "learning_rate": 1.284246575342466e-06, + "loss": 2.3297, + "step": 89 + }, + { + "epoch": 0.09244992295839753, + "grad_norm": 2.4817421436309814, + "learning_rate": 1.3013698630136986e-06, + "loss": 2.2878, + "step": 90 + }, + { + "epoch": 0.09244992295839753, + "eval_loss": 2.4065146446228027, + "eval_runtime": 47.9372, + "eval_samples_per_second": 20.861, + "eval_steps_per_second": 0.876, + "step": 90 + }, + { + "epoch": 0.09347714432460195, + "grad_norm": 2.2219603061676025, + "learning_rate": 1.3184931506849317e-06, + "loss": 2.2978, + "step": 91 + }, + { + "epoch": 0.09450436569080636, + "grad_norm": 2.8790745735168457, + "learning_rate": 1.3356164383561645e-06, + "loss": 2.313, + "step": 92 + }, + { + "epoch": 0.09553158705701079, + "grad_norm": 2.589200496673584, + "learning_rate": 1.3527397260273976e-06, + "loss": 2.2937, + "step": 93 + }, + { + "epoch": 0.09655880842321521, + "grad_norm": 2.760016679763794, + "learning_rate": 1.3698630136986302e-06, + "loss": 2.277, + "step": 94 + }, + { + "epoch": 0.09758602978941962, + "grad_norm": 2.2641494274139404, + "learning_rate": 1.386986301369863e-06, + "loss": 2.3336, + "step": 95 + }, + { + "epoch": 0.09861325115562404, + "grad_norm": 2.2786386013031006, + "learning_rate": 1.4041095890410961e-06, + "loss": 2.233, + "step": 96 + }, + { + "epoch": 0.09964047252182845, + "grad_norm": 2.5590052604675293, + "learning_rate": 1.421232876712329e-06, + "loss": 2.2851, + "step": 97 + }, + { + "epoch": 0.10066769388803287, + "grad_norm": 2.5609471797943115, + "learning_rate": 1.4383561643835616e-06, + "loss": 2.1848, + "step": 98 + }, + { + "epoch": 0.1016949152542373, + "grad_norm": 2.442521810531616, + "learning_rate": 1.4554794520547946e-06, + "loss": 2.2073, + "step": 99 + }, + { + "epoch": 0.1027221366204417, + "grad_norm": 2.3735859394073486, + "learning_rate": 1.4726027397260275e-06, + "loss": 2.2353, + "step": 100 + }, + { + "epoch": 0.1027221366204417, + "eval_loss": 2.3457911014556885, + "eval_runtime": 51.6469, + "eval_samples_per_second": 19.362, + "eval_steps_per_second": 0.813, + "step": 100 + }, + { + "epoch": 0.10374935798664613, + "grad_norm": 2.175354242324829, + "learning_rate": 1.4897260273972605e-06, + "loss": 2.2259, + "step": 101 + }, + { + "epoch": 0.10477657935285054, + "grad_norm": 2.217677116394043, + "learning_rate": 1.5068493150684932e-06, + "loss": 2.2082, + "step": 102 + }, + { + "epoch": 0.10580380071905496, + "grad_norm": 2.4618067741394043, + "learning_rate": 1.5239726027397262e-06, + "loss": 2.1649, + "step": 103 + }, + { + "epoch": 0.10683102208525937, + "grad_norm": 2.232060432434082, + "learning_rate": 1.541095890410959e-06, + "loss": 2.1962, + "step": 104 + }, + { + "epoch": 0.10785824345146379, + "grad_norm": 2.9140872955322266, + "learning_rate": 1.5582191780821921e-06, + "loss": 2.2446, + "step": 105 + }, + { + "epoch": 0.10888546481766821, + "grad_norm": 2.232158660888672, + "learning_rate": 1.5753424657534248e-06, + "loss": 2.2179, + "step": 106 + }, + { + "epoch": 0.10991268618387262, + "grad_norm": 2.804938793182373, + "learning_rate": 1.5924657534246576e-06, + "loss": 2.2449, + "step": 107 + }, + { + "epoch": 0.11093990755007704, + "grad_norm": 2.3690710067749023, + "learning_rate": 1.6095890410958907e-06, + "loss": 2.225, + "step": 108 + }, + { + "epoch": 0.11196712891628145, + "grad_norm": 2.7717416286468506, + "learning_rate": 1.6267123287671235e-06, + "loss": 2.0922, + "step": 109 + }, + { + "epoch": 0.11299435028248588, + "grad_norm": 3.053602933883667, + "learning_rate": 1.6438356164383561e-06, + "loss": 2.1702, + "step": 110 + }, + { + "epoch": 0.11299435028248588, + "eval_loss": 2.298161029815674, + "eval_runtime": 44.0776, + "eval_samples_per_second": 22.687, + "eval_steps_per_second": 0.953, + "step": 110 + }, + { + "epoch": 0.1140215716486903, + "grad_norm": 2.8673532009124756, + "learning_rate": 1.6609589041095892e-06, + "loss": 2.0938, + "step": 111 + }, + { + "epoch": 0.1150487930148947, + "grad_norm": 2.57185959815979, + "learning_rate": 1.678082191780822e-06, + "loss": 2.1892, + "step": 112 + }, + { + "epoch": 0.11607601438109913, + "grad_norm": 2.5278828144073486, + "learning_rate": 1.695205479452055e-06, + "loss": 2.1251, + "step": 113 + }, + { + "epoch": 0.11710323574730354, + "grad_norm": 2.8581748008728027, + "learning_rate": 1.7123287671232877e-06, + "loss": 2.1699, + "step": 114 + }, + { + "epoch": 0.11813045711350796, + "grad_norm": 2.924959897994995, + "learning_rate": 1.7294520547945206e-06, + "loss": 2.1005, + "step": 115 + }, + { + "epoch": 0.11915767847971238, + "grad_norm": 2.477034091949463, + "learning_rate": 1.7465753424657536e-06, + "loss": 2.2016, + "step": 116 + }, + { + "epoch": 0.12018489984591679, + "grad_norm": 2.8719305992126465, + "learning_rate": 1.7636986301369865e-06, + "loss": 2.1588, + "step": 117 + }, + { + "epoch": 0.12121212121212122, + "grad_norm": 2.4403953552246094, + "learning_rate": 1.7808219178082193e-06, + "loss": 2.1445, + "step": 118 + }, + { + "epoch": 0.12223934257832562, + "grad_norm": 2.7994632720947266, + "learning_rate": 1.7979452054794521e-06, + "loss": 2.1369, + "step": 119 + }, + { + "epoch": 0.12326656394453005, + "grad_norm": 2.163196563720703, + "learning_rate": 1.8150684931506852e-06, + "loss": 2.2195, + "step": 120 + }, + { + "epoch": 0.12326656394453005, + "eval_loss": 2.2648775577545166, + "eval_runtime": 49.47, + "eval_samples_per_second": 20.214, + "eval_steps_per_second": 0.849, + "step": 120 + }, + { + "epoch": 0.12429378531073447, + "grad_norm": 2.6743483543395996, + "learning_rate": 1.832191780821918e-06, + "loss": 2.1521, + "step": 121 + }, + { + "epoch": 0.12532100667693888, + "grad_norm": 2.54795503616333, + "learning_rate": 1.8493150684931507e-06, + "loss": 2.1158, + "step": 122 + }, + { + "epoch": 0.1263482280431433, + "grad_norm": 2.9093267917633057, + "learning_rate": 1.8664383561643837e-06, + "loss": 2.0707, + "step": 123 + }, + { + "epoch": 0.12737544940934772, + "grad_norm": 2.416940212249756, + "learning_rate": 1.8835616438356166e-06, + "loss": 2.0901, + "step": 124 + }, + { + "epoch": 0.12840267077555212, + "grad_norm": 2.8333663940429688, + "learning_rate": 1.9006849315068496e-06, + "loss": 2.0545, + "step": 125 + }, + { + "epoch": 0.12942989214175654, + "grad_norm": 2.483051061630249, + "learning_rate": 1.9178082191780823e-06, + "loss": 2.1274, + "step": 126 + }, + { + "epoch": 0.13045711350796096, + "grad_norm": 2.3486859798431396, + "learning_rate": 1.9349315068493153e-06, + "loss": 2.0478, + "step": 127 + }, + { + "epoch": 0.1314843348741654, + "grad_norm": 3.1512887477874756, + "learning_rate": 1.952054794520548e-06, + "loss": 2.1041, + "step": 128 + }, + { + "epoch": 0.1325115562403698, + "grad_norm": 2.902873992919922, + "learning_rate": 1.969178082191781e-06, + "loss": 2.1332, + "step": 129 + }, + { + "epoch": 0.1335387776065742, + "grad_norm": 2.779132127761841, + "learning_rate": 1.9863013698630136e-06, + "loss": 2.1297, + "step": 130 + }, + { + "epoch": 0.1335387776065742, + "eval_loss": 2.2391879558563232, + "eval_runtime": 55.3018, + "eval_samples_per_second": 18.083, + "eval_steps_per_second": 0.759, + "step": 130 + }, + { + "epoch": 0.13456599897277863, + "grad_norm": 2.703019618988037, + "learning_rate": 2.0034246575342467e-06, + "loss": 2.0506, + "step": 131 + }, + { + "epoch": 0.13559322033898305, + "grad_norm": 3.0187575817108154, + "learning_rate": 2.0205479452054797e-06, + "loss": 2.1291, + "step": 132 + }, + { + "epoch": 0.13662044170518747, + "grad_norm": 2.7629971504211426, + "learning_rate": 2.037671232876713e-06, + "loss": 2.1003, + "step": 133 + }, + { + "epoch": 0.1376476630713919, + "grad_norm": 3.175776720046997, + "learning_rate": 2.0547945205479454e-06, + "loss": 2.0429, + "step": 134 + }, + { + "epoch": 0.1386748844375963, + "grad_norm": 3.0342514514923096, + "learning_rate": 2.071917808219178e-06, + "loss": 2.0858, + "step": 135 + }, + { + "epoch": 0.1397021058038007, + "grad_norm": 3.0861928462982178, + "learning_rate": 2.089041095890411e-06, + "loss": 2.0964, + "step": 136 + }, + { + "epoch": 0.14072932717000514, + "grad_norm": 3.151945114135742, + "learning_rate": 2.106164383561644e-06, + "loss": 2.0194, + "step": 137 + }, + { + "epoch": 0.14175654853620956, + "grad_norm": 2.9069676399230957, + "learning_rate": 2.123287671232877e-06, + "loss": 2.1131, + "step": 138 + }, + { + "epoch": 0.14278376990241398, + "grad_norm": 3.2285282611846924, + "learning_rate": 2.14041095890411e-06, + "loss": 2.0643, + "step": 139 + }, + { + "epoch": 0.14381099126861838, + "grad_norm": 2.500481367111206, + "learning_rate": 2.1575342465753425e-06, + "loss": 2.0625, + "step": 140 + }, + { + "epoch": 0.14381099126861838, + "eval_loss": 2.211498975753784, + "eval_runtime": 55.1046, + "eval_samples_per_second": 18.147, + "eval_steps_per_second": 0.762, + "step": 140 + }, + { + "epoch": 0.1448382126348228, + "grad_norm": 4.004592418670654, + "learning_rate": 2.1746575342465755e-06, + "loss": 2.0644, + "step": 141 + }, + { + "epoch": 0.14586543400102722, + "grad_norm": 2.9258391857147217, + "learning_rate": 2.191780821917808e-06, + "loss": 2.0174, + "step": 142 + }, + { + "epoch": 0.14689265536723164, + "grad_norm": 3.0073509216308594, + "learning_rate": 2.2089041095890412e-06, + "loss": 2.1417, + "step": 143 + }, + { + "epoch": 0.14791987673343607, + "grad_norm": 3.040708065032959, + "learning_rate": 2.2260273972602743e-06, + "loss": 2.0357, + "step": 144 + }, + { + "epoch": 0.14894709809964046, + "grad_norm": 2.9062869548797607, + "learning_rate": 2.243150684931507e-06, + "loss": 2.0122, + "step": 145 + }, + { + "epoch": 0.14997431946584489, + "grad_norm": 3.28969669342041, + "learning_rate": 2.26027397260274e-06, + "loss": 2.0255, + "step": 146 + }, + { + "epoch": 0.1510015408320493, + "grad_norm": 3.608752727508545, + "learning_rate": 2.2773972602739726e-06, + "loss": 2.0184, + "step": 147 + }, + { + "epoch": 0.15202876219825373, + "grad_norm": 2.900341033935547, + "learning_rate": 2.2945205479452057e-06, + "loss": 2.0532, + "step": 148 + }, + { + "epoch": 0.15305598356445815, + "grad_norm": 2.881843328475952, + "learning_rate": 2.3116438356164387e-06, + "loss": 2.1336, + "step": 149 + }, + { + "epoch": 0.15408320493066255, + "grad_norm": 3.4043259620666504, + "learning_rate": 2.3287671232876713e-06, + "loss": 2.0536, + "step": 150 + }, + { + "epoch": 0.15408320493066255, + "eval_loss": 2.192880153656006, + "eval_runtime": 53.7838, + "eval_samples_per_second": 18.593, + "eval_steps_per_second": 0.781, + "step": 150 + }, + { + "epoch": 0.15511042629686697, + "grad_norm": 2.693107843399048, + "learning_rate": 2.3458904109589044e-06, + "loss": 1.9503, + "step": 151 + }, + { + "epoch": 0.1561376476630714, + "grad_norm": 2.885260581970215, + "learning_rate": 2.363013698630137e-06, + "loss": 2.0081, + "step": 152 + }, + { + "epoch": 0.15716486902927582, + "grad_norm": 3.4459924697875977, + "learning_rate": 2.38013698630137e-06, + "loss": 2.0649, + "step": 153 + }, + { + "epoch": 0.15819209039548024, + "grad_norm": 2.8447093963623047, + "learning_rate": 2.3972602739726027e-06, + "loss": 2.0476, + "step": 154 + }, + { + "epoch": 0.15921931176168463, + "grad_norm": 2.755176305770874, + "learning_rate": 2.4143835616438358e-06, + "loss": 2.0006, + "step": 155 + }, + { + "epoch": 0.16024653312788906, + "grad_norm": 2.8677561283111572, + "learning_rate": 2.431506849315069e-06, + "loss": 1.9663, + "step": 156 + }, + { + "epoch": 0.16127375449409348, + "grad_norm": 2.783351182937622, + "learning_rate": 2.4486301369863015e-06, + "loss": 2.0301, + "step": 157 + }, + { + "epoch": 0.1623009758602979, + "grad_norm": 2.7545998096466064, + "learning_rate": 2.4657534246575345e-06, + "loss": 2.0438, + "step": 158 + }, + { + "epoch": 0.1633281972265023, + "grad_norm": 2.8140594959259033, + "learning_rate": 2.482876712328767e-06, + "loss": 2.0055, + "step": 159 + }, + { + "epoch": 0.16435541859270672, + "grad_norm": 2.6072335243225098, + "learning_rate": 2.5e-06, + "loss": 2.0262, + "step": 160 + }, + { + "epoch": 0.16435541859270672, + "eval_loss": 2.179842710494995, + "eval_runtime": 52.6159, + "eval_samples_per_second": 19.006, + "eval_steps_per_second": 0.798, + "step": 160 + }, + { + "epoch": 0.16538263995891114, + "grad_norm": 2.830212354660034, + "learning_rate": 2.5171232876712333e-06, + "loss": 2.0154, + "step": 161 + }, + { + "epoch": 0.16640986132511557, + "grad_norm": 2.677140712738037, + "learning_rate": 2.534246575342466e-06, + "loss": 1.9933, + "step": 162 + }, + { + "epoch": 0.16743708269132, + "grad_norm": 2.837759017944336, + "learning_rate": 2.551369863013699e-06, + "loss": 1.9602, + "step": 163 + }, + { + "epoch": 0.16846430405752438, + "grad_norm": 3.9556589126586914, + "learning_rate": 2.568493150684932e-06, + "loss": 2.0121, + "step": 164 + }, + { + "epoch": 0.1694915254237288, + "grad_norm": 2.8715925216674805, + "learning_rate": 2.585616438356164e-06, + "loss": 2.0219, + "step": 165 + }, + { + "epoch": 0.17051874678993323, + "grad_norm": 3.5073623657226562, + "learning_rate": 2.6027397260273973e-06, + "loss": 2.0202, + "step": 166 + }, + { + "epoch": 0.17154596815613765, + "grad_norm": 2.751962900161743, + "learning_rate": 2.6198630136986303e-06, + "loss": 1.9528, + "step": 167 + }, + { + "epoch": 0.17257318952234207, + "grad_norm": 3.7753474712371826, + "learning_rate": 2.6369863013698634e-06, + "loss": 2.0208, + "step": 168 + }, + { + "epoch": 0.17360041088854647, + "grad_norm": 2.642289876937866, + "learning_rate": 2.654109589041096e-06, + "loss": 2.0386, + "step": 169 + }, + { + "epoch": 0.1746276322547509, + "grad_norm": 2.82173490524292, + "learning_rate": 2.671232876712329e-06, + "loss": 2.0355, + "step": 170 + }, + { + "epoch": 0.1746276322547509, + "eval_loss": 2.1699249744415283, + "eval_runtime": 50.0341, + "eval_samples_per_second": 19.986, + "eval_steps_per_second": 0.839, + "step": 170 + }, + { + "epoch": 0.17565485362095531, + "grad_norm": 3.342280149459839, + "learning_rate": 2.688356164383562e-06, + "loss": 1.9672, + "step": 171 + }, + { + "epoch": 0.17668207498715974, + "grad_norm": 2.6548173427581787, + "learning_rate": 2.705479452054795e-06, + "loss": 1.9669, + "step": 172 + }, + { + "epoch": 0.17770929635336416, + "grad_norm": 2.6548173427581787, + "learning_rate": 2.705479452054795e-06, + "loss": 1.961, + "step": 173 + }, + { + "epoch": 0.17873651771956856, + "grad_norm": 3.4306116104125977, + "learning_rate": 2.7226027397260274e-06, + "loss": 1.9173, + "step": 174 + }, + { + "epoch": 0.17976373908577298, + "grad_norm": 2.7991764545440674, + "learning_rate": 2.7397260273972604e-06, + "loss": 2.0339, + "step": 175 + }, + { + "epoch": 0.1807909604519774, + "grad_norm": 3.6040217876434326, + "learning_rate": 2.7568493150684935e-06, + "loss": 2.0218, + "step": 176 + }, + { + "epoch": 0.18181818181818182, + "grad_norm": 3.0253515243530273, + "learning_rate": 2.773972602739726e-06, + "loss": 1.9245, + "step": 177 + }, + { + "epoch": 0.18284540318438625, + "grad_norm": 3.405141830444336, + "learning_rate": 2.791095890410959e-06, + "loss": 2.0136, + "step": 178 + }, + { + "epoch": 0.18387262455059064, + "grad_norm": 2.7518203258514404, + "learning_rate": 2.8082191780821922e-06, + "loss": 1.8925, + "step": 179 + }, + { + "epoch": 0.18489984591679506, + "grad_norm": 3.1903367042541504, + "learning_rate": 2.8253424657534253e-06, + "loss": 2.0523, + "step": 180 + }, + { + "epoch": 0.18489984591679506, + "eval_loss": 2.158029794692993, + "eval_runtime": 48.6162, + "eval_samples_per_second": 20.569, + "eval_steps_per_second": 0.864, + "step": 180 + }, + { + "epoch": 0.1859270672829995, + "grad_norm": 3.0331413745880127, + "learning_rate": 2.842465753424658e-06, + "loss": 2.0167, + "step": 181 + }, + { + "epoch": 0.1869542886492039, + "grad_norm": 2.919433116912842, + "learning_rate": 2.8595890410958905e-06, + "loss": 2.0197, + "step": 182 + }, + { + "epoch": 0.18798151001540833, + "grad_norm": 2.9013049602508545, + "learning_rate": 2.876712328767123e-06, + "loss": 2.0301, + "step": 183 + }, + { + "epoch": 0.18900873138161273, + "grad_norm": 2.738800525665283, + "learning_rate": 2.8938356164383562e-06, + "loss": 1.894, + "step": 184 + }, + { + "epoch": 0.19003595274781715, + "grad_norm": 2.670806407928467, + "learning_rate": 2.9109589041095893e-06, + "loss": 1.9127, + "step": 185 + }, + { + "epoch": 0.19106317411402157, + "grad_norm": 3.0435004234313965, + "learning_rate": 2.9280821917808223e-06, + "loss": 1.8996, + "step": 186 + }, + { + "epoch": 0.192090395480226, + "grad_norm": 3.2139949798583984, + "learning_rate": 2.945205479452055e-06, + "loss": 1.9255, + "step": 187 + }, + { + "epoch": 0.19311761684643042, + "grad_norm": 3.252965211868286, + "learning_rate": 2.962328767123288e-06, + "loss": 1.9809, + "step": 188 + }, + { + "epoch": 0.1941448382126348, + "grad_norm": 2.9325191974639893, + "learning_rate": 2.979452054794521e-06, + "loss": 1.9691, + "step": 189 + }, + { + "epoch": 0.19517205957883924, + "grad_norm": 3.5879862308502197, + "learning_rate": 2.9965753424657533e-06, + "loss": 1.9205, + "step": 190 + }, + { + "epoch": 0.19517205957883924, + "eval_loss": 2.13791561126709, + "eval_runtime": 44.8545, + "eval_samples_per_second": 22.294, + "eval_steps_per_second": 0.936, + "step": 190 + }, + { + "epoch": 0.19619928094504366, + "grad_norm": 3.0025253295898438, + "learning_rate": 3.0136986301369864e-06, + "loss": 1.9717, + "step": 191 + }, + { + "epoch": 0.19722650231124808, + "grad_norm": 3.1785972118377686, + "learning_rate": 3.0308219178082194e-06, + "loss": 1.9535, + "step": 192 + }, + { + "epoch": 0.1982537236774525, + "grad_norm": 3.121065378189087, + "learning_rate": 3.0479452054794525e-06, + "loss": 1.9026, + "step": 193 + }, + { + "epoch": 0.1992809450436569, + "grad_norm": 2.838779926300049, + "learning_rate": 3.065068493150685e-06, + "loss": 1.941, + "step": 194 + }, + { + "epoch": 0.20030816640986132, + "grad_norm": 2.8916969299316406, + "learning_rate": 3.082191780821918e-06, + "loss": 1.8983, + "step": 195 + }, + { + "epoch": 0.20133538777606574, + "grad_norm": 2.9195168018341064, + "learning_rate": 3.099315068493151e-06, + "loss": 1.918, + "step": 196 + }, + { + "epoch": 0.20236260914227017, + "grad_norm": 3.235989809036255, + "learning_rate": 3.1164383561643843e-06, + "loss": 1.9111, + "step": 197 + }, + { + "epoch": 0.2033898305084746, + "grad_norm": 3.0316455364227295, + "learning_rate": 3.1335616438356165e-06, + "loss": 1.8839, + "step": 198 + }, + { + "epoch": 0.20441705187467898, + "grad_norm": 2.963610887527466, + "learning_rate": 3.1506849315068495e-06, + "loss": 2.0303, + "step": 199 + }, + { + "epoch": 0.2054442732408834, + "grad_norm": 2.944054365158081, + "learning_rate": 3.167808219178082e-06, + "loss": 1.8822, + "step": 200 + }, + { + "epoch": 0.2054442732408834, + "eval_loss": 2.1433050632476807, + "eval_runtime": 45.8205, + "eval_samples_per_second": 21.824, + "eval_steps_per_second": 0.917, + "step": 200 + }, + { + "epoch": 0.20647149460708783, + "grad_norm": 2.8648695945739746, + "learning_rate": 3.184931506849315e-06, + "loss": 1.8841, + "step": 201 + }, + { + "epoch": 0.20749871597329225, + "grad_norm": 3.2838797569274902, + "learning_rate": 3.2020547945205483e-06, + "loss": 1.8623, + "step": 202 + }, + { + "epoch": 0.20852593733949668, + "grad_norm": 2.7965173721313477, + "learning_rate": 3.2191780821917813e-06, + "loss": 1.8521, + "step": 203 + }, + { + "epoch": 0.20955315870570107, + "grad_norm": 3.3449349403381348, + "learning_rate": 3.236301369863014e-06, + "loss": 1.84, + "step": 204 + }, + { + "epoch": 0.2105803800719055, + "grad_norm": 3.4888951778411865, + "learning_rate": 3.253424657534247e-06, + "loss": 1.824, + "step": 205 + }, + { + "epoch": 0.21160760143810992, + "grad_norm": 2.991349220275879, + "learning_rate": 3.2705479452054796e-06, + "loss": 1.9101, + "step": 206 + }, + { + "epoch": 0.21263482280431434, + "grad_norm": 3.313509702682495, + "learning_rate": 3.2876712328767123e-06, + "loss": 1.8912, + "step": 207 + }, + { + "epoch": 0.21366204417051873, + "grad_norm": 2.8260746002197266, + "learning_rate": 3.3047945205479453e-06, + "loss": 1.8555, + "step": 208 + }, + { + "epoch": 0.21468926553672316, + "grad_norm": 2.7131688594818115, + "learning_rate": 3.3219178082191784e-06, + "loss": 1.8693, + "step": 209 + }, + { + "epoch": 0.21571648690292758, + "grad_norm": 2.5836338996887207, + "learning_rate": 3.3390410958904114e-06, + "loss": 1.8965, + "step": 210 + }, + { + "epoch": 0.21571648690292758, + "eval_loss": 2.1232542991638184, + "eval_runtime": 42.3352, + "eval_samples_per_second": 23.621, + "eval_steps_per_second": 0.992, + "step": 210 + }, + { + "epoch": 0.216743708269132, + "grad_norm": 2.8341355323791504, + "learning_rate": 3.356164383561644e-06, + "loss": 1.9131, + "step": 211 + }, + { + "epoch": 0.21777092963533642, + "grad_norm": 2.490853786468506, + "learning_rate": 3.373287671232877e-06, + "loss": 1.9626, + "step": 212 + }, + { + "epoch": 0.21879815100154082, + "grad_norm": 2.597956895828247, + "learning_rate": 3.39041095890411e-06, + "loss": 1.9203, + "step": 213 + }, + { + "epoch": 0.21982537236774524, + "grad_norm": 3.1104283332824707, + "learning_rate": 3.4075342465753424e-06, + "loss": 1.9442, + "step": 214 + }, + { + "epoch": 0.22085259373394966, + "grad_norm": 2.5698931217193604, + "learning_rate": 3.4246575342465754e-06, + "loss": 1.8385, + "step": 215 + }, + { + "epoch": 0.2218798151001541, + "grad_norm": 3.1637446880340576, + "learning_rate": 3.4417808219178085e-06, + "loss": 1.9005, + "step": 216 + }, + { + "epoch": 0.2229070364663585, + "grad_norm": 2.764070749282837, + "learning_rate": 3.458904109589041e-06, + "loss": 1.8886, + "step": 217 + }, + { + "epoch": 0.2239342578325629, + "grad_norm": 3.303781032562256, + "learning_rate": 3.476027397260274e-06, + "loss": 1.9085, + "step": 218 + }, + { + "epoch": 0.22496147919876733, + "grad_norm": 2.9543044567108154, + "learning_rate": 3.4931506849315072e-06, + "loss": 1.9014, + "step": 219 + }, + { + "epoch": 0.22598870056497175, + "grad_norm": 2.7277028560638428, + "learning_rate": 3.5102739726027403e-06, + "loss": 1.9017, + "step": 220 + }, + { + "epoch": 0.22598870056497175, + "eval_loss": 2.1024081707000732, + "eval_runtime": 45.4966, + "eval_samples_per_second": 21.98, + "eval_steps_per_second": 0.923, + "step": 220 + }, + { + "epoch": 0.22701592193117617, + "grad_norm": 2.864140033721924, + "learning_rate": 3.527397260273973e-06, + "loss": 1.8935, + "step": 221 + }, + { + "epoch": 0.2280431432973806, + "grad_norm": 3.4133172035217285, + "learning_rate": 3.5445205479452056e-06, + "loss": 1.874, + "step": 222 + }, + { + "epoch": 0.229070364663585, + "grad_norm": 3.33343243598938, + "learning_rate": 3.5616438356164386e-06, + "loss": 1.8789, + "step": 223 + }, + { + "epoch": 0.2300975860297894, + "grad_norm": 3.5947320461273193, + "learning_rate": 3.5787671232876712e-06, + "loss": 1.9529, + "step": 224 + }, + { + "epoch": 0.23112480739599384, + "grad_norm": 3.2194948196411133, + "learning_rate": 3.5958904109589043e-06, + "loss": 1.9205, + "step": 225 + }, + { + "epoch": 0.23215202876219826, + "grad_norm": 3.3137149810791016, + "learning_rate": 3.6130136986301373e-06, + "loss": 1.8584, + "step": 226 + }, + { + "epoch": 0.23317925012840268, + "grad_norm": 3.3358561992645264, + "learning_rate": 3.6301369863013704e-06, + "loss": 1.8817, + "step": 227 + }, + { + "epoch": 0.23420647149460708, + "grad_norm": 3.8612711429595947, + "learning_rate": 3.647260273972603e-06, + "loss": 1.9189, + "step": 228 + }, + { + "epoch": 0.2352336928608115, + "grad_norm": 3.0039336681365967, + "learning_rate": 3.664383561643836e-06, + "loss": 1.92, + "step": 229 + }, + { + "epoch": 0.23626091422701592, + "grad_norm": 3.8983118534088135, + "learning_rate": 3.6815068493150687e-06, + "loss": 1.8512, + "step": 230 + }, + { + "epoch": 0.23626091422701592, + "eval_loss": 2.1237711906433105, + "eval_runtime": 45.6126, + "eval_samples_per_second": 21.924, + "eval_steps_per_second": 0.921, + "step": 230 + }, + { + "epoch": 0.23728813559322035, + "grad_norm": 3.061300039291382, + "learning_rate": 3.6986301369863014e-06, + "loss": 1.8772, + "step": 231 + }, + { + "epoch": 0.23831535695942477, + "grad_norm": 3.3102450370788574, + "learning_rate": 3.7157534246575344e-06, + "loss": 1.8395, + "step": 232 + }, + { + "epoch": 0.23934257832562916, + "grad_norm": 2.9634461402893066, + "learning_rate": 3.7328767123287675e-06, + "loss": 1.9447, + "step": 233 + }, + { + "epoch": 0.24036979969183359, + "grad_norm": 3.0937695503234863, + "learning_rate": 3.7500000000000005e-06, + "loss": 1.8628, + "step": 234 + }, + { + "epoch": 0.241397021058038, + "grad_norm": 2.8129842281341553, + "learning_rate": 3.767123287671233e-06, + "loss": 1.8891, + "step": 235 + }, + { + "epoch": 0.24242424242424243, + "grad_norm": 2.9616780281066895, + "learning_rate": 3.784246575342466e-06, + "loss": 1.9112, + "step": 236 + }, + { + "epoch": 0.24345146379044685, + "grad_norm": 2.837721109390259, + "learning_rate": 3.8013698630136993e-06, + "loss": 1.8514, + "step": 237 + }, + { + "epoch": 0.24447868515665125, + "grad_norm": 3.137784004211426, + "learning_rate": 3.818493150684932e-06, + "loss": 1.8764, + "step": 238 + }, + { + "epoch": 0.24550590652285567, + "grad_norm": 2.9692037105560303, + "learning_rate": 3.8356164383561645e-06, + "loss": 1.8201, + "step": 239 + }, + { + "epoch": 0.2465331278890601, + "grad_norm": 2.6718621253967285, + "learning_rate": 3.852739726027397e-06, + "loss": 1.9309, + "step": 240 + }, + { + "epoch": 0.2465331278890601, + "eval_loss": 2.101331949234009, + "eval_runtime": 45.6346, + "eval_samples_per_second": 21.913, + "eval_steps_per_second": 0.92, + "step": 240 + }, + { + "epoch": 0.24756034925526452, + "grad_norm": 3.411193609237671, + "learning_rate": 3.869863013698631e-06, + "loss": 1.9171, + "step": 241 + }, + { + "epoch": 0.24858757062146894, + "grad_norm": 2.8848962783813477, + "learning_rate": 3.886986301369863e-06, + "loss": 1.8587, + "step": 242 + }, + { + "epoch": 0.24961479198767333, + "grad_norm": 2.756298065185547, + "learning_rate": 3.904109589041096e-06, + "loss": 1.8614, + "step": 243 + }, + { + "epoch": 0.25064201335387776, + "grad_norm": 2.8048956394195557, + "learning_rate": 3.921232876712329e-06, + "loss": 1.8225, + "step": 244 + }, + { + "epoch": 0.2516692347200822, + "grad_norm": 2.9872403144836426, + "learning_rate": 3.938356164383562e-06, + "loss": 1.8367, + "step": 245 + }, + { + "epoch": 0.2526964560862866, + "grad_norm": 2.710156202316284, + "learning_rate": 3.9554794520547955e-06, + "loss": 1.809, + "step": 246 + }, + { + "epoch": 0.253723677452491, + "grad_norm": 3.2177000045776367, + "learning_rate": 3.972602739726027e-06, + "loss": 1.8883, + "step": 247 + }, + { + "epoch": 0.25475089881869545, + "grad_norm": 3.160964250564575, + "learning_rate": 3.989726027397261e-06, + "loss": 1.8267, + "step": 248 + }, + { + "epoch": 0.25577812018489987, + "grad_norm": 2.853498935699463, + "learning_rate": 4.006849315068493e-06, + "loss": 1.8852, + "step": 249 + }, + { + "epoch": 0.25680534155110424, + "grad_norm": 2.571742296218872, + "learning_rate": 4.023972602739726e-06, + "loss": 1.8833, + "step": 250 + }, + { + "epoch": 0.25680534155110424, + "eval_loss": 2.0988597869873047, + "eval_runtime": 47.7045, + "eval_samples_per_second": 20.962, + "eval_steps_per_second": 0.88, + "step": 250 + }, + { + "epoch": 0.25783256291730866, + "grad_norm": 2.776742458343506, + "learning_rate": 4.0410958904109595e-06, + "loss": 1.8662, + "step": 251 + }, + { + "epoch": 0.2588597842835131, + "grad_norm": 2.8038573265075684, + "learning_rate": 4.058219178082192e-06, + "loss": 1.8528, + "step": 252 + }, + { + "epoch": 0.2598870056497175, + "grad_norm": 2.557623863220215, + "learning_rate": 4.075342465753426e-06, + "loss": 1.8765, + "step": 253 + }, + { + "epoch": 0.26091422701592193, + "grad_norm": 3.0477073192596436, + "learning_rate": 4.092465753424658e-06, + "loss": 1.8434, + "step": 254 + }, + { + "epoch": 0.26194144838212635, + "grad_norm": 2.6221649646759033, + "learning_rate": 4.109589041095891e-06, + "loss": 1.8348, + "step": 255 + }, + { + "epoch": 0.2629686697483308, + "grad_norm": 3.0849292278289795, + "learning_rate": 4.1267123287671235e-06, + "loss": 1.9057, + "step": 256 + }, + { + "epoch": 0.2639958911145352, + "grad_norm": 2.795184373855591, + "learning_rate": 4.143835616438356e-06, + "loss": 1.8528, + "step": 257 + }, + { + "epoch": 0.2650231124807396, + "grad_norm": 2.887193441390991, + "learning_rate": 4.16095890410959e-06, + "loss": 1.9176, + "step": 258 + }, + { + "epoch": 0.26605033384694404, + "grad_norm": 2.776801347732544, + "learning_rate": 4.178082191780822e-06, + "loss": 1.8911, + "step": 259 + }, + { + "epoch": 0.2670775552131484, + "grad_norm": 2.980426788330078, + "learning_rate": 4.195205479452055e-06, + "loss": 1.8288, + "step": 260 + }, + { + "epoch": 0.2670775552131484, + "eval_loss": 2.0636370182037354, + "eval_runtime": 49.3081, + "eval_samples_per_second": 20.281, + "eval_steps_per_second": 0.852, + "step": 260 + }, + { + "epoch": 0.26810477657935283, + "grad_norm": 2.7921340465545654, + "learning_rate": 4.212328767123288e-06, + "loss": 1.8405, + "step": 261 + }, + { + "epoch": 0.26913199794555726, + "grad_norm": 2.67714262008667, + "learning_rate": 4.229452054794521e-06, + "loss": 1.8342, + "step": 262 + }, + { + "epoch": 0.2701592193117617, + "grad_norm": 2.819868326187134, + "learning_rate": 4.246575342465754e-06, + "loss": 1.8466, + "step": 263 + }, + { + "epoch": 0.2711864406779661, + "grad_norm": 2.611250638961792, + "learning_rate": 4.263698630136986e-06, + "loss": 1.8314, + "step": 264 + }, + { + "epoch": 0.2722136620441705, + "grad_norm": 3.0406076908111572, + "learning_rate": 4.28082191780822e-06, + "loss": 1.7924, + "step": 265 + }, + { + "epoch": 0.27324088341037495, + "grad_norm": 3.212790012359619, + "learning_rate": 4.297945205479452e-06, + "loss": 1.9479, + "step": 266 + }, + { + "epoch": 0.27426810477657937, + "grad_norm": 2.4767045974731445, + "learning_rate": 4.315068493150685e-06, + "loss": 1.835, + "step": 267 + }, + { + "epoch": 0.2752953261427838, + "grad_norm": 3.191835641860962, + "learning_rate": 4.3321917808219185e-06, + "loss": 1.8261, + "step": 268 + }, + { + "epoch": 0.2763225475089882, + "grad_norm": 2.9970273971557617, + "learning_rate": 4.349315068493151e-06, + "loss": 1.8427, + "step": 269 + }, + { + "epoch": 0.2773497688751926, + "grad_norm": 2.888843297958374, + "learning_rate": 4.3664383561643846e-06, + "loss": 1.819, + "step": 270 + }, + { + "epoch": 0.2773497688751926, + "eval_loss": 2.054966926574707, + "eval_runtime": 45.9212, + "eval_samples_per_second": 21.776, + "eval_steps_per_second": 0.915, + "step": 270 + }, + { + "epoch": 0.278376990241397, + "grad_norm": 2.7771849632263184, + "learning_rate": 4.383561643835616e-06, + "loss": 1.8178, + "step": 271 + }, + { + "epoch": 0.2794042116076014, + "grad_norm": 2.9749014377593994, + "learning_rate": 4.40068493150685e-06, + "loss": 1.8982, + "step": 272 + }, + { + "epoch": 0.28043143297380585, + "grad_norm": 3.076021432876587, + "learning_rate": 4.4178082191780825e-06, + "loss": 1.8575, + "step": 273 + }, + { + "epoch": 0.2814586543400103, + "grad_norm": 3.2113568782806396, + "learning_rate": 4.434931506849315e-06, + "loss": 1.8241, + "step": 274 + }, + { + "epoch": 0.2824858757062147, + "grad_norm": 3.322063684463501, + "learning_rate": 4.4520547945205486e-06, + "loss": 1.7922, + "step": 275 + }, + { + "epoch": 0.2835130970724191, + "grad_norm": 3.2778918743133545, + "learning_rate": 4.469178082191781e-06, + "loss": 1.8412, + "step": 276 + }, + { + "epoch": 0.28454031843862354, + "grad_norm": 3.1802618503570557, + "learning_rate": 4.486301369863014e-06, + "loss": 1.8283, + "step": 277 + }, + { + "epoch": 0.28556753980482796, + "grad_norm": 2.9214258193969727, + "learning_rate": 4.503424657534247e-06, + "loss": 1.8817, + "step": 278 + }, + { + "epoch": 0.28659476117103233, + "grad_norm": 2.8809854984283447, + "learning_rate": 4.52054794520548e-06, + "loss": 1.7662, + "step": 279 + }, + { + "epoch": 0.28762198253723675, + "grad_norm": 2.925537586212158, + "learning_rate": 4.537671232876713e-06, + "loss": 1.8147, + "step": 280 + }, + { + "epoch": 0.28762198253723675, + "eval_loss": 2.054513454437256, + "eval_runtime": 51.5036, + "eval_samples_per_second": 19.416, + "eval_steps_per_second": 0.815, + "step": 280 + }, + { + "epoch": 0.2886492039034412, + "grad_norm": 2.9421887397766113, + "learning_rate": 4.554794520547945e-06, + "loss": 1.8581, + "step": 281 + }, + { + "epoch": 0.2896764252696456, + "grad_norm": 2.9733567237854004, + "learning_rate": 4.571917808219179e-06, + "loss": 1.8703, + "step": 282 + }, + { + "epoch": 0.29070364663585, + "grad_norm": 2.5352156162261963, + "learning_rate": 4.589041095890411e-06, + "loss": 1.8505, + "step": 283 + }, + { + "epoch": 0.29173086800205444, + "grad_norm": 2.773847818374634, + "learning_rate": 4.606164383561644e-06, + "loss": 1.8114, + "step": 284 + }, + { + "epoch": 0.29275808936825887, + "grad_norm": 2.709632635116577, + "learning_rate": 4.6232876712328774e-06, + "loss": 1.815, + "step": 285 + }, + { + "epoch": 0.2937853107344633, + "grad_norm": 3.1234357357025146, + "learning_rate": 4.64041095890411e-06, + "loss": 1.8391, + "step": 286 + }, + { + "epoch": 0.2948125321006677, + "grad_norm": 2.9376144409179688, + "learning_rate": 4.657534246575343e-06, + "loss": 1.7998, + "step": 287 + }, + { + "epoch": 0.29583975346687214, + "grad_norm": 3.0812032222747803, + "learning_rate": 4.674657534246575e-06, + "loss": 1.8462, + "step": 288 + }, + { + "epoch": 0.2968669748330765, + "grad_norm": 2.922722578048706, + "learning_rate": 4.691780821917809e-06, + "loss": 1.8263, + "step": 289 + }, + { + "epoch": 0.2978941961992809, + "grad_norm": 2.9484992027282715, + "learning_rate": 4.7089041095890414e-06, + "loss": 1.8754, + "step": 290 + }, + { + "epoch": 0.2978941961992809, + "eval_loss": 2.032902240753174, + "eval_runtime": 47.8711, + "eval_samples_per_second": 20.889, + "eval_steps_per_second": 0.877, + "step": 290 + }, + { + "epoch": 0.29892141756548535, + "grad_norm": 3.0068976879119873, + "learning_rate": 4.726027397260274e-06, + "loss": 1.8256, + "step": 291 + }, + { + "epoch": 0.29994863893168977, + "grad_norm": 2.6469006538391113, + "learning_rate": 4.7431506849315075e-06, + "loss": 1.8354, + "step": 292 + }, + { + "epoch": 0.3009758602978942, + "grad_norm": 3.06986927986145, + "learning_rate": 4.76027397260274e-06, + "loss": 1.8402, + "step": 293 + }, + { + "epoch": 0.3020030816640986, + "grad_norm": 2.8409981727600098, + "learning_rate": 4.777397260273973e-06, + "loss": 1.7915, + "step": 294 + }, + { + "epoch": 0.30303030303030304, + "grad_norm": 3.1027376651763916, + "learning_rate": 4.7945205479452054e-06, + "loss": 1.8153, + "step": 295 + }, + { + "epoch": 0.30405752439650746, + "grad_norm": 2.7221434116363525, + "learning_rate": 4.811643835616439e-06, + "loss": 1.8098, + "step": 296 + }, + { + "epoch": 0.3050847457627119, + "grad_norm": 3.0559370517730713, + "learning_rate": 4.8287671232876716e-06, + "loss": 1.7504, + "step": 297 + }, + { + "epoch": 0.3061119671289163, + "grad_norm": 2.9628236293792725, + "learning_rate": 4.845890410958904e-06, + "loss": 1.8019, + "step": 298 + }, + { + "epoch": 0.3071391884951207, + "grad_norm": 2.6248221397399902, + "learning_rate": 4.863013698630138e-06, + "loss": 1.8001, + "step": 299 + }, + { + "epoch": 0.3081664098613251, + "grad_norm": 3.0477702617645264, + "learning_rate": 4.88013698630137e-06, + "loss": 1.7915, + "step": 300 + }, + { + "epoch": 0.3081664098613251, + "eval_loss": 2.0522429943084717, + "eval_runtime": 46.796, + "eval_samples_per_second": 21.369, + "eval_steps_per_second": 0.898, + "step": 300 + }, + { + "epoch": 0.3091936312275295, + "grad_norm": 2.784707546234131, + "learning_rate": 4.897260273972603e-06, + "loss": 1.7757, + "step": 301 + }, + { + "epoch": 0.31022085259373394, + "grad_norm": 2.8283841609954834, + "learning_rate": 4.914383561643836e-06, + "loss": 1.8975, + "step": 302 + }, + { + "epoch": 0.31124807395993837, + "grad_norm": 2.879342555999756, + "learning_rate": 4.931506849315069e-06, + "loss": 1.8208, + "step": 303 + }, + { + "epoch": 0.3122752953261428, + "grad_norm": 3.2111175060272217, + "learning_rate": 4.948630136986302e-06, + "loss": 1.8631, + "step": 304 + }, + { + "epoch": 0.3133025166923472, + "grad_norm": 2.876168966293335, + "learning_rate": 4.965753424657534e-06, + "loss": 1.8863, + "step": 305 + }, + { + "epoch": 0.31432973805855163, + "grad_norm": 2.8958466053009033, + "learning_rate": 4.982876712328768e-06, + "loss": 1.7565, + "step": 306 + }, + { + "epoch": 0.31535695942475606, + "grad_norm": 2.6702723503112793, + "learning_rate": 5e-06, + "loss": 1.7844, + "step": 307 + }, + { + "epoch": 0.3163841807909605, + "grad_norm": 2.557966470718384, + "learning_rate": 4.998096688237534e-06, + "loss": 1.7472, + "step": 308 + }, + { + "epoch": 0.31741140215716485, + "grad_norm": 2.9257633686065674, + "learning_rate": 4.996193376475067e-06, + "loss": 1.853, + "step": 309 + }, + { + "epoch": 0.31843862352336927, + "grad_norm": 2.497093439102173, + "learning_rate": 4.9942900647126005e-06, + "loss": 1.7313, + "step": 310 + }, + { + "epoch": 0.31843862352336927, + "eval_loss": 2.0336520671844482, + "eval_runtime": 43.3773, + "eval_samples_per_second": 23.054, + "eval_steps_per_second": 0.968, + "step": 310 + }, + { + "epoch": 0.3194658448895737, + "grad_norm": 3.0313165187835693, + "learning_rate": 4.992386752950134e-06, + "loss": 1.7397, + "step": 311 + }, + { + "epoch": 0.3204930662557781, + "grad_norm": 2.629685163497925, + "learning_rate": 4.990483441187667e-06, + "loss": 1.7839, + "step": 312 + }, + { + "epoch": 0.32152028762198254, + "grad_norm": 2.6772103309631348, + "learning_rate": 4.9885801294252e-06, + "loss": 1.8038, + "step": 313 + }, + { + "epoch": 0.32254750898818696, + "grad_norm": 2.893958568572998, + "learning_rate": 4.986676817662733e-06, + "loss": 1.8072, + "step": 314 + }, + { + "epoch": 0.3235747303543914, + "grad_norm": 2.5317318439483643, + "learning_rate": 4.9847735059002665e-06, + "loss": 1.7632, + "step": 315 + }, + { + "epoch": 0.3246019517205958, + "grad_norm": 2.836867332458496, + "learning_rate": 4.9828701941378e-06, + "loss": 1.8499, + "step": 316 + }, + { + "epoch": 0.3256291730868002, + "grad_norm": 2.5564308166503906, + "learning_rate": 4.980966882375333e-06, + "loss": 1.7849, + "step": 317 + }, + { + "epoch": 0.3266563944530046, + "grad_norm": 2.4335806369781494, + "learning_rate": 4.979063570612867e-06, + "loss": 1.7425, + "step": 318 + }, + { + "epoch": 0.327683615819209, + "grad_norm": 2.7140371799468994, + "learning_rate": 4.9771602588504e-06, + "loss": 1.7173, + "step": 319 + }, + { + "epoch": 0.32871083718541344, + "grad_norm": 2.574345111846924, + "learning_rate": 4.975256947087933e-06, + "loss": 1.8307, + "step": 320 + }, + { + "epoch": 0.32871083718541344, + "eval_loss": 2.026960611343384, + "eval_runtime": 53.8982, + "eval_samples_per_second": 18.553, + "eval_steps_per_second": 0.779, + "step": 320 + }, + { + "epoch": 0.32973805855161786, + "grad_norm": 2.6877281665802, + "learning_rate": 4.973353635325467e-06, + "loss": 1.7216, + "step": 321 + }, + { + "epoch": 0.3307652799178223, + "grad_norm": 2.8152849674224854, + "learning_rate": 4.971450323563e-06, + "loss": 1.7831, + "step": 322 + }, + { + "epoch": 0.3317925012840267, + "grad_norm": 2.4208791255950928, + "learning_rate": 4.969547011800533e-06, + "loss": 1.7266, + "step": 323 + }, + { + "epoch": 0.33281972265023113, + "grad_norm": 2.765162229537964, + "learning_rate": 4.967643700038067e-06, + "loss": 1.7971, + "step": 324 + }, + { + "epoch": 0.33384694401643555, + "grad_norm": 2.5018413066864014, + "learning_rate": 4.9657403882756e-06, + "loss": 1.7842, + "step": 325 + }, + { + "epoch": 0.33487416538264, + "grad_norm": 2.901268482208252, + "learning_rate": 4.9638370765131335e-06, + "loss": 1.7522, + "step": 326 + }, + { + "epoch": 0.3359013867488444, + "grad_norm": 2.40067195892334, + "learning_rate": 4.961933764750667e-06, + "loss": 1.7371, + "step": 327 + }, + { + "epoch": 0.33692860811504877, + "grad_norm": 2.6963725090026855, + "learning_rate": 4.9600304529882e-06, + "loss": 1.8036, + "step": 328 + }, + { + "epoch": 0.3379558294812532, + "grad_norm": 2.483790397644043, + "learning_rate": 4.9581271412257335e-06, + "loss": 1.815, + "step": 329 + }, + { + "epoch": 0.3389830508474576, + "grad_norm": 2.7459559440612793, + "learning_rate": 4.956223829463266e-06, + "loss": 1.777, + "step": 330 + }, + { + "epoch": 0.3389830508474576, + "eval_loss": 2.046339988708496, + "eval_runtime": 53.909, + "eval_samples_per_second": 18.55, + "eval_steps_per_second": 0.779, + "step": 330 + }, + { + "epoch": 0.34001027221366203, + "grad_norm": 2.5724353790283203, + "learning_rate": 4.954320517700799e-06, + "loss": 1.7022, + "step": 331 + }, + { + "epoch": 0.34103749357986646, + "grad_norm": 2.510002613067627, + "learning_rate": 4.952417205938333e-06, + "loss": 1.802, + "step": 332 + }, + { + "epoch": 0.3420647149460709, + "grad_norm": 2.8722083568573, + "learning_rate": 4.950513894175866e-06, + "loss": 1.7772, + "step": 333 + }, + { + "epoch": 0.3430919363122753, + "grad_norm": 2.586747407913208, + "learning_rate": 4.9486105824133995e-06, + "loss": 1.7664, + "step": 334 + }, + { + "epoch": 0.3441191576784797, + "grad_norm": 2.8008205890655518, + "learning_rate": 4.946707270650933e-06, + "loss": 1.7892, + "step": 335 + }, + { + "epoch": 0.34514637904468415, + "grad_norm": 2.632664203643799, + "learning_rate": 4.944803958888466e-06, + "loss": 1.7967, + "step": 336 + }, + { + "epoch": 0.34617360041088857, + "grad_norm": 2.8898301124572754, + "learning_rate": 4.942900647126e-06, + "loss": 1.8126, + "step": 337 + }, + { + "epoch": 0.34720082177709294, + "grad_norm": 2.719019651412964, + "learning_rate": 4.940997335363533e-06, + "loss": 1.7544, + "step": 338 + }, + { + "epoch": 0.34822804314329736, + "grad_norm": 2.9995970726013184, + "learning_rate": 4.939094023601066e-06, + "loss": 1.7142, + "step": 339 + }, + { + "epoch": 0.3492552645095018, + "grad_norm": 2.9737157821655273, + "learning_rate": 4.9371907118386e-06, + "loss": 1.7659, + "step": 340 + }, + { + "epoch": 0.3492552645095018, + "eval_loss": 2.024946928024292, + "eval_runtime": 52.8069, + "eval_samples_per_second": 18.937, + "eval_steps_per_second": 0.795, + "step": 340 + }, + { + "epoch": 0.3502824858757062, + "grad_norm": 2.8651719093322754, + "learning_rate": 4.935287400076133e-06, + "loss": 1.7767, + "step": 341 + }, + { + "epoch": 0.35130970724191063, + "grad_norm": 2.6293981075286865, + "learning_rate": 4.933384088313666e-06, + "loss": 1.7815, + "step": 342 + }, + { + "epoch": 0.35233692860811505, + "grad_norm": 3.230978012084961, + "learning_rate": 4.9314807765512e-06, + "loss": 1.7423, + "step": 343 + }, + { + "epoch": 0.3533641499743195, + "grad_norm": 3.228300094604492, + "learning_rate": 4.929577464788733e-06, + "loss": 1.6845, + "step": 344 + }, + { + "epoch": 0.3543913713405239, + "grad_norm": 3.036733388900757, + "learning_rate": 4.9276741530262665e-06, + "loss": 1.7524, + "step": 345 + }, + { + "epoch": 0.3554185927067283, + "grad_norm": 2.778538227081299, + "learning_rate": 4.9257708412638e-06, + "loss": 1.7433, + "step": 346 + }, + { + "epoch": 0.35644581407293274, + "grad_norm": 3.0958566665649414, + "learning_rate": 4.923867529501332e-06, + "loss": 1.76, + "step": 347 + }, + { + "epoch": 0.3574730354391371, + "grad_norm": 3.1120026111602783, + "learning_rate": 4.921964217738866e-06, + "loss": 1.7399, + "step": 348 + }, + { + "epoch": 0.35850025680534153, + "grad_norm": 2.70768141746521, + "learning_rate": 4.920060905976399e-06, + "loss": 1.8169, + "step": 349 + }, + { + "epoch": 0.35952747817154596, + "grad_norm": 2.917478561401367, + "learning_rate": 4.9181575942139325e-06, + "loss": 1.7948, + "step": 350 + }, + { + "epoch": 0.35952747817154596, + "eval_loss": 2.0208380222320557, + "eval_runtime": 47.4243, + "eval_samples_per_second": 21.086, + "eval_steps_per_second": 0.886, + "step": 350 + }, + { + "epoch": 0.3605546995377504, + "grad_norm": 3.0564756393432617, + "learning_rate": 4.916254282451466e-06, + "loss": 1.7861, + "step": 351 + }, + { + "epoch": 0.3615819209039548, + "grad_norm": 2.709195137023926, + "learning_rate": 4.914350970688999e-06, + "loss": 1.8323, + "step": 352 + }, + { + "epoch": 0.3626091422701592, + "grad_norm": 2.8797669410705566, + "learning_rate": 4.9124476589265325e-06, + "loss": 1.722, + "step": 353 + }, + { + "epoch": 0.36363636363636365, + "grad_norm": 2.560702085494995, + "learning_rate": 4.910544347164066e-06, + "loss": 1.7289, + "step": 354 + }, + { + "epoch": 0.36466358500256807, + "grad_norm": 3.3781847953796387, + "learning_rate": 4.908641035401599e-06, + "loss": 1.8096, + "step": 355 + }, + { + "epoch": 0.3656908063687725, + "grad_norm": 2.582310676574707, + "learning_rate": 4.906737723639133e-06, + "loss": 1.8752, + "step": 356 + }, + { + "epoch": 0.3667180277349769, + "grad_norm": 3.506385564804077, + "learning_rate": 4.904834411876665e-06, + "loss": 1.7404, + "step": 357 + }, + { + "epoch": 0.3677452491011813, + "grad_norm": 2.4360411167144775, + "learning_rate": 4.902931100114199e-06, + "loss": 1.7812, + "step": 358 + }, + { + "epoch": 0.3687724704673857, + "grad_norm": 3.77639102935791, + "learning_rate": 4.901027788351733e-06, + "loss": 1.7103, + "step": 359 + }, + { + "epoch": 0.3697996918335901, + "grad_norm": 2.887723922729492, + "learning_rate": 4.899124476589266e-06, + "loss": 1.7485, + "step": 360 + }, + { + "epoch": 0.3697996918335901, + "eval_loss": 2.0298168659210205, + "eval_runtime": 46.0993, + "eval_samples_per_second": 21.692, + "eval_steps_per_second": 0.911, + "step": 360 + }, + { + "epoch": 0.37082691319979455, + "grad_norm": 3.5861997604370117, + "learning_rate": 4.8972211648267994e-06, + "loss": 1.7189, + "step": 361 + }, + { + "epoch": 0.371854134565999, + "grad_norm": 2.590808868408203, + "learning_rate": 4.895317853064333e-06, + "loss": 1.7683, + "step": 362 + }, + { + "epoch": 0.3728813559322034, + "grad_norm": 3.4451487064361572, + "learning_rate": 4.893414541301866e-06, + "loss": 1.7436, + "step": 363 + }, + { + "epoch": 0.3739085772984078, + "grad_norm": 2.406073808670044, + "learning_rate": 4.891511229539399e-06, + "loss": 1.7378, + "step": 364 + }, + { + "epoch": 0.37493579866461224, + "grad_norm": 2.7162997722625732, + "learning_rate": 4.889607917776932e-06, + "loss": 1.7558, + "step": 365 + }, + { + "epoch": 0.37596302003081666, + "grad_norm": 2.564753770828247, + "learning_rate": 4.887704606014465e-06, + "loss": 1.7559, + "step": 366 + }, + { + "epoch": 0.37699024139702103, + "grad_norm": 2.7196900844573975, + "learning_rate": 4.885801294251999e-06, + "loss": 1.7683, + "step": 367 + }, + { + "epoch": 0.37801746276322545, + "grad_norm": 2.9334428310394287, + "learning_rate": 4.883897982489532e-06, + "loss": 1.6932, + "step": 368 + }, + { + "epoch": 0.3790446841294299, + "grad_norm": 2.4899954795837402, + "learning_rate": 4.8819946707270655e-06, + "loss": 1.7943, + "step": 369 + }, + { + "epoch": 0.3800719054956343, + "grad_norm": 2.820594310760498, + "learning_rate": 4.880091358964599e-06, + "loss": 1.7729, + "step": 370 + }, + { + "epoch": 0.3800719054956343, + "eval_loss": 2.0204291343688965, + "eval_runtime": 44.9993, + "eval_samples_per_second": 22.223, + "eval_steps_per_second": 0.933, + "step": 370 + }, + { + "epoch": 0.3810991268618387, + "grad_norm": 2.4717929363250732, + "learning_rate": 4.878188047202132e-06, + "loss": 1.6813, + "step": 371 + }, + { + "epoch": 0.38212634822804314, + "grad_norm": 3.1815855503082275, + "learning_rate": 4.876284735439666e-06, + "loss": 1.8588, + "step": 372 + }, + { + "epoch": 0.38315356959424757, + "grad_norm": 2.7330422401428223, + "learning_rate": 4.874381423677199e-06, + "loss": 1.8245, + "step": 373 + }, + { + "epoch": 0.384180790960452, + "grad_norm": 3.129444122314453, + "learning_rate": 4.8724781119147315e-06, + "loss": 1.7494, + "step": 374 + }, + { + "epoch": 0.3852080123266564, + "grad_norm": 2.75551700592041, + "learning_rate": 4.870574800152265e-06, + "loss": 1.7761, + "step": 375 + }, + { + "epoch": 0.38623523369286084, + "grad_norm": 2.9313466548919678, + "learning_rate": 4.868671488389798e-06, + "loss": 1.7076, + "step": 376 + }, + { + "epoch": 0.3872624550590652, + "grad_norm": 2.7590556144714355, + "learning_rate": 4.8667681766273316e-06, + "loss": 1.7105, + "step": 377 + }, + { + "epoch": 0.3882896764252696, + "grad_norm": 2.8224680423736572, + "learning_rate": 4.864864864864866e-06, + "loss": 1.6853, + "step": 378 + }, + { + "epoch": 0.38931689779147405, + "grad_norm": 2.5148062705993652, + "learning_rate": 4.862961553102399e-06, + "loss": 1.7477, + "step": 379 + }, + { + "epoch": 0.39034411915767847, + "grad_norm": 3.144247055053711, + "learning_rate": 4.8610582413399325e-06, + "loss": 1.7726, + "step": 380 + }, + { + "epoch": 0.39034411915767847, + "eval_loss": 2.006993532180786, + "eval_runtime": 50.8909, + "eval_samples_per_second": 19.65, + "eval_steps_per_second": 0.825, + "step": 380 + }, + { + "epoch": 0.3913713405238829, + "grad_norm": 2.6818125247955322, + "learning_rate": 4.859154929577465e-06, + "loss": 1.7351, + "step": 381 + }, + { + "epoch": 0.3923985618900873, + "grad_norm": 2.8543903827667236, + "learning_rate": 4.857251617814998e-06, + "loss": 1.7499, + "step": 382 + }, + { + "epoch": 0.39342578325629174, + "grad_norm": 2.5483179092407227, + "learning_rate": 4.855348306052532e-06, + "loss": 1.7849, + "step": 383 + }, + { + "epoch": 0.39445300462249616, + "grad_norm": 2.7169013023376465, + "learning_rate": 4.853444994290065e-06, + "loss": 1.7664, + "step": 384 + }, + { + "epoch": 0.3954802259887006, + "grad_norm": 2.8304696083068848, + "learning_rate": 4.8515416825275985e-06, + "loss": 1.73, + "step": 385 + }, + { + "epoch": 0.396507447354905, + "grad_norm": 2.4217448234558105, + "learning_rate": 4.849638370765132e-06, + "loss": 1.7915, + "step": 386 + }, + { + "epoch": 0.3975346687211094, + "grad_norm": 2.523256301879883, + "learning_rate": 4.847735059002665e-06, + "loss": 1.7195, + "step": 387 + }, + { + "epoch": 0.3985618900873138, + "grad_norm": 2.813638925552368, + "learning_rate": 4.8458317472401985e-06, + "loss": 1.6766, + "step": 388 + }, + { + "epoch": 0.3995891114535182, + "grad_norm": 2.6701548099517822, + "learning_rate": 4.843928435477732e-06, + "loss": 1.7666, + "step": 389 + }, + { + "epoch": 0.40061633281972264, + "grad_norm": 2.5568342208862305, + "learning_rate": 4.842025123715265e-06, + "loss": 1.723, + "step": 390 + }, + { + "epoch": 0.40061633281972264, + "eval_loss": 2.0098180770874023, + "eval_runtime": 44.4503, + "eval_samples_per_second": 22.497, + "eval_steps_per_second": 0.945, + "step": 390 + }, + { + "epoch": 0.40164355418592707, + "grad_norm": 2.6287341117858887, + "learning_rate": 4.840121811952798e-06, + "loss": 1.759, + "step": 391 + }, + { + "epoch": 0.4026707755521315, + "grad_norm": 2.934105157852173, + "learning_rate": 4.838218500190331e-06, + "loss": 1.7501, + "step": 392 + }, + { + "epoch": 0.4036979969183359, + "grad_norm": 2.6563878059387207, + "learning_rate": 4.8363151884278645e-06, + "loss": 1.7624, + "step": 393 + }, + { + "epoch": 0.40472521828454033, + "grad_norm": 2.538278102874756, + "learning_rate": 4.834411876665398e-06, + "loss": 1.7457, + "step": 394 + }, + { + "epoch": 0.40575243965074476, + "grad_norm": 2.801630735397339, + "learning_rate": 4.832508564902931e-06, + "loss": 1.7549, + "step": 395 + }, + { + "epoch": 0.4067796610169492, + "grad_norm": 2.571937322616577, + "learning_rate": 4.830605253140465e-06, + "loss": 1.6954, + "step": 396 + }, + { + "epoch": 0.40780688238315355, + "grad_norm": 2.5652592182159424, + "learning_rate": 4.828701941377998e-06, + "loss": 1.7328, + "step": 397 + }, + { + "epoch": 0.40883410374935797, + "grad_norm": 2.6840121746063232, + "learning_rate": 4.826798629615531e-06, + "loss": 1.6934, + "step": 398 + }, + { + "epoch": 0.4098613251155624, + "grad_norm": 3.0103495121002197, + "learning_rate": 4.824895317853065e-06, + "loss": 1.7086, + "step": 399 + }, + { + "epoch": 0.4108885464817668, + "grad_norm": 2.7361223697662354, + "learning_rate": 4.822992006090598e-06, + "loss": 1.7301, + "step": 400 + }, + { + "epoch": 0.4108885464817668, + "eval_loss": 1.9975080490112305, + "eval_runtime": 44.5413, + "eval_samples_per_second": 22.451, + "eval_steps_per_second": 0.943, + "step": 400 + }, + { + "epoch": 0.41191576784797124, + "grad_norm": 2.6805901527404785, + "learning_rate": 4.821088694328131e-06, + "loss": 1.6896, + "step": 401 + }, + { + "epoch": 0.41294298921417566, + "grad_norm": 3.231706142425537, + "learning_rate": 4.819185382565665e-06, + "loss": 1.7134, + "step": 402 + }, + { + "epoch": 0.4139702105803801, + "grad_norm": 2.5561299324035645, + "learning_rate": 4.817282070803198e-06, + "loss": 1.764, + "step": 403 + }, + { + "epoch": 0.4149974319465845, + "grad_norm": 3.3125340938568115, + "learning_rate": 4.8153787590407315e-06, + "loss": 1.76, + "step": 404 + }, + { + "epoch": 0.41602465331278893, + "grad_norm": 2.750051498413086, + "learning_rate": 4.813475447278265e-06, + "loss": 1.6882, + "step": 405 + }, + { + "epoch": 0.41705187467899335, + "grad_norm": 4.006515979766846, + "learning_rate": 4.811572135515798e-06, + "loss": 1.7892, + "step": 406 + }, + { + "epoch": 0.4180790960451977, + "grad_norm": 2.5276904106140137, + "learning_rate": 4.809668823753332e-06, + "loss": 1.6995, + "step": 407 + }, + { + "epoch": 0.41910631741140214, + "grad_norm": 2.959444046020508, + "learning_rate": 4.807765511990864e-06, + "loss": 1.7058, + "step": 408 + }, + { + "epoch": 0.42013353877760656, + "grad_norm": 2.8472464084625244, + "learning_rate": 4.8058622002283975e-06, + "loss": 1.7258, + "step": 409 + }, + { + "epoch": 0.421160760143811, + "grad_norm": 3.0240273475646973, + "learning_rate": 4.803958888465931e-06, + "loss": 1.7789, + "step": 410 + }, + { + "epoch": 0.421160760143811, + "eval_loss": 1.9919871091842651, + "eval_runtime": 41.727, + "eval_samples_per_second": 23.965, + "eval_steps_per_second": 1.007, + "step": 410 + }, + { + "epoch": 0.4221879815100154, + "grad_norm": 2.6211893558502197, + "learning_rate": 4.802055576703464e-06, + "loss": 1.7315, + "step": 411 + }, + { + "epoch": 0.42321520287621983, + "grad_norm": 2.6072895526885986, + "learning_rate": 4.8001522649409976e-06, + "loss": 1.8176, + "step": 412 + }, + { + "epoch": 0.42424242424242425, + "grad_norm": 2.9197702407836914, + "learning_rate": 4.798248953178531e-06, + "loss": 1.6947, + "step": 413 + }, + { + "epoch": 0.4252696456086287, + "grad_norm": 2.4383556842803955, + "learning_rate": 4.796345641416064e-06, + "loss": 1.6728, + "step": 414 + }, + { + "epoch": 0.4262968669748331, + "grad_norm": 2.9951915740966797, + "learning_rate": 4.794442329653598e-06, + "loss": 1.7249, + "step": 415 + }, + { + "epoch": 0.42732408834103747, + "grad_norm": 2.6095519065856934, + "learning_rate": 4.792539017891131e-06, + "loss": 1.7735, + "step": 416 + }, + { + "epoch": 0.4283513097072419, + "grad_norm": 2.6429390907287598, + "learning_rate": 4.790635706128664e-06, + "loss": 1.704, + "step": 417 + }, + { + "epoch": 0.4293785310734463, + "grad_norm": 2.921812057495117, + "learning_rate": 4.788732394366197e-06, + "loss": 1.7355, + "step": 418 + }, + { + "epoch": 0.43040575243965074, + "grad_norm": 2.6050965785980225, + "learning_rate": 4.786829082603731e-06, + "loss": 1.6636, + "step": 419 + }, + { + "epoch": 0.43143297380585516, + "grad_norm": 2.672070264816284, + "learning_rate": 4.7849257708412645e-06, + "loss": 1.6627, + "step": 420 + }, + { + "epoch": 0.43143297380585516, + "eval_loss": 1.9990200996398926, + "eval_runtime": 45.0672, + "eval_samples_per_second": 22.189, + "eval_steps_per_second": 0.932, + "step": 420 + }, + { + "epoch": 0.4324601951720596, + "grad_norm": 2.6813488006591797, + "learning_rate": 4.783022459078798e-06, + "loss": 1.7453, + "step": 421 + }, + { + "epoch": 0.433487416538264, + "grad_norm": 2.6587321758270264, + "learning_rate": 4.781119147316331e-06, + "loss": 1.6795, + "step": 422 + }, + { + "epoch": 0.4345146379044684, + "grad_norm": 2.827758312225342, + "learning_rate": 4.7792158355538645e-06, + "loss": 1.7098, + "step": 423 + }, + { + "epoch": 0.43554185927067285, + "grad_norm": 2.8615901470184326, + "learning_rate": 4.777312523791398e-06, + "loss": 1.7171, + "step": 424 + }, + { + "epoch": 0.43656908063687727, + "grad_norm": 2.9509294033050537, + "learning_rate": 4.77540921202893e-06, + "loss": 1.7018, + "step": 425 + }, + { + "epoch": 0.43759630200308164, + "grad_norm": 2.7994418144226074, + "learning_rate": 4.773505900266464e-06, + "loss": 1.7328, + "step": 426 + }, + { + "epoch": 0.43862352336928606, + "grad_norm": 2.7476799488067627, + "learning_rate": 4.771602588503997e-06, + "loss": 1.7116, + "step": 427 + }, + { + "epoch": 0.4396507447354905, + "grad_norm": 2.7955048084259033, + "learning_rate": 4.7696992767415305e-06, + "loss": 1.7187, + "step": 428 + }, + { + "epoch": 0.4406779661016949, + "grad_norm": 2.4326071739196777, + "learning_rate": 4.767795964979064e-06, + "loss": 1.75, + "step": 429 + }, + { + "epoch": 0.44170518746789933, + "grad_norm": 2.7708163261413574, + "learning_rate": 4.765892653216597e-06, + "loss": 1.7562, + "step": 430 + }, + { + "epoch": 0.44170518746789933, + "eval_loss": 2.0262997150421143, + "eval_runtime": 46.1649, + "eval_samples_per_second": 21.661, + "eval_steps_per_second": 0.91, + "step": 430 + }, + { + "epoch": 0.44273240883410375, + "grad_norm": 2.872743606567383, + "learning_rate": 4.763989341454131e-06, + "loss": 1.7527, + "step": 431 + }, + { + "epoch": 0.4437596302003082, + "grad_norm": 2.6387035846710205, + "learning_rate": 4.762086029691664e-06, + "loss": 1.6742, + "step": 432 + }, + { + "epoch": 0.4447868515665126, + "grad_norm": 2.6522979736328125, + "learning_rate": 4.760182717929197e-06, + "loss": 1.6811, + "step": 433 + }, + { + "epoch": 0.445814072932717, + "grad_norm": 2.644155263900757, + "learning_rate": 4.758279406166731e-06, + "loss": 1.7083, + "step": 434 + }, + { + "epoch": 0.44684129429892144, + "grad_norm": 2.530224561691284, + "learning_rate": 4.756376094404263e-06, + "loss": 1.6318, + "step": 435 + }, + { + "epoch": 0.4478685156651258, + "grad_norm": 2.431049108505249, + "learning_rate": 4.7544727826417966e-06, + "loss": 1.7162, + "step": 436 + }, + { + "epoch": 0.44889573703133023, + "grad_norm": 2.8091282844543457, + "learning_rate": 4.75256947087933e-06, + "loss": 1.7717, + "step": 437 + }, + { + "epoch": 0.44992295839753466, + "grad_norm": 2.5520501136779785, + "learning_rate": 4.750666159116863e-06, + "loss": 1.7201, + "step": 438 + }, + { + "epoch": 0.4509501797637391, + "grad_norm": 2.596370220184326, + "learning_rate": 4.7487628473543975e-06, + "loss": 1.6882, + "step": 439 + }, + { + "epoch": 0.4519774011299435, + "grad_norm": 2.667034387588501, + "learning_rate": 4.746859535591931e-06, + "loss": 1.6499, + "step": 440 + }, + { + "epoch": 0.4519774011299435, + "eval_loss": 2.008521556854248, + "eval_runtime": 49.3626, + "eval_samples_per_second": 20.258, + "eval_steps_per_second": 0.851, + "step": 440 + }, + { + "epoch": 0.4530046224961479, + "grad_norm": 2.636491537094116, + "learning_rate": 4.744956223829464e-06, + "loss": 1.6987, + "step": 441 + }, + { + "epoch": 0.45403184386235235, + "grad_norm": 2.4653050899505615, + "learning_rate": 4.743052912066997e-06, + "loss": 1.6775, + "step": 442 + }, + { + "epoch": 0.45505906522855677, + "grad_norm": 2.3866467475891113, + "learning_rate": 4.74114960030453e-06, + "loss": 1.7234, + "step": 443 + }, + { + "epoch": 0.4560862865947612, + "grad_norm": 2.675262928009033, + "learning_rate": 4.7392462885420635e-06, + "loss": 1.7363, + "step": 444 + }, + { + "epoch": 0.4571135079609656, + "grad_norm": 2.6233296394348145, + "learning_rate": 4.737342976779597e-06, + "loss": 1.6236, + "step": 445 + }, + { + "epoch": 0.45814072932717, + "grad_norm": 2.5638058185577393, + "learning_rate": 4.73543966501713e-06, + "loss": 1.7097, + "step": 446 + }, + { + "epoch": 0.4591679506933744, + "grad_norm": 2.7206926345825195, + "learning_rate": 4.7335363532546636e-06, + "loss": 1.6909, + "step": 447 + }, + { + "epoch": 0.4601951720595788, + "grad_norm": 2.546241283416748, + "learning_rate": 4.731633041492197e-06, + "loss": 1.6951, + "step": 448 + }, + { + "epoch": 0.46122239342578325, + "grad_norm": 2.6549105644226074, + "learning_rate": 4.72972972972973e-06, + "loss": 1.7052, + "step": 449 + }, + { + "epoch": 0.4622496147919877, + "grad_norm": 2.6649065017700195, + "learning_rate": 4.727826417967264e-06, + "loss": 1.7406, + "step": 450 + }, + { + "epoch": 0.4622496147919877, + "eval_loss": 1.9886436462402344, + "eval_runtime": 47.48, + "eval_samples_per_second": 21.061, + "eval_steps_per_second": 0.885, + "step": 450 + }, + { + "epoch": 0.4632768361581921, + "grad_norm": 2.7138187885284424, + "learning_rate": 4.725923106204797e-06, + "loss": 1.7154, + "step": 451 + }, + { + "epoch": 0.4643040575243965, + "grad_norm": 2.6095476150512695, + "learning_rate": 4.72401979444233e-06, + "loss": 1.686, + "step": 452 + }, + { + "epoch": 0.46533127889060094, + "grad_norm": 2.866363048553467, + "learning_rate": 4.722116482679863e-06, + "loss": 1.721, + "step": 453 + }, + { + "epoch": 0.46635850025680536, + "grad_norm": 2.4947352409362793, + "learning_rate": 4.720213170917396e-06, + "loss": 1.708, + "step": 454 + }, + { + "epoch": 0.46738572162300973, + "grad_norm": 2.3888256549835205, + "learning_rate": 4.71830985915493e-06, + "loss": 1.7356, + "step": 455 + }, + { + "epoch": 0.46841294298921415, + "grad_norm": 2.7065343856811523, + "learning_rate": 4.716406547392463e-06, + "loss": 1.6943, + "step": 456 + }, + { + "epoch": 0.4694401643554186, + "grad_norm": 2.424285888671875, + "learning_rate": 4.714503235629996e-06, + "loss": 1.7385, + "step": 457 + }, + { + "epoch": 0.470467385721623, + "grad_norm": 2.4175925254821777, + "learning_rate": 4.71259992386753e-06, + "loss": 1.6654, + "step": 458 + }, + { + "epoch": 0.4714946070878274, + "grad_norm": 2.3203656673431396, + "learning_rate": 4.710696612105063e-06, + "loss": 1.6977, + "step": 459 + }, + { + "epoch": 0.47252182845403184, + "grad_norm": 2.6660172939300537, + "learning_rate": 4.708793300342596e-06, + "loss": 1.7055, + "step": 460 + }, + { + "epoch": 0.47252182845403184, + "eval_loss": 1.9864659309387207, + "eval_runtime": 47.6232, + "eval_samples_per_second": 20.998, + "eval_steps_per_second": 0.882, + "step": 460 + }, + { + "epoch": 0.47354904982023627, + "grad_norm": 2.515615701675415, + "learning_rate": 4.70688998858013e-06, + "loss": 1.689, + "step": 461 + }, + { + "epoch": 0.4745762711864407, + "grad_norm": 2.429264783859253, + "learning_rate": 4.704986676817663e-06, + "loss": 1.6796, + "step": 462 + }, + { + "epoch": 0.4756034925526451, + "grad_norm": 2.469543933868408, + "learning_rate": 4.7030833650551965e-06, + "loss": 1.7164, + "step": 463 + }, + { + "epoch": 0.47663071391884954, + "grad_norm": 2.5921683311462402, + "learning_rate": 4.70118005329273e-06, + "loss": 1.6746, + "step": 464 + }, + { + "epoch": 0.4776579352850539, + "grad_norm": 2.6612918376922607, + "learning_rate": 4.699276741530263e-06, + "loss": 1.7598, + "step": 465 + }, + { + "epoch": 0.4786851566512583, + "grad_norm": 2.4633305072784424, + "learning_rate": 4.697373429767797e-06, + "loss": 1.7241, + "step": 466 + }, + { + "epoch": 0.47971237801746275, + "grad_norm": 2.613023281097412, + "learning_rate": 4.69547011800533e-06, + "loss": 1.7088, + "step": 467 + }, + { + "epoch": 0.48073959938366717, + "grad_norm": 2.5329010486602783, + "learning_rate": 4.693566806242863e-06, + "loss": 1.7163, + "step": 468 + }, + { + "epoch": 0.4817668207498716, + "grad_norm": 2.7887041568756104, + "learning_rate": 4.691663494480397e-06, + "loss": 1.6969, + "step": 469 + }, + { + "epoch": 0.482794042116076, + "grad_norm": 2.72991943359375, + "learning_rate": 4.689760182717929e-06, + "loss": 1.7332, + "step": 470 + }, + { + "epoch": 0.482794042116076, + "eval_loss": 1.965472936630249, + "eval_runtime": 46.1059, + "eval_samples_per_second": 21.689, + "eval_steps_per_second": 0.911, + "step": 470 + }, + { + "epoch": 0.48382126348228044, + "grad_norm": 2.613785982131958, + "learning_rate": 4.6878568709554626e-06, + "loss": 1.6272, + "step": 471 + }, + { + "epoch": 0.48484848484848486, + "grad_norm": 2.605299472808838, + "learning_rate": 4.685953559192996e-06, + "loss": 1.6966, + "step": 472 + }, + { + "epoch": 0.4858757062146893, + "grad_norm": 2.672013998031616, + "learning_rate": 4.684050247430529e-06, + "loss": 1.7132, + "step": 473 + }, + { + "epoch": 0.4869029275808937, + "grad_norm": 2.576486825942993, + "learning_rate": 4.682146935668063e-06, + "loss": 1.6449, + "step": 474 + }, + { + "epoch": 0.4879301489470981, + "grad_norm": 2.626840114593506, + "learning_rate": 4.680243623905596e-06, + "loss": 1.6993, + "step": 475 + }, + { + "epoch": 0.4889573703133025, + "grad_norm": 2.974418878555298, + "learning_rate": 4.678340312143129e-06, + "loss": 1.7116, + "step": 476 + }, + { + "epoch": 0.4899845916795069, + "grad_norm": 2.541242837905884, + "learning_rate": 4.676437000380663e-06, + "loss": 1.6693, + "step": 477 + }, + { + "epoch": 0.49101181304571134, + "grad_norm": 2.5494985580444336, + "learning_rate": 4.674533688618196e-06, + "loss": 1.7819, + "step": 478 + }, + { + "epoch": 0.49203903441191577, + "grad_norm": 2.606137990951538, + "learning_rate": 4.6726303768557295e-06, + "loss": 1.7278, + "step": 479 + }, + { + "epoch": 0.4930662557781202, + "grad_norm": 2.5032176971435547, + "learning_rate": 4.670727065093263e-06, + "loss": 1.6318, + "step": 480 + }, + { + "epoch": 0.4930662557781202, + "eval_loss": 1.9851018190383911, + "eval_runtime": 46.0386, + "eval_samples_per_second": 21.721, + "eval_steps_per_second": 0.912, + "step": 480 + }, + { + "epoch": 0.4940934771443246, + "grad_norm": 2.703259229660034, + "learning_rate": 4.668823753330796e-06, + "loss": 1.6646, + "step": 481 + }, + { + "epoch": 0.49512069851052903, + "grad_norm": 2.681807041168213, + "learning_rate": 4.6669204415683295e-06, + "loss": 1.6759, + "step": 482 + }, + { + "epoch": 0.49614791987673346, + "grad_norm": 2.7280473709106445, + "learning_rate": 4.665017129805863e-06, + "loss": 1.7388, + "step": 483 + }, + { + "epoch": 0.4971751412429379, + "grad_norm": 2.420909881591797, + "learning_rate": 4.663113818043396e-06, + "loss": 1.729, + "step": 484 + }, + { + "epoch": 0.49820236260914225, + "grad_norm": 2.5764737129211426, + "learning_rate": 4.66121050628093e-06, + "loss": 1.6903, + "step": 485 + }, + { + "epoch": 0.49922958397534667, + "grad_norm": 2.389531135559082, + "learning_rate": 4.659307194518463e-06, + "loss": 1.7191, + "step": 486 + }, + { + "epoch": 0.5002568053415511, + "grad_norm": 2.5207972526550293, + "learning_rate": 4.6574038827559955e-06, + "loss": 1.6892, + "step": 487 + }, + { + "epoch": 0.5012840267077555, + "grad_norm": 2.5662360191345215, + "learning_rate": 4.655500570993529e-06, + "loss": 1.6978, + "step": 488 + }, + { + "epoch": 0.50231124807396, + "grad_norm": 2.50262713432312, + "learning_rate": 4.653597259231062e-06, + "loss": 1.7379, + "step": 489 + }, + { + "epoch": 0.5033384694401644, + "grad_norm": 2.56643009185791, + "learning_rate": 4.651693947468596e-06, + "loss": 1.6233, + "step": 490 + }, + { + "epoch": 0.5033384694401644, + "eval_loss": 1.9815036058425903, + "eval_runtime": 48.0257, + "eval_samples_per_second": 20.822, + "eval_steps_per_second": 0.875, + "step": 490 + }, + { + "epoch": 0.5043656908063687, + "grad_norm": 2.504117965698242, + "learning_rate": 4.649790635706129e-06, + "loss": 1.7664, + "step": 491 + }, + { + "epoch": 0.5053929121725732, + "grad_norm": 2.906843423843384, + "learning_rate": 4.647887323943662e-06, + "loss": 1.6731, + "step": 492 + }, + { + "epoch": 0.5064201335387776, + "grad_norm": 2.586747169494629, + "learning_rate": 4.645984012181196e-06, + "loss": 1.68, + "step": 493 + }, + { + "epoch": 0.507447354904982, + "grad_norm": 2.3300511837005615, + "learning_rate": 4.644080700418729e-06, + "loss": 1.5998, + "step": 494 + }, + { + "epoch": 0.5084745762711864, + "grad_norm": 2.8711049556732178, + "learning_rate": 4.642177388656262e-06, + "loss": 1.7198, + "step": 495 + }, + { + "epoch": 0.5095017976373909, + "grad_norm": 2.6463162899017334, + "learning_rate": 4.640274076893796e-06, + "loss": 1.6814, + "step": 496 + }, + { + "epoch": 0.5105290190035953, + "grad_norm": 2.650481700897217, + "learning_rate": 4.638370765131328e-06, + "loss": 1.7353, + "step": 497 + }, + { + "epoch": 0.5115562403697997, + "grad_norm": 2.4464986324310303, + "learning_rate": 4.636467453368862e-06, + "loss": 1.6579, + "step": 498 + }, + { + "epoch": 0.5125834617360041, + "grad_norm": 2.438021659851074, + "learning_rate": 4.634564141606396e-06, + "loss": 1.6485, + "step": 499 + }, + { + "epoch": 0.5136106831022085, + "grad_norm": 2.5354347229003906, + "learning_rate": 4.632660829843929e-06, + "loss": 1.7123, + "step": 500 + }, + { + "epoch": 0.5136106831022085, + "eval_loss": 1.9738385677337646, + "eval_runtime": 52.5306, + "eval_samples_per_second": 19.037, + "eval_steps_per_second": 0.8, + "step": 500 + }, + { + "epoch": 0.514637904468413, + "grad_norm": 2.3702070713043213, + "learning_rate": 4.630757518081463e-06, + "loss": 1.6423, + "step": 501 + }, + { + "epoch": 0.5156651258346173, + "grad_norm": 2.5336198806762695, + "learning_rate": 4.628854206318996e-06, + "loss": 1.771, + "step": 502 + }, + { + "epoch": 0.5166923472008218, + "grad_norm": 2.572751998901367, + "learning_rate": 4.626950894556529e-06, + "loss": 1.677, + "step": 503 + }, + { + "epoch": 0.5177195685670262, + "grad_norm": 2.3865134716033936, + "learning_rate": 4.625047582794062e-06, + "loss": 1.7111, + "step": 504 + }, + { + "epoch": 0.5187467899332306, + "grad_norm": 2.5152785778045654, + "learning_rate": 4.623144271031595e-06, + "loss": 1.6343, + "step": 505 + }, + { + "epoch": 0.519774011299435, + "grad_norm": 2.5121307373046875, + "learning_rate": 4.6212409592691286e-06, + "loss": 1.6357, + "step": 506 + }, + { + "epoch": 0.5208012326656395, + "grad_norm": 2.506908655166626, + "learning_rate": 4.619337647506662e-06, + "loss": 1.7295, + "step": 507 + }, + { + "epoch": 0.5218284540318439, + "grad_norm": 2.636082172393799, + "learning_rate": 4.617434335744195e-06, + "loss": 1.6916, + "step": 508 + }, + { + "epoch": 0.5228556753980482, + "grad_norm": 2.5987415313720703, + "learning_rate": 4.615531023981729e-06, + "loss": 1.659, + "step": 509 + }, + { + "epoch": 0.5238828967642527, + "grad_norm": 2.642770290374756, + "learning_rate": 4.613627712219262e-06, + "loss": 1.6789, + "step": 510 + }, + { + "epoch": 0.5238828967642527, + "eval_loss": 1.9659523963928223, + "eval_runtime": 42.9453, + "eval_samples_per_second": 23.285, + "eval_steps_per_second": 0.978, + "step": 510 + }, + { + "epoch": 0.5249101181304571, + "grad_norm": 2.438389301300049, + "learning_rate": 4.611724400456795e-06, + "loss": 1.68, + "step": 511 + }, + { + "epoch": 0.5259373394966615, + "grad_norm": 2.536546468734741, + "learning_rate": 4.609821088694329e-06, + "loss": 1.6847, + "step": 512 + }, + { + "epoch": 0.5269645608628659, + "grad_norm": 2.4056830406188965, + "learning_rate": 4.607917776931862e-06, + "loss": 1.6577, + "step": 513 + }, + { + "epoch": 0.5279917822290704, + "grad_norm": 2.81195330619812, + "learning_rate": 4.606014465169395e-06, + "loss": 1.7149, + "step": 514 + }, + { + "epoch": 0.5290190035952748, + "grad_norm": 2.684948205947876, + "learning_rate": 4.604111153406928e-06, + "loss": 1.717, + "step": 515 + }, + { + "epoch": 0.5300462249614792, + "grad_norm": 2.583092212677002, + "learning_rate": 4.602207841644461e-06, + "loss": 1.6238, + "step": 516 + }, + { + "epoch": 0.5310734463276836, + "grad_norm": 3.022528886795044, + "learning_rate": 4.600304529881995e-06, + "loss": 1.7262, + "step": 517 + }, + { + "epoch": 0.5321006676938881, + "grad_norm": 2.6941866874694824, + "learning_rate": 4.598401218119528e-06, + "loss": 1.7055, + "step": 518 + }, + { + "epoch": 0.5331278890600925, + "grad_norm": 2.5207839012145996, + "learning_rate": 4.5964979063570614e-06, + "loss": 1.7117, + "step": 519 + }, + { + "epoch": 0.5341551104262968, + "grad_norm": 2.8056678771972656, + "learning_rate": 4.594594594594596e-06, + "loss": 1.657, + "step": 520 + }, + { + "epoch": 0.5341551104262968, + "eval_loss": 1.9712116718292236, + "eval_runtime": 49.57, + "eval_samples_per_second": 20.173, + "eval_steps_per_second": 0.847, + "step": 520 + }, + { + "epoch": 0.5351823317925013, + "grad_norm": 2.6119930744171143, + "learning_rate": 4.592691282832128e-06, + "loss": 1.6897, + "step": 521 + }, + { + "epoch": 0.5362095531587057, + "grad_norm": 2.7165610790252686, + "learning_rate": 4.5907879710696615e-06, + "loss": 1.7038, + "step": 522 + }, + { + "epoch": 0.5372367745249101, + "grad_norm": 2.3613736629486084, + "learning_rate": 4.588884659307195e-06, + "loss": 1.6222, + "step": 523 + }, + { + "epoch": 0.5382639958911145, + "grad_norm": 2.647672176361084, + "learning_rate": 4.586981347544728e-06, + "loss": 1.702, + "step": 524 + }, + { + "epoch": 0.539291217257319, + "grad_norm": 2.704634428024292, + "learning_rate": 4.585078035782262e-06, + "loss": 1.6566, + "step": 525 + }, + { + "epoch": 0.5403184386235234, + "grad_norm": 2.6634597778320312, + "learning_rate": 4.583174724019795e-06, + "loss": 1.6903, + "step": 526 + }, + { + "epoch": 0.5413456599897278, + "grad_norm": 2.5745162963867188, + "learning_rate": 4.581271412257328e-06, + "loss": 1.6375, + "step": 527 + }, + { + "epoch": 0.5423728813559322, + "grad_norm": 2.5577778816223145, + "learning_rate": 4.579368100494862e-06, + "loss": 1.614, + "step": 528 + }, + { + "epoch": 0.5434001027221366, + "grad_norm": 2.7684454917907715, + "learning_rate": 4.577464788732395e-06, + "loss": 1.7168, + "step": 529 + }, + { + "epoch": 0.544427324088341, + "grad_norm": 2.582108736038208, + "learning_rate": 4.575561476969928e-06, + "loss": 1.6897, + "step": 530 + }, + { + "epoch": 0.544427324088341, + "eval_loss": 1.9821091890335083, + "eval_runtime": 50.9409, + "eval_samples_per_second": 19.631, + "eval_steps_per_second": 0.824, + "step": 530 + }, + { + "epoch": 0.5454545454545454, + "grad_norm": 2.516486406326294, + "learning_rate": 4.573658165207461e-06, + "loss": 1.7273, + "step": 531 + }, + { + "epoch": 0.5464817668207499, + "grad_norm": 2.786806344985962, + "learning_rate": 4.571754853444994e-06, + "loss": 1.6898, + "step": 532 + }, + { + "epoch": 0.5475089881869543, + "grad_norm": 2.3137319087982178, + "learning_rate": 4.569851541682528e-06, + "loss": 1.676, + "step": 533 + }, + { + "epoch": 0.5485362095531587, + "grad_norm": 2.7268903255462646, + "learning_rate": 4.567948229920061e-06, + "loss": 1.6483, + "step": 534 + }, + { + "epoch": 0.5495634309193631, + "grad_norm": 2.4385902881622314, + "learning_rate": 4.566044918157594e-06, + "loss": 1.6172, + "step": 535 + }, + { + "epoch": 0.5505906522855676, + "grad_norm": 2.420576333999634, + "learning_rate": 4.564141606395128e-06, + "loss": 1.6569, + "step": 536 + }, + { + "epoch": 0.551617873651772, + "grad_norm": 2.592290163040161, + "learning_rate": 4.562238294632661e-06, + "loss": 1.7085, + "step": 537 + }, + { + "epoch": 0.5526450950179764, + "grad_norm": 2.485285520553589, + "learning_rate": 4.5603349828701945e-06, + "loss": 1.6699, + "step": 538 + }, + { + "epoch": 0.5536723163841808, + "grad_norm": 2.844710111618042, + "learning_rate": 4.558431671107728e-06, + "loss": 1.6873, + "step": 539 + }, + { + "epoch": 0.5546995377503852, + "grad_norm": 2.5279181003570557, + "learning_rate": 4.556528359345261e-06, + "loss": 1.7285, + "step": 540 + }, + { + "epoch": 0.5546995377503852, + "eval_loss": 1.9825981855392456, + "eval_runtime": 52.461, + "eval_samples_per_second": 19.062, + "eval_steps_per_second": 0.801, + "step": 540 + }, + { + "epoch": 0.5557267591165896, + "grad_norm": 2.673997163772583, + "learning_rate": 4.5546250475827946e-06, + "loss": 1.6658, + "step": 541 + }, + { + "epoch": 0.556753980482794, + "grad_norm": 2.6023194789886475, + "learning_rate": 4.552721735820328e-06, + "loss": 1.6262, + "step": 542 + }, + { + "epoch": 0.5577812018489985, + "grad_norm": 2.6327662467956543, + "learning_rate": 4.550818424057861e-06, + "loss": 1.6574, + "step": 543 + }, + { + "epoch": 0.5588084232152029, + "grad_norm": 2.5532140731811523, + "learning_rate": 4.548915112295395e-06, + "loss": 1.6607, + "step": 544 + }, + { + "epoch": 0.5598356445814073, + "grad_norm": 2.523966073989868, + "learning_rate": 4.547011800532928e-06, + "loss": 1.6752, + "step": 545 + }, + { + "epoch": 0.5608628659476117, + "grad_norm": 2.6338107585906982, + "learning_rate": 4.545108488770461e-06, + "loss": 1.6258, + "step": 546 + }, + { + "epoch": 0.5618900873138162, + "grad_norm": 2.500096559524536, + "learning_rate": 4.543205177007995e-06, + "loss": 1.7189, + "step": 547 + }, + { + "epoch": 0.5629173086800205, + "grad_norm": 2.644413471221924, + "learning_rate": 4.541301865245527e-06, + "loss": 1.6594, + "step": 548 + }, + { + "epoch": 0.5639445300462249, + "grad_norm": 2.451965093612671, + "learning_rate": 4.539398553483061e-06, + "loss": 1.6984, + "step": 549 + }, + { + "epoch": 0.5649717514124294, + "grad_norm": 2.7554423809051514, + "learning_rate": 4.537495241720594e-06, + "loss": 1.6626, + "step": 550 + }, + { + "epoch": 0.5649717514124294, + "eval_loss": 1.9897041320800781, + "eval_runtime": 45.9349, + "eval_samples_per_second": 21.77, + "eval_steps_per_second": 0.914, + "step": 550 + }, + { + "epoch": 0.5659989727786338, + "grad_norm": 2.5962674617767334, + "learning_rate": 4.535591929958127e-06, + "loss": 1.7015, + "step": 551 + }, + { + "epoch": 0.5670261941448382, + "grad_norm": 2.7917397022247314, + "learning_rate": 4.533688618195661e-06, + "loss": 1.6811, + "step": 552 + }, + { + "epoch": 0.5680534155110426, + "grad_norm": 2.4205498695373535, + "learning_rate": 4.531785306433194e-06, + "loss": 1.6659, + "step": 553 + }, + { + "epoch": 0.5690806368772471, + "grad_norm": 2.74898624420166, + "learning_rate": 4.529881994670727e-06, + "loss": 1.6521, + "step": 554 + }, + { + "epoch": 0.5701078582434514, + "grad_norm": 2.859898328781128, + "learning_rate": 4.527978682908261e-06, + "loss": 1.6445, + "step": 555 + }, + { + "epoch": 0.5711350796096559, + "grad_norm": 2.5526154041290283, + "learning_rate": 4.526075371145794e-06, + "loss": 1.6679, + "step": 556 + }, + { + "epoch": 0.5721623009758603, + "grad_norm": 2.96175217628479, + "learning_rate": 4.5241720593833275e-06, + "loss": 1.6784, + "step": 557 + }, + { + "epoch": 0.5731895223420647, + "grad_norm": 2.5037007331848145, + "learning_rate": 4.52226874762086e-06, + "loss": 1.6689, + "step": 558 + }, + { + "epoch": 0.5742167437082691, + "grad_norm": 2.6513259410858154, + "learning_rate": 4.520365435858393e-06, + "loss": 1.7038, + "step": 559 + }, + { + "epoch": 0.5752439650744735, + "grad_norm": 2.762770175933838, + "learning_rate": 4.518462124095928e-06, + "loss": 1.728, + "step": 560 + }, + { + "epoch": 0.5752439650744735, + "eval_loss": 1.972934603691101, + "eval_runtime": 44.2572, + "eval_samples_per_second": 22.595, + "eval_steps_per_second": 0.949, + "step": 560 + }, + { + "epoch": 0.576271186440678, + "grad_norm": 2.637406349182129, + "learning_rate": 4.516558812333461e-06, + "loss": 1.682, + "step": 561 + }, + { + "epoch": 0.5772984078068824, + "grad_norm": 2.545583724975586, + "learning_rate": 4.514655500570994e-06, + "loss": 1.6669, + "step": 562 + }, + { + "epoch": 0.5783256291730868, + "grad_norm": 2.5655033588409424, + "learning_rate": 4.512752188808528e-06, + "loss": 1.7279, + "step": 563 + }, + { + "epoch": 0.5793528505392912, + "grad_norm": 2.6856470108032227, + "learning_rate": 4.510848877046061e-06, + "loss": 1.7331, + "step": 564 + }, + { + "epoch": 0.5803800719054957, + "grad_norm": 2.605252981185913, + "learning_rate": 4.5089455652835936e-06, + "loss": 1.6904, + "step": 565 + }, + { + "epoch": 0.5814072932717, + "grad_norm": 2.610267400741577, + "learning_rate": 4.507042253521127e-06, + "loss": 1.6607, + "step": 566 + }, + { + "epoch": 0.5824345146379045, + "grad_norm": 2.6955432891845703, + "learning_rate": 4.50513894175866e-06, + "loss": 1.6166, + "step": 567 + }, + { + "epoch": 0.5834617360041089, + "grad_norm": 2.5661842823028564, + "learning_rate": 4.503235629996194e-06, + "loss": 1.6955, + "step": 568 + }, + { + "epoch": 0.5844889573703133, + "grad_norm": 2.6393115520477295, + "learning_rate": 4.501332318233727e-06, + "loss": 1.6933, + "step": 569 + }, + { + "epoch": 0.5855161787365177, + "grad_norm": 2.612645387649536, + "learning_rate": 4.49942900647126e-06, + "loss": 1.6585, + "step": 570 + }, + { + "epoch": 0.5855161787365177, + "eval_loss": 1.9599064588546753, + "eval_runtime": 46.0676, + "eval_samples_per_second": 21.707, + "eval_steps_per_second": 0.912, + "step": 570 + }, + { + "epoch": 0.5865434001027221, + "grad_norm": 2.4182891845703125, + "learning_rate": 4.497525694708794e-06, + "loss": 1.707, + "step": 571 + }, + { + "epoch": 0.5875706214689266, + "grad_norm": 2.7214412689208984, + "learning_rate": 4.495622382946327e-06, + "loss": 1.6966, + "step": 572 + }, + { + "epoch": 0.588597842835131, + "grad_norm": 2.5138018131256104, + "learning_rate": 4.4937190711838605e-06, + "loss": 1.6373, + "step": 573 + }, + { + "epoch": 0.5896250642013354, + "grad_norm": 2.5814292430877686, + "learning_rate": 4.491815759421394e-06, + "loss": 1.7313, + "step": 574 + }, + { + "epoch": 0.5906522855675398, + "grad_norm": 2.6236367225646973, + "learning_rate": 4.489912447658926e-06, + "loss": 1.6647, + "step": 575 + }, + { + "epoch": 0.5916795069337443, + "grad_norm": 2.468557357788086, + "learning_rate": 4.48800913589646e-06, + "loss": 1.6412, + "step": 576 + }, + { + "epoch": 0.5927067282999486, + "grad_norm": 2.454141855239868, + "learning_rate": 4.486105824133993e-06, + "loss": 1.6762, + "step": 577 + }, + { + "epoch": 0.593733949666153, + "grad_norm": 3.0009915828704834, + "learning_rate": 4.4842025123715264e-06, + "loss": 1.6847, + "step": 578 + }, + { + "epoch": 0.5947611710323575, + "grad_norm": 2.574394702911377, + "learning_rate": 4.48229920060906e-06, + "loss": 1.6421, + "step": 579 + }, + { + "epoch": 0.5957883923985619, + "grad_norm": 2.6978182792663574, + "learning_rate": 4.480395888846594e-06, + "loss": 1.6198, + "step": 580 + }, + { + "epoch": 0.5957883923985619, + "eval_loss": 1.9570144414901733, + "eval_runtime": 44.7661, + "eval_samples_per_second": 22.338, + "eval_steps_per_second": 0.938, + "step": 580 + }, + { + "epoch": 0.5968156137647663, + "grad_norm": 2.652461051940918, + "learning_rate": 4.478492577084127e-06, + "loss": 1.7334, + "step": 581 + }, + { + "epoch": 0.5978428351309707, + "grad_norm": 2.6292548179626465, + "learning_rate": 4.47658926532166e-06, + "loss": 1.6242, + "step": 582 + }, + { + "epoch": 0.5988700564971752, + "grad_norm": 2.474919080734253, + "learning_rate": 4.474685953559193e-06, + "loss": 1.6732, + "step": 583 + }, + { + "epoch": 0.5998972778633795, + "grad_norm": 2.7282516956329346, + "learning_rate": 4.472782641796727e-06, + "loss": 1.802, + "step": 584 + }, + { + "epoch": 0.600924499229584, + "grad_norm": 2.770307779312134, + "learning_rate": 4.47087933003426e-06, + "loss": 1.6156, + "step": 585 + }, + { + "epoch": 0.6019517205957884, + "grad_norm": 2.570681095123291, + "learning_rate": 4.468976018271793e-06, + "loss": 1.6309, + "step": 586 + }, + { + "epoch": 0.6029789419619929, + "grad_norm": 2.4075732231140137, + "learning_rate": 4.467072706509327e-06, + "loss": 1.6436, + "step": 587 + }, + { + "epoch": 0.6040061633281972, + "grad_norm": 2.4570937156677246, + "learning_rate": 4.46516939474686e-06, + "loss": 1.6737, + "step": 588 + }, + { + "epoch": 0.6050333846944016, + "grad_norm": 2.407274007797241, + "learning_rate": 4.463266082984393e-06, + "loss": 1.6179, + "step": 589 + }, + { + "epoch": 0.6060606060606061, + "grad_norm": 2.745486259460449, + "learning_rate": 4.461362771221927e-06, + "loss": 1.7458, + "step": 590 + }, + { + "epoch": 0.6060606060606061, + "eval_loss": 1.9700438976287842, + "eval_runtime": 45.3018, + "eval_samples_per_second": 22.074, + "eval_steps_per_second": 0.927, + "step": 590 + }, + { + "epoch": 0.6070878274268104, + "grad_norm": 2.4258809089660645, + "learning_rate": 4.45945945945946e-06, + "loss": 1.6372, + "step": 591 + }, + { + "epoch": 0.6081150487930149, + "grad_norm": 2.683662176132202, + "learning_rate": 4.457556147696993e-06, + "loss": 1.6649, + "step": 592 + }, + { + "epoch": 0.6091422701592193, + "grad_norm": 2.879727602005005, + "learning_rate": 4.455652835934526e-06, + "loss": 1.7376, + "step": 593 + }, + { + "epoch": 0.6101694915254238, + "grad_norm": 2.5620689392089844, + "learning_rate": 4.453749524172059e-06, + "loss": 1.7203, + "step": 594 + }, + { + "epoch": 0.6111967128916281, + "grad_norm": 2.662031888961792, + "learning_rate": 4.451846212409593e-06, + "loss": 1.6835, + "step": 595 + }, + { + "epoch": 0.6122239342578326, + "grad_norm": 2.6269922256469727, + "learning_rate": 4.449942900647126e-06, + "loss": 1.6726, + "step": 596 + }, + { + "epoch": 0.613251155624037, + "grad_norm": 2.515852212905884, + "learning_rate": 4.4480395888846595e-06, + "loss": 1.6045, + "step": 597 + }, + { + "epoch": 0.6142783769902413, + "grad_norm": 3.006873607635498, + "learning_rate": 4.446136277122193e-06, + "loss": 1.6827, + "step": 598 + }, + { + "epoch": 0.6153055983564458, + "grad_norm": 2.2904858589172363, + "learning_rate": 4.444232965359726e-06, + "loss": 1.6123, + "step": 599 + }, + { + "epoch": 0.6163328197226502, + "grad_norm": 2.7345471382141113, + "learning_rate": 4.4423296535972596e-06, + "loss": 1.6697, + "step": 600 + }, + { + "epoch": 0.6163328197226502, + "eval_loss": 1.9713997840881348, + "eval_runtime": 45.6715, + "eval_samples_per_second": 21.895, + "eval_steps_per_second": 0.92, + "step": 600 + }, + { + "epoch": 0.6173600410888547, + "grad_norm": 2.955853223800659, + "learning_rate": 4.440426341834793e-06, + "loss": 1.6195, + "step": 601 + }, + { + "epoch": 0.618387262455059, + "grad_norm": 2.4079766273498535, + "learning_rate": 4.438523030072326e-06, + "loss": 1.6589, + "step": 602 + }, + { + "epoch": 0.6194144838212635, + "grad_norm": 2.8158490657806396, + "learning_rate": 4.43661971830986e-06, + "loss": 1.6895, + "step": 603 + }, + { + "epoch": 0.6204417051874679, + "grad_norm": 2.491471529006958, + "learning_rate": 4.434716406547393e-06, + "loss": 1.6351, + "step": 604 + }, + { + "epoch": 0.6214689265536724, + "grad_norm": 2.4093260765075684, + "learning_rate": 4.432813094784926e-06, + "loss": 1.6402, + "step": 605 + }, + { + "epoch": 0.6224961479198767, + "grad_norm": 2.751018762588501, + "learning_rate": 4.43090978302246e-06, + "loss": 1.6962, + "step": 606 + }, + { + "epoch": 0.6235233692860811, + "grad_norm": 2.4410409927368164, + "learning_rate": 4.429006471259993e-06, + "loss": 1.6549, + "step": 607 + }, + { + "epoch": 0.6245505906522856, + "grad_norm": 2.360455274581909, + "learning_rate": 4.4271031594975265e-06, + "loss": 1.6449, + "step": 608 + }, + { + "epoch": 0.6255778120184899, + "grad_norm": 2.5058023929595947, + "learning_rate": 4.425199847735059e-06, + "loss": 1.617, + "step": 609 + }, + { + "epoch": 0.6266050333846944, + "grad_norm": 2.3429927825927734, + "learning_rate": 4.423296535972592e-06, + "loss": 1.6262, + "step": 610 + }, + { + "epoch": 0.6266050333846944, + "eval_loss": 1.9597365856170654, + "eval_runtime": 42.2034, + "eval_samples_per_second": 23.695, + "eval_steps_per_second": 0.995, + "step": 610 + }, + { + "epoch": 0.6276322547508988, + "grad_norm": 2.401761054992676, + "learning_rate": 4.421393224210126e-06, + "loss": 1.6497, + "step": 611 + }, + { + "epoch": 0.6286594761171033, + "grad_norm": 2.433439254760742, + "learning_rate": 4.419489912447659e-06, + "loss": 1.6117, + "step": 612 + }, + { + "epoch": 0.6296866974833076, + "grad_norm": 2.5147452354431152, + "learning_rate": 4.4175866006851924e-06, + "loss": 1.7041, + "step": 613 + }, + { + "epoch": 0.6307139188495121, + "grad_norm": 2.4174447059631348, + "learning_rate": 4.415683288922726e-06, + "loss": 1.6652, + "step": 614 + }, + { + "epoch": 0.6317411402157165, + "grad_norm": 2.424800157546997, + "learning_rate": 4.413779977160259e-06, + "loss": 1.6407, + "step": 615 + }, + { + "epoch": 0.632768361581921, + "grad_norm": 2.6760315895080566, + "learning_rate": 4.4118766653977925e-06, + "loss": 1.6566, + "step": 616 + }, + { + "epoch": 0.6337955829481253, + "grad_norm": 2.694810628890991, + "learning_rate": 4.409973353635326e-06, + "loss": 1.7125, + "step": 617 + }, + { + "epoch": 0.6348228043143297, + "grad_norm": 2.475126266479492, + "learning_rate": 4.408070041872859e-06, + "loss": 1.6663, + "step": 618 + }, + { + "epoch": 0.6358500256805342, + "grad_norm": 2.734738349914551, + "learning_rate": 4.406166730110393e-06, + "loss": 1.6115, + "step": 619 + }, + { + "epoch": 0.6368772470467385, + "grad_norm": 2.4914753437042236, + "learning_rate": 4.404263418347925e-06, + "loss": 1.6096, + "step": 620 + }, + { + "epoch": 0.6368772470467385, + "eval_loss": 1.952081322669983, + "eval_runtime": 45.4451, + "eval_samples_per_second": 22.005, + "eval_steps_per_second": 0.924, + "step": 620 + }, + { + "epoch": 0.637904468412943, + "grad_norm": 2.3727800846099854, + "learning_rate": 4.402360106585459e-06, + "loss": 1.6275, + "step": 621 + }, + { + "epoch": 0.6389316897791474, + "grad_norm": 2.576402187347412, + "learning_rate": 4.400456794822993e-06, + "loss": 1.6252, + "step": 622 + }, + { + "epoch": 0.6399589111453519, + "grad_norm": 2.560185432434082, + "learning_rate": 4.398553483060526e-06, + "loss": 1.6522, + "step": 623 + }, + { + "epoch": 0.6409861325115562, + "grad_norm": 2.462785005569458, + "learning_rate": 4.396650171298059e-06, + "loss": 1.6124, + "step": 624 + }, + { + "epoch": 0.6420133538777607, + "grad_norm": 2.3942930698394775, + "learning_rate": 4.394746859535593e-06, + "loss": 1.6231, + "step": 625 + }, + { + "epoch": 0.6430405752439651, + "grad_norm": 2.4870522022247314, + "learning_rate": 4.392843547773125e-06, + "loss": 1.6086, + "step": 626 + }, + { + "epoch": 0.6440677966101694, + "grad_norm": 2.5702786445617676, + "learning_rate": 4.390940236010659e-06, + "loss": 1.6228, + "step": 627 + }, + { + "epoch": 0.6450950179763739, + "grad_norm": 2.743983268737793, + "learning_rate": 4.389036924248192e-06, + "loss": 1.6394, + "step": 628 + }, + { + "epoch": 0.6461222393425783, + "grad_norm": 2.59942364692688, + "learning_rate": 4.387133612485725e-06, + "loss": 1.6721, + "step": 629 + }, + { + "epoch": 0.6471494607087828, + "grad_norm": 2.4473204612731934, + "learning_rate": 4.385230300723259e-06, + "loss": 1.6407, + "step": 630 + }, + { + "epoch": 0.6471494607087828, + "eval_loss": 1.9426974058151245, + "eval_runtime": 45.2318, + "eval_samples_per_second": 22.108, + "eval_steps_per_second": 0.929, + "step": 630 + }, + { + "epoch": 0.6481766820749871, + "grad_norm": 2.6036875247955322, + "learning_rate": 4.383326988960792e-06, + "loss": 1.6307, + "step": 631 + }, + { + "epoch": 0.6492039034411916, + "grad_norm": 2.691011428833008, + "learning_rate": 4.3814236771983255e-06, + "loss": 1.6059, + "step": 632 + }, + { + "epoch": 0.650231124807396, + "grad_norm": 2.585758924484253, + "learning_rate": 4.379520365435859e-06, + "loss": 1.6781, + "step": 633 + }, + { + "epoch": 0.6512583461736005, + "grad_norm": 2.9522950649261475, + "learning_rate": 4.377617053673392e-06, + "loss": 1.6559, + "step": 634 + }, + { + "epoch": 0.6522855675398048, + "grad_norm": 2.5368709564208984, + "learning_rate": 4.3757137419109256e-06, + "loss": 1.6445, + "step": 635 + }, + { + "epoch": 0.6533127889060092, + "grad_norm": 2.5002150535583496, + "learning_rate": 4.373810430148459e-06, + "loss": 1.7184, + "step": 636 + }, + { + "epoch": 0.6543400102722137, + "grad_norm": 2.7623229026794434, + "learning_rate": 4.3719071183859914e-06, + "loss": 1.6195, + "step": 637 + }, + { + "epoch": 0.655367231638418, + "grad_norm": 2.6188151836395264, + "learning_rate": 4.370003806623525e-06, + "loss": 1.6686, + "step": 638 + }, + { + "epoch": 0.6563944530046225, + "grad_norm": 2.685776948928833, + "learning_rate": 4.368100494861058e-06, + "loss": 1.6134, + "step": 639 + }, + { + "epoch": 0.6574216743708269, + "grad_norm": 2.5703907012939453, + "learning_rate": 4.3661971830985915e-06, + "loss": 1.665, + "step": 640 + }, + { + "epoch": 0.6574216743708269, + "eval_loss": 1.9504806995391846, + "eval_runtime": 47.1947, + "eval_samples_per_second": 21.189, + "eval_steps_per_second": 0.89, + "step": 640 + }, + { + "epoch": 0.6584488957370314, + "grad_norm": 2.511617660522461, + "learning_rate": 4.364293871336126e-06, + "loss": 1.6201, + "step": 641 + }, + { + "epoch": 0.6594761171032357, + "grad_norm": 2.549302339553833, + "learning_rate": 4.362390559573659e-06, + "loss": 1.647, + "step": 642 + }, + { + "epoch": 0.6605033384694402, + "grad_norm": 2.348747730255127, + "learning_rate": 4.3604872478111925e-06, + "loss": 1.5993, + "step": 643 + }, + { + "epoch": 0.6615305598356446, + "grad_norm": 2.7393345832824707, + "learning_rate": 4.358583936048725e-06, + "loss": 1.543, + "step": 644 + }, + { + "epoch": 0.662557781201849, + "grad_norm": 2.4242193698883057, + "learning_rate": 4.356680624286258e-06, + "loss": 1.6726, + "step": 645 + }, + { + "epoch": 0.6635850025680534, + "grad_norm": 2.6446714401245117, + "learning_rate": 4.354777312523792e-06, + "loss": 1.6312, + "step": 646 + }, + { + "epoch": 0.6646122239342578, + "grad_norm": 2.7407124042510986, + "learning_rate": 4.352874000761325e-06, + "loss": 1.6868, + "step": 647 + }, + { + "epoch": 0.6656394453004623, + "grad_norm": 2.3706371784210205, + "learning_rate": 4.3509706889988584e-06, + "loss": 1.6301, + "step": 648 + }, + { + "epoch": 0.6666666666666666, + "grad_norm": 2.5197198390960693, + "learning_rate": 4.349067377236392e-06, + "loss": 1.702, + "step": 649 + }, + { + "epoch": 0.6676938880328711, + "grad_norm": 2.4792377948760986, + "learning_rate": 4.347164065473925e-06, + "loss": 1.6891, + "step": 650 + }, + { + "epoch": 0.6676938880328711, + "eval_loss": 1.9589039087295532, + "eval_runtime": 47.1512, + "eval_samples_per_second": 21.208, + "eval_steps_per_second": 0.891, + "step": 650 + }, + { + "epoch": 0.6687211093990755, + "grad_norm": 2.4003701210021973, + "learning_rate": 4.3452607537114585e-06, + "loss": 1.7015, + "step": 651 + }, + { + "epoch": 0.66974833076528, + "grad_norm": 2.5363786220550537, + "learning_rate": 4.343357441948992e-06, + "loss": 1.6133, + "step": 652 + }, + { + "epoch": 0.6707755521314843, + "grad_norm": 2.5183844566345215, + "learning_rate": 4.341454130186525e-06, + "loss": 1.6574, + "step": 653 + }, + { + "epoch": 0.6718027734976888, + "grad_norm": 2.4668304920196533, + "learning_rate": 4.339550818424058e-06, + "loss": 1.6253, + "step": 654 + }, + { + "epoch": 0.6728299948638932, + "grad_norm": 2.4829814434051514, + "learning_rate": 4.337647506661591e-06, + "loss": 1.6099, + "step": 655 + }, + { + "epoch": 0.6738572162300975, + "grad_norm": 2.3683176040649414, + "learning_rate": 4.3357441948991245e-06, + "loss": 1.6385, + "step": 656 + }, + { + "epoch": 0.674884437596302, + "grad_norm": 2.4906842708587646, + "learning_rate": 4.333840883136658e-06, + "loss": 1.6755, + "step": 657 + }, + { + "epoch": 0.6759116589625064, + "grad_norm": 2.558776617050171, + "learning_rate": 4.331937571374191e-06, + "loss": 1.6152, + "step": 658 + }, + { + "epoch": 0.6769388803287109, + "grad_norm": 2.4081523418426514, + "learning_rate": 4.3300342596117246e-06, + "loss": 1.6224, + "step": 659 + }, + { + "epoch": 0.6779661016949152, + "grad_norm": 2.488370180130005, + "learning_rate": 4.328130947849258e-06, + "loss": 1.6298, + "step": 660 + }, + { + "epoch": 0.6779661016949152, + "eval_loss": 1.9484621286392212, + "eval_runtime": 52.549, + "eval_samples_per_second": 19.03, + "eval_steps_per_second": 0.799, + "step": 660 + }, + { + "epoch": 0.6789933230611197, + "grad_norm": 2.495748281478882, + "learning_rate": 4.326227636086791e-06, + "loss": 1.5852, + "step": 661 + }, + { + "epoch": 0.6800205444273241, + "grad_norm": 2.525031566619873, + "learning_rate": 4.324324324324325e-06, + "loss": 1.6348, + "step": 662 + }, + { + "epoch": 0.6810477657935285, + "grad_norm": 2.427615165710449, + "learning_rate": 4.322421012561858e-06, + "loss": 1.6352, + "step": 663 + }, + { + "epoch": 0.6820749871597329, + "grad_norm": 2.5084927082061768, + "learning_rate": 4.320517700799391e-06, + "loss": 1.6334, + "step": 664 + }, + { + "epoch": 0.6831022085259374, + "grad_norm": 2.6018898487091064, + "learning_rate": 4.318614389036925e-06, + "loss": 1.7212, + "step": 665 + }, + { + "epoch": 0.6841294298921418, + "grad_norm": 2.44184947013855, + "learning_rate": 4.316711077274458e-06, + "loss": 1.6802, + "step": 666 + }, + { + "epoch": 0.6851566512583461, + "grad_norm": 2.542123317718506, + "learning_rate": 4.3148077655119915e-06, + "loss": 1.5912, + "step": 667 + }, + { + "epoch": 0.6861838726245506, + "grad_norm": 2.5208137035369873, + "learning_rate": 4.312904453749525e-06, + "loss": 1.6965, + "step": 668 + }, + { + "epoch": 0.687211093990755, + "grad_norm": 2.391904592514038, + "learning_rate": 4.311001141987058e-06, + "loss": 1.6388, + "step": 669 + }, + { + "epoch": 0.6882383153569595, + "grad_norm": 2.629565477371216, + "learning_rate": 4.3090978302245916e-06, + "loss": 1.6475, + "step": 670 + }, + { + "epoch": 0.6882383153569595, + "eval_loss": 1.9572818279266357, + "eval_runtime": 48.5354, + "eval_samples_per_second": 20.604, + "eval_steps_per_second": 0.865, + "step": 670 + }, + { + "epoch": 0.6892655367231638, + "grad_norm": 2.3132376670837402, + "learning_rate": 4.307194518462124e-06, + "loss": 1.6399, + "step": 671 + }, + { + "epoch": 0.6902927580893683, + "grad_norm": 2.5003411769866943, + "learning_rate": 4.3052912066996574e-06, + "loss": 1.7136, + "step": 672 + }, + { + "epoch": 0.6913199794555727, + "grad_norm": 2.545541286468506, + "learning_rate": 4.303387894937191e-06, + "loss": 1.6825, + "step": 673 + }, + { + "epoch": 0.6923472008217771, + "grad_norm": 2.342423439025879, + "learning_rate": 4.301484583174724e-06, + "loss": 1.6088, + "step": 674 + }, + { + "epoch": 0.6933744221879815, + "grad_norm": 2.5516135692596436, + "learning_rate": 4.2995812714122575e-06, + "loss": 1.5934, + "step": 675 + }, + { + "epoch": 0.6944016435541859, + "grad_norm": 2.438675880432129, + "learning_rate": 4.297677959649791e-06, + "loss": 1.6236, + "step": 676 + }, + { + "epoch": 0.6954288649203904, + "grad_norm": 2.767305612564087, + "learning_rate": 4.295774647887324e-06, + "loss": 1.6514, + "step": 677 + }, + { + "epoch": 0.6964560862865947, + "grad_norm": 2.591994047164917, + "learning_rate": 4.293871336124858e-06, + "loss": 1.6643, + "step": 678 + }, + { + "epoch": 0.6974833076527992, + "grad_norm": 2.4734244346618652, + "learning_rate": 4.291968024362391e-06, + "loss": 1.6362, + "step": 679 + }, + { + "epoch": 0.6985105290190036, + "grad_norm": 2.6181890964508057, + "learning_rate": 4.290064712599924e-06, + "loss": 1.6666, + "step": 680 + }, + { + "epoch": 0.6985105290190036, + "eval_loss": 1.9585388898849487, + "eval_runtime": 46.8094, + "eval_samples_per_second": 21.363, + "eval_steps_per_second": 0.897, + "step": 680 + }, + { + "epoch": 0.699537750385208, + "grad_norm": 2.465571403503418, + "learning_rate": 4.288161400837458e-06, + "loss": 1.6588, + "step": 681 + }, + { + "epoch": 0.7005649717514124, + "grad_norm": 2.664262294769287, + "learning_rate": 4.286258089074991e-06, + "loss": 1.5862, + "step": 682 + }, + { + "epoch": 0.7015921931176169, + "grad_norm": 2.677140474319458, + "learning_rate": 4.284354777312524e-06, + "loss": 1.6549, + "step": 683 + }, + { + "epoch": 0.7026194144838213, + "grad_norm": 2.559051275253296, + "learning_rate": 4.282451465550058e-06, + "loss": 1.609, + "step": 684 + }, + { + "epoch": 0.7036466358500256, + "grad_norm": 2.428424119949341, + "learning_rate": 4.280548153787591e-06, + "loss": 1.6243, + "step": 685 + }, + { + "epoch": 0.7046738572162301, + "grad_norm": 2.54136061668396, + "learning_rate": 4.2786448420251245e-06, + "loss": 1.6301, + "step": 686 + }, + { + "epoch": 0.7057010785824345, + "grad_norm": 2.377699136734009, + "learning_rate": 4.276741530262658e-06, + "loss": 1.5903, + "step": 687 + }, + { + "epoch": 0.706728299948639, + "grad_norm": 2.5548322200775146, + "learning_rate": 4.27483821850019e-06, + "loss": 1.6161, + "step": 688 + }, + { + "epoch": 0.7077555213148433, + "grad_norm": 2.5993409156799316, + "learning_rate": 4.272934906737724e-06, + "loss": 1.6882, + "step": 689 + }, + { + "epoch": 0.7087827426810478, + "grad_norm": 2.4249117374420166, + "learning_rate": 4.271031594975257e-06, + "loss": 1.6723, + "step": 690 + }, + { + "epoch": 0.7087827426810478, + "eval_loss": 1.9526821374893188, + "eval_runtime": 48.2168, + "eval_samples_per_second": 20.74, + "eval_steps_per_second": 0.871, + "step": 690 + }, + { + "epoch": 0.7098099640472522, + "grad_norm": 2.6965081691741943, + "learning_rate": 4.2691282832127905e-06, + "loss": 1.6636, + "step": 691 + }, + { + "epoch": 0.7108371854134566, + "grad_norm": 2.6205356121063232, + "learning_rate": 4.267224971450324e-06, + "loss": 1.6231, + "step": 692 + }, + { + "epoch": 0.711864406779661, + "grad_norm": 2.5924129486083984, + "learning_rate": 4.265321659687857e-06, + "loss": 1.5634, + "step": 693 + }, + { + "epoch": 0.7128916281458655, + "grad_norm": 2.9710943698883057, + "learning_rate": 4.2634183479253906e-06, + "loss": 1.6134, + "step": 694 + }, + { + "epoch": 0.7139188495120699, + "grad_norm": 2.6252782344818115, + "learning_rate": 4.261515036162924e-06, + "loss": 1.5997, + "step": 695 + }, + { + "epoch": 0.7149460708782742, + "grad_norm": 2.6178832054138184, + "learning_rate": 4.259611724400457e-06, + "loss": 1.6255, + "step": 696 + }, + { + "epoch": 0.7159732922444787, + "grad_norm": 2.82548189163208, + "learning_rate": 4.257708412637991e-06, + "loss": 1.6167, + "step": 697 + }, + { + "epoch": 0.7170005136106831, + "grad_norm": 2.6430351734161377, + "learning_rate": 4.255805100875523e-06, + "loss": 1.5585, + "step": 698 + }, + { + "epoch": 0.7180277349768875, + "grad_norm": 2.85046124458313, + "learning_rate": 4.2539017891130565e-06, + "loss": 1.5873, + "step": 699 + }, + { + "epoch": 0.7190549563430919, + "grad_norm": 2.5643832683563232, + "learning_rate": 4.25199847735059e-06, + "loss": 1.6705, + "step": 700 + }, + { + "epoch": 0.7190549563430919, + "eval_loss": 1.945617437362671, + "eval_runtime": 54.2778, + "eval_samples_per_second": 18.424, + "eval_steps_per_second": 0.774, + "step": 700 + }, + { + "epoch": 0.7200821777092964, + "grad_norm": 2.7856431007385254, + "learning_rate": 4.250095165588123e-06, + "loss": 1.5833, + "step": 701 + }, + { + "epoch": 0.7211093990755008, + "grad_norm": 2.9881107807159424, + "learning_rate": 4.2481918538256575e-06, + "loss": 1.6438, + "step": 702 + }, + { + "epoch": 0.7221366204417052, + "grad_norm": 2.602938652038574, + "learning_rate": 4.246288542063191e-06, + "loss": 1.6135, + "step": 703 + }, + { + "epoch": 0.7231638418079096, + "grad_norm": 2.997450590133667, + "learning_rate": 4.244385230300724e-06, + "loss": 1.6419, + "step": 704 + }, + { + "epoch": 0.724191063174114, + "grad_norm": 2.5592496395111084, + "learning_rate": 4.242481918538257e-06, + "loss": 1.5886, + "step": 705 + }, + { + "epoch": 0.7252182845403184, + "grad_norm": 2.4073948860168457, + "learning_rate": 4.24057860677579e-06, + "loss": 1.6219, + "step": 706 + }, + { + "epoch": 0.7262455059065228, + "grad_norm": 2.545997381210327, + "learning_rate": 4.2386752950133234e-06, + "loss": 1.6684, + "step": 707 + }, + { + "epoch": 0.7272727272727273, + "grad_norm": 2.468264579772949, + "learning_rate": 4.236771983250857e-06, + "loss": 1.6155, + "step": 708 + }, + { + "epoch": 0.7282999486389317, + "grad_norm": 2.6635043621063232, + "learning_rate": 4.23486867148839e-06, + "loss": 1.5956, + "step": 709 + }, + { + "epoch": 0.7293271700051361, + "grad_norm": 2.8751156330108643, + "learning_rate": 4.2329653597259235e-06, + "loss": 1.6211, + "step": 710 + }, + { + "epoch": 0.7293271700051361, + "eval_loss": 1.9319628477096558, + "eval_runtime": 43.9701, + "eval_samples_per_second": 22.743, + "eval_steps_per_second": 0.955, + "step": 710 + }, + { + "epoch": 0.7303543913713405, + "grad_norm": 2.5999035835266113, + "learning_rate": 4.231062047963457e-06, + "loss": 1.6442, + "step": 711 + }, + { + "epoch": 0.731381612737545, + "grad_norm": 2.611095666885376, + "learning_rate": 4.22915873620099e-06, + "loss": 1.6745, + "step": 712 + }, + { + "epoch": 0.7324088341037494, + "grad_norm": 2.7235333919525146, + "learning_rate": 4.227255424438524e-06, + "loss": 1.6353, + "step": 713 + }, + { + "epoch": 0.7334360554699538, + "grad_norm": 2.453639507293701, + "learning_rate": 4.225352112676057e-06, + "loss": 1.6657, + "step": 714 + }, + { + "epoch": 0.7344632768361582, + "grad_norm": 2.4031429290771484, + "learning_rate": 4.2234488009135895e-06, + "loss": 1.603, + "step": 715 + }, + { + "epoch": 0.7354904982023626, + "grad_norm": 2.631052017211914, + "learning_rate": 4.221545489151123e-06, + "loss": 1.5681, + "step": 716 + }, + { + "epoch": 0.736517719568567, + "grad_norm": 2.518758773803711, + "learning_rate": 4.219642177388656e-06, + "loss": 1.5941, + "step": 717 + }, + { + "epoch": 0.7375449409347714, + "grad_norm": 2.699491500854492, + "learning_rate": 4.2177388656261896e-06, + "loss": 1.6327, + "step": 718 + }, + { + "epoch": 0.7385721623009759, + "grad_norm": 2.749356985092163, + "learning_rate": 4.215835553863723e-06, + "loss": 1.5864, + "step": 719 + }, + { + "epoch": 0.7395993836671803, + "grad_norm": 2.5699031352996826, + "learning_rate": 4.213932242101256e-06, + "loss": 1.5444, + "step": 720 + }, + { + "epoch": 0.7395993836671803, + "eval_loss": 1.9521623849868774, + "eval_runtime": 46.5749, + "eval_samples_per_second": 21.471, + "eval_steps_per_second": 0.902, + "step": 720 + }, + { + "epoch": 0.7406266050333847, + "grad_norm": 3.197425127029419, + "learning_rate": 4.21202893033879e-06, + "loss": 1.6158, + "step": 721 + }, + { + "epoch": 0.7416538263995891, + "grad_norm": 2.619459390640259, + "learning_rate": 4.210125618576323e-06, + "loss": 1.6328, + "step": 722 + }, + { + "epoch": 0.7426810477657936, + "grad_norm": 2.792086601257324, + "learning_rate": 4.208222306813856e-06, + "loss": 1.6333, + "step": 723 + }, + { + "epoch": 0.743708269131998, + "grad_norm": 2.7023141384124756, + "learning_rate": 4.20631899505139e-06, + "loss": 1.6787, + "step": 724 + }, + { + "epoch": 0.7447354904982023, + "grad_norm": 2.6319735050201416, + "learning_rate": 4.204415683288923e-06, + "loss": 1.54, + "step": 725 + }, + { + "epoch": 0.7457627118644068, + "grad_norm": 2.7032439708709717, + "learning_rate": 4.2025123715264565e-06, + "loss": 1.641, + "step": 726 + }, + { + "epoch": 0.7467899332306112, + "grad_norm": 2.9133267402648926, + "learning_rate": 4.20060905976399e-06, + "loss": 1.5901, + "step": 727 + }, + { + "epoch": 0.7478171545968156, + "grad_norm": 2.492403745651245, + "learning_rate": 4.198705748001523e-06, + "loss": 1.5516, + "step": 728 + }, + { + "epoch": 0.74884437596302, + "grad_norm": 2.604520082473755, + "learning_rate": 4.1968024362390566e-06, + "loss": 1.6574, + "step": 729 + }, + { + "epoch": 0.7498715973292245, + "grad_norm": 2.521969795227051, + "learning_rate": 4.19489912447659e-06, + "loss": 1.6099, + "step": 730 + }, + { + "epoch": 0.7498715973292245, + "eval_loss": 1.9449840784072876, + "eval_runtime": 44.9334, + "eval_samples_per_second": 22.255, + "eval_steps_per_second": 0.935, + "step": 730 + }, + { + "epoch": 0.7508988186954288, + "grad_norm": 2.4645657539367676, + "learning_rate": 4.192995812714123e-06, + "loss": 1.6752, + "step": 731 + }, + { + "epoch": 0.7519260400616333, + "grad_norm": 2.7852981090545654, + "learning_rate": 4.191092500951656e-06, + "loss": 1.6176, + "step": 732 + }, + { + "epoch": 0.7529532614278377, + "grad_norm": 2.2817575931549072, + "learning_rate": 4.189189189189189e-06, + "loss": 1.5934, + "step": 733 + }, + { + "epoch": 0.7539804827940421, + "grad_norm": 2.6218173503875732, + "learning_rate": 4.1872858774267225e-06, + "loss": 1.5902, + "step": 734 + }, + { + "epoch": 0.7550077041602465, + "grad_norm": 2.6671013832092285, + "learning_rate": 4.185382565664256e-06, + "loss": 1.657, + "step": 735 + }, + { + "epoch": 0.7560349255264509, + "grad_norm": 2.669757843017578, + "learning_rate": 4.183479253901789e-06, + "loss": 1.5884, + "step": 736 + }, + { + "epoch": 0.7570621468926554, + "grad_norm": 2.7210471630096436, + "learning_rate": 4.181575942139323e-06, + "loss": 1.5558, + "step": 737 + }, + { + "epoch": 0.7580893682588598, + "grad_norm": 2.414276123046875, + "learning_rate": 4.179672630376856e-06, + "loss": 1.6124, + "step": 738 + }, + { + "epoch": 0.7591165896250642, + "grad_norm": 2.4643094539642334, + "learning_rate": 4.177769318614389e-06, + "loss": 1.5904, + "step": 739 + }, + { + "epoch": 0.7601438109912686, + "grad_norm": 2.5447018146514893, + "learning_rate": 4.175866006851923e-06, + "loss": 1.6705, + "step": 740 + }, + { + "epoch": 0.7601438109912686, + "eval_loss": 1.9437954425811768, + "eval_runtime": 49.2702, + "eval_samples_per_second": 20.296, + "eval_steps_per_second": 0.852, + "step": 740 + }, + { + "epoch": 0.7611710323574731, + "grad_norm": 2.41048002243042, + "learning_rate": 4.173962695089456e-06, + "loss": 1.6398, + "step": 741 + }, + { + "epoch": 0.7621982537236774, + "grad_norm": 2.3949906826019287, + "learning_rate": 4.1720593833269894e-06, + "loss": 1.6198, + "step": 742 + }, + { + "epoch": 0.7632254750898819, + "grad_norm": 2.36966609954834, + "learning_rate": 4.170156071564523e-06, + "loss": 1.6275, + "step": 743 + }, + { + "epoch": 0.7642526964560863, + "grad_norm": 2.623347043991089, + "learning_rate": 4.168252759802056e-06, + "loss": 1.7059, + "step": 744 + }, + { + "epoch": 0.7652799178222907, + "grad_norm": 2.384788751602173, + "learning_rate": 4.1663494480395895e-06, + "loss": 1.6015, + "step": 745 + }, + { + "epoch": 0.7663071391884951, + "grad_norm": 2.586009979248047, + "learning_rate": 4.164446136277123e-06, + "loss": 1.6445, + "step": 746 + }, + { + "epoch": 0.7673343605546995, + "grad_norm": 2.396684169769287, + "learning_rate": 4.162542824514656e-06, + "loss": 1.6539, + "step": 747 + }, + { + "epoch": 0.768361581920904, + "grad_norm": 2.416243553161621, + "learning_rate": 4.16063951275219e-06, + "loss": 1.6142, + "step": 748 + }, + { + "epoch": 0.7693888032871083, + "grad_norm": 2.480912208557129, + "learning_rate": 4.158736200989722e-06, + "loss": 1.5768, + "step": 749 + }, + { + "epoch": 0.7704160246533128, + "grad_norm": 2.5516929626464844, + "learning_rate": 4.1568328892272555e-06, + "loss": 1.6074, + "step": 750 + }, + { + "epoch": 0.7704160246533128, + "eval_loss": 1.9409116506576538, + "eval_runtime": 45.3465, + "eval_samples_per_second": 22.052, + "eval_steps_per_second": 0.926, + "step": 750 + }, + { + "epoch": 0.7714432460195172, + "grad_norm": 2.4248714447021484, + "learning_rate": 4.154929577464789e-06, + "loss": 1.6308, + "step": 751 + }, + { + "epoch": 0.7724704673857217, + "grad_norm": 2.520810604095459, + "learning_rate": 4.153026265702322e-06, + "loss": 1.6227, + "step": 752 + }, + { + "epoch": 0.773497688751926, + "grad_norm": 2.5938422679901123, + "learning_rate": 4.1511229539398556e-06, + "loss": 1.5601, + "step": 753 + }, + { + "epoch": 0.7745249101181304, + "grad_norm": 2.4538512229919434, + "learning_rate": 4.149219642177389e-06, + "loss": 1.6006, + "step": 754 + }, + { + "epoch": 0.7755521314843349, + "grad_norm": 2.4936258792877197, + "learning_rate": 4.147316330414922e-06, + "loss": 1.6281, + "step": 755 + }, + { + "epoch": 0.7765793528505393, + "grad_norm": 2.5774738788604736, + "learning_rate": 4.145413018652456e-06, + "loss": 1.5328, + "step": 756 + }, + { + "epoch": 0.7776065742167437, + "grad_norm": 2.3616905212402344, + "learning_rate": 4.143509706889989e-06, + "loss": 1.629, + "step": 757 + }, + { + "epoch": 0.7786337955829481, + "grad_norm": 2.4983086585998535, + "learning_rate": 4.141606395127522e-06, + "loss": 1.6298, + "step": 758 + }, + { + "epoch": 0.7796610169491526, + "grad_norm": 2.4313876628875732, + "learning_rate": 4.139703083365056e-06, + "loss": 1.6113, + "step": 759 + }, + { + "epoch": 0.7806882383153569, + "grad_norm": 2.6290934085845947, + "learning_rate": 4.137799771602588e-06, + "loss": 1.6632, + "step": 760 + }, + { + "epoch": 0.7806882383153569, + "eval_loss": 1.950036644935608, + "eval_runtime": 47.4523, + "eval_samples_per_second": 21.074, + "eval_steps_per_second": 0.885, + "step": 760 + }, + { + "epoch": 0.7817154596815614, + "grad_norm": 2.3637750148773193, + "learning_rate": 4.135896459840122e-06, + "loss": 1.6268, + "step": 761 + }, + { + "epoch": 0.7827426810477658, + "grad_norm": 2.5537843704223633, + "learning_rate": 4.133993148077656e-06, + "loss": 1.5736, + "step": 762 + }, + { + "epoch": 0.7837699024139703, + "grad_norm": 2.4945273399353027, + "learning_rate": 4.132089836315189e-06, + "loss": 1.612, + "step": 763 + }, + { + "epoch": 0.7847971237801746, + "grad_norm": 2.506983757019043, + "learning_rate": 4.1301865245527226e-06, + "loss": 1.6112, + "step": 764 + }, + { + "epoch": 0.785824345146379, + "grad_norm": 2.599721908569336, + "learning_rate": 4.128283212790256e-06, + "loss": 1.5698, + "step": 765 + }, + { + "epoch": 0.7868515665125835, + "grad_norm": 2.4561266899108887, + "learning_rate": 4.1263799010277884e-06, + "loss": 1.5354, + "step": 766 + }, + { + "epoch": 0.7878787878787878, + "grad_norm": 2.445420026779175, + "learning_rate": 4.124476589265322e-06, + "loss": 1.6555, + "step": 767 + }, + { + "epoch": 0.7889060092449923, + "grad_norm": 2.576533079147339, + "learning_rate": 4.122573277502855e-06, + "loss": 1.5915, + "step": 768 + }, + { + "epoch": 0.7899332306111967, + "grad_norm": 2.503164291381836, + "learning_rate": 4.1206699657403885e-06, + "loss": 1.5893, + "step": 769 + }, + { + "epoch": 0.7909604519774012, + "grad_norm": 2.9463603496551514, + "learning_rate": 4.118766653977922e-06, + "loss": 1.6178, + "step": 770 + }, + { + "epoch": 0.7909604519774012, + "eval_loss": 1.9529738426208496, + "eval_runtime": 45.0922, + "eval_samples_per_second": 22.177, + "eval_steps_per_second": 0.931, + "step": 770 + }, + { + "epoch": 0.7919876733436055, + "grad_norm": 2.47148060798645, + "learning_rate": 4.116863342215455e-06, + "loss": 1.5991, + "step": 771 + }, + { + "epoch": 0.79301489470981, + "grad_norm": 2.515717029571533, + "learning_rate": 4.114960030452989e-06, + "loss": 1.5725, + "step": 772 + }, + { + "epoch": 0.7940421160760144, + "grad_norm": 2.5231993198394775, + "learning_rate": 4.113056718690522e-06, + "loss": 1.6145, + "step": 773 + }, + { + "epoch": 0.7950693374422187, + "grad_norm": 2.479978084564209, + "learning_rate": 4.111153406928055e-06, + "loss": 1.5537, + "step": 774 + }, + { + "epoch": 0.7960965588084232, + "grad_norm": 2.5341250896453857, + "learning_rate": 4.109250095165589e-06, + "loss": 1.6419, + "step": 775 + }, + { + "epoch": 0.7971237801746276, + "grad_norm": 2.6277401447296143, + "learning_rate": 4.107346783403122e-06, + "loss": 1.6505, + "step": 776 + }, + { + "epoch": 0.7981510015408321, + "grad_norm": 2.4688351154327393, + "learning_rate": 4.105443471640655e-06, + "loss": 1.5896, + "step": 777 + }, + { + "epoch": 0.7991782229070364, + "grad_norm": 2.4134581089019775, + "learning_rate": 4.103540159878188e-06, + "loss": 1.5788, + "step": 778 + }, + { + "epoch": 0.8002054442732409, + "grad_norm": 2.5090012550354004, + "learning_rate": 4.101636848115721e-06, + "loss": 1.6162, + "step": 779 + }, + { + "epoch": 0.8012326656394453, + "grad_norm": 2.5123302936553955, + "learning_rate": 4.099733536353255e-06, + "loss": 1.5868, + "step": 780 + }, + { + "epoch": 0.8012326656394453, + "eval_loss": 1.9545625448226929, + "eval_runtime": 45.9316, + "eval_samples_per_second": 21.772, + "eval_steps_per_second": 0.914, + "step": 780 + }, + { + "epoch": 0.8022598870056498, + "grad_norm": 2.787181854248047, + "learning_rate": 4.097830224590788e-06, + "loss": 1.612, + "step": 781 + }, + { + "epoch": 0.8032871083718541, + "grad_norm": 2.414426326751709, + "learning_rate": 4.095926912828322e-06, + "loss": 1.6908, + "step": 782 + }, + { + "epoch": 0.8043143297380585, + "grad_norm": 2.452697992324829, + "learning_rate": 4.094023601065855e-06, + "loss": 1.6575, + "step": 783 + }, + { + "epoch": 0.805341551104263, + "grad_norm": 2.4377169609069824, + "learning_rate": 4.092120289303388e-06, + "loss": 1.6016, + "step": 784 + }, + { + "epoch": 0.8063687724704673, + "grad_norm": 2.369154214859009, + "learning_rate": 4.0902169775409215e-06, + "loss": 1.6133, + "step": 785 + }, + { + "epoch": 0.8073959938366718, + "grad_norm": 2.261869430541992, + "learning_rate": 4.088313665778455e-06, + "loss": 1.5555, + "step": 786 + }, + { + "epoch": 0.8084232152028762, + "grad_norm": 2.7058632373809814, + "learning_rate": 4.086410354015988e-06, + "loss": 1.5957, + "step": 787 + }, + { + "epoch": 0.8094504365690807, + "grad_norm": 2.3907508850097656, + "learning_rate": 4.0845070422535216e-06, + "loss": 1.5676, + "step": 788 + }, + { + "epoch": 0.810477657935285, + "grad_norm": 2.510284900665283, + "learning_rate": 4.082603730491055e-06, + "loss": 1.5872, + "step": 789 + }, + { + "epoch": 0.8115048793014895, + "grad_norm": 2.590810775756836, + "learning_rate": 4.080700418728588e-06, + "loss": 1.6315, + "step": 790 + }, + { + "epoch": 0.8115048793014895, + "eval_loss": 1.9573780298233032, + "eval_runtime": 44.5677, + "eval_samples_per_second": 22.438, + "eval_steps_per_second": 0.942, + "step": 790 + }, + { + "epoch": 0.8125321006676939, + "grad_norm": 2.466134786605835, + "learning_rate": 4.078797106966122e-06, + "loss": 1.6018, + "step": 791 + }, + { + "epoch": 0.8135593220338984, + "grad_norm": 2.474653482437134, + "learning_rate": 4.076893795203655e-06, + "loss": 1.5101, + "step": 792 + }, + { + "epoch": 0.8145865434001027, + "grad_norm": 2.4463517665863037, + "learning_rate": 4.074990483441188e-06, + "loss": 1.5686, + "step": 793 + }, + { + "epoch": 0.8156137647663071, + "grad_norm": 2.5148985385894775, + "learning_rate": 4.073087171678721e-06, + "loss": 1.6544, + "step": 794 + }, + { + "epoch": 0.8166409861325116, + "grad_norm": 2.6183927059173584, + "learning_rate": 4.071183859916254e-06, + "loss": 1.6277, + "step": 795 + }, + { + "epoch": 0.8176682074987159, + "grad_norm": 2.438427686691284, + "learning_rate": 4.069280548153788e-06, + "loss": 1.6002, + "step": 796 + }, + { + "epoch": 0.8186954288649204, + "grad_norm": 2.662855863571167, + "learning_rate": 4.067377236391321e-06, + "loss": 1.5594, + "step": 797 + }, + { + "epoch": 0.8197226502311248, + "grad_norm": 2.64789080619812, + "learning_rate": 4.065473924628854e-06, + "loss": 1.6007, + "step": 798 + }, + { + "epoch": 0.8207498715973293, + "grad_norm": 2.5152997970581055, + "learning_rate": 4.063570612866388e-06, + "loss": 1.6366, + "step": 799 + }, + { + "epoch": 0.8217770929635336, + "grad_norm": 2.468672037124634, + "learning_rate": 4.061667301103921e-06, + "loss": 1.6313, + "step": 800 + }, + { + "epoch": 0.8217770929635336, + "eval_loss": 1.95821213722229, + "eval_runtime": 45.9362, + "eval_samples_per_second": 21.769, + "eval_steps_per_second": 0.914, + "step": 800 + }, + { + "epoch": 0.8228043143297381, + "grad_norm": 2.440307855606079, + "learning_rate": 4.0597639893414544e-06, + "loss": 1.5993, + "step": 801 + }, + { + "epoch": 0.8238315356959425, + "grad_norm": 2.6519880294799805, + "learning_rate": 4.057860677578988e-06, + "loss": 1.6106, + "step": 802 + }, + { + "epoch": 0.8248587570621468, + "grad_norm": 2.4224767684936523, + "learning_rate": 4.055957365816521e-06, + "loss": 1.6247, + "step": 803 + }, + { + "epoch": 0.8258859784283513, + "grad_norm": 2.346858263015747, + "learning_rate": 4.0540540540540545e-06, + "loss": 1.5699, + "step": 804 + }, + { + "epoch": 0.8269131997945557, + "grad_norm": 2.577711582183838, + "learning_rate": 4.052150742291588e-06, + "loss": 1.5665, + "step": 805 + }, + { + "epoch": 0.8279404211607602, + "grad_norm": 2.604501724243164, + "learning_rate": 4.050247430529121e-06, + "loss": 1.561, + "step": 806 + }, + { + "epoch": 0.8289676425269645, + "grad_norm": 2.534820795059204, + "learning_rate": 4.048344118766655e-06, + "loss": 1.6391, + "step": 807 + }, + { + "epoch": 0.829994863893169, + "grad_norm": 2.496765613555908, + "learning_rate": 4.046440807004188e-06, + "loss": 1.6338, + "step": 808 + }, + { + "epoch": 0.8310220852593734, + "grad_norm": 2.587185859680176, + "learning_rate": 4.044537495241721e-06, + "loss": 1.6009, + "step": 809 + }, + { + "epoch": 0.8320493066255779, + "grad_norm": 2.7336575984954834, + "learning_rate": 4.042634183479255e-06, + "loss": 1.5911, + "step": 810 + }, + { + "epoch": 0.8320493066255779, + "eval_loss": 1.951606273651123, + "eval_runtime": 42.4435, + "eval_samples_per_second": 23.561, + "eval_steps_per_second": 0.99, + "step": 810 + }, + { + "epoch": 0.8330765279917822, + "grad_norm": 2.5524845123291016, + "learning_rate": 4.040730871716787e-06, + "loss": 1.6406, + "step": 811 + }, + { + "epoch": 0.8341037493579867, + "grad_norm": 2.4861226081848145, + "learning_rate": 4.038827559954321e-06, + "loss": 1.6092, + "step": 812 + }, + { + "epoch": 0.8351309707241911, + "grad_norm": 2.668187141418457, + "learning_rate": 4.036924248191854e-06, + "loss": 1.6529, + "step": 813 + }, + { + "epoch": 0.8361581920903954, + "grad_norm": 2.5616471767425537, + "learning_rate": 4.035020936429387e-06, + "loss": 1.6253, + "step": 814 + }, + { + "epoch": 0.8371854134565999, + "grad_norm": 2.4261975288391113, + "learning_rate": 4.033117624666921e-06, + "loss": 1.5955, + "step": 815 + }, + { + "epoch": 0.8382126348228043, + "grad_norm": 2.4947080612182617, + "learning_rate": 4.031214312904454e-06, + "loss": 1.5686, + "step": 816 + }, + { + "epoch": 0.8392398561890088, + "grad_norm": 2.747781276702881, + "learning_rate": 4.029311001141987e-06, + "loss": 1.5642, + "step": 817 + }, + { + "epoch": 0.8402670775552131, + "grad_norm": 2.4762001037597656, + "learning_rate": 4.027407689379521e-06, + "loss": 1.5301, + "step": 818 + }, + { + "epoch": 0.8412942989214176, + "grad_norm": 2.6253550052642822, + "learning_rate": 4.025504377617054e-06, + "loss": 1.5885, + "step": 819 + }, + { + "epoch": 0.842321520287622, + "grad_norm": 2.6974875926971436, + "learning_rate": 4.0236010658545875e-06, + "loss": 1.5656, + "step": 820 + }, + { + "epoch": 0.842321520287622, + "eval_loss": 1.9420732259750366, + "eval_runtime": 46.9525, + "eval_samples_per_second": 21.298, + "eval_steps_per_second": 0.895, + "step": 820 + }, + { + "epoch": 0.8433487416538265, + "grad_norm": 2.638587474822998, + "learning_rate": 4.02169775409212e-06, + "loss": 1.6081, + "step": 821 + }, + { + "epoch": 0.8443759630200308, + "grad_norm": 2.963974714279175, + "learning_rate": 4.019794442329653e-06, + "loss": 1.6851, + "step": 822 + }, + { + "epoch": 0.8454031843862352, + "grad_norm": 2.4657533168792725, + "learning_rate": 4.0178911305671876e-06, + "loss": 1.5396, + "step": 823 + }, + { + "epoch": 0.8464304057524397, + "grad_norm": 2.5373876094818115, + "learning_rate": 4.015987818804721e-06, + "loss": 1.6186, + "step": 824 + }, + { + "epoch": 0.847457627118644, + "grad_norm": 2.5873284339904785, + "learning_rate": 4.014084507042254e-06, + "loss": 1.5618, + "step": 825 + }, + { + "epoch": 0.8484848484848485, + "grad_norm": 2.4308598041534424, + "learning_rate": 4.012181195279788e-06, + "loss": 1.5399, + "step": 826 + }, + { + "epoch": 0.8495120698510529, + "grad_norm": 2.5274946689605713, + "learning_rate": 4.010277883517321e-06, + "loss": 1.6375, + "step": 827 + }, + { + "epoch": 0.8505392912172574, + "grad_norm": 2.4370181560516357, + "learning_rate": 4.0083745717548535e-06, + "loss": 1.6313, + "step": 828 + }, + { + "epoch": 0.8515665125834617, + "grad_norm": 2.4524898529052734, + "learning_rate": 4.006471259992387e-06, + "loss": 1.5865, + "step": 829 + }, + { + "epoch": 0.8525937339496662, + "grad_norm": 2.5587961673736572, + "learning_rate": 4.00456794822992e-06, + "loss": 1.608, + "step": 830 + }, + { + "epoch": 0.8525937339496662, + "eval_loss": 1.94539475440979, + "eval_runtime": 49.4543, + "eval_samples_per_second": 20.221, + "eval_steps_per_second": 0.849, + "step": 830 + }, + { + "epoch": 0.8536209553158706, + "grad_norm": 2.512253761291504, + "learning_rate": 4.002664636467454e-06, + "loss": 1.6607, + "step": 831 + }, + { + "epoch": 0.8546481766820749, + "grad_norm": 2.438451051712036, + "learning_rate": 4.000761324704987e-06, + "loss": 1.6153, + "step": 832 + }, + { + "epoch": 0.8556753980482794, + "grad_norm": 2.5550496578216553, + "learning_rate": 3.99885801294252e-06, + "loss": 1.5489, + "step": 833 + }, + { + "epoch": 0.8567026194144838, + "grad_norm": 2.559732437133789, + "learning_rate": 3.996954701180054e-06, + "loss": 1.6045, + "step": 834 + }, + { + "epoch": 0.8577298407806883, + "grad_norm": 2.317692995071411, + "learning_rate": 3.995051389417587e-06, + "loss": 1.5316, + "step": 835 + }, + { + "epoch": 0.8587570621468926, + "grad_norm": 2.3742363452911377, + "learning_rate": 3.9931480776551204e-06, + "loss": 1.553, + "step": 836 + }, + { + "epoch": 0.8597842835130971, + "grad_norm": 2.5645551681518555, + "learning_rate": 3.991244765892654e-06, + "loss": 1.6558, + "step": 837 + }, + { + "epoch": 0.8608115048793015, + "grad_norm": 2.494830846786499, + "learning_rate": 3.989341454130186e-06, + "loss": 1.5713, + "step": 838 + }, + { + "epoch": 0.861838726245506, + "grad_norm": 2.416020393371582, + "learning_rate": 3.98743814236772e-06, + "loss": 1.5383, + "step": 839 + }, + { + "epoch": 0.8628659476117103, + "grad_norm": 2.6063015460968018, + "learning_rate": 3.985534830605253e-06, + "loss": 1.5831, + "step": 840 + }, + { + "epoch": 0.8628659476117103, + "eval_loss": 1.9441509246826172, + "eval_runtime": 46.0083, + "eval_samples_per_second": 21.735, + "eval_steps_per_second": 0.913, + "step": 840 + }, + { + "epoch": 0.8638931689779148, + "grad_norm": 2.5771663188934326, + "learning_rate": 3.983631518842786e-06, + "loss": 1.6077, + "step": 841 + }, + { + "epoch": 0.8649203903441192, + "grad_norm": 2.573119878768921, + "learning_rate": 3.98172820708032e-06, + "loss": 1.6301, + "step": 842 + }, + { + "epoch": 0.8659476117103235, + "grad_norm": 2.41396164894104, + "learning_rate": 3.979824895317854e-06, + "loss": 1.5758, + "step": 843 + }, + { + "epoch": 0.866974833076528, + "grad_norm": 2.532883882522583, + "learning_rate": 3.977921583555387e-06, + "loss": 1.5525, + "step": 844 + }, + { + "epoch": 0.8680020544427324, + "grad_norm": 2.669311285018921, + "learning_rate": 3.97601827179292e-06, + "loss": 1.6183, + "step": 845 + }, + { + "epoch": 0.8690292758089369, + "grad_norm": 2.366861343383789, + "learning_rate": 3.974114960030453e-06, + "loss": 1.5551, + "step": 846 + }, + { + "epoch": 0.8700564971751412, + "grad_norm": 2.3896048069000244, + "learning_rate": 3.9722116482679866e-06, + "loss": 1.5894, + "step": 847 + }, + { + "epoch": 0.8710837185413457, + "grad_norm": 2.6509673595428467, + "learning_rate": 3.97030833650552e-06, + "loss": 1.5813, + "step": 848 + }, + { + "epoch": 0.8721109399075501, + "grad_norm": 2.531726121902466, + "learning_rate": 3.968405024743053e-06, + "loss": 1.5899, + "step": 849 + }, + { + "epoch": 0.8731381612737545, + "grad_norm": 2.4117019176483154, + "learning_rate": 3.966501712980587e-06, + "loss": 1.5117, + "step": 850 + }, + { + "epoch": 0.8731381612737545, + "eval_loss": 1.9224250316619873, + "eval_runtime": 47.7854, + "eval_samples_per_second": 20.927, + "eval_steps_per_second": 0.879, + "step": 850 + }, + { + "epoch": 0.8741653826399589, + "grad_norm": 2.3789749145507812, + "learning_rate": 3.96459840121812e-06, + "loss": 1.5708, + "step": 851 + }, + { + "epoch": 0.8751926040061633, + "grad_norm": 2.386664867401123, + "learning_rate": 3.962695089455653e-06, + "loss": 1.5734, + "step": 852 + }, + { + "epoch": 0.8762198253723678, + "grad_norm": 2.4788997173309326, + "learning_rate": 3.960791777693187e-06, + "loss": 1.6315, + "step": 853 + }, + { + "epoch": 0.8772470467385721, + "grad_norm": 2.630183219909668, + "learning_rate": 3.95888846593072e-06, + "loss": 1.614, + "step": 854 + }, + { + "epoch": 0.8782742681047766, + "grad_norm": 2.3957736492156982, + "learning_rate": 3.956985154168253e-06, + "loss": 1.61, + "step": 855 + }, + { + "epoch": 0.879301489470981, + "grad_norm": 2.652036428451538, + "learning_rate": 3.955081842405786e-06, + "loss": 1.6463, + "step": 856 + }, + { + "epoch": 0.8803287108371854, + "grad_norm": 2.579867124557495, + "learning_rate": 3.953178530643319e-06, + "loss": 1.6624, + "step": 857 + }, + { + "epoch": 0.8813559322033898, + "grad_norm": 2.5118343830108643, + "learning_rate": 3.951275218880853e-06, + "loss": 1.6816, + "step": 858 + }, + { + "epoch": 0.8823831535695943, + "grad_norm": 2.5963029861450195, + "learning_rate": 3.949371907118386e-06, + "loss": 1.5296, + "step": 859 + }, + { + "epoch": 0.8834103749357987, + "grad_norm": 2.571241855621338, + "learning_rate": 3.9474685953559194e-06, + "loss": 1.6362, + "step": 860 + }, + { + "epoch": 0.8834103749357987, + "eval_loss": 1.931492567062378, + "eval_runtime": 52.2417, + "eval_samples_per_second": 19.142, + "eval_steps_per_second": 0.804, + "step": 860 + }, + { + "epoch": 0.884437596302003, + "grad_norm": 2.518036365509033, + "learning_rate": 3.945565283593453e-06, + "loss": 1.5865, + "step": 861 + }, + { + "epoch": 0.8854648176682075, + "grad_norm": 3.14587664604187, + "learning_rate": 3.943661971830986e-06, + "loss": 1.586, + "step": 862 + }, + { + "epoch": 0.8864920390344119, + "grad_norm": 2.686350107192993, + "learning_rate": 3.9417586600685195e-06, + "loss": 1.6385, + "step": 863 + }, + { + "epoch": 0.8875192604006163, + "grad_norm": 3.0324997901916504, + "learning_rate": 3.939855348306053e-06, + "loss": 1.6199, + "step": 864 + }, + { + "epoch": 0.8885464817668207, + "grad_norm": 2.6129438877105713, + "learning_rate": 3.937952036543586e-06, + "loss": 1.6049, + "step": 865 + }, + { + "epoch": 0.8895737031330252, + "grad_norm": 2.795729160308838, + "learning_rate": 3.93604872478112e-06, + "loss": 1.6027, + "step": 866 + }, + { + "epoch": 0.8906009244992296, + "grad_norm": 3.2352302074432373, + "learning_rate": 3.934145413018653e-06, + "loss": 1.6063, + "step": 867 + }, + { + "epoch": 0.891628145865434, + "grad_norm": 2.5050623416900635, + "learning_rate": 3.932242101256186e-06, + "loss": 1.5622, + "step": 868 + }, + { + "epoch": 0.8926553672316384, + "grad_norm": 2.8998777866363525, + "learning_rate": 3.93033878949372e-06, + "loss": 1.5729, + "step": 869 + }, + { + "epoch": 0.8936825885978429, + "grad_norm": 2.60239577293396, + "learning_rate": 3.928435477731253e-06, + "loss": 1.5377, + "step": 870 + }, + { + "epoch": 0.8936825885978429, + "eval_loss": 1.9311563968658447, + "eval_runtime": 46.8098, + "eval_samples_per_second": 21.363, + "eval_steps_per_second": 0.897, + "step": 870 + }, + { + "epoch": 0.8947098099640473, + "grad_norm": 2.5061111450195312, + "learning_rate": 3.9265321659687864e-06, + "loss": 1.6216, + "step": 871 + }, + { + "epoch": 0.8957370313302516, + "grad_norm": 2.7476489543914795, + "learning_rate": 3.924628854206319e-06, + "loss": 1.6357, + "step": 872 + }, + { + "epoch": 0.8967642526964561, + "grad_norm": 2.605536937713623, + "learning_rate": 3.922725542443852e-06, + "loss": 1.5983, + "step": 873 + }, + { + "epoch": 0.8977914740626605, + "grad_norm": 2.6024532318115234, + "learning_rate": 3.920822230681386e-06, + "loss": 1.5571, + "step": 874 + }, + { + "epoch": 0.8988186954288649, + "grad_norm": 2.7357826232910156, + "learning_rate": 3.918918918918919e-06, + "loss": 1.5802, + "step": 875 + }, + { + "epoch": 0.8998459167950693, + "grad_norm": 2.6406939029693604, + "learning_rate": 3.917015607156452e-06, + "loss": 1.6046, + "step": 876 + }, + { + "epoch": 0.9008731381612738, + "grad_norm": 2.4611780643463135, + "learning_rate": 3.915112295393986e-06, + "loss": 1.5803, + "step": 877 + }, + { + "epoch": 0.9019003595274782, + "grad_norm": 3.0608181953430176, + "learning_rate": 3.913208983631519e-06, + "loss": 1.6655, + "step": 878 + }, + { + "epoch": 0.9029275808936826, + "grad_norm": 2.482142448425293, + "learning_rate": 3.9113056718690525e-06, + "loss": 1.6444, + "step": 879 + }, + { + "epoch": 0.903954802259887, + "grad_norm": 2.4936728477478027, + "learning_rate": 3.909402360106586e-06, + "loss": 1.5596, + "step": 880 + }, + { + "epoch": 0.903954802259887, + "eval_loss": 1.933276653289795, + "eval_runtime": 48.0836, + "eval_samples_per_second": 20.797, + "eval_steps_per_second": 0.873, + "step": 880 + }, + { + "epoch": 0.9049820236260914, + "grad_norm": 2.7802248001098633, + "learning_rate": 3.907499048344119e-06, + "loss": 1.6133, + "step": 881 + }, + { + "epoch": 0.9060092449922958, + "grad_norm": 2.486586093902588, + "learning_rate": 3.905595736581652e-06, + "loss": 1.623, + "step": 882 + }, + { + "epoch": 0.9070364663585002, + "grad_norm": 2.477555751800537, + "learning_rate": 3.903692424819186e-06, + "loss": 1.6219, + "step": 883 + }, + { + "epoch": 0.9080636877247047, + "grad_norm": 2.5033347606658936, + "learning_rate": 3.901789113056719e-06, + "loss": 1.5612, + "step": 884 + }, + { + "epoch": 0.9090909090909091, + "grad_norm": 2.410464286804199, + "learning_rate": 3.899885801294253e-06, + "loss": 1.6243, + "step": 885 + }, + { + "epoch": 0.9101181304571135, + "grad_norm": 2.451402425765991, + "learning_rate": 3.897982489531786e-06, + "loss": 1.5961, + "step": 886 + }, + { + "epoch": 0.9111453518233179, + "grad_norm": 2.439730167388916, + "learning_rate": 3.896079177769319e-06, + "loss": 1.6471, + "step": 887 + }, + { + "epoch": 0.9121725731895224, + "grad_norm": 2.527719259262085, + "learning_rate": 3.894175866006853e-06, + "loss": 1.6262, + "step": 888 + }, + { + "epoch": 0.9131997945557268, + "grad_norm": 2.612992763519287, + "learning_rate": 3.892272554244385e-06, + "loss": 1.6533, + "step": 889 + }, + { + "epoch": 0.9142270159219312, + "grad_norm": 2.509989023208618, + "learning_rate": 3.890369242481919e-06, + "loss": 1.5596, + "step": 890 + }, + { + "epoch": 0.9142270159219312, + "eval_loss": 1.9224061965942383, + "eval_runtime": 45.5462, + "eval_samples_per_second": 21.956, + "eval_steps_per_second": 0.922, + "step": 890 + }, + { + "epoch": 0.9152542372881356, + "grad_norm": 2.3849823474884033, + "learning_rate": 3.888465930719452e-06, + "loss": 1.6336, + "step": 891 + }, + { + "epoch": 0.91628145865434, + "grad_norm": 2.4989709854125977, + "learning_rate": 3.886562618956985e-06, + "loss": 1.5733, + "step": 892 + }, + { + "epoch": 0.9173086800205444, + "grad_norm": 2.4675447940826416, + "learning_rate": 3.884659307194519e-06, + "loss": 1.5818, + "step": 893 + }, + { + "epoch": 0.9183359013867488, + "grad_norm": 2.3105645179748535, + "learning_rate": 3.882755995432052e-06, + "loss": 1.5251, + "step": 894 + }, + { + "epoch": 0.9193631227529533, + "grad_norm": 2.5410287380218506, + "learning_rate": 3.8808526836695854e-06, + "loss": 1.6089, + "step": 895 + }, + { + "epoch": 0.9203903441191577, + "grad_norm": 2.617091655731201, + "learning_rate": 3.878949371907119e-06, + "loss": 1.5631, + "step": 896 + }, + { + "epoch": 0.9214175654853621, + "grad_norm": 2.4979820251464844, + "learning_rate": 3.877046060144652e-06, + "loss": 1.5577, + "step": 897 + }, + { + "epoch": 0.9224447868515665, + "grad_norm": 2.4857399463653564, + "learning_rate": 3.8751427483821855e-06, + "loss": 1.56, + "step": 898 + }, + { + "epoch": 0.923472008217771, + "grad_norm": 2.456709146499634, + "learning_rate": 3.873239436619718e-06, + "loss": 1.594, + "step": 899 + }, + { + "epoch": 0.9244992295839753, + "grad_norm": 2.355168342590332, + "learning_rate": 3.871336124857251e-06, + "loss": 1.5371, + "step": 900 + }, + { + "epoch": 0.9244992295839753, + "eval_loss": 1.9357926845550537, + "eval_runtime": 44.4044, + "eval_samples_per_second": 22.52, + "eval_steps_per_second": 0.946, + "step": 900 + }, + { + "epoch": 0.9255264509501797, + "grad_norm": 2.5965569019317627, + "learning_rate": 3.869432813094785e-06, + "loss": 1.5966, + "step": 901 + }, + { + "epoch": 0.9265536723163842, + "grad_norm": 2.443662405014038, + "learning_rate": 3.867529501332318e-06, + "loss": 1.6053, + "step": 902 + }, + { + "epoch": 0.9275808936825886, + "grad_norm": 2.3888511657714844, + "learning_rate": 3.8656261895698515e-06, + "loss": 1.5921, + "step": 903 + }, + { + "epoch": 0.928608115048793, + "grad_norm": 2.4648637771606445, + "learning_rate": 3.863722877807386e-06, + "loss": 1.5879, + "step": 904 + }, + { + "epoch": 0.9296353364149974, + "grad_norm": 2.4636101722717285, + "learning_rate": 3.861819566044919e-06, + "loss": 1.6286, + "step": 905 + }, + { + "epoch": 0.9306625577812019, + "grad_norm": 2.722627878189087, + "learning_rate": 3.859916254282452e-06, + "loss": 1.6666, + "step": 906 + }, + { + "epoch": 0.9316897791474062, + "grad_norm": 2.550255298614502, + "learning_rate": 3.858012942519985e-06, + "loss": 1.5711, + "step": 907 + }, + { + "epoch": 0.9327170005136107, + "grad_norm": 2.510209798812866, + "learning_rate": 3.856109630757518e-06, + "loss": 1.5987, + "step": 908 + }, + { + "epoch": 0.9337442218798151, + "grad_norm": 2.6028876304626465, + "learning_rate": 3.854206318995052e-06, + "loss": 1.5682, + "step": 909 + }, + { + "epoch": 0.9347714432460195, + "grad_norm": 2.487908124923706, + "learning_rate": 3.852303007232585e-06, + "loss": 1.608, + "step": 910 + }, + { + "epoch": 0.9347714432460195, + "eval_loss": 1.93544602394104, + "eval_runtime": 41.5201, + "eval_samples_per_second": 24.085, + "eval_steps_per_second": 1.012, + "step": 910 + }, + { + "epoch": 0.9357986646122239, + "grad_norm": 2.4569835662841797, + "learning_rate": 3.850399695470118e-06, + "loss": 1.5492, + "step": 911 + }, + { + "epoch": 0.9368258859784283, + "grad_norm": 2.572781562805176, + "learning_rate": 3.848496383707652e-06, + "loss": 1.5633, + "step": 912 + }, + { + "epoch": 0.9378531073446328, + "grad_norm": 2.6213486194610596, + "learning_rate": 3.846593071945185e-06, + "loss": 1.5488, + "step": 913 + }, + { + "epoch": 0.9388803287108372, + "grad_norm": 2.6079821586608887, + "learning_rate": 3.8446897601827185e-06, + "loss": 1.5311, + "step": 914 + }, + { + "epoch": 0.9399075500770416, + "grad_norm": 2.6774539947509766, + "learning_rate": 3.842786448420252e-06, + "loss": 1.6564, + "step": 915 + }, + { + "epoch": 0.940934771443246, + "grad_norm": 2.5028746128082275, + "learning_rate": 3.840883136657785e-06, + "loss": 1.6551, + "step": 916 + }, + { + "epoch": 0.9419619928094505, + "grad_norm": 2.5254387855529785, + "learning_rate": 3.838979824895318e-06, + "loss": 1.6492, + "step": 917 + }, + { + "epoch": 0.9429892141756548, + "grad_norm": 2.5451207160949707, + "learning_rate": 3.837076513132851e-06, + "loss": 1.5306, + "step": 918 + }, + { + "epoch": 0.9440164355418593, + "grad_norm": 2.526526689529419, + "learning_rate": 3.8351732013703844e-06, + "loss": 1.6131, + "step": 919 + }, + { + "epoch": 0.9450436569080637, + "grad_norm": 2.454895257949829, + "learning_rate": 3.833269889607918e-06, + "loss": 1.5592, + "step": 920 + }, + { + "epoch": 0.9450436569080637, + "eval_loss": 1.9236185550689697, + "eval_runtime": 44.149, + "eval_samples_per_second": 22.651, + "eval_steps_per_second": 0.951, + "step": 920 + }, + { + "epoch": 0.9460708782742681, + "grad_norm": 2.6773581504821777, + "learning_rate": 3.831366577845451e-06, + "loss": 1.5498, + "step": 921 + }, + { + "epoch": 0.9470980996404725, + "grad_norm": 2.63554310798645, + "learning_rate": 3.8294632660829845e-06, + "loss": 1.5413, + "step": 922 + }, + { + "epoch": 0.9481253210066769, + "grad_norm": 2.7118968963623047, + "learning_rate": 3.827559954320518e-06, + "loss": 1.572, + "step": 923 + }, + { + "epoch": 0.9491525423728814, + "grad_norm": 2.4681396484375, + "learning_rate": 3.825656642558051e-06, + "loss": 1.6003, + "step": 924 + }, + { + "epoch": 0.9501797637390857, + "grad_norm": 2.387608051300049, + "learning_rate": 3.823753330795585e-06, + "loss": 1.6057, + "step": 925 + }, + { + "epoch": 0.9512069851052902, + "grad_norm": 2.426020383834839, + "learning_rate": 3.821850019033118e-06, + "loss": 1.5976, + "step": 926 + }, + { + "epoch": 0.9522342064714946, + "grad_norm": 2.446044445037842, + "learning_rate": 3.819946707270651e-06, + "loss": 1.5363, + "step": 927 + }, + { + "epoch": 0.9532614278376991, + "grad_norm": 2.4415299892425537, + "learning_rate": 3.818043395508185e-06, + "loss": 1.5685, + "step": 928 + }, + { + "epoch": 0.9542886492039034, + "grad_norm": 2.4006199836730957, + "learning_rate": 3.816140083745718e-06, + "loss": 1.5823, + "step": 929 + }, + { + "epoch": 0.9553158705701078, + "grad_norm": 2.6479294300079346, + "learning_rate": 3.8142367719832514e-06, + "loss": 1.6658, + "step": 930 + }, + { + "epoch": 0.9553158705701078, + "eval_loss": 1.9434653520584106, + "eval_runtime": 45.5063, + "eval_samples_per_second": 21.975, + "eval_steps_per_second": 0.923, + "step": 930 + }, + { + "epoch": 0.9563430919363123, + "grad_norm": 2.530367851257324, + "learning_rate": 3.8123334602207844e-06, + "loss": 1.544, + "step": 931 + }, + { + "epoch": 0.9573703133025167, + "grad_norm": 2.641665458679199, + "learning_rate": 3.8104301484583177e-06, + "loss": 1.5658, + "step": 932 + }, + { + "epoch": 0.9583975346687211, + "grad_norm": 2.3992276191711426, + "learning_rate": 3.808526836695851e-06, + "loss": 1.5524, + "step": 933 + }, + { + "epoch": 0.9594247560349255, + "grad_norm": 2.640503168106079, + "learning_rate": 3.8066235249333845e-06, + "loss": 1.6127, + "step": 934 + }, + { + "epoch": 0.96045197740113, + "grad_norm": 2.513735771179199, + "learning_rate": 3.804720213170918e-06, + "loss": 1.5901, + "step": 935 + }, + { + "epoch": 0.9614791987673343, + "grad_norm": 2.74481463432312, + "learning_rate": 3.8028169014084508e-06, + "loss": 1.5792, + "step": 936 + }, + { + "epoch": 0.9625064201335388, + "grad_norm": 2.5238802433013916, + "learning_rate": 3.800913589645984e-06, + "loss": 1.602, + "step": 937 + }, + { + "epoch": 0.9635336414997432, + "grad_norm": 2.5111966133117676, + "learning_rate": 3.7990102778835175e-06, + "loss": 1.607, + "step": 938 + }, + { + "epoch": 0.9645608628659477, + "grad_norm": 2.658449649810791, + "learning_rate": 3.797106966121051e-06, + "loss": 1.593, + "step": 939 + }, + { + "epoch": 0.965588084232152, + "grad_norm": 2.6884148120880127, + "learning_rate": 3.7952036543585842e-06, + "loss": 1.6576, + "step": 940 + }, + { + "epoch": 0.965588084232152, + "eval_loss": 1.9191738367080688, + "eval_runtime": 45.1928, + "eval_samples_per_second": 22.127, + "eval_steps_per_second": 0.929, + "step": 940 + }, + { + "epoch": 0.9666153055983564, + "grad_norm": 2.4238054752349854, + "learning_rate": 3.793300342596117e-06, + "loss": 1.6079, + "step": 941 + }, + { + "epoch": 0.9676425269645609, + "grad_norm": 2.505571126937866, + "learning_rate": 3.7913970308336505e-06, + "loss": 1.6299, + "step": 942 + }, + { + "epoch": 0.9686697483307652, + "grad_norm": 2.4143226146698, + "learning_rate": 3.789493719071184e-06, + "loss": 1.6158, + "step": 943 + }, + { + "epoch": 0.9696969696969697, + "grad_norm": 2.4664456844329834, + "learning_rate": 3.7875904073087177e-06, + "loss": 1.5672, + "step": 944 + }, + { + "epoch": 0.9707241910631741, + "grad_norm": 2.5686466693878174, + "learning_rate": 3.785687095546251e-06, + "loss": 1.5338, + "step": 945 + }, + { + "epoch": 0.9717514124293786, + "grad_norm": 2.4666805267333984, + "learning_rate": 3.7837837837837844e-06, + "loss": 1.5773, + "step": 946 + }, + { + "epoch": 0.9727786337955829, + "grad_norm": 2.6416635513305664, + "learning_rate": 3.7818804720213178e-06, + "loss": 1.5469, + "step": 947 + }, + { + "epoch": 0.9738058551617874, + "grad_norm": 2.721205711364746, + "learning_rate": 3.7799771602588507e-06, + "loss": 1.5278, + "step": 948 + }, + { + "epoch": 0.9748330765279918, + "grad_norm": 2.4834787845611572, + "learning_rate": 3.778073848496384e-06, + "loss": 1.52, + "step": 949 + }, + { + "epoch": 0.9758602978941961, + "grad_norm": 2.42047381401062, + "learning_rate": 3.7761705367339174e-06, + "loss": 1.5814, + "step": 950 + }, + { + "epoch": 0.9758602978941961, + "eval_loss": 1.9303041696548462, + "eval_runtime": 47.3843, + "eval_samples_per_second": 21.104, + "eval_steps_per_second": 0.886, + "step": 950 + }, + { + "epoch": 0.9768875192604006, + "grad_norm": 2.70369815826416, + "learning_rate": 3.7742672249714508e-06, + "loss": 1.6143, + "step": 951 + }, + { + "epoch": 0.977914740626605, + "grad_norm": 2.393588066101074, + "learning_rate": 3.772363913208984e-06, + "loss": 1.6023, + "step": 952 + }, + { + "epoch": 0.9789419619928095, + "grad_norm": 2.361304998397827, + "learning_rate": 3.770460601446517e-06, + "loss": 1.5679, + "step": 953 + }, + { + "epoch": 0.9799691833590138, + "grad_norm": 2.4044971466064453, + "learning_rate": 3.7685572896840504e-06, + "loss": 1.6007, + "step": 954 + }, + { + "epoch": 0.9809964047252183, + "grad_norm": 2.498539447784424, + "learning_rate": 3.766653977921584e-06, + "loss": 1.5429, + "step": 955 + }, + { + "epoch": 0.9820236260914227, + "grad_norm": 2.457162857055664, + "learning_rate": 3.764750666159117e-06, + "loss": 1.4861, + "step": 956 + }, + { + "epoch": 0.9830508474576272, + "grad_norm": 2.52756404876709, + "learning_rate": 3.7628473543966505e-06, + "loss": 1.5848, + "step": 957 + }, + { + "epoch": 0.9840780688238315, + "grad_norm": 2.503998279571533, + "learning_rate": 3.7609440426341835e-06, + "loss": 1.5435, + "step": 958 + }, + { + "epoch": 0.9851052901900359, + "grad_norm": 2.461087226867676, + "learning_rate": 3.759040730871717e-06, + "loss": 1.6344, + "step": 959 + }, + { + "epoch": 0.9861325115562404, + "grad_norm": 2.4149672985076904, + "learning_rate": 3.75713741910925e-06, + "loss": 1.5273, + "step": 960 + }, + { + "epoch": 0.9861325115562404, + "eval_loss": 1.928680658340454, + "eval_runtime": 48.7892, + "eval_samples_per_second": 20.496, + "eval_steps_per_second": 0.861, + "step": 960 + }, + { + "epoch": 0.9871597329224447, + "grad_norm": 2.5741734504699707, + "learning_rate": 3.7552341073467836e-06, + "loss": 1.5244, + "step": 961 + }, + { + "epoch": 0.9881869542886492, + "grad_norm": 2.5517220497131348, + "learning_rate": 3.753330795584317e-06, + "loss": 1.5695, + "step": 962 + }, + { + "epoch": 0.9892141756548536, + "grad_norm": 2.5910379886627197, + "learning_rate": 3.75142748382185e-06, + "loss": 1.6322, + "step": 963 + }, + { + "epoch": 0.9902413970210581, + "grad_norm": 2.4780521392822266, + "learning_rate": 3.749524172059384e-06, + "loss": 1.5292, + "step": 964 + }, + { + "epoch": 0.9912686183872624, + "grad_norm": 2.6141762733459473, + "learning_rate": 3.747620860296917e-06, + "loss": 1.5809, + "step": 965 + }, + { + "epoch": 0.9922958397534669, + "grad_norm": 2.467606544494629, + "learning_rate": 3.7457175485344504e-06, + "loss": 1.6123, + "step": 966 + }, + { + "epoch": 0.9933230611196713, + "grad_norm": 2.6112184524536133, + "learning_rate": 3.7438142367719837e-06, + "loss": 1.6226, + "step": 967 + }, + { + "epoch": 0.9943502824858758, + "grad_norm": 2.352513313293457, + "learning_rate": 3.741910925009517e-06, + "loss": 1.5928, + "step": 968 + }, + { + "epoch": 0.9953775038520801, + "grad_norm": 2.4357361793518066, + "learning_rate": 3.7400076132470505e-06, + "loss": 1.5484, + "step": 969 + }, + { + "epoch": 0.9964047252182845, + "grad_norm": 2.5168254375457764, + "learning_rate": 3.7381043014845834e-06, + "loss": 1.5722, + "step": 970 + }, + { + "epoch": 0.9964047252182845, + "eval_loss": 1.929540753364563, + "eval_runtime": 49.7232, + "eval_samples_per_second": 20.111, + "eval_steps_per_second": 0.845, + "step": 970 + }, + { + "epoch": 0.997431946584489, + "grad_norm": 2.475173234939575, + "learning_rate": 3.7362009897221168e-06, + "loss": 1.5952, + "step": 971 + }, + { + "epoch": 0.9984591679506933, + "grad_norm": 2.5278515815734863, + "learning_rate": 3.73429767795965e-06, + "loss": 1.5769, + "step": 972 + }, + { + "epoch": 0.9994863893168978, + "grad_norm": 2.4988675117492676, + "learning_rate": 3.7323943661971835e-06, + "loss": 1.5912, + "step": 973 + }, + { + "epoch": 1.0005136106831023, + "grad_norm": 2.406353712081909, + "learning_rate": 3.730491054434717e-06, + "loss": 1.5022, + "step": 974 + }, + { + "epoch": 1.0015408320493067, + "grad_norm": 2.611318349838257, + "learning_rate": 3.72858774267225e-06, + "loss": 1.4633, + "step": 975 + }, + { + "epoch": 1.002568053415511, + "grad_norm": 2.504613161087036, + "learning_rate": 3.726684430909783e-06, + "loss": 1.5405, + "step": 976 + }, + { + "epoch": 1.0035952747817154, + "grad_norm": 2.6135828495025635, + "learning_rate": 3.7247811191473165e-06, + "loss": 1.5756, + "step": 977 + }, + { + "epoch": 1.00462249614792, + "grad_norm": 2.692592144012451, + "learning_rate": 3.72287780738485e-06, + "loss": 1.5777, + "step": 978 + }, + { + "epoch": 1.0056497175141244, + "grad_norm": 2.660323143005371, + "learning_rate": 3.7209744956223832e-06, + "loss": 1.5488, + "step": 979 + }, + { + "epoch": 1.0066769388803287, + "grad_norm": 2.647883415222168, + "learning_rate": 3.719071183859916e-06, + "loss": 1.5619, + "step": 980 + }, + { + "epoch": 1.0066769388803287, + "eval_loss": 1.9224871397018433, + "eval_runtime": 52.6149, + "eval_samples_per_second": 19.006, + "eval_steps_per_second": 0.798, + "step": 980 + }, + { + "epoch": 1.007704160246533, + "grad_norm": 2.6186633110046387, + "learning_rate": 3.7171678720974495e-06, + "loss": 1.492, + "step": 981 + }, + { + "epoch": 1.0087313816127375, + "grad_norm": 2.7562549114227295, + "learning_rate": 3.715264560334983e-06, + "loss": 1.5932, + "step": 982 + }, + { + "epoch": 1.009758602978942, + "grad_norm": 2.618856430053711, + "learning_rate": 3.7133612485725163e-06, + "loss": 1.5514, + "step": 983 + }, + { + "epoch": 1.0107858243451464, + "grad_norm": 2.522064685821533, + "learning_rate": 3.71145793681005e-06, + "loss": 1.5699, + "step": 984 + }, + { + "epoch": 1.0118130457113508, + "grad_norm": 2.627108335494995, + "learning_rate": 3.7095546250475834e-06, + "loss": 1.5033, + "step": 985 + }, + { + "epoch": 1.0128402670775551, + "grad_norm": 2.6393308639526367, + "learning_rate": 3.7076513132851168e-06, + "loss": 1.6064, + "step": 986 + }, + { + "epoch": 1.0138674884437597, + "grad_norm": 2.560675621032715, + "learning_rate": 3.7057480015226497e-06, + "loss": 1.5167, + "step": 987 + }, + { + "epoch": 1.014894709809964, + "grad_norm": 2.4677579402923584, + "learning_rate": 3.703844689760183e-06, + "loss": 1.4953, + "step": 988 + }, + { + "epoch": 1.0159219311761685, + "grad_norm": 2.9904487133026123, + "learning_rate": 3.7019413779977164e-06, + "loss": 1.5786, + "step": 989 + }, + { + "epoch": 1.0169491525423728, + "grad_norm": 2.7342119216918945, + "learning_rate": 3.70003806623525e-06, + "loss": 1.5459, + "step": 990 + }, + { + "epoch": 1.0169491525423728, + "eval_loss": 1.929221510887146, + "eval_runtime": 45.9704, + "eval_samples_per_second": 21.753, + "eval_steps_per_second": 0.914, + "step": 990 + }, + { + "epoch": 1.0179763739085772, + "grad_norm": 2.723241090774536, + "learning_rate": 3.698134754472783e-06, + "loss": 1.5317, + "step": 991 + }, + { + "epoch": 1.0190035952747818, + "grad_norm": 2.562356948852539, + "learning_rate": 3.696231442710316e-06, + "loss": 1.5957, + "step": 992 + }, + { + "epoch": 1.0200308166409862, + "grad_norm": 2.635737419128418, + "learning_rate": 3.6943281309478495e-06, + "loss": 1.5404, + "step": 993 + }, + { + "epoch": 1.0210580380071905, + "grad_norm": 2.588714361190796, + "learning_rate": 3.692424819185383e-06, + "loss": 1.5618, + "step": 994 + }, + { + "epoch": 1.022085259373395, + "grad_norm": 2.61643385887146, + "learning_rate": 3.690521507422916e-06, + "loss": 1.5625, + "step": 995 + }, + { + "epoch": 1.0231124807395995, + "grad_norm": 2.566077470779419, + "learning_rate": 3.6886181956604496e-06, + "loss": 1.5051, + "step": 996 + }, + { + "epoch": 1.0241397021058039, + "grad_norm": 2.5667169094085693, + "learning_rate": 3.686714883897983e-06, + "loss": 1.5142, + "step": 997 + }, + { + "epoch": 1.0251669234720082, + "grad_norm": 2.839078903198242, + "learning_rate": 3.684811572135516e-06, + "loss": 1.5833, + "step": 998 + }, + { + "epoch": 1.0261941448382126, + "grad_norm": 2.651643753051758, + "learning_rate": 3.6829082603730492e-06, + "loss": 1.5735, + "step": 999 + }, + { + "epoch": 1.027221366204417, + "grad_norm": 2.6310813426971436, + "learning_rate": 3.6810049486105826e-06, + "loss": 1.5412, + "step": 1000 + }, + { + "epoch": 1.027221366204417, + "eval_loss": 1.9219006299972534, + "eval_runtime": 53.0111, + "eval_samples_per_second": 18.864, + "eval_steps_per_second": 0.792, + "step": 1000 + }, + { + "epoch": 1.0282485875706215, + "grad_norm": 2.721156120300293, + "learning_rate": 3.679101636848116e-06, + "loss": 1.5686, + "step": 1001 + }, + { + "epoch": 1.029275808936826, + "grad_norm": 2.8999130725860596, + "learning_rate": 3.6771983250856493e-06, + "loss": 1.5699, + "step": 1002 + }, + { + "epoch": 1.0303030303030303, + "grad_norm": 2.8879904747009277, + "learning_rate": 3.6752950133231822e-06, + "loss": 1.5353, + "step": 1003 + }, + { + "epoch": 1.0313302516692346, + "grad_norm": 2.4753830432891846, + "learning_rate": 3.6733917015607156e-06, + "loss": 1.5684, + "step": 1004 + }, + { + "epoch": 1.0323574730354392, + "grad_norm": 2.6306774616241455, + "learning_rate": 3.6714883897982494e-06, + "loss": 1.5741, + "step": 1005 + }, + { + "epoch": 1.0333846944016436, + "grad_norm": 2.8131468296051025, + "learning_rate": 3.6695850780357828e-06, + "loss": 1.5127, + "step": 1006 + }, + { + "epoch": 1.034411915767848, + "grad_norm": 2.6741905212402344, + "learning_rate": 3.667681766273316e-06, + "loss": 1.5135, + "step": 1007 + }, + { + "epoch": 1.0354391371340523, + "grad_norm": 2.6110734939575195, + "learning_rate": 3.6657784545108495e-06, + "loss": 1.5287, + "step": 1008 + }, + { + "epoch": 1.0364663585002567, + "grad_norm": 2.6415274143218994, + "learning_rate": 3.6638751427483824e-06, + "loss": 1.5978, + "step": 1009 + }, + { + "epoch": 1.0374935798664613, + "grad_norm": 2.5623817443847656, + "learning_rate": 3.6619718309859158e-06, + "loss": 1.5586, + "step": 1010 + }, + { + "epoch": 1.0374935798664613, + "eval_loss": 1.915805459022522, + "eval_runtime": 43.3255, + "eval_samples_per_second": 23.081, + "eval_steps_per_second": 0.969, + "step": 1010 + }, + { + "epoch": 1.0385208012326657, + "grad_norm": 2.723954916000366, + "learning_rate": 3.660068519223449e-06, + "loss": 1.6013, + "step": 1011 + }, + { + "epoch": 1.03954802259887, + "grad_norm": 2.6564769744873047, + "learning_rate": 3.6581652074609825e-06, + "loss": 1.5653, + "step": 1012 + }, + { + "epoch": 1.0405752439650744, + "grad_norm": 2.6694817543029785, + "learning_rate": 3.656261895698516e-06, + "loss": 1.5297, + "step": 1013 + }, + { + "epoch": 1.041602465331279, + "grad_norm": 2.546647071838379, + "learning_rate": 3.6543585839360492e-06, + "loss": 1.5192, + "step": 1014 + }, + { + "epoch": 1.0426296866974833, + "grad_norm": 2.691000461578369, + "learning_rate": 3.652455272173582e-06, + "loss": 1.5284, + "step": 1015 + }, + { + "epoch": 1.0436569080636877, + "grad_norm": 2.627333641052246, + "learning_rate": 3.6505519604111155e-06, + "loss": 1.5239, + "step": 1016 + }, + { + "epoch": 1.044684129429892, + "grad_norm": 2.6125097274780273, + "learning_rate": 3.648648648648649e-06, + "loss": 1.5357, + "step": 1017 + }, + { + "epoch": 1.0457113507960965, + "grad_norm": 2.6235790252685547, + "learning_rate": 3.6467453368861823e-06, + "loss": 1.5831, + "step": 1018 + }, + { + "epoch": 1.046738572162301, + "grad_norm": 2.5727598667144775, + "learning_rate": 3.6448420251237156e-06, + "loss": 1.5264, + "step": 1019 + }, + { + "epoch": 1.0477657935285054, + "grad_norm": 2.71970534324646, + "learning_rate": 3.6429387133612486e-06, + "loss": 1.5593, + "step": 1020 + }, + { + "epoch": 1.0477657935285054, + "eval_loss": 1.9139868021011353, + "eval_runtime": 46.8291, + "eval_samples_per_second": 21.354, + "eval_steps_per_second": 0.897, + "step": 1020 + }, + { + "epoch": 1.0487930148947098, + "grad_norm": 2.6479101181030273, + "learning_rate": 3.641035401598782e-06, + "loss": 1.5875, + "step": 1021 + }, + { + "epoch": 1.0498202362609141, + "grad_norm": 2.6313018798828125, + "learning_rate": 3.6391320898363153e-06, + "loss": 1.5222, + "step": 1022 + }, + { + "epoch": 1.0508474576271187, + "grad_norm": 2.679009437561035, + "learning_rate": 3.6372287780738487e-06, + "loss": 1.5678, + "step": 1023 + }, + { + "epoch": 1.051874678993323, + "grad_norm": 2.8511667251586914, + "learning_rate": 3.635325466311382e-06, + "loss": 1.5401, + "step": 1024 + }, + { + "epoch": 1.0529019003595275, + "grad_norm": 2.717280864715576, + "learning_rate": 3.633422154548916e-06, + "loss": 1.5686, + "step": 1025 + }, + { + "epoch": 1.0539291217257318, + "grad_norm": 2.6042356491088867, + "learning_rate": 3.6315188427864487e-06, + "loss": 1.5945, + "step": 1026 + }, + { + "epoch": 1.0549563430919364, + "grad_norm": 2.601451873779297, + "learning_rate": 3.629615531023982e-06, + "loss": 1.556, + "step": 1027 + }, + { + "epoch": 1.0559835644581408, + "grad_norm": 2.5195374488830566, + "learning_rate": 3.6277122192615155e-06, + "loss": 1.5026, + "step": 1028 + }, + { + "epoch": 1.0570107858243452, + "grad_norm": 2.604405164718628, + "learning_rate": 3.625808907499049e-06, + "loss": 1.5629, + "step": 1029 + }, + { + "epoch": 1.0580380071905495, + "grad_norm": 2.6300103664398193, + "learning_rate": 3.623905595736582e-06, + "loss": 1.5253, + "step": 1030 + }, + { + "epoch": 1.0580380071905495, + "eval_loss": 1.9272329807281494, + "eval_runtime": 46.1105, + "eval_samples_per_second": 21.687, + "eval_steps_per_second": 0.911, + "step": 1030 + }, + { + "epoch": 1.059065228556754, + "grad_norm": 2.663590908050537, + "learning_rate": 3.6220022839741156e-06, + "loss": 1.5223, + "step": 1031 + }, + { + "epoch": 1.0600924499229585, + "grad_norm": 2.773226499557495, + "learning_rate": 3.6200989722116485e-06, + "loss": 1.5052, + "step": 1032 + }, + { + "epoch": 1.0611196712891628, + "grad_norm": 2.746915817260742, + "learning_rate": 3.618195660449182e-06, + "loss": 1.5371, + "step": 1033 + }, + { + "epoch": 1.0621468926553672, + "grad_norm": 3.0018815994262695, + "learning_rate": 3.6162923486867152e-06, + "loss": 1.5494, + "step": 1034 + }, + { + "epoch": 1.0631741140215716, + "grad_norm": 2.77616810798645, + "learning_rate": 3.6143890369242486e-06, + "loss": 1.4878, + "step": 1035 + }, + { + "epoch": 1.064201335387776, + "grad_norm": 2.638270854949951, + "learning_rate": 3.612485725161782e-06, + "loss": 1.5028, + "step": 1036 + }, + { + "epoch": 1.0652285567539805, + "grad_norm": 2.7400267124176025, + "learning_rate": 3.610582413399315e-06, + "loss": 1.5364, + "step": 1037 + }, + { + "epoch": 1.066255778120185, + "grad_norm": 2.603987455368042, + "learning_rate": 3.6086791016368482e-06, + "loss": 1.5259, + "step": 1038 + }, + { + "epoch": 1.0672829994863893, + "grad_norm": 2.5326781272888184, + "learning_rate": 3.6067757898743816e-06, + "loss": 1.5346, + "step": 1039 + }, + { + "epoch": 1.0683102208525936, + "grad_norm": 2.677980422973633, + "learning_rate": 3.604872478111915e-06, + "loss": 1.6011, + "step": 1040 + }, + { + "epoch": 1.0683102208525936, + "eval_loss": 1.9431331157684326, + "eval_runtime": 46.6449, + "eval_samples_per_second": 21.439, + "eval_steps_per_second": 0.9, + "step": 1040 + }, + { + "epoch": 1.0693374422187982, + "grad_norm": 2.7540171146392822, + "learning_rate": 3.6029691663494483e-06, + "loss": 1.5988, + "step": 1041 + }, + { + "epoch": 1.0703646635850026, + "grad_norm": 2.6210403442382812, + "learning_rate": 3.6010658545869813e-06, + "loss": 1.5373, + "step": 1042 + }, + { + "epoch": 1.071391884951207, + "grad_norm": 2.7578794956207275, + "learning_rate": 3.5991625428245146e-06, + "loss": 1.5707, + "step": 1043 + }, + { + "epoch": 1.0724191063174113, + "grad_norm": 2.787123203277588, + "learning_rate": 3.597259231062048e-06, + "loss": 1.5666, + "step": 1044 + }, + { + "epoch": 1.073446327683616, + "grad_norm": 2.5755369663238525, + "learning_rate": 3.5953559192995818e-06, + "loss": 1.5005, + "step": 1045 + }, + { + "epoch": 1.0744735490498203, + "grad_norm": 2.644008159637451, + "learning_rate": 3.593452607537115e-06, + "loss": 1.5408, + "step": 1046 + }, + { + "epoch": 1.0755007704160247, + "grad_norm": 2.5818886756896973, + "learning_rate": 3.5915492957746485e-06, + "loss": 1.6064, + "step": 1047 + }, + { + "epoch": 1.076527991782229, + "grad_norm": 2.6413795948028564, + "learning_rate": 3.589645984012182e-06, + "loss": 1.5481, + "step": 1048 + }, + { + "epoch": 1.0775552131484334, + "grad_norm": 2.8570823669433594, + "learning_rate": 3.587742672249715e-06, + "loss": 1.5713, + "step": 1049 + }, + { + "epoch": 1.078582434514638, + "grad_norm": 2.6338746547698975, + "learning_rate": 3.585839360487248e-06, + "loss": 1.5252, + "step": 1050 + }, + { + "epoch": 1.078582434514638, + "eval_loss": 1.9467660188674927, + "eval_runtime": 45.8203, + "eval_samples_per_second": 21.824, + "eval_steps_per_second": 0.917, + "step": 1050 + }, + { + "epoch": 1.0796096558808423, + "grad_norm": 2.7295382022857666, + "learning_rate": 3.5839360487247815e-06, + "loss": 1.515, + "step": 1051 + }, + { + "epoch": 1.0806368772470467, + "grad_norm": 2.7476232051849365, + "learning_rate": 3.582032736962315e-06, + "loss": 1.5301, + "step": 1052 + }, + { + "epoch": 1.081664098613251, + "grad_norm": 2.8701605796813965, + "learning_rate": 3.5801294251998483e-06, + "loss": 1.5971, + "step": 1053 + }, + { + "epoch": 1.0826913199794557, + "grad_norm": 2.638195276260376, + "learning_rate": 3.578226113437381e-06, + "loss": 1.4917, + "step": 1054 + }, + { + "epoch": 1.08371854134566, + "grad_norm": 2.587324857711792, + "learning_rate": 3.5763228016749146e-06, + "loss": 1.5349, + "step": 1055 + }, + { + "epoch": 1.0847457627118644, + "grad_norm": 2.645148754119873, + "learning_rate": 3.574419489912448e-06, + "loss": 1.6063, + "step": 1056 + }, + { + "epoch": 1.0857729840780688, + "grad_norm": 2.580125331878662, + "learning_rate": 3.5725161781499813e-06, + "loss": 1.5504, + "step": 1057 + }, + { + "epoch": 1.0868002054442731, + "grad_norm": 2.730672836303711, + "learning_rate": 3.5706128663875147e-06, + "loss": 1.5272, + "step": 1058 + }, + { + "epoch": 1.0878274268104777, + "grad_norm": 2.564277172088623, + "learning_rate": 3.5687095546250476e-06, + "loss": 1.5471, + "step": 1059 + }, + { + "epoch": 1.088854648176682, + "grad_norm": 2.603226661682129, + "learning_rate": 3.566806242862581e-06, + "loss": 1.5759, + "step": 1060 + }, + { + "epoch": 1.088854648176682, + "eval_loss": 1.9485995769500732, + "eval_runtime": 45.476, + "eval_samples_per_second": 21.99, + "eval_steps_per_second": 0.924, + "step": 1060 + }, + { + "epoch": 1.0898818695428865, + "grad_norm": 2.7261931896209717, + "learning_rate": 3.5649029311001143e-06, + "loss": 1.5455, + "step": 1061 + }, + { + "epoch": 1.0909090909090908, + "grad_norm": 2.5638797283172607, + "learning_rate": 3.5629996193376477e-06, + "loss": 1.5236, + "step": 1062 + }, + { + "epoch": 1.0919363122752954, + "grad_norm": 2.8362395763397217, + "learning_rate": 3.561096307575181e-06, + "loss": 1.5376, + "step": 1063 + }, + { + "epoch": 1.0929635336414998, + "grad_norm": 2.6815669536590576, + "learning_rate": 3.559192995812714e-06, + "loss": 1.5681, + "step": 1064 + }, + { + "epoch": 1.0939907550077042, + "grad_norm": 2.691917896270752, + "learning_rate": 3.557289684050248e-06, + "loss": 1.5868, + "step": 1065 + }, + { + "epoch": 1.0950179763739085, + "grad_norm": 2.8439927101135254, + "learning_rate": 3.555386372287781e-06, + "loss": 1.5763, + "step": 1066 + }, + { + "epoch": 1.0960451977401129, + "grad_norm": 2.6857359409332275, + "learning_rate": 3.5534830605253145e-06, + "loss": 1.5784, + "step": 1067 + }, + { + "epoch": 1.0970724191063175, + "grad_norm": 2.532336950302124, + "learning_rate": 3.551579748762848e-06, + "loss": 1.5409, + "step": 1068 + }, + { + "epoch": 1.0980996404725218, + "grad_norm": 2.6567013263702393, + "learning_rate": 3.5496764370003812e-06, + "loss": 1.5484, + "step": 1069 + }, + { + "epoch": 1.0991268618387262, + "grad_norm": 2.800318717956543, + "learning_rate": 3.5477731252379146e-06, + "loss": 1.5247, + "step": 1070 + }, + { + "epoch": 1.0991268618387262, + "eval_loss": 1.9268279075622559, + "eval_runtime": 45.4409, + "eval_samples_per_second": 22.007, + "eval_steps_per_second": 0.924, + "step": 1070 + }, + { + "epoch": 1.1001540832049306, + "grad_norm": 2.6051485538482666, + "learning_rate": 3.5458698134754475e-06, + "loss": 1.5002, + "step": 1071 + }, + { + "epoch": 1.1011813045711352, + "grad_norm": 2.590815544128418, + "learning_rate": 3.543966501712981e-06, + "loss": 1.4982, + "step": 1072 + }, + { + "epoch": 1.1022085259373395, + "grad_norm": 2.6754868030548096, + "learning_rate": 3.5420631899505142e-06, + "loss": 1.5303, + "step": 1073 + }, + { + "epoch": 1.103235747303544, + "grad_norm": 2.8132805824279785, + "learning_rate": 3.5401598781880476e-06, + "loss": 1.6188, + "step": 1074 + }, + { + "epoch": 1.1042629686697483, + "grad_norm": 2.6267812252044678, + "learning_rate": 3.538256566425581e-06, + "loss": 1.442, + "step": 1075 + }, + { + "epoch": 1.1052901900359529, + "grad_norm": 2.945460557937622, + "learning_rate": 3.536353254663114e-06, + "loss": 1.5787, + "step": 1076 + }, + { + "epoch": 1.1063174114021572, + "grad_norm": 2.831702709197998, + "learning_rate": 3.5344499429006473e-06, + "loss": 1.5396, + "step": 1077 + }, + { + "epoch": 1.1073446327683616, + "grad_norm": 2.614482879638672, + "learning_rate": 3.5325466311381806e-06, + "loss": 1.5101, + "step": 1078 + }, + { + "epoch": 1.108371854134566, + "grad_norm": 2.809013605117798, + "learning_rate": 3.530643319375714e-06, + "loss": 1.5678, + "step": 1079 + }, + { + "epoch": 1.1093990755007703, + "grad_norm": 2.6645243167877197, + "learning_rate": 3.5287400076132474e-06, + "loss": 1.4762, + "step": 1080 + }, + { + "epoch": 1.1093990755007703, + "eval_loss": 1.9386240243911743, + "eval_runtime": 43.5259, + "eval_samples_per_second": 22.975, + "eval_steps_per_second": 0.965, + "step": 1080 + }, + { + "epoch": 1.110426296866975, + "grad_norm": 2.5597710609436035, + "learning_rate": 3.5268366958507803e-06, + "loss": 1.5208, + "step": 1081 + }, + { + "epoch": 1.1114535182331793, + "grad_norm": 2.6656508445739746, + "learning_rate": 3.5249333840883137e-06, + "loss": 1.5548, + "step": 1082 + }, + { + "epoch": 1.1124807395993837, + "grad_norm": 2.651195526123047, + "learning_rate": 3.523030072325847e-06, + "loss": 1.546, + "step": 1083 + }, + { + "epoch": 1.113507960965588, + "grad_norm": 2.946263551712036, + "learning_rate": 3.5211267605633804e-06, + "loss": 1.5058, + "step": 1084 + }, + { + "epoch": 1.1145351823317924, + "grad_norm": 2.7462689876556396, + "learning_rate": 3.519223448800914e-06, + "loss": 1.5213, + "step": 1085 + }, + { + "epoch": 1.115562403697997, + "grad_norm": 2.803926944732666, + "learning_rate": 3.5173201370384475e-06, + "loss": 1.5266, + "step": 1086 + }, + { + "epoch": 1.1165896250642013, + "grad_norm": 2.6583237648010254, + "learning_rate": 3.515416825275981e-06, + "loss": 1.5362, + "step": 1087 + }, + { + "epoch": 1.1176168464304057, + "grad_norm": 2.659313201904297, + "learning_rate": 3.513513513513514e-06, + "loss": 1.5365, + "step": 1088 + }, + { + "epoch": 1.11864406779661, + "grad_norm": 2.7181215286254883, + "learning_rate": 3.511610201751047e-06, + "loss": 1.5584, + "step": 1089 + }, + { + "epoch": 1.1196712891628147, + "grad_norm": 2.543297052383423, + "learning_rate": 3.5097068899885806e-06, + "loss": 1.4599, + "step": 1090 + }, + { + "epoch": 1.1196712891628147, + "eval_loss": 1.9393529891967773, + "eval_runtime": 44.5005, + "eval_samples_per_second": 22.472, + "eval_steps_per_second": 0.944, + "step": 1090 + }, + { + "epoch": 1.120698510529019, + "grad_norm": 2.8569319248199463, + "learning_rate": 3.507803578226114e-06, + "loss": 1.5308, + "step": 1091 + }, + { + "epoch": 1.1217257318952234, + "grad_norm": 2.795915365219116, + "learning_rate": 3.5059002664636473e-06, + "loss": 1.6092, + "step": 1092 + }, + { + "epoch": 1.1227529532614278, + "grad_norm": 2.659125566482544, + "learning_rate": 3.5039969547011802e-06, + "loss": 1.4701, + "step": 1093 + }, + { + "epoch": 1.1237801746276324, + "grad_norm": 2.7170166969299316, + "learning_rate": 3.5020936429387136e-06, + "loss": 1.5232, + "step": 1094 + }, + { + "epoch": 1.1248073959938367, + "grad_norm": 2.6485073566436768, + "learning_rate": 3.500190331176247e-06, + "loss": 1.5274, + "step": 1095 + }, + { + "epoch": 1.125834617360041, + "grad_norm": 2.831904888153076, + "learning_rate": 3.4982870194137803e-06, + "loss": 1.5833, + "step": 1096 + }, + { + "epoch": 1.1268618387262455, + "grad_norm": 2.877516984939575, + "learning_rate": 3.4963837076513137e-06, + "loss": 1.5173, + "step": 1097 + }, + { + "epoch": 1.1278890600924498, + "grad_norm": 2.5502521991729736, + "learning_rate": 3.4944803958888466e-06, + "loss": 1.5119, + "step": 1098 + }, + { + "epoch": 1.1289162814586544, + "grad_norm": 2.530945301055908, + "learning_rate": 3.49257708412638e-06, + "loss": 1.4573, + "step": 1099 + }, + { + "epoch": 1.1299435028248588, + "grad_norm": 2.7266969680786133, + "learning_rate": 3.4906737723639133e-06, + "loss": 1.5812, + "step": 1100 + }, + { + "epoch": 1.1299435028248588, + "eval_loss": 1.9370596408843994, + "eval_runtime": 47.2327, + "eval_samples_per_second": 21.172, + "eval_steps_per_second": 0.889, + "step": 1100 + }, + { + "epoch": 1.1309707241910631, + "grad_norm": 2.8846628665924072, + "learning_rate": 3.4887704606014467e-06, + "loss": 1.5945, + "step": 1101 + }, + { + "epoch": 1.1319979455572675, + "grad_norm": 2.5253827571868896, + "learning_rate": 3.48686714883898e-06, + "loss": 1.5391, + "step": 1102 + }, + { + "epoch": 1.133025166923472, + "grad_norm": 2.6929149627685547, + "learning_rate": 3.484963837076513e-06, + "loss": 1.5767, + "step": 1103 + }, + { + "epoch": 1.1340523882896765, + "grad_norm": 2.7702295780181885, + "learning_rate": 3.4830605253140464e-06, + "loss": 1.5206, + "step": 1104 + }, + { + "epoch": 1.1350796096558808, + "grad_norm": 2.6476149559020996, + "learning_rate": 3.4811572135515797e-06, + "loss": 1.4735, + "step": 1105 + }, + { + "epoch": 1.1361068310220852, + "grad_norm": 2.5630619525909424, + "learning_rate": 3.4792539017891135e-06, + "loss": 1.5109, + "step": 1106 + }, + { + "epoch": 1.1371340523882898, + "grad_norm": 2.549581289291382, + "learning_rate": 3.477350590026647e-06, + "loss": 1.4919, + "step": 1107 + }, + { + "epoch": 1.1381612737544942, + "grad_norm": 2.8687074184417725, + "learning_rate": 3.4754472782641802e-06, + "loss": 1.5658, + "step": 1108 + }, + { + "epoch": 1.1391884951206985, + "grad_norm": 2.799149751663208, + "learning_rate": 3.4735439665017136e-06, + "loss": 1.5582, + "step": 1109 + }, + { + "epoch": 1.140215716486903, + "grad_norm": 2.8335328102111816, + "learning_rate": 3.4716406547392465e-06, + "loss": 1.5217, + "step": 1110 + }, + { + "epoch": 1.140215716486903, + "eval_loss": 1.9011008739471436, + "eval_runtime": 42.9222, + "eval_samples_per_second": 23.298, + "eval_steps_per_second": 0.979, + "step": 1110 + }, + { + "epoch": 1.1412429378531073, + "grad_norm": 2.7025723457336426, + "learning_rate": 3.46973734297678e-06, + "loss": 1.5233, + "step": 1111 + }, + { + "epoch": 1.1422701592193119, + "grad_norm": 2.6984663009643555, + "learning_rate": 3.4678340312143133e-06, + "loss": 1.5447, + "step": 1112 + }, + { + "epoch": 1.1432973805855162, + "grad_norm": 2.708031415939331, + "learning_rate": 3.4659307194518466e-06, + "loss": 1.5905, + "step": 1113 + }, + { + "epoch": 1.1443246019517206, + "grad_norm": 2.5180366039276123, + "learning_rate": 3.46402740768938e-06, + "loss": 1.5636, + "step": 1114 + }, + { + "epoch": 1.145351823317925, + "grad_norm": 2.540074348449707, + "learning_rate": 3.462124095926913e-06, + "loss": 1.5045, + "step": 1115 + }, + { + "epoch": 1.1463790446841293, + "grad_norm": 2.4670231342315674, + "learning_rate": 3.4602207841644463e-06, + "loss": 1.4969, + "step": 1116 + }, + { + "epoch": 1.147406266050334, + "grad_norm": 2.7803049087524414, + "learning_rate": 3.4583174724019797e-06, + "loss": 1.5471, + "step": 1117 + }, + { + "epoch": 1.1484334874165383, + "grad_norm": 2.58117413520813, + "learning_rate": 3.456414160639513e-06, + "loss": 1.592, + "step": 1118 + }, + { + "epoch": 1.1494607087827426, + "grad_norm": 2.5810937881469727, + "learning_rate": 3.4545108488770464e-06, + "loss": 1.5542, + "step": 1119 + }, + { + "epoch": 1.150487930148947, + "grad_norm": 2.757809638977051, + "learning_rate": 3.4526075371145793e-06, + "loss": 1.5403, + "step": 1120 + }, + { + "epoch": 1.150487930148947, + "eval_loss": 1.905651569366455, + "eval_runtime": 45.619, + "eval_samples_per_second": 21.921, + "eval_steps_per_second": 0.921, + "step": 1120 + }, + { + "epoch": 1.1515151515151516, + "grad_norm": 2.6740472316741943, + "learning_rate": 3.4507042253521127e-06, + "loss": 1.5422, + "step": 1121 + }, + { + "epoch": 1.152542372881356, + "grad_norm": 2.7041897773742676, + "learning_rate": 3.448800913589646e-06, + "loss": 1.6049, + "step": 1122 + }, + { + "epoch": 1.1535695942475603, + "grad_norm": 2.7759902477264404, + "learning_rate": 3.4468976018271794e-06, + "loss": 1.522, + "step": 1123 + }, + { + "epoch": 1.1545968156137647, + "grad_norm": 2.69626522064209, + "learning_rate": 3.4449942900647128e-06, + "loss": 1.6096, + "step": 1124 + }, + { + "epoch": 1.1556240369799693, + "grad_norm": 2.57637619972229, + "learning_rate": 3.443090978302246e-06, + "loss": 1.5702, + "step": 1125 + }, + { + "epoch": 1.1566512583461737, + "grad_norm": 2.6259238719940186, + "learning_rate": 3.44118766653978e-06, + "loss": 1.5561, + "step": 1126 + }, + { + "epoch": 1.157678479712378, + "grad_norm": 2.7071187496185303, + "learning_rate": 3.439284354777313e-06, + "loss": 1.5545, + "step": 1127 + }, + { + "epoch": 1.1587057010785824, + "grad_norm": 2.7928988933563232, + "learning_rate": 3.4373810430148462e-06, + "loss": 1.5551, + "step": 1128 + }, + { + "epoch": 1.1597329224447868, + "grad_norm": 2.703719139099121, + "learning_rate": 3.4354777312523796e-06, + "loss": 1.4947, + "step": 1129 + }, + { + "epoch": 1.1607601438109914, + "grad_norm": 2.6363744735717773, + "learning_rate": 3.433574419489913e-06, + "loss": 1.511, + "step": 1130 + }, + { + "epoch": 1.1607601438109914, + "eval_loss": 1.9003485441207886, + "eval_runtime": 46.8855, + "eval_samples_per_second": 21.329, + "eval_steps_per_second": 0.896, + "step": 1130 + }, + { + "epoch": 1.1617873651771957, + "grad_norm": 2.7165591716766357, + "learning_rate": 3.4316711077274463e-06, + "loss": 1.5529, + "step": 1131 + }, + { + "epoch": 1.1628145865434, + "grad_norm": 2.6615865230560303, + "learning_rate": 3.4297677959649792e-06, + "loss": 1.5213, + "step": 1132 + }, + { + "epoch": 1.1638418079096045, + "grad_norm": 2.6274473667144775, + "learning_rate": 3.4278644842025126e-06, + "loss": 1.5813, + "step": 1133 + }, + { + "epoch": 1.1648690292758088, + "grad_norm": 2.637418031692505, + "learning_rate": 3.425961172440046e-06, + "loss": 1.4963, + "step": 1134 + }, + { + "epoch": 1.1658962506420134, + "grad_norm": 2.601574182510376, + "learning_rate": 3.4240578606775793e-06, + "loss": 1.6146, + "step": 1135 + }, + { + "epoch": 1.1669234720082178, + "grad_norm": 2.599043369293213, + "learning_rate": 3.4221545489151127e-06, + "loss": 1.5251, + "step": 1136 + }, + { + "epoch": 1.1679506933744221, + "grad_norm": 2.8078091144561768, + "learning_rate": 3.4202512371526456e-06, + "loss": 1.5231, + "step": 1137 + }, + { + "epoch": 1.1689779147406265, + "grad_norm": 2.811053514480591, + "learning_rate": 3.418347925390179e-06, + "loss": 1.6289, + "step": 1138 + }, + { + "epoch": 1.170005136106831, + "grad_norm": 2.7277560234069824, + "learning_rate": 3.4164446136277124e-06, + "loss": 1.5733, + "step": 1139 + }, + { + "epoch": 1.1710323574730355, + "grad_norm": 2.587024211883545, + "learning_rate": 3.4145413018652457e-06, + "loss": 1.5332, + "step": 1140 + }, + { + "epoch": 1.1710323574730355, + "eval_loss": 1.8978687524795532, + "eval_runtime": 49.9935, + "eval_samples_per_second": 20.003, + "eval_steps_per_second": 0.84, + "step": 1140 + }, + { + "epoch": 1.1720595788392398, + "grad_norm": 2.536479949951172, + "learning_rate": 3.412637990102779e-06, + "loss": 1.5601, + "step": 1141 + }, + { + "epoch": 1.1730868002054442, + "grad_norm": 2.7315709590911865, + "learning_rate": 3.4107346783403125e-06, + "loss": 1.5783, + "step": 1142 + }, + { + "epoch": 1.1741140215716488, + "grad_norm": 2.507249593734741, + "learning_rate": 3.4088313665778454e-06, + "loss": 1.5, + "step": 1143 + }, + { + "epoch": 1.1751412429378532, + "grad_norm": 2.625411033630371, + "learning_rate": 3.4069280548153788e-06, + "loss": 1.5328, + "step": 1144 + }, + { + "epoch": 1.1761684643040575, + "grad_norm": 2.591282367706299, + "learning_rate": 3.405024743052912e-06, + "loss": 1.5119, + "step": 1145 + }, + { + "epoch": 1.177195685670262, + "grad_norm": 2.547508955001831, + "learning_rate": 3.403121431290446e-06, + "loss": 1.5361, + "step": 1146 + }, + { + "epoch": 1.1782229070364663, + "grad_norm": 2.797992706298828, + "learning_rate": 3.4012181195279793e-06, + "loss": 1.5221, + "step": 1147 + }, + { + "epoch": 1.1792501284026708, + "grad_norm": 2.905266761779785, + "learning_rate": 3.3993148077655126e-06, + "loss": 1.5209, + "step": 1148 + }, + { + "epoch": 1.1802773497688752, + "grad_norm": 2.9259016513824463, + "learning_rate": 3.3974114960030456e-06, + "loss": 1.5365, + "step": 1149 + }, + { + "epoch": 1.1813045711350796, + "grad_norm": 2.597102165222168, + "learning_rate": 3.395508184240579e-06, + "loss": 1.4663, + "step": 1150 + }, + { + "epoch": 1.1813045711350796, + "eval_loss": 1.8966383934020996, + "eval_runtime": 47.3579, + "eval_samples_per_second": 21.116, + "eval_steps_per_second": 0.887, + "step": 1150 + }, + { + "epoch": 1.182331792501284, + "grad_norm": 2.6672847270965576, + "learning_rate": 3.3936048724781123e-06, + "loss": 1.5115, + "step": 1151 + }, + { + "epoch": 1.1833590138674885, + "grad_norm": 2.870549201965332, + "learning_rate": 3.3917015607156457e-06, + "loss": 1.6097, + "step": 1152 + }, + { + "epoch": 1.184386235233693, + "grad_norm": 2.6559388637542725, + "learning_rate": 3.389798248953179e-06, + "loss": 1.4777, + "step": 1153 + }, + { + "epoch": 1.1854134565998973, + "grad_norm": 2.7968215942382812, + "learning_rate": 3.387894937190712e-06, + "loss": 1.509, + "step": 1154 + }, + { + "epoch": 1.1864406779661016, + "grad_norm": 2.689741611480713, + "learning_rate": 3.3859916254282453e-06, + "loss": 1.5424, + "step": 1155 + }, + { + "epoch": 1.1874678993323062, + "grad_norm": 2.872887134552002, + "learning_rate": 3.3840883136657787e-06, + "loss": 1.5404, + "step": 1156 + }, + { + "epoch": 1.1884951206985106, + "grad_norm": 2.6436610221862793, + "learning_rate": 3.382185001903312e-06, + "loss": 1.5889, + "step": 1157 + }, + { + "epoch": 1.189522342064715, + "grad_norm": 2.697906255722046, + "learning_rate": 3.3802816901408454e-06, + "loss": 1.5086, + "step": 1158 + }, + { + "epoch": 1.1905495634309193, + "grad_norm": 2.8897645473480225, + "learning_rate": 3.3783783783783788e-06, + "loss": 1.5441, + "step": 1159 + }, + { + "epoch": 1.1915767847971237, + "grad_norm": 2.722351312637329, + "learning_rate": 3.3764750666159117e-06, + "loss": 1.527, + "step": 1160 + }, + { + "epoch": 1.1915767847971237, + "eval_loss": 1.8928718566894531, + "eval_runtime": 48.9427, + "eval_samples_per_second": 20.432, + "eval_steps_per_second": 0.858, + "step": 1160 + }, + { + "epoch": 1.1926040061633283, + "grad_norm": 2.5705020427703857, + "learning_rate": 3.374571754853445e-06, + "loss": 1.5304, + "step": 1161 + }, + { + "epoch": 1.1936312275295327, + "grad_norm": 2.571810245513916, + "learning_rate": 3.3726684430909784e-06, + "loss": 1.5242, + "step": 1162 + }, + { + "epoch": 1.194658448895737, + "grad_norm": 2.552725076675415, + "learning_rate": 3.370765131328512e-06, + "loss": 1.4803, + "step": 1163 + }, + { + "epoch": 1.1956856702619414, + "grad_norm": 2.734027862548828, + "learning_rate": 3.368861819566045e-06, + "loss": 1.5041, + "step": 1164 + }, + { + "epoch": 1.1967128916281458, + "grad_norm": 2.594614028930664, + "learning_rate": 3.366958507803578e-06, + "loss": 1.5205, + "step": 1165 + }, + { + "epoch": 1.1977401129943503, + "grad_norm": 2.897482395172119, + "learning_rate": 3.365055196041112e-06, + "loss": 1.5372, + "step": 1166 + }, + { + "epoch": 1.1987673343605547, + "grad_norm": 2.6143221855163574, + "learning_rate": 3.3631518842786452e-06, + "loss": 1.4966, + "step": 1167 + }, + { + "epoch": 1.199794555726759, + "grad_norm": 2.845475912094116, + "learning_rate": 3.3612485725161786e-06, + "loss": 1.5759, + "step": 1168 + }, + { + "epoch": 1.2008217770929634, + "grad_norm": 2.667388916015625, + "learning_rate": 3.359345260753712e-06, + "loss": 1.5014, + "step": 1169 + }, + { + "epoch": 1.201848998459168, + "grad_norm": 2.890197277069092, + "learning_rate": 3.3574419489912453e-06, + "loss": 1.5939, + "step": 1170 + }, + { + "epoch": 1.201848998459168, + "eval_loss": 1.8978379964828491, + "eval_runtime": 48.9839, + "eval_samples_per_second": 20.415, + "eval_steps_per_second": 0.857, + "step": 1170 + }, + { + "epoch": 1.2028762198253724, + "grad_norm": 2.578139066696167, + "learning_rate": 3.3555386372287783e-06, + "loss": 1.5655, + "step": 1171 + }, + { + "epoch": 1.2039034411915768, + "grad_norm": 2.5672738552093506, + "learning_rate": 3.3536353254663116e-06, + "loss": 1.5998, + "step": 1172 + }, + { + "epoch": 1.2049306625577811, + "grad_norm": 2.7297089099884033, + "learning_rate": 3.351732013703845e-06, + "loss": 1.4938, + "step": 1173 + }, + { + "epoch": 1.2059578839239857, + "grad_norm": 2.7455339431762695, + "learning_rate": 3.3498287019413784e-06, + "loss": 1.5505, + "step": 1174 + }, + { + "epoch": 1.20698510529019, + "grad_norm": 2.725179433822632, + "learning_rate": 3.3479253901789117e-06, + "loss": 1.5524, + "step": 1175 + }, + { + "epoch": 1.2080123266563945, + "grad_norm": 2.6953988075256348, + "learning_rate": 3.346022078416445e-06, + "loss": 1.568, + "step": 1176 + }, + { + "epoch": 1.2090395480225988, + "grad_norm": 2.613600492477417, + "learning_rate": 3.344118766653978e-06, + "loss": 1.4608, + "step": 1177 + }, + { + "epoch": 1.2100667693888032, + "grad_norm": 2.7339162826538086, + "learning_rate": 3.3422154548915114e-06, + "loss": 1.5426, + "step": 1178 + }, + { + "epoch": 1.2110939907550078, + "grad_norm": 2.774726152420044, + "learning_rate": 3.3403121431290448e-06, + "loss": 1.5192, + "step": 1179 + }, + { + "epoch": 1.2121212121212122, + "grad_norm": 2.7361602783203125, + "learning_rate": 3.338408831366578e-06, + "loss": 1.5481, + "step": 1180 + }, + { + "epoch": 1.2121212121212122, + "eval_loss": 1.902246117591858, + "eval_runtime": 45.153, + "eval_samples_per_second": 22.147, + "eval_steps_per_second": 0.93, + "step": 1180 + }, + { + "epoch": 1.2131484334874165, + "grad_norm": 2.6936450004577637, + "learning_rate": 3.3365055196041115e-06, + "loss": 1.4821, + "step": 1181 + }, + { + "epoch": 1.214175654853621, + "grad_norm": 2.881898880004883, + "learning_rate": 3.3346022078416444e-06, + "loss": 1.5663, + "step": 1182 + }, + { + "epoch": 1.2152028762198253, + "grad_norm": 2.793069362640381, + "learning_rate": 3.3326988960791778e-06, + "loss": 1.5998, + "step": 1183 + }, + { + "epoch": 1.2162300975860298, + "grad_norm": 2.591367721557617, + "learning_rate": 3.330795584316711e-06, + "loss": 1.5199, + "step": 1184 + }, + { + "epoch": 1.2172573189522342, + "grad_norm": 2.8402175903320312, + "learning_rate": 3.3288922725542445e-06, + "loss": 1.451, + "step": 1185 + }, + { + "epoch": 1.2182845403184386, + "grad_norm": 2.737199068069458, + "learning_rate": 3.326988960791778e-06, + "loss": 1.5117, + "step": 1186 + }, + { + "epoch": 1.219311761684643, + "grad_norm": 2.915210723876953, + "learning_rate": 3.3250856490293117e-06, + "loss": 1.5388, + "step": 1187 + }, + { + "epoch": 1.2203389830508475, + "grad_norm": 3.091784715652466, + "learning_rate": 3.323182337266845e-06, + "loss": 1.5944, + "step": 1188 + }, + { + "epoch": 1.221366204417052, + "grad_norm": 2.770756483078003, + "learning_rate": 3.321279025504378e-06, + "loss": 1.5619, + "step": 1189 + }, + { + "epoch": 1.2223934257832563, + "grad_norm": 2.7989954948425293, + "learning_rate": 3.3193757137419113e-06, + "loss": 1.5581, + "step": 1190 + }, + { + "epoch": 1.2223934257832563, + "eval_loss": 1.907904863357544, + "eval_runtime": 46.8087, + "eval_samples_per_second": 21.364, + "eval_steps_per_second": 0.897, + "step": 1190 + }, + { + "epoch": 1.2234206471494606, + "grad_norm": 2.6740641593933105, + "learning_rate": 3.3174724019794447e-06, + "loss": 1.5556, + "step": 1191 + }, + { + "epoch": 1.2244478685156652, + "grad_norm": 2.760608196258545, + "learning_rate": 3.315569090216978e-06, + "loss": 1.5903, + "step": 1192 + }, + { + "epoch": 1.2254750898818696, + "grad_norm": 2.942370891571045, + "learning_rate": 3.3136657784545114e-06, + "loss": 1.5052, + "step": 1193 + }, + { + "epoch": 1.226502311248074, + "grad_norm": 2.8096179962158203, + "learning_rate": 3.3117624666920443e-06, + "loss": 1.5933, + "step": 1194 + }, + { + "epoch": 1.2275295326142783, + "grad_norm": 2.7357091903686523, + "learning_rate": 3.3098591549295777e-06, + "loss": 1.5074, + "step": 1195 + }, + { + "epoch": 1.2285567539804827, + "grad_norm": 2.728134870529175, + "learning_rate": 3.307955843167111e-06, + "loss": 1.48, + "step": 1196 + }, + { + "epoch": 1.2295839753466873, + "grad_norm": 2.921462059020996, + "learning_rate": 3.3060525314046444e-06, + "loss": 1.5602, + "step": 1197 + }, + { + "epoch": 1.2306111967128917, + "grad_norm": 2.7563631534576416, + "learning_rate": 3.304149219642178e-06, + "loss": 1.5309, + "step": 1198 + }, + { + "epoch": 1.231638418079096, + "grad_norm": 2.533001661300659, + "learning_rate": 3.3022459078797107e-06, + "loss": 1.5242, + "step": 1199 + }, + { + "epoch": 1.2326656394453004, + "grad_norm": 2.5509681701660156, + "learning_rate": 3.300342596117244e-06, + "loss": 1.5488, + "step": 1200 + }, + { + "epoch": 1.2326656394453004, + "eval_loss": 1.8983508348464966, + "eval_runtime": 44.8654, + "eval_samples_per_second": 22.289, + "eval_steps_per_second": 0.936, + "step": 1200 + }, + { + "epoch": 1.2336928608115048, + "grad_norm": 2.688228130340576, + "learning_rate": 3.2984392843547775e-06, + "loss": 1.5344, + "step": 1201 + }, + { + "epoch": 1.2347200821777093, + "grad_norm": 2.567621946334839, + "learning_rate": 3.296535972592311e-06, + "loss": 1.4926, + "step": 1202 + }, + { + "epoch": 1.2357473035439137, + "grad_norm": 2.586637258529663, + "learning_rate": 3.294632660829844e-06, + "loss": 1.4749, + "step": 1203 + }, + { + "epoch": 1.236774524910118, + "grad_norm": 2.697901725769043, + "learning_rate": 3.292729349067377e-06, + "loss": 1.5073, + "step": 1204 + }, + { + "epoch": 1.2378017462763227, + "grad_norm": 2.832059621810913, + "learning_rate": 3.2908260373049105e-06, + "loss": 1.5147, + "step": 1205 + }, + { + "epoch": 1.238828967642527, + "grad_norm": 2.7802395820617676, + "learning_rate": 3.288922725542444e-06, + "loss": 1.5031, + "step": 1206 + }, + { + "epoch": 1.2398561890087314, + "grad_norm": 2.8382885456085205, + "learning_rate": 3.2870194137799776e-06, + "loss": 1.5539, + "step": 1207 + }, + { + "epoch": 1.2408834103749358, + "grad_norm": 2.7725186347961426, + "learning_rate": 3.285116102017511e-06, + "loss": 1.5145, + "step": 1208 + }, + { + "epoch": 1.2419106317411401, + "grad_norm": 2.563866376876831, + "learning_rate": 3.2832127902550444e-06, + "loss": 1.4852, + "step": 1209 + }, + { + "epoch": 1.2429378531073447, + "grad_norm": 2.7052412033081055, + "learning_rate": 3.2813094784925777e-06, + "loss": 1.5615, + "step": 1210 + }, + { + "epoch": 1.2429378531073447, + "eval_loss": 1.9071296453475952, + "eval_runtime": 42.2147, + "eval_samples_per_second": 23.688, + "eval_steps_per_second": 0.995, + "step": 1210 + }, + { + "epoch": 1.243965074473549, + "grad_norm": 2.7794888019561768, + "learning_rate": 3.2794061667301107e-06, + "loss": 1.5445, + "step": 1211 + }, + { + "epoch": 1.2449922958397535, + "grad_norm": 2.6376712322235107, + "learning_rate": 3.277502854967644e-06, + "loss": 1.4728, + "step": 1212 + }, + { + "epoch": 1.2460195172059578, + "grad_norm": 2.5500779151916504, + "learning_rate": 3.2755995432051774e-06, + "loss": 1.5133, + "step": 1213 + }, + { + "epoch": 1.2470467385721622, + "grad_norm": 2.629063606262207, + "learning_rate": 3.2736962314427107e-06, + "loss": 1.5093, + "step": 1214 + }, + { + "epoch": 1.2480739599383668, + "grad_norm": 2.662497043609619, + "learning_rate": 3.271792919680244e-06, + "loss": 1.5237, + "step": 1215 + }, + { + "epoch": 1.2491011813045712, + "grad_norm": 2.5375289916992188, + "learning_rate": 3.269889607917777e-06, + "loss": 1.5211, + "step": 1216 + }, + { + "epoch": 1.2501284026707755, + "grad_norm": 2.8396782875061035, + "learning_rate": 3.2679862961553104e-06, + "loss": 1.5773, + "step": 1217 + }, + { + "epoch": 1.25115562403698, + "grad_norm": 2.8443665504455566, + "learning_rate": 3.2660829843928438e-06, + "loss": 1.4987, + "step": 1218 + }, + { + "epoch": 1.2521828454031843, + "grad_norm": 2.77935528755188, + "learning_rate": 3.264179672630377e-06, + "loss": 1.5344, + "step": 1219 + }, + { + "epoch": 1.2532100667693888, + "grad_norm": 3.0581092834472656, + "learning_rate": 3.2622763608679105e-06, + "loss": 1.5728, + "step": 1220 + }, + { + "epoch": 1.2532100667693888, + "eval_loss": 1.9198752641677856, + "eval_runtime": 45.1116, + "eval_samples_per_second": 22.167, + "eval_steps_per_second": 0.931, + "step": 1220 + }, + { + "epoch": 1.2542372881355932, + "grad_norm": 2.8741002082824707, + "learning_rate": 3.2603730491054434e-06, + "loss": 1.5444, + "step": 1221 + }, + { + "epoch": 1.2552645095017976, + "grad_norm": 2.6263391971588135, + "learning_rate": 3.258469737342977e-06, + "loss": 1.4997, + "step": 1222 + }, + { + "epoch": 1.2562917308680022, + "grad_norm": 2.949795961380005, + "learning_rate": 3.25656642558051e-06, + "loss": 1.5145, + "step": 1223 + }, + { + "epoch": 1.2573189522342065, + "grad_norm": 2.846130609512329, + "learning_rate": 3.2546631138180435e-06, + "loss": 1.4709, + "step": 1224 + }, + { + "epoch": 1.258346173600411, + "grad_norm": 2.6340930461883545, + "learning_rate": 3.252759802055577e-06, + "loss": 1.5087, + "step": 1225 + }, + { + "epoch": 1.2593733949666153, + "grad_norm": 2.9878976345062256, + "learning_rate": 3.25085649029311e-06, + "loss": 1.5142, + "step": 1226 + }, + { + "epoch": 1.2604006163328196, + "grad_norm": 2.9880733489990234, + "learning_rate": 3.248953178530644e-06, + "loss": 1.521, + "step": 1227 + }, + { + "epoch": 1.2614278376990242, + "grad_norm": 2.8507912158966064, + "learning_rate": 3.247049866768177e-06, + "loss": 1.5482, + "step": 1228 + }, + { + "epoch": 1.2624550590652286, + "grad_norm": 2.870678424835205, + "learning_rate": 3.2451465550057103e-06, + "loss": 1.4965, + "step": 1229 + }, + { + "epoch": 1.263482280431433, + "grad_norm": 2.644347906112671, + "learning_rate": 3.2432432432432437e-06, + "loss": 1.4614, + "step": 1230 + }, + { + "epoch": 1.263482280431433, + "eval_loss": 1.9100512266159058, + "eval_runtime": 44.6472, + "eval_samples_per_second": 22.398, + "eval_steps_per_second": 0.941, + "step": 1230 + }, + { + "epoch": 1.2645095017976373, + "grad_norm": 2.5852861404418945, + "learning_rate": 3.241339931480777e-06, + "loss": 1.5122, + "step": 1231 + }, + { + "epoch": 1.2655367231638417, + "grad_norm": 2.9572696685791016, + "learning_rate": 3.2394366197183104e-06, + "loss": 1.5529, + "step": 1232 + }, + { + "epoch": 1.2665639445300463, + "grad_norm": 2.7040176391601562, + "learning_rate": 3.2375333079558434e-06, + "loss": 1.5418, + "step": 1233 + }, + { + "epoch": 1.2675911658962506, + "grad_norm": 2.577706813812256, + "learning_rate": 3.2356299961933767e-06, + "loss": 1.5167, + "step": 1234 + }, + { + "epoch": 1.268618387262455, + "grad_norm": 2.6559629440307617, + "learning_rate": 3.23372668443091e-06, + "loss": 1.5278, + "step": 1235 + }, + { + "epoch": 1.2696456086286596, + "grad_norm": 2.673293113708496, + "learning_rate": 3.2318233726684435e-06, + "loss": 1.5393, + "step": 1236 + }, + { + "epoch": 1.270672829994864, + "grad_norm": 2.507797956466675, + "learning_rate": 3.229920060905977e-06, + "loss": 1.5573, + "step": 1237 + }, + { + "epoch": 1.2717000513610683, + "grad_norm": 2.8361523151397705, + "learning_rate": 3.2280167491435098e-06, + "loss": 1.4911, + "step": 1238 + }, + { + "epoch": 1.2727272727272727, + "grad_norm": 2.7824695110321045, + "learning_rate": 3.226113437381043e-06, + "loss": 1.4949, + "step": 1239 + }, + { + "epoch": 1.273754494093477, + "grad_norm": 2.4863975048065186, + "learning_rate": 3.2242101256185765e-06, + "loss": 1.5017, + "step": 1240 + }, + { + "epoch": 1.273754494093477, + "eval_loss": 1.896610140800476, + "eval_runtime": 46.9004, + "eval_samples_per_second": 21.322, + "eval_steps_per_second": 0.896, + "step": 1240 + }, + { + "epoch": 1.2747817154596817, + "grad_norm": 2.767557144165039, + "learning_rate": 3.22230681385611e-06, + "loss": 1.5137, + "step": 1241 + }, + { + "epoch": 1.275808936825886, + "grad_norm": 2.544274091720581, + "learning_rate": 3.220403502093643e-06, + "loss": 1.5759, + "step": 1242 + }, + { + "epoch": 1.2768361581920904, + "grad_norm": 2.6679677963256836, + "learning_rate": 3.218500190331176e-06, + "loss": 1.533, + "step": 1243 + }, + { + "epoch": 1.2778633795582948, + "grad_norm": 2.634310245513916, + "learning_rate": 3.2165968785687095e-06, + "loss": 1.5605, + "step": 1244 + }, + { + "epoch": 1.2788906009244991, + "grad_norm": 2.6534063816070557, + "learning_rate": 3.214693566806243e-06, + "loss": 1.5129, + "step": 1245 + }, + { + "epoch": 1.2799178222907037, + "grad_norm": 2.6762259006500244, + "learning_rate": 3.2127902550437762e-06, + "loss": 1.5435, + "step": 1246 + }, + { + "epoch": 1.280945043656908, + "grad_norm": 2.70534610748291, + "learning_rate": 3.21088694328131e-06, + "loss": 1.5215, + "step": 1247 + }, + { + "epoch": 1.2819722650231125, + "grad_norm": 2.929002046585083, + "learning_rate": 3.2089836315188434e-06, + "loss": 1.5237, + "step": 1248 + }, + { + "epoch": 1.2829994863893168, + "grad_norm": 2.6163816452026367, + "learning_rate": 3.2070803197563767e-06, + "loss": 1.5114, + "step": 1249 + }, + { + "epoch": 1.2840267077555212, + "grad_norm": 2.6484017372131348, + "learning_rate": 3.2051770079939097e-06, + "loss": 1.5054, + "step": 1250 + }, + { + "epoch": 1.2840267077555212, + "eval_loss": 1.9057985544204712, + "eval_runtime": 45.9877, + "eval_samples_per_second": 21.745, + "eval_steps_per_second": 0.913, + "step": 1250 + }, + { + "epoch": 1.2850539291217258, + "grad_norm": 2.694584608078003, + "learning_rate": 3.203273696231443e-06, + "loss": 1.5272, + "step": 1251 + }, + { + "epoch": 1.2860811504879301, + "grad_norm": 2.699054718017578, + "learning_rate": 3.2013703844689764e-06, + "loss": 1.5632, + "step": 1252 + }, + { + "epoch": 1.2871083718541345, + "grad_norm": 2.765612840652466, + "learning_rate": 3.1994670727065098e-06, + "loss": 1.4927, + "step": 1253 + }, + { + "epoch": 1.288135593220339, + "grad_norm": 2.7850136756896973, + "learning_rate": 3.197563760944043e-06, + "loss": 1.5412, + "step": 1254 + }, + { + "epoch": 1.2891628145865435, + "grad_norm": 2.6565496921539307, + "learning_rate": 3.195660449181576e-06, + "loss": 1.5606, + "step": 1255 + }, + { + "epoch": 1.2901900359527478, + "grad_norm": 2.627163887023926, + "learning_rate": 3.1937571374191094e-06, + "loss": 1.5612, + "step": 1256 + }, + { + "epoch": 1.2912172573189522, + "grad_norm": 2.5778250694274902, + "learning_rate": 3.191853825656643e-06, + "loss": 1.4988, + "step": 1257 + }, + { + "epoch": 1.2922444786851566, + "grad_norm": 2.7875125408172607, + "learning_rate": 3.189950513894176e-06, + "loss": 1.5237, + "step": 1258 + }, + { + "epoch": 1.2932717000513612, + "grad_norm": 2.645707607269287, + "learning_rate": 3.1880472021317095e-06, + "loss": 1.5157, + "step": 1259 + }, + { + "epoch": 1.2942989214175655, + "grad_norm": 2.6457326412200928, + "learning_rate": 3.1861438903692425e-06, + "loss": 1.5109, + "step": 1260 + }, + { + "epoch": 1.2942989214175655, + "eval_loss": 1.896759033203125, + "eval_runtime": 50.1276, + "eval_samples_per_second": 19.949, + "eval_steps_per_second": 0.838, + "step": 1260 + }, + { + "epoch": 1.29532614278377, + "grad_norm": 2.617196798324585, + "learning_rate": 3.184240578606776e-06, + "loss": 1.4646, + "step": 1261 + }, + { + "epoch": 1.2963533641499743, + "grad_norm": 2.618905782699585, + "learning_rate": 3.182337266844309e-06, + "loss": 1.5012, + "step": 1262 + }, + { + "epoch": 1.2973805855161786, + "grad_norm": 2.6222894191741943, + "learning_rate": 3.1804339550818426e-06, + "loss": 1.526, + "step": 1263 + }, + { + "epoch": 1.2984078068823832, + "grad_norm": 2.858421564102173, + "learning_rate": 3.178530643319376e-06, + "loss": 1.5621, + "step": 1264 + }, + { + "epoch": 1.2994350282485876, + "grad_norm": 2.579683303833008, + "learning_rate": 3.176627331556909e-06, + "loss": 1.5217, + "step": 1265 + }, + { + "epoch": 1.300462249614792, + "grad_norm": 2.721665859222412, + "learning_rate": 3.1747240197944422e-06, + "loss": 1.4732, + "step": 1266 + }, + { + "epoch": 1.3014894709809965, + "grad_norm": 2.6431221961975098, + "learning_rate": 3.172820708031976e-06, + "loss": 1.5219, + "step": 1267 + }, + { + "epoch": 1.3025166923472007, + "grad_norm": 2.706671714782715, + "learning_rate": 3.1709173962695094e-06, + "loss": 1.4829, + "step": 1268 + }, + { + "epoch": 1.3035439137134053, + "grad_norm": 2.81333589553833, + "learning_rate": 3.1690140845070427e-06, + "loss": 1.5669, + "step": 1269 + }, + { + "epoch": 1.3045711350796096, + "grad_norm": 2.942777633666992, + "learning_rate": 3.167110772744576e-06, + "loss": 1.5477, + "step": 1270 + }, + { + "epoch": 1.3045711350796096, + "eval_loss": 1.9051777124404907, + "eval_runtime": 48.2881, + "eval_samples_per_second": 20.709, + "eval_steps_per_second": 0.87, + "step": 1270 + }, + { + "epoch": 1.305598356445814, + "grad_norm": 2.721989393234253, + "learning_rate": 3.1652074609821095e-06, + "loss": 1.5269, + "step": 1271 + }, + { + "epoch": 1.3066255778120186, + "grad_norm": 2.8142035007476807, + "learning_rate": 3.1633041492196424e-06, + "loss": 1.5193, + "step": 1272 + }, + { + "epoch": 1.307652799178223, + "grad_norm": 2.796630382537842, + "learning_rate": 3.1614008374571758e-06, + "loss": 1.503, + "step": 1273 + }, + { + "epoch": 1.3086800205444273, + "grad_norm": 2.696720600128174, + "learning_rate": 3.159497525694709e-06, + "loss": 1.5301, + "step": 1274 + }, + { + "epoch": 1.3097072419106317, + "grad_norm": 2.6286635398864746, + "learning_rate": 3.1575942139322425e-06, + "loss": 1.5326, + "step": 1275 + }, + { + "epoch": 1.310734463276836, + "grad_norm": 2.662801504135132, + "learning_rate": 3.155690902169776e-06, + "loss": 1.5048, + "step": 1276 + }, + { + "epoch": 1.3117616846430407, + "grad_norm": 2.838228702545166, + "learning_rate": 3.1537875904073088e-06, + "loss": 1.5321, + "step": 1277 + }, + { + "epoch": 1.312788906009245, + "grad_norm": 2.7021734714508057, + "learning_rate": 3.151884278644842e-06, + "loss": 1.4917, + "step": 1278 + }, + { + "epoch": 1.3138161273754494, + "grad_norm": 2.696664571762085, + "learning_rate": 3.1499809668823755e-06, + "loss": 1.5231, + "step": 1279 + }, + { + "epoch": 1.3148433487416538, + "grad_norm": 2.720292091369629, + "learning_rate": 3.148077655119909e-06, + "loss": 1.497, + "step": 1280 + }, + { + "epoch": 1.3148433487416538, + "eval_loss": 1.8934897184371948, + "eval_runtime": 48.5119, + "eval_samples_per_second": 20.613, + "eval_steps_per_second": 0.866, + "step": 1280 + }, + { + "epoch": 1.3158705701078581, + "grad_norm": 2.6228737831115723, + "learning_rate": 3.1461743433574422e-06, + "loss": 1.5475, + "step": 1281 + }, + { + "epoch": 1.3168977914740627, + "grad_norm": 2.733198881149292, + "learning_rate": 3.144271031594975e-06, + "loss": 1.4909, + "step": 1282 + }, + { + "epoch": 1.317925012840267, + "grad_norm": 2.5863518714904785, + "learning_rate": 3.1423677198325085e-06, + "loss": 1.5446, + "step": 1283 + }, + { + "epoch": 1.3189522342064715, + "grad_norm": 2.775151491165161, + "learning_rate": 3.140464408070042e-06, + "loss": 1.5414, + "step": 1284 + }, + { + "epoch": 1.319979455572676, + "grad_norm": 2.7392020225524902, + "learning_rate": 3.1385610963075753e-06, + "loss": 1.5094, + "step": 1285 + }, + { + "epoch": 1.3210066769388804, + "grad_norm": 2.578563928604126, + "learning_rate": 3.1366577845451086e-06, + "loss": 1.4973, + "step": 1286 + }, + { + "epoch": 1.3220338983050848, + "grad_norm": 2.664562225341797, + "learning_rate": 3.134754472782642e-06, + "loss": 1.4874, + "step": 1287 + }, + { + "epoch": 1.3230611196712891, + "grad_norm": 2.713029146194458, + "learning_rate": 3.1328511610201758e-06, + "loss": 1.479, + "step": 1288 + }, + { + "epoch": 1.3240883410374935, + "grad_norm": 2.569258213043213, + "learning_rate": 3.1309478492577087e-06, + "loss": 1.4641, + "step": 1289 + }, + { + "epoch": 1.325115562403698, + "grad_norm": 2.4881503582000732, + "learning_rate": 3.129044537495242e-06, + "loss": 1.4634, + "step": 1290 + }, + { + "epoch": 1.325115562403698, + "eval_loss": 1.8945759534835815, + "eval_runtime": 48.2935, + "eval_samples_per_second": 20.707, + "eval_steps_per_second": 0.87, + "step": 1290 + }, + { + "epoch": 1.3261427837699025, + "grad_norm": 2.485280752182007, + "learning_rate": 3.1271412257327754e-06, + "loss": 1.507, + "step": 1291 + }, + { + "epoch": 1.3271700051361068, + "grad_norm": 2.7823352813720703, + "learning_rate": 3.125237913970309e-06, + "loss": 1.5414, + "step": 1292 + }, + { + "epoch": 1.3281972265023112, + "grad_norm": 2.516641139984131, + "learning_rate": 3.123334602207842e-06, + "loss": 1.4889, + "step": 1293 + }, + { + "epoch": 1.3292244478685156, + "grad_norm": 2.7155838012695312, + "learning_rate": 3.121431290445375e-06, + "loss": 1.4701, + "step": 1294 + }, + { + "epoch": 1.3302516692347202, + "grad_norm": 2.5513222217559814, + "learning_rate": 3.1195279786829085e-06, + "loss": 1.5514, + "step": 1295 + }, + { + "epoch": 1.3312788906009245, + "grad_norm": 2.7775397300720215, + "learning_rate": 3.117624666920442e-06, + "loss": 1.611, + "step": 1296 + }, + { + "epoch": 1.332306111967129, + "grad_norm": 2.693214178085327, + "learning_rate": 3.115721355157975e-06, + "loss": 1.497, + "step": 1297 + }, + { + "epoch": 1.3333333333333333, + "grad_norm": 2.6052982807159424, + "learning_rate": 3.1138180433955085e-06, + "loss": 1.4911, + "step": 1298 + }, + { + "epoch": 1.3343605546995376, + "grad_norm": 2.578404426574707, + "learning_rate": 3.1119147316330415e-06, + "loss": 1.4497, + "step": 1299 + }, + { + "epoch": 1.3353877760657422, + "grad_norm": 2.588397264480591, + "learning_rate": 3.110011419870575e-06, + "loss": 1.5339, + "step": 1300 + }, + { + "epoch": 1.3353877760657422, + "eval_loss": 1.8876255750656128, + "eval_runtime": 48.2305, + "eval_samples_per_second": 20.734, + "eval_steps_per_second": 0.871, + "step": 1300 + }, + { + "epoch": 1.3364149974319466, + "grad_norm": 2.835510730743408, + "learning_rate": 3.1081081081081082e-06, + "loss": 1.5649, + "step": 1301 + }, + { + "epoch": 1.337442218798151, + "grad_norm": 2.6646556854248047, + "learning_rate": 3.1062047963456416e-06, + "loss": 1.4969, + "step": 1302 + }, + { + "epoch": 1.3384694401643555, + "grad_norm": 2.6960718631744385, + "learning_rate": 3.104301484583175e-06, + "loss": 1.5332, + "step": 1303 + }, + { + "epoch": 1.33949666153056, + "grad_norm": 2.8116049766540527, + "learning_rate": 3.1023981728207083e-06, + "loss": 1.5766, + "step": 1304 + }, + { + "epoch": 1.3405238828967643, + "grad_norm": 2.7282750606536865, + "learning_rate": 3.1004948610582412e-06, + "loss": 1.551, + "step": 1305 + }, + { + "epoch": 1.3415511042629686, + "grad_norm": 2.6903767585754395, + "learning_rate": 3.0985915492957746e-06, + "loss": 1.5053, + "step": 1306 + }, + { + "epoch": 1.342578325629173, + "grad_norm": 2.525789976119995, + "learning_rate": 3.096688237533308e-06, + "loss": 1.4969, + "step": 1307 + }, + { + "epoch": 1.3436055469953776, + "grad_norm": 2.7578322887420654, + "learning_rate": 3.0947849257708418e-06, + "loss": 1.4701, + "step": 1308 + }, + { + "epoch": 1.344632768361582, + "grad_norm": 2.714743137359619, + "learning_rate": 3.092881614008375e-06, + "loss": 1.5628, + "step": 1309 + }, + { + "epoch": 1.3456599897277863, + "grad_norm": 2.5726380348205566, + "learning_rate": 3.0909783022459085e-06, + "loss": 1.5146, + "step": 1310 + }, + { + "epoch": 1.3456599897277863, + "eval_loss": 1.8887813091278076, + "eval_runtime": 44.978, + "eval_samples_per_second": 22.233, + "eval_steps_per_second": 0.934, + "step": 1310 + }, + { + "epoch": 1.3466872110939907, + "grad_norm": 2.82814621925354, + "learning_rate": 3.0890749904834414e-06, + "loss": 1.5684, + "step": 1311 + }, + { + "epoch": 1.347714432460195, + "grad_norm": 2.705440044403076, + "learning_rate": 3.0871716787209748e-06, + "loss": 1.5003, + "step": 1312 + }, + { + "epoch": 1.3487416538263997, + "grad_norm": 2.647636890411377, + "learning_rate": 3.085268366958508e-06, + "loss": 1.5528, + "step": 1313 + }, + { + "epoch": 1.349768875192604, + "grad_norm": 2.954592227935791, + "learning_rate": 3.0833650551960415e-06, + "loss": 1.5274, + "step": 1314 + }, + { + "epoch": 1.3507960965588084, + "grad_norm": 2.984116315841675, + "learning_rate": 3.081461743433575e-06, + "loss": 1.5411, + "step": 1315 + }, + { + "epoch": 1.3518233179250128, + "grad_norm": 2.5633351802825928, + "learning_rate": 3.079558431671108e-06, + "loss": 1.4853, + "step": 1316 + }, + { + "epoch": 1.3528505392912171, + "grad_norm": 2.8512840270996094, + "learning_rate": 3.077655119908641e-06, + "loss": 1.5117, + "step": 1317 + }, + { + "epoch": 1.3538777606574217, + "grad_norm": 2.709902286529541, + "learning_rate": 3.0757518081461745e-06, + "loss": 1.4716, + "step": 1318 + }, + { + "epoch": 1.354904982023626, + "grad_norm": 2.6622109413146973, + "learning_rate": 3.073848496383708e-06, + "loss": 1.5098, + "step": 1319 + }, + { + "epoch": 1.3559322033898304, + "grad_norm": 2.681844472885132, + "learning_rate": 3.0719451846212413e-06, + "loss": 1.5245, + "step": 1320 + }, + { + "epoch": 1.3559322033898304, + "eval_loss": 1.8935153484344482, + "eval_runtime": 51.161, + "eval_samples_per_second": 19.546, + "eval_steps_per_second": 0.821, + "step": 1320 + }, + { + "epoch": 1.356959424756035, + "grad_norm": 2.816667318344116, + "learning_rate": 3.0700418728587746e-06, + "loss": 1.484, + "step": 1321 + }, + { + "epoch": 1.3579866461222394, + "grad_norm": 2.748138427734375, + "learning_rate": 3.0681385610963076e-06, + "loss": 1.5235, + "step": 1322 + }, + { + "epoch": 1.3590138674884438, + "grad_norm": 2.746454954147339, + "learning_rate": 3.066235249333841e-06, + "loss": 1.4929, + "step": 1323 + }, + { + "epoch": 1.3600410888546481, + "grad_norm": 2.8291871547698975, + "learning_rate": 3.0643319375713743e-06, + "loss": 1.5381, + "step": 1324 + }, + { + "epoch": 1.3610683102208525, + "grad_norm": 2.8472836017608643, + "learning_rate": 3.0624286258089076e-06, + "loss": 1.5664, + "step": 1325 + }, + { + "epoch": 1.362095531587057, + "grad_norm": 2.7493903636932373, + "learning_rate": 3.060525314046441e-06, + "loss": 1.5088, + "step": 1326 + }, + { + "epoch": 1.3631227529532615, + "grad_norm": 2.6563873291015625, + "learning_rate": 3.058622002283974e-06, + "loss": 1.4899, + "step": 1327 + }, + { + "epoch": 1.3641499743194658, + "grad_norm": 2.8158602714538574, + "learning_rate": 3.0567186905215077e-06, + "loss": 1.546, + "step": 1328 + }, + { + "epoch": 1.3651771956856702, + "grad_norm": 2.722679615020752, + "learning_rate": 3.054815378759041e-06, + "loss": 1.5572, + "step": 1329 + }, + { + "epoch": 1.3662044170518746, + "grad_norm": 2.572859525680542, + "learning_rate": 3.0529120669965745e-06, + "loss": 1.5441, + "step": 1330 + }, + { + "epoch": 1.3662044170518746, + "eval_loss": 1.8944634199142456, + "eval_runtime": 49.7122, + "eval_samples_per_second": 20.116, + "eval_steps_per_second": 0.845, + "step": 1330 + }, + { + "epoch": 1.3672316384180792, + "grad_norm": 2.6468751430511475, + "learning_rate": 3.051008755234108e-06, + "loss": 1.5507, + "step": 1331 + }, + { + "epoch": 1.3682588597842835, + "grad_norm": 2.618972063064575, + "learning_rate": 3.049105443471641e-06, + "loss": 1.4799, + "step": 1332 + }, + { + "epoch": 1.3692860811504879, + "grad_norm": 2.808328628540039, + "learning_rate": 3.0472021317091745e-06, + "loss": 1.4917, + "step": 1333 + }, + { + "epoch": 1.3703133025166925, + "grad_norm": 2.7831826210021973, + "learning_rate": 3.0452988199467075e-06, + "loss": 1.6242, + "step": 1334 + }, + { + "epoch": 1.3713405238828968, + "grad_norm": 2.6808860301971436, + "learning_rate": 3.043395508184241e-06, + "loss": 1.5455, + "step": 1335 + }, + { + "epoch": 1.3723677452491012, + "grad_norm": 2.5987675189971924, + "learning_rate": 3.041492196421774e-06, + "loss": 1.5406, + "step": 1336 + }, + { + "epoch": 1.3733949666153056, + "grad_norm": 2.6476593017578125, + "learning_rate": 3.0395888846593076e-06, + "loss": 1.487, + "step": 1337 + }, + { + "epoch": 1.37442218798151, + "grad_norm": 2.724715232849121, + "learning_rate": 3.037685572896841e-06, + "loss": 1.5316, + "step": 1338 + }, + { + "epoch": 1.3754494093477145, + "grad_norm": 2.6245994567871094, + "learning_rate": 3.035782261134374e-06, + "loss": 1.5137, + "step": 1339 + }, + { + "epoch": 1.376476630713919, + "grad_norm": 2.676206588745117, + "learning_rate": 3.0338789493719072e-06, + "loss": 1.5321, + "step": 1340 + }, + { + "epoch": 1.376476630713919, + "eval_loss": 1.908184289932251, + "eval_runtime": 49.2092, + "eval_samples_per_second": 20.321, + "eval_steps_per_second": 0.853, + "step": 1340 + }, + { + "epoch": 1.3775038520801233, + "grad_norm": 2.9669554233551025, + "learning_rate": 3.0319756376094406e-06, + "loss": 1.5349, + "step": 1341 + }, + { + "epoch": 1.3785310734463276, + "grad_norm": 2.6755905151367188, + "learning_rate": 3.030072325846974e-06, + "loss": 1.536, + "step": 1342 + }, + { + "epoch": 1.379558294812532, + "grad_norm": 2.5073540210723877, + "learning_rate": 3.0281690140845073e-06, + "loss": 1.5167, + "step": 1343 + }, + { + "epoch": 1.3805855161787366, + "grad_norm": 2.520477533340454, + "learning_rate": 3.0262657023220403e-06, + "loss": 1.4827, + "step": 1344 + }, + { + "epoch": 1.381612737544941, + "grad_norm": 2.6202239990234375, + "learning_rate": 3.0243623905595736e-06, + "loss": 1.5271, + "step": 1345 + }, + { + "epoch": 1.3826399589111453, + "grad_norm": 2.667724609375, + "learning_rate": 3.022459078797107e-06, + "loss": 1.537, + "step": 1346 + }, + { + "epoch": 1.3836671802773497, + "grad_norm": 2.701108455657959, + "learning_rate": 3.0205557670346404e-06, + "loss": 1.53, + "step": 1347 + }, + { + "epoch": 1.384694401643554, + "grad_norm": 2.6228628158569336, + "learning_rate": 3.018652455272174e-06, + "loss": 1.5297, + "step": 1348 + }, + { + "epoch": 1.3857216230097587, + "grad_norm": 2.6589505672454834, + "learning_rate": 3.0167491435097075e-06, + "loss": 1.481, + "step": 1349 + }, + { + "epoch": 1.386748844375963, + "grad_norm": 2.805138349533081, + "learning_rate": 3.014845831747241e-06, + "loss": 1.6176, + "step": 1350 + }, + { + "epoch": 1.386748844375963, + "eval_loss": 1.9109174013137817, + "eval_runtime": 44.6864, + "eval_samples_per_second": 22.378, + "eval_steps_per_second": 0.94, + "step": 1350 + }, + { + "epoch": 1.3877760657421674, + "grad_norm": 2.6587769985198975, + "learning_rate": 3.012942519984774e-06, + "loss": 1.5188, + "step": 1351 + }, + { + "epoch": 1.388803287108372, + "grad_norm": 2.7695798873901367, + "learning_rate": 3.011039208222307e-06, + "loss": 1.4913, + "step": 1352 + }, + { + "epoch": 1.3898305084745763, + "grad_norm": 2.7030696868896484, + "learning_rate": 3.0091358964598405e-06, + "loss": 1.4976, + "step": 1353 + }, + { + "epoch": 1.3908577298407807, + "grad_norm": 2.63368558883667, + "learning_rate": 3.007232584697374e-06, + "loss": 1.4575, + "step": 1354 + }, + { + "epoch": 1.391884951206985, + "grad_norm": 2.6904921531677246, + "learning_rate": 3.0053292729349073e-06, + "loss": 1.5067, + "step": 1355 + }, + { + "epoch": 1.3929121725731894, + "grad_norm": 2.7085046768188477, + "learning_rate": 3.00342596117244e-06, + "loss": 1.5215, + "step": 1356 + }, + { + "epoch": 1.393939393939394, + "grad_norm": 2.744744062423706, + "learning_rate": 3.0015226494099736e-06, + "loss": 1.5213, + "step": 1357 + }, + { + "epoch": 1.3949666153055984, + "grad_norm": 2.8159263134002686, + "learning_rate": 2.999619337647507e-06, + "loss": 1.5091, + "step": 1358 + }, + { + "epoch": 1.3959938366718028, + "grad_norm": 2.7380170822143555, + "learning_rate": 2.9977160258850403e-06, + "loss": 1.4863, + "step": 1359 + }, + { + "epoch": 1.3970210580380071, + "grad_norm": 2.9817709922790527, + "learning_rate": 2.9958127141225736e-06, + "loss": 1.4977, + "step": 1360 + }, + { + "epoch": 1.3970210580380071, + "eval_loss": 1.9042565822601318, + "eval_runtime": 45.13, + "eval_samples_per_second": 22.158, + "eval_steps_per_second": 0.931, + "step": 1360 + }, + { + "epoch": 1.3980482794042115, + "grad_norm": 2.7616889476776123, + "learning_rate": 2.9939094023601066e-06, + "loss": 1.4817, + "step": 1361 + }, + { + "epoch": 1.399075500770416, + "grad_norm": 2.695251226425171, + "learning_rate": 2.99200609059764e-06, + "loss": 1.5512, + "step": 1362 + }, + { + "epoch": 1.4001027221366205, + "grad_norm": 2.645728349685669, + "learning_rate": 2.9901027788351733e-06, + "loss": 1.5232, + "step": 1363 + }, + { + "epoch": 1.4011299435028248, + "grad_norm": 2.5545313358306885, + "learning_rate": 2.9881994670727067e-06, + "loss": 1.4626, + "step": 1364 + }, + { + "epoch": 1.4021571648690292, + "grad_norm": 2.6554672718048096, + "learning_rate": 2.98629615531024e-06, + "loss": 1.4813, + "step": 1365 + }, + { + "epoch": 1.4031843862352336, + "grad_norm": 2.543102979660034, + "learning_rate": 2.984392843547773e-06, + "loss": 1.4954, + "step": 1366 + }, + { + "epoch": 1.4042116076014381, + "grad_norm": 2.5635626316070557, + "learning_rate": 2.9824895317853063e-06, + "loss": 1.4858, + "step": 1367 + }, + { + "epoch": 1.4052388289676425, + "grad_norm": 2.60005259513855, + "learning_rate": 2.98058622002284e-06, + "loss": 1.4991, + "step": 1368 + }, + { + "epoch": 1.4062660503338469, + "grad_norm": 2.664973020553589, + "learning_rate": 2.9786829082603735e-06, + "loss": 1.5083, + "step": 1369 + }, + { + "epoch": 1.4072932717000515, + "grad_norm": 2.642303466796875, + "learning_rate": 2.976779596497907e-06, + "loss": 1.5636, + "step": 1370 + }, + { + "epoch": 1.4072932717000515, + "eval_loss": 1.9043673276901245, + "eval_runtime": 45.8156, + "eval_samples_per_second": 21.827, + "eval_steps_per_second": 0.917, + "step": 1370 + }, + { + "epoch": 1.4083204930662558, + "grad_norm": 2.721719264984131, + "learning_rate": 2.97487628473544e-06, + "loss": 1.5627, + "step": 1371 + }, + { + "epoch": 1.4093477144324602, + "grad_norm": 2.665929079055786, + "learning_rate": 2.9729729729729736e-06, + "loss": 1.5684, + "step": 1372 + }, + { + "epoch": 1.4103749357986646, + "grad_norm": 2.672046661376953, + "learning_rate": 2.9710696612105065e-06, + "loss": 1.518, + "step": 1373 + }, + { + "epoch": 1.411402157164869, + "grad_norm": 2.675283432006836, + "learning_rate": 2.96916634944804e-06, + "loss": 1.5604, + "step": 1374 + }, + { + "epoch": 1.4124293785310735, + "grad_norm": 2.6286461353302, + "learning_rate": 2.9672630376855732e-06, + "loss": 1.5003, + "step": 1375 + }, + { + "epoch": 1.413456599897278, + "grad_norm": 2.6903879642486572, + "learning_rate": 2.9653597259231066e-06, + "loss": 1.5601, + "step": 1376 + }, + { + "epoch": 1.4144838212634823, + "grad_norm": 2.7762627601623535, + "learning_rate": 2.96345641416064e-06, + "loss": 1.5326, + "step": 1377 + }, + { + "epoch": 1.4155110426296866, + "grad_norm": 2.758028030395508, + "learning_rate": 2.961553102398173e-06, + "loss": 1.4837, + "step": 1378 + }, + { + "epoch": 1.416538263995891, + "grad_norm": 2.769880771636963, + "learning_rate": 2.9596497906357063e-06, + "loss": 1.4122, + "step": 1379 + }, + { + "epoch": 1.4175654853620956, + "grad_norm": 2.815645217895508, + "learning_rate": 2.9577464788732396e-06, + "loss": 1.555, + "step": 1380 + }, + { + "epoch": 1.4175654853620956, + "eval_loss": 1.9081141948699951, + "eval_runtime": 44.267, + "eval_samples_per_second": 22.59, + "eval_steps_per_second": 0.949, + "step": 1380 + }, + { + "epoch": 1.4185927067283, + "grad_norm": 2.8118088245391846, + "learning_rate": 2.955843167110773e-06, + "loss": 1.5708, + "step": 1381 + }, + { + "epoch": 1.4196199280945043, + "grad_norm": 2.620800495147705, + "learning_rate": 2.9539398553483064e-06, + "loss": 1.4901, + "step": 1382 + }, + { + "epoch": 1.420647149460709, + "grad_norm": 2.6867423057556152, + "learning_rate": 2.9520365435858393e-06, + "loss": 1.4964, + "step": 1383 + }, + { + "epoch": 1.4216743708269133, + "grad_norm": 2.569495916366577, + "learning_rate": 2.9501332318233727e-06, + "loss": 1.5103, + "step": 1384 + }, + { + "epoch": 1.4227015921931176, + "grad_norm": 2.6389119625091553, + "learning_rate": 2.948229920060906e-06, + "loss": 1.5946, + "step": 1385 + }, + { + "epoch": 1.423728813559322, + "grad_norm": 2.767986536026001, + "learning_rate": 2.9463266082984394e-06, + "loss": 1.5296, + "step": 1386 + }, + { + "epoch": 1.4247560349255264, + "grad_norm": 2.5122997760772705, + "learning_rate": 2.9444232965359727e-06, + "loss": 1.4941, + "step": 1387 + }, + { + "epoch": 1.425783256291731, + "grad_norm": 2.7231523990631104, + "learning_rate": 2.9425199847735057e-06, + "loss": 1.5184, + "step": 1388 + }, + { + "epoch": 1.4268104776579353, + "grad_norm": 2.7313649654388428, + "learning_rate": 2.94061667301104e-06, + "loss": 1.5368, + "step": 1389 + }, + { + "epoch": 1.4278376990241397, + "grad_norm": 2.656062602996826, + "learning_rate": 2.938713361248573e-06, + "loss": 1.4958, + "step": 1390 + }, + { + "epoch": 1.4278376990241397, + "eval_loss": 1.9070981740951538, + "eval_runtime": 43.6226, + "eval_samples_per_second": 22.924, + "eval_steps_per_second": 0.963, + "step": 1390 + }, + { + "epoch": 1.428864920390344, + "grad_norm": 2.6621580123901367, + "learning_rate": 2.936810049486106e-06, + "loss": 1.4315, + "step": 1391 + }, + { + "epoch": 1.4298921417565484, + "grad_norm": 2.7683629989624023, + "learning_rate": 2.9349067377236396e-06, + "loss": 1.5042, + "step": 1392 + }, + { + "epoch": 1.430919363122753, + "grad_norm": 2.695810317993164, + "learning_rate": 2.933003425961173e-06, + "loss": 1.5174, + "step": 1393 + }, + { + "epoch": 1.4319465844889574, + "grad_norm": 2.686978816986084, + "learning_rate": 2.9311001141987063e-06, + "loss": 1.4706, + "step": 1394 + }, + { + "epoch": 1.4329738058551618, + "grad_norm": 2.799234628677368, + "learning_rate": 2.9291968024362392e-06, + "loss": 1.4826, + "step": 1395 + }, + { + "epoch": 1.4340010272213661, + "grad_norm": 2.8211469650268555, + "learning_rate": 2.9272934906737726e-06, + "loss": 1.488, + "step": 1396 + }, + { + "epoch": 1.4350282485875705, + "grad_norm": 2.6225461959838867, + "learning_rate": 2.925390178911306e-06, + "loss": 1.4699, + "step": 1397 + }, + { + "epoch": 1.436055469953775, + "grad_norm": 2.6853697299957275, + "learning_rate": 2.9234868671488393e-06, + "loss": 1.4787, + "step": 1398 + }, + { + "epoch": 1.4370826913199795, + "grad_norm": 2.784473419189453, + "learning_rate": 2.9215835553863727e-06, + "loss": 1.5043, + "step": 1399 + }, + { + "epoch": 1.4381099126861838, + "grad_norm": 2.706911087036133, + "learning_rate": 2.9196802436239056e-06, + "loss": 1.512, + "step": 1400 + }, + { + "epoch": 1.4381099126861838, + "eval_loss": 1.9136111736297607, + "eval_runtime": 43.5675, + "eval_samples_per_second": 22.953, + "eval_steps_per_second": 0.964, + "step": 1400 + }, + { + "epoch": 1.4391371340523884, + "grad_norm": 2.6579840183258057, + "learning_rate": 2.917776931861439e-06, + "loss": 1.5073, + "step": 1401 + }, + { + "epoch": 1.4401643554185928, + "grad_norm": 2.7359421253204346, + "learning_rate": 2.9158736200989723e-06, + "loss": 1.4819, + "step": 1402 + }, + { + "epoch": 1.4411915767847971, + "grad_norm": 2.604591131210327, + "learning_rate": 2.9139703083365057e-06, + "loss": 1.4758, + "step": 1403 + }, + { + "epoch": 1.4422187981510015, + "grad_norm": 2.7267162799835205, + "learning_rate": 2.912066996574039e-06, + "loss": 1.4835, + "step": 1404 + }, + { + "epoch": 1.4432460195172059, + "grad_norm": 2.844331741333008, + "learning_rate": 2.910163684811572e-06, + "loss": 1.5096, + "step": 1405 + }, + { + "epoch": 1.4442732408834105, + "grad_norm": 2.6915996074676514, + "learning_rate": 2.9082603730491054e-06, + "loss": 1.4536, + "step": 1406 + }, + { + "epoch": 1.4453004622496148, + "grad_norm": 2.9222843647003174, + "learning_rate": 2.9063570612866387e-06, + "loss": 1.5769, + "step": 1407 + }, + { + "epoch": 1.4463276836158192, + "grad_norm": 2.667371988296509, + "learning_rate": 2.904453749524172e-06, + "loss": 1.533, + "step": 1408 + }, + { + "epoch": 1.4473549049820236, + "grad_norm": 2.643632411956787, + "learning_rate": 2.902550437761706e-06, + "loss": 1.4666, + "step": 1409 + }, + { + "epoch": 1.448382126348228, + "grad_norm": 2.8633203506469727, + "learning_rate": 2.9006471259992392e-06, + "loss": 1.4826, + "step": 1410 + }, + { + "epoch": 1.448382126348228, + "eval_loss": 1.9127975702285767, + "eval_runtime": 42.2168, + "eval_samples_per_second": 23.687, + "eval_steps_per_second": 0.995, + "step": 1410 + }, + { + "epoch": 1.4494093477144325, + "grad_norm": 2.7763702869415283, + "learning_rate": 2.8987438142367726e-06, + "loss": 1.4969, + "step": 1411 + }, + { + "epoch": 1.450436569080637, + "grad_norm": 2.598337411880493, + "learning_rate": 2.8968405024743055e-06, + "loss": 1.4993, + "step": 1412 + }, + { + "epoch": 1.4514637904468413, + "grad_norm": 2.717944622039795, + "learning_rate": 2.894937190711839e-06, + "loss": 1.5479, + "step": 1413 + }, + { + "epoch": 1.4524910118130456, + "grad_norm": 2.5288076400756836, + "learning_rate": 2.8930338789493723e-06, + "loss": 1.4564, + "step": 1414 + }, + { + "epoch": 1.45351823317925, + "grad_norm": 2.6042704582214355, + "learning_rate": 2.8911305671869056e-06, + "loss": 1.5267, + "step": 1415 + }, + { + "epoch": 1.4545454545454546, + "grad_norm": 2.5768539905548096, + "learning_rate": 2.889227255424439e-06, + "loss": 1.4958, + "step": 1416 + }, + { + "epoch": 1.455572675911659, + "grad_norm": 2.62872576713562, + "learning_rate": 2.887323943661972e-06, + "loss": 1.4407, + "step": 1417 + }, + { + "epoch": 1.4565998972778633, + "grad_norm": 2.8656342029571533, + "learning_rate": 2.8854206318995053e-06, + "loss": 1.4944, + "step": 1418 + }, + { + "epoch": 1.457627118644068, + "grad_norm": 2.7083373069763184, + "learning_rate": 2.8835173201370386e-06, + "loss": 1.474, + "step": 1419 + }, + { + "epoch": 1.4586543400102723, + "grad_norm": 2.712958812713623, + "learning_rate": 2.881614008374572e-06, + "loss": 1.4982, + "step": 1420 + }, + { + "epoch": 1.4586543400102723, + "eval_loss": 1.9154274463653564, + "eval_runtime": 45.8592, + "eval_samples_per_second": 21.806, + "eval_steps_per_second": 0.916, + "step": 1420 + }, + { + "epoch": 1.4596815613764766, + "grad_norm": 3.0084564685821533, + "learning_rate": 2.8797106966121054e-06, + "loss": 1.5505, + "step": 1421 + }, + { + "epoch": 1.460708782742681, + "grad_norm": 2.793546438217163, + "learning_rate": 2.8778073848496383e-06, + "loss": 1.5203, + "step": 1422 + }, + { + "epoch": 1.4617360041088854, + "grad_norm": 2.5944628715515137, + "learning_rate": 2.8759040730871717e-06, + "loss": 1.5016, + "step": 1423 + }, + { + "epoch": 1.46276322547509, + "grad_norm": 2.7558817863464355, + "learning_rate": 2.874000761324705e-06, + "loss": 1.5273, + "step": 1424 + }, + { + "epoch": 1.4637904468412943, + "grad_norm": 2.582310676574707, + "learning_rate": 2.8720974495622384e-06, + "loss": 1.4862, + "step": 1425 + }, + { + "epoch": 1.4648176682074987, + "grad_norm": 2.639944314956665, + "learning_rate": 2.8701941377997718e-06, + "loss": 1.5276, + "step": 1426 + }, + { + "epoch": 1.465844889573703, + "grad_norm": 2.613696575164795, + "learning_rate": 2.8682908260373047e-06, + "loss": 1.5061, + "step": 1427 + }, + { + "epoch": 1.4668721109399074, + "grad_norm": 2.7124717235565186, + "learning_rate": 2.866387514274838e-06, + "loss": 1.5517, + "step": 1428 + }, + { + "epoch": 1.467899332306112, + "grad_norm": 2.7124717235565186, + "learning_rate": 2.866387514274838e-06, + "loss": 1.5288, + "step": 1429 + }, + { + "epoch": 1.4689265536723164, + "grad_norm": 2.722407102584839, + "learning_rate": 2.864484202512372e-06, + "loss": 1.5256, + "step": 1430 + }, + { + "epoch": 1.4689265536723164, + "eval_loss": 1.900185227394104, + "eval_runtime": 47.7724, + "eval_samples_per_second": 20.933, + "eval_steps_per_second": 0.879, + "step": 1430 + }, + { + "epoch": 1.4699537750385208, + "grad_norm": 2.690464973449707, + "learning_rate": 2.8625808907499052e-06, + "loss": 1.4841, + "step": 1431 + }, + { + "epoch": 1.4709809964047253, + "grad_norm": 2.766082525253296, + "learning_rate": 2.8606775789874386e-06, + "loss": 1.5171, + "step": 1432 + }, + { + "epoch": 1.4720082177709297, + "grad_norm": 2.6420655250549316, + "learning_rate": 2.858774267224972e-06, + "loss": 1.4871, + "step": 1433 + }, + { + "epoch": 1.473035439137134, + "grad_norm": 2.687546730041504, + "learning_rate": 2.8568709554625053e-06, + "loss": 1.5116, + "step": 1434 + }, + { + "epoch": 1.4740626605033385, + "grad_norm": 2.6613612174987793, + "learning_rate": 2.8549676437000382e-06, + "loss": 1.5599, + "step": 1435 + }, + { + "epoch": 1.4750898818695428, + "grad_norm": 2.7249948978424072, + "learning_rate": 2.8530643319375716e-06, + "loss": 1.5342, + "step": 1436 + }, + { + "epoch": 1.4761171032357474, + "grad_norm": 2.768580913543701, + "learning_rate": 2.851161020175105e-06, + "loss": 1.5022, + "step": 1437 + }, + { + "epoch": 1.4771443246019518, + "grad_norm": 2.6677844524383545, + "learning_rate": 2.8492577084126383e-06, + "loss": 1.4712, + "step": 1438 + }, + { + "epoch": 1.4781715459681561, + "grad_norm": 2.693856716156006, + "learning_rate": 2.8473543966501717e-06, + "loss": 1.5163, + "step": 1439 + }, + { + "epoch": 1.4791987673343605, + "grad_norm": 2.667125701904297, + "learning_rate": 2.8454510848877046e-06, + "loss": 1.5601, + "step": 1440 + }, + { + "epoch": 1.4791987673343605, + "eval_loss": 1.8993381261825562, + "eval_runtime": 47.2966, + "eval_samples_per_second": 21.143, + "eval_steps_per_second": 0.888, + "step": 1440 + }, + { + "epoch": 1.4802259887005649, + "grad_norm": 2.6483078002929688, + "learning_rate": 2.843547773125238e-06, + "loss": 1.5042, + "step": 1441 + }, + { + "epoch": 1.4812532100667695, + "grad_norm": 2.6776845455169678, + "learning_rate": 2.8416444613627714e-06, + "loss": 1.5096, + "step": 1442 + }, + { + "epoch": 1.4822804314329738, + "grad_norm": 2.7866480350494385, + "learning_rate": 2.8397411496003047e-06, + "loss": 1.4684, + "step": 1443 + }, + { + "epoch": 1.4833076527991782, + "grad_norm": 2.844266176223755, + "learning_rate": 2.837837837837838e-06, + "loss": 1.499, + "step": 1444 + }, + { + "epoch": 1.4843348741653826, + "grad_norm": 2.5618107318878174, + "learning_rate": 2.835934526075371e-06, + "loss": 1.5212, + "step": 1445 + }, + { + "epoch": 1.485362095531587, + "grad_norm": 2.6264190673828125, + "learning_rate": 2.8340312143129044e-06, + "loss": 1.4763, + "step": 1446 + }, + { + "epoch": 1.4863893168977915, + "grad_norm": 2.68026065826416, + "learning_rate": 2.8321279025504377e-06, + "loss": 1.4572, + "step": 1447 + }, + { + "epoch": 1.487416538263996, + "grad_norm": 2.7439587116241455, + "learning_rate": 2.830224590787971e-06, + "loss": 1.5778, + "step": 1448 + }, + { + "epoch": 1.4884437596302003, + "grad_norm": 2.708627700805664, + "learning_rate": 2.8283212790255045e-06, + "loss": 1.5177, + "step": 1449 + }, + { + "epoch": 1.4894709809964048, + "grad_norm": 2.5931451320648193, + "learning_rate": 2.8264179672630383e-06, + "loss": 1.5499, + "step": 1450 + }, + { + "epoch": 1.4894709809964048, + "eval_loss": 1.8863205909729004, + "eval_runtime": 45.8354, + "eval_samples_per_second": 21.817, + "eval_steps_per_second": 0.916, + "step": 1450 + }, + { + "epoch": 1.4904982023626092, + "grad_norm": 2.6775002479553223, + "learning_rate": 2.8245146555005716e-06, + "loss": 1.4987, + "step": 1451 + }, + { + "epoch": 1.4915254237288136, + "grad_norm": 2.568394660949707, + "learning_rate": 2.8226113437381046e-06, + "loss": 1.449, + "step": 1452 + }, + { + "epoch": 1.492552645095018, + "grad_norm": 2.6354329586029053, + "learning_rate": 2.820708031975638e-06, + "loss": 1.5018, + "step": 1453 + }, + { + "epoch": 1.4935798664612223, + "grad_norm": 2.731614112854004, + "learning_rate": 2.8188047202131713e-06, + "loss": 1.5337, + "step": 1454 + }, + { + "epoch": 1.494607087827427, + "grad_norm": 2.662299633026123, + "learning_rate": 2.8169014084507046e-06, + "loss": 1.5449, + "step": 1455 + }, + { + "epoch": 1.4956343091936313, + "grad_norm": 2.6484053134918213, + "learning_rate": 2.814998096688238e-06, + "loss": 1.5048, + "step": 1456 + }, + { + "epoch": 1.4966615305598356, + "grad_norm": 2.5573649406433105, + "learning_rate": 2.813094784925771e-06, + "loss": 1.4417, + "step": 1457 + }, + { + "epoch": 1.49768875192604, + "grad_norm": 2.6440303325653076, + "learning_rate": 2.8111914731633043e-06, + "loss": 1.5545, + "step": 1458 + }, + { + "epoch": 1.4987159732922444, + "grad_norm": 2.6014726161956787, + "learning_rate": 2.8092881614008377e-06, + "loss": 1.4995, + "step": 1459 + }, + { + "epoch": 1.499743194658449, + "grad_norm": 2.8325798511505127, + "learning_rate": 2.807384849638371e-06, + "loss": 1.4762, + "step": 1460 + }, + { + "epoch": 1.499743194658449, + "eval_loss": 1.8842575550079346, + "eval_runtime": 46.058, + "eval_samples_per_second": 21.712, + "eval_steps_per_second": 0.912, + "step": 1460 + }, + { + "epoch": 1.5007704160246533, + "grad_norm": 2.629534959793091, + "learning_rate": 2.8054815378759044e-06, + "loss": 1.5235, + "step": 1461 + }, + { + "epoch": 1.5017976373908577, + "grad_norm": 2.723256826400757, + "learning_rate": 2.8035782261134378e-06, + "loss": 1.4962, + "step": 1462 + }, + { + "epoch": 1.5028248587570623, + "grad_norm": 2.7497754096984863, + "learning_rate": 2.8016749143509707e-06, + "loss": 1.4931, + "step": 1463 + }, + { + "epoch": 1.5038520801232664, + "grad_norm": 2.695408821105957, + "learning_rate": 2.799771602588504e-06, + "loss": 1.4921, + "step": 1464 + }, + { + "epoch": 1.504879301489471, + "grad_norm": 2.6144590377807617, + "learning_rate": 2.7978682908260374e-06, + "loss": 1.5191, + "step": 1465 + }, + { + "epoch": 1.5059065228556754, + "grad_norm": 2.7783288955688477, + "learning_rate": 2.795964979063571e-06, + "loss": 1.5416, + "step": 1466 + }, + { + "epoch": 1.5069337442218798, + "grad_norm": 2.7083182334899902, + "learning_rate": 2.794061667301104e-06, + "loss": 1.5126, + "step": 1467 + }, + { + "epoch": 1.5079609655880843, + "grad_norm": 2.7532167434692383, + "learning_rate": 2.792158355538637e-06, + "loss": 1.5244, + "step": 1468 + }, + { + "epoch": 1.5089881869542885, + "grad_norm": 2.6125259399414062, + "learning_rate": 2.7902550437761705e-06, + "loss": 1.5194, + "step": 1469 + }, + { + "epoch": 1.510015408320493, + "grad_norm": 2.585392713546753, + "learning_rate": 2.7883517320137042e-06, + "loss": 1.474, + "step": 1470 + }, + { + "epoch": 1.510015408320493, + "eval_loss": 1.8812134265899658, + "eval_runtime": 46.7392, + "eval_samples_per_second": 21.395, + "eval_steps_per_second": 0.899, + "step": 1470 + }, + { + "epoch": 1.5110426296866974, + "grad_norm": 2.586333751678467, + "learning_rate": 2.7864484202512376e-06, + "loss": 1.4666, + "step": 1471 + }, + { + "epoch": 1.5120698510529018, + "grad_norm": 2.7523810863494873, + "learning_rate": 2.784545108488771e-06, + "loss": 1.5113, + "step": 1472 + }, + { + "epoch": 1.5130970724191064, + "grad_norm": 2.688206434249878, + "learning_rate": 2.7826417967263043e-06, + "loss": 1.5049, + "step": 1473 + }, + { + "epoch": 1.5141242937853108, + "grad_norm": 2.642054319381714, + "learning_rate": 2.7807384849638373e-06, + "loss": 1.5534, + "step": 1474 + }, + { + "epoch": 1.5151515151515151, + "grad_norm": 2.694291591644287, + "learning_rate": 2.7788351732013706e-06, + "loss": 1.5052, + "step": 1475 + }, + { + "epoch": 1.5161787365177197, + "grad_norm": 2.6969082355499268, + "learning_rate": 2.776931861438904e-06, + "loss": 1.491, + "step": 1476 + }, + { + "epoch": 1.5172059578839239, + "grad_norm": 2.6937081813812256, + "learning_rate": 2.7750285496764374e-06, + "loss": 1.5081, + "step": 1477 + }, + { + "epoch": 1.5182331792501285, + "grad_norm": 2.6754133701324463, + "learning_rate": 2.7731252379139707e-06, + "loss": 1.5107, + "step": 1478 + }, + { + "epoch": 1.5192604006163328, + "grad_norm": 2.586491346359253, + "learning_rate": 2.771221926151504e-06, + "loss": 1.4486, + "step": 1479 + }, + { + "epoch": 1.5202876219825372, + "grad_norm": 2.82452130317688, + "learning_rate": 2.769318614389037e-06, + "loss": 1.5089, + "step": 1480 + }, + { + "epoch": 1.5202876219825372, + "eval_loss": 1.879558801651001, + "eval_runtime": 45.9914, + "eval_samples_per_second": 21.743, + "eval_steps_per_second": 0.913, + "step": 1480 + }, + { + "epoch": 1.5213148433487418, + "grad_norm": 2.69734263420105, + "learning_rate": 2.7674153026265704e-06, + "loss": 1.5069, + "step": 1481 + }, + { + "epoch": 1.522342064714946, + "grad_norm": 2.7553112506866455, + "learning_rate": 2.7655119908641037e-06, + "loss": 1.5211, + "step": 1482 + }, + { + "epoch": 1.5233692860811505, + "grad_norm": 2.659170389175415, + "learning_rate": 2.763608679101637e-06, + "loss": 1.4967, + "step": 1483 + }, + { + "epoch": 1.5243965074473549, + "grad_norm": 2.6266300678253174, + "learning_rate": 2.7617053673391705e-06, + "loss": 1.5376, + "step": 1484 + }, + { + "epoch": 1.5254237288135593, + "grad_norm": 2.6549487113952637, + "learning_rate": 2.7598020555767034e-06, + "loss": 1.546, + "step": 1485 + }, + { + "epoch": 1.5264509501797638, + "grad_norm": 2.651841402053833, + "learning_rate": 2.7578987438142368e-06, + "loss": 1.4625, + "step": 1486 + }, + { + "epoch": 1.5274781715459682, + "grad_norm": 2.783480644226074, + "learning_rate": 2.75599543205177e-06, + "loss": 1.5165, + "step": 1487 + }, + { + "epoch": 1.5285053929121726, + "grad_norm": 2.674964189529419, + "learning_rate": 2.7540921202893035e-06, + "loss": 1.539, + "step": 1488 + }, + { + "epoch": 1.529532614278377, + "grad_norm": 2.7219326496124268, + "learning_rate": 2.752188808526837e-06, + "loss": 1.5151, + "step": 1489 + }, + { + "epoch": 1.5305598356445813, + "grad_norm": 2.6149561405181885, + "learning_rate": 2.75028549676437e-06, + "loss": 1.498, + "step": 1490 + }, + { + "epoch": 1.5305598356445813, + "eval_loss": 1.8944125175476074, + "eval_runtime": 45.5669, + "eval_samples_per_second": 21.946, + "eval_steps_per_second": 0.922, + "step": 1490 + }, + { + "epoch": 1.531587057010786, + "grad_norm": 2.6966514587402344, + "learning_rate": 2.7483821850019036e-06, + "loss": 1.5026, + "step": 1491 + }, + { + "epoch": 1.5326142783769903, + "grad_norm": 2.7445616722106934, + "learning_rate": 2.746478873239437e-06, + "loss": 1.493, + "step": 1492 + }, + { + "epoch": 1.5336414997431946, + "grad_norm": 2.8391168117523193, + "learning_rate": 2.7445755614769703e-06, + "loss": 1.4918, + "step": 1493 + }, + { + "epoch": 1.5346687211093992, + "grad_norm": 2.79190731048584, + "learning_rate": 2.7426722497145037e-06, + "loss": 1.4654, + "step": 1494 + }, + { + "epoch": 1.5356959424756034, + "grad_norm": 2.7497293949127197, + "learning_rate": 2.740768937952037e-06, + "loss": 1.4702, + "step": 1495 + }, + { + "epoch": 1.536723163841808, + "grad_norm": 2.7948496341705322, + "learning_rate": 2.7388656261895704e-06, + "loss": 1.4879, + "step": 1496 + }, + { + "epoch": 1.5377503852080123, + "grad_norm": 2.9910364151000977, + "learning_rate": 2.7369623144271033e-06, + "loss": 1.5689, + "step": 1497 + }, + { + "epoch": 1.5387776065742167, + "grad_norm": 2.726814031600952, + "learning_rate": 2.7350590026646367e-06, + "loss": 1.492, + "step": 1498 + }, + { + "epoch": 1.5398048279404213, + "grad_norm": 2.839493751525879, + "learning_rate": 2.73315569090217e-06, + "loss": 1.4539, + "step": 1499 + }, + { + "epoch": 1.5408320493066254, + "grad_norm": 2.776416063308716, + "learning_rate": 2.7312523791397034e-06, + "loss": 1.5509, + "step": 1500 + }, + { + "epoch": 1.5408320493066254, + "eval_loss": 1.9009771347045898, + "eval_runtime": 51.232, + "eval_samples_per_second": 19.519, + "eval_steps_per_second": 0.82, + "step": 1500 + }, + { + "epoch": 1.54185927067283, + "grad_norm": 2.8426146507263184, + "learning_rate": 2.7293490673772368e-06, + "loss": 1.5696, + "step": 1501 + }, + { + "epoch": 1.5428864920390344, + "grad_norm": 2.8076331615448, + "learning_rate": 2.7274457556147697e-06, + "loss": 1.5216, + "step": 1502 + }, + { + "epoch": 1.5439137134052388, + "grad_norm": 2.800166368484497, + "learning_rate": 2.725542443852303e-06, + "loss": 1.5094, + "step": 1503 + }, + { + "epoch": 1.5449409347714433, + "grad_norm": 2.597820281982422, + "learning_rate": 2.7236391320898365e-06, + "loss": 1.5149, + "step": 1504 + }, + { + "epoch": 1.5459681561376477, + "grad_norm": 2.557790517807007, + "learning_rate": 2.72173582032737e-06, + "loss": 1.4817, + "step": 1505 + }, + { + "epoch": 1.546995377503852, + "grad_norm": 2.6866400241851807, + "learning_rate": 2.719832508564903e-06, + "loss": 1.5024, + "step": 1506 + }, + { + "epoch": 1.5480225988700564, + "grad_norm": 2.739360809326172, + "learning_rate": 2.717929196802436e-06, + "loss": 1.5151, + "step": 1507 + }, + { + "epoch": 1.5490498202362608, + "grad_norm": 2.694063186645508, + "learning_rate": 2.7160258850399695e-06, + "loss": 1.4815, + "step": 1508 + }, + { + "epoch": 1.5500770416024654, + "grad_norm": 2.4691152572631836, + "learning_rate": 2.714122573277503e-06, + "loss": 1.4989, + "step": 1509 + }, + { + "epoch": 1.5511042629686698, + "grad_norm": 2.594684362411499, + "learning_rate": 2.712219261515036e-06, + "loss": 1.499, + "step": 1510 + }, + { + "epoch": 1.5511042629686698, + "eval_loss": 1.891904592514038, + "eval_runtime": 44.1399, + "eval_samples_per_second": 22.655, + "eval_steps_per_second": 0.952, + "step": 1510 + }, + { + "epoch": 1.5521314843348741, + "grad_norm": 2.621588706970215, + "learning_rate": 2.71031594975257e-06, + "loss": 1.516, + "step": 1511 + }, + { + "epoch": 1.5531587057010787, + "grad_norm": 2.677280902862549, + "learning_rate": 2.7084126379901034e-06, + "loss": 1.5192, + "step": 1512 + }, + { + "epoch": 1.5541859270672829, + "grad_norm": 2.7408759593963623, + "learning_rate": 2.7065093262276367e-06, + "loss": 1.5377, + "step": 1513 + }, + { + "epoch": 1.5552131484334875, + "grad_norm": 2.622382402420044, + "learning_rate": 2.7046060144651697e-06, + "loss": 1.5437, + "step": 1514 + }, + { + "epoch": 1.5562403697996918, + "grad_norm": 2.8121471405029297, + "learning_rate": 2.702702702702703e-06, + "loss": 1.5698, + "step": 1515 + }, + { + "epoch": 1.5572675911658962, + "grad_norm": 2.5897836685180664, + "learning_rate": 2.7007993909402364e-06, + "loss": 1.4295, + "step": 1516 + }, + { + "epoch": 1.5582948125321008, + "grad_norm": 2.793713092803955, + "learning_rate": 2.6988960791777697e-06, + "loss": 1.5575, + "step": 1517 + }, + { + "epoch": 1.559322033898305, + "grad_norm": 2.767385244369507, + "learning_rate": 2.696992767415303e-06, + "loss": 1.5007, + "step": 1518 + }, + { + "epoch": 1.5603492552645095, + "grad_norm": 2.6531004905700684, + "learning_rate": 2.695089455652836e-06, + "loss": 1.4977, + "step": 1519 + }, + { + "epoch": 1.5613764766307139, + "grad_norm": 2.686450719833374, + "learning_rate": 2.6931861438903694e-06, + "loss": 1.479, + "step": 1520 + }, + { + "epoch": 1.5613764766307139, + "eval_loss": 1.88840651512146, + "eval_runtime": 44.5631, + "eval_samples_per_second": 22.44, + "eval_steps_per_second": 0.942, + "step": 1520 + }, + { + "epoch": 1.5624036979969183, + "grad_norm": 2.6664063930511475, + "learning_rate": 2.6912828321279028e-06, + "loss": 1.5267, + "step": 1521 + }, + { + "epoch": 1.5634309193631228, + "grad_norm": 2.676542282104492, + "learning_rate": 2.689379520365436e-06, + "loss": 1.5005, + "step": 1522 + }, + { + "epoch": 1.5644581407293272, + "grad_norm": 2.8331971168518066, + "learning_rate": 2.6874762086029695e-06, + "loss": 1.5308, + "step": 1523 + }, + { + "epoch": 1.5654853620955316, + "grad_norm": 2.6427881717681885, + "learning_rate": 2.6855728968405024e-06, + "loss": 1.4813, + "step": 1524 + }, + { + "epoch": 1.5665125834617362, + "grad_norm": 2.7058424949645996, + "learning_rate": 2.683669585078036e-06, + "loss": 1.5149, + "step": 1525 + }, + { + "epoch": 1.5675398048279403, + "grad_norm": 2.7069029808044434, + "learning_rate": 2.681766273315569e-06, + "loss": 1.4588, + "step": 1526 + }, + { + "epoch": 1.568567026194145, + "grad_norm": 2.7757389545440674, + "learning_rate": 2.6798629615531025e-06, + "loss": 1.4953, + "step": 1527 + }, + { + "epoch": 1.5695942475603493, + "grad_norm": 2.788466215133667, + "learning_rate": 2.677959649790636e-06, + "loss": 1.5028, + "step": 1528 + }, + { + "epoch": 1.5706214689265536, + "grad_norm": 2.659156322479248, + "learning_rate": 2.676056338028169e-06, + "loss": 1.5365, + "step": 1529 + }, + { + "epoch": 1.5716486902927582, + "grad_norm": 2.7179627418518066, + "learning_rate": 2.674153026265702e-06, + "loss": 1.5232, + "step": 1530 + }, + { + "epoch": 1.5716486902927582, + "eval_loss": 1.8908981084823608, + "eval_runtime": 44.1745, + "eval_samples_per_second": 22.637, + "eval_steps_per_second": 0.951, + "step": 1530 + }, + { + "epoch": 1.5726759116589624, + "grad_norm": 2.676546812057495, + "learning_rate": 2.672249714503236e-06, + "loss": 1.5247, + "step": 1531 + }, + { + "epoch": 1.573703133025167, + "grad_norm": 2.629784345626831, + "learning_rate": 2.6703464027407693e-06, + "loss": 1.488, + "step": 1532 + }, + { + "epoch": 1.5747303543913713, + "grad_norm": 2.7101712226867676, + "learning_rate": 2.6684430909783027e-06, + "loss": 1.5135, + "step": 1533 + }, + { + "epoch": 1.5757575757575757, + "grad_norm": 2.8025403022766113, + "learning_rate": 2.666539779215836e-06, + "loss": 1.5024, + "step": 1534 + }, + { + "epoch": 1.5767847971237803, + "grad_norm": 2.6695220470428467, + "learning_rate": 2.6646364674533694e-06, + "loss": 1.4844, + "step": 1535 + }, + { + "epoch": 1.5778120184899846, + "grad_norm": 2.610874652862549, + "learning_rate": 2.6627331556909024e-06, + "loss": 1.3949, + "step": 1536 + }, + { + "epoch": 1.578839239856189, + "grad_norm": 2.926985025405884, + "learning_rate": 2.6608298439284357e-06, + "loss": 1.4941, + "step": 1537 + }, + { + "epoch": 1.5798664612223934, + "grad_norm": 2.610355854034424, + "learning_rate": 2.658926532165969e-06, + "loss": 1.4895, + "step": 1538 + }, + { + "epoch": 1.5808936825885977, + "grad_norm": 2.6426877975463867, + "learning_rate": 2.6570232204035024e-06, + "loss": 1.4483, + "step": 1539 + }, + { + "epoch": 1.5819209039548023, + "grad_norm": 2.7410695552825928, + "learning_rate": 2.655119908641036e-06, + "loss": 1.4931, + "step": 1540 + }, + { + "epoch": 1.5819209039548023, + "eval_loss": 1.903024673461914, + "eval_runtime": 43.387, + "eval_samples_per_second": 23.048, + "eval_steps_per_second": 0.968, + "step": 1540 + }, + { + "epoch": 1.5829481253210067, + "grad_norm": 2.6797378063201904, + "learning_rate": 2.6532165968785687e-06, + "loss": 1.497, + "step": 1541 + }, + { + "epoch": 1.583975346687211, + "grad_norm": 2.697439670562744, + "learning_rate": 2.651313285116102e-06, + "loss": 1.4596, + "step": 1542 + }, + { + "epoch": 1.5850025680534157, + "grad_norm": 2.6819119453430176, + "learning_rate": 2.6494099733536355e-06, + "loss": 1.4712, + "step": 1543 + }, + { + "epoch": 1.5860297894196198, + "grad_norm": 2.9221737384796143, + "learning_rate": 2.647506661591169e-06, + "loss": 1.6449, + "step": 1544 + }, + { + "epoch": 1.5870570107858244, + "grad_norm": 2.757866859436035, + "learning_rate": 2.645603349828702e-06, + "loss": 1.5251, + "step": 1545 + }, + { + "epoch": 1.5880842321520288, + "grad_norm": 2.693976402282715, + "learning_rate": 2.643700038066235e-06, + "loss": 1.462, + "step": 1546 + }, + { + "epoch": 1.5891114535182331, + "grad_norm": 2.732808828353882, + "learning_rate": 2.6417967263037685e-06, + "loss": 1.5115, + "step": 1547 + }, + { + "epoch": 1.5901386748844377, + "grad_norm": 2.5835330486297607, + "learning_rate": 2.639893414541302e-06, + "loss": 1.533, + "step": 1548 + }, + { + "epoch": 1.5911658962506419, + "grad_norm": 2.6625828742980957, + "learning_rate": 2.6379901027788352e-06, + "loss": 1.555, + "step": 1549 + }, + { + "epoch": 1.5921931176168465, + "grad_norm": 2.5941221714019775, + "learning_rate": 2.6360867910163686e-06, + "loss": 1.4881, + "step": 1550 + }, + { + "epoch": 1.5921931176168465, + "eval_loss": 1.9022443294525146, + "eval_runtime": 46.3086, + "eval_samples_per_second": 21.594, + "eval_steps_per_second": 0.907, + "step": 1550 + }, + { + "epoch": 1.5932203389830508, + "grad_norm": 2.6347010135650635, + "learning_rate": 2.6341834792539024e-06, + "loss": 1.5118, + "step": 1551 + }, + { + "epoch": 1.5942475603492552, + "grad_norm": 2.6772608757019043, + "learning_rate": 2.6322801674914357e-06, + "loss": 1.5107, + "step": 1552 + }, + { + "epoch": 1.5952747817154598, + "grad_norm": 2.819559097290039, + "learning_rate": 2.6303768557289687e-06, + "loss": 1.5008, + "step": 1553 + }, + { + "epoch": 1.5963020030816641, + "grad_norm": 2.5580296516418457, + "learning_rate": 2.628473543966502e-06, + "loss": 1.4561, + "step": 1554 + }, + { + "epoch": 1.5973292244478685, + "grad_norm": 2.6931111812591553, + "learning_rate": 2.6265702322040354e-06, + "loss": 1.5222, + "step": 1555 + }, + { + "epoch": 1.5983564458140729, + "grad_norm": 2.6396803855895996, + "learning_rate": 2.6246669204415688e-06, + "loss": 1.4791, + "step": 1556 + }, + { + "epoch": 1.5993836671802772, + "grad_norm": 2.6619536876678467, + "learning_rate": 2.622763608679102e-06, + "loss": 1.471, + "step": 1557 + }, + { + "epoch": 1.6004108885464818, + "grad_norm": 2.578331232070923, + "learning_rate": 2.620860296916635e-06, + "loss": 1.4787, + "step": 1558 + }, + { + "epoch": 1.6014381099126862, + "grad_norm": 2.764533042907715, + "learning_rate": 2.6189569851541684e-06, + "loss": 1.4305, + "step": 1559 + }, + { + "epoch": 1.6024653312788906, + "grad_norm": 2.673753499984741, + "learning_rate": 2.617053673391702e-06, + "loss": 1.536, + "step": 1560 + }, + { + "epoch": 1.6024653312788906, + "eval_loss": 1.8979606628417969, + "eval_runtime": 46.5101, + "eval_samples_per_second": 21.501, + "eval_steps_per_second": 0.903, + "step": 1560 + }, + { + "epoch": 1.6034925526450952, + "grad_norm": 2.7593281269073486, + "learning_rate": 2.615150361629235e-06, + "loss": 1.5721, + "step": 1561 + }, + { + "epoch": 1.6045197740112993, + "grad_norm": 2.72963285446167, + "learning_rate": 2.6132470498667685e-06, + "loss": 1.5198, + "step": 1562 + }, + { + "epoch": 1.605546995377504, + "grad_norm": 2.6438205242156982, + "learning_rate": 2.6113437381043015e-06, + "loss": 1.4922, + "step": 1563 + }, + { + "epoch": 1.6065742167437083, + "grad_norm": 2.6523141860961914, + "learning_rate": 2.609440426341835e-06, + "loss": 1.4775, + "step": 1564 + }, + { + "epoch": 1.6076014381099126, + "grad_norm": 2.7506513595581055, + "learning_rate": 2.607537114579368e-06, + "loss": 1.524, + "step": 1565 + }, + { + "epoch": 1.6086286594761172, + "grad_norm": 2.589853286743164, + "learning_rate": 2.6056338028169015e-06, + "loss": 1.4946, + "step": 1566 + }, + { + "epoch": 1.6096558808423214, + "grad_norm": 2.6591460704803467, + "learning_rate": 2.603730491054435e-06, + "loss": 1.4686, + "step": 1567 + }, + { + "epoch": 1.610683102208526, + "grad_norm": 2.8163933753967285, + "learning_rate": 2.601827179291968e-06, + "loss": 1.5196, + "step": 1568 + }, + { + "epoch": 1.6117103235747303, + "grad_norm": 2.6938319206237793, + "learning_rate": 2.599923867529501e-06, + "loss": 1.5082, + "step": 1569 + }, + { + "epoch": 1.6127375449409347, + "grad_norm": 2.7489402294158936, + "learning_rate": 2.5980205557670346e-06, + "loss": 1.4978, + "step": 1570 + }, + { + "epoch": 1.6127375449409347, + "eval_loss": 1.897243618965149, + "eval_runtime": 45.6931, + "eval_samples_per_second": 21.885, + "eval_steps_per_second": 0.919, + "step": 1570 + }, + { + "epoch": 1.6137647663071393, + "grad_norm": 2.763204336166382, + "learning_rate": 2.5961172440045684e-06, + "loss": 1.5382, + "step": 1571 + }, + { + "epoch": 1.6147919876733436, + "grad_norm": 2.726813316345215, + "learning_rate": 2.5942139322421017e-06, + "loss": 1.5377, + "step": 1572 + }, + { + "epoch": 1.615819209039548, + "grad_norm": 2.6197266578674316, + "learning_rate": 2.592310620479635e-06, + "loss": 1.4887, + "step": 1573 + }, + { + "epoch": 1.6168464304057526, + "grad_norm": 2.739987373352051, + "learning_rate": 2.5904073087171684e-06, + "loss": 1.4546, + "step": 1574 + }, + { + "epoch": 1.6178736517719567, + "grad_norm": 2.6571943759918213, + "learning_rate": 2.5885039969547014e-06, + "loss": 1.5129, + "step": 1575 + }, + { + "epoch": 1.6189008731381613, + "grad_norm": 2.7591447830200195, + "learning_rate": 2.5866006851922347e-06, + "loss": 1.4627, + "step": 1576 + }, + { + "epoch": 1.6199280945043657, + "grad_norm": 2.6228175163269043, + "learning_rate": 2.584697373429768e-06, + "loss": 1.4712, + "step": 1577 + }, + { + "epoch": 1.62095531587057, + "grad_norm": 2.7786920070648193, + "learning_rate": 2.5827940616673015e-06, + "loss": 1.4995, + "step": 1578 + }, + { + "epoch": 1.6219825372367747, + "grad_norm": 2.782848596572876, + "learning_rate": 2.580890749904835e-06, + "loss": 1.5387, + "step": 1579 + }, + { + "epoch": 1.6230097586029788, + "grad_norm": 2.7152798175811768, + "learning_rate": 2.5789874381423678e-06, + "loss": 1.4813, + "step": 1580 + }, + { + "epoch": 1.6230097586029788, + "eval_loss": 1.8939763307571411, + "eval_runtime": 46.2339, + "eval_samples_per_second": 21.629, + "eval_steps_per_second": 0.908, + "step": 1580 + }, + { + "epoch": 1.6240369799691834, + "grad_norm": 2.6328132152557373, + "learning_rate": 2.577084126379901e-06, + "loss": 1.4801, + "step": 1581 + }, + { + "epoch": 1.6250642013353878, + "grad_norm": 2.6837892532348633, + "learning_rate": 2.5751808146174345e-06, + "loss": 1.453, + "step": 1582 + }, + { + "epoch": 1.6260914227015921, + "grad_norm": 2.6900155544281006, + "learning_rate": 2.573277502854968e-06, + "loss": 1.5044, + "step": 1583 + }, + { + "epoch": 1.6271186440677967, + "grad_norm": 2.6830708980560303, + "learning_rate": 2.5713741910925012e-06, + "loss": 1.4662, + "step": 1584 + }, + { + "epoch": 1.6281458654340009, + "grad_norm": 2.6451282501220703, + "learning_rate": 2.569470879330034e-06, + "loss": 1.4749, + "step": 1585 + }, + { + "epoch": 1.6291730868002054, + "grad_norm": 2.8980915546417236, + "learning_rate": 2.5675675675675675e-06, + "loss": 1.5729, + "step": 1586 + }, + { + "epoch": 1.6302003081664098, + "grad_norm": 2.715513229370117, + "learning_rate": 2.565664255805101e-06, + "loss": 1.4848, + "step": 1587 + }, + { + "epoch": 1.6312275295326142, + "grad_norm": 2.8286707401275635, + "learning_rate": 2.5637609440426343e-06, + "loss": 1.5164, + "step": 1588 + }, + { + "epoch": 1.6322547508988188, + "grad_norm": 2.7614853382110596, + "learning_rate": 2.5618576322801676e-06, + "loss": 1.4623, + "step": 1589 + }, + { + "epoch": 1.6332819722650231, + "grad_norm": 2.826911687850952, + "learning_rate": 2.5599543205177006e-06, + "loss": 1.5593, + "step": 1590 + }, + { + "epoch": 1.6332819722650231, + "eval_loss": 1.8977031707763672, + "eval_runtime": 46.7228, + "eval_samples_per_second": 21.403, + "eval_steps_per_second": 0.899, + "step": 1590 + }, + { + "epoch": 1.6343091936312275, + "grad_norm": 2.707435131072998, + "learning_rate": 2.558051008755234e-06, + "loss": 1.5186, + "step": 1591 + }, + { + "epoch": 1.635336414997432, + "grad_norm": 2.701657295227051, + "learning_rate": 2.5561476969927677e-06, + "loss": 1.5643, + "step": 1592 + }, + { + "epoch": 1.6363636363636362, + "grad_norm": 2.7254910469055176, + "learning_rate": 2.554244385230301e-06, + "loss": 1.4701, + "step": 1593 + }, + { + "epoch": 1.6373908577298408, + "grad_norm": 2.6841659545898438, + "learning_rate": 2.5523410734678344e-06, + "loss": 1.4949, + "step": 1594 + }, + { + "epoch": 1.6384180790960452, + "grad_norm": 2.746089220046997, + "learning_rate": 2.550437761705368e-06, + "loss": 1.435, + "step": 1595 + }, + { + "epoch": 1.6394453004622496, + "grad_norm": 2.6010215282440186, + "learning_rate": 2.548534449942901e-06, + "loss": 1.4594, + "step": 1596 + }, + { + "epoch": 1.6404725218284542, + "grad_norm": 2.9612057209014893, + "learning_rate": 2.546631138180434e-06, + "loss": 1.5084, + "step": 1597 + }, + { + "epoch": 1.6414997431946583, + "grad_norm": 3.105278253555298, + "learning_rate": 2.5447278264179675e-06, + "loss": 1.5041, + "step": 1598 + }, + { + "epoch": 1.642526964560863, + "grad_norm": 2.714082956314087, + "learning_rate": 2.542824514655501e-06, + "loss": 1.462, + "step": 1599 + }, + { + "epoch": 1.6435541859270673, + "grad_norm": 2.8359603881835938, + "learning_rate": 2.540921202893034e-06, + "loss": 1.5545, + "step": 1600 + }, + { + "epoch": 1.6435541859270673, + "eval_loss": 1.9013447761535645, + "eval_runtime": 47.8392, + "eval_samples_per_second": 20.903, + "eval_steps_per_second": 0.878, + "step": 1600 + }, + { + "epoch": 1.6445814072932716, + "grad_norm": 2.8133420944213867, + "learning_rate": 2.5390178911305675e-06, + "loss": 1.5362, + "step": 1601 + }, + { + "epoch": 1.6456086286594762, + "grad_norm": 2.7947328090667725, + "learning_rate": 2.5371145793681005e-06, + "loss": 1.5067, + "step": 1602 + }, + { + "epoch": 1.6466358500256806, + "grad_norm": 2.705209732055664, + "learning_rate": 2.535211267605634e-06, + "loss": 1.4921, + "step": 1603 + }, + { + "epoch": 1.647663071391885, + "grad_norm": 2.7078511714935303, + "learning_rate": 2.533307955843167e-06, + "loss": 1.4398, + "step": 1604 + }, + { + "epoch": 1.6486902927580893, + "grad_norm": 2.749514579772949, + "learning_rate": 2.5314046440807006e-06, + "loss": 1.4931, + "step": 1605 + }, + { + "epoch": 1.6497175141242937, + "grad_norm": 2.703160285949707, + "learning_rate": 2.529501332318234e-06, + "loss": 1.4693, + "step": 1606 + }, + { + "epoch": 1.6507447354904983, + "grad_norm": 2.8278794288635254, + "learning_rate": 2.5275980205557673e-06, + "loss": 1.5353, + "step": 1607 + }, + { + "epoch": 1.6517719568567026, + "grad_norm": 2.927949905395508, + "learning_rate": 2.5256947087933002e-06, + "loss": 1.5665, + "step": 1608 + }, + { + "epoch": 1.652799178222907, + "grad_norm": 2.714650869369507, + "learning_rate": 2.5237913970308336e-06, + "loss": 1.5503, + "step": 1609 + }, + { + "epoch": 1.6538263995891116, + "grad_norm": 2.7319588661193848, + "learning_rate": 2.521888085268367e-06, + "loss": 1.5018, + "step": 1610 + }, + { + "epoch": 1.6538263995891116, + "eval_loss": 1.8987922668457031, + "eval_runtime": 43.4436, + "eval_samples_per_second": 23.018, + "eval_steps_per_second": 0.967, + "step": 1610 + }, + { + "epoch": 1.6548536209553157, + "grad_norm": 2.746079921722412, + "learning_rate": 2.5199847735059003e-06, + "loss": 1.5447, + "step": 1611 + }, + { + "epoch": 1.6558808423215203, + "grad_norm": 2.7670583724975586, + "learning_rate": 2.518081461743434e-06, + "loss": 1.4723, + "step": 1612 + }, + { + "epoch": 1.6569080636877247, + "grad_norm": 2.680903196334839, + "learning_rate": 2.5161781499809675e-06, + "loss": 1.5116, + "step": 1613 + }, + { + "epoch": 1.657935285053929, + "grad_norm": 2.5936760902404785, + "learning_rate": 2.5142748382185004e-06, + "loss": 1.4607, + "step": 1614 + }, + { + "epoch": 1.6589625064201337, + "grad_norm": 2.742476224899292, + "learning_rate": 2.5123715264560338e-06, + "loss": 1.4579, + "step": 1615 + }, + { + "epoch": 1.6599897277863378, + "grad_norm": 2.9192259311676025, + "learning_rate": 2.510468214693567e-06, + "loss": 1.4815, + "step": 1616 + }, + { + "epoch": 1.6610169491525424, + "grad_norm": 2.7400426864624023, + "learning_rate": 2.5085649029311005e-06, + "loss": 1.4268, + "step": 1617 + }, + { + "epoch": 1.6620441705187468, + "grad_norm": 2.6352646350860596, + "learning_rate": 2.506661591168634e-06, + "loss": 1.3921, + "step": 1618 + }, + { + "epoch": 1.6630713918849511, + "grad_norm": 2.6632823944091797, + "learning_rate": 2.504758279406167e-06, + "loss": 1.4754, + "step": 1619 + }, + { + "epoch": 1.6640986132511557, + "grad_norm": 2.8280115127563477, + "learning_rate": 2.5028549676437e-06, + "loss": 1.5208, + "step": 1620 + }, + { + "epoch": 1.6640986132511557, + "eval_loss": 1.8962472677230835, + "eval_runtime": 45.5434, + "eval_samples_per_second": 21.957, + "eval_steps_per_second": 0.922, + "step": 1620 + }, + { + "epoch": 1.66512583461736, + "grad_norm": 2.84045672416687, + "learning_rate": 2.5009516558812335e-06, + "loss": 1.4669, + "step": 1621 + }, + { + "epoch": 1.6661530559835644, + "grad_norm": 2.564673900604248, + "learning_rate": 2.499048344118767e-06, + "loss": 1.5092, + "step": 1622 + }, + { + "epoch": 1.667180277349769, + "grad_norm": 2.622130870819092, + "learning_rate": 2.4971450323563002e-06, + "loss": 1.5606, + "step": 1623 + }, + { + "epoch": 1.6682074987159732, + "grad_norm": 2.5972182750701904, + "learning_rate": 2.4952417205938336e-06, + "loss": 1.4751, + "step": 1624 + }, + { + "epoch": 1.6692347200821778, + "grad_norm": 2.6662817001342773, + "learning_rate": 2.4933384088313666e-06, + "loss": 1.4789, + "step": 1625 + }, + { + "epoch": 1.6702619414483821, + "grad_norm": 2.6823441982269287, + "learning_rate": 2.4914350970689e-06, + "loss": 1.462, + "step": 1626 + }, + { + "epoch": 1.6712891628145865, + "grad_norm": 2.7047946453094482, + "learning_rate": 2.4895317853064337e-06, + "loss": 1.4809, + "step": 1627 + }, + { + "epoch": 1.672316384180791, + "grad_norm": 2.7356603145599365, + "learning_rate": 2.4876284735439666e-06, + "loss": 1.5316, + "step": 1628 + }, + { + "epoch": 1.6733436055469952, + "grad_norm": 2.8151392936706543, + "learning_rate": 2.4857251617815e-06, + "loss": 1.4969, + "step": 1629 + }, + { + "epoch": 1.6743708269131998, + "grad_norm": 2.8269588947296143, + "learning_rate": 2.4838218500190334e-06, + "loss": 1.5041, + "step": 1630 + }, + { + "epoch": 1.6743708269131998, + "eval_loss": 1.9033366441726685, + "eval_runtime": 47.514, + "eval_samples_per_second": 21.046, + "eval_steps_per_second": 0.884, + "step": 1630 + }, + { + "epoch": 1.6753980482794042, + "grad_norm": 2.7226245403289795, + "learning_rate": 2.4819185382565667e-06, + "loss": 1.4924, + "step": 1631 + }, + { + "epoch": 1.6764252696456086, + "grad_norm": 2.6129348278045654, + "learning_rate": 2.4800152264941e-06, + "loss": 1.4829, + "step": 1632 + }, + { + "epoch": 1.6774524910118132, + "grad_norm": 2.618772029876709, + "learning_rate": 2.478111914731633e-06, + "loss": 1.452, + "step": 1633 + }, + { + "epoch": 1.6784797123780173, + "grad_norm": 2.7626497745513916, + "learning_rate": 2.4762086029691664e-06, + "loss": 1.4875, + "step": 1634 + }, + { + "epoch": 1.6795069337442219, + "grad_norm": 3.0622305870056152, + "learning_rate": 2.4743052912066998e-06, + "loss": 1.6208, + "step": 1635 + }, + { + "epoch": 1.6805341551104263, + "grad_norm": 2.6085500717163086, + "learning_rate": 2.472401979444233e-06, + "loss": 1.457, + "step": 1636 + }, + { + "epoch": 1.6815613764766306, + "grad_norm": 2.7662956714630127, + "learning_rate": 2.4704986676817665e-06, + "loss": 1.5572, + "step": 1637 + }, + { + "epoch": 1.6825885978428352, + "grad_norm": 2.8282175064086914, + "learning_rate": 2.4685953559193e-06, + "loss": 1.4922, + "step": 1638 + }, + { + "epoch": 1.6836158192090396, + "grad_norm": 2.930837631225586, + "learning_rate": 2.466692044156833e-06, + "loss": 1.5296, + "step": 1639 + }, + { + "epoch": 1.684643040575244, + "grad_norm": 2.7347614765167236, + "learning_rate": 2.4647887323943666e-06, + "loss": 1.4739, + "step": 1640 + }, + { + "epoch": 1.684643040575244, + "eval_loss": 1.8892933130264282, + "eval_runtime": 48.2635, + "eval_samples_per_second": 20.72, + "eval_steps_per_second": 0.87, + "step": 1640 + }, + { + "epoch": 1.6856702619414485, + "grad_norm": 2.6786420345306396, + "learning_rate": 2.4628854206319e-06, + "loss": 1.4429, + "step": 1641 + }, + { + "epoch": 1.6866974833076527, + "grad_norm": 2.754531145095825, + "learning_rate": 2.460982108869433e-06, + "loss": 1.4375, + "step": 1642 + }, + { + "epoch": 1.6877247046738573, + "grad_norm": 2.8799307346343994, + "learning_rate": 2.4590787971069662e-06, + "loss": 1.5438, + "step": 1643 + }, + { + "epoch": 1.6887519260400616, + "grad_norm": 2.6701910495758057, + "learning_rate": 2.4571754853444996e-06, + "loss": 1.4691, + "step": 1644 + }, + { + "epoch": 1.689779147406266, + "grad_norm": 2.760960340499878, + "learning_rate": 2.455272173582033e-06, + "loss": 1.4607, + "step": 1645 + }, + { + "epoch": 1.6908063687724706, + "grad_norm": 2.7178521156311035, + "learning_rate": 2.4533688618195663e-06, + "loss": 1.466, + "step": 1646 + }, + { + "epoch": 1.6918335901386747, + "grad_norm": 2.7641139030456543, + "learning_rate": 2.4514655500570997e-06, + "loss": 1.4065, + "step": 1647 + }, + { + "epoch": 1.6928608115048793, + "grad_norm": 2.7858457565307617, + "learning_rate": 2.449562238294633e-06, + "loss": 1.4587, + "step": 1648 + }, + { + "epoch": 1.6938880328710837, + "grad_norm": 2.6738100051879883, + "learning_rate": 2.4476589265321664e-06, + "loss": 1.4205, + "step": 1649 + }, + { + "epoch": 1.694915254237288, + "grad_norm": 2.748340129852295, + "learning_rate": 2.4457556147696993e-06, + "loss": 1.5212, + "step": 1650 + }, + { + "epoch": 1.694915254237288, + "eval_loss": 1.9066362380981445, + "eval_runtime": 52.2773, + "eval_samples_per_second": 19.129, + "eval_steps_per_second": 0.803, + "step": 1650 + }, + { + "epoch": 1.6959424756034926, + "grad_norm": 2.7664692401885986, + "learning_rate": 2.4438523030072327e-06, + "loss": 1.5045, + "step": 1651 + }, + { + "epoch": 1.696969696969697, + "grad_norm": 2.7928621768951416, + "learning_rate": 2.441948991244766e-06, + "loss": 1.5134, + "step": 1652 + }, + { + "epoch": 1.6979969183359014, + "grad_norm": 2.7065975666046143, + "learning_rate": 2.4400456794822994e-06, + "loss": 1.5072, + "step": 1653 + }, + { + "epoch": 1.6990241397021058, + "grad_norm": 2.7586593627929688, + "learning_rate": 2.438142367719833e-06, + "loss": 1.5276, + "step": 1654 + }, + { + "epoch": 1.7000513610683101, + "grad_norm": 2.689526081085205, + "learning_rate": 2.4362390559573657e-06, + "loss": 1.4902, + "step": 1655 + }, + { + "epoch": 1.7010785824345147, + "grad_norm": 2.716808795928955, + "learning_rate": 2.434335744194899e-06, + "loss": 1.4772, + "step": 1656 + }, + { + "epoch": 1.702105803800719, + "grad_norm": 2.825169086456299, + "learning_rate": 2.432432432432433e-06, + "loss": 1.5185, + "step": 1657 + }, + { + "epoch": 1.7031330251669234, + "grad_norm": 2.727210283279419, + "learning_rate": 2.4305291206699662e-06, + "loss": 1.4353, + "step": 1658 + }, + { + "epoch": 1.704160246533128, + "grad_norm": 2.753382921218872, + "learning_rate": 2.428625808907499e-06, + "loss": 1.5161, + "step": 1659 + }, + { + "epoch": 1.7051874678993322, + "grad_norm": 2.7188405990600586, + "learning_rate": 2.4267224971450325e-06, + "loss": 1.5324, + "step": 1660 + }, + { + "epoch": 1.7051874678993322, + "eval_loss": 1.8945986032485962, + "eval_runtime": 49.8237, + "eval_samples_per_second": 20.071, + "eval_steps_per_second": 0.843, + "step": 1660 + }, + { + "epoch": 1.7062146892655368, + "grad_norm": 2.702894687652588, + "learning_rate": 2.424819185382566e-06, + "loss": 1.4915, + "step": 1661 + }, + { + "epoch": 1.7072419106317411, + "grad_norm": 2.6755716800689697, + "learning_rate": 2.4229158736200993e-06, + "loss": 1.4471, + "step": 1662 + }, + { + "epoch": 1.7082691319979455, + "grad_norm": 2.637878656387329, + "learning_rate": 2.4210125618576326e-06, + "loss": 1.4488, + "step": 1663 + }, + { + "epoch": 1.70929635336415, + "grad_norm": 2.6714439392089844, + "learning_rate": 2.4191092500951656e-06, + "loss": 1.4708, + "step": 1664 + }, + { + "epoch": 1.7103235747303542, + "grad_norm": 2.703479528427124, + "learning_rate": 2.417205938332699e-06, + "loss": 1.5039, + "step": 1665 + }, + { + "epoch": 1.7113507960965588, + "grad_norm": 2.7005186080932617, + "learning_rate": 2.4153026265702323e-06, + "loss": 1.4242, + "step": 1666 + }, + { + "epoch": 1.7123780174627632, + "grad_norm": 2.7262654304504395, + "learning_rate": 2.4133993148077657e-06, + "loss": 1.4649, + "step": 1667 + }, + { + "epoch": 1.7134052388289676, + "grad_norm": 2.912041425704956, + "learning_rate": 2.411496003045299e-06, + "loss": 1.5247, + "step": 1668 + }, + { + "epoch": 1.7144324601951721, + "grad_norm": 2.713007926940918, + "learning_rate": 2.4095926912828324e-06, + "loss": 1.4585, + "step": 1669 + }, + { + "epoch": 1.7154596815613765, + "grad_norm": 2.6106162071228027, + "learning_rate": 2.4076893795203658e-06, + "loss": 1.4932, + "step": 1670 + }, + { + "epoch": 1.7154596815613765, + "eval_loss": 1.8863319158554077, + "eval_runtime": 45.1847, + "eval_samples_per_second": 22.131, + "eval_steps_per_second": 0.93, + "step": 1670 + }, + { + "epoch": 1.7164869029275809, + "grad_norm": 2.9076762199401855, + "learning_rate": 2.405786067757899e-06, + "loss": 1.4725, + "step": 1671 + }, + { + "epoch": 1.7175141242937855, + "grad_norm": 2.9677484035491943, + "learning_rate": 2.403882755995432e-06, + "loss": 1.551, + "step": 1672 + }, + { + "epoch": 1.7185413456599896, + "grad_norm": 2.6974384784698486, + "learning_rate": 2.4019794442329654e-06, + "loss": 1.4897, + "step": 1673 + }, + { + "epoch": 1.7195685670261942, + "grad_norm": 2.796464443206787, + "learning_rate": 2.4000761324704988e-06, + "loss": 1.4573, + "step": 1674 + }, + { + "epoch": 1.7205957883923986, + "grad_norm": 2.653627872467041, + "learning_rate": 2.398172820708032e-06, + "loss": 1.5025, + "step": 1675 + }, + { + "epoch": 1.721623009758603, + "grad_norm": 2.716545343399048, + "learning_rate": 2.3962695089455655e-06, + "loss": 1.4687, + "step": 1676 + }, + { + "epoch": 1.7226502311248075, + "grad_norm": 2.737752676010132, + "learning_rate": 2.3943661971830984e-06, + "loss": 1.4515, + "step": 1677 + }, + { + "epoch": 1.7236774524910117, + "grad_norm": 2.7596747875213623, + "learning_rate": 2.3924628854206322e-06, + "loss": 1.4712, + "step": 1678 + }, + { + "epoch": 1.7247046738572163, + "grad_norm": 2.9202370643615723, + "learning_rate": 2.3905595736581656e-06, + "loss": 1.4818, + "step": 1679 + }, + { + "epoch": 1.7257318952234206, + "grad_norm": 2.7129971981048584, + "learning_rate": 2.388656261895699e-06, + "loss": 1.5269, + "step": 1680 + }, + { + "epoch": 1.7257318952234206, + "eval_loss": 1.8941866159439087, + "eval_runtime": 45.4463, + "eval_samples_per_second": 22.004, + "eval_steps_per_second": 0.924, + "step": 1680 + }, + { + "epoch": 1.726759116589625, + "grad_norm": 2.956148862838745, + "learning_rate": 2.386752950133232e-06, + "loss": 1.4977, + "step": 1681 + }, + { + "epoch": 1.7277863379558296, + "grad_norm": 2.8728253841400146, + "learning_rate": 2.3848496383707653e-06, + "loss": 1.4961, + "step": 1682 + }, + { + "epoch": 1.7288135593220337, + "grad_norm": 2.7839162349700928, + "learning_rate": 2.3829463266082986e-06, + "loss": 1.4757, + "step": 1683 + }, + { + "epoch": 1.7298407806882383, + "grad_norm": 2.7487802505493164, + "learning_rate": 2.381043014845832e-06, + "loss": 1.5048, + "step": 1684 + }, + { + "epoch": 1.7308680020544427, + "grad_norm": 2.80683970451355, + "learning_rate": 2.3791397030833653e-06, + "loss": 1.4916, + "step": 1685 + }, + { + "epoch": 1.731895223420647, + "grad_norm": 2.6910040378570557, + "learning_rate": 2.3772363913208983e-06, + "loss": 1.4711, + "step": 1686 + }, + { + "epoch": 1.7329224447868516, + "grad_norm": 2.594284772872925, + "learning_rate": 2.3753330795584316e-06, + "loss": 1.4558, + "step": 1687 + }, + { + "epoch": 1.733949666153056, + "grad_norm": 2.7420263290405273, + "learning_rate": 2.3734297677959654e-06, + "loss": 1.4769, + "step": 1688 + }, + { + "epoch": 1.7349768875192604, + "grad_norm": 2.80002760887146, + "learning_rate": 2.3715264560334984e-06, + "loss": 1.5289, + "step": 1689 + }, + { + "epoch": 1.736004108885465, + "grad_norm": 2.799618721008301, + "learning_rate": 2.3696231442710317e-06, + "loss": 1.5131, + "step": 1690 + }, + { + "epoch": 1.736004108885465, + "eval_loss": 1.8925220966339111, + "eval_runtime": 45.89, + "eval_samples_per_second": 21.791, + "eval_steps_per_second": 0.915, + "step": 1690 + }, + { + "epoch": 1.7370313302516691, + "grad_norm": 2.701648712158203, + "learning_rate": 2.367719832508565e-06, + "loss": 1.5076, + "step": 1691 + }, + { + "epoch": 1.7380585516178737, + "grad_norm": 2.6925787925720215, + "learning_rate": 2.3658165207460985e-06, + "loss": 1.5331, + "step": 1692 + }, + { + "epoch": 1.739085772984078, + "grad_norm": 2.8448283672332764, + "learning_rate": 2.363913208983632e-06, + "loss": 1.4933, + "step": 1693 + }, + { + "epoch": 1.7401129943502824, + "grad_norm": 2.8252811431884766, + "learning_rate": 2.362009897221165e-06, + "loss": 1.5205, + "step": 1694 + }, + { + "epoch": 1.741140215716487, + "grad_norm": 2.675318956375122, + "learning_rate": 2.360106585458698e-06, + "loss": 1.4958, + "step": 1695 + }, + { + "epoch": 1.7421674370826912, + "grad_norm": 2.8392488956451416, + "learning_rate": 2.3582032736962315e-06, + "loss": 1.4522, + "step": 1696 + }, + { + "epoch": 1.7431946584488958, + "grad_norm": 2.8221116065979004, + "learning_rate": 2.356299961933765e-06, + "loss": 1.4924, + "step": 1697 + }, + { + "epoch": 1.7442218798151001, + "grad_norm": 2.685828447341919, + "learning_rate": 2.354396650171298e-06, + "loss": 1.4668, + "step": 1698 + }, + { + "epoch": 1.7452491011813045, + "grad_norm": 2.622481107711792, + "learning_rate": 2.3524933384088316e-06, + "loss": 1.451, + "step": 1699 + }, + { + "epoch": 1.746276322547509, + "grad_norm": 2.79746413230896, + "learning_rate": 2.350590026646365e-06, + "loss": 1.5473, + "step": 1700 + }, + { + "epoch": 1.746276322547509, + "eval_loss": 1.898282766342163, + "eval_runtime": 45.0673, + "eval_samples_per_second": 22.189, + "eval_steps_per_second": 0.932, + "step": 1700 + }, + { + "epoch": 1.7473035439137135, + "grad_norm": 2.8120598793029785, + "learning_rate": 2.3486867148838983e-06, + "loss": 1.4692, + "step": 1701 + }, + { + "epoch": 1.7483307652799178, + "grad_norm": 2.7376883029937744, + "learning_rate": 2.3467834031214317e-06, + "loss": 1.4749, + "step": 1702 + }, + { + "epoch": 1.7493579866461222, + "grad_norm": 2.7579424381256104, + "learning_rate": 2.3448800913589646e-06, + "loss": 1.4726, + "step": 1703 + }, + { + "epoch": 1.7503852080123266, + "grad_norm": 2.648026466369629, + "learning_rate": 2.342976779596498e-06, + "loss": 1.4938, + "step": 1704 + }, + { + "epoch": 1.7514124293785311, + "grad_norm": 2.6477997303009033, + "learning_rate": 2.3410734678340313e-06, + "loss": 1.4813, + "step": 1705 + }, + { + "epoch": 1.7524396507447355, + "grad_norm": 2.742249011993408, + "learning_rate": 2.3391701560715647e-06, + "loss": 1.5257, + "step": 1706 + }, + { + "epoch": 1.7534668721109399, + "grad_norm": 2.6430649757385254, + "learning_rate": 2.337266844309098e-06, + "loss": 1.4781, + "step": 1707 + }, + { + "epoch": 1.7544940934771445, + "grad_norm": 2.6746671199798584, + "learning_rate": 2.3353635325466314e-06, + "loss": 1.4807, + "step": 1708 + }, + { + "epoch": 1.7555213148433486, + "grad_norm": 2.8837218284606934, + "learning_rate": 2.3334602207841648e-06, + "loss": 1.4932, + "step": 1709 + }, + { + "epoch": 1.7565485362095532, + "grad_norm": 2.703761577606201, + "learning_rate": 2.331556909021698e-06, + "loss": 1.4969, + "step": 1710 + }, + { + "epoch": 1.7565485362095532, + "eval_loss": 1.8966599702835083, + "eval_runtime": 42.6325, + "eval_samples_per_second": 23.456, + "eval_steps_per_second": 0.985, + "step": 1710 + }, + { + "epoch": 1.7575757575757576, + "grad_norm": 2.760502338409424, + "learning_rate": 2.3296535972592315e-06, + "loss": 1.5148, + "step": 1711 + }, + { + "epoch": 1.758602978941962, + "grad_norm": 2.696356773376465, + "learning_rate": 2.3277502854967644e-06, + "loss": 1.4604, + "step": 1712 + }, + { + "epoch": 1.7596302003081665, + "grad_norm": 2.7124855518341064, + "learning_rate": 2.325846973734298e-06, + "loss": 1.5178, + "step": 1713 + }, + { + "epoch": 1.7606574216743707, + "grad_norm": 2.624866008758545, + "learning_rate": 2.323943661971831e-06, + "loss": 1.4786, + "step": 1714 + }, + { + "epoch": 1.7616846430405753, + "grad_norm": 2.785651922225952, + "learning_rate": 2.3220403502093645e-06, + "loss": 1.5077, + "step": 1715 + }, + { + "epoch": 1.7627118644067796, + "grad_norm": 2.7688722610473633, + "learning_rate": 2.320137038446898e-06, + "loss": 1.5299, + "step": 1716 + }, + { + "epoch": 1.763739085772984, + "grad_norm": 2.72312593460083, + "learning_rate": 2.318233726684431e-06, + "loss": 1.5352, + "step": 1717 + }, + { + "epoch": 1.7647663071391886, + "grad_norm": 2.7877933979034424, + "learning_rate": 2.3163304149219646e-06, + "loss": 1.5148, + "step": 1718 + }, + { + "epoch": 1.765793528505393, + "grad_norm": 2.6651675701141357, + "learning_rate": 2.314427103159498e-06, + "loss": 1.4979, + "step": 1719 + }, + { + "epoch": 1.7668207498715973, + "grad_norm": 2.747264862060547, + "learning_rate": 2.312523791397031e-06, + "loss": 1.4785, + "step": 1720 + }, + { + "epoch": 1.7668207498715973, + "eval_loss": 1.9028254747390747, + "eval_runtime": 47.0728, + "eval_samples_per_second": 21.244, + "eval_steps_per_second": 0.892, + "step": 1720 + }, + { + "epoch": 1.767847971237802, + "grad_norm": 2.6331684589385986, + "learning_rate": 2.3106204796345643e-06, + "loss": 1.4479, + "step": 1721 + }, + { + "epoch": 1.768875192604006, + "grad_norm": 2.662079095840454, + "learning_rate": 2.3087171678720976e-06, + "loss": 1.4827, + "step": 1722 + }, + { + "epoch": 1.7699024139702106, + "grad_norm": 2.735499382019043, + "learning_rate": 2.306813856109631e-06, + "loss": 1.5324, + "step": 1723 + }, + { + "epoch": 1.770929635336415, + "grad_norm": 2.66086745262146, + "learning_rate": 2.3049105443471644e-06, + "loss": 1.5002, + "step": 1724 + }, + { + "epoch": 1.7719568567026194, + "grad_norm": 2.7453548908233643, + "learning_rate": 2.3030072325846973e-06, + "loss": 1.48, + "step": 1725 + }, + { + "epoch": 1.772984078068824, + "grad_norm": 2.701510429382324, + "learning_rate": 2.3011039208222307e-06, + "loss": 1.4642, + "step": 1726 + }, + { + "epoch": 1.774011299435028, + "grad_norm": 2.6908695697784424, + "learning_rate": 2.299200609059764e-06, + "loss": 1.4607, + "step": 1727 + }, + { + "epoch": 1.7750385208012327, + "grad_norm": 2.7793400287628174, + "learning_rate": 2.297297297297298e-06, + "loss": 1.4756, + "step": 1728 + }, + { + "epoch": 1.776065742167437, + "grad_norm": 2.6336896419525146, + "learning_rate": 2.2953939855348308e-06, + "loss": 1.4336, + "step": 1729 + }, + { + "epoch": 1.7770929635336414, + "grad_norm": 2.82281231880188, + "learning_rate": 2.293490673772364e-06, + "loss": 1.5273, + "step": 1730 + }, + { + "epoch": 1.7770929635336414, + "eval_loss": 1.9012898206710815, + "eval_runtime": 46.5443, + "eval_samples_per_second": 21.485, + "eval_steps_per_second": 0.902, + "step": 1730 + }, + { + "epoch": 1.778120184899846, + "grad_norm": 2.7541351318359375, + "learning_rate": 2.2915873620098975e-06, + "loss": 1.5071, + "step": 1731 + }, + { + "epoch": 1.7791474062660502, + "grad_norm": 2.771777629852295, + "learning_rate": 2.289684050247431e-06, + "loss": 1.503, + "step": 1732 + }, + { + "epoch": 1.7801746276322548, + "grad_norm": 2.5365493297576904, + "learning_rate": 2.287780738484964e-06, + "loss": 1.4635, + "step": 1733 + }, + { + "epoch": 1.7812018489984591, + "grad_norm": 2.691694498062134, + "learning_rate": 2.285877426722497e-06, + "loss": 1.5027, + "step": 1734 + }, + { + "epoch": 1.7822290703646635, + "grad_norm": 2.782968282699585, + "learning_rate": 2.2839741149600305e-06, + "loss": 1.5204, + "step": 1735 + }, + { + "epoch": 1.783256291730868, + "grad_norm": 2.6389904022216797, + "learning_rate": 2.282070803197564e-06, + "loss": 1.5135, + "step": 1736 + }, + { + "epoch": 1.7842835130970724, + "grad_norm": 2.8609848022460938, + "learning_rate": 2.2801674914350972e-06, + "loss": 1.5013, + "step": 1737 + }, + { + "epoch": 1.7853107344632768, + "grad_norm": 2.6814842224121094, + "learning_rate": 2.2782641796726306e-06, + "loss": 1.4874, + "step": 1738 + }, + { + "epoch": 1.7863379558294814, + "grad_norm": 2.7633700370788574, + "learning_rate": 2.276360867910164e-06, + "loss": 1.5129, + "step": 1739 + }, + { + "epoch": 1.7873651771956856, + "grad_norm": 2.637755870819092, + "learning_rate": 2.2744575561476973e-06, + "loss": 1.4486, + "step": 1740 + }, + { + "epoch": 1.7873651771956856, + "eval_loss": 1.8927199840545654, + "eval_runtime": 46.5881, + "eval_samples_per_second": 21.465, + "eval_steps_per_second": 0.902, + "step": 1740 + }, + { + "epoch": 1.7883923985618901, + "grad_norm": 2.815883159637451, + "learning_rate": 2.2725542443852307e-06, + "loss": 1.4815, + "step": 1741 + }, + { + "epoch": 1.7894196199280945, + "grad_norm": 2.758363723754883, + "learning_rate": 2.2706509326227636e-06, + "loss": 1.5235, + "step": 1742 + }, + { + "epoch": 1.7904468412942989, + "grad_norm": 2.6607120037078857, + "learning_rate": 2.268747620860297e-06, + "loss": 1.503, + "step": 1743 + }, + { + "epoch": 1.7914740626605035, + "grad_norm": 2.7479255199432373, + "learning_rate": 2.2668443090978303e-06, + "loss": 1.503, + "step": 1744 + }, + { + "epoch": 1.7925012840267076, + "grad_norm": 2.746718168258667, + "learning_rate": 2.2649409973353637e-06, + "loss": 1.5142, + "step": 1745 + }, + { + "epoch": 1.7935285053929122, + "grad_norm": 2.6848037242889404, + "learning_rate": 2.263037685572897e-06, + "loss": 1.4585, + "step": 1746 + }, + { + "epoch": 1.7945557267591166, + "grad_norm": 2.6677210330963135, + "learning_rate": 2.26113437381043e-06, + "loss": 1.4617, + "step": 1747 + }, + { + "epoch": 1.795582948125321, + "grad_norm": 2.866431474685669, + "learning_rate": 2.259231062047964e-06, + "loss": 1.4945, + "step": 1748 + }, + { + "epoch": 1.7966101694915255, + "grad_norm": 2.953123092651367, + "learning_rate": 2.257327750285497e-06, + "loss": 1.5225, + "step": 1749 + }, + { + "epoch": 1.7976373908577299, + "grad_norm": 2.6214499473571777, + "learning_rate": 2.2554244385230305e-06, + "loss": 1.4707, + "step": 1750 + }, + { + "epoch": 1.7976373908577299, + "eval_loss": 1.8950202465057373, + "eval_runtime": 44.7553, + "eval_samples_per_second": 22.344, + "eval_steps_per_second": 0.938, + "step": 1750 + }, + { + "epoch": 1.7986646122239343, + "grad_norm": 2.7714295387268066, + "learning_rate": 2.2535211267605635e-06, + "loss": 1.4788, + "step": 1751 + }, + { + "epoch": 1.7996918335901386, + "grad_norm": 2.840985059738159, + "learning_rate": 2.251617814998097e-06, + "loss": 1.5004, + "step": 1752 + }, + { + "epoch": 1.800719054956343, + "grad_norm": 2.6215717792510986, + "learning_rate": 2.24971450323563e-06, + "loss": 1.4063, + "step": 1753 + }, + { + "epoch": 1.8017462763225476, + "grad_norm": 2.7458207607269287, + "learning_rate": 2.2478111914731636e-06, + "loss": 1.5045, + "step": 1754 + }, + { + "epoch": 1.802773497688752, + "grad_norm": 2.5622992515563965, + "learning_rate": 2.245907879710697e-06, + "loss": 1.3671, + "step": 1755 + }, + { + "epoch": 1.8038007190549563, + "grad_norm": 3.09394907951355, + "learning_rate": 2.24400456794823e-06, + "loss": 1.5151, + "step": 1756 + }, + { + "epoch": 1.804827940421161, + "grad_norm": 2.740785837173462, + "learning_rate": 2.2421012561857632e-06, + "loss": 1.4635, + "step": 1757 + }, + { + "epoch": 1.805855161787365, + "grad_norm": 2.8012168407440186, + "learning_rate": 2.240197944423297e-06, + "loss": 1.4694, + "step": 1758 + }, + { + "epoch": 1.8068823831535696, + "grad_norm": 2.799586534500122, + "learning_rate": 2.23829463266083e-06, + "loss": 1.4627, + "step": 1759 + }, + { + "epoch": 1.807909604519774, + "grad_norm": 2.6345391273498535, + "learning_rate": 2.2363913208983633e-06, + "loss": 1.4498, + "step": 1760 + }, + { + "epoch": 1.807909604519774, + "eval_loss": 1.9042582511901855, + "eval_runtime": 47.4919, + "eval_samples_per_second": 21.056, + "eval_steps_per_second": 0.884, + "step": 1760 + }, + { + "epoch": 1.8089368258859784, + "grad_norm": 2.715463638305664, + "learning_rate": 2.2344880091358967e-06, + "loss": 1.493, + "step": 1761 + }, + { + "epoch": 1.809964047252183, + "grad_norm": 2.6666505336761475, + "learning_rate": 2.23258469737343e-06, + "loss": 1.4678, + "step": 1762 + }, + { + "epoch": 1.810991268618387, + "grad_norm": 2.6605517864227295, + "learning_rate": 2.2306813856109634e-06, + "loss": 1.5154, + "step": 1763 + }, + { + "epoch": 1.8120184899845917, + "grad_norm": 2.621842384338379, + "learning_rate": 2.2287780738484963e-06, + "loss": 1.4626, + "step": 1764 + }, + { + "epoch": 1.813045711350796, + "grad_norm": 2.7204320430755615, + "learning_rate": 2.2268747620860297e-06, + "loss": 1.5461, + "step": 1765 + }, + { + "epoch": 1.8140729327170004, + "grad_norm": 2.8174679279327393, + "learning_rate": 2.224971450323563e-06, + "loss": 1.4881, + "step": 1766 + }, + { + "epoch": 1.815100154083205, + "grad_norm": 2.707890272140503, + "learning_rate": 2.2230681385610964e-06, + "loss": 1.4123, + "step": 1767 + }, + { + "epoch": 1.8161273754494094, + "grad_norm": 2.753523111343384, + "learning_rate": 2.2211648267986298e-06, + "loss": 1.4783, + "step": 1768 + }, + { + "epoch": 1.8171545968156138, + "grad_norm": 2.830095052719116, + "learning_rate": 2.219261515036163e-06, + "loss": 1.4536, + "step": 1769 + }, + { + "epoch": 1.8181818181818183, + "grad_norm": 2.803318738937378, + "learning_rate": 2.2173582032736965e-06, + "loss": 1.4939, + "step": 1770 + }, + { + "epoch": 1.8181818181818183, + "eval_loss": 1.899889349937439, + "eval_runtime": 46.7673, + "eval_samples_per_second": 21.382, + "eval_steps_per_second": 0.898, + "step": 1770 + }, + { + "epoch": 1.8192090395480225, + "grad_norm": 2.8235929012298584, + "learning_rate": 2.21545489151123e-06, + "loss": 1.5176, + "step": 1771 + }, + { + "epoch": 1.820236260914227, + "grad_norm": 2.761497974395752, + "learning_rate": 2.2135515797487632e-06, + "loss": 1.4685, + "step": 1772 + }, + { + "epoch": 1.8212634822804314, + "grad_norm": 2.589054822921753, + "learning_rate": 2.211648267986296e-06, + "loss": 1.4673, + "step": 1773 + }, + { + "epoch": 1.8222907036466358, + "grad_norm": 2.763709306716919, + "learning_rate": 2.2097449562238295e-06, + "loss": 1.4727, + "step": 1774 + }, + { + "epoch": 1.8233179250128404, + "grad_norm": 2.7250869274139404, + "learning_rate": 2.207841644461363e-06, + "loss": 1.4729, + "step": 1775 + }, + { + "epoch": 1.8243451463790445, + "grad_norm": 2.7706706523895264, + "learning_rate": 2.2059383326988963e-06, + "loss": 1.5382, + "step": 1776 + }, + { + "epoch": 1.8253723677452491, + "grad_norm": 2.9034879207611084, + "learning_rate": 2.2040350209364296e-06, + "loss": 1.4936, + "step": 1777 + }, + { + "epoch": 1.8263995891114535, + "grad_norm": 2.625908851623535, + "learning_rate": 2.2021317091739626e-06, + "loss": 1.4601, + "step": 1778 + }, + { + "epoch": 1.8274268104776579, + "grad_norm": 2.703167676925659, + "learning_rate": 2.2002283974114963e-06, + "loss": 1.4876, + "step": 1779 + }, + { + "epoch": 1.8284540318438625, + "grad_norm": 2.8047842979431152, + "learning_rate": 2.1983250856490297e-06, + "loss": 1.428, + "step": 1780 + }, + { + "epoch": 1.8284540318438625, + "eval_loss": 1.8968960046768188, + "eval_runtime": 46.5826, + "eval_samples_per_second": 21.467, + "eval_steps_per_second": 0.902, + "step": 1780 + }, + { + "epoch": 1.8294812532100666, + "grad_norm": 2.8730416297912598, + "learning_rate": 2.1964217738865626e-06, + "loss": 1.4697, + "step": 1781 + }, + { + "epoch": 1.8305084745762712, + "grad_norm": 2.958203077316284, + "learning_rate": 2.194518462124096e-06, + "loss": 1.4955, + "step": 1782 + }, + { + "epoch": 1.8315356959424756, + "grad_norm": 2.8280398845672607, + "learning_rate": 2.1926151503616294e-06, + "loss": 1.5391, + "step": 1783 + }, + { + "epoch": 1.83256291730868, + "grad_norm": 2.776029348373413, + "learning_rate": 2.1907118385991627e-06, + "loss": 1.4945, + "step": 1784 + }, + { + "epoch": 1.8335901386748845, + "grad_norm": 2.7049412727355957, + "learning_rate": 2.188808526836696e-06, + "loss": 1.4481, + "step": 1785 + }, + { + "epoch": 1.8346173600410889, + "grad_norm": 2.71523118019104, + "learning_rate": 2.1869052150742295e-06, + "loss": 1.4756, + "step": 1786 + }, + { + "epoch": 1.8356445814072933, + "grad_norm": 2.7659823894500732, + "learning_rate": 2.1850019033117624e-06, + "loss": 1.4065, + "step": 1787 + }, + { + "epoch": 1.8366718027734978, + "grad_norm": 2.679553508758545, + "learning_rate": 2.1830985915492958e-06, + "loss": 1.4744, + "step": 1788 + }, + { + "epoch": 1.837699024139702, + "grad_norm": 2.6452908515930176, + "learning_rate": 2.1811952797868295e-06, + "loss": 1.4922, + "step": 1789 + }, + { + "epoch": 1.8387262455059066, + "grad_norm": 2.780548095703125, + "learning_rate": 2.1792919680243625e-06, + "loss": 1.458, + "step": 1790 + }, + { + "epoch": 1.8387262455059066, + "eval_loss": 1.8877090215682983, + "eval_runtime": 47.6844, + "eval_samples_per_second": 20.971, + "eval_steps_per_second": 0.881, + "step": 1790 + }, + { + "epoch": 1.839753466872111, + "grad_norm": 2.8762049674987793, + "learning_rate": 2.177388656261896e-06, + "loss": 1.489, + "step": 1791 + }, + { + "epoch": 1.8407806882383153, + "grad_norm": 3.0053279399871826, + "learning_rate": 2.1754853444994292e-06, + "loss": 1.5681, + "step": 1792 + }, + { + "epoch": 1.84180790960452, + "grad_norm": 2.737626791000366, + "learning_rate": 2.1735820327369626e-06, + "loss": 1.3764, + "step": 1793 + }, + { + "epoch": 1.842835130970724, + "grad_norm": 2.6998884677886963, + "learning_rate": 2.171678720974496e-06, + "loss": 1.5099, + "step": 1794 + }, + { + "epoch": 1.8438623523369286, + "grad_norm": 2.8110108375549316, + "learning_rate": 2.169775409212029e-06, + "loss": 1.5079, + "step": 1795 + }, + { + "epoch": 1.844889573703133, + "grad_norm": 2.7312750816345215, + "learning_rate": 2.1678720974495622e-06, + "loss": 1.5136, + "step": 1796 + }, + { + "epoch": 1.8459167950693374, + "grad_norm": 2.692768096923828, + "learning_rate": 2.1659687856870956e-06, + "loss": 1.5005, + "step": 1797 + }, + { + "epoch": 1.846944016435542, + "grad_norm": 2.673879623413086, + "learning_rate": 2.164065473924629e-06, + "loss": 1.5127, + "step": 1798 + }, + { + "epoch": 1.8479712378017463, + "grad_norm": 2.6218647956848145, + "learning_rate": 2.1621621621621623e-06, + "loss": 1.4529, + "step": 1799 + }, + { + "epoch": 1.8489984591679507, + "grad_norm": 2.7149012088775635, + "learning_rate": 2.1602588503996957e-06, + "loss": 1.5048, + "step": 1800 + }, + { + "epoch": 1.8489984591679507, + "eval_loss": 1.883215069770813, + "eval_runtime": 50.8526, + "eval_samples_per_second": 19.665, + "eval_steps_per_second": 0.826, + "step": 1800 + }, + { + "epoch": 1.850025680534155, + "grad_norm": 2.60640025138855, + "learning_rate": 2.158355538637229e-06, + "loss": 1.4827, + "step": 1801 + }, + { + "epoch": 1.8510529019003594, + "grad_norm": 2.7535810470581055, + "learning_rate": 2.1564522268747624e-06, + "loss": 1.4709, + "step": 1802 + }, + { + "epoch": 1.852080123266564, + "grad_norm": 2.8784072399139404, + "learning_rate": 2.1545489151122958e-06, + "loss": 1.548, + "step": 1803 + }, + { + "epoch": 1.8531073446327684, + "grad_norm": 2.7007954120635986, + "learning_rate": 2.1526456033498287e-06, + "loss": 1.4926, + "step": 1804 + }, + { + "epoch": 1.8541345659989728, + "grad_norm": 2.699535369873047, + "learning_rate": 2.150742291587362e-06, + "loss": 1.4993, + "step": 1805 + }, + { + "epoch": 1.8551617873651773, + "grad_norm": 2.777305841445923, + "learning_rate": 2.1488389798248954e-06, + "loss": 1.4822, + "step": 1806 + }, + { + "epoch": 1.8561890087313815, + "grad_norm": 2.6989753246307373, + "learning_rate": 2.146935668062429e-06, + "loss": 1.515, + "step": 1807 + }, + { + "epoch": 1.857216230097586, + "grad_norm": 2.7000770568847656, + "learning_rate": 2.145032356299962e-06, + "loss": 1.4278, + "step": 1808 + }, + { + "epoch": 1.8582434514637904, + "grad_norm": 2.706331729888916, + "learning_rate": 2.1431290445374955e-06, + "loss": 1.4846, + "step": 1809 + }, + { + "epoch": 1.8592706728299948, + "grad_norm": 2.683302402496338, + "learning_rate": 2.141225732775029e-06, + "loss": 1.4501, + "step": 1810 + }, + { + "epoch": 1.8592706728299948, + "eval_loss": 1.8919310569763184, + "eval_runtime": 43.2772, + "eval_samples_per_second": 23.107, + "eval_steps_per_second": 0.97, + "step": 1810 + }, + { + "epoch": 1.8602978941961994, + "grad_norm": 2.9528095722198486, + "learning_rate": 2.1393224210125623e-06, + "loss": 1.4895, + "step": 1811 + }, + { + "epoch": 1.8613251155624035, + "grad_norm": 2.6797986030578613, + "learning_rate": 2.137419109250095e-06, + "loss": 1.4815, + "step": 1812 + }, + { + "epoch": 1.8623523369286081, + "grad_norm": 2.753924608230591, + "learning_rate": 2.1355157974876286e-06, + "loss": 1.5224, + "step": 1813 + }, + { + "epoch": 1.8633795582948125, + "grad_norm": 2.8037896156311035, + "learning_rate": 2.133612485725162e-06, + "loss": 1.5203, + "step": 1814 + }, + { + "epoch": 1.8644067796610169, + "grad_norm": 2.7934775352478027, + "learning_rate": 2.1317091739626953e-06, + "loss": 1.4307, + "step": 1815 + }, + { + "epoch": 1.8654340010272215, + "grad_norm": 2.919184684753418, + "learning_rate": 2.1298058622002286e-06, + "loss": 1.4654, + "step": 1816 + }, + { + "epoch": 1.8664612223934258, + "grad_norm": 2.671267509460449, + "learning_rate": 2.1279025504377616e-06, + "loss": 1.4776, + "step": 1817 + }, + { + "epoch": 1.8674884437596302, + "grad_norm": 2.6870152950286865, + "learning_rate": 2.125999238675295e-06, + "loss": 1.4759, + "step": 1818 + }, + { + "epoch": 1.8685156651258348, + "grad_norm": 2.8859241008758545, + "learning_rate": 2.1240959269128287e-06, + "loss": 1.515, + "step": 1819 + }, + { + "epoch": 1.869542886492039, + "grad_norm": 2.773226261138916, + "learning_rate": 2.122192615150362e-06, + "loss": 1.4743, + "step": 1820 + }, + { + "epoch": 1.869542886492039, + "eval_loss": 1.8901108503341675, + "eval_runtime": 45.5952, + "eval_samples_per_second": 21.932, + "eval_steps_per_second": 0.921, + "step": 1820 + }, + { + "epoch": 1.8705701078582435, + "grad_norm": 2.591169595718384, + "learning_rate": 2.120289303387895e-06, + "loss": 1.4245, + "step": 1821 + }, + { + "epoch": 1.8715973292244479, + "grad_norm": 2.977644205093384, + "learning_rate": 2.1183859916254284e-06, + "loss": 1.4755, + "step": 1822 + }, + { + "epoch": 1.8726245505906522, + "grad_norm": 2.7917141914367676, + "learning_rate": 2.1164826798629618e-06, + "loss": 1.5061, + "step": 1823 + }, + { + "epoch": 1.8736517719568568, + "grad_norm": 2.6995670795440674, + "learning_rate": 2.114579368100495e-06, + "loss": 1.4627, + "step": 1824 + }, + { + "epoch": 1.874678993323061, + "grad_norm": 2.7223551273345947, + "learning_rate": 2.1126760563380285e-06, + "loss": 1.4495, + "step": 1825 + }, + { + "epoch": 1.8757062146892656, + "grad_norm": 2.749028205871582, + "learning_rate": 2.1107727445755614e-06, + "loss": 1.4262, + "step": 1826 + }, + { + "epoch": 1.87673343605547, + "grad_norm": 3.0993876457214355, + "learning_rate": 2.1088694328130948e-06, + "loss": 1.5554, + "step": 1827 + }, + { + "epoch": 1.8777606574216743, + "grad_norm": 2.6830661296844482, + "learning_rate": 2.106966121050628e-06, + "loss": 1.4404, + "step": 1828 + }, + { + "epoch": 1.878787878787879, + "grad_norm": 2.7380309104919434, + "learning_rate": 2.1050628092881615e-06, + "loss": 1.5113, + "step": 1829 + }, + { + "epoch": 1.879815100154083, + "grad_norm": 2.873230457305908, + "learning_rate": 2.103159497525695e-06, + "loss": 1.545, + "step": 1830 + }, + { + "epoch": 1.879815100154083, + "eval_loss": 1.8950482606887817, + "eval_runtime": 46.8358, + "eval_samples_per_second": 21.351, + "eval_steps_per_second": 0.897, + "step": 1830 + }, + { + "epoch": 1.8808423215202876, + "grad_norm": 2.805757522583008, + "learning_rate": 2.1012561857632282e-06, + "loss": 1.4593, + "step": 1831 + }, + { + "epoch": 1.881869542886492, + "grad_norm": 2.7594594955444336, + "learning_rate": 2.0993528740007616e-06, + "loss": 1.5167, + "step": 1832 + }, + { + "epoch": 1.8828967642526964, + "grad_norm": 2.810119152069092, + "learning_rate": 2.097449562238295e-06, + "loss": 1.5199, + "step": 1833 + }, + { + "epoch": 1.883923985618901, + "grad_norm": 2.6915268898010254, + "learning_rate": 2.095546250475828e-06, + "loss": 1.47, + "step": 1834 + }, + { + "epoch": 1.8849512069851053, + "grad_norm": 2.966466188430786, + "learning_rate": 2.0936429387133613e-06, + "loss": 1.4815, + "step": 1835 + }, + { + "epoch": 1.8859784283513097, + "grad_norm": 2.719665050506592, + "learning_rate": 2.0917396269508946e-06, + "loss": 1.537, + "step": 1836 + }, + { + "epoch": 1.8870056497175143, + "grad_norm": 2.7843384742736816, + "learning_rate": 2.089836315188428e-06, + "loss": 1.5102, + "step": 1837 + }, + { + "epoch": 1.8880328710837184, + "grad_norm": 2.7335705757141113, + "learning_rate": 2.0879330034259614e-06, + "loss": 1.5199, + "step": 1838 + }, + { + "epoch": 1.889060092449923, + "grad_norm": 2.9173038005828857, + "learning_rate": 2.0860296916634947e-06, + "loss": 1.4605, + "step": 1839 + }, + { + "epoch": 1.8900873138161274, + "grad_norm": 2.7791881561279297, + "learning_rate": 2.084126379901028e-06, + "loss": 1.5482, + "step": 1840 + }, + { + "epoch": 1.8900873138161274, + "eval_loss": 1.8914867639541626, + "eval_runtime": 49.5764, + "eval_samples_per_second": 20.171, + "eval_steps_per_second": 0.847, + "step": 1840 + }, + { + "epoch": 1.8911145351823317, + "grad_norm": 2.700892686843872, + "learning_rate": 2.0822230681385614e-06, + "loss": 1.4509, + "step": 1841 + }, + { + "epoch": 1.8921417565485363, + "grad_norm": 2.9125330448150635, + "learning_rate": 2.080319756376095e-06, + "loss": 1.5081, + "step": 1842 + }, + { + "epoch": 1.8931689779147405, + "grad_norm": 2.762075424194336, + "learning_rate": 2.0784164446136277e-06, + "loss": 1.4989, + "step": 1843 + }, + { + "epoch": 1.894196199280945, + "grad_norm": 2.782738447189331, + "learning_rate": 2.076513132851161e-06, + "loss": 1.4643, + "step": 1844 + }, + { + "epoch": 1.8952234206471494, + "grad_norm": 2.7540040016174316, + "learning_rate": 2.0746098210886945e-06, + "loss": 1.5029, + "step": 1845 + }, + { + "epoch": 1.8962506420133538, + "grad_norm": 2.6205410957336426, + "learning_rate": 2.072706509326228e-06, + "loss": 1.4674, + "step": 1846 + }, + { + "epoch": 1.8972778633795584, + "grad_norm": 2.6946957111358643, + "learning_rate": 2.070803197563761e-06, + "loss": 1.4747, + "step": 1847 + }, + { + "epoch": 1.8983050847457628, + "grad_norm": 2.9114229679107666, + "learning_rate": 2.068899885801294e-06, + "loss": 1.4809, + "step": 1848 + }, + { + "epoch": 1.8993323061119671, + "grad_norm": 2.823211431503296, + "learning_rate": 2.066996574038828e-06, + "loss": 1.4887, + "step": 1849 + }, + { + "epoch": 1.9003595274781715, + "grad_norm": 2.779381513595581, + "learning_rate": 2.0650932622763613e-06, + "loss": 1.5217, + "step": 1850 + }, + { + "epoch": 1.9003595274781715, + "eval_loss": 1.8863105773925781, + "eval_runtime": 46.2064, + "eval_samples_per_second": 21.642, + "eval_steps_per_second": 0.909, + "step": 1850 + }, + { + "epoch": 1.9013867488443759, + "grad_norm": 2.7235658168792725, + "learning_rate": 2.0631899505138942e-06, + "loss": 1.4221, + "step": 1851 + }, + { + "epoch": 1.9024139702105805, + "grad_norm": 2.6994757652282715, + "learning_rate": 2.0612866387514276e-06, + "loss": 1.461, + "step": 1852 + }, + { + "epoch": 1.9034411915767848, + "grad_norm": 2.7634339332580566, + "learning_rate": 2.059383326988961e-06, + "loss": 1.4975, + "step": 1853 + }, + { + "epoch": 1.9044684129429892, + "grad_norm": 2.78421688079834, + "learning_rate": 2.0574800152264943e-06, + "loss": 1.4757, + "step": 1854 + }, + { + "epoch": 1.9054956343091938, + "grad_norm": 2.7357447147369385, + "learning_rate": 2.0555767034640277e-06, + "loss": 1.4524, + "step": 1855 + }, + { + "epoch": 1.906522855675398, + "grad_norm": 2.712550163269043, + "learning_rate": 2.053673391701561e-06, + "loss": 1.4232, + "step": 1856 + }, + { + "epoch": 1.9075500770416025, + "grad_norm": 2.7226524353027344, + "learning_rate": 2.051770079939094e-06, + "loss": 1.4846, + "step": 1857 + }, + { + "epoch": 1.9085772984078069, + "grad_norm": 2.762127637863159, + "learning_rate": 2.0498667681766273e-06, + "loss": 1.447, + "step": 1858 + }, + { + "epoch": 1.9096045197740112, + "grad_norm": 2.7561259269714355, + "learning_rate": 2.047963456414161e-06, + "loss": 1.4512, + "step": 1859 + }, + { + "epoch": 1.9106317411402158, + "grad_norm": 2.7032148838043213, + "learning_rate": 2.046060144651694e-06, + "loss": 1.4783, + "step": 1860 + }, + { + "epoch": 1.9106317411402158, + "eval_loss": 1.8838176727294922, + "eval_runtime": 46.0974, + "eval_samples_per_second": 21.693, + "eval_steps_per_second": 0.911, + "step": 1860 + }, + { + "epoch": 1.91165896250642, + "grad_norm": 2.8010241985321045, + "learning_rate": 2.0441568328892274e-06, + "loss": 1.4898, + "step": 1861 + }, + { + "epoch": 1.9126861838726246, + "grad_norm": 2.7780871391296387, + "learning_rate": 2.0422535211267608e-06, + "loss": 1.4848, + "step": 1862 + }, + { + "epoch": 1.913713405238829, + "grad_norm": 2.712059259414673, + "learning_rate": 2.040350209364294e-06, + "loss": 1.459, + "step": 1863 + }, + { + "epoch": 1.9147406266050333, + "grad_norm": 2.642489194869995, + "learning_rate": 2.0384468976018275e-06, + "loss": 1.4534, + "step": 1864 + }, + { + "epoch": 1.915767847971238, + "grad_norm": 2.645132064819336, + "learning_rate": 2.0365435858393604e-06, + "loss": 1.4573, + "step": 1865 + }, + { + "epoch": 1.9167950693374423, + "grad_norm": 2.747443914413452, + "learning_rate": 2.034640274076894e-06, + "loss": 1.4505, + "step": 1866 + }, + { + "epoch": 1.9178222907036466, + "grad_norm": 2.660188674926758, + "learning_rate": 2.032736962314427e-06, + "loss": 1.5146, + "step": 1867 + }, + { + "epoch": 1.9188495120698512, + "grad_norm": 2.7249484062194824, + "learning_rate": 2.0308336505519605e-06, + "loss": 1.4553, + "step": 1868 + }, + { + "epoch": 1.9198767334360554, + "grad_norm": 2.670896530151367, + "learning_rate": 2.028930338789494e-06, + "loss": 1.5204, + "step": 1869 + }, + { + "epoch": 1.92090395480226, + "grad_norm": 2.858780860900879, + "learning_rate": 2.0270270270270273e-06, + "loss": 1.446, + "step": 1870 + }, + { + "epoch": 1.92090395480226, + "eval_loss": 1.8811051845550537, + "eval_runtime": 45.7097, + "eval_samples_per_second": 21.877, + "eval_steps_per_second": 0.919, + "step": 1870 + }, + { + "epoch": 1.9219311761684643, + "grad_norm": 2.744133949279785, + "learning_rate": 2.0251237152645606e-06, + "loss": 1.4666, + "step": 1871 + }, + { + "epoch": 1.9229583975346687, + "grad_norm": 2.747067928314209, + "learning_rate": 2.023220403502094e-06, + "loss": 1.4466, + "step": 1872 + }, + { + "epoch": 1.9239856189008733, + "grad_norm": 2.7786505222320557, + "learning_rate": 2.0213170917396273e-06, + "loss": 1.4233, + "step": 1873 + }, + { + "epoch": 1.9250128402670774, + "grad_norm": 2.793522357940674, + "learning_rate": 2.0194137799771603e-06, + "loss": 1.4683, + "step": 1874 + }, + { + "epoch": 1.926040061633282, + "grad_norm": 2.7715535163879395, + "learning_rate": 2.0175104682146937e-06, + "loss": 1.4696, + "step": 1875 + }, + { + "epoch": 1.9270672829994864, + "grad_norm": 2.7226290702819824, + "learning_rate": 2.015607156452227e-06, + "loss": 1.412, + "step": 1876 + }, + { + "epoch": 1.9280945043656907, + "grad_norm": 2.722703456878662, + "learning_rate": 2.0137038446897604e-06, + "loss": 1.4503, + "step": 1877 + }, + { + "epoch": 1.9291217257318953, + "grad_norm": 2.717355251312256, + "learning_rate": 2.0118005329272937e-06, + "loss": 1.4423, + "step": 1878 + }, + { + "epoch": 1.9301489470980995, + "grad_norm": 2.719848394393921, + "learning_rate": 2.0098972211648267e-06, + "loss": 1.4609, + "step": 1879 + }, + { + "epoch": 1.931176168464304, + "grad_norm": 2.693844795227051, + "learning_rate": 2.0079939094023605e-06, + "loss": 1.4606, + "step": 1880 + }, + { + "epoch": 1.931176168464304, + "eval_loss": 1.8892173767089844, + "eval_runtime": 45.2703, + "eval_samples_per_second": 22.09, + "eval_steps_per_second": 0.928, + "step": 1880 + }, + { + "epoch": 1.9322033898305084, + "grad_norm": 2.649186372756958, + "learning_rate": 2.006090597639894e-06, + "loss": 1.5056, + "step": 1881 + }, + { + "epoch": 1.9332306111967128, + "grad_norm": 2.738987922668457, + "learning_rate": 2.0041872858774268e-06, + "loss": 1.5321, + "step": 1882 + }, + { + "epoch": 1.9342578325629174, + "grad_norm": 2.6583259105682373, + "learning_rate": 2.00228397411496e-06, + "loss": 1.5004, + "step": 1883 + }, + { + "epoch": 1.9352850539291218, + "grad_norm": 2.625006675720215, + "learning_rate": 2.0003806623524935e-06, + "loss": 1.4897, + "step": 1884 + }, + { + "epoch": 1.9363122752953261, + "grad_norm": 2.6956043243408203, + "learning_rate": 1.998477350590027e-06, + "loss": 1.4049, + "step": 1885 + }, + { + "epoch": 1.9373394966615307, + "grad_norm": 2.668865203857422, + "learning_rate": 1.9965740388275602e-06, + "loss": 1.4999, + "step": 1886 + }, + { + "epoch": 1.9383667180277349, + "grad_norm": 2.623006820678711, + "learning_rate": 1.994670727065093e-06, + "loss": 1.4742, + "step": 1887 + }, + { + "epoch": 1.9393939393939394, + "grad_norm": 2.7890007495880127, + "learning_rate": 1.9927674153026265e-06, + "loss": 1.4332, + "step": 1888 + }, + { + "epoch": 1.9404211607601438, + "grad_norm": 2.7803823947906494, + "learning_rate": 1.99086410354016e-06, + "loss": 1.5016, + "step": 1889 + }, + { + "epoch": 1.9414483821263482, + "grad_norm": 2.801680088043213, + "learning_rate": 1.9889607917776937e-06, + "loss": 1.4577, + "step": 1890 + }, + { + "epoch": 1.9414483821263482, + "eval_loss": 1.8877300024032593, + "eval_runtime": 43.6778, + "eval_samples_per_second": 22.895, + "eval_steps_per_second": 0.962, + "step": 1890 + }, + { + "epoch": 1.9424756034925528, + "grad_norm": 2.954679489135742, + "learning_rate": 1.9870574800152266e-06, + "loss": 1.4811, + "step": 1891 + }, + { + "epoch": 1.943502824858757, + "grad_norm": 2.6189258098602295, + "learning_rate": 1.98515416825276e-06, + "loss": 1.4955, + "step": 1892 + }, + { + "epoch": 1.9445300462249615, + "grad_norm": 2.7194509506225586, + "learning_rate": 1.9832508564902933e-06, + "loss": 1.4833, + "step": 1893 + }, + { + "epoch": 1.9455572675911659, + "grad_norm": 2.5767593383789062, + "learning_rate": 1.9813475447278267e-06, + "loss": 1.3973, + "step": 1894 + }, + { + "epoch": 1.9465844889573702, + "grad_norm": 2.7513067722320557, + "learning_rate": 1.97944423296536e-06, + "loss": 1.4814, + "step": 1895 + }, + { + "epoch": 1.9476117103235748, + "grad_norm": 2.712155818939209, + "learning_rate": 1.977540921202893e-06, + "loss": 1.4686, + "step": 1896 + }, + { + "epoch": 1.9486389316897792, + "grad_norm": 2.582233190536499, + "learning_rate": 1.9756376094404264e-06, + "loss": 1.4418, + "step": 1897 + }, + { + "epoch": 1.9496661530559836, + "grad_norm": 2.6065144538879395, + "learning_rate": 1.9737342976779597e-06, + "loss": 1.4198, + "step": 1898 + }, + { + "epoch": 1.950693374422188, + "grad_norm": 2.7891361713409424, + "learning_rate": 1.971830985915493e-06, + "loss": 1.5054, + "step": 1899 + }, + { + "epoch": 1.9517205957883923, + "grad_norm": 2.7199501991271973, + "learning_rate": 1.9699276741530264e-06, + "loss": 1.462, + "step": 1900 + }, + { + "epoch": 1.9517205957883923, + "eval_loss": 1.8931901454925537, + "eval_runtime": 44.0781, + "eval_samples_per_second": 22.687, + "eval_steps_per_second": 0.953, + "step": 1900 + }, + { + "epoch": 1.9527478171545969, + "grad_norm": 2.742461681365967, + "learning_rate": 1.96802436239056e-06, + "loss": 1.4716, + "step": 1901 + }, + { + "epoch": 1.9537750385208013, + "grad_norm": 2.689281702041626, + "learning_rate": 1.966121050628093e-06, + "loss": 1.4613, + "step": 1902 + }, + { + "epoch": 1.9548022598870056, + "grad_norm": 2.722919225692749, + "learning_rate": 1.9642177388656265e-06, + "loss": 1.4901, + "step": 1903 + }, + { + "epoch": 1.9558294812532102, + "grad_norm": 2.773724317550659, + "learning_rate": 1.9623144271031595e-06, + "loss": 1.5116, + "step": 1904 + }, + { + "epoch": 1.9568567026194144, + "grad_norm": 2.7967145442962646, + "learning_rate": 1.960411115340693e-06, + "loss": 1.5249, + "step": 1905 + }, + { + "epoch": 1.957883923985619, + "grad_norm": 2.7624928951263428, + "learning_rate": 1.958507803578226e-06, + "loss": 1.4751, + "step": 1906 + }, + { + "epoch": 1.9589111453518233, + "grad_norm": 2.787565231323242, + "learning_rate": 1.9566044918157596e-06, + "loss": 1.4878, + "step": 1907 + }, + { + "epoch": 1.9599383667180277, + "grad_norm": 2.6808853149414062, + "learning_rate": 1.954701180053293e-06, + "loss": 1.4963, + "step": 1908 + }, + { + "epoch": 1.9609655880842323, + "grad_norm": 2.7684340476989746, + "learning_rate": 1.952797868290826e-06, + "loss": 1.4528, + "step": 1909 + }, + { + "epoch": 1.9619928094504364, + "grad_norm": 2.7576661109924316, + "learning_rate": 1.9508945565283596e-06, + "loss": 1.5019, + "step": 1910 + }, + { + "epoch": 1.9619928094504364, + "eval_loss": 1.8857277631759644, + "eval_runtime": 43.0174, + "eval_samples_per_second": 23.246, + "eval_steps_per_second": 0.976, + "step": 1910 + }, + { + "epoch": 1.963020030816641, + "grad_norm": 2.6631407737731934, + "learning_rate": 1.948991244765893e-06, + "loss": 1.4638, + "step": 1911 + }, + { + "epoch": 1.9640472521828454, + "grad_norm": 2.720525026321411, + "learning_rate": 1.9470879330034264e-06, + "loss": 1.4951, + "step": 1912 + }, + { + "epoch": 1.9650744735490497, + "grad_norm": 2.780932903289795, + "learning_rate": 1.9451846212409593e-06, + "loss": 1.4351, + "step": 1913 + }, + { + "epoch": 1.9661016949152543, + "grad_norm": 2.8517262935638428, + "learning_rate": 1.9432813094784927e-06, + "loss": 1.4941, + "step": 1914 + }, + { + "epoch": 1.9671289162814587, + "grad_norm": 2.7812676429748535, + "learning_rate": 1.941377997716026e-06, + "loss": 1.4383, + "step": 1915 + }, + { + "epoch": 1.968156137647663, + "grad_norm": 2.815864086151123, + "learning_rate": 1.9394746859535594e-06, + "loss": 1.4562, + "step": 1916 + }, + { + "epoch": 1.9691833590138677, + "grad_norm": 2.841853618621826, + "learning_rate": 1.9375713741910928e-06, + "loss": 1.4289, + "step": 1917 + }, + { + "epoch": 1.9702105803800718, + "grad_norm": 2.751080274581909, + "learning_rate": 1.9356680624286257e-06, + "loss": 1.4567, + "step": 1918 + }, + { + "epoch": 1.9712378017462764, + "grad_norm": 2.7139065265655518, + "learning_rate": 1.933764750666159e-06, + "loss": 1.4158, + "step": 1919 + }, + { + "epoch": 1.9722650231124808, + "grad_norm": 2.6788330078125, + "learning_rate": 1.931861438903693e-06, + "loss": 1.4535, + "step": 1920 + }, + { + "epoch": 1.9722650231124808, + "eval_loss": 1.8828307390213013, + "eval_runtime": 44.8729, + "eval_samples_per_second": 22.285, + "eval_steps_per_second": 0.936, + "step": 1920 + }, + { + "epoch": 1.9732922444786851, + "grad_norm": 2.695631742477417, + "learning_rate": 1.929958127141226e-06, + "loss": 1.4348, + "step": 1921 + }, + { + "epoch": 1.9743194658448897, + "grad_norm": 2.6957788467407227, + "learning_rate": 1.928054815378759e-06, + "loss": 1.4627, + "step": 1922 + }, + { + "epoch": 1.9753466872110939, + "grad_norm": 2.8832242488861084, + "learning_rate": 1.9261515036162925e-06, + "loss": 1.5429, + "step": 1923 + }, + { + "epoch": 1.9763739085772984, + "grad_norm": 2.744378089904785, + "learning_rate": 1.924248191853826e-06, + "loss": 1.4973, + "step": 1924 + }, + { + "epoch": 1.9774011299435028, + "grad_norm": 2.695059299468994, + "learning_rate": 1.9223448800913592e-06, + "loss": 1.484, + "step": 1925 + }, + { + "epoch": 1.9784283513097072, + "grad_norm": 2.719637870788574, + "learning_rate": 1.9204415683288926e-06, + "loss": 1.4536, + "step": 1926 + }, + { + "epoch": 1.9794555726759118, + "grad_norm": 2.7248711585998535, + "learning_rate": 1.9185382565664255e-06, + "loss": 1.502, + "step": 1927 + }, + { + "epoch": 1.980482794042116, + "grad_norm": 2.6778104305267334, + "learning_rate": 1.916634944803959e-06, + "loss": 1.5065, + "step": 1928 + }, + { + "epoch": 1.9815100154083205, + "grad_norm": 2.859102964401245, + "learning_rate": 1.9147316330414923e-06, + "loss": 1.4791, + "step": 1929 + }, + { + "epoch": 1.9825372367745249, + "grad_norm": 2.708515167236328, + "learning_rate": 1.9128283212790256e-06, + "loss": 1.4693, + "step": 1930 + }, + { + "epoch": 1.9825372367745249, + "eval_loss": 1.8875067234039307, + "eval_runtime": 44.2114, + "eval_samples_per_second": 22.619, + "eval_steps_per_second": 0.95, + "step": 1930 + }, + { + "epoch": 1.9835644581407292, + "grad_norm": 2.6972970962524414, + "learning_rate": 1.910925009516559e-06, + "loss": 1.5149, + "step": 1931 + }, + { + "epoch": 1.9845916795069338, + "grad_norm": 2.800899028778076, + "learning_rate": 1.9090216977540924e-06, + "loss": 1.4785, + "step": 1932 + }, + { + "epoch": 1.9856189008731382, + "grad_norm": 2.8238046169281006, + "learning_rate": 1.9071183859916257e-06, + "loss": 1.5203, + "step": 1933 + }, + { + "epoch": 1.9866461222393426, + "grad_norm": 2.6949214935302734, + "learning_rate": 1.9052150742291589e-06, + "loss": 1.4919, + "step": 1934 + }, + { + "epoch": 1.9876733436055471, + "grad_norm": 2.7611637115478516, + "learning_rate": 1.9033117624666922e-06, + "loss": 1.4889, + "step": 1935 + }, + { + "epoch": 1.9887005649717513, + "grad_norm": 2.704080581665039, + "learning_rate": 1.9014084507042254e-06, + "loss": 1.4598, + "step": 1936 + }, + { + "epoch": 1.9897277863379559, + "grad_norm": 2.8081612586975098, + "learning_rate": 1.8995051389417587e-06, + "loss": 1.4324, + "step": 1937 + }, + { + "epoch": 1.9907550077041603, + "grad_norm": 2.597720146179199, + "learning_rate": 1.8976018271792921e-06, + "loss": 1.4196, + "step": 1938 + }, + { + "epoch": 1.9917822290703646, + "grad_norm": 2.716373920440674, + "learning_rate": 1.8956985154168253e-06, + "loss": 1.4795, + "step": 1939 + }, + { + "epoch": 1.9928094504365692, + "grad_norm": 2.7661051750183105, + "learning_rate": 1.8937952036543588e-06, + "loss": 1.457, + "step": 1940 + }, + { + "epoch": 1.9928094504365692, + "eval_loss": 1.8831615447998047, + "eval_runtime": 46.9632, + "eval_samples_per_second": 21.293, + "eval_steps_per_second": 0.894, + "step": 1940 + }, + { + "epoch": 1.9938366718027734, + "grad_norm": 2.66249942779541, + "learning_rate": 1.8918918918918922e-06, + "loss": 1.4213, + "step": 1941 + }, + { + "epoch": 1.994863893168978, + "grad_norm": 2.7329354286193848, + "learning_rate": 1.8899885801294253e-06, + "loss": 1.4707, + "step": 1942 + }, + { + "epoch": 1.9958911145351823, + "grad_norm": 2.844203472137451, + "learning_rate": 1.8880852683669587e-06, + "loss": 1.4847, + "step": 1943 + }, + { + "epoch": 1.9969183359013867, + "grad_norm": 2.9034390449523926, + "learning_rate": 1.886181956604492e-06, + "loss": 1.4605, + "step": 1944 + }, + { + "epoch": 1.9979455572675913, + "grad_norm": 2.8142337799072266, + "learning_rate": 1.8842786448420252e-06, + "loss": 1.4529, + "step": 1945 + }, + { + "epoch": 1.9989727786337956, + "grad_norm": 2.8397836685180664, + "learning_rate": 1.8823753330795586e-06, + "loss": 1.4863, + "step": 1946 + }, + { + "epoch": 2.0, + "grad_norm": 2.79947829246521, + "learning_rate": 1.8804720213170917e-06, + "loss": 1.3827, + "step": 1947 + }, + { + "epoch": 2.0010272213662046, + "grad_norm": 2.900805711746216, + "learning_rate": 1.878568709554625e-06, + "loss": 1.4183, + "step": 1948 + }, + { + "epoch": 2.0020544427324087, + "grad_norm": 2.8362643718719482, + "learning_rate": 1.8766653977921585e-06, + "loss": 1.468, + "step": 1949 + }, + { + "epoch": 2.0030816640986133, + "grad_norm": 2.774122476577759, + "learning_rate": 1.874762086029692e-06, + "loss": 1.4352, + "step": 1950 + }, + { + "epoch": 2.0030816640986133, + "eval_loss": 1.8892052173614502, + "eval_runtime": 49.6699, + "eval_samples_per_second": 20.133, + "eval_steps_per_second": 0.846, + "step": 1950 + }, + { + "epoch": 2.0041088854648175, + "grad_norm": 3.0558278560638428, + "learning_rate": 1.8728587742672252e-06, + "loss": 1.4593, + "step": 1951 + }, + { + "epoch": 2.005136106831022, + "grad_norm": 3.044110059738159, + "learning_rate": 1.8709554625047585e-06, + "loss": 1.4679, + "step": 1952 + }, + { + "epoch": 2.0061633281972266, + "grad_norm": 3.015784740447998, + "learning_rate": 1.8690521507422917e-06, + "loss": 1.4053, + "step": 1953 + }, + { + "epoch": 2.007190549563431, + "grad_norm": 3.1507861614227295, + "learning_rate": 1.867148838979825e-06, + "loss": 1.4289, + "step": 1954 + }, + { + "epoch": 2.0082177709296354, + "grad_norm": 2.9931421279907227, + "learning_rate": 1.8652455272173584e-06, + "loss": 1.4559, + "step": 1955 + }, + { + "epoch": 2.00924499229584, + "grad_norm": 3.0338571071624756, + "learning_rate": 1.8633422154548916e-06, + "loss": 1.473, + "step": 1956 + }, + { + "epoch": 2.010272213662044, + "grad_norm": 3.1007537841796875, + "learning_rate": 1.861438903692425e-06, + "loss": 1.3715, + "step": 1957 + }, + { + "epoch": 2.0112994350282487, + "grad_norm": 3.1187386512756348, + "learning_rate": 1.859535591929958e-06, + "loss": 1.4007, + "step": 1958 + }, + { + "epoch": 2.012326656394453, + "grad_norm": 3.045088529586792, + "learning_rate": 1.8576322801674915e-06, + "loss": 1.4866, + "step": 1959 + }, + { + "epoch": 2.0133538777606574, + "grad_norm": 2.9711339473724365, + "learning_rate": 1.855728968405025e-06, + "loss": 1.4193, + "step": 1960 + }, + { + "epoch": 2.0133538777606574, + "eval_loss": 1.8801918029785156, + "eval_runtime": 51.4737, + "eval_samples_per_second": 19.427, + "eval_steps_per_second": 0.816, + "step": 1960 + }, + { + "epoch": 2.014381099126862, + "grad_norm": 3.008835554122925, + "learning_rate": 1.8538256566425584e-06, + "loss": 1.4409, + "step": 1961 + }, + { + "epoch": 2.015408320493066, + "grad_norm": 3.013974189758301, + "learning_rate": 1.8519223448800915e-06, + "loss": 1.4708, + "step": 1962 + }, + { + "epoch": 2.0164355418592708, + "grad_norm": 3.00911545753479, + "learning_rate": 1.850019033117625e-06, + "loss": 1.4169, + "step": 1963 + }, + { + "epoch": 2.017462763225475, + "grad_norm": 2.9539992809295654, + "learning_rate": 1.848115721355158e-06, + "loss": 1.4172, + "step": 1964 + }, + { + "epoch": 2.0184899845916795, + "grad_norm": 3.2330169677734375, + "learning_rate": 1.8462124095926914e-06, + "loss": 1.4372, + "step": 1965 + }, + { + "epoch": 2.019517205957884, + "grad_norm": 3.1095242500305176, + "learning_rate": 1.8443090978302248e-06, + "loss": 1.4376, + "step": 1966 + }, + { + "epoch": 2.0205444273240882, + "grad_norm": 3.032968044281006, + "learning_rate": 1.842405786067758e-06, + "loss": 1.4784, + "step": 1967 + }, + { + "epoch": 2.021571648690293, + "grad_norm": 3.0439071655273438, + "learning_rate": 1.8405024743052913e-06, + "loss": 1.418, + "step": 1968 + }, + { + "epoch": 2.022598870056497, + "grad_norm": 3.0753591060638428, + "learning_rate": 1.8385991625428247e-06, + "loss": 1.4687, + "step": 1969 + }, + { + "epoch": 2.0236260914227016, + "grad_norm": 3.0760700702667236, + "learning_rate": 1.8366958507803578e-06, + "loss": 1.397, + "step": 1970 + }, + { + "epoch": 2.0236260914227016, + "eval_loss": 1.885207176208496, + "eval_runtime": 51.5554, + "eval_samples_per_second": 19.397, + "eval_steps_per_second": 0.815, + "step": 1970 + }, + { + "epoch": 2.024653312788906, + "grad_norm": 3.0004823207855225, + "learning_rate": 1.8347925390178914e-06, + "loss": 1.4413, + "step": 1971 + }, + { + "epoch": 2.0256805341551103, + "grad_norm": 2.861964702606201, + "learning_rate": 1.8328892272554247e-06, + "loss": 1.3885, + "step": 1972 + }, + { + "epoch": 2.026707755521315, + "grad_norm": 3.034388542175293, + "learning_rate": 1.8309859154929579e-06, + "loss": 1.4579, + "step": 1973 + }, + { + "epoch": 2.0277349768875195, + "grad_norm": 3.1062700748443604, + "learning_rate": 1.8290826037304913e-06, + "loss": 1.4396, + "step": 1974 + }, + { + "epoch": 2.0287621982537236, + "grad_norm": 2.986480712890625, + "learning_rate": 1.8271792919680246e-06, + "loss": 1.4048, + "step": 1975 + }, + { + "epoch": 2.029789419619928, + "grad_norm": 3.0232272148132324, + "learning_rate": 1.8252759802055578e-06, + "loss": 1.4107, + "step": 1976 + }, + { + "epoch": 2.0308166409861323, + "grad_norm": 3.131545305252075, + "learning_rate": 1.8233726684430911e-06, + "loss": 1.4532, + "step": 1977 + }, + { + "epoch": 2.031843862352337, + "grad_norm": 3.1322755813598633, + "learning_rate": 1.8214693566806243e-06, + "loss": 1.3931, + "step": 1978 + }, + { + "epoch": 2.0328710837185415, + "grad_norm": 3.06524395942688, + "learning_rate": 1.8195660449181576e-06, + "loss": 1.4203, + "step": 1979 + }, + { + "epoch": 2.0338983050847457, + "grad_norm": 3.1509804725646973, + "learning_rate": 1.817662733155691e-06, + "loss": 1.4451, + "step": 1980 + }, + { + "epoch": 2.0338983050847457, + "eval_loss": 1.8945039510726929, + "eval_runtime": 47.4904, + "eval_samples_per_second": 21.057, + "eval_steps_per_second": 0.884, + "step": 1980 + }, + { + "epoch": 2.0349255264509503, + "grad_norm": 3.01600980758667, + "learning_rate": 1.8157594213932244e-06, + "loss": 1.4225, + "step": 1981 + }, + { + "epoch": 2.0359527478171544, + "grad_norm": 3.092637777328491, + "learning_rate": 1.8138561096307577e-06, + "loss": 1.4495, + "step": 1982 + }, + { + "epoch": 2.036979969183359, + "grad_norm": 2.8800899982452393, + "learning_rate": 1.811952797868291e-06, + "loss": 1.3792, + "step": 1983 + }, + { + "epoch": 2.0380071905495636, + "grad_norm": 2.91048002243042, + "learning_rate": 1.8100494861058242e-06, + "loss": 1.427, + "step": 1984 + }, + { + "epoch": 2.0390344119157677, + "grad_norm": 2.9576427936553955, + "learning_rate": 1.8081461743433576e-06, + "loss": 1.4328, + "step": 1985 + }, + { + "epoch": 2.0400616332819723, + "grad_norm": 2.950599193572998, + "learning_rate": 1.806242862580891e-06, + "loss": 1.3877, + "step": 1986 + }, + { + "epoch": 2.0410888546481765, + "grad_norm": 3.0447452068328857, + "learning_rate": 1.8043395508184241e-06, + "loss": 1.4442, + "step": 1987 + }, + { + "epoch": 2.042116076014381, + "grad_norm": 3.058427095413208, + "learning_rate": 1.8024362390559575e-06, + "loss": 1.438, + "step": 1988 + }, + { + "epoch": 2.0431432973805856, + "grad_norm": 3.0351500511169434, + "learning_rate": 1.8005329272934906e-06, + "loss": 1.4007, + "step": 1989 + }, + { + "epoch": 2.04417051874679, + "grad_norm": 3.2143735885620117, + "learning_rate": 1.798629615531024e-06, + "loss": 1.4968, + "step": 1990 + }, + { + "epoch": 2.04417051874679, + "eval_loss": 1.895965337753296, + "eval_runtime": 47.6538, + "eval_samples_per_second": 20.985, + "eval_steps_per_second": 0.881, + "step": 1990 + }, + { + "epoch": 2.0451977401129944, + "grad_norm": 3.207671642303467, + "learning_rate": 1.7967263037685576e-06, + "loss": 1.419, + "step": 1991 + }, + { + "epoch": 2.046224961479199, + "grad_norm": 3.161807060241699, + "learning_rate": 1.794822992006091e-06, + "loss": 1.4193, + "step": 1992 + }, + { + "epoch": 2.047252182845403, + "grad_norm": 3.1695258617401123, + "learning_rate": 1.792919680243624e-06, + "loss": 1.4708, + "step": 1993 + }, + { + "epoch": 2.0482794042116077, + "grad_norm": 3.241506814956665, + "learning_rate": 1.7910163684811574e-06, + "loss": 1.3984, + "step": 1994 + }, + { + "epoch": 2.049306625577812, + "grad_norm": 3.014082193374634, + "learning_rate": 1.7891130567186906e-06, + "loss": 1.4208, + "step": 1995 + }, + { + "epoch": 2.0503338469440164, + "grad_norm": 3.0239710807800293, + "learning_rate": 1.787209744956224e-06, + "loss": 1.4036, + "step": 1996 + }, + { + "epoch": 2.051361068310221, + "grad_norm": 3.0406906604766846, + "learning_rate": 1.7853064331937573e-06, + "loss": 1.4266, + "step": 1997 + }, + { + "epoch": 2.052388289676425, + "grad_norm": 3.000096559524536, + "learning_rate": 1.7834031214312905e-06, + "loss": 1.4619, + "step": 1998 + }, + { + "epoch": 2.0534155110426298, + "grad_norm": 3.042457103729248, + "learning_rate": 1.7814998096688238e-06, + "loss": 1.3968, + "step": 1999 + }, + { + "epoch": 2.054442732408834, + "grad_norm": 3.0115537643432617, + "learning_rate": 1.779596497906357e-06, + "loss": 1.4404, + "step": 2000 + }, + { + "epoch": 2.054442732408834, + "eval_loss": 1.8855949640274048, + "eval_runtime": 49.9663, + "eval_samples_per_second": 20.013, + "eval_steps_per_second": 0.841, + "step": 2000 + }, + { + "epoch": 2.0554699537750385, + "grad_norm": 3.127013921737671, + "learning_rate": 1.7776931861438906e-06, + "loss": 1.4841, + "step": 2001 + }, + { + "epoch": 2.056497175141243, + "grad_norm": 3.080503463745117, + "learning_rate": 1.775789874381424e-06, + "loss": 1.477, + "step": 2002 + }, + { + "epoch": 2.0575243965074472, + "grad_norm": 3.193058967590332, + "learning_rate": 1.7738865626189573e-06, + "loss": 1.4539, + "step": 2003 + }, + { + "epoch": 2.058551617873652, + "grad_norm": 3.063589334487915, + "learning_rate": 1.7719832508564904e-06, + "loss": 1.3891, + "step": 2004 + }, + { + "epoch": 2.0595788392398564, + "grad_norm": 3.1374058723449707, + "learning_rate": 1.7700799390940238e-06, + "loss": 1.4131, + "step": 2005 + }, + { + "epoch": 2.0606060606060606, + "grad_norm": 3.029297351837158, + "learning_rate": 1.768176627331557e-06, + "loss": 1.4519, + "step": 2006 + }, + { + "epoch": 2.061633281972265, + "grad_norm": 3.0728025436401367, + "learning_rate": 1.7662733155690903e-06, + "loss": 1.434, + "step": 2007 + }, + { + "epoch": 2.0626605033384693, + "grad_norm": 3.0600240230560303, + "learning_rate": 1.7643700038066237e-06, + "loss": 1.464, + "step": 2008 + }, + { + "epoch": 2.063687724704674, + "grad_norm": 2.991814136505127, + "learning_rate": 1.7624666920441568e-06, + "loss": 1.3938, + "step": 2009 + }, + { + "epoch": 2.0647149460708785, + "grad_norm": 2.9615304470062256, + "learning_rate": 1.7605633802816902e-06, + "loss": 1.3883, + "step": 2010 + }, + { + "epoch": 2.0647149460708785, + "eval_loss": 1.8773977756500244, + "eval_runtime": 43.5833, + "eval_samples_per_second": 22.945, + "eval_steps_per_second": 0.964, + "step": 2010 + }, + { + "epoch": 2.0657421674370826, + "grad_norm": 3.108255624771118, + "learning_rate": 1.7586600685192238e-06, + "loss": 1.4956, + "step": 2011 + }, + { + "epoch": 2.066769388803287, + "grad_norm": 3.156050205230713, + "learning_rate": 1.756756756756757e-06, + "loss": 1.4922, + "step": 2012 + }, + { + "epoch": 2.0677966101694913, + "grad_norm": 3.24027943611145, + "learning_rate": 1.7548534449942903e-06, + "loss": 1.4504, + "step": 2013 + }, + { + "epoch": 2.068823831535696, + "grad_norm": 3.1180083751678467, + "learning_rate": 1.7529501332318236e-06, + "loss": 1.4183, + "step": 2014 + }, + { + "epoch": 2.0698510529019005, + "grad_norm": 3.178832769393921, + "learning_rate": 1.7510468214693568e-06, + "loss": 1.4178, + "step": 2015 + }, + { + "epoch": 2.0708782742681047, + "grad_norm": 3.058708906173706, + "learning_rate": 1.7491435097068902e-06, + "loss": 1.404, + "step": 2016 + }, + { + "epoch": 2.0719054956343093, + "grad_norm": 3.176765203475952, + "learning_rate": 1.7472401979444233e-06, + "loss": 1.4439, + "step": 2017 + }, + { + "epoch": 2.0729327170005134, + "grad_norm": 3.0423498153686523, + "learning_rate": 1.7453368861819567e-06, + "loss": 1.3661, + "step": 2018 + }, + { + "epoch": 2.073959938366718, + "grad_norm": 3.056716203689575, + "learning_rate": 1.74343357441949e-06, + "loss": 1.4196, + "step": 2019 + }, + { + "epoch": 2.0749871597329226, + "grad_norm": 3.0644123554229736, + "learning_rate": 1.7415302626570232e-06, + "loss": 1.4825, + "step": 2020 + }, + { + "epoch": 2.0749871597329226, + "eval_loss": 1.89279043674469, + "eval_runtime": 46.9776, + "eval_samples_per_second": 21.287, + "eval_steps_per_second": 0.894, + "step": 2020 + }, + { + "epoch": 2.0760143810991267, + "grad_norm": 2.9771924018859863, + "learning_rate": 1.7396269508945568e-06, + "loss": 1.4163, + "step": 2021 + }, + { + "epoch": 2.0770416024653313, + "grad_norm": 2.9193663597106934, + "learning_rate": 1.7377236391320901e-06, + "loss": 1.386, + "step": 2022 + }, + { + "epoch": 2.078068823831536, + "grad_norm": 3.194058418273926, + "learning_rate": 1.7358203273696233e-06, + "loss": 1.4696, + "step": 2023 + }, + { + "epoch": 2.07909604519774, + "grad_norm": 3.1295535564422607, + "learning_rate": 1.7339170156071566e-06, + "loss": 1.4252, + "step": 2024 + }, + { + "epoch": 2.0801232665639446, + "grad_norm": 2.948204755783081, + "learning_rate": 1.73201370384469e-06, + "loss": 1.416, + "step": 2025 + }, + { + "epoch": 2.081150487930149, + "grad_norm": 3.064087390899658, + "learning_rate": 1.7301103920822231e-06, + "loss": 1.431, + "step": 2026 + }, + { + "epoch": 2.0821777092963534, + "grad_norm": 3.230952739715576, + "learning_rate": 1.7282070803197565e-06, + "loss": 1.4454, + "step": 2027 + }, + { + "epoch": 2.083204930662558, + "grad_norm": 3.1373074054718018, + "learning_rate": 1.7263037685572897e-06, + "loss": 1.4218, + "step": 2028 + }, + { + "epoch": 2.084232152028762, + "grad_norm": 3.2426247596740723, + "learning_rate": 1.724400456794823e-06, + "loss": 1.4481, + "step": 2029 + }, + { + "epoch": 2.0852593733949667, + "grad_norm": 2.8744704723358154, + "learning_rate": 1.7224971450323564e-06, + "loss": 1.406, + "step": 2030 + }, + { + "epoch": 2.0852593733949667, + "eval_loss": 1.8881611824035645, + "eval_runtime": 49.2991, + "eval_samples_per_second": 20.284, + "eval_steps_per_second": 0.852, + "step": 2030 + }, + { + "epoch": 2.086286594761171, + "grad_norm": 3.0375564098358154, + "learning_rate": 1.72059383326989e-06, + "loss": 1.4187, + "step": 2031 + }, + { + "epoch": 2.0873138161273754, + "grad_norm": 3.069279909133911, + "learning_rate": 1.7186905215074231e-06, + "loss": 1.4847, + "step": 2032 + }, + { + "epoch": 2.08834103749358, + "grad_norm": 3.0621039867401123, + "learning_rate": 1.7167872097449565e-06, + "loss": 1.4048, + "step": 2033 + }, + { + "epoch": 2.089368258859784, + "grad_norm": 3.137110948562622, + "learning_rate": 1.7148838979824896e-06, + "loss": 1.4272, + "step": 2034 + }, + { + "epoch": 2.0903954802259888, + "grad_norm": 3.1717305183410645, + "learning_rate": 1.712980586220023e-06, + "loss": 1.424, + "step": 2035 + }, + { + "epoch": 2.091422701592193, + "grad_norm": 3.0675606727600098, + "learning_rate": 1.7110772744575563e-06, + "loss": 1.4076, + "step": 2036 + }, + { + "epoch": 2.0924499229583975, + "grad_norm": 3.0585849285125732, + "learning_rate": 1.7091739626950895e-06, + "loss": 1.4205, + "step": 2037 + }, + { + "epoch": 2.093477144324602, + "grad_norm": 3.1450932025909424, + "learning_rate": 1.7072706509326229e-06, + "loss": 1.4557, + "step": 2038 + }, + { + "epoch": 2.0945043656908062, + "grad_norm": 3.0912528038024902, + "learning_rate": 1.7053673391701562e-06, + "loss": 1.4335, + "step": 2039 + }, + { + "epoch": 2.095531587057011, + "grad_norm": 2.9372971057891846, + "learning_rate": 1.7034640274076894e-06, + "loss": 1.3959, + "step": 2040 + }, + { + "epoch": 2.095531587057011, + "eval_loss": 1.8857965469360352, + "eval_runtime": 50.3859, + "eval_samples_per_second": 19.847, + "eval_steps_per_second": 0.834, + "step": 2040 + }, + { + "epoch": 2.0965588084232154, + "grad_norm": 3.127795457839966, + "learning_rate": 1.701560715645223e-06, + "loss": 1.4741, + "step": 2041 + }, + { + "epoch": 2.0975860297894195, + "grad_norm": 3.1425771713256836, + "learning_rate": 1.6996574038827563e-06, + "loss": 1.3868, + "step": 2042 + }, + { + "epoch": 2.098613251155624, + "grad_norm": 3.158665180206299, + "learning_rate": 1.6977540921202895e-06, + "loss": 1.434, + "step": 2043 + }, + { + "epoch": 2.0996404725218283, + "grad_norm": 3.060809850692749, + "learning_rate": 1.6958507803578228e-06, + "loss": 1.4093, + "step": 2044 + }, + { + "epoch": 2.100667693888033, + "grad_norm": 3.138719320297241, + "learning_rate": 1.693947468595356e-06, + "loss": 1.4494, + "step": 2045 + }, + { + "epoch": 2.1016949152542375, + "grad_norm": 3.235567092895508, + "learning_rate": 1.6920441568328893e-06, + "loss": 1.4497, + "step": 2046 + }, + { + "epoch": 2.1027221366204416, + "grad_norm": 3.1233010292053223, + "learning_rate": 1.6901408450704227e-06, + "loss": 1.3903, + "step": 2047 + }, + { + "epoch": 2.103749357986646, + "grad_norm": 3.095599412918091, + "learning_rate": 1.6882375333079559e-06, + "loss": 1.416, + "step": 2048 + }, + { + "epoch": 2.1047765793528503, + "grad_norm": 3.266810178756714, + "learning_rate": 1.6863342215454892e-06, + "loss": 1.4676, + "step": 2049 + }, + { + "epoch": 2.105803800719055, + "grad_norm": 3.2977075576782227, + "learning_rate": 1.6844309097830226e-06, + "loss": 1.4641, + "step": 2050 + }, + { + "epoch": 2.105803800719055, + "eval_loss": 1.8875017166137695, + "eval_runtime": 44.2952, + "eval_samples_per_second": 22.576, + "eval_steps_per_second": 0.948, + "step": 2050 + }, + { + "epoch": 2.1068310220852595, + "grad_norm": 3.06746244430542, + "learning_rate": 1.682527598020556e-06, + "loss": 1.4443, + "step": 2051 + }, + { + "epoch": 2.1078582434514637, + "grad_norm": 3.0481631755828857, + "learning_rate": 1.6806242862580893e-06, + "loss": 1.4728, + "step": 2052 + }, + { + "epoch": 2.1088854648176683, + "grad_norm": 3.090182065963745, + "learning_rate": 1.6787209744956227e-06, + "loss": 1.3925, + "step": 2053 + }, + { + "epoch": 2.109912686183873, + "grad_norm": 3.0113272666931152, + "learning_rate": 1.6768176627331558e-06, + "loss": 1.3924, + "step": 2054 + }, + { + "epoch": 2.110939907550077, + "grad_norm": 3.0159542560577393, + "learning_rate": 1.6749143509706892e-06, + "loss": 1.4297, + "step": 2055 + }, + { + "epoch": 2.1119671289162816, + "grad_norm": 3.0569703578948975, + "learning_rate": 1.6730110392082225e-06, + "loss": 1.425, + "step": 2056 + }, + { + "epoch": 2.1129943502824857, + "grad_norm": 3.1391549110412598, + "learning_rate": 1.6711077274457557e-06, + "loss": 1.4357, + "step": 2057 + }, + { + "epoch": 2.1140215716486903, + "grad_norm": 3.0112249851226807, + "learning_rate": 1.669204415683289e-06, + "loss": 1.3948, + "step": 2058 + }, + { + "epoch": 2.115048793014895, + "grad_norm": 3.143165111541748, + "learning_rate": 1.6673011039208222e-06, + "loss": 1.4502, + "step": 2059 + }, + { + "epoch": 2.116076014381099, + "grad_norm": 3.030191659927368, + "learning_rate": 1.6653977921583556e-06, + "loss": 1.43, + "step": 2060 + }, + { + "epoch": 2.116076014381099, + "eval_loss": 1.885430932044983, + "eval_runtime": 44.6532, + "eval_samples_per_second": 22.395, + "eval_steps_per_second": 0.941, + "step": 2060 + }, + { + "epoch": 2.1171032357473036, + "grad_norm": 3.1698837280273438, + "learning_rate": 1.663494480395889e-06, + "loss": 1.4461, + "step": 2061 + }, + { + "epoch": 2.118130457113508, + "grad_norm": 3.0927395820617676, + "learning_rate": 1.6615911686334225e-06, + "loss": 1.3957, + "step": 2062 + }, + { + "epoch": 2.1191576784797124, + "grad_norm": 3.049832582473755, + "learning_rate": 1.6596878568709557e-06, + "loss": 1.4269, + "step": 2063 + }, + { + "epoch": 2.120184899845917, + "grad_norm": 3.031977653503418, + "learning_rate": 1.657784545108489e-06, + "loss": 1.4141, + "step": 2064 + }, + { + "epoch": 2.121212121212121, + "grad_norm": 3.0946176052093506, + "learning_rate": 1.6558812333460222e-06, + "loss": 1.4677, + "step": 2065 + }, + { + "epoch": 2.1222393425783257, + "grad_norm": 3.1852216720581055, + "learning_rate": 1.6539779215835555e-06, + "loss": 1.4476, + "step": 2066 + }, + { + "epoch": 2.12326656394453, + "grad_norm": 3.1202027797698975, + "learning_rate": 1.652074609821089e-06, + "loss": 1.4107, + "step": 2067 + }, + { + "epoch": 2.1242937853107344, + "grad_norm": 2.9965174198150635, + "learning_rate": 1.650171298058622e-06, + "loss": 1.4118, + "step": 2068 + }, + { + "epoch": 2.125321006676939, + "grad_norm": 3.142627239227295, + "learning_rate": 1.6482679862961554e-06, + "loss": 1.4148, + "step": 2069 + }, + { + "epoch": 2.126348228043143, + "grad_norm": 3.114840030670166, + "learning_rate": 1.6463646745336886e-06, + "loss": 1.4104, + "step": 2070 + }, + { + "epoch": 2.126348228043143, + "eval_loss": 1.8847553730010986, + "eval_runtime": 45.4766, + "eval_samples_per_second": 21.989, + "eval_steps_per_second": 0.924, + "step": 2070 + }, + { + "epoch": 2.1273754494093478, + "grad_norm": 3.0997679233551025, + "learning_rate": 1.644461362771222e-06, + "loss": 1.3903, + "step": 2071 + }, + { + "epoch": 2.128402670775552, + "grad_norm": 3.053497791290283, + "learning_rate": 1.6425580510087555e-06, + "loss": 1.4362, + "step": 2072 + }, + { + "epoch": 2.1294298921417565, + "grad_norm": 3.1247191429138184, + "learning_rate": 1.6406547392462889e-06, + "loss": 1.395, + "step": 2073 + }, + { + "epoch": 2.130457113507961, + "grad_norm": 3.0684449672698975, + "learning_rate": 1.638751427483822e-06, + "loss": 1.4, + "step": 2074 + }, + { + "epoch": 2.131484334874165, + "grad_norm": 3.0759975910186768, + "learning_rate": 1.6368481157213554e-06, + "loss": 1.4301, + "step": 2075 + }, + { + "epoch": 2.13251155624037, + "grad_norm": 3.1344354152679443, + "learning_rate": 1.6349448039588885e-06, + "loss": 1.4165, + "step": 2076 + }, + { + "epoch": 2.1335387776065744, + "grad_norm": 3.1342315673828125, + "learning_rate": 1.6330414921964219e-06, + "loss": 1.4811, + "step": 2077 + }, + { + "epoch": 2.1345659989727785, + "grad_norm": 2.9999516010284424, + "learning_rate": 1.6311381804339553e-06, + "loss": 1.3974, + "step": 2078 + }, + { + "epoch": 2.135593220338983, + "grad_norm": 3.10968279838562, + "learning_rate": 1.6292348686714884e-06, + "loss": 1.4308, + "step": 2079 + }, + { + "epoch": 2.1366204417051873, + "grad_norm": 3.123739004135132, + "learning_rate": 1.6273315569090218e-06, + "loss": 1.4116, + "step": 2080 + }, + { + "epoch": 2.1366204417051873, + "eval_loss": 1.8859853744506836, + "eval_runtime": 43.5954, + "eval_samples_per_second": 22.938, + "eval_steps_per_second": 0.963, + "step": 2080 + }, + { + "epoch": 2.137647663071392, + "grad_norm": 3.1350209712982178, + "learning_rate": 1.625428245146555e-06, + "loss": 1.4914, + "step": 2081 + }, + { + "epoch": 2.1386748844375965, + "grad_norm": 3.1349897384643555, + "learning_rate": 1.6235249333840885e-06, + "loss": 1.4681, + "step": 2082 + }, + { + "epoch": 2.1397021058038006, + "grad_norm": 3.1517136096954346, + "learning_rate": 1.6216216216216219e-06, + "loss": 1.4287, + "step": 2083 + }, + { + "epoch": 2.140729327170005, + "grad_norm": 3.109863758087158, + "learning_rate": 1.6197183098591552e-06, + "loss": 1.4557, + "step": 2084 + }, + { + "epoch": 2.1417565485362093, + "grad_norm": 3.017601728439331, + "learning_rate": 1.6178149980966884e-06, + "loss": 1.435, + "step": 2085 + }, + { + "epoch": 2.142783769902414, + "grad_norm": 3.0578653812408447, + "learning_rate": 1.6159116863342217e-06, + "loss": 1.4078, + "step": 2086 + }, + { + "epoch": 2.1438109912686185, + "grad_norm": 3.1092753410339355, + "learning_rate": 1.6140083745717549e-06, + "loss": 1.4809, + "step": 2087 + }, + { + "epoch": 2.1448382126348227, + "grad_norm": 3.012550115585327, + "learning_rate": 1.6121050628092882e-06, + "loss": 1.3883, + "step": 2088 + }, + { + "epoch": 2.1458654340010272, + "grad_norm": 3.239058256149292, + "learning_rate": 1.6102017510468216e-06, + "loss": 1.4733, + "step": 2089 + }, + { + "epoch": 2.146892655367232, + "grad_norm": 3.0761263370513916, + "learning_rate": 1.6082984392843548e-06, + "loss": 1.377, + "step": 2090 + }, + { + "epoch": 2.146892655367232, + "eval_loss": 1.8865137100219727, + "eval_runtime": 43.4996, + "eval_samples_per_second": 22.989, + "eval_steps_per_second": 0.966, + "step": 2090 + }, + { + "epoch": 2.147919876733436, + "grad_norm": 3.125427007675171, + "learning_rate": 1.6063951275218881e-06, + "loss": 1.4381, + "step": 2091 + }, + { + "epoch": 2.1489470980996406, + "grad_norm": 3.1415202617645264, + "learning_rate": 1.6044918157594217e-06, + "loss": 1.4823, + "step": 2092 + }, + { + "epoch": 2.1499743194658447, + "grad_norm": 3.061314821243286, + "learning_rate": 1.6025885039969548e-06, + "loss": 1.4177, + "step": 2093 + }, + { + "epoch": 2.1510015408320493, + "grad_norm": 3.0278573036193848, + "learning_rate": 1.6006851922344882e-06, + "loss": 1.4283, + "step": 2094 + }, + { + "epoch": 2.152028762198254, + "grad_norm": 3.0583488941192627, + "learning_rate": 1.5987818804720216e-06, + "loss": 1.4453, + "step": 2095 + }, + { + "epoch": 2.153055983564458, + "grad_norm": 3.149545907974243, + "learning_rate": 1.5968785687095547e-06, + "loss": 1.4481, + "step": 2096 + }, + { + "epoch": 2.1540832049306626, + "grad_norm": 3.031721591949463, + "learning_rate": 1.594975256947088e-06, + "loss": 1.3923, + "step": 2097 + }, + { + "epoch": 2.1551104262968668, + "grad_norm": 3.0269832611083984, + "learning_rate": 1.5930719451846212e-06, + "loss": 1.4489, + "step": 2098 + }, + { + "epoch": 2.1561376476630714, + "grad_norm": 3.0105485916137695, + "learning_rate": 1.5911686334221546e-06, + "loss": 1.3967, + "step": 2099 + }, + { + "epoch": 2.157164869029276, + "grad_norm": 3.170759916305542, + "learning_rate": 1.589265321659688e-06, + "loss": 1.4467, + "step": 2100 + }, + { + "epoch": 2.157164869029276, + "eval_loss": 1.887832760810852, + "eval_runtime": 44.2197, + "eval_samples_per_second": 22.614, + "eval_steps_per_second": 0.95, + "step": 2100 + }, + { + "epoch": 2.15819209039548, + "grad_norm": 2.9956490993499756, + "learning_rate": 1.5873620098972211e-06, + "loss": 1.3601, + "step": 2101 + }, + { + "epoch": 2.1592193117616847, + "grad_norm": 3.118528366088867, + "learning_rate": 1.5854586981347547e-06, + "loss": 1.4053, + "step": 2102 + }, + { + "epoch": 2.1602465331278893, + "grad_norm": 3.3036599159240723, + "learning_rate": 1.583555386372288e-06, + "loss": 1.4523, + "step": 2103 + }, + { + "epoch": 2.1612737544940934, + "grad_norm": 3.2302451133728027, + "learning_rate": 1.5816520746098212e-06, + "loss": 1.4646, + "step": 2104 + }, + { + "epoch": 2.162300975860298, + "grad_norm": 3.1311371326446533, + "learning_rate": 1.5797487628473546e-06, + "loss": 1.4233, + "step": 2105 + }, + { + "epoch": 2.163328197226502, + "grad_norm": 3.024657726287842, + "learning_rate": 1.577845451084888e-06, + "loss": 1.3791, + "step": 2106 + }, + { + "epoch": 2.1643554185927067, + "grad_norm": 3.1913206577301025, + "learning_rate": 1.575942139322421e-06, + "loss": 1.4284, + "step": 2107 + }, + { + "epoch": 2.1653826399589113, + "grad_norm": 3.010667085647583, + "learning_rate": 1.5740388275599544e-06, + "loss": 1.352, + "step": 2108 + }, + { + "epoch": 2.1664098613251155, + "grad_norm": 3.0839853286743164, + "learning_rate": 1.5721355157974876e-06, + "loss": 1.4342, + "step": 2109 + }, + { + "epoch": 2.16743708269132, + "grad_norm": 3.115177631378174, + "learning_rate": 1.570232204035021e-06, + "loss": 1.4188, + "step": 2110 + }, + { + "epoch": 2.16743708269132, + "eval_loss": 1.8869224786758423, + "eval_runtime": 42.2937, + "eval_samples_per_second": 23.644, + "eval_steps_per_second": 0.993, + "step": 2110 + }, + { + "epoch": 2.168464304057524, + "grad_norm": 3.1874091625213623, + "learning_rate": 1.5683288922725543e-06, + "loss": 1.4252, + "step": 2111 + }, + { + "epoch": 2.169491525423729, + "grad_norm": 3.149386405944824, + "learning_rate": 1.5664255805100879e-06, + "loss": 1.4252, + "step": 2112 + }, + { + "epoch": 2.1705187467899334, + "grad_norm": 3.1509039402008057, + "learning_rate": 1.564522268747621e-06, + "loss": 1.4135, + "step": 2113 + }, + { + "epoch": 2.1715459681561375, + "grad_norm": 3.1496448516845703, + "learning_rate": 1.5626189569851544e-06, + "loss": 1.4384, + "step": 2114 + }, + { + "epoch": 2.172573189522342, + "grad_norm": 2.9863228797912598, + "learning_rate": 1.5607156452226875e-06, + "loss": 1.3569, + "step": 2115 + }, + { + "epoch": 2.1736004108885463, + "grad_norm": 3.1577210426330566, + "learning_rate": 1.558812333460221e-06, + "loss": 1.4263, + "step": 2116 + }, + { + "epoch": 2.174627632254751, + "grad_norm": 3.079028844833374, + "learning_rate": 1.5569090216977543e-06, + "loss": 1.4363, + "step": 2117 + }, + { + "epoch": 2.1756548536209555, + "grad_norm": 3.0446267127990723, + "learning_rate": 1.5550057099352874e-06, + "loss": 1.4287, + "step": 2118 + }, + { + "epoch": 2.1766820749871596, + "grad_norm": 3.0535521507263184, + "learning_rate": 1.5531023981728208e-06, + "loss": 1.4457, + "step": 2119 + }, + { + "epoch": 2.177709296353364, + "grad_norm": 2.997999429702759, + "learning_rate": 1.5511990864103542e-06, + "loss": 1.4013, + "step": 2120 + }, + { + "epoch": 2.177709296353364, + "eval_loss": 1.8781853914260864, + "eval_runtime": 46.565, + "eval_samples_per_second": 21.475, + "eval_steps_per_second": 0.902, + "step": 2120 + }, + { + "epoch": 2.1787365177195683, + "grad_norm": 3.021111249923706, + "learning_rate": 1.5492957746478873e-06, + "loss": 1.4337, + "step": 2121 + }, + { + "epoch": 2.179763739085773, + "grad_norm": 3.147423267364502, + "learning_rate": 1.5473924628854209e-06, + "loss": 1.4566, + "step": 2122 + }, + { + "epoch": 2.1807909604519775, + "grad_norm": 3.2015597820281982, + "learning_rate": 1.5454891511229542e-06, + "loss": 1.4183, + "step": 2123 + }, + { + "epoch": 2.1818181818181817, + "grad_norm": 3.027254819869995, + "learning_rate": 1.5435858393604874e-06, + "loss": 1.4211, + "step": 2124 + }, + { + "epoch": 2.1828454031843862, + "grad_norm": 3.140397787094116, + "learning_rate": 1.5416825275980208e-06, + "loss": 1.4545, + "step": 2125 + }, + { + "epoch": 2.183872624550591, + "grad_norm": 3.087702751159668, + "learning_rate": 1.539779215835554e-06, + "loss": 1.4038, + "step": 2126 + }, + { + "epoch": 2.184899845916795, + "grad_norm": 3.1732728481292725, + "learning_rate": 1.5378759040730873e-06, + "loss": 1.4156, + "step": 2127 + }, + { + "epoch": 2.1859270672829996, + "grad_norm": 3.181870460510254, + "learning_rate": 1.5359725923106206e-06, + "loss": 1.468, + "step": 2128 + }, + { + "epoch": 2.1869542886492037, + "grad_norm": 3.111034393310547, + "learning_rate": 1.5340692805481538e-06, + "loss": 1.4402, + "step": 2129 + }, + { + "epoch": 2.1879815100154083, + "grad_norm": 3.07132887840271, + "learning_rate": 1.5321659687856871e-06, + "loss": 1.4072, + "step": 2130 + }, + { + "epoch": 2.1879815100154083, + "eval_loss": 1.8848137855529785, + "eval_runtime": 48.9628, + "eval_samples_per_second": 20.424, + "eval_steps_per_second": 0.858, + "step": 2130 + }, + { + "epoch": 2.189008731381613, + "grad_norm": 3.078758716583252, + "learning_rate": 1.5302626570232205e-06, + "loss": 1.4022, + "step": 2131 + }, + { + "epoch": 2.190035952747817, + "grad_norm": 3.104029655456543, + "learning_rate": 1.5283593452607539e-06, + "loss": 1.3619, + "step": 2132 + }, + { + "epoch": 2.1910631741140216, + "grad_norm": 3.167560577392578, + "learning_rate": 1.5264560334982872e-06, + "loss": 1.4784, + "step": 2133 + }, + { + "epoch": 2.1920903954802258, + "grad_norm": 3.055535078048706, + "learning_rate": 1.5245527217358206e-06, + "loss": 1.4053, + "step": 2134 + }, + { + "epoch": 2.1931176168464304, + "grad_norm": 3.1337225437164307, + "learning_rate": 1.5226494099733537e-06, + "loss": 1.4459, + "step": 2135 + }, + { + "epoch": 2.194144838212635, + "grad_norm": 3.0870962142944336, + "learning_rate": 1.520746098210887e-06, + "loss": 1.3966, + "step": 2136 + }, + { + "epoch": 2.195172059578839, + "grad_norm": 3.0952706336975098, + "learning_rate": 1.5188427864484205e-06, + "loss": 1.4625, + "step": 2137 + }, + { + "epoch": 2.1961992809450437, + "grad_norm": 3.064378023147583, + "learning_rate": 1.5169394746859536e-06, + "loss": 1.3913, + "step": 2138 + }, + { + "epoch": 2.1972265023112483, + "grad_norm": 3.181436538696289, + "learning_rate": 1.515036162923487e-06, + "loss": 1.4114, + "step": 2139 + }, + { + "epoch": 2.1982537236774524, + "grad_norm": 3.0721826553344727, + "learning_rate": 1.5131328511610201e-06, + "loss": 1.4337, + "step": 2140 + }, + { + "epoch": 2.1982537236774524, + "eval_loss": 1.8966083526611328, + "eval_runtime": 46.5515, + "eval_samples_per_second": 21.482, + "eval_steps_per_second": 0.902, + "step": 2140 + }, + { + "epoch": 2.199280945043657, + "grad_norm": 3.3798177242279053, + "learning_rate": 1.5112295393985535e-06, + "loss": 1.465, + "step": 2141 + }, + { + "epoch": 2.200308166409861, + "grad_norm": 3.2605955600738525, + "learning_rate": 1.509326227636087e-06, + "loss": 1.4434, + "step": 2142 + }, + { + "epoch": 2.2013353877760657, + "grad_norm": 3.14622163772583, + "learning_rate": 1.5074229158736204e-06, + "loss": 1.3743, + "step": 2143 + }, + { + "epoch": 2.2023626091422703, + "grad_norm": 3.0676040649414062, + "learning_rate": 1.5055196041111536e-06, + "loss": 1.4221, + "step": 2144 + }, + { + "epoch": 2.2033898305084745, + "grad_norm": 3.2191338539123535, + "learning_rate": 1.503616292348687e-06, + "loss": 1.4318, + "step": 2145 + }, + { + "epoch": 2.204417051874679, + "grad_norm": 3.293288230895996, + "learning_rate": 1.50171298058622e-06, + "loss": 1.4646, + "step": 2146 + }, + { + "epoch": 2.205444273240883, + "grad_norm": 3.0085601806640625, + "learning_rate": 1.4998096688237535e-06, + "loss": 1.3608, + "step": 2147 + }, + { + "epoch": 2.206471494607088, + "grad_norm": 3.1057655811309814, + "learning_rate": 1.4979063570612868e-06, + "loss": 1.4342, + "step": 2148 + }, + { + "epoch": 2.2074987159732924, + "grad_norm": 3.1611974239349365, + "learning_rate": 1.49600304529882e-06, + "loss": 1.4303, + "step": 2149 + }, + { + "epoch": 2.2085259373394965, + "grad_norm": 3.0769052505493164, + "learning_rate": 1.4940997335363533e-06, + "loss": 1.4049, + "step": 2150 + }, + { + "epoch": 2.2085259373394965, + "eval_loss": 1.8821698427200317, + "eval_runtime": 47.8527, + "eval_samples_per_second": 20.897, + "eval_steps_per_second": 0.878, + "step": 2150 + }, + { + "epoch": 2.209553158705701, + "grad_norm": 3.3353095054626465, + "learning_rate": 1.4921964217738865e-06, + "loss": 1.4316, + "step": 2151 + }, + { + "epoch": 2.2105803800719057, + "grad_norm": 3.109008550643921, + "learning_rate": 1.49029311001142e-06, + "loss": 1.4088, + "step": 2152 + }, + { + "epoch": 2.21160760143811, + "grad_norm": 3.315708875656128, + "learning_rate": 1.4883897982489534e-06, + "loss": 1.4372, + "step": 2153 + }, + { + "epoch": 2.2126348228043144, + "grad_norm": 3.41266131401062, + "learning_rate": 1.4864864864864868e-06, + "loss": 1.4516, + "step": 2154 + }, + { + "epoch": 2.2136620441705186, + "grad_norm": 3.206641435623169, + "learning_rate": 1.48458317472402e-06, + "loss": 1.4198, + "step": 2155 + }, + { + "epoch": 2.214689265536723, + "grad_norm": 3.3165881633758545, + "learning_rate": 1.4826798629615533e-06, + "loss": 1.4229, + "step": 2156 + }, + { + "epoch": 2.2157164869029278, + "grad_norm": 3.113398313522339, + "learning_rate": 1.4807765511990864e-06, + "loss": 1.3785, + "step": 2157 + }, + { + "epoch": 2.216743708269132, + "grad_norm": 3.155613422393799, + "learning_rate": 1.4788732394366198e-06, + "loss": 1.3985, + "step": 2158 + }, + { + "epoch": 2.2177709296353365, + "grad_norm": 3.1063179969787598, + "learning_rate": 1.4769699276741532e-06, + "loss": 1.4282, + "step": 2159 + }, + { + "epoch": 2.2187981510015407, + "grad_norm": 3.1550281047821045, + "learning_rate": 1.4750666159116863e-06, + "loss": 1.4086, + "step": 2160 + }, + { + "epoch": 2.2187981510015407, + "eval_loss": 1.8825618028640747, + "eval_runtime": 45.5113, + "eval_samples_per_second": 21.973, + "eval_steps_per_second": 0.923, + "step": 2160 + }, + { + "epoch": 2.2198253723677452, + "grad_norm": 3.0101521015167236, + "learning_rate": 1.4731633041492197e-06, + "loss": 1.3705, + "step": 2161 + }, + { + "epoch": 2.22085259373395, + "grad_norm": 3.0926272869110107, + "learning_rate": 1.4712599923867528e-06, + "loss": 1.443, + "step": 2162 + }, + { + "epoch": 2.221879815100154, + "grad_norm": 3.1270558834075928, + "learning_rate": 1.4693566806242864e-06, + "loss": 1.4641, + "step": 2163 + }, + { + "epoch": 2.2229070364663586, + "grad_norm": 2.9442856311798096, + "learning_rate": 1.4674533688618198e-06, + "loss": 1.3873, + "step": 2164 + }, + { + "epoch": 2.2239342578325627, + "grad_norm": 3.0549123287200928, + "learning_rate": 1.4655500570993531e-06, + "loss": 1.4236, + "step": 2165 + }, + { + "epoch": 2.2249614791987673, + "grad_norm": 3.1002092361450195, + "learning_rate": 1.4636467453368863e-06, + "loss": 1.4174, + "step": 2166 + }, + { + "epoch": 2.225988700564972, + "grad_norm": 3.118391990661621, + "learning_rate": 1.4617434335744197e-06, + "loss": 1.4783, + "step": 2167 + }, + { + "epoch": 2.227015921931176, + "grad_norm": 3.1835744380950928, + "learning_rate": 1.4598401218119528e-06, + "loss": 1.4542, + "step": 2168 + }, + { + "epoch": 2.2280431432973806, + "grad_norm": 3.1558971405029297, + "learning_rate": 1.4579368100494862e-06, + "loss": 1.4443, + "step": 2169 + }, + { + "epoch": 2.2290703646635848, + "grad_norm": 3.0535943508148193, + "learning_rate": 1.4560334982870195e-06, + "loss": 1.4209, + "step": 2170 + }, + { + "epoch": 2.2290703646635848, + "eval_loss": 1.882952094078064, + "eval_runtime": 47.6589, + "eval_samples_per_second": 20.982, + "eval_steps_per_second": 0.881, + "step": 2170 + }, + { + "epoch": 2.2300975860297894, + "grad_norm": 3.110412836074829, + "learning_rate": 1.4541301865245527e-06, + "loss": 1.432, + "step": 2171 + }, + { + "epoch": 2.231124807395994, + "grad_norm": 3.0594677925109863, + "learning_rate": 1.452226874762086e-06, + "loss": 1.4466, + "step": 2172 + }, + { + "epoch": 2.232152028762198, + "grad_norm": 3.1013123989105225, + "learning_rate": 1.4503235629996196e-06, + "loss": 1.4038, + "step": 2173 + }, + { + "epoch": 2.2331792501284027, + "grad_norm": 3.327085018157959, + "learning_rate": 1.4484202512371528e-06, + "loss": 1.4712, + "step": 2174 + }, + { + "epoch": 2.2342064714946073, + "grad_norm": 3.1586358547210693, + "learning_rate": 1.4465169394746861e-06, + "loss": 1.3606, + "step": 2175 + }, + { + "epoch": 2.2352336928608114, + "grad_norm": 3.200604200363159, + "learning_rate": 1.4446136277122195e-06, + "loss": 1.4792, + "step": 2176 + }, + { + "epoch": 2.236260914227016, + "grad_norm": 3.0358104705810547, + "learning_rate": 1.4427103159497526e-06, + "loss": 1.3812, + "step": 2177 + }, + { + "epoch": 2.23728813559322, + "grad_norm": 3.103911876678467, + "learning_rate": 1.440807004187286e-06, + "loss": 1.4049, + "step": 2178 + }, + { + "epoch": 2.2383153569594247, + "grad_norm": 3.0799686908721924, + "learning_rate": 1.4389036924248192e-06, + "loss": 1.437, + "step": 2179 + }, + { + "epoch": 2.2393425783256293, + "grad_norm": 3.0964274406433105, + "learning_rate": 1.4370003806623525e-06, + "loss": 1.4501, + "step": 2180 + }, + { + "epoch": 2.2393425783256293, + "eval_loss": 1.8832465410232544, + "eval_runtime": 50.3011, + "eval_samples_per_second": 19.88, + "eval_steps_per_second": 0.835, + "step": 2180 + }, + { + "epoch": 2.2403697996918335, + "grad_norm": 3.1642727851867676, + "learning_rate": 1.4350970688998859e-06, + "loss": 1.4198, + "step": 2181 + }, + { + "epoch": 2.241397021058038, + "grad_norm": 3.0271341800689697, + "learning_rate": 1.433193757137419e-06, + "loss": 1.4648, + "step": 2182 + }, + { + "epoch": 2.242424242424242, + "grad_norm": 2.9605119228363037, + "learning_rate": 1.4312904453749526e-06, + "loss": 1.4079, + "step": 2183 + }, + { + "epoch": 2.243451463790447, + "grad_norm": 3.034020185470581, + "learning_rate": 1.429387133612486e-06, + "loss": 1.3797, + "step": 2184 + }, + { + "epoch": 2.2444786851566514, + "grad_norm": 3.0324604511260986, + "learning_rate": 1.4274838218500191e-06, + "loss": 1.4268, + "step": 2185 + }, + { + "epoch": 2.2455059065228555, + "grad_norm": 3.260662317276001, + "learning_rate": 1.4255805100875525e-06, + "loss": 1.4302, + "step": 2186 + }, + { + "epoch": 2.24653312788906, + "grad_norm": 3.2712326049804688, + "learning_rate": 1.4236771983250858e-06, + "loss": 1.4863, + "step": 2187 + }, + { + "epoch": 2.2475603492552647, + "grad_norm": 3.1835451126098633, + "learning_rate": 1.421773886562619e-06, + "loss": 1.4176, + "step": 2188 + }, + { + "epoch": 2.248587570621469, + "grad_norm": 3.0387771129608154, + "learning_rate": 1.4198705748001524e-06, + "loss": 1.3861, + "step": 2189 + }, + { + "epoch": 2.2496147919876734, + "grad_norm": 3.074028968811035, + "learning_rate": 1.4179672630376855e-06, + "loss": 1.4074, + "step": 2190 + }, + { + "epoch": 2.2496147919876734, + "eval_loss": 1.8878331184387207, + "eval_runtime": 46.7284, + "eval_samples_per_second": 21.4, + "eval_steps_per_second": 0.899, + "step": 2190 + }, + { + "epoch": 2.2506420133538776, + "grad_norm": 3.131228446960449, + "learning_rate": 1.4160639512752189e-06, + "loss": 1.3913, + "step": 2191 + }, + { + "epoch": 2.251669234720082, + "grad_norm": 3.049567699432373, + "learning_rate": 1.4141606395127522e-06, + "loss": 1.4114, + "step": 2192 + }, + { + "epoch": 2.2526964560862868, + "grad_norm": 3.0833427906036377, + "learning_rate": 1.4122573277502858e-06, + "loss": 1.3855, + "step": 2193 + }, + { + "epoch": 2.253723677452491, + "grad_norm": 3.0678162574768066, + "learning_rate": 1.410354015987819e-06, + "loss": 1.4452, + "step": 2194 + }, + { + "epoch": 2.2547508988186955, + "grad_norm": 3.1660776138305664, + "learning_rate": 1.4084507042253523e-06, + "loss": 1.4169, + "step": 2195 + }, + { + "epoch": 2.2557781201848996, + "grad_norm": 2.9469871520996094, + "learning_rate": 1.4065473924628855e-06, + "loss": 1.443, + "step": 2196 + }, + { + "epoch": 2.2568053415511042, + "grad_norm": 2.9388298988342285, + "learning_rate": 1.4046440807004188e-06, + "loss": 1.411, + "step": 2197 + }, + { + "epoch": 2.257832562917309, + "grad_norm": 3.080441951751709, + "learning_rate": 1.4027407689379522e-06, + "loss": 1.4163, + "step": 2198 + }, + { + "epoch": 2.258859784283513, + "grad_norm": 3.086993932723999, + "learning_rate": 1.4008374571754854e-06, + "loss": 1.3996, + "step": 2199 + }, + { + "epoch": 2.2598870056497176, + "grad_norm": 3.1033880710601807, + "learning_rate": 1.3989341454130187e-06, + "loss": 1.4271, + "step": 2200 + }, + { + "epoch": 2.2598870056497176, + "eval_loss": 1.8867006301879883, + "eval_runtime": 47.3101, + "eval_samples_per_second": 21.137, + "eval_steps_per_second": 0.888, + "step": 2200 + }, + { + "epoch": 2.260914227015922, + "grad_norm": 3.137422561645508, + "learning_rate": 1.397030833650552e-06, + "loss": 1.4244, + "step": 2201 + }, + { + "epoch": 2.2619414483821263, + "grad_norm": 3.1550681591033936, + "learning_rate": 1.3951275218880852e-06, + "loss": 1.4401, + "step": 2202 + }, + { + "epoch": 2.262968669748331, + "grad_norm": 3.341092824935913, + "learning_rate": 1.3932242101256188e-06, + "loss": 1.505, + "step": 2203 + }, + { + "epoch": 2.263995891114535, + "grad_norm": 3.0959739685058594, + "learning_rate": 1.3913208983631522e-06, + "loss": 1.3924, + "step": 2204 + }, + { + "epoch": 2.2650231124807396, + "grad_norm": 3.121640920639038, + "learning_rate": 1.3894175866006853e-06, + "loss": 1.4263, + "step": 2205 + }, + { + "epoch": 2.266050333846944, + "grad_norm": 3.0325067043304443, + "learning_rate": 1.3875142748382187e-06, + "loss": 1.3724, + "step": 2206 + }, + { + "epoch": 2.2670775552131484, + "grad_norm": 3.130715847015381, + "learning_rate": 1.385610963075752e-06, + "loss": 1.3632, + "step": 2207 + }, + { + "epoch": 2.268104776579353, + "grad_norm": 3.10500431060791, + "learning_rate": 1.3837076513132852e-06, + "loss": 1.4227, + "step": 2208 + }, + { + "epoch": 2.269131997945557, + "grad_norm": 3.247288465499878, + "learning_rate": 1.3818043395508186e-06, + "loss": 1.4617, + "step": 2209 + }, + { + "epoch": 2.2701592193117617, + "grad_norm": 3.2476606369018555, + "learning_rate": 1.3799010277883517e-06, + "loss": 1.4325, + "step": 2210 + }, + { + "epoch": 2.2701592193117617, + "eval_loss": 1.8897888660430908, + "eval_runtime": 42.9294, + "eval_samples_per_second": 23.294, + "eval_steps_per_second": 0.978, + "step": 2210 + }, + { + "epoch": 2.2711864406779663, + "grad_norm": 3.0068607330322266, + "learning_rate": 1.377997716025885e-06, + "loss": 1.4119, + "step": 2211 + }, + { + "epoch": 2.2722136620441704, + "grad_norm": 3.185081958770752, + "learning_rate": 1.3760944042634184e-06, + "loss": 1.4334, + "step": 2212 + }, + { + "epoch": 2.273240883410375, + "grad_norm": 3.091458559036255, + "learning_rate": 1.3741910925009518e-06, + "loss": 1.4496, + "step": 2213 + }, + { + "epoch": 2.2742681047765796, + "grad_norm": 3.07018780708313, + "learning_rate": 1.3722877807384852e-06, + "loss": 1.4325, + "step": 2214 + }, + { + "epoch": 2.2752953261427837, + "grad_norm": 3.116154909133911, + "learning_rate": 1.3703844689760185e-06, + "loss": 1.4392, + "step": 2215 + }, + { + "epoch": 2.2763225475089883, + "grad_norm": 3.1154232025146484, + "learning_rate": 1.3684811572135517e-06, + "loss": 1.4062, + "step": 2216 + }, + { + "epoch": 2.2773497688751925, + "grad_norm": 3.0303139686584473, + "learning_rate": 1.366577845451085e-06, + "loss": 1.3569, + "step": 2217 + }, + { + "epoch": 2.278376990241397, + "grad_norm": 3.2322487831115723, + "learning_rate": 1.3646745336886184e-06, + "loss": 1.4498, + "step": 2218 + }, + { + "epoch": 2.279404211607601, + "grad_norm": 3.241386651992798, + "learning_rate": 1.3627712219261515e-06, + "loss": 1.4373, + "step": 2219 + }, + { + "epoch": 2.280431432973806, + "grad_norm": 3.052802801132202, + "learning_rate": 1.360867910163685e-06, + "loss": 1.3918, + "step": 2220 + }, + { + "epoch": 2.280431432973806, + "eval_loss": 1.882577657699585, + "eval_runtime": 48.945, + "eval_samples_per_second": 20.431, + "eval_steps_per_second": 0.858, + "step": 2220 + }, + { + "epoch": 2.2814586543400104, + "grad_norm": 3.256002426147461, + "learning_rate": 1.358964598401218e-06, + "loss": 1.4227, + "step": 2221 + }, + { + "epoch": 2.2824858757062145, + "grad_norm": 3.185438394546509, + "learning_rate": 1.3570612866387514e-06, + "loss": 1.423, + "step": 2222 + }, + { + "epoch": 2.283513097072419, + "grad_norm": 3.1974103450775146, + "learning_rate": 1.355157974876285e-06, + "loss": 1.4046, + "step": 2223 + }, + { + "epoch": 2.2845403184386237, + "grad_norm": 3.2578117847442627, + "learning_rate": 1.3532546631138184e-06, + "loss": 1.4692, + "step": 2224 + }, + { + "epoch": 2.285567539804828, + "grad_norm": 3.166027784347534, + "learning_rate": 1.3513513513513515e-06, + "loss": 1.4298, + "step": 2225 + }, + { + "epoch": 2.2865947611710324, + "grad_norm": 3.2637693881988525, + "learning_rate": 1.3494480395888849e-06, + "loss": 1.4132, + "step": 2226 + }, + { + "epoch": 2.2876219825372366, + "grad_norm": 3.124319314956665, + "learning_rate": 1.347544727826418e-06, + "loss": 1.4097, + "step": 2227 + }, + { + "epoch": 2.288649203903441, + "grad_norm": 3.164501428604126, + "learning_rate": 1.3456414160639514e-06, + "loss": 1.3947, + "step": 2228 + }, + { + "epoch": 2.2896764252696458, + "grad_norm": 3.2155773639678955, + "learning_rate": 1.3437381043014847e-06, + "loss": 1.4347, + "step": 2229 + }, + { + "epoch": 2.29070364663585, + "grad_norm": 3.2893707752227783, + "learning_rate": 1.341834792539018e-06, + "loss": 1.4258, + "step": 2230 + }, + { + "epoch": 2.29070364663585, + "eval_loss": 1.879811406135559, + "eval_runtime": 47.4777, + "eval_samples_per_second": 21.063, + "eval_steps_per_second": 0.885, + "step": 2230 + }, + { + "epoch": 2.2917308680020545, + "grad_norm": 3.095522403717041, + "learning_rate": 1.3399314807765513e-06, + "loss": 1.4227, + "step": 2231 + }, + { + "epoch": 2.2927580893682586, + "grad_norm": 3.1545920372009277, + "learning_rate": 1.3380281690140844e-06, + "loss": 1.4344, + "step": 2232 + }, + { + "epoch": 2.2937853107344632, + "grad_norm": 3.2042078971862793, + "learning_rate": 1.336124857251618e-06, + "loss": 1.4322, + "step": 2233 + }, + { + "epoch": 2.294812532100668, + "grad_norm": 3.162391424179077, + "learning_rate": 1.3342215454891513e-06, + "loss": 1.3924, + "step": 2234 + }, + { + "epoch": 2.295839753466872, + "grad_norm": 3.0768051147460938, + "learning_rate": 1.3323182337266847e-06, + "loss": 1.4, + "step": 2235 + }, + { + "epoch": 2.2968669748330766, + "grad_norm": 3.0797791481018066, + "learning_rate": 1.3304149219642179e-06, + "loss": 1.4268, + "step": 2236 + }, + { + "epoch": 2.297894196199281, + "grad_norm": 3.1587228775024414, + "learning_rate": 1.3285116102017512e-06, + "loss": 1.4665, + "step": 2237 + }, + { + "epoch": 2.2989214175654853, + "grad_norm": 3.0737040042877197, + "learning_rate": 1.3266082984392844e-06, + "loss": 1.4149, + "step": 2238 + }, + { + "epoch": 2.29994863893169, + "grad_norm": 3.1760494709014893, + "learning_rate": 1.3247049866768177e-06, + "loss": 1.4254, + "step": 2239 + }, + { + "epoch": 2.300975860297894, + "grad_norm": 3.1770107746124268, + "learning_rate": 1.322801674914351e-06, + "loss": 1.3979, + "step": 2240 + }, + { + "epoch": 2.300975860297894, + "eval_loss": 1.8801367282867432, + "eval_runtime": 50.137, + "eval_samples_per_second": 19.945, + "eval_steps_per_second": 0.838, + "step": 2240 + }, + { + "epoch": 2.3020030816640986, + "grad_norm": 3.215686559677124, + "learning_rate": 1.3208983631518843e-06, + "loss": 1.3948, + "step": 2241 + }, + { + "epoch": 2.303030303030303, + "grad_norm": 3.1983041763305664, + "learning_rate": 1.3189950513894176e-06, + "loss": 1.4276, + "step": 2242 + }, + { + "epoch": 2.3040575243965074, + "grad_norm": 3.2098212242126465, + "learning_rate": 1.3170917396269512e-06, + "loss": 1.4144, + "step": 2243 + }, + { + "epoch": 2.305084745762712, + "grad_norm": 3.170607089996338, + "learning_rate": 1.3151884278644843e-06, + "loss": 1.429, + "step": 2244 + }, + { + "epoch": 2.306111967128916, + "grad_norm": 3.1504342555999756, + "learning_rate": 1.3132851161020177e-06, + "loss": 1.4446, + "step": 2245 + }, + { + "epoch": 2.3071391884951207, + "grad_norm": 3.012388229370117, + "learning_rate": 1.311381804339551e-06, + "loss": 1.3926, + "step": 2246 + }, + { + "epoch": 2.3081664098613253, + "grad_norm": 3.1543102264404297, + "learning_rate": 1.3094784925770842e-06, + "loss": 1.4796, + "step": 2247 + }, + { + "epoch": 2.3091936312275294, + "grad_norm": 2.9769937992095947, + "learning_rate": 1.3075751808146176e-06, + "loss": 1.3916, + "step": 2248 + }, + { + "epoch": 2.310220852593734, + "grad_norm": 3.0640859603881836, + "learning_rate": 1.3056718690521507e-06, + "loss": 1.4255, + "step": 2249 + }, + { + "epoch": 2.3112480739599386, + "grad_norm": 2.9977078437805176, + "learning_rate": 1.303768557289684e-06, + "loss": 1.3964, + "step": 2250 + }, + { + "epoch": 2.3112480739599386, + "eval_loss": 1.8834030628204346, + "eval_runtime": 47.1318, + "eval_samples_per_second": 21.217, + "eval_steps_per_second": 0.891, + "step": 2250 + }, + { + "epoch": 2.3122752953261427, + "grad_norm": 3.0376105308532715, + "learning_rate": 1.3018652455272175e-06, + "loss": 1.4033, + "step": 2251 + }, + { + "epoch": 2.3133025166923473, + "grad_norm": 3.154709815979004, + "learning_rate": 1.2999619337647506e-06, + "loss": 1.4132, + "step": 2252 + }, + { + "epoch": 2.3143297380585515, + "grad_norm": 3.1585581302642822, + "learning_rate": 1.2980586220022842e-06, + "loss": 1.4045, + "step": 2253 + }, + { + "epoch": 2.315356959424756, + "grad_norm": 3.2512333393096924, + "learning_rate": 1.2961553102398175e-06, + "loss": 1.4431, + "step": 2254 + }, + { + "epoch": 2.3163841807909606, + "grad_norm": 3.093665599822998, + "learning_rate": 1.2942519984773507e-06, + "loss": 1.4519, + "step": 2255 + }, + { + "epoch": 2.317411402157165, + "grad_norm": 3.1790826320648193, + "learning_rate": 1.292348686714884e-06, + "loss": 1.4384, + "step": 2256 + }, + { + "epoch": 2.3184386235233694, + "grad_norm": 3.0824191570281982, + "learning_rate": 1.2904453749524174e-06, + "loss": 1.4011, + "step": 2257 + }, + { + "epoch": 2.3194658448895735, + "grad_norm": 3.218667507171631, + "learning_rate": 1.2885420631899506e-06, + "loss": 1.4446, + "step": 2258 + }, + { + "epoch": 2.320493066255778, + "grad_norm": 3.0636513233184814, + "learning_rate": 1.286638751427484e-06, + "loss": 1.3601, + "step": 2259 + }, + { + "epoch": 2.3215202876219827, + "grad_norm": 3.137908935546875, + "learning_rate": 1.284735439665017e-06, + "loss": 1.4033, + "step": 2260 + }, + { + "epoch": 2.3215202876219827, + "eval_loss": 1.8799747228622437, + "eval_runtime": 49.6332, + "eval_samples_per_second": 20.148, + "eval_steps_per_second": 0.846, + "step": 2260 + }, + { + "epoch": 2.322547508988187, + "grad_norm": 3.1707377433776855, + "learning_rate": 1.2828321279025504e-06, + "loss": 1.4306, + "step": 2261 + }, + { + "epoch": 2.3235747303543914, + "grad_norm": 3.2448198795318604, + "learning_rate": 1.2809288161400838e-06, + "loss": 1.4392, + "step": 2262 + }, + { + "epoch": 2.324601951720596, + "grad_norm": 3.115103006362915, + "learning_rate": 1.279025504377617e-06, + "loss": 1.4179, + "step": 2263 + }, + { + "epoch": 2.3256291730868, + "grad_norm": 3.1800553798675537, + "learning_rate": 1.2771221926151505e-06, + "loss": 1.3979, + "step": 2264 + }, + { + "epoch": 2.3266563944530048, + "grad_norm": 3.2946650981903076, + "learning_rate": 1.275218880852684e-06, + "loss": 1.4626, + "step": 2265 + }, + { + "epoch": 2.327683615819209, + "grad_norm": 3.1193125247955322, + "learning_rate": 1.273315569090217e-06, + "loss": 1.406, + "step": 2266 + }, + { + "epoch": 2.3287108371854135, + "grad_norm": 3.2309799194335938, + "learning_rate": 1.2714122573277504e-06, + "loss": 1.4179, + "step": 2267 + }, + { + "epoch": 2.3297380585516176, + "grad_norm": 3.1344757080078125, + "learning_rate": 1.2695089455652838e-06, + "loss": 1.4717, + "step": 2268 + }, + { + "epoch": 2.3307652799178222, + "grad_norm": 3.0551161766052246, + "learning_rate": 1.267605633802817e-06, + "loss": 1.4706, + "step": 2269 + }, + { + "epoch": 2.331792501284027, + "grad_norm": 3.1680619716644287, + "learning_rate": 1.2657023220403503e-06, + "loss": 1.4447, + "step": 2270 + }, + { + "epoch": 2.331792501284027, + "eval_loss": 1.872067928314209, + "eval_runtime": 44.2239, + "eval_samples_per_second": 22.612, + "eval_steps_per_second": 0.95, + "step": 2270 + }, + { + "epoch": 2.332819722650231, + "grad_norm": 3.0938210487365723, + "learning_rate": 1.2637990102778836e-06, + "loss": 1.4318, + "step": 2271 + }, + { + "epoch": 2.3338469440164356, + "grad_norm": 3.1710591316223145, + "learning_rate": 1.2618956985154168e-06, + "loss": 1.4182, + "step": 2272 + }, + { + "epoch": 2.33487416538264, + "grad_norm": 3.192415714263916, + "learning_rate": 1.2599923867529502e-06, + "loss": 1.423, + "step": 2273 + }, + { + "epoch": 2.3359013867488443, + "grad_norm": 3.219047784805298, + "learning_rate": 1.2580890749904837e-06, + "loss": 1.4457, + "step": 2274 + }, + { + "epoch": 2.336928608115049, + "grad_norm": 3.0913424491882324, + "learning_rate": 1.2561857632280169e-06, + "loss": 1.4174, + "step": 2275 + }, + { + "epoch": 2.337955829481253, + "grad_norm": 3.2987632751464844, + "learning_rate": 1.2542824514655502e-06, + "loss": 1.4551, + "step": 2276 + }, + { + "epoch": 2.3389830508474576, + "grad_norm": 3.5121114253997803, + "learning_rate": 1.2523791397030834e-06, + "loss": 1.4793, + "step": 2277 + }, + { + "epoch": 2.340010272213662, + "grad_norm": 3.3527626991271973, + "learning_rate": 1.2504758279406168e-06, + "loss": 1.4445, + "step": 2278 + }, + { + "epoch": 2.3410374935798663, + "grad_norm": 3.1356029510498047, + "learning_rate": 1.2485725161781501e-06, + "loss": 1.4062, + "step": 2279 + }, + { + "epoch": 2.342064714946071, + "grad_norm": 3.326028823852539, + "learning_rate": 1.2466692044156833e-06, + "loss": 1.4499, + "step": 2280 + }, + { + "epoch": 2.342064714946071, + "eval_loss": 1.8791024684906006, + "eval_runtime": 44.2336, + "eval_samples_per_second": 22.607, + "eval_steps_per_second": 0.95, + "step": 2280 + }, + { + "epoch": 2.343091936312275, + "grad_norm": 3.4187798500061035, + "learning_rate": 1.2447658926532168e-06, + "loss": 1.5179, + "step": 2281 + }, + { + "epoch": 2.3441191576784797, + "grad_norm": 3.1069252490997314, + "learning_rate": 1.24286258089075e-06, + "loss": 1.3825, + "step": 2282 + }, + { + "epoch": 2.3451463790446843, + "grad_norm": 2.9901397228240967, + "learning_rate": 1.2409592691282834e-06, + "loss": 1.4218, + "step": 2283 + }, + { + "epoch": 2.3461736004108884, + "grad_norm": 3.057490587234497, + "learning_rate": 1.2390559573658165e-06, + "loss": 1.4043, + "step": 2284 + }, + { + "epoch": 2.347200821777093, + "grad_norm": 3.265415906906128, + "learning_rate": 1.2371526456033499e-06, + "loss": 1.4315, + "step": 2285 + }, + { + "epoch": 2.3482280431432976, + "grad_norm": 3.2394862174987793, + "learning_rate": 1.2352493338408832e-06, + "loss": 1.4267, + "step": 2286 + }, + { + "epoch": 2.3492552645095017, + "grad_norm": 3.0940380096435547, + "learning_rate": 1.2333460220784166e-06, + "loss": 1.3947, + "step": 2287 + }, + { + "epoch": 2.3502824858757063, + "grad_norm": 3.230414628982544, + "learning_rate": 1.23144271031595e-06, + "loss": 1.5144, + "step": 2288 + }, + { + "epoch": 2.3513097072419105, + "grad_norm": 3.3229947090148926, + "learning_rate": 1.2295393985534831e-06, + "loss": 1.4413, + "step": 2289 + }, + { + "epoch": 2.352336928608115, + "grad_norm": 3.337873935699463, + "learning_rate": 1.2276360867910165e-06, + "loss": 1.461, + "step": 2290 + }, + { + "epoch": 2.352336928608115, + "eval_loss": 1.881198525428772, + "eval_runtime": 45.0014, + "eval_samples_per_second": 22.222, + "eval_steps_per_second": 0.933, + "step": 2290 + }, + { + "epoch": 2.3533641499743196, + "grad_norm": 3.064849615097046, + "learning_rate": 1.2257327750285498e-06, + "loss": 1.3404, + "step": 2291 + }, + { + "epoch": 2.354391371340524, + "grad_norm": 3.2958126068115234, + "learning_rate": 1.2238294632660832e-06, + "loss": 1.472, + "step": 2292 + }, + { + "epoch": 2.3554185927067284, + "grad_norm": 3.1702499389648438, + "learning_rate": 1.2219261515036164e-06, + "loss": 1.3842, + "step": 2293 + }, + { + "epoch": 2.3564458140729325, + "grad_norm": 3.022472620010376, + "learning_rate": 1.2200228397411497e-06, + "loss": 1.3946, + "step": 2294 + }, + { + "epoch": 2.357473035439137, + "grad_norm": 3.2721829414367676, + "learning_rate": 1.2181195279786829e-06, + "loss": 1.4444, + "step": 2295 + }, + { + "epoch": 2.3585002568053417, + "grad_norm": 3.1014726161956787, + "learning_rate": 1.2162162162162164e-06, + "loss": 1.478, + "step": 2296 + }, + { + "epoch": 2.359527478171546, + "grad_norm": 3.1011557579040527, + "learning_rate": 1.2143129044537496e-06, + "loss": 1.4067, + "step": 2297 + }, + { + "epoch": 2.3605546995377504, + "grad_norm": 3.09366512298584, + "learning_rate": 1.212409592691283e-06, + "loss": 1.3998, + "step": 2298 + }, + { + "epoch": 2.361581920903955, + "grad_norm": 3.1009769439697266, + "learning_rate": 1.2105062809288163e-06, + "loss": 1.4072, + "step": 2299 + }, + { + "epoch": 2.362609142270159, + "grad_norm": 3.1026828289031982, + "learning_rate": 1.2086029691663495e-06, + "loss": 1.4213, + "step": 2300 + }, + { + "epoch": 2.362609142270159, + "eval_loss": 1.8786081075668335, + "eval_runtime": 44.8368, + "eval_samples_per_second": 22.303, + "eval_steps_per_second": 0.937, + "step": 2300 + }, + { + "epoch": 2.3636363636363638, + "grad_norm": 3.214540481567383, + "learning_rate": 1.2066996574038828e-06, + "loss": 1.4397, + "step": 2301 + }, + { + "epoch": 2.364663585002568, + "grad_norm": 3.0264482498168945, + "learning_rate": 1.2047963456414162e-06, + "loss": 1.3706, + "step": 2302 + }, + { + "epoch": 2.3656908063687725, + "grad_norm": 3.361640453338623, + "learning_rate": 1.2028930338789496e-06, + "loss": 1.4427, + "step": 2303 + }, + { + "epoch": 2.366718027734977, + "grad_norm": 3.1837921142578125, + "learning_rate": 1.2009897221164827e-06, + "loss": 1.4235, + "step": 2304 + }, + { + "epoch": 2.3677452491011812, + "grad_norm": 3.1406962871551514, + "learning_rate": 1.199086410354016e-06, + "loss": 1.3666, + "step": 2305 + }, + { + "epoch": 2.368772470467386, + "grad_norm": 3.189342975616455, + "learning_rate": 1.1971830985915492e-06, + "loss": 1.4383, + "step": 2306 + }, + { + "epoch": 2.36979969183359, + "grad_norm": 3.163757801055908, + "learning_rate": 1.1952797868290828e-06, + "loss": 1.4025, + "step": 2307 + }, + { + "epoch": 2.3708269131997946, + "grad_norm": 3.415684223175049, + "learning_rate": 1.193376475066616e-06, + "loss": 1.4026, + "step": 2308 + }, + { + "epoch": 2.371854134565999, + "grad_norm": 3.28863787651062, + "learning_rate": 1.1914731633041493e-06, + "loss": 1.4249, + "step": 2309 + }, + { + "epoch": 2.3728813559322033, + "grad_norm": 3.2626993656158447, + "learning_rate": 1.1895698515416827e-06, + "loss": 1.4409, + "step": 2310 + }, + { + "epoch": 2.3728813559322033, + "eval_loss": 1.8808053731918335, + "eval_runtime": 41.9147, + "eval_samples_per_second": 23.858, + "eval_steps_per_second": 1.002, + "step": 2310 + }, + { + "epoch": 2.373908577298408, + "grad_norm": 3.252225160598755, + "learning_rate": 1.1876665397792158e-06, + "loss": 1.3958, + "step": 2311 + }, + { + "epoch": 2.3749357986646125, + "grad_norm": 3.0922813415527344, + "learning_rate": 1.1857632280167492e-06, + "loss": 1.3902, + "step": 2312 + }, + { + "epoch": 2.3759630200308166, + "grad_norm": 3.242309808731079, + "learning_rate": 1.1838599162542825e-06, + "loss": 1.4393, + "step": 2313 + }, + { + "epoch": 2.376990241397021, + "grad_norm": 3.0868842601776123, + "learning_rate": 1.181956604491816e-06, + "loss": 1.3587, + "step": 2314 + }, + { + "epoch": 2.3780174627632253, + "grad_norm": 3.1234078407287598, + "learning_rate": 1.180053292729349e-06, + "loss": 1.4387, + "step": 2315 + }, + { + "epoch": 2.37904468412943, + "grad_norm": 3.1602776050567627, + "learning_rate": 1.1781499809668824e-06, + "loss": 1.4151, + "step": 2316 + }, + { + "epoch": 2.380071905495634, + "grad_norm": 3.359309673309326, + "learning_rate": 1.1762466692044158e-06, + "loss": 1.4918, + "step": 2317 + }, + { + "epoch": 2.3810991268618387, + "grad_norm": 3.204411268234253, + "learning_rate": 1.1743433574419491e-06, + "loss": 1.4199, + "step": 2318 + }, + { + "epoch": 2.3821263482280433, + "grad_norm": 3.1535773277282715, + "learning_rate": 1.1724400456794823e-06, + "loss": 1.4494, + "step": 2319 + }, + { + "epoch": 2.3831535695942474, + "grad_norm": 3.124729871749878, + "learning_rate": 1.1705367339170157e-06, + "loss": 1.3915, + "step": 2320 + }, + { + "epoch": 2.3831535695942474, + "eval_loss": 1.8865554332733154, + "eval_runtime": 43.2617, + "eval_samples_per_second": 23.115, + "eval_steps_per_second": 0.971, + "step": 2320 + }, + { + "epoch": 2.384180790960452, + "grad_norm": 3.1918749809265137, + "learning_rate": 1.168633422154549e-06, + "loss": 1.4486, + "step": 2321 + }, + { + "epoch": 2.3852080123266566, + "grad_norm": 3.362424850463867, + "learning_rate": 1.1667301103920824e-06, + "loss": 1.4313, + "step": 2322 + }, + { + "epoch": 2.3862352336928607, + "grad_norm": 3.0750956535339355, + "learning_rate": 1.1648267986296157e-06, + "loss": 1.3796, + "step": 2323 + }, + { + "epoch": 2.3872624550590653, + "grad_norm": 3.081636428833008, + "learning_rate": 1.162923486867149e-06, + "loss": 1.3569, + "step": 2324 + }, + { + "epoch": 2.3882896764252695, + "grad_norm": 3.1504268646240234, + "learning_rate": 1.1610201751046823e-06, + "loss": 1.4058, + "step": 2325 + }, + { + "epoch": 2.389316897791474, + "grad_norm": 3.1414480209350586, + "learning_rate": 1.1591168633422154e-06, + "loss": 1.4215, + "step": 2326 + }, + { + "epoch": 2.3903441191576786, + "grad_norm": 3.262014389038086, + "learning_rate": 1.157213551579749e-06, + "loss": 1.4678, + "step": 2327 + }, + { + "epoch": 2.391371340523883, + "grad_norm": 3.163388729095459, + "learning_rate": 1.1553102398172821e-06, + "loss": 1.4479, + "step": 2328 + }, + { + "epoch": 2.3923985618900874, + "grad_norm": 3.0833804607391357, + "learning_rate": 1.1534069280548155e-06, + "loss": 1.4164, + "step": 2329 + }, + { + "epoch": 2.3934257832562915, + "grad_norm": 3.127962589263916, + "learning_rate": 1.1515036162923487e-06, + "loss": 1.4022, + "step": 2330 + }, + { + "epoch": 2.3934257832562915, + "eval_loss": 1.8800996541976929, + "eval_runtime": 46.0509, + "eval_samples_per_second": 21.715, + "eval_steps_per_second": 0.912, + "step": 2330 + }, + { + "epoch": 2.394453004622496, + "grad_norm": 3.239804983139038, + "learning_rate": 1.149600304529882e-06, + "loss": 1.4315, + "step": 2331 + }, + { + "epoch": 2.3954802259887007, + "grad_norm": 3.0544466972351074, + "learning_rate": 1.1476969927674154e-06, + "loss": 1.3993, + "step": 2332 + }, + { + "epoch": 2.396507447354905, + "grad_norm": 3.2019660472869873, + "learning_rate": 1.1457936810049487e-06, + "loss": 1.4055, + "step": 2333 + }, + { + "epoch": 2.3975346687211094, + "grad_norm": 3.1437528133392334, + "learning_rate": 1.143890369242482e-06, + "loss": 1.4109, + "step": 2334 + }, + { + "epoch": 2.398561890087314, + "grad_norm": 3.274785041809082, + "learning_rate": 1.1419870574800153e-06, + "loss": 1.4603, + "step": 2335 + }, + { + "epoch": 2.399589111453518, + "grad_norm": 3.129502058029175, + "learning_rate": 1.1400837457175486e-06, + "loss": 1.3705, + "step": 2336 + }, + { + "epoch": 2.4006163328197228, + "grad_norm": 3.1008121967315674, + "learning_rate": 1.138180433955082e-06, + "loss": 1.4041, + "step": 2337 + }, + { + "epoch": 2.401643554185927, + "grad_norm": 3.138871431350708, + "learning_rate": 1.1362771221926153e-06, + "loss": 1.375, + "step": 2338 + }, + { + "epoch": 2.4026707755521315, + "grad_norm": 3.1868844032287598, + "learning_rate": 1.1343738104301485e-06, + "loss": 1.4419, + "step": 2339 + }, + { + "epoch": 2.403697996918336, + "grad_norm": 3.2045419216156006, + "learning_rate": 1.1324704986676819e-06, + "loss": 1.4233, + "step": 2340 + }, + { + "epoch": 2.403697996918336, + "eval_loss": 1.8847991228103638, + "eval_runtime": 45.2785, + "eval_samples_per_second": 22.086, + "eval_steps_per_second": 0.928, + "step": 2340 + }, + { + "epoch": 2.4047252182845402, + "grad_norm": 3.190446615219116, + "learning_rate": 1.130567186905215e-06, + "loss": 1.3998, + "step": 2341 + }, + { + "epoch": 2.405752439650745, + "grad_norm": 3.1729485988616943, + "learning_rate": 1.1286638751427486e-06, + "loss": 1.4609, + "step": 2342 + }, + { + "epoch": 2.406779661016949, + "grad_norm": 3.0363855361938477, + "learning_rate": 1.1267605633802817e-06, + "loss": 1.3987, + "step": 2343 + }, + { + "epoch": 2.4078068823831535, + "grad_norm": 3.087101459503174, + "learning_rate": 1.124857251617815e-06, + "loss": 1.4006, + "step": 2344 + }, + { + "epoch": 2.408834103749358, + "grad_norm": 3.1294353008270264, + "learning_rate": 1.1229539398553485e-06, + "loss": 1.4475, + "step": 2345 + }, + { + "epoch": 2.4098613251155623, + "grad_norm": 3.2285165786743164, + "learning_rate": 1.1210506280928816e-06, + "loss": 1.4778, + "step": 2346 + }, + { + "epoch": 2.410888546481767, + "grad_norm": 3.1207237243652344, + "learning_rate": 1.119147316330415e-06, + "loss": 1.4174, + "step": 2347 + }, + { + "epoch": 2.4119157678479715, + "grad_norm": 3.248936653137207, + "learning_rate": 1.1172440045679483e-06, + "loss": 1.4532, + "step": 2348 + }, + { + "epoch": 2.4129429892141756, + "grad_norm": 3.1401236057281494, + "learning_rate": 1.1153406928054817e-06, + "loss": 1.4282, + "step": 2349 + }, + { + "epoch": 2.41397021058038, + "grad_norm": 3.179011821746826, + "learning_rate": 1.1134373810430148e-06, + "loss": 1.4169, + "step": 2350 + }, + { + "epoch": 2.41397021058038, + "eval_loss": 1.8840751647949219, + "eval_runtime": 44.964, + "eval_samples_per_second": 22.24, + "eval_steps_per_second": 0.934, + "step": 2350 + }, + { + "epoch": 2.4149974319465843, + "grad_norm": 3.1402602195739746, + "learning_rate": 1.1115340692805482e-06, + "loss": 1.4533, + "step": 2351 + }, + { + "epoch": 2.416024653312789, + "grad_norm": 3.315294027328491, + "learning_rate": 1.1096307575180816e-06, + "loss": 1.4546, + "step": 2352 + }, + { + "epoch": 2.4170518746789935, + "grad_norm": 3.172640085220337, + "learning_rate": 1.107727445755615e-06, + "loss": 1.4203, + "step": 2353 + }, + { + "epoch": 2.4180790960451977, + "grad_norm": 3.1118452548980713, + "learning_rate": 1.105824133993148e-06, + "loss": 1.3931, + "step": 2354 + }, + { + "epoch": 2.4191063174114023, + "grad_norm": 3.1557371616363525, + "learning_rate": 1.1039208222306814e-06, + "loss": 1.3937, + "step": 2355 + }, + { + "epoch": 2.4201335387776064, + "grad_norm": 3.1625313758850098, + "learning_rate": 1.1020175104682148e-06, + "loss": 1.4184, + "step": 2356 + }, + { + "epoch": 2.421160760143811, + "grad_norm": 3.0680766105651855, + "learning_rate": 1.1001141987057482e-06, + "loss": 1.4219, + "step": 2357 + }, + { + "epoch": 2.4221879815100156, + "grad_norm": 2.9973316192626953, + "learning_rate": 1.0982108869432813e-06, + "loss": 1.4075, + "step": 2358 + }, + { + "epoch": 2.4232152028762197, + "grad_norm": 3.202812433242798, + "learning_rate": 1.0963075751808147e-06, + "loss": 1.4227, + "step": 2359 + }, + { + "epoch": 2.4242424242424243, + "grad_norm": 3.296248197555542, + "learning_rate": 1.094404263418348e-06, + "loss": 1.4486, + "step": 2360 + }, + { + "epoch": 2.4242424242424243, + "eval_loss": 1.8838396072387695, + "eval_runtime": 45.3465, + "eval_samples_per_second": 22.052, + "eval_steps_per_second": 0.926, + "step": 2360 + }, + { + "epoch": 2.425269645608629, + "grad_norm": 3.1101279258728027, + "learning_rate": 1.0925009516558812e-06, + "loss": 1.413, + "step": 2361 + }, + { + "epoch": 2.426296866974833, + "grad_norm": 3.070812463760376, + "learning_rate": 1.0905976398934148e-06, + "loss": 1.4198, + "step": 2362 + }, + { + "epoch": 2.4273240883410376, + "grad_norm": 3.0629656314849854, + "learning_rate": 1.088694328130948e-06, + "loss": 1.4121, + "step": 2363 + }, + { + "epoch": 2.428351309707242, + "grad_norm": 3.228043556213379, + "learning_rate": 1.0867910163684813e-06, + "loss": 1.4292, + "step": 2364 + }, + { + "epoch": 2.4293785310734464, + "grad_norm": 3.068103790283203, + "learning_rate": 1.0848877046060144e-06, + "loss": 1.3747, + "step": 2365 + }, + { + "epoch": 2.4304057524396505, + "grad_norm": 3.105422019958496, + "learning_rate": 1.0829843928435478e-06, + "loss": 1.4128, + "step": 2366 + }, + { + "epoch": 2.431432973805855, + "grad_norm": 3.2109289169311523, + "learning_rate": 1.0810810810810812e-06, + "loss": 1.4051, + "step": 2367 + }, + { + "epoch": 2.4324601951720597, + "grad_norm": 3.1759419441223145, + "learning_rate": 1.0791777693186145e-06, + "loss": 1.4164, + "step": 2368 + }, + { + "epoch": 2.433487416538264, + "grad_norm": 3.2143466472625732, + "learning_rate": 1.0772744575561479e-06, + "loss": 1.4363, + "step": 2369 + }, + { + "epoch": 2.4345146379044684, + "grad_norm": 3.155259132385254, + "learning_rate": 1.075371145793681e-06, + "loss": 1.4089, + "step": 2370 + }, + { + "epoch": 2.4345146379044684, + "eval_loss": 1.8794890642166138, + "eval_runtime": 48.175, + "eval_samples_per_second": 20.758, + "eval_steps_per_second": 0.872, + "step": 2370 + }, + { + "epoch": 2.435541859270673, + "grad_norm": 3.2817187309265137, + "learning_rate": 1.0734678340312144e-06, + "loss": 1.404, + "step": 2371 + }, + { + "epoch": 2.436569080636877, + "grad_norm": 3.2742044925689697, + "learning_rate": 1.0715645222687478e-06, + "loss": 1.4048, + "step": 2372 + }, + { + "epoch": 2.4375963020030817, + "grad_norm": 3.111154556274414, + "learning_rate": 1.0696612105062811e-06, + "loss": 1.3966, + "step": 2373 + }, + { + "epoch": 2.438623523369286, + "grad_norm": 3.083907127380371, + "learning_rate": 1.0677578987438143e-06, + "loss": 1.3931, + "step": 2374 + }, + { + "epoch": 2.4396507447354905, + "grad_norm": 3.164278030395508, + "learning_rate": 1.0658545869813476e-06, + "loss": 1.4373, + "step": 2375 + }, + { + "epoch": 2.440677966101695, + "grad_norm": 3.17753005027771, + "learning_rate": 1.0639512752188808e-06, + "loss": 1.3874, + "step": 2376 + }, + { + "epoch": 2.441705187467899, + "grad_norm": 3.0339651107788086, + "learning_rate": 1.0620479634564144e-06, + "loss": 1.401, + "step": 2377 + }, + { + "epoch": 2.442732408834104, + "grad_norm": 3.1768946647644043, + "learning_rate": 1.0601446516939475e-06, + "loss": 1.4083, + "step": 2378 + }, + { + "epoch": 2.443759630200308, + "grad_norm": 3.0505385398864746, + "learning_rate": 1.0582413399314809e-06, + "loss": 1.3733, + "step": 2379 + }, + { + "epoch": 2.4447868515665125, + "grad_norm": 3.064157485961914, + "learning_rate": 1.0563380281690142e-06, + "loss": 1.3782, + "step": 2380 + }, + { + "epoch": 2.4447868515665125, + "eval_loss": 1.879464864730835, + "eval_runtime": 50.9219, + "eval_samples_per_second": 19.638, + "eval_steps_per_second": 0.825, + "step": 2380 + }, + { + "epoch": 2.445814072932717, + "grad_norm": 3.1522152423858643, + "learning_rate": 1.0544347164065474e-06, + "loss": 1.3926, + "step": 2381 + }, + { + "epoch": 2.4468412942989213, + "grad_norm": 3.073568820953369, + "learning_rate": 1.0525314046440808e-06, + "loss": 1.399, + "step": 2382 + }, + { + "epoch": 2.447868515665126, + "grad_norm": 3.171452522277832, + "learning_rate": 1.0506280928816141e-06, + "loss": 1.4208, + "step": 2383 + }, + { + "epoch": 2.4488957370313305, + "grad_norm": 3.1839122772216797, + "learning_rate": 1.0487247811191475e-06, + "loss": 1.3923, + "step": 2384 + }, + { + "epoch": 2.4499229583975346, + "grad_norm": 3.0783913135528564, + "learning_rate": 1.0468214693566806e-06, + "loss": 1.3984, + "step": 2385 + }, + { + "epoch": 2.450950179763739, + "grad_norm": 3.203411817550659, + "learning_rate": 1.044918157594214e-06, + "loss": 1.4106, + "step": 2386 + }, + { + "epoch": 2.4519774011299433, + "grad_norm": 3.254432201385498, + "learning_rate": 1.0430148458317474e-06, + "loss": 1.4426, + "step": 2387 + }, + { + "epoch": 2.453004622496148, + "grad_norm": 3.2302334308624268, + "learning_rate": 1.0411115340692807e-06, + "loss": 1.3964, + "step": 2388 + }, + { + "epoch": 2.4540318438623525, + "grad_norm": 3.0067214965820312, + "learning_rate": 1.0392082223068139e-06, + "loss": 1.3749, + "step": 2389 + }, + { + "epoch": 2.4550590652285567, + "grad_norm": 3.2783524990081787, + "learning_rate": 1.0373049105443472e-06, + "loss": 1.4603, + "step": 2390 + }, + { + "epoch": 2.4550590652285567, + "eval_loss": 1.8760793209075928, + "eval_runtime": 47.0458, + "eval_samples_per_second": 21.256, + "eval_steps_per_second": 0.893, + "step": 2390 + }, + { + "epoch": 2.4560862865947612, + "grad_norm": 3.2079291343688965, + "learning_rate": 1.0354015987818806e-06, + "loss": 1.4176, + "step": 2391 + }, + { + "epoch": 2.4571135079609654, + "grad_norm": 3.0905849933624268, + "learning_rate": 1.033498287019414e-06, + "loss": 1.428, + "step": 2392 + }, + { + "epoch": 2.45814072932717, + "grad_norm": 3.034698963165283, + "learning_rate": 1.0315949752569471e-06, + "loss": 1.3884, + "step": 2393 + }, + { + "epoch": 2.4591679506933746, + "grad_norm": 3.2759525775909424, + "learning_rate": 1.0296916634944805e-06, + "loss": 1.4455, + "step": 2394 + }, + { + "epoch": 2.4601951720595787, + "grad_norm": 3.2399072647094727, + "learning_rate": 1.0277883517320138e-06, + "loss": 1.3684, + "step": 2395 + }, + { + "epoch": 2.4612223934257833, + "grad_norm": 3.173449993133545, + "learning_rate": 1.025885039969547e-06, + "loss": 1.4227, + "step": 2396 + }, + { + "epoch": 2.462249614791988, + "grad_norm": 3.244391679763794, + "learning_rate": 1.0239817282070806e-06, + "loss": 1.4525, + "step": 2397 + }, + { + "epoch": 2.463276836158192, + "grad_norm": 3.0395283699035645, + "learning_rate": 1.0220784164446137e-06, + "loss": 1.3501, + "step": 2398 + }, + { + "epoch": 2.4643040575243966, + "grad_norm": 3.0880112648010254, + "learning_rate": 1.020175104682147e-06, + "loss": 1.395, + "step": 2399 + }, + { + "epoch": 2.4653312788906008, + "grad_norm": 3.1978697776794434, + "learning_rate": 1.0182717929196802e-06, + "loss": 1.3984, + "step": 2400 + }, + { + "epoch": 2.4653312788906008, + "eval_loss": 1.8770432472229004, + "eval_runtime": 45.2381, + "eval_samples_per_second": 22.105, + "eval_steps_per_second": 0.928, + "step": 2400 + }, + { + "epoch": 2.4663585002568054, + "grad_norm": 3.132573366165161, + "learning_rate": 1.0163684811572136e-06, + "loss": 1.413, + "step": 2401 + }, + { + "epoch": 2.4673857216230095, + "grad_norm": 3.2909750938415527, + "learning_rate": 1.014465169394747e-06, + "loss": 1.4493, + "step": 2402 + }, + { + "epoch": 2.468412942989214, + "grad_norm": 3.1101572513580322, + "learning_rate": 1.0125618576322803e-06, + "loss": 1.3656, + "step": 2403 + }, + { + "epoch": 2.4694401643554187, + "grad_norm": 3.202353000640869, + "learning_rate": 1.0106585458698137e-06, + "loss": 1.4134, + "step": 2404 + }, + { + "epoch": 2.470467385721623, + "grad_norm": 3.1683971881866455, + "learning_rate": 1.0087552341073468e-06, + "loss": 1.3877, + "step": 2405 + }, + { + "epoch": 2.4714946070878274, + "grad_norm": 3.1234371662139893, + "learning_rate": 1.0068519223448802e-06, + "loss": 1.3931, + "step": 2406 + }, + { + "epoch": 2.472521828454032, + "grad_norm": 3.2071094512939453, + "learning_rate": 1.0049486105824133e-06, + "loss": 1.4685, + "step": 2407 + }, + { + "epoch": 2.473549049820236, + "grad_norm": 3.1973717212677, + "learning_rate": 1.003045298819947e-06, + "loss": 1.4531, + "step": 2408 + }, + { + "epoch": 2.4745762711864407, + "grad_norm": 3.134814500808716, + "learning_rate": 1.00114198705748e-06, + "loss": 1.3936, + "step": 2409 + }, + { + "epoch": 2.4756034925526453, + "grad_norm": 3.1617591381073, + "learning_rate": 9.992386752950134e-07, + "loss": 1.3855, + "step": 2410 + }, + { + "epoch": 2.4756034925526453, + "eval_loss": 1.8814165592193604, + "eval_runtime": 43.582, + "eval_samples_per_second": 22.945, + "eval_steps_per_second": 0.964, + "step": 2410 + }, + { + "epoch": 2.4766307139188495, + "grad_norm": 3.1300113201141357, + "learning_rate": 9.973353635325466e-07, + "loss": 1.38, + "step": 2411 + }, + { + "epoch": 2.477657935285054, + "grad_norm": 3.1064655780792236, + "learning_rate": 9.9543205177008e-07, + "loss": 1.3716, + "step": 2412 + }, + { + "epoch": 2.478685156651258, + "grad_norm": 3.210845470428467, + "learning_rate": 9.935287400076133e-07, + "loss": 1.3765, + "step": 2413 + }, + { + "epoch": 2.479712378017463, + "grad_norm": 3.077348470687866, + "learning_rate": 9.916254282451467e-07, + "loss": 1.3906, + "step": 2414 + }, + { + "epoch": 2.480739599383667, + "grad_norm": 3.149536371231079, + "learning_rate": 9.8972211648268e-07, + "loss": 1.3691, + "step": 2415 + }, + { + "epoch": 2.4817668207498715, + "grad_norm": 3.229722261428833, + "learning_rate": 9.878188047202132e-07, + "loss": 1.407, + "step": 2416 + }, + { + "epoch": 2.482794042116076, + "grad_norm": 3.180819034576416, + "learning_rate": 9.859154929577465e-07, + "loss": 1.4256, + "step": 2417 + }, + { + "epoch": 2.4838212634822803, + "grad_norm": 3.1556131839752197, + "learning_rate": 9.8401218119528e-07, + "loss": 1.3997, + "step": 2418 + }, + { + "epoch": 2.484848484848485, + "grad_norm": 3.1113390922546387, + "learning_rate": 9.821088694328133e-07, + "loss": 1.3784, + "step": 2419 + }, + { + "epoch": 2.4858757062146895, + "grad_norm": 3.226135492324829, + "learning_rate": 9.802055576703464e-07, + "loss": 1.3974, + "step": 2420 + }, + { + "epoch": 2.4858757062146895, + "eval_loss": 1.8805677890777588, + "eval_runtime": 46.1144, + "eval_samples_per_second": 21.685, + "eval_steps_per_second": 0.911, + "step": 2420 + }, + { + "epoch": 2.4869029275808936, + "grad_norm": 3.1731371879577637, + "learning_rate": 9.783022459078798e-07, + "loss": 1.4439, + "step": 2421 + }, + { + "epoch": 2.487930148947098, + "grad_norm": 3.254910469055176, + "learning_rate": 9.76398934145413e-07, + "loss": 1.4279, + "step": 2422 + }, + { + "epoch": 2.4889573703133023, + "grad_norm": 3.0564846992492676, + "learning_rate": 9.744956223829465e-07, + "loss": 1.4086, + "step": 2423 + }, + { + "epoch": 2.489984591679507, + "grad_norm": 3.267612934112549, + "learning_rate": 9.725923106204797e-07, + "loss": 1.4294, + "step": 2424 + }, + { + "epoch": 2.4910118130457115, + "grad_norm": 3.2458605766296387, + "learning_rate": 9.70688998858013e-07, + "loss": 1.4338, + "step": 2425 + }, + { + "epoch": 2.4920390344119157, + "grad_norm": 3.357015609741211, + "learning_rate": 9.687856870955464e-07, + "loss": 1.4688, + "step": 2426 + }, + { + "epoch": 2.4930662557781202, + "grad_norm": 3.1078314781188965, + "learning_rate": 9.668823753330795e-07, + "loss": 1.4341, + "step": 2427 + }, + { + "epoch": 2.4940934771443244, + "grad_norm": 3.1556806564331055, + "learning_rate": 9.64979063570613e-07, + "loss": 1.4264, + "step": 2428 + }, + { + "epoch": 2.495120698510529, + "grad_norm": 3.076340436935425, + "learning_rate": 9.630757518081463e-07, + "loss": 1.3715, + "step": 2429 + }, + { + "epoch": 2.4961479198767336, + "grad_norm": 3.1635854244232178, + "learning_rate": 9.611724400456796e-07, + "loss": 1.4048, + "step": 2430 + }, + { + "epoch": 2.4961479198767336, + "eval_loss": 1.8799227476119995, + "eval_runtime": 51.3108, + "eval_samples_per_second": 19.489, + "eval_steps_per_second": 0.819, + "step": 2430 + }, + { + "epoch": 2.4971751412429377, + "grad_norm": 3.0959770679473877, + "learning_rate": 9.592691282832128e-07, + "loss": 1.4043, + "step": 2431 + }, + { + "epoch": 2.4982023626091423, + "grad_norm": 3.1850130558013916, + "learning_rate": 9.573658165207461e-07, + "loss": 1.3823, + "step": 2432 + }, + { + "epoch": 2.499229583975347, + "grad_norm": 3.17700457572937, + "learning_rate": 9.554625047582795e-07, + "loss": 1.4171, + "step": 2433 + }, + { + "epoch": 2.500256805341551, + "grad_norm": 3.17700457572937, + "learning_rate": 9.554625047582795e-07, + "loss": 1.4311, + "step": 2434 + }, + { + "epoch": 2.5012840267077556, + "grad_norm": 3.066020965576172, + "learning_rate": 9.535591929958129e-07, + "loss": 1.3727, + "step": 2435 + }, + { + "epoch": 2.50231124807396, + "grad_norm": 3.2088215351104736, + "learning_rate": 9.516558812333461e-07, + "loss": 1.4193, + "step": 2436 + }, + { + "epoch": 2.5033384694401644, + "grad_norm": 3.231025457382202, + "learning_rate": 9.497525694708794e-07, + "loss": 1.4222, + "step": 2437 + }, + { + "epoch": 2.5043656908063685, + "grad_norm": 3.115729331970215, + "learning_rate": 9.478492577084126e-07, + "loss": 1.3897, + "step": 2438 + }, + { + "epoch": 2.505392912172573, + "grad_norm": 3.0405523777008057, + "learning_rate": 9.459459459459461e-07, + "loss": 1.3993, + "step": 2439 + }, + { + "epoch": 2.5064201335387777, + "grad_norm": 3.1454200744628906, + "learning_rate": 9.440426341834794e-07, + "loss": 1.3896, + "step": 2440 + }, + { + "epoch": 2.5064201335387777, + "eval_loss": 1.8846684694290161, + "eval_runtime": 48.722, + "eval_samples_per_second": 20.525, + "eval_steps_per_second": 0.862, + "step": 2440 + }, + { + "epoch": 2.507447354904982, + "grad_norm": 3.172870397567749, + "learning_rate": 9.421393224210126e-07, + "loss": 1.4045, + "step": 2441 + }, + { + "epoch": 2.5084745762711864, + "grad_norm": 3.237417459487915, + "learning_rate": 9.402360106585459e-07, + "loss": 1.4074, + "step": 2442 + }, + { + "epoch": 2.509501797637391, + "grad_norm": 3.1962826251983643, + "learning_rate": 9.383326988960792e-07, + "loss": 1.3629, + "step": 2443 + }, + { + "epoch": 2.510529019003595, + "grad_norm": 3.230431079864502, + "learning_rate": 9.364293871336126e-07, + "loss": 1.3665, + "step": 2444 + }, + { + "epoch": 2.5115562403697997, + "grad_norm": 3.0424323081970215, + "learning_rate": 9.345260753711458e-07, + "loss": 1.3862, + "step": 2445 + }, + { + "epoch": 2.5125834617360043, + "grad_norm": 3.291992425918579, + "learning_rate": 9.326227636086792e-07, + "loss": 1.4165, + "step": 2446 + }, + { + "epoch": 2.5136106831022085, + "grad_norm": 3.060072660446167, + "learning_rate": 9.307194518462125e-07, + "loss": 1.3607, + "step": 2447 + }, + { + "epoch": 2.514637904468413, + "grad_norm": 3.2155919075012207, + "learning_rate": 9.288161400837457e-07, + "loss": 1.4896, + "step": 2448 + }, + { + "epoch": 2.515665125834617, + "grad_norm": 3.0423970222473145, + "learning_rate": 9.269128283212792e-07, + "loss": 1.3669, + "step": 2449 + }, + { + "epoch": 2.516692347200822, + "grad_norm": 3.080432653427124, + "learning_rate": 9.250095165588125e-07, + "loss": 1.3987, + "step": 2450 + }, + { + "epoch": 2.516692347200822, + "eval_loss": 1.8857015371322632, + "eval_runtime": 47.8346, + "eval_samples_per_second": 20.905, + "eval_steps_per_second": 0.878, + "step": 2450 + }, + { + "epoch": 2.517719568567026, + "grad_norm": 3.2272510528564453, + "learning_rate": 9.231062047963457e-07, + "loss": 1.4614, + "step": 2451 + }, + { + "epoch": 2.5187467899332305, + "grad_norm": 3.181736707687378, + "learning_rate": 9.21202893033879e-07, + "loss": 1.4657, + "step": 2452 + }, + { + "epoch": 2.519774011299435, + "grad_norm": 3.119811773300171, + "learning_rate": 9.192995812714123e-07, + "loss": 1.4099, + "step": 2453 + }, + { + "epoch": 2.5208012326656393, + "grad_norm": 3.149948835372925, + "learning_rate": 9.173962695089457e-07, + "loss": 1.4194, + "step": 2454 + }, + { + "epoch": 2.521828454031844, + "grad_norm": 3.099076747894287, + "learning_rate": 9.154929577464789e-07, + "loss": 1.4063, + "step": 2455 + }, + { + "epoch": 2.5228556753980484, + "grad_norm": 3.166980743408203, + "learning_rate": 9.135896459840123e-07, + "loss": 1.4279, + "step": 2456 + }, + { + "epoch": 2.5238828967642526, + "grad_norm": 3.2286252975463867, + "learning_rate": 9.116863342215456e-07, + "loss": 1.4189, + "step": 2457 + }, + { + "epoch": 2.524910118130457, + "grad_norm": 3.112492799758911, + "learning_rate": 9.097830224590788e-07, + "loss": 1.4488, + "step": 2458 + }, + { + "epoch": 2.5259373394966618, + "grad_norm": 3.03338360786438, + "learning_rate": 9.078797106966122e-07, + "loss": 1.3641, + "step": 2459 + }, + { + "epoch": 2.526964560862866, + "grad_norm": 3.1647260189056396, + "learning_rate": 9.059763989341455e-07, + "loss": 1.4406, + "step": 2460 + }, + { + "epoch": 2.526964560862866, + "eval_loss": 1.8841506242752075, + "eval_runtime": 47.3542, + "eval_samples_per_second": 21.117, + "eval_steps_per_second": 0.887, + "step": 2460 + }, + { + "epoch": 2.5279917822290705, + "grad_norm": 3.1184210777282715, + "learning_rate": 9.040730871716788e-07, + "loss": 1.3837, + "step": 2461 + }, + { + "epoch": 2.5290190035952747, + "grad_norm": 3.0378000736236572, + "learning_rate": 9.021697754092121e-07, + "loss": 1.3998, + "step": 2462 + }, + { + "epoch": 2.5300462249614792, + "grad_norm": 3.1301138401031494, + "learning_rate": 9.002664636467453e-07, + "loss": 1.4128, + "step": 2463 + }, + { + "epoch": 2.5310734463276834, + "grad_norm": 3.2062747478485107, + "learning_rate": 8.983631518842788e-07, + "loss": 1.4288, + "step": 2464 + }, + { + "epoch": 2.532100667693888, + "grad_norm": 3.296144962310791, + "learning_rate": 8.96459840121812e-07, + "loss": 1.4513, + "step": 2465 + }, + { + "epoch": 2.5331278890600926, + "grad_norm": 3.1658542156219482, + "learning_rate": 8.945565283593453e-07, + "loss": 1.4035, + "step": 2466 + }, + { + "epoch": 2.5341551104262967, + "grad_norm": 3.2297472953796387, + "learning_rate": 8.926532165968787e-07, + "loss": 1.4241, + "step": 2467 + }, + { + "epoch": 2.5351823317925013, + "grad_norm": 3.257892370223999, + "learning_rate": 8.907499048344119e-07, + "loss": 1.4206, + "step": 2468 + }, + { + "epoch": 2.536209553158706, + "grad_norm": 3.136453866958618, + "learning_rate": 8.888465930719453e-07, + "loss": 1.3672, + "step": 2469 + }, + { + "epoch": 2.53723677452491, + "grad_norm": 3.0918774604797363, + "learning_rate": 8.869432813094786e-07, + "loss": 1.377, + "step": 2470 + }, + { + "epoch": 2.53723677452491, + "eval_loss": 1.885011911392212, + "eval_runtime": 46.9296, + "eval_samples_per_second": 21.309, + "eval_steps_per_second": 0.895, + "step": 2470 + }, + { + "epoch": 2.5382639958911146, + "grad_norm": 3.1935360431671143, + "learning_rate": 8.850399695470119e-07, + "loss": 1.4104, + "step": 2471 + }, + { + "epoch": 2.539291217257319, + "grad_norm": 3.2117061614990234, + "learning_rate": 8.831366577845452e-07, + "loss": 1.43, + "step": 2472 + }, + { + "epoch": 2.5403184386235234, + "grad_norm": 3.1783292293548584, + "learning_rate": 8.812333460220784e-07, + "loss": 1.3886, + "step": 2473 + }, + { + "epoch": 2.541345659989728, + "grad_norm": 3.118147373199463, + "learning_rate": 8.793300342596119e-07, + "loss": 1.3557, + "step": 2474 + }, + { + "epoch": 2.542372881355932, + "grad_norm": 3.154395341873169, + "learning_rate": 8.774267224971451e-07, + "loss": 1.4161, + "step": 2475 + }, + { + "epoch": 2.5434001027221367, + "grad_norm": 3.2336461544036865, + "learning_rate": 8.755234107346784e-07, + "loss": 1.4484, + "step": 2476 + }, + { + "epoch": 2.544427324088341, + "grad_norm": 3.058476686477661, + "learning_rate": 8.736200989722117e-07, + "loss": 1.3745, + "step": 2477 + }, + { + "epoch": 2.5454545454545454, + "grad_norm": 3.0842530727386475, + "learning_rate": 8.71716787209745e-07, + "loss": 1.4058, + "step": 2478 + }, + { + "epoch": 2.54648176682075, + "grad_norm": 3.1283812522888184, + "learning_rate": 8.698134754472784e-07, + "loss": 1.4268, + "step": 2479 + }, + { + "epoch": 2.547508988186954, + "grad_norm": 2.9921720027923584, + "learning_rate": 8.679101636848116e-07, + "loss": 1.3503, + "step": 2480 + }, + { + "epoch": 2.547508988186954, + "eval_loss": 1.882778286933899, + "eval_runtime": 46.3444, + "eval_samples_per_second": 21.578, + "eval_steps_per_second": 0.906, + "step": 2480 + }, + { + "epoch": 2.5485362095531587, + "grad_norm": 3.073862075805664, + "learning_rate": 8.66006851922345e-07, + "loss": 1.4279, + "step": 2481 + }, + { + "epoch": 2.5495634309193633, + "grad_norm": 3.0918421745300293, + "learning_rate": 8.641035401598783e-07, + "loss": 1.4119, + "step": 2482 + }, + { + "epoch": 2.5505906522855675, + "grad_norm": 3.090078830718994, + "learning_rate": 8.622002283974115e-07, + "loss": 1.4044, + "step": 2483 + }, + { + "epoch": 2.551617873651772, + "grad_norm": 3.176739454269409, + "learning_rate": 8.60296916634945e-07, + "loss": 1.4354, + "step": 2484 + }, + { + "epoch": 2.5526450950179767, + "grad_norm": 3.0653412342071533, + "learning_rate": 8.583936048724782e-07, + "loss": 1.3854, + "step": 2485 + }, + { + "epoch": 2.553672316384181, + "grad_norm": 3.1199936866760254, + "learning_rate": 8.564902931100115e-07, + "loss": 1.3905, + "step": 2486 + }, + { + "epoch": 2.554699537750385, + "grad_norm": 3.121417999267578, + "learning_rate": 8.545869813475448e-07, + "loss": 1.4175, + "step": 2487 + }, + { + "epoch": 2.5557267591165895, + "grad_norm": 3.208608865737915, + "learning_rate": 8.526836695850781e-07, + "loss": 1.4184, + "step": 2488 + }, + { + "epoch": 2.556753980482794, + "grad_norm": 3.204195261001587, + "learning_rate": 8.507803578226115e-07, + "loss": 1.4309, + "step": 2489 + }, + { + "epoch": 2.5577812018489983, + "grad_norm": 3.159266710281372, + "learning_rate": 8.488770460601447e-07, + "loss": 1.418, + "step": 2490 + }, + { + "epoch": 2.5577812018489983, + "eval_loss": 1.8883039951324463, + "eval_runtime": 49.2187, + "eval_samples_per_second": 20.317, + "eval_steps_per_second": 0.853, + "step": 2490 + }, + { + "epoch": 2.558808423215203, + "grad_norm": 3.1951637268066406, + "learning_rate": 8.46973734297678e-07, + "loss": 1.3811, + "step": 2491 + }, + { + "epoch": 2.5598356445814074, + "grad_norm": 3.0998926162719727, + "learning_rate": 8.450704225352114e-07, + "loss": 1.392, + "step": 2492 + }, + { + "epoch": 2.5608628659476116, + "grad_norm": 3.141263484954834, + "learning_rate": 8.431671107727446e-07, + "loss": 1.3954, + "step": 2493 + }, + { + "epoch": 2.561890087313816, + "grad_norm": 3.2348129749298096, + "learning_rate": 8.41263799010278e-07, + "loss": 1.4165, + "step": 2494 + }, + { + "epoch": 2.5629173086800208, + "grad_norm": 3.1451265811920166, + "learning_rate": 8.393604872478113e-07, + "loss": 1.4087, + "step": 2495 + }, + { + "epoch": 2.563944530046225, + "grad_norm": 3.095964193344116, + "learning_rate": 8.374571754853446e-07, + "loss": 1.3961, + "step": 2496 + }, + { + "epoch": 2.5649717514124295, + "grad_norm": 3.1411421298980713, + "learning_rate": 8.355538637228778e-07, + "loss": 1.4003, + "step": 2497 + }, + { + "epoch": 2.5659989727786336, + "grad_norm": 3.1239426136016846, + "learning_rate": 8.336505519604111e-07, + "loss": 1.4028, + "step": 2498 + }, + { + "epoch": 2.5670261941448382, + "grad_norm": 3.1425132751464844, + "learning_rate": 8.317472401979445e-07, + "loss": 1.4278, + "step": 2499 + }, + { + "epoch": 2.5680534155110424, + "grad_norm": 3.2860255241394043, + "learning_rate": 8.298439284354778e-07, + "loss": 1.4556, + "step": 2500 + }, + { + "epoch": 2.5680534155110424, + "eval_loss": 1.886796236038208, + "eval_runtime": 48.3725, + "eval_samples_per_second": 20.673, + "eval_steps_per_second": 0.868, + "step": 2500 + }, + { + "epoch": 2.569080636877247, + "grad_norm": 3.1930577754974365, + "learning_rate": 8.279406166730111e-07, + "loss": 1.4292, + "step": 2501 + }, + { + "epoch": 2.5701078582434516, + "grad_norm": 3.344346523284912, + "learning_rate": 8.260373049105444e-07, + "loss": 1.4317, + "step": 2502 + }, + { + "epoch": 2.5711350796096557, + "grad_norm": 3.1397411823272705, + "learning_rate": 8.241339931480777e-07, + "loss": 1.4064, + "step": 2503 + }, + { + "epoch": 2.5721623009758603, + "grad_norm": 3.1194756031036377, + "learning_rate": 8.22230681385611e-07, + "loss": 1.4517, + "step": 2504 + }, + { + "epoch": 2.573189522342065, + "grad_norm": 3.0981686115264893, + "learning_rate": 8.203273696231444e-07, + "loss": 1.4199, + "step": 2505 + }, + { + "epoch": 2.574216743708269, + "grad_norm": 3.1898980140686035, + "learning_rate": 8.184240578606777e-07, + "loss": 1.3732, + "step": 2506 + }, + { + "epoch": 2.5752439650744736, + "grad_norm": 3.166059732437134, + "learning_rate": 8.165207460982109e-07, + "loss": 1.4206, + "step": 2507 + }, + { + "epoch": 2.576271186440678, + "grad_norm": 3.1356523036956787, + "learning_rate": 8.146174343357442e-07, + "loss": 1.3839, + "step": 2508 + }, + { + "epoch": 2.5772984078068824, + "grad_norm": 3.2097487449645996, + "learning_rate": 8.127141225732775e-07, + "loss": 1.4226, + "step": 2509 + }, + { + "epoch": 2.578325629173087, + "grad_norm": 3.091526508331299, + "learning_rate": 8.108108108108109e-07, + "loss": 1.4282, + "step": 2510 + }, + { + "epoch": 2.578325629173087, + "eval_loss": 1.883396863937378, + "eval_runtime": 42.1441, + "eval_samples_per_second": 23.728, + "eval_steps_per_second": 0.997, + "step": 2510 + }, + { + "epoch": 2.579352850539291, + "grad_norm": 3.3042099475860596, + "learning_rate": 8.089074990483442e-07, + "loss": 1.4124, + "step": 2511 + }, + { + "epoch": 2.5803800719054957, + "grad_norm": 3.1862287521362305, + "learning_rate": 8.070041872858774e-07, + "loss": 1.4072, + "step": 2512 + }, + { + "epoch": 2.5814072932717, + "grad_norm": 3.2045605182647705, + "learning_rate": 8.051008755234108e-07, + "loss": 1.4188, + "step": 2513 + }, + { + "epoch": 2.5824345146379044, + "grad_norm": 3.3298919200897217, + "learning_rate": 8.031975637609441e-07, + "loss": 1.3571, + "step": 2514 + }, + { + "epoch": 2.583461736004109, + "grad_norm": 3.354160785675049, + "learning_rate": 8.012942519984774e-07, + "loss": 1.4135, + "step": 2515 + }, + { + "epoch": 2.584488957370313, + "grad_norm": 3.162343740463257, + "learning_rate": 7.993909402360108e-07, + "loss": 1.426, + "step": 2516 + }, + { + "epoch": 2.5855161787365177, + "grad_norm": 3.238799571990967, + "learning_rate": 7.97487628473544e-07, + "loss": 1.4207, + "step": 2517 + }, + { + "epoch": 2.5865434001027223, + "grad_norm": 3.31868839263916, + "learning_rate": 7.955843167110773e-07, + "loss": 1.4177, + "step": 2518 + }, + { + "epoch": 2.5875706214689265, + "grad_norm": 3.076633930206299, + "learning_rate": 7.936810049486106e-07, + "loss": 1.3662, + "step": 2519 + }, + { + "epoch": 2.588597842835131, + "grad_norm": 3.271270275115967, + "learning_rate": 7.91777693186144e-07, + "loss": 1.3868, + "step": 2520 + }, + { + "epoch": 2.588597842835131, + "eval_loss": 1.88602876663208, + "eval_runtime": 44.2042, + "eval_samples_per_second": 22.622, + "eval_steps_per_second": 0.95, + "step": 2520 + }, + { + "epoch": 2.5896250642013356, + "grad_norm": 3.1954116821289062, + "learning_rate": 7.898743814236773e-07, + "loss": 1.4183, + "step": 2521 + }, + { + "epoch": 2.59065228556754, + "grad_norm": 3.1537485122680664, + "learning_rate": 7.879710696612105e-07, + "loss": 1.3986, + "step": 2522 + }, + { + "epoch": 2.5916795069337444, + "grad_norm": 3.363866090774536, + "learning_rate": 7.860677578987438e-07, + "loss": 1.458, + "step": 2523 + }, + { + "epoch": 2.5927067282999485, + "grad_norm": 3.083444595336914, + "learning_rate": 7.841644461362772e-07, + "loss": 1.411, + "step": 2524 + }, + { + "epoch": 2.593733949666153, + "grad_norm": 3.1068875789642334, + "learning_rate": 7.822611343738105e-07, + "loss": 1.3865, + "step": 2525 + }, + { + "epoch": 2.5947611710323573, + "grad_norm": 3.2236549854278564, + "learning_rate": 7.803578226113438e-07, + "loss": 1.436, + "step": 2526 + }, + { + "epoch": 2.595788392398562, + "grad_norm": 3.3439364433288574, + "learning_rate": 7.784545108488771e-07, + "loss": 1.3893, + "step": 2527 + }, + { + "epoch": 2.5968156137647664, + "grad_norm": 3.0914306640625, + "learning_rate": 7.765511990864104e-07, + "loss": 1.4133, + "step": 2528 + }, + { + "epoch": 2.5978428351309706, + "grad_norm": 3.302943229675293, + "learning_rate": 7.746478873239437e-07, + "loss": 1.4607, + "step": 2529 + }, + { + "epoch": 2.598870056497175, + "grad_norm": 3.1647748947143555, + "learning_rate": 7.727445755614771e-07, + "loss": 1.409, + "step": 2530 + }, + { + "epoch": 2.598870056497175, + "eval_loss": 1.8868354558944702, + "eval_runtime": 43.4815, + "eval_samples_per_second": 22.998, + "eval_steps_per_second": 0.966, + "step": 2530 + }, + { + "epoch": 2.5998972778633798, + "grad_norm": 3.4687814712524414, + "learning_rate": 7.708412637990104e-07, + "loss": 1.4242, + "step": 2531 + }, + { + "epoch": 2.600924499229584, + "grad_norm": 3.2370734214782715, + "learning_rate": 7.689379520365436e-07, + "loss": 1.4234, + "step": 2532 + }, + { + "epoch": 2.6019517205957885, + "grad_norm": 3.1242740154266357, + "learning_rate": 7.670346402740769e-07, + "loss": 1.3895, + "step": 2533 + }, + { + "epoch": 2.602978941961993, + "grad_norm": 3.1856257915496826, + "learning_rate": 7.651313285116103e-07, + "loss": 1.4233, + "step": 2534 + }, + { + "epoch": 2.6040061633281972, + "grad_norm": 3.1666159629821777, + "learning_rate": 7.632280167491436e-07, + "loss": 1.4356, + "step": 2535 + }, + { + "epoch": 2.6050333846944014, + "grad_norm": 3.181180715560913, + "learning_rate": 7.613247049866769e-07, + "loss": 1.3741, + "step": 2536 + }, + { + "epoch": 2.606060606060606, + "grad_norm": 3.263272523880005, + "learning_rate": 7.594213932242102e-07, + "loss": 1.3484, + "step": 2537 + }, + { + "epoch": 2.6070878274268106, + "grad_norm": 3.1280739307403564, + "learning_rate": 7.575180814617435e-07, + "loss": 1.3781, + "step": 2538 + }, + { + "epoch": 2.6081150487930147, + "grad_norm": 3.1342055797576904, + "learning_rate": 7.556147696992767e-07, + "loss": 1.4089, + "step": 2539 + }, + { + "epoch": 2.6091422701592193, + "grad_norm": 3.1810309886932373, + "learning_rate": 7.537114579368102e-07, + "loss": 1.4095, + "step": 2540 + }, + { + "epoch": 2.6091422701592193, + "eval_loss": 1.8847311735153198, + "eval_runtime": 44.5386, + "eval_samples_per_second": 22.452, + "eval_steps_per_second": 0.943, + "step": 2540 + }, + { + "epoch": 2.610169491525424, + "grad_norm": 3.1380860805511475, + "learning_rate": 7.518081461743435e-07, + "loss": 1.3777, + "step": 2541 + }, + { + "epoch": 2.611196712891628, + "grad_norm": 3.141103982925415, + "learning_rate": 7.499048344118767e-07, + "loss": 1.3942, + "step": 2542 + }, + { + "epoch": 2.6122239342578326, + "grad_norm": 3.170095205307007, + "learning_rate": 7.4800152264941e-07, + "loss": 1.4267, + "step": 2543 + }, + { + "epoch": 2.613251155624037, + "grad_norm": 3.089294672012329, + "learning_rate": 7.460982108869432e-07, + "loss": 1.4008, + "step": 2544 + }, + { + "epoch": 2.6142783769902413, + "grad_norm": 3.211880922317505, + "learning_rate": 7.441948991244767e-07, + "loss": 1.3855, + "step": 2545 + }, + { + "epoch": 2.615305598356446, + "grad_norm": 3.350862741470337, + "learning_rate": 7.4229158736201e-07, + "loss": 1.4182, + "step": 2546 + }, + { + "epoch": 2.61633281972265, + "grad_norm": 3.2299282550811768, + "learning_rate": 7.403882755995432e-07, + "loss": 1.4363, + "step": 2547 + }, + { + "epoch": 2.6173600410888547, + "grad_norm": 3.04044508934021, + "learning_rate": 7.384849638370766e-07, + "loss": 1.3784, + "step": 2548 + }, + { + "epoch": 2.618387262455059, + "grad_norm": 3.066154718399048, + "learning_rate": 7.365816520746098e-07, + "loss": 1.3338, + "step": 2549 + }, + { + "epoch": 2.6194144838212634, + "grad_norm": 3.2050673961639404, + "learning_rate": 7.346783403121432e-07, + "loss": 1.4053, + "step": 2550 + }, + { + "epoch": 2.6194144838212634, + "eval_loss": 1.8817609548568726, + "eval_runtime": 45.452, + "eval_samples_per_second": 22.001, + "eval_steps_per_second": 0.924, + "step": 2550 + }, + { + "epoch": 2.620441705187468, + "grad_norm": 3.338341474533081, + "learning_rate": 7.327750285496766e-07, + "loss": 1.387, + "step": 2551 + }, + { + "epoch": 2.621468926553672, + "grad_norm": 3.3431243896484375, + "learning_rate": 7.308717167872098e-07, + "loss": 1.3971, + "step": 2552 + }, + { + "epoch": 2.6224961479198767, + "grad_norm": 3.0557432174682617, + "learning_rate": 7.289684050247431e-07, + "loss": 1.356, + "step": 2553 + }, + { + "epoch": 2.6235233692860813, + "grad_norm": 3.3249623775482178, + "learning_rate": 7.270650932622763e-07, + "loss": 1.4557, + "step": 2554 + }, + { + "epoch": 2.6245505906522855, + "grad_norm": 3.073441743850708, + "learning_rate": 7.251617814998098e-07, + "loss": 1.3434, + "step": 2555 + }, + { + "epoch": 2.62557781201849, + "grad_norm": 3.118184804916382, + "learning_rate": 7.232584697373431e-07, + "loss": 1.4243, + "step": 2556 + }, + { + "epoch": 2.6266050333846946, + "grad_norm": 3.0361053943634033, + "learning_rate": 7.213551579748763e-07, + "loss": 1.3718, + "step": 2557 + }, + { + "epoch": 2.627632254750899, + "grad_norm": 3.147110939025879, + "learning_rate": 7.194518462124096e-07, + "loss": 1.4078, + "step": 2558 + }, + { + "epoch": 2.6286594761171034, + "grad_norm": 3.3143343925476074, + "learning_rate": 7.175485344499429e-07, + "loss": 1.4862, + "step": 2559 + }, + { + "epoch": 2.6296866974833075, + "grad_norm": 3.27156662940979, + "learning_rate": 7.156452226874763e-07, + "loss": 1.3821, + "step": 2560 + }, + { + "epoch": 2.6296866974833075, + "eval_loss": 1.8816218376159668, + "eval_runtime": 45.9219, + "eval_samples_per_second": 21.776, + "eval_steps_per_second": 0.915, + "step": 2560 + }, + { + "epoch": 2.630713918849512, + "grad_norm": 3.228637933731079, + "learning_rate": 7.137419109250096e-07, + "loss": 1.4853, + "step": 2561 + }, + { + "epoch": 2.6317411402157163, + "grad_norm": 3.15281343460083, + "learning_rate": 7.118385991625429e-07, + "loss": 1.3904, + "step": 2562 + }, + { + "epoch": 2.632768361581921, + "grad_norm": 3.1659791469573975, + "learning_rate": 7.099352874000762e-07, + "loss": 1.3844, + "step": 2563 + }, + { + "epoch": 2.6337955829481254, + "grad_norm": 3.201091766357422, + "learning_rate": 7.080319756376094e-07, + "loss": 1.3931, + "step": 2564 + }, + { + "epoch": 2.6348228043143296, + "grad_norm": 3.1422455310821533, + "learning_rate": 7.061286638751429e-07, + "loss": 1.3694, + "step": 2565 + }, + { + "epoch": 2.635850025680534, + "grad_norm": 3.1560862064361572, + "learning_rate": 7.042253521126762e-07, + "loss": 1.402, + "step": 2566 + }, + { + "epoch": 2.6368772470467388, + "grad_norm": 3.2023425102233887, + "learning_rate": 7.023220403502094e-07, + "loss": 1.3594, + "step": 2567 + }, + { + "epoch": 2.637904468412943, + "grad_norm": 3.3929922580718994, + "learning_rate": 7.004187285877427e-07, + "loss": 1.4435, + "step": 2568 + }, + { + "epoch": 2.6389316897791475, + "grad_norm": 3.2756786346435547, + "learning_rate": 6.98515416825276e-07, + "loss": 1.4351, + "step": 2569 + }, + { + "epoch": 2.639958911145352, + "grad_norm": 3.148296594619751, + "learning_rate": 6.966121050628094e-07, + "loss": 1.3976, + "step": 2570 + }, + { + "epoch": 2.639958911145352, + "eval_loss": 1.8865966796875, + "eval_runtime": 43.2127, + "eval_samples_per_second": 23.141, + "eval_steps_per_second": 0.972, + "step": 2570 + }, + { + "epoch": 2.6409861325115562, + "grad_norm": 3.149265766143799, + "learning_rate": 6.947087933003427e-07, + "loss": 1.451, + "step": 2571 + }, + { + "epoch": 2.642013353877761, + "grad_norm": 3.1604983806610107, + "learning_rate": 6.92805481537876e-07, + "loss": 1.411, + "step": 2572 + }, + { + "epoch": 2.643040575243965, + "grad_norm": 3.1915225982666016, + "learning_rate": 6.909021697754093e-07, + "loss": 1.4526, + "step": 2573 + }, + { + "epoch": 2.6440677966101696, + "grad_norm": 3.0936203002929688, + "learning_rate": 6.889988580129425e-07, + "loss": 1.4229, + "step": 2574 + }, + { + "epoch": 2.6450950179763737, + "grad_norm": 3.2255401611328125, + "learning_rate": 6.870955462504759e-07, + "loss": 1.3829, + "step": 2575 + }, + { + "epoch": 2.6461222393425783, + "grad_norm": 3.3044967651367188, + "learning_rate": 6.851922344880093e-07, + "loss": 1.431, + "step": 2576 + }, + { + "epoch": 2.647149460708783, + "grad_norm": 3.1123087406158447, + "learning_rate": 6.832889227255425e-07, + "loss": 1.4137, + "step": 2577 + }, + { + "epoch": 2.648176682074987, + "grad_norm": 3.199956178665161, + "learning_rate": 6.813856109630758e-07, + "loss": 1.4252, + "step": 2578 + }, + { + "epoch": 2.6492039034411916, + "grad_norm": 3.1218457221984863, + "learning_rate": 6.79482299200609e-07, + "loss": 1.4155, + "step": 2579 + }, + { + "epoch": 2.650231124807396, + "grad_norm": 3.1815218925476074, + "learning_rate": 6.775789874381425e-07, + "loss": 1.3842, + "step": 2580 + }, + { + "epoch": 2.650231124807396, + "eval_loss": 1.8833224773406982, + "eval_runtime": 44.6273, + "eval_samples_per_second": 22.408, + "eval_steps_per_second": 0.941, + "step": 2580 + }, + { + "epoch": 2.6512583461736003, + "grad_norm": 3.3460185527801514, + "learning_rate": 6.756756756756758e-07, + "loss": 1.4572, + "step": 2581 + }, + { + "epoch": 2.652285567539805, + "grad_norm": 3.2174816131591797, + "learning_rate": 6.73772363913209e-07, + "loss": 1.436, + "step": 2582 + }, + { + "epoch": 2.653312788906009, + "grad_norm": 3.125154733657837, + "learning_rate": 6.718690521507424e-07, + "loss": 1.4199, + "step": 2583 + }, + { + "epoch": 2.6543400102722137, + "grad_norm": 3.1968681812286377, + "learning_rate": 6.699657403882756e-07, + "loss": 1.3772, + "step": 2584 + }, + { + "epoch": 2.655367231638418, + "grad_norm": 3.24725604057312, + "learning_rate": 6.68062428625809e-07, + "loss": 1.429, + "step": 2585 + }, + { + "epoch": 2.6563944530046224, + "grad_norm": 3.229875087738037, + "learning_rate": 6.661591168633424e-07, + "loss": 1.4417, + "step": 2586 + }, + { + "epoch": 2.657421674370827, + "grad_norm": 3.186762809753418, + "learning_rate": 6.642558051008756e-07, + "loss": 1.4265, + "step": 2587 + }, + { + "epoch": 2.658448895737031, + "grad_norm": 3.1768531799316406, + "learning_rate": 6.623524933384089e-07, + "loss": 1.4261, + "step": 2588 + }, + { + "epoch": 2.6594761171032357, + "grad_norm": 3.239741086959839, + "learning_rate": 6.604491815759421e-07, + "loss": 1.4194, + "step": 2589 + }, + { + "epoch": 2.6605033384694403, + "grad_norm": 3.1606059074401855, + "learning_rate": 6.585458698134756e-07, + "loss": 1.3795, + "step": 2590 + }, + { + "epoch": 2.6605033384694403, + "eval_loss": 1.881027102470398, + "eval_runtime": 43.3141, + "eval_samples_per_second": 23.087, + "eval_steps_per_second": 0.97, + "step": 2590 + }, + { + "epoch": 2.6615305598356445, + "grad_norm": 3.0860469341278076, + "learning_rate": 6.566425580510089e-07, + "loss": 1.3524, + "step": 2591 + }, + { + "epoch": 2.662557781201849, + "grad_norm": 3.168243408203125, + "learning_rate": 6.547392462885421e-07, + "loss": 1.4167, + "step": 2592 + }, + { + "epoch": 2.6635850025680536, + "grad_norm": 3.0896172523498535, + "learning_rate": 6.528359345260754e-07, + "loss": 1.4433, + "step": 2593 + }, + { + "epoch": 2.664612223934258, + "grad_norm": 3.0680644512176514, + "learning_rate": 6.509326227636087e-07, + "loss": 1.4016, + "step": 2594 + }, + { + "epoch": 2.6656394453004624, + "grad_norm": 3.0809788703918457, + "learning_rate": 6.490293110011421e-07, + "loss": 1.4101, + "step": 2595 + }, + { + "epoch": 2.6666666666666665, + "grad_norm": 3.1185302734375, + "learning_rate": 6.471259992386753e-07, + "loss": 1.4054, + "step": 2596 + }, + { + "epoch": 2.667693888032871, + "grad_norm": 3.0916998386383057, + "learning_rate": 6.452226874762087e-07, + "loss": 1.4549, + "step": 2597 + }, + { + "epoch": 2.6687211093990753, + "grad_norm": 3.1180596351623535, + "learning_rate": 6.43319375713742e-07, + "loss": 1.4029, + "step": 2598 + }, + { + "epoch": 2.66974833076528, + "grad_norm": 3.1555957794189453, + "learning_rate": 6.414160639512752e-07, + "loss": 1.4269, + "step": 2599 + }, + { + "epoch": 2.6707755521314844, + "grad_norm": 3.2724740505218506, + "learning_rate": 6.395127521888085e-07, + "loss": 1.4113, + "step": 2600 + }, + { + "epoch": 2.6707755521314844, + "eval_loss": 1.8863404989242554, + "eval_runtime": 43.3384, + "eval_samples_per_second": 23.074, + "eval_steps_per_second": 0.969, + "step": 2600 + }, + { + "epoch": 2.6718027734976886, + "grad_norm": 3.1361265182495117, + "learning_rate": 6.37609440426342e-07, + "loss": 1.3885, + "step": 2601 + }, + { + "epoch": 2.672829994863893, + "grad_norm": 3.2071163654327393, + "learning_rate": 6.357061286638752e-07, + "loss": 1.4656, + "step": 2602 + }, + { + "epoch": 2.6738572162300978, + "grad_norm": 3.2071850299835205, + "learning_rate": 6.338028169014085e-07, + "loss": 1.416, + "step": 2603 + }, + { + "epoch": 2.674884437596302, + "grad_norm": 3.256455421447754, + "learning_rate": 6.318995051389418e-07, + "loss": 1.4543, + "step": 2604 + }, + { + "epoch": 2.6759116589625065, + "grad_norm": 3.2594852447509766, + "learning_rate": 6.299961933764751e-07, + "loss": 1.3285, + "step": 2605 + }, + { + "epoch": 2.676938880328711, + "grad_norm": 3.18585467338562, + "learning_rate": 6.280928816140084e-07, + "loss": 1.4307, + "step": 2606 + }, + { + "epoch": 2.6779661016949152, + "grad_norm": 3.420093059539795, + "learning_rate": 6.261895698515417e-07, + "loss": 1.4935, + "step": 2607 + }, + { + "epoch": 2.67899332306112, + "grad_norm": 3.2765462398529053, + "learning_rate": 6.242862580890751e-07, + "loss": 1.4333, + "step": 2608 + }, + { + "epoch": 2.680020544427324, + "grad_norm": 3.171917676925659, + "learning_rate": 6.223829463266084e-07, + "loss": 1.3841, + "step": 2609 + }, + { + "epoch": 2.6810477657935285, + "grad_norm": 3.1052963733673096, + "learning_rate": 6.204796345641417e-07, + "loss": 1.3436, + "step": 2610 + }, + { + "epoch": 2.6810477657935285, + "eval_loss": 1.8882763385772705, + "eval_runtime": 42.4433, + "eval_samples_per_second": 23.561, + "eval_steps_per_second": 0.99, + "step": 2610 + }, + { + "epoch": 2.6820749871597327, + "grad_norm": 3.3243353366851807, + "learning_rate": 6.185763228016749e-07, + "loss": 1.4047, + "step": 2611 + }, + { + "epoch": 2.6831022085259373, + "grad_norm": 3.3923227787017822, + "learning_rate": 6.166730110392083e-07, + "loss": 1.4434, + "step": 2612 + }, + { + "epoch": 2.684129429892142, + "grad_norm": 3.2917444705963135, + "learning_rate": 6.147696992767416e-07, + "loss": 1.4588, + "step": 2613 + }, + { + "epoch": 2.685156651258346, + "grad_norm": 3.1572155952453613, + "learning_rate": 6.128663875142749e-07, + "loss": 1.4321, + "step": 2614 + }, + { + "epoch": 2.6861838726245506, + "grad_norm": 3.2125251293182373, + "learning_rate": 6.109630757518082e-07, + "loss": 1.411, + "step": 2615 + }, + { + "epoch": 2.687211093990755, + "grad_norm": 3.2413294315338135, + "learning_rate": 6.090597639893414e-07, + "loss": 1.4796, + "step": 2616 + }, + { + "epoch": 2.6882383153569593, + "grad_norm": 3.16619610786438, + "learning_rate": 6.071564522268748e-07, + "loss": 1.3741, + "step": 2617 + }, + { + "epoch": 2.689265536723164, + "grad_norm": 3.1315054893493652, + "learning_rate": 6.052531404644082e-07, + "loss": 1.4188, + "step": 2618 + }, + { + "epoch": 2.6902927580893685, + "grad_norm": 3.5836288928985596, + "learning_rate": 6.033498287019414e-07, + "loss": 1.5302, + "step": 2619 + }, + { + "epoch": 2.6913199794555727, + "grad_norm": 3.1081457138061523, + "learning_rate": 6.014465169394748e-07, + "loss": 1.3928, + "step": 2620 + }, + { + "epoch": 2.6913199794555727, + "eval_loss": 1.8860454559326172, + "eval_runtime": 44.7057, + "eval_samples_per_second": 22.368, + "eval_steps_per_second": 0.939, + "step": 2620 + }, + { + "epoch": 2.6923472008217773, + "grad_norm": 3.1687238216400146, + "learning_rate": 5.99543205177008e-07, + "loss": 1.4031, + "step": 2621 + }, + { + "epoch": 2.6933744221879814, + "grad_norm": 3.111706018447876, + "learning_rate": 5.976398934145414e-07, + "loss": 1.3888, + "step": 2622 + }, + { + "epoch": 2.694401643554186, + "grad_norm": 3.29876446723938, + "learning_rate": 5.957365816520747e-07, + "loss": 1.3879, + "step": 2623 + }, + { + "epoch": 2.69542886492039, + "grad_norm": 3.1588897705078125, + "learning_rate": 5.938332698896079e-07, + "loss": 1.4158, + "step": 2624 + }, + { + "epoch": 2.6964560862865947, + "grad_norm": 3.068836212158203, + "learning_rate": 5.919299581271413e-07, + "loss": 1.3904, + "step": 2625 + }, + { + "epoch": 2.6974833076527993, + "grad_norm": 3.184014081954956, + "learning_rate": 5.900266463646745e-07, + "loss": 1.4163, + "step": 2626 + }, + { + "epoch": 2.6985105290190035, + "grad_norm": 3.123582363128662, + "learning_rate": 5.881233346022079e-07, + "loss": 1.3922, + "step": 2627 + }, + { + "epoch": 2.699537750385208, + "grad_norm": 3.287302255630493, + "learning_rate": 5.862200228397411e-07, + "loss": 1.4418, + "step": 2628 + }, + { + "epoch": 2.7005649717514126, + "grad_norm": 3.357113838195801, + "learning_rate": 5.843167110772745e-07, + "loss": 1.4614, + "step": 2629 + }, + { + "epoch": 2.701592193117617, + "grad_norm": 3.2212133407592773, + "learning_rate": 5.824133993148079e-07, + "loss": 1.4174, + "step": 2630 + }, + { + "epoch": 2.701592193117617, + "eval_loss": 1.8899319171905518, + "eval_runtime": 46.2517, + "eval_samples_per_second": 21.621, + "eval_steps_per_second": 0.908, + "step": 2630 + }, + { + "epoch": 2.7026194144838214, + "grad_norm": 3.1398634910583496, + "learning_rate": 5.805100875523411e-07, + "loss": 1.3816, + "step": 2631 + }, + { + "epoch": 2.7036466358500255, + "grad_norm": 3.1249663829803467, + "learning_rate": 5.786067757898745e-07, + "loss": 1.3558, + "step": 2632 + }, + { + "epoch": 2.70467385721623, + "grad_norm": 3.1791367530822754, + "learning_rate": 5.767034640274078e-07, + "loss": 1.3876, + "step": 2633 + }, + { + "epoch": 2.7057010785824342, + "grad_norm": 3.312366485595703, + "learning_rate": 5.74800152264941e-07, + "loss": 1.4819, + "step": 2634 + }, + { + "epoch": 2.706728299948639, + "grad_norm": 3.259929656982422, + "learning_rate": 5.728968405024744e-07, + "loss": 1.374, + "step": 2635 + }, + { + "epoch": 2.7077555213148434, + "grad_norm": 3.272461414337158, + "learning_rate": 5.709935287400076e-07, + "loss": 1.4565, + "step": 2636 + }, + { + "epoch": 2.7087827426810476, + "grad_norm": 3.2284653186798096, + "learning_rate": 5.69090216977541e-07, + "loss": 1.4273, + "step": 2637 + }, + { + "epoch": 2.709809964047252, + "grad_norm": 3.1966605186462402, + "learning_rate": 5.671869052150742e-07, + "loss": 1.3866, + "step": 2638 + }, + { + "epoch": 2.7108371854134568, + "grad_norm": 3.1381869316101074, + "learning_rate": 5.652835934526075e-07, + "loss": 1.393, + "step": 2639 + }, + { + "epoch": 2.711864406779661, + "grad_norm": 3.2234206199645996, + "learning_rate": 5.633802816901409e-07, + "loss": 1.4028, + "step": 2640 + }, + { + "epoch": 2.711864406779661, + "eval_loss": 1.8904504776000977, + "eval_runtime": 46.7218, + "eval_samples_per_second": 21.403, + "eval_steps_per_second": 0.899, + "step": 2640 + }, + { + "epoch": 2.7128916281458655, + "grad_norm": 3.1707582473754883, + "learning_rate": 5.614769699276742e-07, + "loss": 1.387, + "step": 2641 + }, + { + "epoch": 2.71391884951207, + "grad_norm": 3.2514328956604004, + "learning_rate": 5.595736581652075e-07, + "loss": 1.4018, + "step": 2642 + }, + { + "epoch": 2.714946070878274, + "grad_norm": 3.07417631149292, + "learning_rate": 5.576703464027408e-07, + "loss": 1.3989, + "step": 2643 + }, + { + "epoch": 2.715973292244479, + "grad_norm": 3.29019832611084, + "learning_rate": 5.557670346402741e-07, + "loss": 1.4312, + "step": 2644 + }, + { + "epoch": 2.717000513610683, + "grad_norm": 3.2749898433685303, + "learning_rate": 5.538637228778075e-07, + "loss": 1.4063, + "step": 2645 + }, + { + "epoch": 2.7180277349768875, + "grad_norm": 3.1695713996887207, + "learning_rate": 5.519604111153407e-07, + "loss": 1.4396, + "step": 2646 + }, + { + "epoch": 2.7190549563430917, + "grad_norm": 3.1040000915527344, + "learning_rate": 5.500570993528741e-07, + "loss": 1.3656, + "step": 2647 + }, + { + "epoch": 2.7200821777092963, + "grad_norm": 3.1661036014556885, + "learning_rate": 5.481537875904073e-07, + "loss": 1.4545, + "step": 2648 + }, + { + "epoch": 2.721109399075501, + "grad_norm": 3.2728803157806396, + "learning_rate": 5.462504758279406e-07, + "loss": 1.4743, + "step": 2649 + }, + { + "epoch": 2.722136620441705, + "grad_norm": 3.124732732772827, + "learning_rate": 5.44347164065474e-07, + "loss": 1.3938, + "step": 2650 + }, + { + "epoch": 2.722136620441705, + "eval_loss": 1.8866426944732666, + "eval_runtime": 51.4642, + "eval_samples_per_second": 19.431, + "eval_steps_per_second": 0.816, + "step": 2650 + }, + { + "epoch": 2.7231638418079096, + "grad_norm": 3.113888740539551, + "learning_rate": 5.424438523030072e-07, + "loss": 1.4075, + "step": 2651 + }, + { + "epoch": 2.724191063174114, + "grad_norm": 3.2134811878204346, + "learning_rate": 5.405405405405406e-07, + "loss": 1.4673, + "step": 2652 + }, + { + "epoch": 2.7252182845403183, + "grad_norm": 3.100324869155884, + "learning_rate": 5.386372287780739e-07, + "loss": 1.427, + "step": 2653 + }, + { + "epoch": 2.726245505906523, + "grad_norm": 3.0996758937835693, + "learning_rate": 5.367339170156072e-07, + "loss": 1.4449, + "step": 2654 + }, + { + "epoch": 2.7272727272727275, + "grad_norm": 3.247462749481201, + "learning_rate": 5.348306052531406e-07, + "loss": 1.4031, + "step": 2655 + }, + { + "epoch": 2.7282999486389317, + "grad_norm": 3.2123465538024902, + "learning_rate": 5.329272934906738e-07, + "loss": 1.4347, + "step": 2656 + }, + { + "epoch": 2.7293271700051362, + "grad_norm": 3.0907955169677734, + "learning_rate": 5.310239817282072e-07, + "loss": 1.3706, + "step": 2657 + }, + { + "epoch": 2.7303543913713404, + "grad_norm": 3.341423988342285, + "learning_rate": 5.291206699657404e-07, + "loss": 1.4589, + "step": 2658 + }, + { + "epoch": 2.731381612737545, + "grad_norm": 3.167069435119629, + "learning_rate": 5.272173582032737e-07, + "loss": 1.398, + "step": 2659 + }, + { + "epoch": 2.732408834103749, + "grad_norm": 3.2320168018341064, + "learning_rate": 5.253140464408071e-07, + "loss": 1.4048, + "step": 2660 + }, + { + "epoch": 2.732408834103749, + "eval_loss": 1.8844244480133057, + "eval_runtime": 49.2994, + "eval_samples_per_second": 20.284, + "eval_steps_per_second": 0.852, + "step": 2660 + }, + { + "epoch": 2.7334360554699537, + "grad_norm": 3.1752889156341553, + "learning_rate": 5.234107346783403e-07, + "loss": 1.3898, + "step": 2661 + }, + { + "epoch": 2.7344632768361583, + "grad_norm": 3.2391722202301025, + "learning_rate": 5.215074229158737e-07, + "loss": 1.3757, + "step": 2662 + }, + { + "epoch": 2.7354904982023625, + "grad_norm": 3.2485735416412354, + "learning_rate": 5.196041111534069e-07, + "loss": 1.3953, + "step": 2663 + }, + { + "epoch": 2.736517719568567, + "grad_norm": 3.242967367172241, + "learning_rate": 5.177007993909403e-07, + "loss": 1.4104, + "step": 2664 + }, + { + "epoch": 2.7375449409347716, + "grad_norm": 3.2751710414886475, + "learning_rate": 5.157974876284736e-07, + "loss": 1.4286, + "step": 2665 + }, + { + "epoch": 2.7385721623009758, + "grad_norm": 3.0682404041290283, + "learning_rate": 5.138941758660069e-07, + "loss": 1.4188, + "step": 2666 + }, + { + "epoch": 2.7395993836671804, + "grad_norm": 3.1645498275756836, + "learning_rate": 5.119908641035403e-07, + "loss": 1.4001, + "step": 2667 + }, + { + "epoch": 2.740626605033385, + "grad_norm": 3.130505323410034, + "learning_rate": 5.100875523410735e-07, + "loss": 1.3692, + "step": 2668 + }, + { + "epoch": 2.741653826399589, + "grad_norm": 3.0741729736328125, + "learning_rate": 5.081842405786068e-07, + "loss": 1.3776, + "step": 2669 + }, + { + "epoch": 2.7426810477657937, + "grad_norm": 3.2680647373199463, + "learning_rate": 5.062809288161402e-07, + "loss": 1.4062, + "step": 2670 + }, + { + "epoch": 2.7426810477657937, + "eval_loss": 1.8841313123703003, + "eval_runtime": 48.6426, + "eval_samples_per_second": 20.558, + "eval_steps_per_second": 0.863, + "step": 2670 + }, + { + "epoch": 2.743708269131998, + "grad_norm": 3.240175485610962, + "learning_rate": 5.043776170536734e-07, + "loss": 1.4245, + "step": 2671 + }, + { + "epoch": 2.7447354904982024, + "grad_norm": 3.2721028327941895, + "learning_rate": 5.024743052912067e-07, + "loss": 1.4405, + "step": 2672 + }, + { + "epoch": 2.7457627118644066, + "grad_norm": 3.1048996448516846, + "learning_rate": 5.0057099352874e-07, + "loss": 1.4201, + "step": 2673 + }, + { + "epoch": 2.746789933230611, + "grad_norm": 3.232975959777832, + "learning_rate": 4.986676817662733e-07, + "loss": 1.4438, + "step": 2674 + }, + { + "epoch": 2.7478171545968157, + "grad_norm": 3.1347835063934326, + "learning_rate": 4.967643700038067e-07, + "loss": 1.4133, + "step": 2675 + }, + { + "epoch": 2.74884437596302, + "grad_norm": 3.240530014038086, + "learning_rate": 4.9486105824134e-07, + "loss": 1.432, + "step": 2676 + }, + { + "epoch": 2.7498715973292245, + "grad_norm": 3.234736919403076, + "learning_rate": 4.929577464788733e-07, + "loss": 1.4611, + "step": 2677 + }, + { + "epoch": 2.750898818695429, + "grad_norm": 3.24591064453125, + "learning_rate": 4.910544347164066e-07, + "loss": 1.4343, + "step": 2678 + }, + { + "epoch": 2.751926040061633, + "grad_norm": 3.1365740299224854, + "learning_rate": 4.891511229539399e-07, + "loss": 1.3578, + "step": 2679 + }, + { + "epoch": 2.752953261427838, + "grad_norm": 3.1024372577667236, + "learning_rate": 4.872478111914733e-07, + "loss": 1.4216, + "step": 2680 + }, + { + "epoch": 2.752953261427838, + "eval_loss": 1.8822519779205322, + "eval_runtime": 51.5985, + "eval_samples_per_second": 19.38, + "eval_steps_per_second": 0.814, + "step": 2680 + }, + { + "epoch": 2.753980482794042, + "grad_norm": 3.1567726135253906, + "learning_rate": 4.853444994290065e-07, + "loss": 1.3821, + "step": 2681 + }, + { + "epoch": 2.7550077041602465, + "grad_norm": 3.2141427993774414, + "learning_rate": 4.834411876665398e-07, + "loss": 1.4061, + "step": 2682 + }, + { + "epoch": 2.7560349255264507, + "grad_norm": 3.19643235206604, + "learning_rate": 4.815378759040731e-07, + "loss": 1.3828, + "step": 2683 + }, + { + "epoch": 2.7570621468926553, + "grad_norm": 3.276965618133545, + "learning_rate": 4.796345641416064e-07, + "loss": 1.4314, + "step": 2684 + }, + { + "epoch": 2.75808936825886, + "grad_norm": 3.077977418899536, + "learning_rate": 4.777312523791397e-07, + "loss": 1.3956, + "step": 2685 + }, + { + "epoch": 2.759116589625064, + "grad_norm": 3.151568651199341, + "learning_rate": 4.7582794061667306e-07, + "loss": 1.4186, + "step": 2686 + }, + { + "epoch": 2.7601438109912686, + "grad_norm": 3.2300612926483154, + "learning_rate": 4.739246288542063e-07, + "loss": 1.4284, + "step": 2687 + }, + { + "epoch": 2.761171032357473, + "grad_norm": 3.2608187198638916, + "learning_rate": 4.720213170917397e-07, + "loss": 1.4125, + "step": 2688 + }, + { + "epoch": 2.7621982537236773, + "grad_norm": 3.3204030990600586, + "learning_rate": 4.7011800532927293e-07, + "loss": 1.4771, + "step": 2689 + }, + { + "epoch": 2.763225475089882, + "grad_norm": 3.1368515491485596, + "learning_rate": 4.682146935668063e-07, + "loss": 1.3892, + "step": 2690 + }, + { + "epoch": 2.763225475089882, + "eval_loss": 1.8787988424301147, + "eval_runtime": 50.2559, + "eval_samples_per_second": 19.898, + "eval_steps_per_second": 0.836, + "step": 2690 + }, + { + "epoch": 2.7642526964560865, + "grad_norm": 3.166334390640259, + "learning_rate": 4.663113818043396e-07, + "loss": 1.4253, + "step": 2691 + }, + { + "epoch": 2.7652799178222907, + "grad_norm": 3.2137794494628906, + "learning_rate": 4.6440807004187286e-07, + "loss": 1.4191, + "step": 2692 + }, + { + "epoch": 2.7663071391884952, + "grad_norm": 3.1977455615997314, + "learning_rate": 4.625047582794062e-07, + "loss": 1.4164, + "step": 2693 + }, + { + "epoch": 2.7673343605546994, + "grad_norm": 3.184385061264038, + "learning_rate": 4.606014465169395e-07, + "loss": 1.3935, + "step": 2694 + }, + { + "epoch": 2.768361581920904, + "grad_norm": 3.205763816833496, + "learning_rate": 4.5869813475447284e-07, + "loss": 1.4069, + "step": 2695 + }, + { + "epoch": 2.769388803287108, + "grad_norm": 3.0484068393707275, + "learning_rate": 4.5679482299200615e-07, + "loss": 1.4166, + "step": 2696 + }, + { + "epoch": 2.7704160246533127, + "grad_norm": 3.2038793563842773, + "learning_rate": 4.548915112295394e-07, + "loss": 1.4172, + "step": 2697 + }, + { + "epoch": 2.7714432460195173, + "grad_norm": 3.012335777282715, + "learning_rate": 4.529881994670728e-07, + "loss": 1.3436, + "step": 2698 + }, + { + "epoch": 2.7724704673857214, + "grad_norm": 3.259434700012207, + "learning_rate": 4.5108488770460603e-07, + "loss": 1.4123, + "step": 2699 + }, + { + "epoch": 2.773497688751926, + "grad_norm": 3.145287275314331, + "learning_rate": 4.491815759421394e-07, + "loss": 1.4153, + "step": 2700 + }, + { + "epoch": 2.773497688751926, + "eval_loss": 1.8797712326049805, + "eval_runtime": 46.756, + "eval_samples_per_second": 21.388, + "eval_steps_per_second": 0.898, + "step": 2700 + } + ], + "logging_steps": 1, + "max_steps": 2919, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 100, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 8.416735405277184e+17, + "train_batch_size": 3, + "trial_name": null, + "trial_params": null +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/zero_to_fp32.py b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-2700/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fb0f9a53a507ffcdb60410deb71d60ef801bf350 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/config.json @@ -0,0 +1,36 @@ +{ + "_name_or_path": "meta-llama/Llama-3.2-3B", + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "eos_token_id": 128001, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 3072, + "initializer_range": 0.02, + "intermediate_size": 8192, + "max_position_embeddings": 131072, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 24, + "num_hidden_layers": 28, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": { + "factor": 32.0, + "high_freq_factor": 4.0, + "low_freq_factor": 1.0, + "original_max_position_embeddings": 8192, + "rope_type": "llama3" + }, + "rope_theta": 500000.0, + "tie_word_embeddings": true, + "torch_dtype": "float16", + "transformers_version": "4.45.1", + "use_cache": true, + "vocab_size": 128256 +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/generation_config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..9c389b841a9a29ddd904e02d1eb0e08dcce82ad9 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/generation_config.json @@ -0,0 +1,9 @@ +{ + "_from_model_config": true, + "bos_token_id": 128000, + "do_sample": true, + "eos_token_id": 128001, + "temperature": 0.6, + "top_p": 0.9, + "transformers_version": "4.45.1" +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/latest b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/latest new file mode 100644 index 0000000000000000000000000000000000000000..57729c0be88118cbd582c8c68b4149cee821f0b4 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/latest @@ -0,0 +1 @@ +global_step800 \ No newline at end of file diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/model.safetensors.index.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..ed64de846d720b9a7859dc20575fea8e8ca51940 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/model.safetensors.index.json @@ -0,0 +1,262 @@ +{ + "metadata": { + "total_size": 7213504512 + }, + "weight_map": { + "lm_head.weight": "model-00002-of-00002.safetensors", + "model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.norm.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/special_tokens_map.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..6949c5975ee0e961ef61cf31010dce04df0a03f8 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/special_tokens_map.json @@ -0,0 +1,23 @@ +{ + "bos_token": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/tokenizer_config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..0b782d4602b361f7577abb52b710a8c0e088588d --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/tokenizer_config.json @@ -0,0 +1,2078 @@ +{ + "added_tokens_decoder": { + "128000": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128001": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128002": { + "content": "<|reserved_special_token_0|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128003": { + "content": "<|reserved_special_token_1|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128004": { + "content": "<|finetune_right_pad_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128005": { + "content": "<|reserved_special_token_2|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128006": { + "content": "<|start_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128007": { + "content": "<|end_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128008": { + "content": "<|eom_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128009": { + "content": "<|eot_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128010": { + "content": "<|python_tag|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128011": { + "content": "<|reserved_special_token_3|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128012": { + "content": "<|reserved_special_token_4|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128013": { + "content": "<|reserved_special_token_5|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128014": { + "content": "<|reserved_special_token_6|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128015": { + "content": "<|reserved_special_token_7|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128016": { + "content": "<|reserved_special_token_8|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128017": { + "content": "<|reserved_special_token_9|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128018": { + "content": "<|reserved_special_token_10|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128019": { + "content": "<|reserved_special_token_11|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128020": { + "content": "<|reserved_special_token_12|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128021": { + "content": "<|reserved_special_token_13|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128022": { + "content": "<|reserved_special_token_14|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128023": { + "content": "<|reserved_special_token_15|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128024": { + "content": "<|reserved_special_token_16|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128025": { + "content": "<|reserved_special_token_17|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128026": { + "content": "<|reserved_special_token_18|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128027": { + "content": "<|reserved_special_token_19|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128028": { + "content": "<|reserved_special_token_20|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128029": { + "content": "<|reserved_special_token_21|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128030": { + "content": "<|reserved_special_token_22|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128031": { + "content": "<|reserved_special_token_23|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128032": { + "content": "<|reserved_special_token_24|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128033": { + "content": "<|reserved_special_token_25|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128034": { + "content": "<|reserved_special_token_26|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128035": { + "content": "<|reserved_special_token_27|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128036": { + "content": "<|reserved_special_token_28|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128037": { + "content": "<|reserved_special_token_29|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128038": { + "content": "<|reserved_special_token_30|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128039": { + "content": "<|reserved_special_token_31|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128040": { + "content": "<|reserved_special_token_32|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128041": { + "content": "<|reserved_special_token_33|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128042": { + "content": "<|reserved_special_token_34|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128043": { + "content": "<|reserved_special_token_35|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128044": { + "content": "<|reserved_special_token_36|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128045": { + "content": "<|reserved_special_token_37|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128046": { + "content": "<|reserved_special_token_38|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128047": { + "content": "<|reserved_special_token_39|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128048": { + "content": "<|reserved_special_token_40|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128049": { + "content": "<|reserved_special_token_41|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128050": { + "content": "<|reserved_special_token_42|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128051": { + "content": "<|reserved_special_token_43|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128052": { + "content": "<|reserved_special_token_44|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128053": { + "content": "<|reserved_special_token_45|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128054": { + "content": "<|reserved_special_token_46|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128055": { + "content": "<|reserved_special_token_47|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128056": { + "content": "<|reserved_special_token_48|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128057": { + "content": "<|reserved_special_token_49|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128058": { + "content": "<|reserved_special_token_50|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128059": { + "content": "<|reserved_special_token_51|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128060": { + "content": "<|reserved_special_token_52|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128061": { + "content": "<|reserved_special_token_53|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128062": { + "content": "<|reserved_special_token_54|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128063": { + "content": "<|reserved_special_token_55|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128064": { + "content": "<|reserved_special_token_56|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128065": { + "content": "<|reserved_special_token_57|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128066": { + "content": "<|reserved_special_token_58|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128067": { + "content": "<|reserved_special_token_59|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128068": { + "content": "<|reserved_special_token_60|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128069": { + "content": "<|reserved_special_token_61|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128070": { + "content": "<|reserved_special_token_62|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128071": { + "content": "<|reserved_special_token_63|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128072": { + "content": "<|reserved_special_token_64|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128073": { + "content": "<|reserved_special_token_65|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128074": { + "content": "<|reserved_special_token_66|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128075": { + "content": "<|reserved_special_token_67|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128076": { + "content": "<|reserved_special_token_68|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128077": { + "content": "<|reserved_special_token_69|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128078": { + "content": "<|reserved_special_token_70|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128079": { + "content": "<|reserved_special_token_71|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128080": { + "content": "<|reserved_special_token_72|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128081": { + "content": "<|reserved_special_token_73|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128082": { + "content": "<|reserved_special_token_74|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128083": { + "content": "<|reserved_special_token_75|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128084": { + "content": "<|reserved_special_token_76|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128085": { + "content": "<|reserved_special_token_77|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128086": { + "content": "<|reserved_special_token_78|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128087": { + "content": "<|reserved_special_token_79|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128088": { + "content": "<|reserved_special_token_80|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128089": { + "content": "<|reserved_special_token_81|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128090": { + "content": "<|reserved_special_token_82|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128091": { + "content": "<|reserved_special_token_83|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128092": { + "content": "<|reserved_special_token_84|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128093": { + "content": "<|reserved_special_token_85|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128094": { + "content": "<|reserved_special_token_86|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128095": { + "content": "<|reserved_special_token_87|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128096": { + "content": "<|reserved_special_token_88|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128097": { + "content": "<|reserved_special_token_89|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128098": { + "content": "<|reserved_special_token_90|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128099": { + "content": "<|reserved_special_token_91|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128100": { + "content": "<|reserved_special_token_92|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128101": { + "content": "<|reserved_special_token_93|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128102": { + "content": "<|reserved_special_token_94|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128103": { + "content": "<|reserved_special_token_95|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128104": { + "content": "<|reserved_special_token_96|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128105": { + "content": "<|reserved_special_token_97|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128106": { + "content": "<|reserved_special_token_98|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128107": { + "content": "<|reserved_special_token_99|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128108": { + "content": "<|reserved_special_token_100|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128109": { + "content": "<|reserved_special_token_101|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128110": { + "content": "<|reserved_special_token_102|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128111": { + "content": "<|reserved_special_token_103|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128112": { + "content": "<|reserved_special_token_104|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128113": { + "content": "<|reserved_special_token_105|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128114": { + "content": "<|reserved_special_token_106|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128115": { + "content": "<|reserved_special_token_107|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128116": { + "content": "<|reserved_special_token_108|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128117": { + "content": "<|reserved_special_token_109|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128118": { + "content": "<|reserved_special_token_110|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128119": { + "content": "<|reserved_special_token_111|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128120": { + "content": "<|reserved_special_token_112|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128121": { + "content": "<|reserved_special_token_113|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128122": { + "content": "<|reserved_special_token_114|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128123": { + "content": "<|reserved_special_token_115|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128124": { + "content": "<|reserved_special_token_116|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128125": { + "content": "<|reserved_special_token_117|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128126": { + "content": "<|reserved_special_token_118|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128127": { + "content": "<|reserved_special_token_119|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128128": { + "content": "<|reserved_special_token_120|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128129": { + "content": "<|reserved_special_token_121|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128130": { + "content": "<|reserved_special_token_122|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128131": { + "content": "<|reserved_special_token_123|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128132": { + "content": "<|reserved_special_token_124|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128133": { + "content": "<|reserved_special_token_125|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128134": { + "content": "<|reserved_special_token_126|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128135": { + "content": "<|reserved_special_token_127|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128136": { + "content": "<|reserved_special_token_128|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128137": { + "content": "<|reserved_special_token_129|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128138": { + "content": "<|reserved_special_token_130|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128139": { + "content": "<|reserved_special_token_131|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128140": { + "content": "<|reserved_special_token_132|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128141": { + "content": "<|reserved_special_token_133|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128142": { + "content": "<|reserved_special_token_134|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128143": { + "content": "<|reserved_special_token_135|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128144": { + "content": "<|reserved_special_token_136|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128145": { + "content": "<|reserved_special_token_137|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128146": { + "content": "<|reserved_special_token_138|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128147": { + "content": "<|reserved_special_token_139|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128148": { + "content": "<|reserved_special_token_140|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128149": { + "content": "<|reserved_special_token_141|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128150": { + "content": "<|reserved_special_token_142|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128151": { + "content": "<|reserved_special_token_143|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128152": { + "content": "<|reserved_special_token_144|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128153": { + "content": "<|reserved_special_token_145|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128154": { + "content": "<|reserved_special_token_146|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128155": { + "content": "<|reserved_special_token_147|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128156": { + "content": "<|reserved_special_token_148|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128157": { + "content": "<|reserved_special_token_149|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128158": { + "content": "<|reserved_special_token_150|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128159": { + "content": "<|reserved_special_token_151|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128160": { + "content": "<|reserved_special_token_152|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128161": { + "content": "<|reserved_special_token_153|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128162": { + "content": "<|reserved_special_token_154|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128163": { + "content": "<|reserved_special_token_155|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128164": { + "content": "<|reserved_special_token_156|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128165": { + "content": "<|reserved_special_token_157|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128166": { + "content": "<|reserved_special_token_158|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128167": { + "content": "<|reserved_special_token_159|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128168": { + "content": "<|reserved_special_token_160|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128169": { + "content": "<|reserved_special_token_161|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128170": { + "content": "<|reserved_special_token_162|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128171": { + "content": "<|reserved_special_token_163|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128172": { + "content": "<|reserved_special_token_164|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128173": { + "content": "<|reserved_special_token_165|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128174": { + "content": "<|reserved_special_token_166|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128175": { + "content": "<|reserved_special_token_167|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128176": { + "content": "<|reserved_special_token_168|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128177": { + "content": "<|reserved_special_token_169|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128178": { + "content": "<|reserved_special_token_170|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128179": { + "content": "<|reserved_special_token_171|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128180": { + "content": "<|reserved_special_token_172|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128181": { + "content": "<|reserved_special_token_173|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128182": { + "content": "<|reserved_special_token_174|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128183": { + "content": "<|reserved_special_token_175|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128184": { + "content": "<|reserved_special_token_176|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128185": { + "content": "<|reserved_special_token_177|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128186": { + "content": "<|reserved_special_token_178|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128187": { + "content": "<|reserved_special_token_179|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128188": { + "content": "<|reserved_special_token_180|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128189": { + "content": "<|reserved_special_token_181|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128190": { + "content": "<|reserved_special_token_182|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128191": { + "content": "<|reserved_special_token_183|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128192": { + "content": "<|reserved_special_token_184|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128193": { + "content": "<|reserved_special_token_185|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128194": { + "content": "<|reserved_special_token_186|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128195": { + "content": "<|reserved_special_token_187|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128196": { + "content": "<|reserved_special_token_188|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128197": { + "content": "<|reserved_special_token_189|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128198": { + "content": "<|reserved_special_token_190|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128199": { + "content": "<|reserved_special_token_191|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128200": { + "content": "<|reserved_special_token_192|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128201": { + "content": "<|reserved_special_token_193|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128202": { + "content": "<|reserved_special_token_194|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128203": { + "content": "<|reserved_special_token_195|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128204": { + "content": "<|reserved_special_token_196|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128205": { + "content": "<|reserved_special_token_197|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128206": { + "content": "<|reserved_special_token_198|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128207": { + "content": "<|reserved_special_token_199|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128208": { + "content": "<|reserved_special_token_200|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128209": { + "content": "<|reserved_special_token_201|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128210": { + "content": "<|reserved_special_token_202|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128211": { + "content": "<|reserved_special_token_203|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128212": { + "content": "<|reserved_special_token_204|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128213": { + "content": "<|reserved_special_token_205|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128214": { + "content": "<|reserved_special_token_206|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128215": { + "content": "<|reserved_special_token_207|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128216": { + "content": "<|reserved_special_token_208|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128217": { + "content": "<|reserved_special_token_209|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128218": { + "content": "<|reserved_special_token_210|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128219": { + "content": "<|reserved_special_token_211|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128220": { + "content": "<|reserved_special_token_212|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128221": { + "content": "<|reserved_special_token_213|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128222": { + "content": "<|reserved_special_token_214|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128223": { + "content": "<|reserved_special_token_215|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128224": { + "content": "<|reserved_special_token_216|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128225": { + "content": "<|reserved_special_token_217|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128226": { + "content": "<|reserved_special_token_218|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128227": { + "content": "<|reserved_special_token_219|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128228": { + "content": "<|reserved_special_token_220|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128229": { + "content": "<|reserved_special_token_221|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128230": { + "content": "<|reserved_special_token_222|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128231": { + "content": "<|reserved_special_token_223|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128232": { + "content": "<|reserved_special_token_224|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128233": { + "content": "<|reserved_special_token_225|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128234": { + "content": "<|reserved_special_token_226|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128235": { + "content": "<|reserved_special_token_227|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128236": { + "content": "<|reserved_special_token_228|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128237": { + "content": "<|reserved_special_token_229|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128238": { + "content": "<|reserved_special_token_230|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128239": { + "content": "<|reserved_special_token_231|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128240": { + "content": "<|reserved_special_token_232|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128241": { + "content": "<|reserved_special_token_233|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128242": { + "content": "<|reserved_special_token_234|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128243": { + "content": "<|reserved_special_token_235|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128244": { + "content": "<|reserved_special_token_236|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128245": { + "content": "<|reserved_special_token_237|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128246": { + "content": "<|reserved_special_token_238|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128247": { + "content": "<|reserved_special_token_239|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128248": { + "content": "<|reserved_special_token_240|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128249": { + "content": "<|reserved_special_token_241|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128250": { + "content": "<|reserved_special_token_242|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128251": { + "content": "<|reserved_special_token_243|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128252": { + "content": "<|reserved_special_token_244|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128253": { + "content": "<|reserved_special_token_245|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128254": { + "content": "<|reserved_special_token_246|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128255": { + "content": "<|reserved_special_token_247|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128256": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128257": { + "content": "🅁", + "lstrip": true, + "normalized": true, + "rstrip": false, + "single_word": false, + "special": false + } + }, + "bos_token": "<|begin_of_text|>", + "clean_up_tokenization_spaces": true, + "eos_token": "<|end_of_text|>", + "model_input_names": [ + "input_ids", + "attention_mask" + ], + "model_max_length": 131072, + "pad_token": "[PAD]", + "tokenizer_class": "PreTrainedTokenizerFast" +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/trainer_state.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..c5529528dc6c8f67400f64c0ea94452f4b2f9f66 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/trainer_state.json @@ -0,0 +1,6273 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.8217770929635336, + "eval_steps": 10, + "global_step": 800, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.001027221366204417, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7873, + "step": 1 + }, + { + "epoch": 0.002054442732408834, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7904, + "step": 2 + }, + { + "epoch": 0.0030816640986132513, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8348, + "step": 3 + }, + { + "epoch": 0.004108885464817668, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8357, + "step": 4 + }, + { + "epoch": 0.005136106831022085, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8153, + "step": 5 + }, + { + "epoch": 0.0061633281972265025, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.825, + "step": 6 + }, + { + "epoch": 0.007190549563430919, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8002, + "step": 7 + }, + { + "epoch": 0.008217770929635337, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8413, + "step": 8 + }, + { + "epoch": 0.009244992295839754, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8157, + "step": 9 + }, + { + "epoch": 0.01027221366204417, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8249, + "step": 10 + }, + { + "epoch": 0.01027221366204417, + "eval_loss": 2.8518412113189697, + "eval_runtime": 43.212, + "eval_samples_per_second": 23.142, + "eval_steps_per_second": 0.972, + "step": 10 + }, + { + "epoch": 0.011299435028248588, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7829, + "step": 11 + }, + { + "epoch": 0.012326656394453005, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7819, + "step": 12 + }, + { + "epoch": 0.01335387776065742, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8355, + "step": 13 + }, + { + "epoch": 0.014381099126861838, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8138, + "step": 14 + }, + { + "epoch": 0.015408320493066256, + "grad_norm": 5.162191390991211, + "learning_rate": 1.7123287671232876e-08, + "loss": 2.8088, + "step": 15 + }, + { + "epoch": 0.016435541859270673, + "grad_norm": 5.5954084396362305, + "learning_rate": 3.424657534246575e-08, + "loss": 2.7883, + "step": 16 + }, + { + "epoch": 0.01746276322547509, + "grad_norm": 5.542337417602539, + "learning_rate": 5.136986301369863e-08, + "loss": 2.7625, + "step": 17 + }, + { + "epoch": 0.01848998459167951, + "grad_norm": 5.300518989562988, + "learning_rate": 6.84931506849315e-08, + "loss": 2.84, + "step": 18 + }, + { + "epoch": 0.019517205957883924, + "grad_norm": 5.6646881103515625, + "learning_rate": 8.561643835616439e-08, + "loss": 2.761, + "step": 19 + }, + { + "epoch": 0.02054442732408834, + "grad_norm": 5.375557899475098, + "learning_rate": 1.0273972602739726e-07, + "loss": 2.8215, + "step": 20 + }, + { + "epoch": 0.02054442732408834, + "eval_loss": 2.8507144451141357, + "eval_runtime": 45.1133, + "eval_samples_per_second": 22.166, + "eval_steps_per_second": 0.931, + "step": 20 + }, + { + "epoch": 0.02157164869029276, + "grad_norm": 5.1731977462768555, + "learning_rate": 1.1986301369863014e-07, + "loss": 2.8348, + "step": 21 + }, + { + "epoch": 0.022598870056497175, + "grad_norm": 5.370945930480957, + "learning_rate": 1.36986301369863e-07, + "loss": 2.822, + "step": 22 + }, + { + "epoch": 0.02362609142270159, + "grad_norm": 5.550389289855957, + "learning_rate": 1.541095890410959e-07, + "loss": 2.7904, + "step": 23 + }, + { + "epoch": 0.02465331278890601, + "grad_norm": 5.339347839355469, + "learning_rate": 1.7123287671232878e-07, + "loss": 2.8066, + "step": 24 + }, + { + "epoch": 0.025680534155110426, + "grad_norm": 5.2319722175598145, + "learning_rate": 1.8835616438356165e-07, + "loss": 2.7976, + "step": 25 + }, + { + "epoch": 0.02670775552131484, + "grad_norm": 5.207149982452393, + "learning_rate": 2.0547945205479452e-07, + "loss": 2.8162, + "step": 26 + }, + { + "epoch": 0.02773497688751926, + "grad_norm": 4.400204181671143, + "learning_rate": 2.226027397260274e-07, + "loss": 2.7808, + "step": 27 + }, + { + "epoch": 0.028762198253723677, + "grad_norm": 4.21675968170166, + "learning_rate": 2.397260273972603e-07, + "loss": 2.8044, + "step": 28 + }, + { + "epoch": 0.029789419619928096, + "grad_norm": 4.406878471374512, + "learning_rate": 2.568493150684932e-07, + "loss": 2.8162, + "step": 29 + }, + { + "epoch": 0.030816640986132512, + "grad_norm": 4.024835586547852, + "learning_rate": 2.73972602739726e-07, + "loss": 2.7994, + "step": 30 + }, + { + "epoch": 0.030816640986132512, + "eval_loss": 2.8124029636383057, + "eval_runtime": 43.1772, + "eval_samples_per_second": 23.16, + "eval_steps_per_second": 0.973, + "step": 30 + }, + { + "epoch": 0.03184386235233693, + "grad_norm": 3.5378901958465576, + "learning_rate": 2.910958904109589e-07, + "loss": 2.7637, + "step": 31 + }, + { + "epoch": 0.03287108371854135, + "grad_norm": 3.6289916038513184, + "learning_rate": 3.082191780821918e-07, + "loss": 2.8017, + "step": 32 + }, + { + "epoch": 0.03389830508474576, + "grad_norm": 3.6789698600769043, + "learning_rate": 3.2534246575342466e-07, + "loss": 2.7723, + "step": 33 + }, + { + "epoch": 0.03492552645095018, + "grad_norm": 3.491389751434326, + "learning_rate": 3.4246575342465755e-07, + "loss": 2.7528, + "step": 34 + }, + { + "epoch": 0.035952747817154594, + "grad_norm": 3.383080005645752, + "learning_rate": 3.595890410958904e-07, + "loss": 2.7418, + "step": 35 + }, + { + "epoch": 0.03697996918335902, + "grad_norm": 3.4823594093322754, + "learning_rate": 3.767123287671233e-07, + "loss": 2.6662, + "step": 36 + }, + { + "epoch": 0.03800719054956343, + "grad_norm": 3.2146108150482178, + "learning_rate": 3.938356164383562e-07, + "loss": 2.7659, + "step": 37 + }, + { + "epoch": 0.03903441191576785, + "grad_norm": 2.9549412727355957, + "learning_rate": 4.1095890410958903e-07, + "loss": 2.6838, + "step": 38 + }, + { + "epoch": 0.040061633281972264, + "grad_norm": 3.0744080543518066, + "learning_rate": 4.2808219178082193e-07, + "loss": 2.6879, + "step": 39 + }, + { + "epoch": 0.04108885464817668, + "grad_norm": 3.19840669631958, + "learning_rate": 4.452054794520548e-07, + "loss": 2.7443, + "step": 40 + }, + { + "epoch": 0.04108885464817668, + "eval_loss": 2.7469444274902344, + "eval_runtime": 46.4676, + "eval_samples_per_second": 21.52, + "eval_steps_per_second": 0.904, + "step": 40 + }, + { + "epoch": 0.042116076014381096, + "grad_norm": 3.178406000137329, + "learning_rate": 4.6232876712328767e-07, + "loss": 2.7036, + "step": 41 + }, + { + "epoch": 0.04314329738058552, + "grad_norm": 3.2032673358917236, + "learning_rate": 4.794520547945206e-07, + "loss": 2.6465, + "step": 42 + }, + { + "epoch": 0.044170518746789934, + "grad_norm": 3.1525962352752686, + "learning_rate": 4.965753424657534e-07, + "loss": 2.6996, + "step": 43 + }, + { + "epoch": 0.04519774011299435, + "grad_norm": 2.907524585723877, + "learning_rate": 5.136986301369864e-07, + "loss": 2.6581, + "step": 44 + }, + { + "epoch": 0.046224961479198766, + "grad_norm": 2.5885250568389893, + "learning_rate": 5.308219178082192e-07, + "loss": 2.6773, + "step": 45 + }, + { + "epoch": 0.04725218284540318, + "grad_norm": 2.569068431854248, + "learning_rate": 5.47945205479452e-07, + "loss": 2.6635, + "step": 46 + }, + { + "epoch": 0.048279404211607604, + "grad_norm": 2.5313518047332764, + "learning_rate": 5.65068493150685e-07, + "loss": 2.706, + "step": 47 + }, + { + "epoch": 0.04930662557781202, + "grad_norm": 2.615442991256714, + "learning_rate": 5.821917808219178e-07, + "loss": 2.638, + "step": 48 + }, + { + "epoch": 0.050333846944016436, + "grad_norm": 2.7420806884765625, + "learning_rate": 5.993150684931507e-07, + "loss": 2.6398, + "step": 49 + }, + { + "epoch": 0.05136106831022085, + "grad_norm": 2.5246975421905518, + "learning_rate": 6.164383561643836e-07, + "loss": 2.5916, + "step": 50 + }, + { + "epoch": 0.05136106831022085, + "eval_loss": 2.6723692417144775, + "eval_runtime": 46.7206, + "eval_samples_per_second": 21.404, + "eval_steps_per_second": 0.899, + "step": 50 + }, + { + "epoch": 0.05238828967642527, + "grad_norm": 2.390143632888794, + "learning_rate": 6.335616438356165e-07, + "loss": 2.5773, + "step": 51 + }, + { + "epoch": 0.05341551104262968, + "grad_norm": 2.0792484283447266, + "learning_rate": 6.506849315068493e-07, + "loss": 2.6249, + "step": 52 + }, + { + "epoch": 0.054442732408834106, + "grad_norm": 1.9995019435882568, + "learning_rate": 6.678082191780823e-07, + "loss": 2.5425, + "step": 53 + }, + { + "epoch": 0.05546995377503852, + "grad_norm": 1.985452651977539, + "learning_rate": 6.849315068493151e-07, + "loss": 2.5986, + "step": 54 + }, + { + "epoch": 0.05649717514124294, + "grad_norm": 2.10184907913208, + "learning_rate": 7.020547945205481e-07, + "loss": 2.5713, + "step": 55 + }, + { + "epoch": 0.05752439650744735, + "grad_norm": 2.0898985862731934, + "learning_rate": 7.191780821917808e-07, + "loss": 2.535, + "step": 56 + }, + { + "epoch": 0.05855161787365177, + "grad_norm": 2.0239579677581787, + "learning_rate": 7.363013698630137e-07, + "loss": 2.5667, + "step": 57 + }, + { + "epoch": 0.05957883923985619, + "grad_norm": 2.207080125808716, + "learning_rate": 7.534246575342466e-07, + "loss": 2.4999, + "step": 58 + }, + { + "epoch": 0.06060606060606061, + "grad_norm": 1.8541918992996216, + "learning_rate": 7.705479452054795e-07, + "loss": 2.5582, + "step": 59 + }, + { + "epoch": 0.061633281972265024, + "grad_norm": 1.8530476093292236, + "learning_rate": 7.876712328767124e-07, + "loss": 2.5535, + "step": 60 + }, + { + "epoch": 0.061633281972265024, + "eval_loss": 2.607963800430298, + "eval_runtime": 45.9754, + "eval_samples_per_second": 21.751, + "eval_steps_per_second": 0.914, + "step": 60 + }, + { + "epoch": 0.06266050333846944, + "grad_norm": 1.7978090047836304, + "learning_rate": 8.047945205479453e-07, + "loss": 2.53, + "step": 61 + }, + { + "epoch": 0.06368772470467386, + "grad_norm": 1.8173762559890747, + "learning_rate": 8.219178082191781e-07, + "loss": 2.5144, + "step": 62 + }, + { + "epoch": 0.06471494607087827, + "grad_norm": 2.0847485065460205, + "learning_rate": 8.39041095890411e-07, + "loss": 2.5525, + "step": 63 + }, + { + "epoch": 0.0657421674370827, + "grad_norm": 1.8691741228103638, + "learning_rate": 8.561643835616439e-07, + "loss": 2.5451, + "step": 64 + }, + { + "epoch": 0.0667693888032871, + "grad_norm": 2.0242726802825928, + "learning_rate": 8.732876712328768e-07, + "loss": 2.4145, + "step": 65 + }, + { + "epoch": 0.06779661016949153, + "grad_norm": 1.8268908262252808, + "learning_rate": 8.904109589041097e-07, + "loss": 2.5024, + "step": 66 + }, + { + "epoch": 0.06882383153569595, + "grad_norm": 1.8703057765960693, + "learning_rate": 9.075342465753426e-07, + "loss": 2.461, + "step": 67 + }, + { + "epoch": 0.06985105290190036, + "grad_norm": 1.8747144937515259, + "learning_rate": 9.246575342465753e-07, + "loss": 2.5122, + "step": 68 + }, + { + "epoch": 0.07087827426810478, + "grad_norm": 1.744924783706665, + "learning_rate": 9.417808219178083e-07, + "loss": 2.4673, + "step": 69 + }, + { + "epoch": 0.07190549563430919, + "grad_norm": 1.7492460012435913, + "learning_rate": 9.589041095890411e-07, + "loss": 2.5598, + "step": 70 + }, + { + "epoch": 0.07190549563430919, + "eval_loss": 2.547417640686035, + "eval_runtime": 49.7488, + "eval_samples_per_second": 20.101, + "eval_steps_per_second": 0.844, + "step": 70 + }, + { + "epoch": 0.07293271700051361, + "grad_norm": 1.9218230247497559, + "learning_rate": 9.76027397260274e-07, + "loss": 2.4562, + "step": 71 + }, + { + "epoch": 0.07395993836671803, + "grad_norm": 2.0127363204956055, + "learning_rate": 9.931506849315068e-07, + "loss": 2.4141, + "step": 72 + }, + { + "epoch": 0.07498715973292244, + "grad_norm": 1.8700369596481323, + "learning_rate": 1.0102739726027399e-06, + "loss": 2.4696, + "step": 73 + }, + { + "epoch": 0.07601438109912687, + "grad_norm": 2.1706016063690186, + "learning_rate": 1.0273972602739727e-06, + "loss": 2.4404, + "step": 74 + }, + { + "epoch": 0.07704160246533127, + "grad_norm": 2.470412254333496, + "learning_rate": 1.0445205479452056e-06, + "loss": 2.4409, + "step": 75 + }, + { + "epoch": 0.0780688238315357, + "grad_norm": 1.906569242477417, + "learning_rate": 1.0616438356164384e-06, + "loss": 2.4313, + "step": 76 + }, + { + "epoch": 0.07909604519774012, + "grad_norm": 2.122936487197876, + "learning_rate": 1.0787671232876712e-06, + "loss": 2.4369, + "step": 77 + }, + { + "epoch": 0.08012326656394453, + "grad_norm": 2.079127311706543, + "learning_rate": 1.095890410958904e-06, + "loss": 2.449, + "step": 78 + }, + { + "epoch": 0.08115048793014895, + "grad_norm": 1.9173146486282349, + "learning_rate": 1.1130136986301371e-06, + "loss": 2.3782, + "step": 79 + }, + { + "epoch": 0.08217770929635336, + "grad_norm": 1.9856903553009033, + "learning_rate": 1.13013698630137e-06, + "loss": 2.4215, + "step": 80 + }, + { + "epoch": 0.08217770929635336, + "eval_loss": 2.479673385620117, + "eval_runtime": 46.2211, + "eval_samples_per_second": 21.635, + "eval_steps_per_second": 0.909, + "step": 80 + }, + { + "epoch": 0.08320493066255778, + "grad_norm": 2.3764865398406982, + "learning_rate": 1.1472602739726028e-06, + "loss": 2.3744, + "step": 81 + }, + { + "epoch": 0.08423215202876219, + "grad_norm": 1.980668544769287, + "learning_rate": 1.1643835616438357e-06, + "loss": 2.3856, + "step": 82 + }, + { + "epoch": 0.08525937339496661, + "grad_norm": 2.2589704990386963, + "learning_rate": 1.1815068493150685e-06, + "loss": 2.3853, + "step": 83 + }, + { + "epoch": 0.08628659476117104, + "grad_norm": 2.171780824661255, + "learning_rate": 1.1986301369863014e-06, + "loss": 2.3603, + "step": 84 + }, + { + "epoch": 0.08731381612737545, + "grad_norm": 2.25736403465271, + "learning_rate": 1.2157534246575344e-06, + "loss": 2.3189, + "step": 85 + }, + { + "epoch": 0.08834103749357987, + "grad_norm": 2.3643763065338135, + "learning_rate": 1.2328767123287673e-06, + "loss": 2.3795, + "step": 86 + }, + { + "epoch": 0.08936825885978428, + "grad_norm": 2.384516954421997, + "learning_rate": 1.25e-06, + "loss": 2.3257, + "step": 87 + }, + { + "epoch": 0.0903954802259887, + "grad_norm": 2.5072567462921143, + "learning_rate": 1.267123287671233e-06, + "loss": 2.3125, + "step": 88 + }, + { + "epoch": 0.09142270159219312, + "grad_norm": 2.2726807594299316, + "learning_rate": 1.284246575342466e-06, + "loss": 2.3297, + "step": 89 + }, + { + "epoch": 0.09244992295839753, + "grad_norm": 2.4817421436309814, + "learning_rate": 1.3013698630136986e-06, + "loss": 2.2878, + "step": 90 + }, + { + "epoch": 0.09244992295839753, + "eval_loss": 2.4065146446228027, + "eval_runtime": 47.9372, + "eval_samples_per_second": 20.861, + "eval_steps_per_second": 0.876, + "step": 90 + }, + { + "epoch": 0.09347714432460195, + "grad_norm": 2.2219603061676025, + "learning_rate": 1.3184931506849317e-06, + "loss": 2.2978, + "step": 91 + }, + { + "epoch": 0.09450436569080636, + "grad_norm": 2.8790745735168457, + "learning_rate": 1.3356164383561645e-06, + "loss": 2.313, + "step": 92 + }, + { + "epoch": 0.09553158705701079, + "grad_norm": 2.589200496673584, + "learning_rate": 1.3527397260273976e-06, + "loss": 2.2937, + "step": 93 + }, + { + "epoch": 0.09655880842321521, + "grad_norm": 2.760016679763794, + "learning_rate": 1.3698630136986302e-06, + "loss": 2.277, + "step": 94 + }, + { + "epoch": 0.09758602978941962, + "grad_norm": 2.2641494274139404, + "learning_rate": 1.386986301369863e-06, + "loss": 2.3336, + "step": 95 + }, + { + "epoch": 0.09861325115562404, + "grad_norm": 2.2786386013031006, + "learning_rate": 1.4041095890410961e-06, + "loss": 2.233, + "step": 96 + }, + { + "epoch": 0.09964047252182845, + "grad_norm": 2.5590052604675293, + "learning_rate": 1.421232876712329e-06, + "loss": 2.2851, + "step": 97 + }, + { + "epoch": 0.10066769388803287, + "grad_norm": 2.5609471797943115, + "learning_rate": 1.4383561643835616e-06, + "loss": 2.1848, + "step": 98 + }, + { + "epoch": 0.1016949152542373, + "grad_norm": 2.442521810531616, + "learning_rate": 1.4554794520547946e-06, + "loss": 2.2073, + "step": 99 + }, + { + "epoch": 0.1027221366204417, + "grad_norm": 2.3735859394073486, + "learning_rate": 1.4726027397260275e-06, + "loss": 2.2353, + "step": 100 + }, + { + "epoch": 0.1027221366204417, + "eval_loss": 2.3457911014556885, + "eval_runtime": 51.6469, + "eval_samples_per_second": 19.362, + "eval_steps_per_second": 0.813, + "step": 100 + }, + { + "epoch": 0.10374935798664613, + "grad_norm": 2.175354242324829, + "learning_rate": 1.4897260273972605e-06, + "loss": 2.2259, + "step": 101 + }, + { + "epoch": 0.10477657935285054, + "grad_norm": 2.217677116394043, + "learning_rate": 1.5068493150684932e-06, + "loss": 2.2082, + "step": 102 + }, + { + "epoch": 0.10580380071905496, + "grad_norm": 2.4618067741394043, + "learning_rate": 1.5239726027397262e-06, + "loss": 2.1649, + "step": 103 + }, + { + "epoch": 0.10683102208525937, + "grad_norm": 2.232060432434082, + "learning_rate": 1.541095890410959e-06, + "loss": 2.1962, + "step": 104 + }, + { + "epoch": 0.10785824345146379, + "grad_norm": 2.9140872955322266, + "learning_rate": 1.5582191780821921e-06, + "loss": 2.2446, + "step": 105 + }, + { + "epoch": 0.10888546481766821, + "grad_norm": 2.232158660888672, + "learning_rate": 1.5753424657534248e-06, + "loss": 2.2179, + "step": 106 + }, + { + "epoch": 0.10991268618387262, + "grad_norm": 2.804938793182373, + "learning_rate": 1.5924657534246576e-06, + "loss": 2.2449, + "step": 107 + }, + { + "epoch": 0.11093990755007704, + "grad_norm": 2.3690710067749023, + "learning_rate": 1.6095890410958907e-06, + "loss": 2.225, + "step": 108 + }, + { + "epoch": 0.11196712891628145, + "grad_norm": 2.7717416286468506, + "learning_rate": 1.6267123287671235e-06, + "loss": 2.0922, + "step": 109 + }, + { + "epoch": 0.11299435028248588, + "grad_norm": 3.053602933883667, + "learning_rate": 1.6438356164383561e-06, + "loss": 2.1702, + "step": 110 + }, + { + "epoch": 0.11299435028248588, + "eval_loss": 2.298161029815674, + "eval_runtime": 44.0776, + "eval_samples_per_second": 22.687, + "eval_steps_per_second": 0.953, + "step": 110 + }, + { + "epoch": 0.1140215716486903, + "grad_norm": 2.8673532009124756, + "learning_rate": 1.6609589041095892e-06, + "loss": 2.0938, + "step": 111 + }, + { + "epoch": 0.1150487930148947, + "grad_norm": 2.57185959815979, + "learning_rate": 1.678082191780822e-06, + "loss": 2.1892, + "step": 112 + }, + { + "epoch": 0.11607601438109913, + "grad_norm": 2.5278828144073486, + "learning_rate": 1.695205479452055e-06, + "loss": 2.1251, + "step": 113 + }, + { + "epoch": 0.11710323574730354, + "grad_norm": 2.8581748008728027, + "learning_rate": 1.7123287671232877e-06, + "loss": 2.1699, + "step": 114 + }, + { + "epoch": 0.11813045711350796, + "grad_norm": 2.924959897994995, + "learning_rate": 1.7294520547945206e-06, + "loss": 2.1005, + "step": 115 + }, + { + "epoch": 0.11915767847971238, + "grad_norm": 2.477034091949463, + "learning_rate": 1.7465753424657536e-06, + "loss": 2.2016, + "step": 116 + }, + { + "epoch": 0.12018489984591679, + "grad_norm": 2.8719305992126465, + "learning_rate": 1.7636986301369865e-06, + "loss": 2.1588, + "step": 117 + }, + { + "epoch": 0.12121212121212122, + "grad_norm": 2.4403953552246094, + "learning_rate": 1.7808219178082193e-06, + "loss": 2.1445, + "step": 118 + }, + { + "epoch": 0.12223934257832562, + "grad_norm": 2.7994632720947266, + "learning_rate": 1.7979452054794521e-06, + "loss": 2.1369, + "step": 119 + }, + { + "epoch": 0.12326656394453005, + "grad_norm": 2.163196563720703, + "learning_rate": 1.8150684931506852e-06, + "loss": 2.2195, + "step": 120 + }, + { + "epoch": 0.12326656394453005, + "eval_loss": 2.2648775577545166, + "eval_runtime": 49.47, + "eval_samples_per_second": 20.214, + "eval_steps_per_second": 0.849, + "step": 120 + }, + { + "epoch": 0.12429378531073447, + "grad_norm": 2.6743483543395996, + "learning_rate": 1.832191780821918e-06, + "loss": 2.1521, + "step": 121 + }, + { + "epoch": 0.12532100667693888, + "grad_norm": 2.54795503616333, + "learning_rate": 1.8493150684931507e-06, + "loss": 2.1158, + "step": 122 + }, + { + "epoch": 0.1263482280431433, + "grad_norm": 2.9093267917633057, + "learning_rate": 1.8664383561643837e-06, + "loss": 2.0707, + "step": 123 + }, + { + "epoch": 0.12737544940934772, + "grad_norm": 2.416940212249756, + "learning_rate": 1.8835616438356166e-06, + "loss": 2.0901, + "step": 124 + }, + { + "epoch": 0.12840267077555212, + "grad_norm": 2.8333663940429688, + "learning_rate": 1.9006849315068496e-06, + "loss": 2.0545, + "step": 125 + }, + { + "epoch": 0.12942989214175654, + "grad_norm": 2.483051061630249, + "learning_rate": 1.9178082191780823e-06, + "loss": 2.1274, + "step": 126 + }, + { + "epoch": 0.13045711350796096, + "grad_norm": 2.3486859798431396, + "learning_rate": 1.9349315068493153e-06, + "loss": 2.0478, + "step": 127 + }, + { + "epoch": 0.1314843348741654, + "grad_norm": 3.1512887477874756, + "learning_rate": 1.952054794520548e-06, + "loss": 2.1041, + "step": 128 + }, + { + "epoch": 0.1325115562403698, + "grad_norm": 2.902873992919922, + "learning_rate": 1.969178082191781e-06, + "loss": 2.1332, + "step": 129 + }, + { + "epoch": 0.1335387776065742, + "grad_norm": 2.779132127761841, + "learning_rate": 1.9863013698630136e-06, + "loss": 2.1297, + "step": 130 + }, + { + "epoch": 0.1335387776065742, + "eval_loss": 2.2391879558563232, + "eval_runtime": 55.3018, + "eval_samples_per_second": 18.083, + "eval_steps_per_second": 0.759, + "step": 130 + }, + { + "epoch": 0.13456599897277863, + "grad_norm": 2.703019618988037, + "learning_rate": 2.0034246575342467e-06, + "loss": 2.0506, + "step": 131 + }, + { + "epoch": 0.13559322033898305, + "grad_norm": 3.0187575817108154, + "learning_rate": 2.0205479452054797e-06, + "loss": 2.1291, + "step": 132 + }, + { + "epoch": 0.13662044170518747, + "grad_norm": 2.7629971504211426, + "learning_rate": 2.037671232876713e-06, + "loss": 2.1003, + "step": 133 + }, + { + "epoch": 0.1376476630713919, + "grad_norm": 3.175776720046997, + "learning_rate": 2.0547945205479454e-06, + "loss": 2.0429, + "step": 134 + }, + { + "epoch": 0.1386748844375963, + "grad_norm": 3.0342514514923096, + "learning_rate": 2.071917808219178e-06, + "loss": 2.0858, + "step": 135 + }, + { + "epoch": 0.1397021058038007, + "grad_norm": 3.0861928462982178, + "learning_rate": 2.089041095890411e-06, + "loss": 2.0964, + "step": 136 + }, + { + "epoch": 0.14072932717000514, + "grad_norm": 3.151945114135742, + "learning_rate": 2.106164383561644e-06, + "loss": 2.0194, + "step": 137 + }, + { + "epoch": 0.14175654853620956, + "grad_norm": 2.9069676399230957, + "learning_rate": 2.123287671232877e-06, + "loss": 2.1131, + "step": 138 + }, + { + "epoch": 0.14278376990241398, + "grad_norm": 3.2285282611846924, + "learning_rate": 2.14041095890411e-06, + "loss": 2.0643, + "step": 139 + }, + { + "epoch": 0.14381099126861838, + "grad_norm": 2.500481367111206, + "learning_rate": 2.1575342465753425e-06, + "loss": 2.0625, + "step": 140 + }, + { + "epoch": 0.14381099126861838, + "eval_loss": 2.211498975753784, + "eval_runtime": 55.1046, + "eval_samples_per_second": 18.147, + "eval_steps_per_second": 0.762, + "step": 140 + }, + { + "epoch": 0.1448382126348228, + "grad_norm": 4.004592418670654, + "learning_rate": 2.1746575342465755e-06, + "loss": 2.0644, + "step": 141 + }, + { + "epoch": 0.14586543400102722, + "grad_norm": 2.9258391857147217, + "learning_rate": 2.191780821917808e-06, + "loss": 2.0174, + "step": 142 + }, + { + "epoch": 0.14689265536723164, + "grad_norm": 3.0073509216308594, + "learning_rate": 2.2089041095890412e-06, + "loss": 2.1417, + "step": 143 + }, + { + "epoch": 0.14791987673343607, + "grad_norm": 3.040708065032959, + "learning_rate": 2.2260273972602743e-06, + "loss": 2.0357, + "step": 144 + }, + { + "epoch": 0.14894709809964046, + "grad_norm": 2.9062869548797607, + "learning_rate": 2.243150684931507e-06, + "loss": 2.0122, + "step": 145 + }, + { + "epoch": 0.14997431946584489, + "grad_norm": 3.28969669342041, + "learning_rate": 2.26027397260274e-06, + "loss": 2.0255, + "step": 146 + }, + { + "epoch": 0.1510015408320493, + "grad_norm": 3.608752727508545, + "learning_rate": 2.2773972602739726e-06, + "loss": 2.0184, + "step": 147 + }, + { + "epoch": 0.15202876219825373, + "grad_norm": 2.900341033935547, + "learning_rate": 2.2945205479452057e-06, + "loss": 2.0532, + "step": 148 + }, + { + "epoch": 0.15305598356445815, + "grad_norm": 2.881843328475952, + "learning_rate": 2.3116438356164387e-06, + "loss": 2.1336, + "step": 149 + }, + { + "epoch": 0.15408320493066255, + "grad_norm": 3.4043259620666504, + "learning_rate": 2.3287671232876713e-06, + "loss": 2.0536, + "step": 150 + }, + { + "epoch": 0.15408320493066255, + "eval_loss": 2.192880153656006, + "eval_runtime": 53.7838, + "eval_samples_per_second": 18.593, + "eval_steps_per_second": 0.781, + "step": 150 + }, + { + "epoch": 0.15511042629686697, + "grad_norm": 2.693107843399048, + "learning_rate": 2.3458904109589044e-06, + "loss": 1.9503, + "step": 151 + }, + { + "epoch": 0.1561376476630714, + "grad_norm": 2.885260581970215, + "learning_rate": 2.363013698630137e-06, + "loss": 2.0081, + "step": 152 + }, + { + "epoch": 0.15716486902927582, + "grad_norm": 3.4459924697875977, + "learning_rate": 2.38013698630137e-06, + "loss": 2.0649, + "step": 153 + }, + { + "epoch": 0.15819209039548024, + "grad_norm": 2.8447093963623047, + "learning_rate": 2.3972602739726027e-06, + "loss": 2.0476, + "step": 154 + }, + { + "epoch": 0.15921931176168463, + "grad_norm": 2.755176305770874, + "learning_rate": 2.4143835616438358e-06, + "loss": 2.0006, + "step": 155 + }, + { + "epoch": 0.16024653312788906, + "grad_norm": 2.8677561283111572, + "learning_rate": 2.431506849315069e-06, + "loss": 1.9663, + "step": 156 + }, + { + "epoch": 0.16127375449409348, + "grad_norm": 2.783351182937622, + "learning_rate": 2.4486301369863015e-06, + "loss": 2.0301, + "step": 157 + }, + { + "epoch": 0.1623009758602979, + "grad_norm": 2.7545998096466064, + "learning_rate": 2.4657534246575345e-06, + "loss": 2.0438, + "step": 158 + }, + { + "epoch": 0.1633281972265023, + "grad_norm": 2.8140594959259033, + "learning_rate": 2.482876712328767e-06, + "loss": 2.0055, + "step": 159 + }, + { + "epoch": 0.16435541859270672, + "grad_norm": 2.6072335243225098, + "learning_rate": 2.5e-06, + "loss": 2.0262, + "step": 160 + }, + { + "epoch": 0.16435541859270672, + "eval_loss": 2.179842710494995, + "eval_runtime": 52.6159, + "eval_samples_per_second": 19.006, + "eval_steps_per_second": 0.798, + "step": 160 + }, + { + "epoch": 0.16538263995891114, + "grad_norm": 2.830212354660034, + "learning_rate": 2.5171232876712333e-06, + "loss": 2.0154, + "step": 161 + }, + { + "epoch": 0.16640986132511557, + "grad_norm": 2.677140712738037, + "learning_rate": 2.534246575342466e-06, + "loss": 1.9933, + "step": 162 + }, + { + "epoch": 0.16743708269132, + "grad_norm": 2.837759017944336, + "learning_rate": 2.551369863013699e-06, + "loss": 1.9602, + "step": 163 + }, + { + "epoch": 0.16846430405752438, + "grad_norm": 3.9556589126586914, + "learning_rate": 2.568493150684932e-06, + "loss": 2.0121, + "step": 164 + }, + { + "epoch": 0.1694915254237288, + "grad_norm": 2.8715925216674805, + "learning_rate": 2.585616438356164e-06, + "loss": 2.0219, + "step": 165 + }, + { + "epoch": 0.17051874678993323, + "grad_norm": 3.5073623657226562, + "learning_rate": 2.6027397260273973e-06, + "loss": 2.0202, + "step": 166 + }, + { + "epoch": 0.17154596815613765, + "grad_norm": 2.751962900161743, + "learning_rate": 2.6198630136986303e-06, + "loss": 1.9528, + "step": 167 + }, + { + "epoch": 0.17257318952234207, + "grad_norm": 3.7753474712371826, + "learning_rate": 2.6369863013698634e-06, + "loss": 2.0208, + "step": 168 + }, + { + "epoch": 0.17360041088854647, + "grad_norm": 2.642289876937866, + "learning_rate": 2.654109589041096e-06, + "loss": 2.0386, + "step": 169 + }, + { + "epoch": 0.1746276322547509, + "grad_norm": 2.82173490524292, + "learning_rate": 2.671232876712329e-06, + "loss": 2.0355, + "step": 170 + }, + { + "epoch": 0.1746276322547509, + "eval_loss": 2.1699249744415283, + "eval_runtime": 50.0341, + "eval_samples_per_second": 19.986, + "eval_steps_per_second": 0.839, + "step": 170 + }, + { + "epoch": 0.17565485362095531, + "grad_norm": 3.342280149459839, + "learning_rate": 2.688356164383562e-06, + "loss": 1.9672, + "step": 171 + }, + { + "epoch": 0.17668207498715974, + "grad_norm": 2.6548173427581787, + "learning_rate": 2.705479452054795e-06, + "loss": 1.9669, + "step": 172 + }, + { + "epoch": 0.17770929635336416, + "grad_norm": 2.6548173427581787, + "learning_rate": 2.705479452054795e-06, + "loss": 1.961, + "step": 173 + }, + { + "epoch": 0.17873651771956856, + "grad_norm": 3.4306116104125977, + "learning_rate": 2.7226027397260274e-06, + "loss": 1.9173, + "step": 174 + }, + { + "epoch": 0.17976373908577298, + "grad_norm": 2.7991764545440674, + "learning_rate": 2.7397260273972604e-06, + "loss": 2.0339, + "step": 175 + }, + { + "epoch": 0.1807909604519774, + "grad_norm": 3.6040217876434326, + "learning_rate": 2.7568493150684935e-06, + "loss": 2.0218, + "step": 176 + }, + { + "epoch": 0.18181818181818182, + "grad_norm": 3.0253515243530273, + "learning_rate": 2.773972602739726e-06, + "loss": 1.9245, + "step": 177 + }, + { + "epoch": 0.18284540318438625, + "grad_norm": 3.405141830444336, + "learning_rate": 2.791095890410959e-06, + "loss": 2.0136, + "step": 178 + }, + { + "epoch": 0.18387262455059064, + "grad_norm": 2.7518203258514404, + "learning_rate": 2.8082191780821922e-06, + "loss": 1.8925, + "step": 179 + }, + { + "epoch": 0.18489984591679506, + "grad_norm": 3.1903367042541504, + "learning_rate": 2.8253424657534253e-06, + "loss": 2.0523, + "step": 180 + }, + { + "epoch": 0.18489984591679506, + "eval_loss": 2.158029794692993, + "eval_runtime": 48.6162, + "eval_samples_per_second": 20.569, + "eval_steps_per_second": 0.864, + "step": 180 + }, + { + "epoch": 0.1859270672829995, + "grad_norm": 3.0331413745880127, + "learning_rate": 2.842465753424658e-06, + "loss": 2.0167, + "step": 181 + }, + { + "epoch": 0.1869542886492039, + "grad_norm": 2.919433116912842, + "learning_rate": 2.8595890410958905e-06, + "loss": 2.0197, + "step": 182 + }, + { + "epoch": 0.18798151001540833, + "grad_norm": 2.9013049602508545, + "learning_rate": 2.876712328767123e-06, + "loss": 2.0301, + "step": 183 + }, + { + "epoch": 0.18900873138161273, + "grad_norm": 2.738800525665283, + "learning_rate": 2.8938356164383562e-06, + "loss": 1.894, + "step": 184 + }, + { + "epoch": 0.19003595274781715, + "grad_norm": 2.670806407928467, + "learning_rate": 2.9109589041095893e-06, + "loss": 1.9127, + "step": 185 + }, + { + "epoch": 0.19106317411402157, + "grad_norm": 3.0435004234313965, + "learning_rate": 2.9280821917808223e-06, + "loss": 1.8996, + "step": 186 + }, + { + "epoch": 0.192090395480226, + "grad_norm": 3.2139949798583984, + "learning_rate": 2.945205479452055e-06, + "loss": 1.9255, + "step": 187 + }, + { + "epoch": 0.19311761684643042, + "grad_norm": 3.252965211868286, + "learning_rate": 2.962328767123288e-06, + "loss": 1.9809, + "step": 188 + }, + { + "epoch": 0.1941448382126348, + "grad_norm": 2.9325191974639893, + "learning_rate": 2.979452054794521e-06, + "loss": 1.9691, + "step": 189 + }, + { + "epoch": 0.19517205957883924, + "grad_norm": 3.5879862308502197, + "learning_rate": 2.9965753424657533e-06, + "loss": 1.9205, + "step": 190 + }, + { + "epoch": 0.19517205957883924, + "eval_loss": 2.13791561126709, + "eval_runtime": 44.8545, + "eval_samples_per_second": 22.294, + "eval_steps_per_second": 0.936, + "step": 190 + }, + { + "epoch": 0.19619928094504366, + "grad_norm": 3.0025253295898438, + "learning_rate": 3.0136986301369864e-06, + "loss": 1.9717, + "step": 191 + }, + { + "epoch": 0.19722650231124808, + "grad_norm": 3.1785972118377686, + "learning_rate": 3.0308219178082194e-06, + "loss": 1.9535, + "step": 192 + }, + { + "epoch": 0.1982537236774525, + "grad_norm": 3.121065378189087, + "learning_rate": 3.0479452054794525e-06, + "loss": 1.9026, + "step": 193 + }, + { + "epoch": 0.1992809450436569, + "grad_norm": 2.838779926300049, + "learning_rate": 3.065068493150685e-06, + "loss": 1.941, + "step": 194 + }, + { + "epoch": 0.20030816640986132, + "grad_norm": 2.8916969299316406, + "learning_rate": 3.082191780821918e-06, + "loss": 1.8983, + "step": 195 + }, + { + "epoch": 0.20133538777606574, + "grad_norm": 2.9195168018341064, + "learning_rate": 3.099315068493151e-06, + "loss": 1.918, + "step": 196 + }, + { + "epoch": 0.20236260914227017, + "grad_norm": 3.235989809036255, + "learning_rate": 3.1164383561643843e-06, + "loss": 1.9111, + "step": 197 + }, + { + "epoch": 0.2033898305084746, + "grad_norm": 3.0316455364227295, + "learning_rate": 3.1335616438356165e-06, + "loss": 1.8839, + "step": 198 + }, + { + "epoch": 0.20441705187467898, + "grad_norm": 2.963610887527466, + "learning_rate": 3.1506849315068495e-06, + "loss": 2.0303, + "step": 199 + }, + { + "epoch": 0.2054442732408834, + "grad_norm": 2.944054365158081, + "learning_rate": 3.167808219178082e-06, + "loss": 1.8822, + "step": 200 + }, + { + "epoch": 0.2054442732408834, + "eval_loss": 2.1433050632476807, + "eval_runtime": 45.8205, + "eval_samples_per_second": 21.824, + "eval_steps_per_second": 0.917, + "step": 200 + }, + { + "epoch": 0.20647149460708783, + "grad_norm": 2.8648695945739746, + "learning_rate": 3.184931506849315e-06, + "loss": 1.8841, + "step": 201 + }, + { + "epoch": 0.20749871597329225, + "grad_norm": 3.2838797569274902, + "learning_rate": 3.2020547945205483e-06, + "loss": 1.8623, + "step": 202 + }, + { + "epoch": 0.20852593733949668, + "grad_norm": 2.7965173721313477, + "learning_rate": 3.2191780821917813e-06, + "loss": 1.8521, + "step": 203 + }, + { + "epoch": 0.20955315870570107, + "grad_norm": 3.3449349403381348, + "learning_rate": 3.236301369863014e-06, + "loss": 1.84, + "step": 204 + }, + { + "epoch": 0.2105803800719055, + "grad_norm": 3.4888951778411865, + "learning_rate": 3.253424657534247e-06, + "loss": 1.824, + "step": 205 + }, + { + "epoch": 0.21160760143810992, + "grad_norm": 2.991349220275879, + "learning_rate": 3.2705479452054796e-06, + "loss": 1.9101, + "step": 206 + }, + { + "epoch": 0.21263482280431434, + "grad_norm": 3.313509702682495, + "learning_rate": 3.2876712328767123e-06, + "loss": 1.8912, + "step": 207 + }, + { + "epoch": 0.21366204417051873, + "grad_norm": 2.8260746002197266, + "learning_rate": 3.3047945205479453e-06, + "loss": 1.8555, + "step": 208 + }, + { + "epoch": 0.21468926553672316, + "grad_norm": 2.7131688594818115, + "learning_rate": 3.3219178082191784e-06, + "loss": 1.8693, + "step": 209 + }, + { + "epoch": 0.21571648690292758, + "grad_norm": 2.5836338996887207, + "learning_rate": 3.3390410958904114e-06, + "loss": 1.8965, + "step": 210 + }, + { + "epoch": 0.21571648690292758, + "eval_loss": 2.1232542991638184, + "eval_runtime": 42.3352, + "eval_samples_per_second": 23.621, + "eval_steps_per_second": 0.992, + "step": 210 + }, + { + "epoch": 0.216743708269132, + "grad_norm": 2.8341355323791504, + "learning_rate": 3.356164383561644e-06, + "loss": 1.9131, + "step": 211 + }, + { + "epoch": 0.21777092963533642, + "grad_norm": 2.490853786468506, + "learning_rate": 3.373287671232877e-06, + "loss": 1.9626, + "step": 212 + }, + { + "epoch": 0.21879815100154082, + "grad_norm": 2.597956895828247, + "learning_rate": 3.39041095890411e-06, + "loss": 1.9203, + "step": 213 + }, + { + "epoch": 0.21982537236774524, + "grad_norm": 3.1104283332824707, + "learning_rate": 3.4075342465753424e-06, + "loss": 1.9442, + "step": 214 + }, + { + "epoch": 0.22085259373394966, + "grad_norm": 2.5698931217193604, + "learning_rate": 3.4246575342465754e-06, + "loss": 1.8385, + "step": 215 + }, + { + "epoch": 0.2218798151001541, + "grad_norm": 3.1637446880340576, + "learning_rate": 3.4417808219178085e-06, + "loss": 1.9005, + "step": 216 + }, + { + "epoch": 0.2229070364663585, + "grad_norm": 2.764070749282837, + "learning_rate": 3.458904109589041e-06, + "loss": 1.8886, + "step": 217 + }, + { + "epoch": 0.2239342578325629, + "grad_norm": 3.303781032562256, + "learning_rate": 3.476027397260274e-06, + "loss": 1.9085, + "step": 218 + }, + { + "epoch": 0.22496147919876733, + "grad_norm": 2.9543044567108154, + "learning_rate": 3.4931506849315072e-06, + "loss": 1.9014, + "step": 219 + }, + { + "epoch": 0.22598870056497175, + "grad_norm": 2.7277028560638428, + "learning_rate": 3.5102739726027403e-06, + "loss": 1.9017, + "step": 220 + }, + { + "epoch": 0.22598870056497175, + "eval_loss": 2.1024081707000732, + "eval_runtime": 45.4966, + "eval_samples_per_second": 21.98, + "eval_steps_per_second": 0.923, + "step": 220 + }, + { + "epoch": 0.22701592193117617, + "grad_norm": 2.864140033721924, + "learning_rate": 3.527397260273973e-06, + "loss": 1.8935, + "step": 221 + }, + { + "epoch": 0.2280431432973806, + "grad_norm": 3.4133172035217285, + "learning_rate": 3.5445205479452056e-06, + "loss": 1.874, + "step": 222 + }, + { + "epoch": 0.229070364663585, + "grad_norm": 3.33343243598938, + "learning_rate": 3.5616438356164386e-06, + "loss": 1.8789, + "step": 223 + }, + { + "epoch": 0.2300975860297894, + "grad_norm": 3.5947320461273193, + "learning_rate": 3.5787671232876712e-06, + "loss": 1.9529, + "step": 224 + }, + { + "epoch": 0.23112480739599384, + "grad_norm": 3.2194948196411133, + "learning_rate": 3.5958904109589043e-06, + "loss": 1.9205, + "step": 225 + }, + { + "epoch": 0.23215202876219826, + "grad_norm": 3.3137149810791016, + "learning_rate": 3.6130136986301373e-06, + "loss": 1.8584, + "step": 226 + }, + { + "epoch": 0.23317925012840268, + "grad_norm": 3.3358561992645264, + "learning_rate": 3.6301369863013704e-06, + "loss": 1.8817, + "step": 227 + }, + { + "epoch": 0.23420647149460708, + "grad_norm": 3.8612711429595947, + "learning_rate": 3.647260273972603e-06, + "loss": 1.9189, + "step": 228 + }, + { + "epoch": 0.2352336928608115, + "grad_norm": 3.0039336681365967, + "learning_rate": 3.664383561643836e-06, + "loss": 1.92, + "step": 229 + }, + { + "epoch": 0.23626091422701592, + "grad_norm": 3.8983118534088135, + "learning_rate": 3.6815068493150687e-06, + "loss": 1.8512, + "step": 230 + }, + { + "epoch": 0.23626091422701592, + "eval_loss": 2.1237711906433105, + "eval_runtime": 45.6126, + "eval_samples_per_second": 21.924, + "eval_steps_per_second": 0.921, + "step": 230 + }, + { + "epoch": 0.23728813559322035, + "grad_norm": 3.061300039291382, + "learning_rate": 3.6986301369863014e-06, + "loss": 1.8772, + "step": 231 + }, + { + "epoch": 0.23831535695942477, + "grad_norm": 3.3102450370788574, + "learning_rate": 3.7157534246575344e-06, + "loss": 1.8395, + "step": 232 + }, + { + "epoch": 0.23934257832562916, + "grad_norm": 2.9634461402893066, + "learning_rate": 3.7328767123287675e-06, + "loss": 1.9447, + "step": 233 + }, + { + "epoch": 0.24036979969183359, + "grad_norm": 3.0937695503234863, + "learning_rate": 3.7500000000000005e-06, + "loss": 1.8628, + "step": 234 + }, + { + "epoch": 0.241397021058038, + "grad_norm": 2.8129842281341553, + "learning_rate": 3.767123287671233e-06, + "loss": 1.8891, + "step": 235 + }, + { + "epoch": 0.24242424242424243, + "grad_norm": 2.9616780281066895, + "learning_rate": 3.784246575342466e-06, + "loss": 1.9112, + "step": 236 + }, + { + "epoch": 0.24345146379044685, + "grad_norm": 2.837721109390259, + "learning_rate": 3.8013698630136993e-06, + "loss": 1.8514, + "step": 237 + }, + { + "epoch": 0.24447868515665125, + "grad_norm": 3.137784004211426, + "learning_rate": 3.818493150684932e-06, + "loss": 1.8764, + "step": 238 + }, + { + "epoch": 0.24550590652285567, + "grad_norm": 2.9692037105560303, + "learning_rate": 3.8356164383561645e-06, + "loss": 1.8201, + "step": 239 + }, + { + "epoch": 0.2465331278890601, + "grad_norm": 2.6718621253967285, + "learning_rate": 3.852739726027397e-06, + "loss": 1.9309, + "step": 240 + }, + { + "epoch": 0.2465331278890601, + "eval_loss": 2.101331949234009, + "eval_runtime": 45.6346, + "eval_samples_per_second": 21.913, + "eval_steps_per_second": 0.92, + "step": 240 + }, + { + "epoch": 0.24756034925526452, + "grad_norm": 3.411193609237671, + "learning_rate": 3.869863013698631e-06, + "loss": 1.9171, + "step": 241 + }, + { + "epoch": 0.24858757062146894, + "grad_norm": 2.8848962783813477, + "learning_rate": 3.886986301369863e-06, + "loss": 1.8587, + "step": 242 + }, + { + "epoch": 0.24961479198767333, + "grad_norm": 2.756298065185547, + "learning_rate": 3.904109589041096e-06, + "loss": 1.8614, + "step": 243 + }, + { + "epoch": 0.25064201335387776, + "grad_norm": 2.8048956394195557, + "learning_rate": 3.921232876712329e-06, + "loss": 1.8225, + "step": 244 + }, + { + "epoch": 0.2516692347200822, + "grad_norm": 2.9872403144836426, + "learning_rate": 3.938356164383562e-06, + "loss": 1.8367, + "step": 245 + }, + { + "epoch": 0.2526964560862866, + "grad_norm": 2.710156202316284, + "learning_rate": 3.9554794520547955e-06, + "loss": 1.809, + "step": 246 + }, + { + "epoch": 0.253723677452491, + "grad_norm": 3.2177000045776367, + "learning_rate": 3.972602739726027e-06, + "loss": 1.8883, + "step": 247 + }, + { + "epoch": 0.25475089881869545, + "grad_norm": 3.160964250564575, + "learning_rate": 3.989726027397261e-06, + "loss": 1.8267, + "step": 248 + }, + { + "epoch": 0.25577812018489987, + "grad_norm": 2.853498935699463, + "learning_rate": 4.006849315068493e-06, + "loss": 1.8852, + "step": 249 + }, + { + "epoch": 0.25680534155110424, + "grad_norm": 2.571742296218872, + "learning_rate": 4.023972602739726e-06, + "loss": 1.8833, + "step": 250 + }, + { + "epoch": 0.25680534155110424, + "eval_loss": 2.0988597869873047, + "eval_runtime": 47.7045, + "eval_samples_per_second": 20.962, + "eval_steps_per_second": 0.88, + "step": 250 + }, + { + "epoch": 0.25783256291730866, + "grad_norm": 2.776742458343506, + "learning_rate": 4.0410958904109595e-06, + "loss": 1.8662, + "step": 251 + }, + { + "epoch": 0.2588597842835131, + "grad_norm": 2.8038573265075684, + "learning_rate": 4.058219178082192e-06, + "loss": 1.8528, + "step": 252 + }, + { + "epoch": 0.2598870056497175, + "grad_norm": 2.557623863220215, + "learning_rate": 4.075342465753426e-06, + "loss": 1.8765, + "step": 253 + }, + { + "epoch": 0.26091422701592193, + "grad_norm": 3.0477073192596436, + "learning_rate": 4.092465753424658e-06, + "loss": 1.8434, + "step": 254 + }, + { + "epoch": 0.26194144838212635, + "grad_norm": 2.6221649646759033, + "learning_rate": 4.109589041095891e-06, + "loss": 1.8348, + "step": 255 + }, + { + "epoch": 0.2629686697483308, + "grad_norm": 3.0849292278289795, + "learning_rate": 4.1267123287671235e-06, + "loss": 1.9057, + "step": 256 + }, + { + "epoch": 0.2639958911145352, + "grad_norm": 2.795184373855591, + "learning_rate": 4.143835616438356e-06, + "loss": 1.8528, + "step": 257 + }, + { + "epoch": 0.2650231124807396, + "grad_norm": 2.887193441390991, + "learning_rate": 4.16095890410959e-06, + "loss": 1.9176, + "step": 258 + }, + { + "epoch": 0.26605033384694404, + "grad_norm": 2.776801347732544, + "learning_rate": 4.178082191780822e-06, + "loss": 1.8911, + "step": 259 + }, + { + "epoch": 0.2670775552131484, + "grad_norm": 2.980426788330078, + "learning_rate": 4.195205479452055e-06, + "loss": 1.8288, + "step": 260 + }, + { + "epoch": 0.2670775552131484, + "eval_loss": 2.0636370182037354, + "eval_runtime": 49.3081, + "eval_samples_per_second": 20.281, + "eval_steps_per_second": 0.852, + "step": 260 + }, + { + "epoch": 0.26810477657935283, + "grad_norm": 2.7921340465545654, + "learning_rate": 4.212328767123288e-06, + "loss": 1.8405, + "step": 261 + }, + { + "epoch": 0.26913199794555726, + "grad_norm": 2.67714262008667, + "learning_rate": 4.229452054794521e-06, + "loss": 1.8342, + "step": 262 + }, + { + "epoch": 0.2701592193117617, + "grad_norm": 2.819868326187134, + "learning_rate": 4.246575342465754e-06, + "loss": 1.8466, + "step": 263 + }, + { + "epoch": 0.2711864406779661, + "grad_norm": 2.611250638961792, + "learning_rate": 4.263698630136986e-06, + "loss": 1.8314, + "step": 264 + }, + { + "epoch": 0.2722136620441705, + "grad_norm": 3.0406076908111572, + "learning_rate": 4.28082191780822e-06, + "loss": 1.7924, + "step": 265 + }, + { + "epoch": 0.27324088341037495, + "grad_norm": 3.212790012359619, + "learning_rate": 4.297945205479452e-06, + "loss": 1.9479, + "step": 266 + }, + { + "epoch": 0.27426810477657937, + "grad_norm": 2.4767045974731445, + "learning_rate": 4.315068493150685e-06, + "loss": 1.835, + "step": 267 + }, + { + "epoch": 0.2752953261427838, + "grad_norm": 3.191835641860962, + "learning_rate": 4.3321917808219185e-06, + "loss": 1.8261, + "step": 268 + }, + { + "epoch": 0.2763225475089882, + "grad_norm": 2.9970273971557617, + "learning_rate": 4.349315068493151e-06, + "loss": 1.8427, + "step": 269 + }, + { + "epoch": 0.2773497688751926, + "grad_norm": 2.888843297958374, + "learning_rate": 4.3664383561643846e-06, + "loss": 1.819, + "step": 270 + }, + { + "epoch": 0.2773497688751926, + "eval_loss": 2.054966926574707, + "eval_runtime": 45.9212, + "eval_samples_per_second": 21.776, + "eval_steps_per_second": 0.915, + "step": 270 + }, + { + "epoch": 0.278376990241397, + "grad_norm": 2.7771849632263184, + "learning_rate": 4.383561643835616e-06, + "loss": 1.8178, + "step": 271 + }, + { + "epoch": 0.2794042116076014, + "grad_norm": 2.9749014377593994, + "learning_rate": 4.40068493150685e-06, + "loss": 1.8982, + "step": 272 + }, + { + "epoch": 0.28043143297380585, + "grad_norm": 3.076021432876587, + "learning_rate": 4.4178082191780825e-06, + "loss": 1.8575, + "step": 273 + }, + { + "epoch": 0.2814586543400103, + "grad_norm": 3.2113568782806396, + "learning_rate": 4.434931506849315e-06, + "loss": 1.8241, + "step": 274 + }, + { + "epoch": 0.2824858757062147, + "grad_norm": 3.322063684463501, + "learning_rate": 4.4520547945205486e-06, + "loss": 1.7922, + "step": 275 + }, + { + "epoch": 0.2835130970724191, + "grad_norm": 3.2778918743133545, + "learning_rate": 4.469178082191781e-06, + "loss": 1.8412, + "step": 276 + }, + { + "epoch": 0.28454031843862354, + "grad_norm": 3.1802618503570557, + "learning_rate": 4.486301369863014e-06, + "loss": 1.8283, + "step": 277 + }, + { + "epoch": 0.28556753980482796, + "grad_norm": 2.9214258193969727, + "learning_rate": 4.503424657534247e-06, + "loss": 1.8817, + "step": 278 + }, + { + "epoch": 0.28659476117103233, + "grad_norm": 2.8809854984283447, + "learning_rate": 4.52054794520548e-06, + "loss": 1.7662, + "step": 279 + }, + { + "epoch": 0.28762198253723675, + "grad_norm": 2.925537586212158, + "learning_rate": 4.537671232876713e-06, + "loss": 1.8147, + "step": 280 + }, + { + "epoch": 0.28762198253723675, + "eval_loss": 2.054513454437256, + "eval_runtime": 51.5036, + "eval_samples_per_second": 19.416, + "eval_steps_per_second": 0.815, + "step": 280 + }, + { + "epoch": 0.2886492039034412, + "grad_norm": 2.9421887397766113, + "learning_rate": 4.554794520547945e-06, + "loss": 1.8581, + "step": 281 + }, + { + "epoch": 0.2896764252696456, + "grad_norm": 2.9733567237854004, + "learning_rate": 4.571917808219179e-06, + "loss": 1.8703, + "step": 282 + }, + { + "epoch": 0.29070364663585, + "grad_norm": 2.5352156162261963, + "learning_rate": 4.589041095890411e-06, + "loss": 1.8505, + "step": 283 + }, + { + "epoch": 0.29173086800205444, + "grad_norm": 2.773847818374634, + "learning_rate": 4.606164383561644e-06, + "loss": 1.8114, + "step": 284 + }, + { + "epoch": 0.29275808936825887, + "grad_norm": 2.709632635116577, + "learning_rate": 4.6232876712328774e-06, + "loss": 1.815, + "step": 285 + }, + { + "epoch": 0.2937853107344633, + "grad_norm": 3.1234357357025146, + "learning_rate": 4.64041095890411e-06, + "loss": 1.8391, + "step": 286 + }, + { + "epoch": 0.2948125321006677, + "grad_norm": 2.9376144409179688, + "learning_rate": 4.657534246575343e-06, + "loss": 1.7998, + "step": 287 + }, + { + "epoch": 0.29583975346687214, + "grad_norm": 3.0812032222747803, + "learning_rate": 4.674657534246575e-06, + "loss": 1.8462, + "step": 288 + }, + { + "epoch": 0.2968669748330765, + "grad_norm": 2.922722578048706, + "learning_rate": 4.691780821917809e-06, + "loss": 1.8263, + "step": 289 + }, + { + "epoch": 0.2978941961992809, + "grad_norm": 2.9484992027282715, + "learning_rate": 4.7089041095890414e-06, + "loss": 1.8754, + "step": 290 + }, + { + "epoch": 0.2978941961992809, + "eval_loss": 2.032902240753174, + "eval_runtime": 47.8711, + "eval_samples_per_second": 20.889, + "eval_steps_per_second": 0.877, + "step": 290 + }, + { + "epoch": 0.29892141756548535, + "grad_norm": 3.0068976879119873, + "learning_rate": 4.726027397260274e-06, + "loss": 1.8256, + "step": 291 + }, + { + "epoch": 0.29994863893168977, + "grad_norm": 2.6469006538391113, + "learning_rate": 4.7431506849315075e-06, + "loss": 1.8354, + "step": 292 + }, + { + "epoch": 0.3009758602978942, + "grad_norm": 3.06986927986145, + "learning_rate": 4.76027397260274e-06, + "loss": 1.8402, + "step": 293 + }, + { + "epoch": 0.3020030816640986, + "grad_norm": 2.8409981727600098, + "learning_rate": 4.777397260273973e-06, + "loss": 1.7915, + "step": 294 + }, + { + "epoch": 0.30303030303030304, + "grad_norm": 3.1027376651763916, + "learning_rate": 4.7945205479452054e-06, + "loss": 1.8153, + "step": 295 + }, + { + "epoch": 0.30405752439650746, + "grad_norm": 2.7221434116363525, + "learning_rate": 4.811643835616439e-06, + "loss": 1.8098, + "step": 296 + }, + { + "epoch": 0.3050847457627119, + "grad_norm": 3.0559370517730713, + "learning_rate": 4.8287671232876716e-06, + "loss": 1.7504, + "step": 297 + }, + { + "epoch": 0.3061119671289163, + "grad_norm": 2.9628236293792725, + "learning_rate": 4.845890410958904e-06, + "loss": 1.8019, + "step": 298 + }, + { + "epoch": 0.3071391884951207, + "grad_norm": 2.6248221397399902, + "learning_rate": 4.863013698630138e-06, + "loss": 1.8001, + "step": 299 + }, + { + "epoch": 0.3081664098613251, + "grad_norm": 3.0477702617645264, + "learning_rate": 4.88013698630137e-06, + "loss": 1.7915, + "step": 300 + }, + { + "epoch": 0.3081664098613251, + "eval_loss": 2.0522429943084717, + "eval_runtime": 46.796, + "eval_samples_per_second": 21.369, + "eval_steps_per_second": 0.898, + "step": 300 + }, + { + "epoch": 0.3091936312275295, + "grad_norm": 2.784707546234131, + "learning_rate": 4.897260273972603e-06, + "loss": 1.7757, + "step": 301 + }, + { + "epoch": 0.31022085259373394, + "grad_norm": 2.8283841609954834, + "learning_rate": 4.914383561643836e-06, + "loss": 1.8975, + "step": 302 + }, + { + "epoch": 0.31124807395993837, + "grad_norm": 2.879342555999756, + "learning_rate": 4.931506849315069e-06, + "loss": 1.8208, + "step": 303 + }, + { + "epoch": 0.3122752953261428, + "grad_norm": 3.2111175060272217, + "learning_rate": 4.948630136986302e-06, + "loss": 1.8631, + "step": 304 + }, + { + "epoch": 0.3133025166923472, + "grad_norm": 2.876168966293335, + "learning_rate": 4.965753424657534e-06, + "loss": 1.8863, + "step": 305 + }, + { + "epoch": 0.31432973805855163, + "grad_norm": 2.8958466053009033, + "learning_rate": 4.982876712328768e-06, + "loss": 1.7565, + "step": 306 + }, + { + "epoch": 0.31535695942475606, + "grad_norm": 2.6702723503112793, + "learning_rate": 5e-06, + "loss": 1.7844, + "step": 307 + }, + { + "epoch": 0.3163841807909605, + "grad_norm": 2.557966470718384, + "learning_rate": 4.998096688237534e-06, + "loss": 1.7472, + "step": 308 + }, + { + "epoch": 0.31741140215716485, + "grad_norm": 2.9257633686065674, + "learning_rate": 4.996193376475067e-06, + "loss": 1.853, + "step": 309 + }, + { + "epoch": 0.31843862352336927, + "grad_norm": 2.497093439102173, + "learning_rate": 4.9942900647126005e-06, + "loss": 1.7313, + "step": 310 + }, + { + "epoch": 0.31843862352336927, + "eval_loss": 2.0336520671844482, + "eval_runtime": 43.3773, + "eval_samples_per_second": 23.054, + "eval_steps_per_second": 0.968, + "step": 310 + }, + { + "epoch": 0.3194658448895737, + "grad_norm": 3.0313165187835693, + "learning_rate": 4.992386752950134e-06, + "loss": 1.7397, + "step": 311 + }, + { + "epoch": 0.3204930662557781, + "grad_norm": 2.629685163497925, + "learning_rate": 4.990483441187667e-06, + "loss": 1.7839, + "step": 312 + }, + { + "epoch": 0.32152028762198254, + "grad_norm": 2.6772103309631348, + "learning_rate": 4.9885801294252e-06, + "loss": 1.8038, + "step": 313 + }, + { + "epoch": 0.32254750898818696, + "grad_norm": 2.893958568572998, + "learning_rate": 4.986676817662733e-06, + "loss": 1.8072, + "step": 314 + }, + { + "epoch": 0.3235747303543914, + "grad_norm": 2.5317318439483643, + "learning_rate": 4.9847735059002665e-06, + "loss": 1.7632, + "step": 315 + }, + { + "epoch": 0.3246019517205958, + "grad_norm": 2.836867332458496, + "learning_rate": 4.9828701941378e-06, + "loss": 1.8499, + "step": 316 + }, + { + "epoch": 0.3256291730868002, + "grad_norm": 2.5564308166503906, + "learning_rate": 4.980966882375333e-06, + "loss": 1.7849, + "step": 317 + }, + { + "epoch": 0.3266563944530046, + "grad_norm": 2.4335806369781494, + "learning_rate": 4.979063570612867e-06, + "loss": 1.7425, + "step": 318 + }, + { + "epoch": 0.327683615819209, + "grad_norm": 2.7140371799468994, + "learning_rate": 4.9771602588504e-06, + "loss": 1.7173, + "step": 319 + }, + { + "epoch": 0.32871083718541344, + "grad_norm": 2.574345111846924, + "learning_rate": 4.975256947087933e-06, + "loss": 1.8307, + "step": 320 + }, + { + "epoch": 0.32871083718541344, + "eval_loss": 2.026960611343384, + "eval_runtime": 53.8982, + "eval_samples_per_second": 18.553, + "eval_steps_per_second": 0.779, + "step": 320 + }, + { + "epoch": 0.32973805855161786, + "grad_norm": 2.6877281665802, + "learning_rate": 4.973353635325467e-06, + "loss": 1.7216, + "step": 321 + }, + { + "epoch": 0.3307652799178223, + "grad_norm": 2.8152849674224854, + "learning_rate": 4.971450323563e-06, + "loss": 1.7831, + "step": 322 + }, + { + "epoch": 0.3317925012840267, + "grad_norm": 2.4208791255950928, + "learning_rate": 4.969547011800533e-06, + "loss": 1.7266, + "step": 323 + }, + { + "epoch": 0.33281972265023113, + "grad_norm": 2.765162229537964, + "learning_rate": 4.967643700038067e-06, + "loss": 1.7971, + "step": 324 + }, + { + "epoch": 0.33384694401643555, + "grad_norm": 2.5018413066864014, + "learning_rate": 4.9657403882756e-06, + "loss": 1.7842, + "step": 325 + }, + { + "epoch": 0.33487416538264, + "grad_norm": 2.901268482208252, + "learning_rate": 4.9638370765131335e-06, + "loss": 1.7522, + "step": 326 + }, + { + "epoch": 0.3359013867488444, + "grad_norm": 2.40067195892334, + "learning_rate": 4.961933764750667e-06, + "loss": 1.7371, + "step": 327 + }, + { + "epoch": 0.33692860811504877, + "grad_norm": 2.6963725090026855, + "learning_rate": 4.9600304529882e-06, + "loss": 1.8036, + "step": 328 + }, + { + "epoch": 0.3379558294812532, + "grad_norm": 2.483790397644043, + "learning_rate": 4.9581271412257335e-06, + "loss": 1.815, + "step": 329 + }, + { + "epoch": 0.3389830508474576, + "grad_norm": 2.7459559440612793, + "learning_rate": 4.956223829463266e-06, + "loss": 1.777, + "step": 330 + }, + { + "epoch": 0.3389830508474576, + "eval_loss": 2.046339988708496, + "eval_runtime": 53.909, + "eval_samples_per_second": 18.55, + "eval_steps_per_second": 0.779, + "step": 330 + }, + { + "epoch": 0.34001027221366203, + "grad_norm": 2.5724353790283203, + "learning_rate": 4.954320517700799e-06, + "loss": 1.7022, + "step": 331 + }, + { + "epoch": 0.34103749357986646, + "grad_norm": 2.510002613067627, + "learning_rate": 4.952417205938333e-06, + "loss": 1.802, + "step": 332 + }, + { + "epoch": 0.3420647149460709, + "grad_norm": 2.8722083568573, + "learning_rate": 4.950513894175866e-06, + "loss": 1.7772, + "step": 333 + }, + { + "epoch": 0.3430919363122753, + "grad_norm": 2.586747407913208, + "learning_rate": 4.9486105824133995e-06, + "loss": 1.7664, + "step": 334 + }, + { + "epoch": 0.3441191576784797, + "grad_norm": 2.8008205890655518, + "learning_rate": 4.946707270650933e-06, + "loss": 1.7892, + "step": 335 + }, + { + "epoch": 0.34514637904468415, + "grad_norm": 2.632664203643799, + "learning_rate": 4.944803958888466e-06, + "loss": 1.7967, + "step": 336 + }, + { + "epoch": 0.34617360041088857, + "grad_norm": 2.8898301124572754, + "learning_rate": 4.942900647126e-06, + "loss": 1.8126, + "step": 337 + }, + { + "epoch": 0.34720082177709294, + "grad_norm": 2.719019651412964, + "learning_rate": 4.940997335363533e-06, + "loss": 1.7544, + "step": 338 + }, + { + "epoch": 0.34822804314329736, + "grad_norm": 2.9995970726013184, + "learning_rate": 4.939094023601066e-06, + "loss": 1.7142, + "step": 339 + }, + { + "epoch": 0.3492552645095018, + "grad_norm": 2.9737157821655273, + "learning_rate": 4.9371907118386e-06, + "loss": 1.7659, + "step": 340 + }, + { + "epoch": 0.3492552645095018, + "eval_loss": 2.024946928024292, + "eval_runtime": 52.8069, + "eval_samples_per_second": 18.937, + "eval_steps_per_second": 0.795, + "step": 340 + }, + { + "epoch": 0.3502824858757062, + "grad_norm": 2.8651719093322754, + "learning_rate": 4.935287400076133e-06, + "loss": 1.7767, + "step": 341 + }, + { + "epoch": 0.35130970724191063, + "grad_norm": 2.6293981075286865, + "learning_rate": 4.933384088313666e-06, + "loss": 1.7815, + "step": 342 + }, + { + "epoch": 0.35233692860811505, + "grad_norm": 3.230978012084961, + "learning_rate": 4.9314807765512e-06, + "loss": 1.7423, + "step": 343 + }, + { + "epoch": 0.3533641499743195, + "grad_norm": 3.228300094604492, + "learning_rate": 4.929577464788733e-06, + "loss": 1.6845, + "step": 344 + }, + { + "epoch": 0.3543913713405239, + "grad_norm": 3.036733388900757, + "learning_rate": 4.9276741530262665e-06, + "loss": 1.7524, + "step": 345 + }, + { + "epoch": 0.3554185927067283, + "grad_norm": 2.778538227081299, + "learning_rate": 4.9257708412638e-06, + "loss": 1.7433, + "step": 346 + }, + { + "epoch": 0.35644581407293274, + "grad_norm": 3.0958566665649414, + "learning_rate": 4.923867529501332e-06, + "loss": 1.76, + "step": 347 + }, + { + "epoch": 0.3574730354391371, + "grad_norm": 3.1120026111602783, + "learning_rate": 4.921964217738866e-06, + "loss": 1.7399, + "step": 348 + }, + { + "epoch": 0.35850025680534153, + "grad_norm": 2.70768141746521, + "learning_rate": 4.920060905976399e-06, + "loss": 1.8169, + "step": 349 + }, + { + "epoch": 0.35952747817154596, + "grad_norm": 2.917478561401367, + "learning_rate": 4.9181575942139325e-06, + "loss": 1.7948, + "step": 350 + }, + { + "epoch": 0.35952747817154596, + "eval_loss": 2.0208380222320557, + "eval_runtime": 47.4243, + "eval_samples_per_second": 21.086, + "eval_steps_per_second": 0.886, + "step": 350 + }, + { + "epoch": 0.3605546995377504, + "grad_norm": 3.0564756393432617, + "learning_rate": 4.916254282451466e-06, + "loss": 1.7861, + "step": 351 + }, + { + "epoch": 0.3615819209039548, + "grad_norm": 2.709195137023926, + "learning_rate": 4.914350970688999e-06, + "loss": 1.8323, + "step": 352 + }, + { + "epoch": 0.3626091422701592, + "grad_norm": 2.8797669410705566, + "learning_rate": 4.9124476589265325e-06, + "loss": 1.722, + "step": 353 + }, + { + "epoch": 0.36363636363636365, + "grad_norm": 2.560702085494995, + "learning_rate": 4.910544347164066e-06, + "loss": 1.7289, + "step": 354 + }, + { + "epoch": 0.36466358500256807, + "grad_norm": 3.3781847953796387, + "learning_rate": 4.908641035401599e-06, + "loss": 1.8096, + "step": 355 + }, + { + "epoch": 0.3656908063687725, + "grad_norm": 2.582310676574707, + "learning_rate": 4.906737723639133e-06, + "loss": 1.8752, + "step": 356 + }, + { + "epoch": 0.3667180277349769, + "grad_norm": 3.506385564804077, + "learning_rate": 4.904834411876665e-06, + "loss": 1.7404, + "step": 357 + }, + { + "epoch": 0.3677452491011813, + "grad_norm": 2.4360411167144775, + "learning_rate": 4.902931100114199e-06, + "loss": 1.7812, + "step": 358 + }, + { + "epoch": 0.3687724704673857, + "grad_norm": 3.77639102935791, + "learning_rate": 4.901027788351733e-06, + "loss": 1.7103, + "step": 359 + }, + { + "epoch": 0.3697996918335901, + "grad_norm": 2.887723922729492, + "learning_rate": 4.899124476589266e-06, + "loss": 1.7485, + "step": 360 + }, + { + "epoch": 0.3697996918335901, + "eval_loss": 2.0298168659210205, + "eval_runtime": 46.0993, + "eval_samples_per_second": 21.692, + "eval_steps_per_second": 0.911, + "step": 360 + }, + { + "epoch": 0.37082691319979455, + "grad_norm": 3.5861997604370117, + "learning_rate": 4.8972211648267994e-06, + "loss": 1.7189, + "step": 361 + }, + { + "epoch": 0.371854134565999, + "grad_norm": 2.590808868408203, + "learning_rate": 4.895317853064333e-06, + "loss": 1.7683, + "step": 362 + }, + { + "epoch": 0.3728813559322034, + "grad_norm": 3.4451487064361572, + "learning_rate": 4.893414541301866e-06, + "loss": 1.7436, + "step": 363 + }, + { + "epoch": 0.3739085772984078, + "grad_norm": 2.406073808670044, + "learning_rate": 4.891511229539399e-06, + "loss": 1.7378, + "step": 364 + }, + { + "epoch": 0.37493579866461224, + "grad_norm": 2.7162997722625732, + "learning_rate": 4.889607917776932e-06, + "loss": 1.7558, + "step": 365 + }, + { + "epoch": 0.37596302003081666, + "grad_norm": 2.564753770828247, + "learning_rate": 4.887704606014465e-06, + "loss": 1.7559, + "step": 366 + }, + { + "epoch": 0.37699024139702103, + "grad_norm": 2.7196900844573975, + "learning_rate": 4.885801294251999e-06, + "loss": 1.7683, + "step": 367 + }, + { + "epoch": 0.37801746276322545, + "grad_norm": 2.9334428310394287, + "learning_rate": 4.883897982489532e-06, + "loss": 1.6932, + "step": 368 + }, + { + "epoch": 0.3790446841294299, + "grad_norm": 2.4899954795837402, + "learning_rate": 4.8819946707270655e-06, + "loss": 1.7943, + "step": 369 + }, + { + "epoch": 0.3800719054956343, + "grad_norm": 2.820594310760498, + "learning_rate": 4.880091358964599e-06, + "loss": 1.7729, + "step": 370 + }, + { + "epoch": 0.3800719054956343, + "eval_loss": 2.0204291343688965, + "eval_runtime": 44.9993, + "eval_samples_per_second": 22.223, + "eval_steps_per_second": 0.933, + "step": 370 + }, + { + "epoch": 0.3810991268618387, + "grad_norm": 2.4717929363250732, + "learning_rate": 4.878188047202132e-06, + "loss": 1.6813, + "step": 371 + }, + { + "epoch": 0.38212634822804314, + "grad_norm": 3.1815855503082275, + "learning_rate": 4.876284735439666e-06, + "loss": 1.8588, + "step": 372 + }, + { + "epoch": 0.38315356959424757, + "grad_norm": 2.7330422401428223, + "learning_rate": 4.874381423677199e-06, + "loss": 1.8245, + "step": 373 + }, + { + "epoch": 0.384180790960452, + "grad_norm": 3.129444122314453, + "learning_rate": 4.8724781119147315e-06, + "loss": 1.7494, + "step": 374 + }, + { + "epoch": 0.3852080123266564, + "grad_norm": 2.75551700592041, + "learning_rate": 4.870574800152265e-06, + "loss": 1.7761, + "step": 375 + }, + { + "epoch": 0.38623523369286084, + "grad_norm": 2.9313466548919678, + "learning_rate": 4.868671488389798e-06, + "loss": 1.7076, + "step": 376 + }, + { + "epoch": 0.3872624550590652, + "grad_norm": 2.7590556144714355, + "learning_rate": 4.8667681766273316e-06, + "loss": 1.7105, + "step": 377 + }, + { + "epoch": 0.3882896764252696, + "grad_norm": 2.8224680423736572, + "learning_rate": 4.864864864864866e-06, + "loss": 1.6853, + "step": 378 + }, + { + "epoch": 0.38931689779147405, + "grad_norm": 2.5148062705993652, + "learning_rate": 4.862961553102399e-06, + "loss": 1.7477, + "step": 379 + }, + { + "epoch": 0.39034411915767847, + "grad_norm": 3.144247055053711, + "learning_rate": 4.8610582413399325e-06, + "loss": 1.7726, + "step": 380 + }, + { + "epoch": 0.39034411915767847, + "eval_loss": 2.006993532180786, + "eval_runtime": 50.8909, + "eval_samples_per_second": 19.65, + "eval_steps_per_second": 0.825, + "step": 380 + }, + { + "epoch": 0.3913713405238829, + "grad_norm": 2.6818125247955322, + "learning_rate": 4.859154929577465e-06, + "loss": 1.7351, + "step": 381 + }, + { + "epoch": 0.3923985618900873, + "grad_norm": 2.8543903827667236, + "learning_rate": 4.857251617814998e-06, + "loss": 1.7499, + "step": 382 + }, + { + "epoch": 0.39342578325629174, + "grad_norm": 2.5483179092407227, + "learning_rate": 4.855348306052532e-06, + "loss": 1.7849, + "step": 383 + }, + { + "epoch": 0.39445300462249616, + "grad_norm": 2.7169013023376465, + "learning_rate": 4.853444994290065e-06, + "loss": 1.7664, + "step": 384 + }, + { + "epoch": 0.3954802259887006, + "grad_norm": 2.8304696083068848, + "learning_rate": 4.8515416825275985e-06, + "loss": 1.73, + "step": 385 + }, + { + "epoch": 0.396507447354905, + "grad_norm": 2.4217448234558105, + "learning_rate": 4.849638370765132e-06, + "loss": 1.7915, + "step": 386 + }, + { + "epoch": 0.3975346687211094, + "grad_norm": 2.523256301879883, + "learning_rate": 4.847735059002665e-06, + "loss": 1.7195, + "step": 387 + }, + { + "epoch": 0.3985618900873138, + "grad_norm": 2.813638925552368, + "learning_rate": 4.8458317472401985e-06, + "loss": 1.6766, + "step": 388 + }, + { + "epoch": 0.3995891114535182, + "grad_norm": 2.6701548099517822, + "learning_rate": 4.843928435477732e-06, + "loss": 1.7666, + "step": 389 + }, + { + "epoch": 0.40061633281972264, + "grad_norm": 2.5568342208862305, + "learning_rate": 4.842025123715265e-06, + "loss": 1.723, + "step": 390 + }, + { + "epoch": 0.40061633281972264, + "eval_loss": 2.0098180770874023, + "eval_runtime": 44.4503, + "eval_samples_per_second": 22.497, + "eval_steps_per_second": 0.945, + "step": 390 + }, + { + "epoch": 0.40164355418592707, + "grad_norm": 2.6287341117858887, + "learning_rate": 4.840121811952798e-06, + "loss": 1.759, + "step": 391 + }, + { + "epoch": 0.4026707755521315, + "grad_norm": 2.934105157852173, + "learning_rate": 4.838218500190331e-06, + "loss": 1.7501, + "step": 392 + }, + { + "epoch": 0.4036979969183359, + "grad_norm": 2.6563878059387207, + "learning_rate": 4.8363151884278645e-06, + "loss": 1.7624, + "step": 393 + }, + { + "epoch": 0.40472521828454033, + "grad_norm": 2.538278102874756, + "learning_rate": 4.834411876665398e-06, + "loss": 1.7457, + "step": 394 + }, + { + "epoch": 0.40575243965074476, + "grad_norm": 2.801630735397339, + "learning_rate": 4.832508564902931e-06, + "loss": 1.7549, + "step": 395 + }, + { + "epoch": 0.4067796610169492, + "grad_norm": 2.571937322616577, + "learning_rate": 4.830605253140465e-06, + "loss": 1.6954, + "step": 396 + }, + { + "epoch": 0.40780688238315355, + "grad_norm": 2.5652592182159424, + "learning_rate": 4.828701941377998e-06, + "loss": 1.7328, + "step": 397 + }, + { + "epoch": 0.40883410374935797, + "grad_norm": 2.6840121746063232, + "learning_rate": 4.826798629615531e-06, + "loss": 1.6934, + "step": 398 + }, + { + "epoch": 0.4098613251155624, + "grad_norm": 3.0103495121002197, + "learning_rate": 4.824895317853065e-06, + "loss": 1.7086, + "step": 399 + }, + { + "epoch": 0.4108885464817668, + "grad_norm": 2.7361223697662354, + "learning_rate": 4.822992006090598e-06, + "loss": 1.7301, + "step": 400 + }, + { + "epoch": 0.4108885464817668, + "eval_loss": 1.9975080490112305, + "eval_runtime": 44.5413, + "eval_samples_per_second": 22.451, + "eval_steps_per_second": 0.943, + "step": 400 + }, + { + "epoch": 0.41191576784797124, + "grad_norm": 2.6805901527404785, + "learning_rate": 4.821088694328131e-06, + "loss": 1.6896, + "step": 401 + }, + { + "epoch": 0.41294298921417566, + "grad_norm": 3.231706142425537, + "learning_rate": 4.819185382565665e-06, + "loss": 1.7134, + "step": 402 + }, + { + "epoch": 0.4139702105803801, + "grad_norm": 2.5561299324035645, + "learning_rate": 4.817282070803198e-06, + "loss": 1.764, + "step": 403 + }, + { + "epoch": 0.4149974319465845, + "grad_norm": 3.3125340938568115, + "learning_rate": 4.8153787590407315e-06, + "loss": 1.76, + "step": 404 + }, + { + "epoch": 0.41602465331278893, + "grad_norm": 2.750051498413086, + "learning_rate": 4.813475447278265e-06, + "loss": 1.6882, + "step": 405 + }, + { + "epoch": 0.41705187467899335, + "grad_norm": 4.006515979766846, + "learning_rate": 4.811572135515798e-06, + "loss": 1.7892, + "step": 406 + }, + { + "epoch": 0.4180790960451977, + "grad_norm": 2.5276904106140137, + "learning_rate": 4.809668823753332e-06, + "loss": 1.6995, + "step": 407 + }, + { + "epoch": 0.41910631741140214, + "grad_norm": 2.959444046020508, + "learning_rate": 4.807765511990864e-06, + "loss": 1.7058, + "step": 408 + }, + { + "epoch": 0.42013353877760656, + "grad_norm": 2.8472464084625244, + "learning_rate": 4.8058622002283975e-06, + "loss": 1.7258, + "step": 409 + }, + { + "epoch": 0.421160760143811, + "grad_norm": 3.0240273475646973, + "learning_rate": 4.803958888465931e-06, + "loss": 1.7789, + "step": 410 + }, + { + "epoch": 0.421160760143811, + "eval_loss": 1.9919871091842651, + "eval_runtime": 41.727, + "eval_samples_per_second": 23.965, + "eval_steps_per_second": 1.007, + "step": 410 + }, + { + "epoch": 0.4221879815100154, + "grad_norm": 2.6211893558502197, + "learning_rate": 4.802055576703464e-06, + "loss": 1.7315, + "step": 411 + }, + { + "epoch": 0.42321520287621983, + "grad_norm": 2.6072895526885986, + "learning_rate": 4.8001522649409976e-06, + "loss": 1.8176, + "step": 412 + }, + { + "epoch": 0.42424242424242425, + "grad_norm": 2.9197702407836914, + "learning_rate": 4.798248953178531e-06, + "loss": 1.6947, + "step": 413 + }, + { + "epoch": 0.4252696456086287, + "grad_norm": 2.4383556842803955, + "learning_rate": 4.796345641416064e-06, + "loss": 1.6728, + "step": 414 + }, + { + "epoch": 0.4262968669748331, + "grad_norm": 2.9951915740966797, + "learning_rate": 4.794442329653598e-06, + "loss": 1.7249, + "step": 415 + }, + { + "epoch": 0.42732408834103747, + "grad_norm": 2.6095519065856934, + "learning_rate": 4.792539017891131e-06, + "loss": 1.7735, + "step": 416 + }, + { + "epoch": 0.4283513097072419, + "grad_norm": 2.6429390907287598, + "learning_rate": 4.790635706128664e-06, + "loss": 1.704, + "step": 417 + }, + { + "epoch": 0.4293785310734463, + "grad_norm": 2.921812057495117, + "learning_rate": 4.788732394366197e-06, + "loss": 1.7355, + "step": 418 + }, + { + "epoch": 0.43040575243965074, + "grad_norm": 2.6050965785980225, + "learning_rate": 4.786829082603731e-06, + "loss": 1.6636, + "step": 419 + }, + { + "epoch": 0.43143297380585516, + "grad_norm": 2.672070264816284, + "learning_rate": 4.7849257708412645e-06, + "loss": 1.6627, + "step": 420 + }, + { + "epoch": 0.43143297380585516, + "eval_loss": 1.9990200996398926, + "eval_runtime": 45.0672, + "eval_samples_per_second": 22.189, + "eval_steps_per_second": 0.932, + "step": 420 + }, + { + "epoch": 0.4324601951720596, + "grad_norm": 2.6813488006591797, + "learning_rate": 4.783022459078798e-06, + "loss": 1.7453, + "step": 421 + }, + { + "epoch": 0.433487416538264, + "grad_norm": 2.6587321758270264, + "learning_rate": 4.781119147316331e-06, + "loss": 1.6795, + "step": 422 + }, + { + "epoch": 0.4345146379044684, + "grad_norm": 2.827758312225342, + "learning_rate": 4.7792158355538645e-06, + "loss": 1.7098, + "step": 423 + }, + { + "epoch": 0.43554185927067285, + "grad_norm": 2.8615901470184326, + "learning_rate": 4.777312523791398e-06, + "loss": 1.7171, + "step": 424 + }, + { + "epoch": 0.43656908063687727, + "grad_norm": 2.9509294033050537, + "learning_rate": 4.77540921202893e-06, + "loss": 1.7018, + "step": 425 + }, + { + "epoch": 0.43759630200308164, + "grad_norm": 2.7994418144226074, + "learning_rate": 4.773505900266464e-06, + "loss": 1.7328, + "step": 426 + }, + { + "epoch": 0.43862352336928606, + "grad_norm": 2.7476799488067627, + "learning_rate": 4.771602588503997e-06, + "loss": 1.7116, + "step": 427 + }, + { + "epoch": 0.4396507447354905, + "grad_norm": 2.7955048084259033, + "learning_rate": 4.7696992767415305e-06, + "loss": 1.7187, + "step": 428 + }, + { + "epoch": 0.4406779661016949, + "grad_norm": 2.4326071739196777, + "learning_rate": 4.767795964979064e-06, + "loss": 1.75, + "step": 429 + }, + { + "epoch": 0.44170518746789933, + "grad_norm": 2.7708163261413574, + "learning_rate": 4.765892653216597e-06, + "loss": 1.7562, + "step": 430 + }, + { + "epoch": 0.44170518746789933, + "eval_loss": 2.0262997150421143, + "eval_runtime": 46.1649, + "eval_samples_per_second": 21.661, + "eval_steps_per_second": 0.91, + "step": 430 + }, + { + "epoch": 0.44273240883410375, + "grad_norm": 2.872743606567383, + "learning_rate": 4.763989341454131e-06, + "loss": 1.7527, + "step": 431 + }, + { + "epoch": 0.4437596302003082, + "grad_norm": 2.6387035846710205, + "learning_rate": 4.762086029691664e-06, + "loss": 1.6742, + "step": 432 + }, + { + "epoch": 0.4447868515665126, + "grad_norm": 2.6522979736328125, + "learning_rate": 4.760182717929197e-06, + "loss": 1.6811, + "step": 433 + }, + { + "epoch": 0.445814072932717, + "grad_norm": 2.644155263900757, + "learning_rate": 4.758279406166731e-06, + "loss": 1.7083, + "step": 434 + }, + { + "epoch": 0.44684129429892144, + "grad_norm": 2.530224561691284, + "learning_rate": 4.756376094404263e-06, + "loss": 1.6318, + "step": 435 + }, + { + "epoch": 0.4478685156651258, + "grad_norm": 2.431049108505249, + "learning_rate": 4.7544727826417966e-06, + "loss": 1.7162, + "step": 436 + }, + { + "epoch": 0.44889573703133023, + "grad_norm": 2.8091282844543457, + "learning_rate": 4.75256947087933e-06, + "loss": 1.7717, + "step": 437 + }, + { + "epoch": 0.44992295839753466, + "grad_norm": 2.5520501136779785, + "learning_rate": 4.750666159116863e-06, + "loss": 1.7201, + "step": 438 + }, + { + "epoch": 0.4509501797637391, + "grad_norm": 2.596370220184326, + "learning_rate": 4.7487628473543975e-06, + "loss": 1.6882, + "step": 439 + }, + { + "epoch": 0.4519774011299435, + "grad_norm": 2.667034387588501, + "learning_rate": 4.746859535591931e-06, + "loss": 1.6499, + "step": 440 + }, + { + "epoch": 0.4519774011299435, + "eval_loss": 2.008521556854248, + "eval_runtime": 49.3626, + "eval_samples_per_second": 20.258, + "eval_steps_per_second": 0.851, + "step": 440 + }, + { + "epoch": 0.4530046224961479, + "grad_norm": 2.636491537094116, + "learning_rate": 4.744956223829464e-06, + "loss": 1.6987, + "step": 441 + }, + { + "epoch": 0.45403184386235235, + "grad_norm": 2.4653050899505615, + "learning_rate": 4.743052912066997e-06, + "loss": 1.6775, + "step": 442 + }, + { + "epoch": 0.45505906522855677, + "grad_norm": 2.3866467475891113, + "learning_rate": 4.74114960030453e-06, + "loss": 1.7234, + "step": 443 + }, + { + "epoch": 0.4560862865947612, + "grad_norm": 2.675262928009033, + "learning_rate": 4.7392462885420635e-06, + "loss": 1.7363, + "step": 444 + }, + { + "epoch": 0.4571135079609656, + "grad_norm": 2.6233296394348145, + "learning_rate": 4.737342976779597e-06, + "loss": 1.6236, + "step": 445 + }, + { + "epoch": 0.45814072932717, + "grad_norm": 2.5638058185577393, + "learning_rate": 4.73543966501713e-06, + "loss": 1.7097, + "step": 446 + }, + { + "epoch": 0.4591679506933744, + "grad_norm": 2.7206926345825195, + "learning_rate": 4.7335363532546636e-06, + "loss": 1.6909, + "step": 447 + }, + { + "epoch": 0.4601951720595788, + "grad_norm": 2.546241283416748, + "learning_rate": 4.731633041492197e-06, + "loss": 1.6951, + "step": 448 + }, + { + "epoch": 0.46122239342578325, + "grad_norm": 2.6549105644226074, + "learning_rate": 4.72972972972973e-06, + "loss": 1.7052, + "step": 449 + }, + { + "epoch": 0.4622496147919877, + "grad_norm": 2.6649065017700195, + "learning_rate": 4.727826417967264e-06, + "loss": 1.7406, + "step": 450 + }, + { + "epoch": 0.4622496147919877, + "eval_loss": 1.9886436462402344, + "eval_runtime": 47.48, + "eval_samples_per_second": 21.061, + "eval_steps_per_second": 0.885, + "step": 450 + }, + { + "epoch": 0.4632768361581921, + "grad_norm": 2.7138187885284424, + "learning_rate": 4.725923106204797e-06, + "loss": 1.7154, + "step": 451 + }, + { + "epoch": 0.4643040575243965, + "grad_norm": 2.6095476150512695, + "learning_rate": 4.72401979444233e-06, + "loss": 1.686, + "step": 452 + }, + { + "epoch": 0.46533127889060094, + "grad_norm": 2.866363048553467, + "learning_rate": 4.722116482679863e-06, + "loss": 1.721, + "step": 453 + }, + { + "epoch": 0.46635850025680536, + "grad_norm": 2.4947352409362793, + "learning_rate": 4.720213170917396e-06, + "loss": 1.708, + "step": 454 + }, + { + "epoch": 0.46738572162300973, + "grad_norm": 2.3888256549835205, + "learning_rate": 4.71830985915493e-06, + "loss": 1.7356, + "step": 455 + }, + { + "epoch": 0.46841294298921415, + "grad_norm": 2.7065343856811523, + "learning_rate": 4.716406547392463e-06, + "loss": 1.6943, + "step": 456 + }, + { + "epoch": 0.4694401643554186, + "grad_norm": 2.424285888671875, + "learning_rate": 4.714503235629996e-06, + "loss": 1.7385, + "step": 457 + }, + { + "epoch": 0.470467385721623, + "grad_norm": 2.4175925254821777, + "learning_rate": 4.71259992386753e-06, + "loss": 1.6654, + "step": 458 + }, + { + "epoch": 0.4714946070878274, + "grad_norm": 2.3203656673431396, + "learning_rate": 4.710696612105063e-06, + "loss": 1.6977, + "step": 459 + }, + { + "epoch": 0.47252182845403184, + "grad_norm": 2.6660172939300537, + "learning_rate": 4.708793300342596e-06, + "loss": 1.7055, + "step": 460 + }, + { + "epoch": 0.47252182845403184, + "eval_loss": 1.9864659309387207, + "eval_runtime": 47.6232, + "eval_samples_per_second": 20.998, + "eval_steps_per_second": 0.882, + "step": 460 + }, + { + "epoch": 0.47354904982023627, + "grad_norm": 2.515615701675415, + "learning_rate": 4.70688998858013e-06, + "loss": 1.689, + "step": 461 + }, + { + "epoch": 0.4745762711864407, + "grad_norm": 2.429264783859253, + "learning_rate": 4.704986676817663e-06, + "loss": 1.6796, + "step": 462 + }, + { + "epoch": 0.4756034925526451, + "grad_norm": 2.469543933868408, + "learning_rate": 4.7030833650551965e-06, + "loss": 1.7164, + "step": 463 + }, + { + "epoch": 0.47663071391884954, + "grad_norm": 2.5921683311462402, + "learning_rate": 4.70118005329273e-06, + "loss": 1.6746, + "step": 464 + }, + { + "epoch": 0.4776579352850539, + "grad_norm": 2.6612918376922607, + "learning_rate": 4.699276741530263e-06, + "loss": 1.7598, + "step": 465 + }, + { + "epoch": 0.4786851566512583, + "grad_norm": 2.4633305072784424, + "learning_rate": 4.697373429767797e-06, + "loss": 1.7241, + "step": 466 + }, + { + "epoch": 0.47971237801746275, + "grad_norm": 2.613023281097412, + "learning_rate": 4.69547011800533e-06, + "loss": 1.7088, + "step": 467 + }, + { + "epoch": 0.48073959938366717, + "grad_norm": 2.5329010486602783, + "learning_rate": 4.693566806242863e-06, + "loss": 1.7163, + "step": 468 + }, + { + "epoch": 0.4817668207498716, + "grad_norm": 2.7887041568756104, + "learning_rate": 4.691663494480397e-06, + "loss": 1.6969, + "step": 469 + }, + { + "epoch": 0.482794042116076, + "grad_norm": 2.72991943359375, + "learning_rate": 4.689760182717929e-06, + "loss": 1.7332, + "step": 470 + }, + { + "epoch": 0.482794042116076, + "eval_loss": 1.965472936630249, + "eval_runtime": 46.1059, + "eval_samples_per_second": 21.689, + "eval_steps_per_second": 0.911, + "step": 470 + }, + { + "epoch": 0.48382126348228044, + "grad_norm": 2.613785982131958, + "learning_rate": 4.6878568709554626e-06, + "loss": 1.6272, + "step": 471 + }, + { + "epoch": 0.48484848484848486, + "grad_norm": 2.605299472808838, + "learning_rate": 4.685953559192996e-06, + "loss": 1.6966, + "step": 472 + }, + { + "epoch": 0.4858757062146893, + "grad_norm": 2.672013998031616, + "learning_rate": 4.684050247430529e-06, + "loss": 1.7132, + "step": 473 + }, + { + "epoch": 0.4869029275808937, + "grad_norm": 2.576486825942993, + "learning_rate": 4.682146935668063e-06, + "loss": 1.6449, + "step": 474 + }, + { + "epoch": 0.4879301489470981, + "grad_norm": 2.626840114593506, + "learning_rate": 4.680243623905596e-06, + "loss": 1.6993, + "step": 475 + }, + { + "epoch": 0.4889573703133025, + "grad_norm": 2.974418878555298, + "learning_rate": 4.678340312143129e-06, + "loss": 1.7116, + "step": 476 + }, + { + "epoch": 0.4899845916795069, + "grad_norm": 2.541242837905884, + "learning_rate": 4.676437000380663e-06, + "loss": 1.6693, + "step": 477 + }, + { + "epoch": 0.49101181304571134, + "grad_norm": 2.5494985580444336, + "learning_rate": 4.674533688618196e-06, + "loss": 1.7819, + "step": 478 + }, + { + "epoch": 0.49203903441191577, + "grad_norm": 2.606137990951538, + "learning_rate": 4.6726303768557295e-06, + "loss": 1.7278, + "step": 479 + }, + { + "epoch": 0.4930662557781202, + "grad_norm": 2.5032176971435547, + "learning_rate": 4.670727065093263e-06, + "loss": 1.6318, + "step": 480 + }, + { + "epoch": 0.4930662557781202, + "eval_loss": 1.9851018190383911, + "eval_runtime": 46.0386, + "eval_samples_per_second": 21.721, + "eval_steps_per_second": 0.912, + "step": 480 + }, + { + "epoch": 0.4940934771443246, + "grad_norm": 2.703259229660034, + "learning_rate": 4.668823753330796e-06, + "loss": 1.6646, + "step": 481 + }, + { + "epoch": 0.49512069851052903, + "grad_norm": 2.681807041168213, + "learning_rate": 4.6669204415683295e-06, + "loss": 1.6759, + "step": 482 + }, + { + "epoch": 0.49614791987673346, + "grad_norm": 2.7280473709106445, + "learning_rate": 4.665017129805863e-06, + "loss": 1.7388, + "step": 483 + }, + { + "epoch": 0.4971751412429379, + "grad_norm": 2.420909881591797, + "learning_rate": 4.663113818043396e-06, + "loss": 1.729, + "step": 484 + }, + { + "epoch": 0.49820236260914225, + "grad_norm": 2.5764737129211426, + "learning_rate": 4.66121050628093e-06, + "loss": 1.6903, + "step": 485 + }, + { + "epoch": 0.49922958397534667, + "grad_norm": 2.389531135559082, + "learning_rate": 4.659307194518463e-06, + "loss": 1.7191, + "step": 486 + }, + { + "epoch": 0.5002568053415511, + "grad_norm": 2.5207972526550293, + "learning_rate": 4.6574038827559955e-06, + "loss": 1.6892, + "step": 487 + }, + { + "epoch": 0.5012840267077555, + "grad_norm": 2.5662360191345215, + "learning_rate": 4.655500570993529e-06, + "loss": 1.6978, + "step": 488 + }, + { + "epoch": 0.50231124807396, + "grad_norm": 2.50262713432312, + "learning_rate": 4.653597259231062e-06, + "loss": 1.7379, + "step": 489 + }, + { + "epoch": 0.5033384694401644, + "grad_norm": 2.56643009185791, + "learning_rate": 4.651693947468596e-06, + "loss": 1.6233, + "step": 490 + }, + { + "epoch": 0.5033384694401644, + "eval_loss": 1.9815036058425903, + "eval_runtime": 48.0257, + "eval_samples_per_second": 20.822, + "eval_steps_per_second": 0.875, + "step": 490 + }, + { + "epoch": 0.5043656908063687, + "grad_norm": 2.504117965698242, + "learning_rate": 4.649790635706129e-06, + "loss": 1.7664, + "step": 491 + }, + { + "epoch": 0.5053929121725732, + "grad_norm": 2.906843423843384, + "learning_rate": 4.647887323943662e-06, + "loss": 1.6731, + "step": 492 + }, + { + "epoch": 0.5064201335387776, + "grad_norm": 2.586747169494629, + "learning_rate": 4.645984012181196e-06, + "loss": 1.68, + "step": 493 + }, + { + "epoch": 0.507447354904982, + "grad_norm": 2.3300511837005615, + "learning_rate": 4.644080700418729e-06, + "loss": 1.5998, + "step": 494 + }, + { + "epoch": 0.5084745762711864, + "grad_norm": 2.8711049556732178, + "learning_rate": 4.642177388656262e-06, + "loss": 1.7198, + "step": 495 + }, + { + "epoch": 0.5095017976373909, + "grad_norm": 2.6463162899017334, + "learning_rate": 4.640274076893796e-06, + "loss": 1.6814, + "step": 496 + }, + { + "epoch": 0.5105290190035953, + "grad_norm": 2.650481700897217, + "learning_rate": 4.638370765131328e-06, + "loss": 1.7353, + "step": 497 + }, + { + "epoch": 0.5115562403697997, + "grad_norm": 2.4464986324310303, + "learning_rate": 4.636467453368862e-06, + "loss": 1.6579, + "step": 498 + }, + { + "epoch": 0.5125834617360041, + "grad_norm": 2.438021659851074, + "learning_rate": 4.634564141606396e-06, + "loss": 1.6485, + "step": 499 + }, + { + "epoch": 0.5136106831022085, + "grad_norm": 2.5354347229003906, + "learning_rate": 4.632660829843929e-06, + "loss": 1.7123, + "step": 500 + }, + { + "epoch": 0.5136106831022085, + "eval_loss": 1.9738385677337646, + "eval_runtime": 52.5306, + "eval_samples_per_second": 19.037, + "eval_steps_per_second": 0.8, + "step": 500 + }, + { + "epoch": 0.514637904468413, + "grad_norm": 2.3702070713043213, + "learning_rate": 4.630757518081463e-06, + "loss": 1.6423, + "step": 501 + }, + { + "epoch": 0.5156651258346173, + "grad_norm": 2.5336198806762695, + "learning_rate": 4.628854206318996e-06, + "loss": 1.771, + "step": 502 + }, + { + "epoch": 0.5166923472008218, + "grad_norm": 2.572751998901367, + "learning_rate": 4.626950894556529e-06, + "loss": 1.677, + "step": 503 + }, + { + "epoch": 0.5177195685670262, + "grad_norm": 2.3865134716033936, + "learning_rate": 4.625047582794062e-06, + "loss": 1.7111, + "step": 504 + }, + { + "epoch": 0.5187467899332306, + "grad_norm": 2.5152785778045654, + "learning_rate": 4.623144271031595e-06, + "loss": 1.6343, + "step": 505 + }, + { + "epoch": 0.519774011299435, + "grad_norm": 2.5121307373046875, + "learning_rate": 4.6212409592691286e-06, + "loss": 1.6357, + "step": 506 + }, + { + "epoch": 0.5208012326656395, + "grad_norm": 2.506908655166626, + "learning_rate": 4.619337647506662e-06, + "loss": 1.7295, + "step": 507 + }, + { + "epoch": 0.5218284540318439, + "grad_norm": 2.636082172393799, + "learning_rate": 4.617434335744195e-06, + "loss": 1.6916, + "step": 508 + }, + { + "epoch": 0.5228556753980482, + "grad_norm": 2.5987415313720703, + "learning_rate": 4.615531023981729e-06, + "loss": 1.659, + "step": 509 + }, + { + "epoch": 0.5238828967642527, + "grad_norm": 2.642770290374756, + "learning_rate": 4.613627712219262e-06, + "loss": 1.6789, + "step": 510 + }, + { + "epoch": 0.5238828967642527, + "eval_loss": 1.9659523963928223, + "eval_runtime": 42.9453, + "eval_samples_per_second": 23.285, + "eval_steps_per_second": 0.978, + "step": 510 + }, + { + "epoch": 0.5249101181304571, + "grad_norm": 2.438389301300049, + "learning_rate": 4.611724400456795e-06, + "loss": 1.68, + "step": 511 + }, + { + "epoch": 0.5259373394966615, + "grad_norm": 2.536546468734741, + "learning_rate": 4.609821088694329e-06, + "loss": 1.6847, + "step": 512 + }, + { + "epoch": 0.5269645608628659, + "grad_norm": 2.4056830406188965, + "learning_rate": 4.607917776931862e-06, + "loss": 1.6577, + "step": 513 + }, + { + "epoch": 0.5279917822290704, + "grad_norm": 2.81195330619812, + "learning_rate": 4.606014465169395e-06, + "loss": 1.7149, + "step": 514 + }, + { + "epoch": 0.5290190035952748, + "grad_norm": 2.684948205947876, + "learning_rate": 4.604111153406928e-06, + "loss": 1.717, + "step": 515 + }, + { + "epoch": 0.5300462249614792, + "grad_norm": 2.583092212677002, + "learning_rate": 4.602207841644461e-06, + "loss": 1.6238, + "step": 516 + }, + { + "epoch": 0.5310734463276836, + "grad_norm": 3.022528886795044, + "learning_rate": 4.600304529881995e-06, + "loss": 1.7262, + "step": 517 + }, + { + "epoch": 0.5321006676938881, + "grad_norm": 2.6941866874694824, + "learning_rate": 4.598401218119528e-06, + "loss": 1.7055, + "step": 518 + }, + { + "epoch": 0.5331278890600925, + "grad_norm": 2.5207839012145996, + "learning_rate": 4.5964979063570614e-06, + "loss": 1.7117, + "step": 519 + }, + { + "epoch": 0.5341551104262968, + "grad_norm": 2.8056678771972656, + "learning_rate": 4.594594594594596e-06, + "loss": 1.657, + "step": 520 + }, + { + "epoch": 0.5341551104262968, + "eval_loss": 1.9712116718292236, + "eval_runtime": 49.57, + "eval_samples_per_second": 20.173, + "eval_steps_per_second": 0.847, + "step": 520 + }, + { + "epoch": 0.5351823317925013, + "grad_norm": 2.6119930744171143, + "learning_rate": 4.592691282832128e-06, + "loss": 1.6897, + "step": 521 + }, + { + "epoch": 0.5362095531587057, + "grad_norm": 2.7165610790252686, + "learning_rate": 4.5907879710696615e-06, + "loss": 1.7038, + "step": 522 + }, + { + "epoch": 0.5372367745249101, + "grad_norm": 2.3613736629486084, + "learning_rate": 4.588884659307195e-06, + "loss": 1.6222, + "step": 523 + }, + { + "epoch": 0.5382639958911145, + "grad_norm": 2.647672176361084, + "learning_rate": 4.586981347544728e-06, + "loss": 1.702, + "step": 524 + }, + { + "epoch": 0.539291217257319, + "grad_norm": 2.704634428024292, + "learning_rate": 4.585078035782262e-06, + "loss": 1.6566, + "step": 525 + }, + { + "epoch": 0.5403184386235234, + "grad_norm": 2.6634597778320312, + "learning_rate": 4.583174724019795e-06, + "loss": 1.6903, + "step": 526 + }, + { + "epoch": 0.5413456599897278, + "grad_norm": 2.5745162963867188, + "learning_rate": 4.581271412257328e-06, + "loss": 1.6375, + "step": 527 + }, + { + "epoch": 0.5423728813559322, + "grad_norm": 2.5577778816223145, + "learning_rate": 4.579368100494862e-06, + "loss": 1.614, + "step": 528 + }, + { + "epoch": 0.5434001027221366, + "grad_norm": 2.7684454917907715, + "learning_rate": 4.577464788732395e-06, + "loss": 1.7168, + "step": 529 + }, + { + "epoch": 0.544427324088341, + "grad_norm": 2.582108736038208, + "learning_rate": 4.575561476969928e-06, + "loss": 1.6897, + "step": 530 + }, + { + "epoch": 0.544427324088341, + "eval_loss": 1.9821091890335083, + "eval_runtime": 50.9409, + "eval_samples_per_second": 19.631, + "eval_steps_per_second": 0.824, + "step": 530 + }, + { + "epoch": 0.5454545454545454, + "grad_norm": 2.516486406326294, + "learning_rate": 4.573658165207461e-06, + "loss": 1.7273, + "step": 531 + }, + { + "epoch": 0.5464817668207499, + "grad_norm": 2.786806344985962, + "learning_rate": 4.571754853444994e-06, + "loss": 1.6898, + "step": 532 + }, + { + "epoch": 0.5475089881869543, + "grad_norm": 2.3137319087982178, + "learning_rate": 4.569851541682528e-06, + "loss": 1.676, + "step": 533 + }, + { + "epoch": 0.5485362095531587, + "grad_norm": 2.7268903255462646, + "learning_rate": 4.567948229920061e-06, + "loss": 1.6483, + "step": 534 + }, + { + "epoch": 0.5495634309193631, + "grad_norm": 2.4385902881622314, + "learning_rate": 4.566044918157594e-06, + "loss": 1.6172, + "step": 535 + }, + { + "epoch": 0.5505906522855676, + "grad_norm": 2.420576333999634, + "learning_rate": 4.564141606395128e-06, + "loss": 1.6569, + "step": 536 + }, + { + "epoch": 0.551617873651772, + "grad_norm": 2.592290163040161, + "learning_rate": 4.562238294632661e-06, + "loss": 1.7085, + "step": 537 + }, + { + "epoch": 0.5526450950179764, + "grad_norm": 2.485285520553589, + "learning_rate": 4.5603349828701945e-06, + "loss": 1.6699, + "step": 538 + }, + { + "epoch": 0.5536723163841808, + "grad_norm": 2.844710111618042, + "learning_rate": 4.558431671107728e-06, + "loss": 1.6873, + "step": 539 + }, + { + "epoch": 0.5546995377503852, + "grad_norm": 2.5279181003570557, + "learning_rate": 4.556528359345261e-06, + "loss": 1.7285, + "step": 540 + }, + { + "epoch": 0.5546995377503852, + "eval_loss": 1.9825981855392456, + "eval_runtime": 52.461, + "eval_samples_per_second": 19.062, + "eval_steps_per_second": 0.801, + "step": 540 + }, + { + "epoch": 0.5557267591165896, + "grad_norm": 2.673997163772583, + "learning_rate": 4.5546250475827946e-06, + "loss": 1.6658, + "step": 541 + }, + { + "epoch": 0.556753980482794, + "grad_norm": 2.6023194789886475, + "learning_rate": 4.552721735820328e-06, + "loss": 1.6262, + "step": 542 + }, + { + "epoch": 0.5577812018489985, + "grad_norm": 2.6327662467956543, + "learning_rate": 4.550818424057861e-06, + "loss": 1.6574, + "step": 543 + }, + { + "epoch": 0.5588084232152029, + "grad_norm": 2.5532140731811523, + "learning_rate": 4.548915112295395e-06, + "loss": 1.6607, + "step": 544 + }, + { + "epoch": 0.5598356445814073, + "grad_norm": 2.523966073989868, + "learning_rate": 4.547011800532928e-06, + "loss": 1.6752, + "step": 545 + }, + { + "epoch": 0.5608628659476117, + "grad_norm": 2.6338107585906982, + "learning_rate": 4.545108488770461e-06, + "loss": 1.6258, + "step": 546 + }, + { + "epoch": 0.5618900873138162, + "grad_norm": 2.500096559524536, + "learning_rate": 4.543205177007995e-06, + "loss": 1.7189, + "step": 547 + }, + { + "epoch": 0.5629173086800205, + "grad_norm": 2.644413471221924, + "learning_rate": 4.541301865245527e-06, + "loss": 1.6594, + "step": 548 + }, + { + "epoch": 0.5639445300462249, + "grad_norm": 2.451965093612671, + "learning_rate": 4.539398553483061e-06, + "loss": 1.6984, + "step": 549 + }, + { + "epoch": 0.5649717514124294, + "grad_norm": 2.7554423809051514, + "learning_rate": 4.537495241720594e-06, + "loss": 1.6626, + "step": 550 + }, + { + "epoch": 0.5649717514124294, + "eval_loss": 1.9897041320800781, + "eval_runtime": 45.9349, + "eval_samples_per_second": 21.77, + "eval_steps_per_second": 0.914, + "step": 550 + }, + { + "epoch": 0.5659989727786338, + "grad_norm": 2.5962674617767334, + "learning_rate": 4.535591929958127e-06, + "loss": 1.7015, + "step": 551 + }, + { + "epoch": 0.5670261941448382, + "grad_norm": 2.7917397022247314, + "learning_rate": 4.533688618195661e-06, + "loss": 1.6811, + "step": 552 + }, + { + "epoch": 0.5680534155110426, + "grad_norm": 2.4205498695373535, + "learning_rate": 4.531785306433194e-06, + "loss": 1.6659, + "step": 553 + }, + { + "epoch": 0.5690806368772471, + "grad_norm": 2.74898624420166, + "learning_rate": 4.529881994670727e-06, + "loss": 1.6521, + "step": 554 + }, + { + "epoch": 0.5701078582434514, + "grad_norm": 2.859898328781128, + "learning_rate": 4.527978682908261e-06, + "loss": 1.6445, + "step": 555 + }, + { + "epoch": 0.5711350796096559, + "grad_norm": 2.5526154041290283, + "learning_rate": 4.526075371145794e-06, + "loss": 1.6679, + "step": 556 + }, + { + "epoch": 0.5721623009758603, + "grad_norm": 2.96175217628479, + "learning_rate": 4.5241720593833275e-06, + "loss": 1.6784, + "step": 557 + }, + { + "epoch": 0.5731895223420647, + "grad_norm": 2.5037007331848145, + "learning_rate": 4.52226874762086e-06, + "loss": 1.6689, + "step": 558 + }, + { + "epoch": 0.5742167437082691, + "grad_norm": 2.6513259410858154, + "learning_rate": 4.520365435858393e-06, + "loss": 1.7038, + "step": 559 + }, + { + "epoch": 0.5752439650744735, + "grad_norm": 2.762770175933838, + "learning_rate": 4.518462124095928e-06, + "loss": 1.728, + "step": 560 + }, + { + "epoch": 0.5752439650744735, + "eval_loss": 1.972934603691101, + "eval_runtime": 44.2572, + "eval_samples_per_second": 22.595, + "eval_steps_per_second": 0.949, + "step": 560 + }, + { + "epoch": 0.576271186440678, + "grad_norm": 2.637406349182129, + "learning_rate": 4.516558812333461e-06, + "loss": 1.682, + "step": 561 + }, + { + "epoch": 0.5772984078068824, + "grad_norm": 2.545583724975586, + "learning_rate": 4.514655500570994e-06, + "loss": 1.6669, + "step": 562 + }, + { + "epoch": 0.5783256291730868, + "grad_norm": 2.5655033588409424, + "learning_rate": 4.512752188808528e-06, + "loss": 1.7279, + "step": 563 + }, + { + "epoch": 0.5793528505392912, + "grad_norm": 2.6856470108032227, + "learning_rate": 4.510848877046061e-06, + "loss": 1.7331, + "step": 564 + }, + { + "epoch": 0.5803800719054957, + "grad_norm": 2.605252981185913, + "learning_rate": 4.5089455652835936e-06, + "loss": 1.6904, + "step": 565 + }, + { + "epoch": 0.5814072932717, + "grad_norm": 2.610267400741577, + "learning_rate": 4.507042253521127e-06, + "loss": 1.6607, + "step": 566 + }, + { + "epoch": 0.5824345146379045, + "grad_norm": 2.6955432891845703, + "learning_rate": 4.50513894175866e-06, + "loss": 1.6166, + "step": 567 + }, + { + "epoch": 0.5834617360041089, + "grad_norm": 2.5661842823028564, + "learning_rate": 4.503235629996194e-06, + "loss": 1.6955, + "step": 568 + }, + { + "epoch": 0.5844889573703133, + "grad_norm": 2.6393115520477295, + "learning_rate": 4.501332318233727e-06, + "loss": 1.6933, + "step": 569 + }, + { + "epoch": 0.5855161787365177, + "grad_norm": 2.612645387649536, + "learning_rate": 4.49942900647126e-06, + "loss": 1.6585, + "step": 570 + }, + { + "epoch": 0.5855161787365177, + "eval_loss": 1.9599064588546753, + "eval_runtime": 46.0676, + "eval_samples_per_second": 21.707, + "eval_steps_per_second": 0.912, + "step": 570 + }, + { + "epoch": 0.5865434001027221, + "grad_norm": 2.4182891845703125, + "learning_rate": 4.497525694708794e-06, + "loss": 1.707, + "step": 571 + }, + { + "epoch": 0.5875706214689266, + "grad_norm": 2.7214412689208984, + "learning_rate": 4.495622382946327e-06, + "loss": 1.6966, + "step": 572 + }, + { + "epoch": 0.588597842835131, + "grad_norm": 2.5138018131256104, + "learning_rate": 4.4937190711838605e-06, + "loss": 1.6373, + "step": 573 + }, + { + "epoch": 0.5896250642013354, + "grad_norm": 2.5814292430877686, + "learning_rate": 4.491815759421394e-06, + "loss": 1.7313, + "step": 574 + }, + { + "epoch": 0.5906522855675398, + "grad_norm": 2.6236367225646973, + "learning_rate": 4.489912447658926e-06, + "loss": 1.6647, + "step": 575 + }, + { + "epoch": 0.5916795069337443, + "grad_norm": 2.468557357788086, + "learning_rate": 4.48800913589646e-06, + "loss": 1.6412, + "step": 576 + }, + { + "epoch": 0.5927067282999486, + "grad_norm": 2.454141855239868, + "learning_rate": 4.486105824133993e-06, + "loss": 1.6762, + "step": 577 + }, + { + "epoch": 0.593733949666153, + "grad_norm": 3.0009915828704834, + "learning_rate": 4.4842025123715264e-06, + "loss": 1.6847, + "step": 578 + }, + { + "epoch": 0.5947611710323575, + "grad_norm": 2.574394702911377, + "learning_rate": 4.48229920060906e-06, + "loss": 1.6421, + "step": 579 + }, + { + "epoch": 0.5957883923985619, + "grad_norm": 2.6978182792663574, + "learning_rate": 4.480395888846594e-06, + "loss": 1.6198, + "step": 580 + }, + { + "epoch": 0.5957883923985619, + "eval_loss": 1.9570144414901733, + "eval_runtime": 44.7661, + "eval_samples_per_second": 22.338, + "eval_steps_per_second": 0.938, + "step": 580 + }, + { + "epoch": 0.5968156137647663, + "grad_norm": 2.652461051940918, + "learning_rate": 4.478492577084127e-06, + "loss": 1.7334, + "step": 581 + }, + { + "epoch": 0.5978428351309707, + "grad_norm": 2.6292548179626465, + "learning_rate": 4.47658926532166e-06, + "loss": 1.6242, + "step": 582 + }, + { + "epoch": 0.5988700564971752, + "grad_norm": 2.474919080734253, + "learning_rate": 4.474685953559193e-06, + "loss": 1.6732, + "step": 583 + }, + { + "epoch": 0.5998972778633795, + "grad_norm": 2.7282516956329346, + "learning_rate": 4.472782641796727e-06, + "loss": 1.802, + "step": 584 + }, + { + "epoch": 0.600924499229584, + "grad_norm": 2.770307779312134, + "learning_rate": 4.47087933003426e-06, + "loss": 1.6156, + "step": 585 + }, + { + "epoch": 0.6019517205957884, + "grad_norm": 2.570681095123291, + "learning_rate": 4.468976018271793e-06, + "loss": 1.6309, + "step": 586 + }, + { + "epoch": 0.6029789419619929, + "grad_norm": 2.4075732231140137, + "learning_rate": 4.467072706509327e-06, + "loss": 1.6436, + "step": 587 + }, + { + "epoch": 0.6040061633281972, + "grad_norm": 2.4570937156677246, + "learning_rate": 4.46516939474686e-06, + "loss": 1.6737, + "step": 588 + }, + { + "epoch": 0.6050333846944016, + "grad_norm": 2.407274007797241, + "learning_rate": 4.463266082984393e-06, + "loss": 1.6179, + "step": 589 + }, + { + "epoch": 0.6060606060606061, + "grad_norm": 2.745486259460449, + "learning_rate": 4.461362771221927e-06, + "loss": 1.7458, + "step": 590 + }, + { + "epoch": 0.6060606060606061, + "eval_loss": 1.9700438976287842, + "eval_runtime": 45.3018, + "eval_samples_per_second": 22.074, + "eval_steps_per_second": 0.927, + "step": 590 + }, + { + "epoch": 0.6070878274268104, + "grad_norm": 2.4258809089660645, + "learning_rate": 4.45945945945946e-06, + "loss": 1.6372, + "step": 591 + }, + { + "epoch": 0.6081150487930149, + "grad_norm": 2.683662176132202, + "learning_rate": 4.457556147696993e-06, + "loss": 1.6649, + "step": 592 + }, + { + "epoch": 0.6091422701592193, + "grad_norm": 2.879727602005005, + "learning_rate": 4.455652835934526e-06, + "loss": 1.7376, + "step": 593 + }, + { + "epoch": 0.6101694915254238, + "grad_norm": 2.5620689392089844, + "learning_rate": 4.453749524172059e-06, + "loss": 1.7203, + "step": 594 + }, + { + "epoch": 0.6111967128916281, + "grad_norm": 2.662031888961792, + "learning_rate": 4.451846212409593e-06, + "loss": 1.6835, + "step": 595 + }, + { + "epoch": 0.6122239342578326, + "grad_norm": 2.6269922256469727, + "learning_rate": 4.449942900647126e-06, + "loss": 1.6726, + "step": 596 + }, + { + "epoch": 0.613251155624037, + "grad_norm": 2.515852212905884, + "learning_rate": 4.4480395888846595e-06, + "loss": 1.6045, + "step": 597 + }, + { + "epoch": 0.6142783769902413, + "grad_norm": 3.006873607635498, + "learning_rate": 4.446136277122193e-06, + "loss": 1.6827, + "step": 598 + }, + { + "epoch": 0.6153055983564458, + "grad_norm": 2.2904858589172363, + "learning_rate": 4.444232965359726e-06, + "loss": 1.6123, + "step": 599 + }, + { + "epoch": 0.6163328197226502, + "grad_norm": 2.7345471382141113, + "learning_rate": 4.4423296535972596e-06, + "loss": 1.6697, + "step": 600 + }, + { + "epoch": 0.6163328197226502, + "eval_loss": 1.9713997840881348, + "eval_runtime": 45.6715, + "eval_samples_per_second": 21.895, + "eval_steps_per_second": 0.92, + "step": 600 + }, + { + "epoch": 0.6173600410888547, + "grad_norm": 2.955853223800659, + "learning_rate": 4.440426341834793e-06, + "loss": 1.6195, + "step": 601 + }, + { + "epoch": 0.618387262455059, + "grad_norm": 2.4079766273498535, + "learning_rate": 4.438523030072326e-06, + "loss": 1.6589, + "step": 602 + }, + { + "epoch": 0.6194144838212635, + "grad_norm": 2.8158490657806396, + "learning_rate": 4.43661971830986e-06, + "loss": 1.6895, + "step": 603 + }, + { + "epoch": 0.6204417051874679, + "grad_norm": 2.491471529006958, + "learning_rate": 4.434716406547393e-06, + "loss": 1.6351, + "step": 604 + }, + { + "epoch": 0.6214689265536724, + "grad_norm": 2.4093260765075684, + "learning_rate": 4.432813094784926e-06, + "loss": 1.6402, + "step": 605 + }, + { + "epoch": 0.6224961479198767, + "grad_norm": 2.751018762588501, + "learning_rate": 4.43090978302246e-06, + "loss": 1.6962, + "step": 606 + }, + { + "epoch": 0.6235233692860811, + "grad_norm": 2.4410409927368164, + "learning_rate": 4.429006471259993e-06, + "loss": 1.6549, + "step": 607 + }, + { + "epoch": 0.6245505906522856, + "grad_norm": 2.360455274581909, + "learning_rate": 4.4271031594975265e-06, + "loss": 1.6449, + "step": 608 + }, + { + "epoch": 0.6255778120184899, + "grad_norm": 2.5058023929595947, + "learning_rate": 4.425199847735059e-06, + "loss": 1.617, + "step": 609 + }, + { + "epoch": 0.6266050333846944, + "grad_norm": 2.3429927825927734, + "learning_rate": 4.423296535972592e-06, + "loss": 1.6262, + "step": 610 + }, + { + "epoch": 0.6266050333846944, + "eval_loss": 1.9597365856170654, + "eval_runtime": 42.2034, + "eval_samples_per_second": 23.695, + "eval_steps_per_second": 0.995, + "step": 610 + }, + { + "epoch": 0.6276322547508988, + "grad_norm": 2.401761054992676, + "learning_rate": 4.421393224210126e-06, + "loss": 1.6497, + "step": 611 + }, + { + "epoch": 0.6286594761171033, + "grad_norm": 2.433439254760742, + "learning_rate": 4.419489912447659e-06, + "loss": 1.6117, + "step": 612 + }, + { + "epoch": 0.6296866974833076, + "grad_norm": 2.5147452354431152, + "learning_rate": 4.4175866006851924e-06, + "loss": 1.7041, + "step": 613 + }, + { + "epoch": 0.6307139188495121, + "grad_norm": 2.4174447059631348, + "learning_rate": 4.415683288922726e-06, + "loss": 1.6652, + "step": 614 + }, + { + "epoch": 0.6317411402157165, + "grad_norm": 2.424800157546997, + "learning_rate": 4.413779977160259e-06, + "loss": 1.6407, + "step": 615 + }, + { + "epoch": 0.632768361581921, + "grad_norm": 2.6760315895080566, + "learning_rate": 4.4118766653977925e-06, + "loss": 1.6566, + "step": 616 + }, + { + "epoch": 0.6337955829481253, + "grad_norm": 2.694810628890991, + "learning_rate": 4.409973353635326e-06, + "loss": 1.7125, + "step": 617 + }, + { + "epoch": 0.6348228043143297, + "grad_norm": 2.475126266479492, + "learning_rate": 4.408070041872859e-06, + "loss": 1.6663, + "step": 618 + }, + { + "epoch": 0.6358500256805342, + "grad_norm": 2.734738349914551, + "learning_rate": 4.406166730110393e-06, + "loss": 1.6115, + "step": 619 + }, + { + "epoch": 0.6368772470467385, + "grad_norm": 2.4914753437042236, + "learning_rate": 4.404263418347925e-06, + "loss": 1.6096, + "step": 620 + }, + { + "epoch": 0.6368772470467385, + "eval_loss": 1.952081322669983, + "eval_runtime": 45.4451, + "eval_samples_per_second": 22.005, + "eval_steps_per_second": 0.924, + "step": 620 + }, + { + "epoch": 0.637904468412943, + "grad_norm": 2.3727800846099854, + "learning_rate": 4.402360106585459e-06, + "loss": 1.6275, + "step": 621 + }, + { + "epoch": 0.6389316897791474, + "grad_norm": 2.576402187347412, + "learning_rate": 4.400456794822993e-06, + "loss": 1.6252, + "step": 622 + }, + { + "epoch": 0.6399589111453519, + "grad_norm": 2.560185432434082, + "learning_rate": 4.398553483060526e-06, + "loss": 1.6522, + "step": 623 + }, + { + "epoch": 0.6409861325115562, + "grad_norm": 2.462785005569458, + "learning_rate": 4.396650171298059e-06, + "loss": 1.6124, + "step": 624 + }, + { + "epoch": 0.6420133538777607, + "grad_norm": 2.3942930698394775, + "learning_rate": 4.394746859535593e-06, + "loss": 1.6231, + "step": 625 + }, + { + "epoch": 0.6430405752439651, + "grad_norm": 2.4870522022247314, + "learning_rate": 4.392843547773125e-06, + "loss": 1.6086, + "step": 626 + }, + { + "epoch": 0.6440677966101694, + "grad_norm": 2.5702786445617676, + "learning_rate": 4.390940236010659e-06, + "loss": 1.6228, + "step": 627 + }, + { + "epoch": 0.6450950179763739, + "grad_norm": 2.743983268737793, + "learning_rate": 4.389036924248192e-06, + "loss": 1.6394, + "step": 628 + }, + { + "epoch": 0.6461222393425783, + "grad_norm": 2.59942364692688, + "learning_rate": 4.387133612485725e-06, + "loss": 1.6721, + "step": 629 + }, + { + "epoch": 0.6471494607087828, + "grad_norm": 2.4473204612731934, + "learning_rate": 4.385230300723259e-06, + "loss": 1.6407, + "step": 630 + }, + { + "epoch": 0.6471494607087828, + "eval_loss": 1.9426974058151245, + "eval_runtime": 45.2318, + "eval_samples_per_second": 22.108, + "eval_steps_per_second": 0.929, + "step": 630 + }, + { + "epoch": 0.6481766820749871, + "grad_norm": 2.6036875247955322, + "learning_rate": 4.383326988960792e-06, + "loss": 1.6307, + "step": 631 + }, + { + "epoch": 0.6492039034411916, + "grad_norm": 2.691011428833008, + "learning_rate": 4.3814236771983255e-06, + "loss": 1.6059, + "step": 632 + }, + { + "epoch": 0.650231124807396, + "grad_norm": 2.585758924484253, + "learning_rate": 4.379520365435859e-06, + "loss": 1.6781, + "step": 633 + }, + { + "epoch": 0.6512583461736005, + "grad_norm": 2.9522950649261475, + "learning_rate": 4.377617053673392e-06, + "loss": 1.6559, + "step": 634 + }, + { + "epoch": 0.6522855675398048, + "grad_norm": 2.5368709564208984, + "learning_rate": 4.3757137419109256e-06, + "loss": 1.6445, + "step": 635 + }, + { + "epoch": 0.6533127889060092, + "grad_norm": 2.5002150535583496, + "learning_rate": 4.373810430148459e-06, + "loss": 1.7184, + "step": 636 + }, + { + "epoch": 0.6543400102722137, + "grad_norm": 2.7623229026794434, + "learning_rate": 4.3719071183859914e-06, + "loss": 1.6195, + "step": 637 + }, + { + "epoch": 0.655367231638418, + "grad_norm": 2.6188151836395264, + "learning_rate": 4.370003806623525e-06, + "loss": 1.6686, + "step": 638 + }, + { + "epoch": 0.6563944530046225, + "grad_norm": 2.685776948928833, + "learning_rate": 4.368100494861058e-06, + "loss": 1.6134, + "step": 639 + }, + { + "epoch": 0.6574216743708269, + "grad_norm": 2.5703907012939453, + "learning_rate": 4.3661971830985915e-06, + "loss": 1.665, + "step": 640 + }, + { + "epoch": 0.6574216743708269, + "eval_loss": 1.9504806995391846, + "eval_runtime": 47.1947, + "eval_samples_per_second": 21.189, + "eval_steps_per_second": 0.89, + "step": 640 + }, + { + "epoch": 0.6584488957370314, + "grad_norm": 2.511617660522461, + "learning_rate": 4.364293871336126e-06, + "loss": 1.6201, + "step": 641 + }, + { + "epoch": 0.6594761171032357, + "grad_norm": 2.549302339553833, + "learning_rate": 4.362390559573659e-06, + "loss": 1.647, + "step": 642 + }, + { + "epoch": 0.6605033384694402, + "grad_norm": 2.348747730255127, + "learning_rate": 4.3604872478111925e-06, + "loss": 1.5993, + "step": 643 + }, + { + "epoch": 0.6615305598356446, + "grad_norm": 2.7393345832824707, + "learning_rate": 4.358583936048725e-06, + "loss": 1.543, + "step": 644 + }, + { + "epoch": 0.662557781201849, + "grad_norm": 2.4242193698883057, + "learning_rate": 4.356680624286258e-06, + "loss": 1.6726, + "step": 645 + }, + { + "epoch": 0.6635850025680534, + "grad_norm": 2.6446714401245117, + "learning_rate": 4.354777312523792e-06, + "loss": 1.6312, + "step": 646 + }, + { + "epoch": 0.6646122239342578, + "grad_norm": 2.7407124042510986, + "learning_rate": 4.352874000761325e-06, + "loss": 1.6868, + "step": 647 + }, + { + "epoch": 0.6656394453004623, + "grad_norm": 2.3706371784210205, + "learning_rate": 4.3509706889988584e-06, + "loss": 1.6301, + "step": 648 + }, + { + "epoch": 0.6666666666666666, + "grad_norm": 2.5197198390960693, + "learning_rate": 4.349067377236392e-06, + "loss": 1.702, + "step": 649 + }, + { + "epoch": 0.6676938880328711, + "grad_norm": 2.4792377948760986, + "learning_rate": 4.347164065473925e-06, + "loss": 1.6891, + "step": 650 + }, + { + "epoch": 0.6676938880328711, + "eval_loss": 1.9589039087295532, + "eval_runtime": 47.1512, + "eval_samples_per_second": 21.208, + "eval_steps_per_second": 0.891, + "step": 650 + }, + { + "epoch": 0.6687211093990755, + "grad_norm": 2.4003701210021973, + "learning_rate": 4.3452607537114585e-06, + "loss": 1.7015, + "step": 651 + }, + { + "epoch": 0.66974833076528, + "grad_norm": 2.5363786220550537, + "learning_rate": 4.343357441948992e-06, + "loss": 1.6133, + "step": 652 + }, + { + "epoch": 0.6707755521314843, + "grad_norm": 2.5183844566345215, + "learning_rate": 4.341454130186525e-06, + "loss": 1.6574, + "step": 653 + }, + { + "epoch": 0.6718027734976888, + "grad_norm": 2.4668304920196533, + "learning_rate": 4.339550818424058e-06, + "loss": 1.6253, + "step": 654 + }, + { + "epoch": 0.6728299948638932, + "grad_norm": 2.4829814434051514, + "learning_rate": 4.337647506661591e-06, + "loss": 1.6099, + "step": 655 + }, + { + "epoch": 0.6738572162300975, + "grad_norm": 2.3683176040649414, + "learning_rate": 4.3357441948991245e-06, + "loss": 1.6385, + "step": 656 + }, + { + "epoch": 0.674884437596302, + "grad_norm": 2.4906842708587646, + "learning_rate": 4.333840883136658e-06, + "loss": 1.6755, + "step": 657 + }, + { + "epoch": 0.6759116589625064, + "grad_norm": 2.558776617050171, + "learning_rate": 4.331937571374191e-06, + "loss": 1.6152, + "step": 658 + }, + { + "epoch": 0.6769388803287109, + "grad_norm": 2.4081523418426514, + "learning_rate": 4.3300342596117246e-06, + "loss": 1.6224, + "step": 659 + }, + { + "epoch": 0.6779661016949152, + "grad_norm": 2.488370180130005, + "learning_rate": 4.328130947849258e-06, + "loss": 1.6298, + "step": 660 + }, + { + "epoch": 0.6779661016949152, + "eval_loss": 1.9484621286392212, + "eval_runtime": 52.549, + "eval_samples_per_second": 19.03, + "eval_steps_per_second": 0.799, + "step": 660 + }, + { + "epoch": 0.6789933230611197, + "grad_norm": 2.495748281478882, + "learning_rate": 4.326227636086791e-06, + "loss": 1.5852, + "step": 661 + }, + { + "epoch": 0.6800205444273241, + "grad_norm": 2.525031566619873, + "learning_rate": 4.324324324324325e-06, + "loss": 1.6348, + "step": 662 + }, + { + "epoch": 0.6810477657935285, + "grad_norm": 2.427615165710449, + "learning_rate": 4.322421012561858e-06, + "loss": 1.6352, + "step": 663 + }, + { + "epoch": 0.6820749871597329, + "grad_norm": 2.5084927082061768, + "learning_rate": 4.320517700799391e-06, + "loss": 1.6334, + "step": 664 + }, + { + "epoch": 0.6831022085259374, + "grad_norm": 2.6018898487091064, + "learning_rate": 4.318614389036925e-06, + "loss": 1.7212, + "step": 665 + }, + { + "epoch": 0.6841294298921418, + "grad_norm": 2.44184947013855, + "learning_rate": 4.316711077274458e-06, + "loss": 1.6802, + "step": 666 + }, + { + "epoch": 0.6851566512583461, + "grad_norm": 2.542123317718506, + "learning_rate": 4.3148077655119915e-06, + "loss": 1.5912, + "step": 667 + }, + { + "epoch": 0.6861838726245506, + "grad_norm": 2.5208137035369873, + "learning_rate": 4.312904453749525e-06, + "loss": 1.6965, + "step": 668 + }, + { + "epoch": 0.687211093990755, + "grad_norm": 2.391904592514038, + "learning_rate": 4.311001141987058e-06, + "loss": 1.6388, + "step": 669 + }, + { + "epoch": 0.6882383153569595, + "grad_norm": 2.629565477371216, + "learning_rate": 4.3090978302245916e-06, + "loss": 1.6475, + "step": 670 + }, + { + "epoch": 0.6882383153569595, + "eval_loss": 1.9572818279266357, + "eval_runtime": 48.5354, + "eval_samples_per_second": 20.604, + "eval_steps_per_second": 0.865, + "step": 670 + }, + { + "epoch": 0.6892655367231638, + "grad_norm": 2.3132376670837402, + "learning_rate": 4.307194518462124e-06, + "loss": 1.6399, + "step": 671 + }, + { + "epoch": 0.6902927580893683, + "grad_norm": 2.5003411769866943, + "learning_rate": 4.3052912066996574e-06, + "loss": 1.7136, + "step": 672 + }, + { + "epoch": 0.6913199794555727, + "grad_norm": 2.545541286468506, + "learning_rate": 4.303387894937191e-06, + "loss": 1.6825, + "step": 673 + }, + { + "epoch": 0.6923472008217771, + "grad_norm": 2.342423439025879, + "learning_rate": 4.301484583174724e-06, + "loss": 1.6088, + "step": 674 + }, + { + "epoch": 0.6933744221879815, + "grad_norm": 2.5516135692596436, + "learning_rate": 4.2995812714122575e-06, + "loss": 1.5934, + "step": 675 + }, + { + "epoch": 0.6944016435541859, + "grad_norm": 2.438675880432129, + "learning_rate": 4.297677959649791e-06, + "loss": 1.6236, + "step": 676 + }, + { + "epoch": 0.6954288649203904, + "grad_norm": 2.767305612564087, + "learning_rate": 4.295774647887324e-06, + "loss": 1.6514, + "step": 677 + }, + { + "epoch": 0.6964560862865947, + "grad_norm": 2.591994047164917, + "learning_rate": 4.293871336124858e-06, + "loss": 1.6643, + "step": 678 + }, + { + "epoch": 0.6974833076527992, + "grad_norm": 2.4734244346618652, + "learning_rate": 4.291968024362391e-06, + "loss": 1.6362, + "step": 679 + }, + { + "epoch": 0.6985105290190036, + "grad_norm": 2.6181890964508057, + "learning_rate": 4.290064712599924e-06, + "loss": 1.6666, + "step": 680 + }, + { + "epoch": 0.6985105290190036, + "eval_loss": 1.9585388898849487, + "eval_runtime": 46.8094, + "eval_samples_per_second": 21.363, + "eval_steps_per_second": 0.897, + "step": 680 + }, + { + "epoch": 0.699537750385208, + "grad_norm": 2.465571403503418, + "learning_rate": 4.288161400837458e-06, + "loss": 1.6588, + "step": 681 + }, + { + "epoch": 0.7005649717514124, + "grad_norm": 2.664262294769287, + "learning_rate": 4.286258089074991e-06, + "loss": 1.5862, + "step": 682 + }, + { + "epoch": 0.7015921931176169, + "grad_norm": 2.677140474319458, + "learning_rate": 4.284354777312524e-06, + "loss": 1.6549, + "step": 683 + }, + { + "epoch": 0.7026194144838213, + "grad_norm": 2.559051275253296, + "learning_rate": 4.282451465550058e-06, + "loss": 1.609, + "step": 684 + }, + { + "epoch": 0.7036466358500256, + "grad_norm": 2.428424119949341, + "learning_rate": 4.280548153787591e-06, + "loss": 1.6243, + "step": 685 + }, + { + "epoch": 0.7046738572162301, + "grad_norm": 2.54136061668396, + "learning_rate": 4.2786448420251245e-06, + "loss": 1.6301, + "step": 686 + }, + { + "epoch": 0.7057010785824345, + "grad_norm": 2.377699136734009, + "learning_rate": 4.276741530262658e-06, + "loss": 1.5903, + "step": 687 + }, + { + "epoch": 0.706728299948639, + "grad_norm": 2.5548322200775146, + "learning_rate": 4.27483821850019e-06, + "loss": 1.6161, + "step": 688 + }, + { + "epoch": 0.7077555213148433, + "grad_norm": 2.5993409156799316, + "learning_rate": 4.272934906737724e-06, + "loss": 1.6882, + "step": 689 + }, + { + "epoch": 0.7087827426810478, + "grad_norm": 2.4249117374420166, + "learning_rate": 4.271031594975257e-06, + "loss": 1.6723, + "step": 690 + }, + { + "epoch": 0.7087827426810478, + "eval_loss": 1.9526821374893188, + "eval_runtime": 48.2168, + "eval_samples_per_second": 20.74, + "eval_steps_per_second": 0.871, + "step": 690 + }, + { + "epoch": 0.7098099640472522, + "grad_norm": 2.6965081691741943, + "learning_rate": 4.2691282832127905e-06, + "loss": 1.6636, + "step": 691 + }, + { + "epoch": 0.7108371854134566, + "grad_norm": 2.6205356121063232, + "learning_rate": 4.267224971450324e-06, + "loss": 1.6231, + "step": 692 + }, + { + "epoch": 0.711864406779661, + "grad_norm": 2.5924129486083984, + "learning_rate": 4.265321659687857e-06, + "loss": 1.5634, + "step": 693 + }, + { + "epoch": 0.7128916281458655, + "grad_norm": 2.9710943698883057, + "learning_rate": 4.2634183479253906e-06, + "loss": 1.6134, + "step": 694 + }, + { + "epoch": 0.7139188495120699, + "grad_norm": 2.6252782344818115, + "learning_rate": 4.261515036162924e-06, + "loss": 1.5997, + "step": 695 + }, + { + "epoch": 0.7149460708782742, + "grad_norm": 2.6178832054138184, + "learning_rate": 4.259611724400457e-06, + "loss": 1.6255, + "step": 696 + }, + { + "epoch": 0.7159732922444787, + "grad_norm": 2.82548189163208, + "learning_rate": 4.257708412637991e-06, + "loss": 1.6167, + "step": 697 + }, + { + "epoch": 0.7170005136106831, + "grad_norm": 2.6430351734161377, + "learning_rate": 4.255805100875523e-06, + "loss": 1.5585, + "step": 698 + }, + { + "epoch": 0.7180277349768875, + "grad_norm": 2.85046124458313, + "learning_rate": 4.2539017891130565e-06, + "loss": 1.5873, + "step": 699 + }, + { + "epoch": 0.7190549563430919, + "grad_norm": 2.5643832683563232, + "learning_rate": 4.25199847735059e-06, + "loss": 1.6705, + "step": 700 + }, + { + "epoch": 0.7190549563430919, + "eval_loss": 1.945617437362671, + "eval_runtime": 54.2778, + "eval_samples_per_second": 18.424, + "eval_steps_per_second": 0.774, + "step": 700 + }, + { + "epoch": 0.7200821777092964, + "grad_norm": 2.7856431007385254, + "learning_rate": 4.250095165588123e-06, + "loss": 1.5833, + "step": 701 + }, + { + "epoch": 0.7211093990755008, + "grad_norm": 2.9881107807159424, + "learning_rate": 4.2481918538256575e-06, + "loss": 1.6438, + "step": 702 + }, + { + "epoch": 0.7221366204417052, + "grad_norm": 2.602938652038574, + "learning_rate": 4.246288542063191e-06, + "loss": 1.6135, + "step": 703 + }, + { + "epoch": 0.7231638418079096, + "grad_norm": 2.997450590133667, + "learning_rate": 4.244385230300724e-06, + "loss": 1.6419, + "step": 704 + }, + { + "epoch": 0.724191063174114, + "grad_norm": 2.5592496395111084, + "learning_rate": 4.242481918538257e-06, + "loss": 1.5886, + "step": 705 + }, + { + "epoch": 0.7252182845403184, + "grad_norm": 2.4073948860168457, + "learning_rate": 4.24057860677579e-06, + "loss": 1.6219, + "step": 706 + }, + { + "epoch": 0.7262455059065228, + "grad_norm": 2.545997381210327, + "learning_rate": 4.2386752950133234e-06, + "loss": 1.6684, + "step": 707 + }, + { + "epoch": 0.7272727272727273, + "grad_norm": 2.468264579772949, + "learning_rate": 4.236771983250857e-06, + "loss": 1.6155, + "step": 708 + }, + { + "epoch": 0.7282999486389317, + "grad_norm": 2.6635043621063232, + "learning_rate": 4.23486867148839e-06, + "loss": 1.5956, + "step": 709 + }, + { + "epoch": 0.7293271700051361, + "grad_norm": 2.8751156330108643, + "learning_rate": 4.2329653597259235e-06, + "loss": 1.6211, + "step": 710 + }, + { + "epoch": 0.7293271700051361, + "eval_loss": 1.9319628477096558, + "eval_runtime": 43.9701, + "eval_samples_per_second": 22.743, + "eval_steps_per_second": 0.955, + "step": 710 + }, + { + "epoch": 0.7303543913713405, + "grad_norm": 2.5999035835266113, + "learning_rate": 4.231062047963457e-06, + "loss": 1.6442, + "step": 711 + }, + { + "epoch": 0.731381612737545, + "grad_norm": 2.611095666885376, + "learning_rate": 4.22915873620099e-06, + "loss": 1.6745, + "step": 712 + }, + { + "epoch": 0.7324088341037494, + "grad_norm": 2.7235333919525146, + "learning_rate": 4.227255424438524e-06, + "loss": 1.6353, + "step": 713 + }, + { + "epoch": 0.7334360554699538, + "grad_norm": 2.453639507293701, + "learning_rate": 4.225352112676057e-06, + "loss": 1.6657, + "step": 714 + }, + { + "epoch": 0.7344632768361582, + "grad_norm": 2.4031429290771484, + "learning_rate": 4.2234488009135895e-06, + "loss": 1.603, + "step": 715 + }, + { + "epoch": 0.7354904982023626, + "grad_norm": 2.631052017211914, + "learning_rate": 4.221545489151123e-06, + "loss": 1.5681, + "step": 716 + }, + { + "epoch": 0.736517719568567, + "grad_norm": 2.518758773803711, + "learning_rate": 4.219642177388656e-06, + "loss": 1.5941, + "step": 717 + }, + { + "epoch": 0.7375449409347714, + "grad_norm": 2.699491500854492, + "learning_rate": 4.2177388656261896e-06, + "loss": 1.6327, + "step": 718 + }, + { + "epoch": 0.7385721623009759, + "grad_norm": 2.749356985092163, + "learning_rate": 4.215835553863723e-06, + "loss": 1.5864, + "step": 719 + }, + { + "epoch": 0.7395993836671803, + "grad_norm": 2.5699031352996826, + "learning_rate": 4.213932242101256e-06, + "loss": 1.5444, + "step": 720 + }, + { + "epoch": 0.7395993836671803, + "eval_loss": 1.9521623849868774, + "eval_runtime": 46.5749, + "eval_samples_per_second": 21.471, + "eval_steps_per_second": 0.902, + "step": 720 + }, + { + "epoch": 0.7406266050333847, + "grad_norm": 3.197425127029419, + "learning_rate": 4.21202893033879e-06, + "loss": 1.6158, + "step": 721 + }, + { + "epoch": 0.7416538263995891, + "grad_norm": 2.619459390640259, + "learning_rate": 4.210125618576323e-06, + "loss": 1.6328, + "step": 722 + }, + { + "epoch": 0.7426810477657936, + "grad_norm": 2.792086601257324, + "learning_rate": 4.208222306813856e-06, + "loss": 1.6333, + "step": 723 + }, + { + "epoch": 0.743708269131998, + "grad_norm": 2.7023141384124756, + "learning_rate": 4.20631899505139e-06, + "loss": 1.6787, + "step": 724 + }, + { + "epoch": 0.7447354904982023, + "grad_norm": 2.6319735050201416, + "learning_rate": 4.204415683288923e-06, + "loss": 1.54, + "step": 725 + }, + { + "epoch": 0.7457627118644068, + "grad_norm": 2.7032439708709717, + "learning_rate": 4.2025123715264565e-06, + "loss": 1.641, + "step": 726 + }, + { + "epoch": 0.7467899332306112, + "grad_norm": 2.9133267402648926, + "learning_rate": 4.20060905976399e-06, + "loss": 1.5901, + "step": 727 + }, + { + "epoch": 0.7478171545968156, + "grad_norm": 2.492403745651245, + "learning_rate": 4.198705748001523e-06, + "loss": 1.5516, + "step": 728 + }, + { + "epoch": 0.74884437596302, + "grad_norm": 2.604520082473755, + "learning_rate": 4.1968024362390566e-06, + "loss": 1.6574, + "step": 729 + }, + { + "epoch": 0.7498715973292245, + "grad_norm": 2.521969795227051, + "learning_rate": 4.19489912447659e-06, + "loss": 1.6099, + "step": 730 + }, + { + "epoch": 0.7498715973292245, + "eval_loss": 1.9449840784072876, + "eval_runtime": 44.9334, + "eval_samples_per_second": 22.255, + "eval_steps_per_second": 0.935, + "step": 730 + }, + { + "epoch": 0.7508988186954288, + "grad_norm": 2.4645657539367676, + "learning_rate": 4.192995812714123e-06, + "loss": 1.6752, + "step": 731 + }, + { + "epoch": 0.7519260400616333, + "grad_norm": 2.7852981090545654, + "learning_rate": 4.191092500951656e-06, + "loss": 1.6176, + "step": 732 + }, + { + "epoch": 0.7529532614278377, + "grad_norm": 2.2817575931549072, + "learning_rate": 4.189189189189189e-06, + "loss": 1.5934, + "step": 733 + }, + { + "epoch": 0.7539804827940421, + "grad_norm": 2.6218173503875732, + "learning_rate": 4.1872858774267225e-06, + "loss": 1.5902, + "step": 734 + }, + { + "epoch": 0.7550077041602465, + "grad_norm": 2.6671013832092285, + "learning_rate": 4.185382565664256e-06, + "loss": 1.657, + "step": 735 + }, + { + "epoch": 0.7560349255264509, + "grad_norm": 2.669757843017578, + "learning_rate": 4.183479253901789e-06, + "loss": 1.5884, + "step": 736 + }, + { + "epoch": 0.7570621468926554, + "grad_norm": 2.7210471630096436, + "learning_rate": 4.181575942139323e-06, + "loss": 1.5558, + "step": 737 + }, + { + "epoch": 0.7580893682588598, + "grad_norm": 2.414276123046875, + "learning_rate": 4.179672630376856e-06, + "loss": 1.6124, + "step": 738 + }, + { + "epoch": 0.7591165896250642, + "grad_norm": 2.4643094539642334, + "learning_rate": 4.177769318614389e-06, + "loss": 1.5904, + "step": 739 + }, + { + "epoch": 0.7601438109912686, + "grad_norm": 2.5447018146514893, + "learning_rate": 4.175866006851923e-06, + "loss": 1.6705, + "step": 740 + }, + { + "epoch": 0.7601438109912686, + "eval_loss": 1.9437954425811768, + "eval_runtime": 49.2702, + "eval_samples_per_second": 20.296, + "eval_steps_per_second": 0.852, + "step": 740 + }, + { + "epoch": 0.7611710323574731, + "grad_norm": 2.41048002243042, + "learning_rate": 4.173962695089456e-06, + "loss": 1.6398, + "step": 741 + }, + { + "epoch": 0.7621982537236774, + "grad_norm": 2.3949906826019287, + "learning_rate": 4.1720593833269894e-06, + "loss": 1.6198, + "step": 742 + }, + { + "epoch": 0.7632254750898819, + "grad_norm": 2.36966609954834, + "learning_rate": 4.170156071564523e-06, + "loss": 1.6275, + "step": 743 + }, + { + "epoch": 0.7642526964560863, + "grad_norm": 2.623347043991089, + "learning_rate": 4.168252759802056e-06, + "loss": 1.7059, + "step": 744 + }, + { + "epoch": 0.7652799178222907, + "grad_norm": 2.384788751602173, + "learning_rate": 4.1663494480395895e-06, + "loss": 1.6015, + "step": 745 + }, + { + "epoch": 0.7663071391884951, + "grad_norm": 2.586009979248047, + "learning_rate": 4.164446136277123e-06, + "loss": 1.6445, + "step": 746 + }, + { + "epoch": 0.7673343605546995, + "grad_norm": 2.396684169769287, + "learning_rate": 4.162542824514656e-06, + "loss": 1.6539, + "step": 747 + }, + { + "epoch": 0.768361581920904, + "grad_norm": 2.416243553161621, + "learning_rate": 4.16063951275219e-06, + "loss": 1.6142, + "step": 748 + }, + { + "epoch": 0.7693888032871083, + "grad_norm": 2.480912208557129, + "learning_rate": 4.158736200989722e-06, + "loss": 1.5768, + "step": 749 + }, + { + "epoch": 0.7704160246533128, + "grad_norm": 2.5516929626464844, + "learning_rate": 4.1568328892272555e-06, + "loss": 1.6074, + "step": 750 + }, + { + "epoch": 0.7704160246533128, + "eval_loss": 1.9409116506576538, + "eval_runtime": 45.3465, + "eval_samples_per_second": 22.052, + "eval_steps_per_second": 0.926, + "step": 750 + }, + { + "epoch": 0.7714432460195172, + "grad_norm": 2.4248714447021484, + "learning_rate": 4.154929577464789e-06, + "loss": 1.6308, + "step": 751 + }, + { + "epoch": 0.7724704673857217, + "grad_norm": 2.520810604095459, + "learning_rate": 4.153026265702322e-06, + "loss": 1.6227, + "step": 752 + }, + { + "epoch": 0.773497688751926, + "grad_norm": 2.5938422679901123, + "learning_rate": 4.1511229539398556e-06, + "loss": 1.5601, + "step": 753 + }, + { + "epoch": 0.7745249101181304, + "grad_norm": 2.4538512229919434, + "learning_rate": 4.149219642177389e-06, + "loss": 1.6006, + "step": 754 + }, + { + "epoch": 0.7755521314843349, + "grad_norm": 2.4936258792877197, + "learning_rate": 4.147316330414922e-06, + "loss": 1.6281, + "step": 755 + }, + { + "epoch": 0.7765793528505393, + "grad_norm": 2.5774738788604736, + "learning_rate": 4.145413018652456e-06, + "loss": 1.5328, + "step": 756 + }, + { + "epoch": 0.7776065742167437, + "grad_norm": 2.3616905212402344, + "learning_rate": 4.143509706889989e-06, + "loss": 1.629, + "step": 757 + }, + { + "epoch": 0.7786337955829481, + "grad_norm": 2.4983086585998535, + "learning_rate": 4.141606395127522e-06, + "loss": 1.6298, + "step": 758 + }, + { + "epoch": 0.7796610169491526, + "grad_norm": 2.4313876628875732, + "learning_rate": 4.139703083365056e-06, + "loss": 1.6113, + "step": 759 + }, + { + "epoch": 0.7806882383153569, + "grad_norm": 2.6290934085845947, + "learning_rate": 4.137799771602588e-06, + "loss": 1.6632, + "step": 760 + }, + { + "epoch": 0.7806882383153569, + "eval_loss": 1.950036644935608, + "eval_runtime": 47.4523, + "eval_samples_per_second": 21.074, + "eval_steps_per_second": 0.885, + "step": 760 + }, + { + "epoch": 0.7817154596815614, + "grad_norm": 2.3637750148773193, + "learning_rate": 4.135896459840122e-06, + "loss": 1.6268, + "step": 761 + }, + { + "epoch": 0.7827426810477658, + "grad_norm": 2.5537843704223633, + "learning_rate": 4.133993148077656e-06, + "loss": 1.5736, + "step": 762 + }, + { + "epoch": 0.7837699024139703, + "grad_norm": 2.4945273399353027, + "learning_rate": 4.132089836315189e-06, + "loss": 1.612, + "step": 763 + }, + { + "epoch": 0.7847971237801746, + "grad_norm": 2.506983757019043, + "learning_rate": 4.1301865245527226e-06, + "loss": 1.6112, + "step": 764 + }, + { + "epoch": 0.785824345146379, + "grad_norm": 2.599721908569336, + "learning_rate": 4.128283212790256e-06, + "loss": 1.5698, + "step": 765 + }, + { + "epoch": 0.7868515665125835, + "grad_norm": 2.4561266899108887, + "learning_rate": 4.1263799010277884e-06, + "loss": 1.5354, + "step": 766 + }, + { + "epoch": 0.7878787878787878, + "grad_norm": 2.445420026779175, + "learning_rate": 4.124476589265322e-06, + "loss": 1.6555, + "step": 767 + }, + { + "epoch": 0.7889060092449923, + "grad_norm": 2.576533079147339, + "learning_rate": 4.122573277502855e-06, + "loss": 1.5915, + "step": 768 + }, + { + "epoch": 0.7899332306111967, + "grad_norm": 2.503164291381836, + "learning_rate": 4.1206699657403885e-06, + "loss": 1.5893, + "step": 769 + }, + { + "epoch": 0.7909604519774012, + "grad_norm": 2.9463603496551514, + "learning_rate": 4.118766653977922e-06, + "loss": 1.6178, + "step": 770 + }, + { + "epoch": 0.7909604519774012, + "eval_loss": 1.9529738426208496, + "eval_runtime": 45.0922, + "eval_samples_per_second": 22.177, + "eval_steps_per_second": 0.931, + "step": 770 + }, + { + "epoch": 0.7919876733436055, + "grad_norm": 2.47148060798645, + "learning_rate": 4.116863342215455e-06, + "loss": 1.5991, + "step": 771 + }, + { + "epoch": 0.79301489470981, + "grad_norm": 2.515717029571533, + "learning_rate": 4.114960030452989e-06, + "loss": 1.5725, + "step": 772 + }, + { + "epoch": 0.7940421160760144, + "grad_norm": 2.5231993198394775, + "learning_rate": 4.113056718690522e-06, + "loss": 1.6145, + "step": 773 + }, + { + "epoch": 0.7950693374422187, + "grad_norm": 2.479978084564209, + "learning_rate": 4.111153406928055e-06, + "loss": 1.5537, + "step": 774 + }, + { + "epoch": 0.7960965588084232, + "grad_norm": 2.5341250896453857, + "learning_rate": 4.109250095165589e-06, + "loss": 1.6419, + "step": 775 + }, + { + "epoch": 0.7971237801746276, + "grad_norm": 2.6277401447296143, + "learning_rate": 4.107346783403122e-06, + "loss": 1.6505, + "step": 776 + }, + { + "epoch": 0.7981510015408321, + "grad_norm": 2.4688351154327393, + "learning_rate": 4.105443471640655e-06, + "loss": 1.5896, + "step": 777 + }, + { + "epoch": 0.7991782229070364, + "grad_norm": 2.4134581089019775, + "learning_rate": 4.103540159878188e-06, + "loss": 1.5788, + "step": 778 + }, + { + "epoch": 0.8002054442732409, + "grad_norm": 2.5090012550354004, + "learning_rate": 4.101636848115721e-06, + "loss": 1.6162, + "step": 779 + }, + { + "epoch": 0.8012326656394453, + "grad_norm": 2.5123302936553955, + "learning_rate": 4.099733536353255e-06, + "loss": 1.5868, + "step": 780 + }, + { + "epoch": 0.8012326656394453, + "eval_loss": 1.9545625448226929, + "eval_runtime": 45.9316, + "eval_samples_per_second": 21.772, + "eval_steps_per_second": 0.914, + "step": 780 + }, + { + "epoch": 0.8022598870056498, + "grad_norm": 2.787181854248047, + "learning_rate": 4.097830224590788e-06, + "loss": 1.612, + "step": 781 + }, + { + "epoch": 0.8032871083718541, + "grad_norm": 2.414426326751709, + "learning_rate": 4.095926912828322e-06, + "loss": 1.6908, + "step": 782 + }, + { + "epoch": 0.8043143297380585, + "grad_norm": 2.452697992324829, + "learning_rate": 4.094023601065855e-06, + "loss": 1.6575, + "step": 783 + }, + { + "epoch": 0.805341551104263, + "grad_norm": 2.4377169609069824, + "learning_rate": 4.092120289303388e-06, + "loss": 1.6016, + "step": 784 + }, + { + "epoch": 0.8063687724704673, + "grad_norm": 2.369154214859009, + "learning_rate": 4.0902169775409215e-06, + "loss": 1.6133, + "step": 785 + }, + { + "epoch": 0.8073959938366718, + "grad_norm": 2.261869430541992, + "learning_rate": 4.088313665778455e-06, + "loss": 1.5555, + "step": 786 + }, + { + "epoch": 0.8084232152028762, + "grad_norm": 2.7058632373809814, + "learning_rate": 4.086410354015988e-06, + "loss": 1.5957, + "step": 787 + }, + { + "epoch": 0.8094504365690807, + "grad_norm": 2.3907508850097656, + "learning_rate": 4.0845070422535216e-06, + "loss": 1.5676, + "step": 788 + }, + { + "epoch": 0.810477657935285, + "grad_norm": 2.510284900665283, + "learning_rate": 4.082603730491055e-06, + "loss": 1.5872, + "step": 789 + }, + { + "epoch": 0.8115048793014895, + "grad_norm": 2.590810775756836, + "learning_rate": 4.080700418728588e-06, + "loss": 1.6315, + "step": 790 + }, + { + "epoch": 0.8115048793014895, + "eval_loss": 1.9573780298233032, + "eval_runtime": 44.5677, + "eval_samples_per_second": 22.438, + "eval_steps_per_second": 0.942, + "step": 790 + }, + { + "epoch": 0.8125321006676939, + "grad_norm": 2.466134786605835, + "learning_rate": 4.078797106966122e-06, + "loss": 1.6018, + "step": 791 + }, + { + "epoch": 0.8135593220338984, + "grad_norm": 2.474653482437134, + "learning_rate": 4.076893795203655e-06, + "loss": 1.5101, + "step": 792 + }, + { + "epoch": 0.8145865434001027, + "grad_norm": 2.4463517665863037, + "learning_rate": 4.074990483441188e-06, + "loss": 1.5686, + "step": 793 + }, + { + "epoch": 0.8156137647663071, + "grad_norm": 2.5148985385894775, + "learning_rate": 4.073087171678721e-06, + "loss": 1.6544, + "step": 794 + }, + { + "epoch": 0.8166409861325116, + "grad_norm": 2.6183927059173584, + "learning_rate": 4.071183859916254e-06, + "loss": 1.6277, + "step": 795 + }, + { + "epoch": 0.8176682074987159, + "grad_norm": 2.438427686691284, + "learning_rate": 4.069280548153788e-06, + "loss": 1.6002, + "step": 796 + }, + { + "epoch": 0.8186954288649204, + "grad_norm": 2.662855863571167, + "learning_rate": 4.067377236391321e-06, + "loss": 1.5594, + "step": 797 + }, + { + "epoch": 0.8197226502311248, + "grad_norm": 2.64789080619812, + "learning_rate": 4.065473924628854e-06, + "loss": 1.6007, + "step": 798 + }, + { + "epoch": 0.8207498715973293, + "grad_norm": 2.5152997970581055, + "learning_rate": 4.063570612866388e-06, + "loss": 1.6366, + "step": 799 + }, + { + "epoch": 0.8217770929635336, + "grad_norm": 2.468672037124634, + "learning_rate": 4.061667301103921e-06, + "loss": 1.6313, + "step": 800 + }, + { + "epoch": 0.8217770929635336, + "eval_loss": 1.95821213722229, + "eval_runtime": 45.9362, + "eval_samples_per_second": 21.769, + "eval_steps_per_second": 0.914, + "step": 800 + } + ], + "logging_steps": 1, + "max_steps": 2919, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 100, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 2.493847527489536e+17, + "train_batch_size": 3, + "trial_name": null, + "trial_params": null +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/zero_to_fp32.py b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-800/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fb0f9a53a507ffcdb60410deb71d60ef801bf350 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/config.json @@ -0,0 +1,36 @@ +{ + "_name_or_path": "meta-llama/Llama-3.2-3B", + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "eos_token_id": 128001, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 3072, + "initializer_range": 0.02, + "intermediate_size": 8192, + "max_position_embeddings": 131072, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 24, + "num_hidden_layers": 28, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": { + "factor": 32.0, + "high_freq_factor": 4.0, + "low_freq_factor": 1.0, + "original_max_position_embeddings": 8192, + "rope_type": "llama3" + }, + "rope_theta": 500000.0, + "tie_word_embeddings": true, + "torch_dtype": "float16", + "transformers_version": "4.45.1", + "use_cache": true, + "vocab_size": 128256 +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/generation_config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..9c389b841a9a29ddd904e02d1eb0e08dcce82ad9 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/generation_config.json @@ -0,0 +1,9 @@ +{ + "_from_model_config": true, + "bos_token_id": 128000, + "do_sample": true, + "eos_token_id": 128001, + "temperature": 0.6, + "top_p": 0.9, + "transformers_version": "4.45.1" +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/latest b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/latest new file mode 100644 index 0000000000000000000000000000000000000000..4b10acccf3e8395339ff8799cea202bbc54d7f7d --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/latest @@ -0,0 +1 @@ +global_step900 \ No newline at end of file diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/model.safetensors.index.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..ed64de846d720b9a7859dc20575fea8e8ca51940 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/model.safetensors.index.json @@ -0,0 +1,262 @@ +{ + "metadata": { + "total_size": 7213504512 + }, + "weight_map": { + "lm_head.weight": "model-00002-of-00002.safetensors", + "model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors", + "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "model.norm.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/special_tokens_map.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..6949c5975ee0e961ef61cf31010dce04df0a03f8 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/special_tokens_map.json @@ -0,0 +1,23 @@ +{ + "bos_token": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/tokenizer_config.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..0b782d4602b361f7577abb52b710a8c0e088588d --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/tokenizer_config.json @@ -0,0 +1,2078 @@ +{ + "added_tokens_decoder": { + "128000": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128001": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128002": { + "content": "<|reserved_special_token_0|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128003": { + "content": "<|reserved_special_token_1|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128004": { + "content": "<|finetune_right_pad_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128005": { + "content": "<|reserved_special_token_2|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128006": { + "content": "<|start_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128007": { + "content": "<|end_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128008": { + "content": "<|eom_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128009": { + "content": "<|eot_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128010": { + "content": "<|python_tag|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128011": { + "content": "<|reserved_special_token_3|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128012": { + "content": "<|reserved_special_token_4|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128013": { + "content": "<|reserved_special_token_5|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128014": { + "content": "<|reserved_special_token_6|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128015": { + "content": "<|reserved_special_token_7|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128016": { + "content": "<|reserved_special_token_8|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128017": { + "content": "<|reserved_special_token_9|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128018": { + "content": "<|reserved_special_token_10|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128019": { + "content": "<|reserved_special_token_11|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128020": { + "content": "<|reserved_special_token_12|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128021": { + "content": "<|reserved_special_token_13|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128022": { + "content": "<|reserved_special_token_14|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128023": { + "content": "<|reserved_special_token_15|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128024": { + "content": "<|reserved_special_token_16|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128025": { + "content": "<|reserved_special_token_17|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128026": { + "content": "<|reserved_special_token_18|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128027": { + "content": "<|reserved_special_token_19|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128028": { + "content": "<|reserved_special_token_20|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128029": { + "content": "<|reserved_special_token_21|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128030": { + "content": "<|reserved_special_token_22|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128031": { + "content": "<|reserved_special_token_23|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128032": { + "content": "<|reserved_special_token_24|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128033": { + "content": "<|reserved_special_token_25|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128034": { + "content": "<|reserved_special_token_26|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128035": { + "content": "<|reserved_special_token_27|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128036": { + "content": "<|reserved_special_token_28|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128037": { + "content": "<|reserved_special_token_29|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128038": { + "content": "<|reserved_special_token_30|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128039": { + "content": "<|reserved_special_token_31|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128040": { + "content": "<|reserved_special_token_32|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128041": { + "content": "<|reserved_special_token_33|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128042": { + "content": "<|reserved_special_token_34|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128043": { + "content": "<|reserved_special_token_35|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128044": { + "content": "<|reserved_special_token_36|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128045": { + "content": "<|reserved_special_token_37|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128046": { + "content": "<|reserved_special_token_38|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128047": { + "content": "<|reserved_special_token_39|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128048": { + "content": "<|reserved_special_token_40|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128049": { + "content": "<|reserved_special_token_41|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128050": { + "content": "<|reserved_special_token_42|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128051": { + "content": "<|reserved_special_token_43|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128052": { + "content": "<|reserved_special_token_44|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128053": { + "content": "<|reserved_special_token_45|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128054": { + "content": "<|reserved_special_token_46|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128055": { + "content": "<|reserved_special_token_47|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128056": { + "content": "<|reserved_special_token_48|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128057": { + "content": "<|reserved_special_token_49|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128058": { + "content": "<|reserved_special_token_50|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128059": { + "content": "<|reserved_special_token_51|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128060": { + "content": "<|reserved_special_token_52|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128061": { + "content": "<|reserved_special_token_53|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128062": { + "content": "<|reserved_special_token_54|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128063": { + "content": "<|reserved_special_token_55|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128064": { + "content": "<|reserved_special_token_56|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128065": { + "content": "<|reserved_special_token_57|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128066": { + "content": "<|reserved_special_token_58|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128067": { + "content": "<|reserved_special_token_59|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128068": { + "content": "<|reserved_special_token_60|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128069": { + "content": "<|reserved_special_token_61|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128070": { + "content": "<|reserved_special_token_62|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128071": { + "content": "<|reserved_special_token_63|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128072": { + "content": "<|reserved_special_token_64|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128073": { + "content": "<|reserved_special_token_65|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128074": { + "content": "<|reserved_special_token_66|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128075": { + "content": "<|reserved_special_token_67|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128076": { + "content": "<|reserved_special_token_68|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128077": { + "content": "<|reserved_special_token_69|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128078": { + "content": "<|reserved_special_token_70|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128079": { + "content": "<|reserved_special_token_71|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128080": { + "content": "<|reserved_special_token_72|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128081": { + "content": "<|reserved_special_token_73|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128082": { + "content": "<|reserved_special_token_74|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128083": { + "content": "<|reserved_special_token_75|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128084": { + "content": "<|reserved_special_token_76|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128085": { + "content": "<|reserved_special_token_77|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128086": { + "content": "<|reserved_special_token_78|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128087": { + "content": "<|reserved_special_token_79|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128088": { + "content": "<|reserved_special_token_80|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128089": { + "content": "<|reserved_special_token_81|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128090": { + "content": "<|reserved_special_token_82|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128091": { + "content": "<|reserved_special_token_83|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128092": { + "content": "<|reserved_special_token_84|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128093": { + "content": "<|reserved_special_token_85|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128094": { + "content": "<|reserved_special_token_86|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128095": { + "content": "<|reserved_special_token_87|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128096": { + "content": "<|reserved_special_token_88|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128097": { + "content": "<|reserved_special_token_89|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128098": { + "content": "<|reserved_special_token_90|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128099": { + "content": "<|reserved_special_token_91|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128100": { + "content": "<|reserved_special_token_92|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128101": { + "content": "<|reserved_special_token_93|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128102": { + "content": "<|reserved_special_token_94|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128103": { + "content": "<|reserved_special_token_95|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128104": { + "content": "<|reserved_special_token_96|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128105": { + "content": "<|reserved_special_token_97|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128106": { + "content": "<|reserved_special_token_98|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128107": { + "content": "<|reserved_special_token_99|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128108": { + "content": "<|reserved_special_token_100|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128109": { + "content": "<|reserved_special_token_101|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128110": { + "content": "<|reserved_special_token_102|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128111": { + "content": "<|reserved_special_token_103|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128112": { + "content": "<|reserved_special_token_104|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128113": { + "content": "<|reserved_special_token_105|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128114": { + "content": "<|reserved_special_token_106|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128115": { + "content": "<|reserved_special_token_107|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128116": { + "content": "<|reserved_special_token_108|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128117": { + "content": "<|reserved_special_token_109|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128118": { + "content": "<|reserved_special_token_110|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128119": { + "content": "<|reserved_special_token_111|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128120": { + "content": "<|reserved_special_token_112|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128121": { + "content": "<|reserved_special_token_113|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128122": { + "content": "<|reserved_special_token_114|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128123": { + "content": "<|reserved_special_token_115|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128124": { + "content": "<|reserved_special_token_116|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128125": { + "content": "<|reserved_special_token_117|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128126": { + "content": "<|reserved_special_token_118|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128127": { + "content": "<|reserved_special_token_119|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128128": { + "content": "<|reserved_special_token_120|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128129": { + "content": "<|reserved_special_token_121|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128130": { + "content": "<|reserved_special_token_122|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128131": { + "content": "<|reserved_special_token_123|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128132": { + "content": "<|reserved_special_token_124|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128133": { + "content": "<|reserved_special_token_125|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128134": { + "content": "<|reserved_special_token_126|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128135": { + "content": "<|reserved_special_token_127|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128136": { + "content": "<|reserved_special_token_128|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128137": { + "content": "<|reserved_special_token_129|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128138": { + "content": "<|reserved_special_token_130|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128139": { + "content": "<|reserved_special_token_131|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128140": { + "content": "<|reserved_special_token_132|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128141": { + "content": "<|reserved_special_token_133|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128142": { + "content": "<|reserved_special_token_134|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128143": { + "content": "<|reserved_special_token_135|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128144": { + "content": "<|reserved_special_token_136|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128145": { + "content": "<|reserved_special_token_137|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128146": { + "content": "<|reserved_special_token_138|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128147": { + "content": "<|reserved_special_token_139|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128148": { + "content": "<|reserved_special_token_140|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128149": { + "content": "<|reserved_special_token_141|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128150": { + "content": "<|reserved_special_token_142|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128151": { + "content": "<|reserved_special_token_143|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128152": { + "content": "<|reserved_special_token_144|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128153": { + "content": "<|reserved_special_token_145|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128154": { + "content": "<|reserved_special_token_146|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128155": { + "content": "<|reserved_special_token_147|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128156": { + "content": "<|reserved_special_token_148|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128157": { + "content": "<|reserved_special_token_149|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128158": { + "content": "<|reserved_special_token_150|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128159": { + "content": "<|reserved_special_token_151|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128160": { + "content": "<|reserved_special_token_152|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128161": { + "content": "<|reserved_special_token_153|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128162": { + "content": "<|reserved_special_token_154|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128163": { + "content": "<|reserved_special_token_155|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128164": { + "content": "<|reserved_special_token_156|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128165": { + "content": "<|reserved_special_token_157|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128166": { + "content": "<|reserved_special_token_158|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128167": { + "content": "<|reserved_special_token_159|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128168": { + "content": "<|reserved_special_token_160|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128169": { + "content": "<|reserved_special_token_161|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128170": { + "content": "<|reserved_special_token_162|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128171": { + "content": "<|reserved_special_token_163|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128172": { + "content": "<|reserved_special_token_164|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128173": { + "content": "<|reserved_special_token_165|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128174": { + "content": "<|reserved_special_token_166|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128175": { + "content": "<|reserved_special_token_167|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128176": { + "content": "<|reserved_special_token_168|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128177": { + "content": "<|reserved_special_token_169|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128178": { + "content": "<|reserved_special_token_170|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128179": { + "content": "<|reserved_special_token_171|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128180": { + "content": "<|reserved_special_token_172|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128181": { + "content": "<|reserved_special_token_173|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128182": { + "content": "<|reserved_special_token_174|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128183": { + "content": "<|reserved_special_token_175|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128184": { + "content": "<|reserved_special_token_176|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128185": { + "content": "<|reserved_special_token_177|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128186": { + "content": "<|reserved_special_token_178|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128187": { + "content": "<|reserved_special_token_179|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128188": { + "content": "<|reserved_special_token_180|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128189": { + "content": "<|reserved_special_token_181|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128190": { + "content": "<|reserved_special_token_182|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128191": { + "content": "<|reserved_special_token_183|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128192": { + "content": "<|reserved_special_token_184|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128193": { + "content": "<|reserved_special_token_185|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128194": { + "content": "<|reserved_special_token_186|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128195": { + "content": "<|reserved_special_token_187|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128196": { + "content": "<|reserved_special_token_188|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128197": { + "content": "<|reserved_special_token_189|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128198": { + "content": "<|reserved_special_token_190|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128199": { + "content": "<|reserved_special_token_191|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128200": { + "content": "<|reserved_special_token_192|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128201": { + "content": "<|reserved_special_token_193|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128202": { + "content": "<|reserved_special_token_194|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128203": { + "content": "<|reserved_special_token_195|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128204": { + "content": "<|reserved_special_token_196|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128205": { + "content": "<|reserved_special_token_197|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128206": { + "content": "<|reserved_special_token_198|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128207": { + "content": "<|reserved_special_token_199|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128208": { + "content": "<|reserved_special_token_200|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128209": { + "content": "<|reserved_special_token_201|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128210": { + "content": "<|reserved_special_token_202|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128211": { + "content": "<|reserved_special_token_203|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128212": { + "content": "<|reserved_special_token_204|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128213": { + "content": "<|reserved_special_token_205|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128214": { + "content": "<|reserved_special_token_206|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128215": { + "content": "<|reserved_special_token_207|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128216": { + "content": "<|reserved_special_token_208|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128217": { + "content": "<|reserved_special_token_209|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128218": { + "content": "<|reserved_special_token_210|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128219": { + "content": "<|reserved_special_token_211|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128220": { + "content": "<|reserved_special_token_212|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128221": { + "content": "<|reserved_special_token_213|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128222": { + "content": "<|reserved_special_token_214|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128223": { + "content": "<|reserved_special_token_215|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128224": { + "content": "<|reserved_special_token_216|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128225": { + "content": "<|reserved_special_token_217|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128226": { + "content": "<|reserved_special_token_218|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128227": { + "content": "<|reserved_special_token_219|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128228": { + "content": "<|reserved_special_token_220|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128229": { + "content": "<|reserved_special_token_221|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128230": { + "content": "<|reserved_special_token_222|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128231": { + "content": "<|reserved_special_token_223|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128232": { + "content": "<|reserved_special_token_224|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128233": { + "content": "<|reserved_special_token_225|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128234": { + "content": "<|reserved_special_token_226|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128235": { + "content": "<|reserved_special_token_227|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128236": { + "content": "<|reserved_special_token_228|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128237": { + "content": "<|reserved_special_token_229|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128238": { + "content": "<|reserved_special_token_230|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128239": { + "content": "<|reserved_special_token_231|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128240": { + "content": "<|reserved_special_token_232|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128241": { + "content": "<|reserved_special_token_233|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128242": { + "content": "<|reserved_special_token_234|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128243": { + "content": "<|reserved_special_token_235|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128244": { + "content": "<|reserved_special_token_236|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128245": { + "content": "<|reserved_special_token_237|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128246": { + "content": "<|reserved_special_token_238|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128247": { + "content": "<|reserved_special_token_239|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128248": { + "content": "<|reserved_special_token_240|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128249": { + "content": "<|reserved_special_token_241|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128250": { + "content": "<|reserved_special_token_242|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128251": { + "content": "<|reserved_special_token_243|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128252": { + "content": "<|reserved_special_token_244|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128253": { + "content": "<|reserved_special_token_245|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128254": { + "content": "<|reserved_special_token_246|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128255": { + "content": "<|reserved_special_token_247|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128256": { + "content": "[PAD]", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128257": { + "content": "🅁", + "lstrip": true, + "normalized": true, + "rstrip": false, + "single_word": false, + "special": false + } + }, + "bos_token": "<|begin_of_text|>", + "clean_up_tokenization_spaces": true, + "eos_token": "<|end_of_text|>", + "model_input_names": [ + "input_ids", + "attention_mask" + ], + "model_max_length": 131072, + "pad_token": "[PAD]", + "tokenizer_class": "PreTrainedTokenizerFast" +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/trainer_state.json b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..05a068f02f071b75bcc9dcf3b3dc93a6b504942a --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/trainer_state.json @@ -0,0 +1,7053 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.9244992295839753, + "eval_steps": 10, + "global_step": 900, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.001027221366204417, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7873, + "step": 1 + }, + { + "epoch": 0.002054442732408834, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7904, + "step": 2 + }, + { + "epoch": 0.0030816640986132513, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8348, + "step": 3 + }, + { + "epoch": 0.004108885464817668, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8357, + "step": 4 + }, + { + "epoch": 0.005136106831022085, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8153, + "step": 5 + }, + { + "epoch": 0.0061633281972265025, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.825, + "step": 6 + }, + { + "epoch": 0.007190549563430919, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8002, + "step": 7 + }, + { + "epoch": 0.008217770929635337, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8413, + "step": 8 + }, + { + "epoch": 0.009244992295839754, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8157, + "step": 9 + }, + { + "epoch": 0.01027221366204417, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8249, + "step": 10 + }, + { + "epoch": 0.01027221366204417, + "eval_loss": 2.8518412113189697, + "eval_runtime": 43.212, + "eval_samples_per_second": 23.142, + "eval_steps_per_second": 0.972, + "step": 10 + }, + { + "epoch": 0.011299435028248588, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7829, + "step": 11 + }, + { + "epoch": 0.012326656394453005, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.7819, + "step": 12 + }, + { + "epoch": 0.01335387776065742, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8355, + "step": 13 + }, + { + "epoch": 0.014381099126861838, + "grad_norm": 0.0, + "learning_rate": 0.0, + "loss": 2.8138, + "step": 14 + }, + { + "epoch": 0.015408320493066256, + "grad_norm": 5.162191390991211, + "learning_rate": 1.7123287671232876e-08, + "loss": 2.8088, + "step": 15 + }, + { + "epoch": 0.016435541859270673, + "grad_norm": 5.5954084396362305, + "learning_rate": 3.424657534246575e-08, + "loss": 2.7883, + "step": 16 + }, + { + "epoch": 0.01746276322547509, + "grad_norm": 5.542337417602539, + "learning_rate": 5.136986301369863e-08, + "loss": 2.7625, + "step": 17 + }, + { + "epoch": 0.01848998459167951, + "grad_norm": 5.300518989562988, + "learning_rate": 6.84931506849315e-08, + "loss": 2.84, + "step": 18 + }, + { + "epoch": 0.019517205957883924, + "grad_norm": 5.6646881103515625, + "learning_rate": 8.561643835616439e-08, + "loss": 2.761, + "step": 19 + }, + { + "epoch": 0.02054442732408834, + "grad_norm": 5.375557899475098, + "learning_rate": 1.0273972602739726e-07, + "loss": 2.8215, + "step": 20 + }, + { + "epoch": 0.02054442732408834, + "eval_loss": 2.8507144451141357, + "eval_runtime": 45.1133, + "eval_samples_per_second": 22.166, + "eval_steps_per_second": 0.931, + "step": 20 + }, + { + "epoch": 0.02157164869029276, + "grad_norm": 5.1731977462768555, + "learning_rate": 1.1986301369863014e-07, + "loss": 2.8348, + "step": 21 + }, + { + "epoch": 0.022598870056497175, + "grad_norm": 5.370945930480957, + "learning_rate": 1.36986301369863e-07, + "loss": 2.822, + "step": 22 + }, + { + "epoch": 0.02362609142270159, + "grad_norm": 5.550389289855957, + "learning_rate": 1.541095890410959e-07, + "loss": 2.7904, + "step": 23 + }, + { + "epoch": 0.02465331278890601, + "grad_norm": 5.339347839355469, + "learning_rate": 1.7123287671232878e-07, + "loss": 2.8066, + "step": 24 + }, + { + "epoch": 0.025680534155110426, + "grad_norm": 5.2319722175598145, + "learning_rate": 1.8835616438356165e-07, + "loss": 2.7976, + "step": 25 + }, + { + "epoch": 0.02670775552131484, + "grad_norm": 5.207149982452393, + "learning_rate": 2.0547945205479452e-07, + "loss": 2.8162, + "step": 26 + }, + { + "epoch": 0.02773497688751926, + "grad_norm": 4.400204181671143, + "learning_rate": 2.226027397260274e-07, + "loss": 2.7808, + "step": 27 + }, + { + "epoch": 0.028762198253723677, + "grad_norm": 4.21675968170166, + "learning_rate": 2.397260273972603e-07, + "loss": 2.8044, + "step": 28 + }, + { + "epoch": 0.029789419619928096, + "grad_norm": 4.406878471374512, + "learning_rate": 2.568493150684932e-07, + "loss": 2.8162, + "step": 29 + }, + { + "epoch": 0.030816640986132512, + "grad_norm": 4.024835586547852, + "learning_rate": 2.73972602739726e-07, + "loss": 2.7994, + "step": 30 + }, + { + "epoch": 0.030816640986132512, + "eval_loss": 2.8124029636383057, + "eval_runtime": 43.1772, + "eval_samples_per_second": 23.16, + "eval_steps_per_second": 0.973, + "step": 30 + }, + { + "epoch": 0.03184386235233693, + "grad_norm": 3.5378901958465576, + "learning_rate": 2.910958904109589e-07, + "loss": 2.7637, + "step": 31 + }, + { + "epoch": 0.03287108371854135, + "grad_norm": 3.6289916038513184, + "learning_rate": 3.082191780821918e-07, + "loss": 2.8017, + "step": 32 + }, + { + "epoch": 0.03389830508474576, + "grad_norm": 3.6789698600769043, + "learning_rate": 3.2534246575342466e-07, + "loss": 2.7723, + "step": 33 + }, + { + "epoch": 0.03492552645095018, + "grad_norm": 3.491389751434326, + "learning_rate": 3.4246575342465755e-07, + "loss": 2.7528, + "step": 34 + }, + { + "epoch": 0.035952747817154594, + "grad_norm": 3.383080005645752, + "learning_rate": 3.595890410958904e-07, + "loss": 2.7418, + "step": 35 + }, + { + "epoch": 0.03697996918335902, + "grad_norm": 3.4823594093322754, + "learning_rate": 3.767123287671233e-07, + "loss": 2.6662, + "step": 36 + }, + { + "epoch": 0.03800719054956343, + "grad_norm": 3.2146108150482178, + "learning_rate": 3.938356164383562e-07, + "loss": 2.7659, + "step": 37 + }, + { + "epoch": 0.03903441191576785, + "grad_norm": 2.9549412727355957, + "learning_rate": 4.1095890410958903e-07, + "loss": 2.6838, + "step": 38 + }, + { + "epoch": 0.040061633281972264, + "grad_norm": 3.0744080543518066, + "learning_rate": 4.2808219178082193e-07, + "loss": 2.6879, + "step": 39 + }, + { + "epoch": 0.04108885464817668, + "grad_norm": 3.19840669631958, + "learning_rate": 4.452054794520548e-07, + "loss": 2.7443, + "step": 40 + }, + { + "epoch": 0.04108885464817668, + "eval_loss": 2.7469444274902344, + "eval_runtime": 46.4676, + "eval_samples_per_second": 21.52, + "eval_steps_per_second": 0.904, + "step": 40 + }, + { + "epoch": 0.042116076014381096, + "grad_norm": 3.178406000137329, + "learning_rate": 4.6232876712328767e-07, + "loss": 2.7036, + "step": 41 + }, + { + "epoch": 0.04314329738058552, + "grad_norm": 3.2032673358917236, + "learning_rate": 4.794520547945206e-07, + "loss": 2.6465, + "step": 42 + }, + { + "epoch": 0.044170518746789934, + "grad_norm": 3.1525962352752686, + "learning_rate": 4.965753424657534e-07, + "loss": 2.6996, + "step": 43 + }, + { + "epoch": 0.04519774011299435, + "grad_norm": 2.907524585723877, + "learning_rate": 5.136986301369864e-07, + "loss": 2.6581, + "step": 44 + }, + { + "epoch": 0.046224961479198766, + "grad_norm": 2.5885250568389893, + "learning_rate": 5.308219178082192e-07, + "loss": 2.6773, + "step": 45 + }, + { + "epoch": 0.04725218284540318, + "grad_norm": 2.569068431854248, + "learning_rate": 5.47945205479452e-07, + "loss": 2.6635, + "step": 46 + }, + { + "epoch": 0.048279404211607604, + "grad_norm": 2.5313518047332764, + "learning_rate": 5.65068493150685e-07, + "loss": 2.706, + "step": 47 + }, + { + "epoch": 0.04930662557781202, + "grad_norm": 2.615442991256714, + "learning_rate": 5.821917808219178e-07, + "loss": 2.638, + "step": 48 + }, + { + "epoch": 0.050333846944016436, + "grad_norm": 2.7420806884765625, + "learning_rate": 5.993150684931507e-07, + "loss": 2.6398, + "step": 49 + }, + { + "epoch": 0.05136106831022085, + "grad_norm": 2.5246975421905518, + "learning_rate": 6.164383561643836e-07, + "loss": 2.5916, + "step": 50 + }, + { + "epoch": 0.05136106831022085, + "eval_loss": 2.6723692417144775, + "eval_runtime": 46.7206, + "eval_samples_per_second": 21.404, + "eval_steps_per_second": 0.899, + "step": 50 + }, + { + "epoch": 0.05238828967642527, + "grad_norm": 2.390143632888794, + "learning_rate": 6.335616438356165e-07, + "loss": 2.5773, + "step": 51 + }, + { + "epoch": 0.05341551104262968, + "grad_norm": 2.0792484283447266, + "learning_rate": 6.506849315068493e-07, + "loss": 2.6249, + "step": 52 + }, + { + "epoch": 0.054442732408834106, + "grad_norm": 1.9995019435882568, + "learning_rate": 6.678082191780823e-07, + "loss": 2.5425, + "step": 53 + }, + { + "epoch": 0.05546995377503852, + "grad_norm": 1.985452651977539, + "learning_rate": 6.849315068493151e-07, + "loss": 2.5986, + "step": 54 + }, + { + "epoch": 0.05649717514124294, + "grad_norm": 2.10184907913208, + "learning_rate": 7.020547945205481e-07, + "loss": 2.5713, + "step": 55 + }, + { + "epoch": 0.05752439650744735, + "grad_norm": 2.0898985862731934, + "learning_rate": 7.191780821917808e-07, + "loss": 2.535, + "step": 56 + }, + { + "epoch": 0.05855161787365177, + "grad_norm": 2.0239579677581787, + "learning_rate": 7.363013698630137e-07, + "loss": 2.5667, + "step": 57 + }, + { + "epoch": 0.05957883923985619, + "grad_norm": 2.207080125808716, + "learning_rate": 7.534246575342466e-07, + "loss": 2.4999, + "step": 58 + }, + { + "epoch": 0.06060606060606061, + "grad_norm": 1.8541918992996216, + "learning_rate": 7.705479452054795e-07, + "loss": 2.5582, + "step": 59 + }, + { + "epoch": 0.061633281972265024, + "grad_norm": 1.8530476093292236, + "learning_rate": 7.876712328767124e-07, + "loss": 2.5535, + "step": 60 + }, + { + "epoch": 0.061633281972265024, + "eval_loss": 2.607963800430298, + "eval_runtime": 45.9754, + "eval_samples_per_second": 21.751, + "eval_steps_per_second": 0.914, + "step": 60 + }, + { + "epoch": 0.06266050333846944, + "grad_norm": 1.7978090047836304, + "learning_rate": 8.047945205479453e-07, + "loss": 2.53, + "step": 61 + }, + { + "epoch": 0.06368772470467386, + "grad_norm": 1.8173762559890747, + "learning_rate": 8.219178082191781e-07, + "loss": 2.5144, + "step": 62 + }, + { + "epoch": 0.06471494607087827, + "grad_norm": 2.0847485065460205, + "learning_rate": 8.39041095890411e-07, + "loss": 2.5525, + "step": 63 + }, + { + "epoch": 0.0657421674370827, + "grad_norm": 1.8691741228103638, + "learning_rate": 8.561643835616439e-07, + "loss": 2.5451, + "step": 64 + }, + { + "epoch": 0.0667693888032871, + "grad_norm": 2.0242726802825928, + "learning_rate": 8.732876712328768e-07, + "loss": 2.4145, + "step": 65 + }, + { + "epoch": 0.06779661016949153, + "grad_norm": 1.8268908262252808, + "learning_rate": 8.904109589041097e-07, + "loss": 2.5024, + "step": 66 + }, + { + "epoch": 0.06882383153569595, + "grad_norm": 1.8703057765960693, + "learning_rate": 9.075342465753426e-07, + "loss": 2.461, + "step": 67 + }, + { + "epoch": 0.06985105290190036, + "grad_norm": 1.8747144937515259, + "learning_rate": 9.246575342465753e-07, + "loss": 2.5122, + "step": 68 + }, + { + "epoch": 0.07087827426810478, + "grad_norm": 1.744924783706665, + "learning_rate": 9.417808219178083e-07, + "loss": 2.4673, + "step": 69 + }, + { + "epoch": 0.07190549563430919, + "grad_norm": 1.7492460012435913, + "learning_rate": 9.589041095890411e-07, + "loss": 2.5598, + "step": 70 + }, + { + "epoch": 0.07190549563430919, + "eval_loss": 2.547417640686035, + "eval_runtime": 49.7488, + "eval_samples_per_second": 20.101, + "eval_steps_per_second": 0.844, + "step": 70 + }, + { + "epoch": 0.07293271700051361, + "grad_norm": 1.9218230247497559, + "learning_rate": 9.76027397260274e-07, + "loss": 2.4562, + "step": 71 + }, + { + "epoch": 0.07395993836671803, + "grad_norm": 2.0127363204956055, + "learning_rate": 9.931506849315068e-07, + "loss": 2.4141, + "step": 72 + }, + { + "epoch": 0.07498715973292244, + "grad_norm": 1.8700369596481323, + "learning_rate": 1.0102739726027399e-06, + "loss": 2.4696, + "step": 73 + }, + { + "epoch": 0.07601438109912687, + "grad_norm": 2.1706016063690186, + "learning_rate": 1.0273972602739727e-06, + "loss": 2.4404, + "step": 74 + }, + { + "epoch": 0.07704160246533127, + "grad_norm": 2.470412254333496, + "learning_rate": 1.0445205479452056e-06, + "loss": 2.4409, + "step": 75 + }, + { + "epoch": 0.0780688238315357, + "grad_norm": 1.906569242477417, + "learning_rate": 1.0616438356164384e-06, + "loss": 2.4313, + "step": 76 + }, + { + "epoch": 0.07909604519774012, + "grad_norm": 2.122936487197876, + "learning_rate": 1.0787671232876712e-06, + "loss": 2.4369, + "step": 77 + }, + { + "epoch": 0.08012326656394453, + "grad_norm": 2.079127311706543, + "learning_rate": 1.095890410958904e-06, + "loss": 2.449, + "step": 78 + }, + { + "epoch": 0.08115048793014895, + "grad_norm": 1.9173146486282349, + "learning_rate": 1.1130136986301371e-06, + "loss": 2.3782, + "step": 79 + }, + { + "epoch": 0.08217770929635336, + "grad_norm": 1.9856903553009033, + "learning_rate": 1.13013698630137e-06, + "loss": 2.4215, + "step": 80 + }, + { + "epoch": 0.08217770929635336, + "eval_loss": 2.479673385620117, + "eval_runtime": 46.2211, + "eval_samples_per_second": 21.635, + "eval_steps_per_second": 0.909, + "step": 80 + }, + { + "epoch": 0.08320493066255778, + "grad_norm": 2.3764865398406982, + "learning_rate": 1.1472602739726028e-06, + "loss": 2.3744, + "step": 81 + }, + { + "epoch": 0.08423215202876219, + "grad_norm": 1.980668544769287, + "learning_rate": 1.1643835616438357e-06, + "loss": 2.3856, + "step": 82 + }, + { + "epoch": 0.08525937339496661, + "grad_norm": 2.2589704990386963, + "learning_rate": 1.1815068493150685e-06, + "loss": 2.3853, + "step": 83 + }, + { + "epoch": 0.08628659476117104, + "grad_norm": 2.171780824661255, + "learning_rate": 1.1986301369863014e-06, + "loss": 2.3603, + "step": 84 + }, + { + "epoch": 0.08731381612737545, + "grad_norm": 2.25736403465271, + "learning_rate": 1.2157534246575344e-06, + "loss": 2.3189, + "step": 85 + }, + { + "epoch": 0.08834103749357987, + "grad_norm": 2.3643763065338135, + "learning_rate": 1.2328767123287673e-06, + "loss": 2.3795, + "step": 86 + }, + { + "epoch": 0.08936825885978428, + "grad_norm": 2.384516954421997, + "learning_rate": 1.25e-06, + "loss": 2.3257, + "step": 87 + }, + { + "epoch": 0.0903954802259887, + "grad_norm": 2.5072567462921143, + "learning_rate": 1.267123287671233e-06, + "loss": 2.3125, + "step": 88 + }, + { + "epoch": 0.09142270159219312, + "grad_norm": 2.2726807594299316, + "learning_rate": 1.284246575342466e-06, + "loss": 2.3297, + "step": 89 + }, + { + "epoch": 0.09244992295839753, + "grad_norm": 2.4817421436309814, + "learning_rate": 1.3013698630136986e-06, + "loss": 2.2878, + "step": 90 + }, + { + "epoch": 0.09244992295839753, + "eval_loss": 2.4065146446228027, + "eval_runtime": 47.9372, + "eval_samples_per_second": 20.861, + "eval_steps_per_second": 0.876, + "step": 90 + }, + { + "epoch": 0.09347714432460195, + "grad_norm": 2.2219603061676025, + "learning_rate": 1.3184931506849317e-06, + "loss": 2.2978, + "step": 91 + }, + { + "epoch": 0.09450436569080636, + "grad_norm": 2.8790745735168457, + "learning_rate": 1.3356164383561645e-06, + "loss": 2.313, + "step": 92 + }, + { + "epoch": 0.09553158705701079, + "grad_norm": 2.589200496673584, + "learning_rate": 1.3527397260273976e-06, + "loss": 2.2937, + "step": 93 + }, + { + "epoch": 0.09655880842321521, + "grad_norm": 2.760016679763794, + "learning_rate": 1.3698630136986302e-06, + "loss": 2.277, + "step": 94 + }, + { + "epoch": 0.09758602978941962, + "grad_norm": 2.2641494274139404, + "learning_rate": 1.386986301369863e-06, + "loss": 2.3336, + "step": 95 + }, + { + "epoch": 0.09861325115562404, + "grad_norm": 2.2786386013031006, + "learning_rate": 1.4041095890410961e-06, + "loss": 2.233, + "step": 96 + }, + { + "epoch": 0.09964047252182845, + "grad_norm": 2.5590052604675293, + "learning_rate": 1.421232876712329e-06, + "loss": 2.2851, + "step": 97 + }, + { + "epoch": 0.10066769388803287, + "grad_norm": 2.5609471797943115, + "learning_rate": 1.4383561643835616e-06, + "loss": 2.1848, + "step": 98 + }, + { + "epoch": 0.1016949152542373, + "grad_norm": 2.442521810531616, + "learning_rate": 1.4554794520547946e-06, + "loss": 2.2073, + "step": 99 + }, + { + "epoch": 0.1027221366204417, + "grad_norm": 2.3735859394073486, + "learning_rate": 1.4726027397260275e-06, + "loss": 2.2353, + "step": 100 + }, + { + "epoch": 0.1027221366204417, + "eval_loss": 2.3457911014556885, + "eval_runtime": 51.6469, + "eval_samples_per_second": 19.362, + "eval_steps_per_second": 0.813, + "step": 100 + }, + { + "epoch": 0.10374935798664613, + "grad_norm": 2.175354242324829, + "learning_rate": 1.4897260273972605e-06, + "loss": 2.2259, + "step": 101 + }, + { + "epoch": 0.10477657935285054, + "grad_norm": 2.217677116394043, + "learning_rate": 1.5068493150684932e-06, + "loss": 2.2082, + "step": 102 + }, + { + "epoch": 0.10580380071905496, + "grad_norm": 2.4618067741394043, + "learning_rate": 1.5239726027397262e-06, + "loss": 2.1649, + "step": 103 + }, + { + "epoch": 0.10683102208525937, + "grad_norm": 2.232060432434082, + "learning_rate": 1.541095890410959e-06, + "loss": 2.1962, + "step": 104 + }, + { + "epoch": 0.10785824345146379, + "grad_norm": 2.9140872955322266, + "learning_rate": 1.5582191780821921e-06, + "loss": 2.2446, + "step": 105 + }, + { + "epoch": 0.10888546481766821, + "grad_norm": 2.232158660888672, + "learning_rate": 1.5753424657534248e-06, + "loss": 2.2179, + "step": 106 + }, + { + "epoch": 0.10991268618387262, + "grad_norm": 2.804938793182373, + "learning_rate": 1.5924657534246576e-06, + "loss": 2.2449, + "step": 107 + }, + { + "epoch": 0.11093990755007704, + "grad_norm": 2.3690710067749023, + "learning_rate": 1.6095890410958907e-06, + "loss": 2.225, + "step": 108 + }, + { + "epoch": 0.11196712891628145, + "grad_norm": 2.7717416286468506, + "learning_rate": 1.6267123287671235e-06, + "loss": 2.0922, + "step": 109 + }, + { + "epoch": 0.11299435028248588, + "grad_norm": 3.053602933883667, + "learning_rate": 1.6438356164383561e-06, + "loss": 2.1702, + "step": 110 + }, + { + "epoch": 0.11299435028248588, + "eval_loss": 2.298161029815674, + "eval_runtime": 44.0776, + "eval_samples_per_second": 22.687, + "eval_steps_per_second": 0.953, + "step": 110 + }, + { + "epoch": 0.1140215716486903, + "grad_norm": 2.8673532009124756, + "learning_rate": 1.6609589041095892e-06, + "loss": 2.0938, + "step": 111 + }, + { + "epoch": 0.1150487930148947, + "grad_norm": 2.57185959815979, + "learning_rate": 1.678082191780822e-06, + "loss": 2.1892, + "step": 112 + }, + { + "epoch": 0.11607601438109913, + "grad_norm": 2.5278828144073486, + "learning_rate": 1.695205479452055e-06, + "loss": 2.1251, + "step": 113 + }, + { + "epoch": 0.11710323574730354, + "grad_norm": 2.8581748008728027, + "learning_rate": 1.7123287671232877e-06, + "loss": 2.1699, + "step": 114 + }, + { + "epoch": 0.11813045711350796, + "grad_norm": 2.924959897994995, + "learning_rate": 1.7294520547945206e-06, + "loss": 2.1005, + "step": 115 + }, + { + "epoch": 0.11915767847971238, + "grad_norm": 2.477034091949463, + "learning_rate": 1.7465753424657536e-06, + "loss": 2.2016, + "step": 116 + }, + { + "epoch": 0.12018489984591679, + "grad_norm": 2.8719305992126465, + "learning_rate": 1.7636986301369865e-06, + "loss": 2.1588, + "step": 117 + }, + { + "epoch": 0.12121212121212122, + "grad_norm": 2.4403953552246094, + "learning_rate": 1.7808219178082193e-06, + "loss": 2.1445, + "step": 118 + }, + { + "epoch": 0.12223934257832562, + "grad_norm": 2.7994632720947266, + "learning_rate": 1.7979452054794521e-06, + "loss": 2.1369, + "step": 119 + }, + { + "epoch": 0.12326656394453005, + "grad_norm": 2.163196563720703, + "learning_rate": 1.8150684931506852e-06, + "loss": 2.2195, + "step": 120 + }, + { + "epoch": 0.12326656394453005, + "eval_loss": 2.2648775577545166, + "eval_runtime": 49.47, + "eval_samples_per_second": 20.214, + "eval_steps_per_second": 0.849, + "step": 120 + }, + { + "epoch": 0.12429378531073447, + "grad_norm": 2.6743483543395996, + "learning_rate": 1.832191780821918e-06, + "loss": 2.1521, + "step": 121 + }, + { + "epoch": 0.12532100667693888, + "grad_norm": 2.54795503616333, + "learning_rate": 1.8493150684931507e-06, + "loss": 2.1158, + "step": 122 + }, + { + "epoch": 0.1263482280431433, + "grad_norm": 2.9093267917633057, + "learning_rate": 1.8664383561643837e-06, + "loss": 2.0707, + "step": 123 + }, + { + "epoch": 0.12737544940934772, + "grad_norm": 2.416940212249756, + "learning_rate": 1.8835616438356166e-06, + "loss": 2.0901, + "step": 124 + }, + { + "epoch": 0.12840267077555212, + "grad_norm": 2.8333663940429688, + "learning_rate": 1.9006849315068496e-06, + "loss": 2.0545, + "step": 125 + }, + { + "epoch": 0.12942989214175654, + "grad_norm": 2.483051061630249, + "learning_rate": 1.9178082191780823e-06, + "loss": 2.1274, + "step": 126 + }, + { + "epoch": 0.13045711350796096, + "grad_norm": 2.3486859798431396, + "learning_rate": 1.9349315068493153e-06, + "loss": 2.0478, + "step": 127 + }, + { + "epoch": 0.1314843348741654, + "grad_norm": 3.1512887477874756, + "learning_rate": 1.952054794520548e-06, + "loss": 2.1041, + "step": 128 + }, + { + "epoch": 0.1325115562403698, + "grad_norm": 2.902873992919922, + "learning_rate": 1.969178082191781e-06, + "loss": 2.1332, + "step": 129 + }, + { + "epoch": 0.1335387776065742, + "grad_norm": 2.779132127761841, + "learning_rate": 1.9863013698630136e-06, + "loss": 2.1297, + "step": 130 + }, + { + "epoch": 0.1335387776065742, + "eval_loss": 2.2391879558563232, + "eval_runtime": 55.3018, + "eval_samples_per_second": 18.083, + "eval_steps_per_second": 0.759, + "step": 130 + }, + { + "epoch": 0.13456599897277863, + "grad_norm": 2.703019618988037, + "learning_rate": 2.0034246575342467e-06, + "loss": 2.0506, + "step": 131 + }, + { + "epoch": 0.13559322033898305, + "grad_norm": 3.0187575817108154, + "learning_rate": 2.0205479452054797e-06, + "loss": 2.1291, + "step": 132 + }, + { + "epoch": 0.13662044170518747, + "grad_norm": 2.7629971504211426, + "learning_rate": 2.037671232876713e-06, + "loss": 2.1003, + "step": 133 + }, + { + "epoch": 0.1376476630713919, + "grad_norm": 3.175776720046997, + "learning_rate": 2.0547945205479454e-06, + "loss": 2.0429, + "step": 134 + }, + { + "epoch": 0.1386748844375963, + "grad_norm": 3.0342514514923096, + "learning_rate": 2.071917808219178e-06, + "loss": 2.0858, + "step": 135 + }, + { + "epoch": 0.1397021058038007, + "grad_norm": 3.0861928462982178, + "learning_rate": 2.089041095890411e-06, + "loss": 2.0964, + "step": 136 + }, + { + "epoch": 0.14072932717000514, + "grad_norm": 3.151945114135742, + "learning_rate": 2.106164383561644e-06, + "loss": 2.0194, + "step": 137 + }, + { + "epoch": 0.14175654853620956, + "grad_norm": 2.9069676399230957, + "learning_rate": 2.123287671232877e-06, + "loss": 2.1131, + "step": 138 + }, + { + "epoch": 0.14278376990241398, + "grad_norm": 3.2285282611846924, + "learning_rate": 2.14041095890411e-06, + "loss": 2.0643, + "step": 139 + }, + { + "epoch": 0.14381099126861838, + "grad_norm": 2.500481367111206, + "learning_rate": 2.1575342465753425e-06, + "loss": 2.0625, + "step": 140 + }, + { + "epoch": 0.14381099126861838, + "eval_loss": 2.211498975753784, + "eval_runtime": 55.1046, + "eval_samples_per_second": 18.147, + "eval_steps_per_second": 0.762, + "step": 140 + }, + { + "epoch": 0.1448382126348228, + "grad_norm": 4.004592418670654, + "learning_rate": 2.1746575342465755e-06, + "loss": 2.0644, + "step": 141 + }, + { + "epoch": 0.14586543400102722, + "grad_norm": 2.9258391857147217, + "learning_rate": 2.191780821917808e-06, + "loss": 2.0174, + "step": 142 + }, + { + "epoch": 0.14689265536723164, + "grad_norm": 3.0073509216308594, + "learning_rate": 2.2089041095890412e-06, + "loss": 2.1417, + "step": 143 + }, + { + "epoch": 0.14791987673343607, + "grad_norm": 3.040708065032959, + "learning_rate": 2.2260273972602743e-06, + "loss": 2.0357, + "step": 144 + }, + { + "epoch": 0.14894709809964046, + "grad_norm": 2.9062869548797607, + "learning_rate": 2.243150684931507e-06, + "loss": 2.0122, + "step": 145 + }, + { + "epoch": 0.14997431946584489, + "grad_norm": 3.28969669342041, + "learning_rate": 2.26027397260274e-06, + "loss": 2.0255, + "step": 146 + }, + { + "epoch": 0.1510015408320493, + "grad_norm": 3.608752727508545, + "learning_rate": 2.2773972602739726e-06, + "loss": 2.0184, + "step": 147 + }, + { + "epoch": 0.15202876219825373, + "grad_norm": 2.900341033935547, + "learning_rate": 2.2945205479452057e-06, + "loss": 2.0532, + "step": 148 + }, + { + "epoch": 0.15305598356445815, + "grad_norm": 2.881843328475952, + "learning_rate": 2.3116438356164387e-06, + "loss": 2.1336, + "step": 149 + }, + { + "epoch": 0.15408320493066255, + "grad_norm": 3.4043259620666504, + "learning_rate": 2.3287671232876713e-06, + "loss": 2.0536, + "step": 150 + }, + { + "epoch": 0.15408320493066255, + "eval_loss": 2.192880153656006, + "eval_runtime": 53.7838, + "eval_samples_per_second": 18.593, + "eval_steps_per_second": 0.781, + "step": 150 + }, + { + "epoch": 0.15511042629686697, + "grad_norm": 2.693107843399048, + "learning_rate": 2.3458904109589044e-06, + "loss": 1.9503, + "step": 151 + }, + { + "epoch": 0.1561376476630714, + "grad_norm": 2.885260581970215, + "learning_rate": 2.363013698630137e-06, + "loss": 2.0081, + "step": 152 + }, + { + "epoch": 0.15716486902927582, + "grad_norm": 3.4459924697875977, + "learning_rate": 2.38013698630137e-06, + "loss": 2.0649, + "step": 153 + }, + { + "epoch": 0.15819209039548024, + "grad_norm": 2.8447093963623047, + "learning_rate": 2.3972602739726027e-06, + "loss": 2.0476, + "step": 154 + }, + { + "epoch": 0.15921931176168463, + "grad_norm": 2.755176305770874, + "learning_rate": 2.4143835616438358e-06, + "loss": 2.0006, + "step": 155 + }, + { + "epoch": 0.16024653312788906, + "grad_norm": 2.8677561283111572, + "learning_rate": 2.431506849315069e-06, + "loss": 1.9663, + "step": 156 + }, + { + "epoch": 0.16127375449409348, + "grad_norm": 2.783351182937622, + "learning_rate": 2.4486301369863015e-06, + "loss": 2.0301, + "step": 157 + }, + { + "epoch": 0.1623009758602979, + "grad_norm": 2.7545998096466064, + "learning_rate": 2.4657534246575345e-06, + "loss": 2.0438, + "step": 158 + }, + { + "epoch": 0.1633281972265023, + "grad_norm": 2.8140594959259033, + "learning_rate": 2.482876712328767e-06, + "loss": 2.0055, + "step": 159 + }, + { + "epoch": 0.16435541859270672, + "grad_norm": 2.6072335243225098, + "learning_rate": 2.5e-06, + "loss": 2.0262, + "step": 160 + }, + { + "epoch": 0.16435541859270672, + "eval_loss": 2.179842710494995, + "eval_runtime": 52.6159, + "eval_samples_per_second": 19.006, + "eval_steps_per_second": 0.798, + "step": 160 + }, + { + "epoch": 0.16538263995891114, + "grad_norm": 2.830212354660034, + "learning_rate": 2.5171232876712333e-06, + "loss": 2.0154, + "step": 161 + }, + { + "epoch": 0.16640986132511557, + "grad_norm": 2.677140712738037, + "learning_rate": 2.534246575342466e-06, + "loss": 1.9933, + "step": 162 + }, + { + "epoch": 0.16743708269132, + "grad_norm": 2.837759017944336, + "learning_rate": 2.551369863013699e-06, + "loss": 1.9602, + "step": 163 + }, + { + "epoch": 0.16846430405752438, + "grad_norm": 3.9556589126586914, + "learning_rate": 2.568493150684932e-06, + "loss": 2.0121, + "step": 164 + }, + { + "epoch": 0.1694915254237288, + "grad_norm": 2.8715925216674805, + "learning_rate": 2.585616438356164e-06, + "loss": 2.0219, + "step": 165 + }, + { + "epoch": 0.17051874678993323, + "grad_norm": 3.5073623657226562, + "learning_rate": 2.6027397260273973e-06, + "loss": 2.0202, + "step": 166 + }, + { + "epoch": 0.17154596815613765, + "grad_norm": 2.751962900161743, + "learning_rate": 2.6198630136986303e-06, + "loss": 1.9528, + "step": 167 + }, + { + "epoch": 0.17257318952234207, + "grad_norm": 3.7753474712371826, + "learning_rate": 2.6369863013698634e-06, + "loss": 2.0208, + "step": 168 + }, + { + "epoch": 0.17360041088854647, + "grad_norm": 2.642289876937866, + "learning_rate": 2.654109589041096e-06, + "loss": 2.0386, + "step": 169 + }, + { + "epoch": 0.1746276322547509, + "grad_norm": 2.82173490524292, + "learning_rate": 2.671232876712329e-06, + "loss": 2.0355, + "step": 170 + }, + { + "epoch": 0.1746276322547509, + "eval_loss": 2.1699249744415283, + "eval_runtime": 50.0341, + "eval_samples_per_second": 19.986, + "eval_steps_per_second": 0.839, + "step": 170 + }, + { + "epoch": 0.17565485362095531, + "grad_norm": 3.342280149459839, + "learning_rate": 2.688356164383562e-06, + "loss": 1.9672, + "step": 171 + }, + { + "epoch": 0.17668207498715974, + "grad_norm": 2.6548173427581787, + "learning_rate": 2.705479452054795e-06, + "loss": 1.9669, + "step": 172 + }, + { + "epoch": 0.17770929635336416, + "grad_norm": 2.6548173427581787, + "learning_rate": 2.705479452054795e-06, + "loss": 1.961, + "step": 173 + }, + { + "epoch": 0.17873651771956856, + "grad_norm": 3.4306116104125977, + "learning_rate": 2.7226027397260274e-06, + "loss": 1.9173, + "step": 174 + }, + { + "epoch": 0.17976373908577298, + "grad_norm": 2.7991764545440674, + "learning_rate": 2.7397260273972604e-06, + "loss": 2.0339, + "step": 175 + }, + { + "epoch": 0.1807909604519774, + "grad_norm": 3.6040217876434326, + "learning_rate": 2.7568493150684935e-06, + "loss": 2.0218, + "step": 176 + }, + { + "epoch": 0.18181818181818182, + "grad_norm": 3.0253515243530273, + "learning_rate": 2.773972602739726e-06, + "loss": 1.9245, + "step": 177 + }, + { + "epoch": 0.18284540318438625, + "grad_norm": 3.405141830444336, + "learning_rate": 2.791095890410959e-06, + "loss": 2.0136, + "step": 178 + }, + { + "epoch": 0.18387262455059064, + "grad_norm": 2.7518203258514404, + "learning_rate": 2.8082191780821922e-06, + "loss": 1.8925, + "step": 179 + }, + { + "epoch": 0.18489984591679506, + "grad_norm": 3.1903367042541504, + "learning_rate": 2.8253424657534253e-06, + "loss": 2.0523, + "step": 180 + }, + { + "epoch": 0.18489984591679506, + "eval_loss": 2.158029794692993, + "eval_runtime": 48.6162, + "eval_samples_per_second": 20.569, + "eval_steps_per_second": 0.864, + "step": 180 + }, + { + "epoch": 0.1859270672829995, + "grad_norm": 3.0331413745880127, + "learning_rate": 2.842465753424658e-06, + "loss": 2.0167, + "step": 181 + }, + { + "epoch": 0.1869542886492039, + "grad_norm": 2.919433116912842, + "learning_rate": 2.8595890410958905e-06, + "loss": 2.0197, + "step": 182 + }, + { + "epoch": 0.18798151001540833, + "grad_norm": 2.9013049602508545, + "learning_rate": 2.876712328767123e-06, + "loss": 2.0301, + "step": 183 + }, + { + "epoch": 0.18900873138161273, + "grad_norm": 2.738800525665283, + "learning_rate": 2.8938356164383562e-06, + "loss": 1.894, + "step": 184 + }, + { + "epoch": 0.19003595274781715, + "grad_norm": 2.670806407928467, + "learning_rate": 2.9109589041095893e-06, + "loss": 1.9127, + "step": 185 + }, + { + "epoch": 0.19106317411402157, + "grad_norm": 3.0435004234313965, + "learning_rate": 2.9280821917808223e-06, + "loss": 1.8996, + "step": 186 + }, + { + "epoch": 0.192090395480226, + "grad_norm": 3.2139949798583984, + "learning_rate": 2.945205479452055e-06, + "loss": 1.9255, + "step": 187 + }, + { + "epoch": 0.19311761684643042, + "grad_norm": 3.252965211868286, + "learning_rate": 2.962328767123288e-06, + "loss": 1.9809, + "step": 188 + }, + { + "epoch": 0.1941448382126348, + "grad_norm": 2.9325191974639893, + "learning_rate": 2.979452054794521e-06, + "loss": 1.9691, + "step": 189 + }, + { + "epoch": 0.19517205957883924, + "grad_norm": 3.5879862308502197, + "learning_rate": 2.9965753424657533e-06, + "loss": 1.9205, + "step": 190 + }, + { + "epoch": 0.19517205957883924, + "eval_loss": 2.13791561126709, + "eval_runtime": 44.8545, + "eval_samples_per_second": 22.294, + "eval_steps_per_second": 0.936, + "step": 190 + }, + { + "epoch": 0.19619928094504366, + "grad_norm": 3.0025253295898438, + "learning_rate": 3.0136986301369864e-06, + "loss": 1.9717, + "step": 191 + }, + { + "epoch": 0.19722650231124808, + "grad_norm": 3.1785972118377686, + "learning_rate": 3.0308219178082194e-06, + "loss": 1.9535, + "step": 192 + }, + { + "epoch": 0.1982537236774525, + "grad_norm": 3.121065378189087, + "learning_rate": 3.0479452054794525e-06, + "loss": 1.9026, + "step": 193 + }, + { + "epoch": 0.1992809450436569, + "grad_norm": 2.838779926300049, + "learning_rate": 3.065068493150685e-06, + "loss": 1.941, + "step": 194 + }, + { + "epoch": 0.20030816640986132, + "grad_norm": 2.8916969299316406, + "learning_rate": 3.082191780821918e-06, + "loss": 1.8983, + "step": 195 + }, + { + "epoch": 0.20133538777606574, + "grad_norm": 2.9195168018341064, + "learning_rate": 3.099315068493151e-06, + "loss": 1.918, + "step": 196 + }, + { + "epoch": 0.20236260914227017, + "grad_norm": 3.235989809036255, + "learning_rate": 3.1164383561643843e-06, + "loss": 1.9111, + "step": 197 + }, + { + "epoch": 0.2033898305084746, + "grad_norm": 3.0316455364227295, + "learning_rate": 3.1335616438356165e-06, + "loss": 1.8839, + "step": 198 + }, + { + "epoch": 0.20441705187467898, + "grad_norm": 2.963610887527466, + "learning_rate": 3.1506849315068495e-06, + "loss": 2.0303, + "step": 199 + }, + { + "epoch": 0.2054442732408834, + "grad_norm": 2.944054365158081, + "learning_rate": 3.167808219178082e-06, + "loss": 1.8822, + "step": 200 + }, + { + "epoch": 0.2054442732408834, + "eval_loss": 2.1433050632476807, + "eval_runtime": 45.8205, + "eval_samples_per_second": 21.824, + "eval_steps_per_second": 0.917, + "step": 200 + }, + { + "epoch": 0.20647149460708783, + "grad_norm": 2.8648695945739746, + "learning_rate": 3.184931506849315e-06, + "loss": 1.8841, + "step": 201 + }, + { + "epoch": 0.20749871597329225, + "grad_norm": 3.2838797569274902, + "learning_rate": 3.2020547945205483e-06, + "loss": 1.8623, + "step": 202 + }, + { + "epoch": 0.20852593733949668, + "grad_norm": 2.7965173721313477, + "learning_rate": 3.2191780821917813e-06, + "loss": 1.8521, + "step": 203 + }, + { + "epoch": 0.20955315870570107, + "grad_norm": 3.3449349403381348, + "learning_rate": 3.236301369863014e-06, + "loss": 1.84, + "step": 204 + }, + { + "epoch": 0.2105803800719055, + "grad_norm": 3.4888951778411865, + "learning_rate": 3.253424657534247e-06, + "loss": 1.824, + "step": 205 + }, + { + "epoch": 0.21160760143810992, + "grad_norm": 2.991349220275879, + "learning_rate": 3.2705479452054796e-06, + "loss": 1.9101, + "step": 206 + }, + { + "epoch": 0.21263482280431434, + "grad_norm": 3.313509702682495, + "learning_rate": 3.2876712328767123e-06, + "loss": 1.8912, + "step": 207 + }, + { + "epoch": 0.21366204417051873, + "grad_norm": 2.8260746002197266, + "learning_rate": 3.3047945205479453e-06, + "loss": 1.8555, + "step": 208 + }, + { + "epoch": 0.21468926553672316, + "grad_norm": 2.7131688594818115, + "learning_rate": 3.3219178082191784e-06, + "loss": 1.8693, + "step": 209 + }, + { + "epoch": 0.21571648690292758, + "grad_norm": 2.5836338996887207, + "learning_rate": 3.3390410958904114e-06, + "loss": 1.8965, + "step": 210 + }, + { + "epoch": 0.21571648690292758, + "eval_loss": 2.1232542991638184, + "eval_runtime": 42.3352, + "eval_samples_per_second": 23.621, + "eval_steps_per_second": 0.992, + "step": 210 + }, + { + "epoch": 0.216743708269132, + "grad_norm": 2.8341355323791504, + "learning_rate": 3.356164383561644e-06, + "loss": 1.9131, + "step": 211 + }, + { + "epoch": 0.21777092963533642, + "grad_norm": 2.490853786468506, + "learning_rate": 3.373287671232877e-06, + "loss": 1.9626, + "step": 212 + }, + { + "epoch": 0.21879815100154082, + "grad_norm": 2.597956895828247, + "learning_rate": 3.39041095890411e-06, + "loss": 1.9203, + "step": 213 + }, + { + "epoch": 0.21982537236774524, + "grad_norm": 3.1104283332824707, + "learning_rate": 3.4075342465753424e-06, + "loss": 1.9442, + "step": 214 + }, + { + "epoch": 0.22085259373394966, + "grad_norm": 2.5698931217193604, + "learning_rate": 3.4246575342465754e-06, + "loss": 1.8385, + "step": 215 + }, + { + "epoch": 0.2218798151001541, + "grad_norm": 3.1637446880340576, + "learning_rate": 3.4417808219178085e-06, + "loss": 1.9005, + "step": 216 + }, + { + "epoch": 0.2229070364663585, + "grad_norm": 2.764070749282837, + "learning_rate": 3.458904109589041e-06, + "loss": 1.8886, + "step": 217 + }, + { + "epoch": 0.2239342578325629, + "grad_norm": 3.303781032562256, + "learning_rate": 3.476027397260274e-06, + "loss": 1.9085, + "step": 218 + }, + { + "epoch": 0.22496147919876733, + "grad_norm": 2.9543044567108154, + "learning_rate": 3.4931506849315072e-06, + "loss": 1.9014, + "step": 219 + }, + { + "epoch": 0.22598870056497175, + "grad_norm": 2.7277028560638428, + "learning_rate": 3.5102739726027403e-06, + "loss": 1.9017, + "step": 220 + }, + { + "epoch": 0.22598870056497175, + "eval_loss": 2.1024081707000732, + "eval_runtime": 45.4966, + "eval_samples_per_second": 21.98, + "eval_steps_per_second": 0.923, + "step": 220 + }, + { + "epoch": 0.22701592193117617, + "grad_norm": 2.864140033721924, + "learning_rate": 3.527397260273973e-06, + "loss": 1.8935, + "step": 221 + }, + { + "epoch": 0.2280431432973806, + "grad_norm": 3.4133172035217285, + "learning_rate": 3.5445205479452056e-06, + "loss": 1.874, + "step": 222 + }, + { + "epoch": 0.229070364663585, + "grad_norm": 3.33343243598938, + "learning_rate": 3.5616438356164386e-06, + "loss": 1.8789, + "step": 223 + }, + { + "epoch": 0.2300975860297894, + "grad_norm": 3.5947320461273193, + "learning_rate": 3.5787671232876712e-06, + "loss": 1.9529, + "step": 224 + }, + { + "epoch": 0.23112480739599384, + "grad_norm": 3.2194948196411133, + "learning_rate": 3.5958904109589043e-06, + "loss": 1.9205, + "step": 225 + }, + { + "epoch": 0.23215202876219826, + "grad_norm": 3.3137149810791016, + "learning_rate": 3.6130136986301373e-06, + "loss": 1.8584, + "step": 226 + }, + { + "epoch": 0.23317925012840268, + "grad_norm": 3.3358561992645264, + "learning_rate": 3.6301369863013704e-06, + "loss": 1.8817, + "step": 227 + }, + { + "epoch": 0.23420647149460708, + "grad_norm": 3.8612711429595947, + "learning_rate": 3.647260273972603e-06, + "loss": 1.9189, + "step": 228 + }, + { + "epoch": 0.2352336928608115, + "grad_norm": 3.0039336681365967, + "learning_rate": 3.664383561643836e-06, + "loss": 1.92, + "step": 229 + }, + { + "epoch": 0.23626091422701592, + "grad_norm": 3.8983118534088135, + "learning_rate": 3.6815068493150687e-06, + "loss": 1.8512, + "step": 230 + }, + { + "epoch": 0.23626091422701592, + "eval_loss": 2.1237711906433105, + "eval_runtime": 45.6126, + "eval_samples_per_second": 21.924, + "eval_steps_per_second": 0.921, + "step": 230 + }, + { + "epoch": 0.23728813559322035, + "grad_norm": 3.061300039291382, + "learning_rate": 3.6986301369863014e-06, + "loss": 1.8772, + "step": 231 + }, + { + "epoch": 0.23831535695942477, + "grad_norm": 3.3102450370788574, + "learning_rate": 3.7157534246575344e-06, + "loss": 1.8395, + "step": 232 + }, + { + "epoch": 0.23934257832562916, + "grad_norm": 2.9634461402893066, + "learning_rate": 3.7328767123287675e-06, + "loss": 1.9447, + "step": 233 + }, + { + "epoch": 0.24036979969183359, + "grad_norm": 3.0937695503234863, + "learning_rate": 3.7500000000000005e-06, + "loss": 1.8628, + "step": 234 + }, + { + "epoch": 0.241397021058038, + "grad_norm": 2.8129842281341553, + "learning_rate": 3.767123287671233e-06, + "loss": 1.8891, + "step": 235 + }, + { + "epoch": 0.24242424242424243, + "grad_norm": 2.9616780281066895, + "learning_rate": 3.784246575342466e-06, + "loss": 1.9112, + "step": 236 + }, + { + "epoch": 0.24345146379044685, + "grad_norm": 2.837721109390259, + "learning_rate": 3.8013698630136993e-06, + "loss": 1.8514, + "step": 237 + }, + { + "epoch": 0.24447868515665125, + "grad_norm": 3.137784004211426, + "learning_rate": 3.818493150684932e-06, + "loss": 1.8764, + "step": 238 + }, + { + "epoch": 0.24550590652285567, + "grad_norm": 2.9692037105560303, + "learning_rate": 3.8356164383561645e-06, + "loss": 1.8201, + "step": 239 + }, + { + "epoch": 0.2465331278890601, + "grad_norm": 2.6718621253967285, + "learning_rate": 3.852739726027397e-06, + "loss": 1.9309, + "step": 240 + }, + { + "epoch": 0.2465331278890601, + "eval_loss": 2.101331949234009, + "eval_runtime": 45.6346, + "eval_samples_per_second": 21.913, + "eval_steps_per_second": 0.92, + "step": 240 + }, + { + "epoch": 0.24756034925526452, + "grad_norm": 3.411193609237671, + "learning_rate": 3.869863013698631e-06, + "loss": 1.9171, + "step": 241 + }, + { + "epoch": 0.24858757062146894, + "grad_norm": 2.8848962783813477, + "learning_rate": 3.886986301369863e-06, + "loss": 1.8587, + "step": 242 + }, + { + "epoch": 0.24961479198767333, + "grad_norm": 2.756298065185547, + "learning_rate": 3.904109589041096e-06, + "loss": 1.8614, + "step": 243 + }, + { + "epoch": 0.25064201335387776, + "grad_norm": 2.8048956394195557, + "learning_rate": 3.921232876712329e-06, + "loss": 1.8225, + "step": 244 + }, + { + "epoch": 0.2516692347200822, + "grad_norm": 2.9872403144836426, + "learning_rate": 3.938356164383562e-06, + "loss": 1.8367, + "step": 245 + }, + { + "epoch": 0.2526964560862866, + "grad_norm": 2.710156202316284, + "learning_rate": 3.9554794520547955e-06, + "loss": 1.809, + "step": 246 + }, + { + "epoch": 0.253723677452491, + "grad_norm": 3.2177000045776367, + "learning_rate": 3.972602739726027e-06, + "loss": 1.8883, + "step": 247 + }, + { + "epoch": 0.25475089881869545, + "grad_norm": 3.160964250564575, + "learning_rate": 3.989726027397261e-06, + "loss": 1.8267, + "step": 248 + }, + { + "epoch": 0.25577812018489987, + "grad_norm": 2.853498935699463, + "learning_rate": 4.006849315068493e-06, + "loss": 1.8852, + "step": 249 + }, + { + "epoch": 0.25680534155110424, + "grad_norm": 2.571742296218872, + "learning_rate": 4.023972602739726e-06, + "loss": 1.8833, + "step": 250 + }, + { + "epoch": 0.25680534155110424, + "eval_loss": 2.0988597869873047, + "eval_runtime": 47.7045, + "eval_samples_per_second": 20.962, + "eval_steps_per_second": 0.88, + "step": 250 + }, + { + "epoch": 0.25783256291730866, + "grad_norm": 2.776742458343506, + "learning_rate": 4.0410958904109595e-06, + "loss": 1.8662, + "step": 251 + }, + { + "epoch": 0.2588597842835131, + "grad_norm": 2.8038573265075684, + "learning_rate": 4.058219178082192e-06, + "loss": 1.8528, + "step": 252 + }, + { + "epoch": 0.2598870056497175, + "grad_norm": 2.557623863220215, + "learning_rate": 4.075342465753426e-06, + "loss": 1.8765, + "step": 253 + }, + { + "epoch": 0.26091422701592193, + "grad_norm": 3.0477073192596436, + "learning_rate": 4.092465753424658e-06, + "loss": 1.8434, + "step": 254 + }, + { + "epoch": 0.26194144838212635, + "grad_norm": 2.6221649646759033, + "learning_rate": 4.109589041095891e-06, + "loss": 1.8348, + "step": 255 + }, + { + "epoch": 0.2629686697483308, + "grad_norm": 3.0849292278289795, + "learning_rate": 4.1267123287671235e-06, + "loss": 1.9057, + "step": 256 + }, + { + "epoch": 0.2639958911145352, + "grad_norm": 2.795184373855591, + "learning_rate": 4.143835616438356e-06, + "loss": 1.8528, + "step": 257 + }, + { + "epoch": 0.2650231124807396, + "grad_norm": 2.887193441390991, + "learning_rate": 4.16095890410959e-06, + "loss": 1.9176, + "step": 258 + }, + { + "epoch": 0.26605033384694404, + "grad_norm": 2.776801347732544, + "learning_rate": 4.178082191780822e-06, + "loss": 1.8911, + "step": 259 + }, + { + "epoch": 0.2670775552131484, + "grad_norm": 2.980426788330078, + "learning_rate": 4.195205479452055e-06, + "loss": 1.8288, + "step": 260 + }, + { + "epoch": 0.2670775552131484, + "eval_loss": 2.0636370182037354, + "eval_runtime": 49.3081, + "eval_samples_per_second": 20.281, + "eval_steps_per_second": 0.852, + "step": 260 + }, + { + "epoch": 0.26810477657935283, + "grad_norm": 2.7921340465545654, + "learning_rate": 4.212328767123288e-06, + "loss": 1.8405, + "step": 261 + }, + { + "epoch": 0.26913199794555726, + "grad_norm": 2.67714262008667, + "learning_rate": 4.229452054794521e-06, + "loss": 1.8342, + "step": 262 + }, + { + "epoch": 0.2701592193117617, + "grad_norm": 2.819868326187134, + "learning_rate": 4.246575342465754e-06, + "loss": 1.8466, + "step": 263 + }, + { + "epoch": 0.2711864406779661, + "grad_norm": 2.611250638961792, + "learning_rate": 4.263698630136986e-06, + "loss": 1.8314, + "step": 264 + }, + { + "epoch": 0.2722136620441705, + "grad_norm": 3.0406076908111572, + "learning_rate": 4.28082191780822e-06, + "loss": 1.7924, + "step": 265 + }, + { + "epoch": 0.27324088341037495, + "grad_norm": 3.212790012359619, + "learning_rate": 4.297945205479452e-06, + "loss": 1.9479, + "step": 266 + }, + { + "epoch": 0.27426810477657937, + "grad_norm": 2.4767045974731445, + "learning_rate": 4.315068493150685e-06, + "loss": 1.835, + "step": 267 + }, + { + "epoch": 0.2752953261427838, + "grad_norm": 3.191835641860962, + "learning_rate": 4.3321917808219185e-06, + "loss": 1.8261, + "step": 268 + }, + { + "epoch": 0.2763225475089882, + "grad_norm": 2.9970273971557617, + "learning_rate": 4.349315068493151e-06, + "loss": 1.8427, + "step": 269 + }, + { + "epoch": 0.2773497688751926, + "grad_norm": 2.888843297958374, + "learning_rate": 4.3664383561643846e-06, + "loss": 1.819, + "step": 270 + }, + { + "epoch": 0.2773497688751926, + "eval_loss": 2.054966926574707, + "eval_runtime": 45.9212, + "eval_samples_per_second": 21.776, + "eval_steps_per_second": 0.915, + "step": 270 + }, + { + "epoch": 0.278376990241397, + "grad_norm": 2.7771849632263184, + "learning_rate": 4.383561643835616e-06, + "loss": 1.8178, + "step": 271 + }, + { + "epoch": 0.2794042116076014, + "grad_norm": 2.9749014377593994, + "learning_rate": 4.40068493150685e-06, + "loss": 1.8982, + "step": 272 + }, + { + "epoch": 0.28043143297380585, + "grad_norm": 3.076021432876587, + "learning_rate": 4.4178082191780825e-06, + "loss": 1.8575, + "step": 273 + }, + { + "epoch": 0.2814586543400103, + "grad_norm": 3.2113568782806396, + "learning_rate": 4.434931506849315e-06, + "loss": 1.8241, + "step": 274 + }, + { + "epoch": 0.2824858757062147, + "grad_norm": 3.322063684463501, + "learning_rate": 4.4520547945205486e-06, + "loss": 1.7922, + "step": 275 + }, + { + "epoch": 0.2835130970724191, + "grad_norm": 3.2778918743133545, + "learning_rate": 4.469178082191781e-06, + "loss": 1.8412, + "step": 276 + }, + { + "epoch": 0.28454031843862354, + "grad_norm": 3.1802618503570557, + "learning_rate": 4.486301369863014e-06, + "loss": 1.8283, + "step": 277 + }, + { + "epoch": 0.28556753980482796, + "grad_norm": 2.9214258193969727, + "learning_rate": 4.503424657534247e-06, + "loss": 1.8817, + "step": 278 + }, + { + "epoch": 0.28659476117103233, + "grad_norm": 2.8809854984283447, + "learning_rate": 4.52054794520548e-06, + "loss": 1.7662, + "step": 279 + }, + { + "epoch": 0.28762198253723675, + "grad_norm": 2.925537586212158, + "learning_rate": 4.537671232876713e-06, + "loss": 1.8147, + "step": 280 + }, + { + "epoch": 0.28762198253723675, + "eval_loss": 2.054513454437256, + "eval_runtime": 51.5036, + "eval_samples_per_second": 19.416, + "eval_steps_per_second": 0.815, + "step": 280 + }, + { + "epoch": 0.2886492039034412, + "grad_norm": 2.9421887397766113, + "learning_rate": 4.554794520547945e-06, + "loss": 1.8581, + "step": 281 + }, + { + "epoch": 0.2896764252696456, + "grad_norm": 2.9733567237854004, + "learning_rate": 4.571917808219179e-06, + "loss": 1.8703, + "step": 282 + }, + { + "epoch": 0.29070364663585, + "grad_norm": 2.5352156162261963, + "learning_rate": 4.589041095890411e-06, + "loss": 1.8505, + "step": 283 + }, + { + "epoch": 0.29173086800205444, + "grad_norm": 2.773847818374634, + "learning_rate": 4.606164383561644e-06, + "loss": 1.8114, + "step": 284 + }, + { + "epoch": 0.29275808936825887, + "grad_norm": 2.709632635116577, + "learning_rate": 4.6232876712328774e-06, + "loss": 1.815, + "step": 285 + }, + { + "epoch": 0.2937853107344633, + "grad_norm": 3.1234357357025146, + "learning_rate": 4.64041095890411e-06, + "loss": 1.8391, + "step": 286 + }, + { + "epoch": 0.2948125321006677, + "grad_norm": 2.9376144409179688, + "learning_rate": 4.657534246575343e-06, + "loss": 1.7998, + "step": 287 + }, + { + "epoch": 0.29583975346687214, + "grad_norm": 3.0812032222747803, + "learning_rate": 4.674657534246575e-06, + "loss": 1.8462, + "step": 288 + }, + { + "epoch": 0.2968669748330765, + "grad_norm": 2.922722578048706, + "learning_rate": 4.691780821917809e-06, + "loss": 1.8263, + "step": 289 + }, + { + "epoch": 0.2978941961992809, + "grad_norm": 2.9484992027282715, + "learning_rate": 4.7089041095890414e-06, + "loss": 1.8754, + "step": 290 + }, + { + "epoch": 0.2978941961992809, + "eval_loss": 2.032902240753174, + "eval_runtime": 47.8711, + "eval_samples_per_second": 20.889, + "eval_steps_per_second": 0.877, + "step": 290 + }, + { + "epoch": 0.29892141756548535, + "grad_norm": 3.0068976879119873, + "learning_rate": 4.726027397260274e-06, + "loss": 1.8256, + "step": 291 + }, + { + "epoch": 0.29994863893168977, + "grad_norm": 2.6469006538391113, + "learning_rate": 4.7431506849315075e-06, + "loss": 1.8354, + "step": 292 + }, + { + "epoch": 0.3009758602978942, + "grad_norm": 3.06986927986145, + "learning_rate": 4.76027397260274e-06, + "loss": 1.8402, + "step": 293 + }, + { + "epoch": 0.3020030816640986, + "grad_norm": 2.8409981727600098, + "learning_rate": 4.777397260273973e-06, + "loss": 1.7915, + "step": 294 + }, + { + "epoch": 0.30303030303030304, + "grad_norm": 3.1027376651763916, + "learning_rate": 4.7945205479452054e-06, + "loss": 1.8153, + "step": 295 + }, + { + "epoch": 0.30405752439650746, + "grad_norm": 2.7221434116363525, + "learning_rate": 4.811643835616439e-06, + "loss": 1.8098, + "step": 296 + }, + { + "epoch": 0.3050847457627119, + "grad_norm": 3.0559370517730713, + "learning_rate": 4.8287671232876716e-06, + "loss": 1.7504, + "step": 297 + }, + { + "epoch": 0.3061119671289163, + "grad_norm": 2.9628236293792725, + "learning_rate": 4.845890410958904e-06, + "loss": 1.8019, + "step": 298 + }, + { + "epoch": 0.3071391884951207, + "grad_norm": 2.6248221397399902, + "learning_rate": 4.863013698630138e-06, + "loss": 1.8001, + "step": 299 + }, + { + "epoch": 0.3081664098613251, + "grad_norm": 3.0477702617645264, + "learning_rate": 4.88013698630137e-06, + "loss": 1.7915, + "step": 300 + }, + { + "epoch": 0.3081664098613251, + "eval_loss": 2.0522429943084717, + "eval_runtime": 46.796, + "eval_samples_per_second": 21.369, + "eval_steps_per_second": 0.898, + "step": 300 + }, + { + "epoch": 0.3091936312275295, + "grad_norm": 2.784707546234131, + "learning_rate": 4.897260273972603e-06, + "loss": 1.7757, + "step": 301 + }, + { + "epoch": 0.31022085259373394, + "grad_norm": 2.8283841609954834, + "learning_rate": 4.914383561643836e-06, + "loss": 1.8975, + "step": 302 + }, + { + "epoch": 0.31124807395993837, + "grad_norm": 2.879342555999756, + "learning_rate": 4.931506849315069e-06, + "loss": 1.8208, + "step": 303 + }, + { + "epoch": 0.3122752953261428, + "grad_norm": 3.2111175060272217, + "learning_rate": 4.948630136986302e-06, + "loss": 1.8631, + "step": 304 + }, + { + "epoch": 0.3133025166923472, + "grad_norm": 2.876168966293335, + "learning_rate": 4.965753424657534e-06, + "loss": 1.8863, + "step": 305 + }, + { + "epoch": 0.31432973805855163, + "grad_norm": 2.8958466053009033, + "learning_rate": 4.982876712328768e-06, + "loss": 1.7565, + "step": 306 + }, + { + "epoch": 0.31535695942475606, + "grad_norm": 2.6702723503112793, + "learning_rate": 5e-06, + "loss": 1.7844, + "step": 307 + }, + { + "epoch": 0.3163841807909605, + "grad_norm": 2.557966470718384, + "learning_rate": 4.998096688237534e-06, + "loss": 1.7472, + "step": 308 + }, + { + "epoch": 0.31741140215716485, + "grad_norm": 2.9257633686065674, + "learning_rate": 4.996193376475067e-06, + "loss": 1.853, + "step": 309 + }, + { + "epoch": 0.31843862352336927, + "grad_norm": 2.497093439102173, + "learning_rate": 4.9942900647126005e-06, + "loss": 1.7313, + "step": 310 + }, + { + "epoch": 0.31843862352336927, + "eval_loss": 2.0336520671844482, + "eval_runtime": 43.3773, + "eval_samples_per_second": 23.054, + "eval_steps_per_second": 0.968, + "step": 310 + }, + { + "epoch": 0.3194658448895737, + "grad_norm": 3.0313165187835693, + "learning_rate": 4.992386752950134e-06, + "loss": 1.7397, + "step": 311 + }, + { + "epoch": 0.3204930662557781, + "grad_norm": 2.629685163497925, + "learning_rate": 4.990483441187667e-06, + "loss": 1.7839, + "step": 312 + }, + { + "epoch": 0.32152028762198254, + "grad_norm": 2.6772103309631348, + "learning_rate": 4.9885801294252e-06, + "loss": 1.8038, + "step": 313 + }, + { + "epoch": 0.32254750898818696, + "grad_norm": 2.893958568572998, + "learning_rate": 4.986676817662733e-06, + "loss": 1.8072, + "step": 314 + }, + { + "epoch": 0.3235747303543914, + "grad_norm": 2.5317318439483643, + "learning_rate": 4.9847735059002665e-06, + "loss": 1.7632, + "step": 315 + }, + { + "epoch": 0.3246019517205958, + "grad_norm": 2.836867332458496, + "learning_rate": 4.9828701941378e-06, + "loss": 1.8499, + "step": 316 + }, + { + "epoch": 0.3256291730868002, + "grad_norm": 2.5564308166503906, + "learning_rate": 4.980966882375333e-06, + "loss": 1.7849, + "step": 317 + }, + { + "epoch": 0.3266563944530046, + "grad_norm": 2.4335806369781494, + "learning_rate": 4.979063570612867e-06, + "loss": 1.7425, + "step": 318 + }, + { + "epoch": 0.327683615819209, + "grad_norm": 2.7140371799468994, + "learning_rate": 4.9771602588504e-06, + "loss": 1.7173, + "step": 319 + }, + { + "epoch": 0.32871083718541344, + "grad_norm": 2.574345111846924, + "learning_rate": 4.975256947087933e-06, + "loss": 1.8307, + "step": 320 + }, + { + "epoch": 0.32871083718541344, + "eval_loss": 2.026960611343384, + "eval_runtime": 53.8982, + "eval_samples_per_second": 18.553, + "eval_steps_per_second": 0.779, + "step": 320 + }, + { + "epoch": 0.32973805855161786, + "grad_norm": 2.6877281665802, + "learning_rate": 4.973353635325467e-06, + "loss": 1.7216, + "step": 321 + }, + { + "epoch": 0.3307652799178223, + "grad_norm": 2.8152849674224854, + "learning_rate": 4.971450323563e-06, + "loss": 1.7831, + "step": 322 + }, + { + "epoch": 0.3317925012840267, + "grad_norm": 2.4208791255950928, + "learning_rate": 4.969547011800533e-06, + "loss": 1.7266, + "step": 323 + }, + { + "epoch": 0.33281972265023113, + "grad_norm": 2.765162229537964, + "learning_rate": 4.967643700038067e-06, + "loss": 1.7971, + "step": 324 + }, + { + "epoch": 0.33384694401643555, + "grad_norm": 2.5018413066864014, + "learning_rate": 4.9657403882756e-06, + "loss": 1.7842, + "step": 325 + }, + { + "epoch": 0.33487416538264, + "grad_norm": 2.901268482208252, + "learning_rate": 4.9638370765131335e-06, + "loss": 1.7522, + "step": 326 + }, + { + "epoch": 0.3359013867488444, + "grad_norm": 2.40067195892334, + "learning_rate": 4.961933764750667e-06, + "loss": 1.7371, + "step": 327 + }, + { + "epoch": 0.33692860811504877, + "grad_norm": 2.6963725090026855, + "learning_rate": 4.9600304529882e-06, + "loss": 1.8036, + "step": 328 + }, + { + "epoch": 0.3379558294812532, + "grad_norm": 2.483790397644043, + "learning_rate": 4.9581271412257335e-06, + "loss": 1.815, + "step": 329 + }, + { + "epoch": 0.3389830508474576, + "grad_norm": 2.7459559440612793, + "learning_rate": 4.956223829463266e-06, + "loss": 1.777, + "step": 330 + }, + { + "epoch": 0.3389830508474576, + "eval_loss": 2.046339988708496, + "eval_runtime": 53.909, + "eval_samples_per_second": 18.55, + "eval_steps_per_second": 0.779, + "step": 330 + }, + { + "epoch": 0.34001027221366203, + "grad_norm": 2.5724353790283203, + "learning_rate": 4.954320517700799e-06, + "loss": 1.7022, + "step": 331 + }, + { + "epoch": 0.34103749357986646, + "grad_norm": 2.510002613067627, + "learning_rate": 4.952417205938333e-06, + "loss": 1.802, + "step": 332 + }, + { + "epoch": 0.3420647149460709, + "grad_norm": 2.8722083568573, + "learning_rate": 4.950513894175866e-06, + "loss": 1.7772, + "step": 333 + }, + { + "epoch": 0.3430919363122753, + "grad_norm": 2.586747407913208, + "learning_rate": 4.9486105824133995e-06, + "loss": 1.7664, + "step": 334 + }, + { + "epoch": 0.3441191576784797, + "grad_norm": 2.8008205890655518, + "learning_rate": 4.946707270650933e-06, + "loss": 1.7892, + "step": 335 + }, + { + "epoch": 0.34514637904468415, + "grad_norm": 2.632664203643799, + "learning_rate": 4.944803958888466e-06, + "loss": 1.7967, + "step": 336 + }, + { + "epoch": 0.34617360041088857, + "grad_norm": 2.8898301124572754, + "learning_rate": 4.942900647126e-06, + "loss": 1.8126, + "step": 337 + }, + { + "epoch": 0.34720082177709294, + "grad_norm": 2.719019651412964, + "learning_rate": 4.940997335363533e-06, + "loss": 1.7544, + "step": 338 + }, + { + "epoch": 0.34822804314329736, + "grad_norm": 2.9995970726013184, + "learning_rate": 4.939094023601066e-06, + "loss": 1.7142, + "step": 339 + }, + { + "epoch": 0.3492552645095018, + "grad_norm": 2.9737157821655273, + "learning_rate": 4.9371907118386e-06, + "loss": 1.7659, + "step": 340 + }, + { + "epoch": 0.3492552645095018, + "eval_loss": 2.024946928024292, + "eval_runtime": 52.8069, + "eval_samples_per_second": 18.937, + "eval_steps_per_second": 0.795, + "step": 340 + }, + { + "epoch": 0.3502824858757062, + "grad_norm": 2.8651719093322754, + "learning_rate": 4.935287400076133e-06, + "loss": 1.7767, + "step": 341 + }, + { + "epoch": 0.35130970724191063, + "grad_norm": 2.6293981075286865, + "learning_rate": 4.933384088313666e-06, + "loss": 1.7815, + "step": 342 + }, + { + "epoch": 0.35233692860811505, + "grad_norm": 3.230978012084961, + "learning_rate": 4.9314807765512e-06, + "loss": 1.7423, + "step": 343 + }, + { + "epoch": 0.3533641499743195, + "grad_norm": 3.228300094604492, + "learning_rate": 4.929577464788733e-06, + "loss": 1.6845, + "step": 344 + }, + { + "epoch": 0.3543913713405239, + "grad_norm": 3.036733388900757, + "learning_rate": 4.9276741530262665e-06, + "loss": 1.7524, + "step": 345 + }, + { + "epoch": 0.3554185927067283, + "grad_norm": 2.778538227081299, + "learning_rate": 4.9257708412638e-06, + "loss": 1.7433, + "step": 346 + }, + { + "epoch": 0.35644581407293274, + "grad_norm": 3.0958566665649414, + "learning_rate": 4.923867529501332e-06, + "loss": 1.76, + "step": 347 + }, + { + "epoch": 0.3574730354391371, + "grad_norm": 3.1120026111602783, + "learning_rate": 4.921964217738866e-06, + "loss": 1.7399, + "step": 348 + }, + { + "epoch": 0.35850025680534153, + "grad_norm": 2.70768141746521, + "learning_rate": 4.920060905976399e-06, + "loss": 1.8169, + "step": 349 + }, + { + "epoch": 0.35952747817154596, + "grad_norm": 2.917478561401367, + "learning_rate": 4.9181575942139325e-06, + "loss": 1.7948, + "step": 350 + }, + { + "epoch": 0.35952747817154596, + "eval_loss": 2.0208380222320557, + "eval_runtime": 47.4243, + "eval_samples_per_second": 21.086, + "eval_steps_per_second": 0.886, + "step": 350 + }, + { + "epoch": 0.3605546995377504, + "grad_norm": 3.0564756393432617, + "learning_rate": 4.916254282451466e-06, + "loss": 1.7861, + "step": 351 + }, + { + "epoch": 0.3615819209039548, + "grad_norm": 2.709195137023926, + "learning_rate": 4.914350970688999e-06, + "loss": 1.8323, + "step": 352 + }, + { + "epoch": 0.3626091422701592, + "grad_norm": 2.8797669410705566, + "learning_rate": 4.9124476589265325e-06, + "loss": 1.722, + "step": 353 + }, + { + "epoch": 0.36363636363636365, + "grad_norm": 2.560702085494995, + "learning_rate": 4.910544347164066e-06, + "loss": 1.7289, + "step": 354 + }, + { + "epoch": 0.36466358500256807, + "grad_norm": 3.3781847953796387, + "learning_rate": 4.908641035401599e-06, + "loss": 1.8096, + "step": 355 + }, + { + "epoch": 0.3656908063687725, + "grad_norm": 2.582310676574707, + "learning_rate": 4.906737723639133e-06, + "loss": 1.8752, + "step": 356 + }, + { + "epoch": 0.3667180277349769, + "grad_norm": 3.506385564804077, + "learning_rate": 4.904834411876665e-06, + "loss": 1.7404, + "step": 357 + }, + { + "epoch": 0.3677452491011813, + "grad_norm": 2.4360411167144775, + "learning_rate": 4.902931100114199e-06, + "loss": 1.7812, + "step": 358 + }, + { + "epoch": 0.3687724704673857, + "grad_norm": 3.77639102935791, + "learning_rate": 4.901027788351733e-06, + "loss": 1.7103, + "step": 359 + }, + { + "epoch": 0.3697996918335901, + "grad_norm": 2.887723922729492, + "learning_rate": 4.899124476589266e-06, + "loss": 1.7485, + "step": 360 + }, + { + "epoch": 0.3697996918335901, + "eval_loss": 2.0298168659210205, + "eval_runtime": 46.0993, + "eval_samples_per_second": 21.692, + "eval_steps_per_second": 0.911, + "step": 360 + }, + { + "epoch": 0.37082691319979455, + "grad_norm": 3.5861997604370117, + "learning_rate": 4.8972211648267994e-06, + "loss": 1.7189, + "step": 361 + }, + { + "epoch": 0.371854134565999, + "grad_norm": 2.590808868408203, + "learning_rate": 4.895317853064333e-06, + "loss": 1.7683, + "step": 362 + }, + { + "epoch": 0.3728813559322034, + "grad_norm": 3.4451487064361572, + "learning_rate": 4.893414541301866e-06, + "loss": 1.7436, + "step": 363 + }, + { + "epoch": 0.3739085772984078, + "grad_norm": 2.406073808670044, + "learning_rate": 4.891511229539399e-06, + "loss": 1.7378, + "step": 364 + }, + { + "epoch": 0.37493579866461224, + "grad_norm": 2.7162997722625732, + "learning_rate": 4.889607917776932e-06, + "loss": 1.7558, + "step": 365 + }, + { + "epoch": 0.37596302003081666, + "grad_norm": 2.564753770828247, + "learning_rate": 4.887704606014465e-06, + "loss": 1.7559, + "step": 366 + }, + { + "epoch": 0.37699024139702103, + "grad_norm": 2.7196900844573975, + "learning_rate": 4.885801294251999e-06, + "loss": 1.7683, + "step": 367 + }, + { + "epoch": 0.37801746276322545, + "grad_norm": 2.9334428310394287, + "learning_rate": 4.883897982489532e-06, + "loss": 1.6932, + "step": 368 + }, + { + "epoch": 0.3790446841294299, + "grad_norm": 2.4899954795837402, + "learning_rate": 4.8819946707270655e-06, + "loss": 1.7943, + "step": 369 + }, + { + "epoch": 0.3800719054956343, + "grad_norm": 2.820594310760498, + "learning_rate": 4.880091358964599e-06, + "loss": 1.7729, + "step": 370 + }, + { + "epoch": 0.3800719054956343, + "eval_loss": 2.0204291343688965, + "eval_runtime": 44.9993, + "eval_samples_per_second": 22.223, + "eval_steps_per_second": 0.933, + "step": 370 + }, + { + "epoch": 0.3810991268618387, + "grad_norm": 2.4717929363250732, + "learning_rate": 4.878188047202132e-06, + "loss": 1.6813, + "step": 371 + }, + { + "epoch": 0.38212634822804314, + "grad_norm": 3.1815855503082275, + "learning_rate": 4.876284735439666e-06, + "loss": 1.8588, + "step": 372 + }, + { + "epoch": 0.38315356959424757, + "grad_norm": 2.7330422401428223, + "learning_rate": 4.874381423677199e-06, + "loss": 1.8245, + "step": 373 + }, + { + "epoch": 0.384180790960452, + "grad_norm": 3.129444122314453, + "learning_rate": 4.8724781119147315e-06, + "loss": 1.7494, + "step": 374 + }, + { + "epoch": 0.3852080123266564, + "grad_norm": 2.75551700592041, + "learning_rate": 4.870574800152265e-06, + "loss": 1.7761, + "step": 375 + }, + { + "epoch": 0.38623523369286084, + "grad_norm": 2.9313466548919678, + "learning_rate": 4.868671488389798e-06, + "loss": 1.7076, + "step": 376 + }, + { + "epoch": 0.3872624550590652, + "grad_norm": 2.7590556144714355, + "learning_rate": 4.8667681766273316e-06, + "loss": 1.7105, + "step": 377 + }, + { + "epoch": 0.3882896764252696, + "grad_norm": 2.8224680423736572, + "learning_rate": 4.864864864864866e-06, + "loss": 1.6853, + "step": 378 + }, + { + "epoch": 0.38931689779147405, + "grad_norm": 2.5148062705993652, + "learning_rate": 4.862961553102399e-06, + "loss": 1.7477, + "step": 379 + }, + { + "epoch": 0.39034411915767847, + "grad_norm": 3.144247055053711, + "learning_rate": 4.8610582413399325e-06, + "loss": 1.7726, + "step": 380 + }, + { + "epoch": 0.39034411915767847, + "eval_loss": 2.006993532180786, + "eval_runtime": 50.8909, + "eval_samples_per_second": 19.65, + "eval_steps_per_second": 0.825, + "step": 380 + }, + { + "epoch": 0.3913713405238829, + "grad_norm": 2.6818125247955322, + "learning_rate": 4.859154929577465e-06, + "loss": 1.7351, + "step": 381 + }, + { + "epoch": 0.3923985618900873, + "grad_norm": 2.8543903827667236, + "learning_rate": 4.857251617814998e-06, + "loss": 1.7499, + "step": 382 + }, + { + "epoch": 0.39342578325629174, + "grad_norm": 2.5483179092407227, + "learning_rate": 4.855348306052532e-06, + "loss": 1.7849, + "step": 383 + }, + { + "epoch": 0.39445300462249616, + "grad_norm": 2.7169013023376465, + "learning_rate": 4.853444994290065e-06, + "loss": 1.7664, + "step": 384 + }, + { + "epoch": 0.3954802259887006, + "grad_norm": 2.8304696083068848, + "learning_rate": 4.8515416825275985e-06, + "loss": 1.73, + "step": 385 + }, + { + "epoch": 0.396507447354905, + "grad_norm": 2.4217448234558105, + "learning_rate": 4.849638370765132e-06, + "loss": 1.7915, + "step": 386 + }, + { + "epoch": 0.3975346687211094, + "grad_norm": 2.523256301879883, + "learning_rate": 4.847735059002665e-06, + "loss": 1.7195, + "step": 387 + }, + { + "epoch": 0.3985618900873138, + "grad_norm": 2.813638925552368, + "learning_rate": 4.8458317472401985e-06, + "loss": 1.6766, + "step": 388 + }, + { + "epoch": 0.3995891114535182, + "grad_norm": 2.6701548099517822, + "learning_rate": 4.843928435477732e-06, + "loss": 1.7666, + "step": 389 + }, + { + "epoch": 0.40061633281972264, + "grad_norm": 2.5568342208862305, + "learning_rate": 4.842025123715265e-06, + "loss": 1.723, + "step": 390 + }, + { + "epoch": 0.40061633281972264, + "eval_loss": 2.0098180770874023, + "eval_runtime": 44.4503, + "eval_samples_per_second": 22.497, + "eval_steps_per_second": 0.945, + "step": 390 + }, + { + "epoch": 0.40164355418592707, + "grad_norm": 2.6287341117858887, + "learning_rate": 4.840121811952798e-06, + "loss": 1.759, + "step": 391 + }, + { + "epoch": 0.4026707755521315, + "grad_norm": 2.934105157852173, + "learning_rate": 4.838218500190331e-06, + "loss": 1.7501, + "step": 392 + }, + { + "epoch": 0.4036979969183359, + "grad_norm": 2.6563878059387207, + "learning_rate": 4.8363151884278645e-06, + "loss": 1.7624, + "step": 393 + }, + { + "epoch": 0.40472521828454033, + "grad_norm": 2.538278102874756, + "learning_rate": 4.834411876665398e-06, + "loss": 1.7457, + "step": 394 + }, + { + "epoch": 0.40575243965074476, + "grad_norm": 2.801630735397339, + "learning_rate": 4.832508564902931e-06, + "loss": 1.7549, + "step": 395 + }, + { + "epoch": 0.4067796610169492, + "grad_norm": 2.571937322616577, + "learning_rate": 4.830605253140465e-06, + "loss": 1.6954, + "step": 396 + }, + { + "epoch": 0.40780688238315355, + "grad_norm": 2.5652592182159424, + "learning_rate": 4.828701941377998e-06, + "loss": 1.7328, + "step": 397 + }, + { + "epoch": 0.40883410374935797, + "grad_norm": 2.6840121746063232, + "learning_rate": 4.826798629615531e-06, + "loss": 1.6934, + "step": 398 + }, + { + "epoch": 0.4098613251155624, + "grad_norm": 3.0103495121002197, + "learning_rate": 4.824895317853065e-06, + "loss": 1.7086, + "step": 399 + }, + { + "epoch": 0.4108885464817668, + "grad_norm": 2.7361223697662354, + "learning_rate": 4.822992006090598e-06, + "loss": 1.7301, + "step": 400 + }, + { + "epoch": 0.4108885464817668, + "eval_loss": 1.9975080490112305, + "eval_runtime": 44.5413, + "eval_samples_per_second": 22.451, + "eval_steps_per_second": 0.943, + "step": 400 + }, + { + "epoch": 0.41191576784797124, + "grad_norm": 2.6805901527404785, + "learning_rate": 4.821088694328131e-06, + "loss": 1.6896, + "step": 401 + }, + { + "epoch": 0.41294298921417566, + "grad_norm": 3.231706142425537, + "learning_rate": 4.819185382565665e-06, + "loss": 1.7134, + "step": 402 + }, + { + "epoch": 0.4139702105803801, + "grad_norm": 2.5561299324035645, + "learning_rate": 4.817282070803198e-06, + "loss": 1.764, + "step": 403 + }, + { + "epoch": 0.4149974319465845, + "grad_norm": 3.3125340938568115, + "learning_rate": 4.8153787590407315e-06, + "loss": 1.76, + "step": 404 + }, + { + "epoch": 0.41602465331278893, + "grad_norm": 2.750051498413086, + "learning_rate": 4.813475447278265e-06, + "loss": 1.6882, + "step": 405 + }, + { + "epoch": 0.41705187467899335, + "grad_norm": 4.006515979766846, + "learning_rate": 4.811572135515798e-06, + "loss": 1.7892, + "step": 406 + }, + { + "epoch": 0.4180790960451977, + "grad_norm": 2.5276904106140137, + "learning_rate": 4.809668823753332e-06, + "loss": 1.6995, + "step": 407 + }, + { + "epoch": 0.41910631741140214, + "grad_norm": 2.959444046020508, + "learning_rate": 4.807765511990864e-06, + "loss": 1.7058, + "step": 408 + }, + { + "epoch": 0.42013353877760656, + "grad_norm": 2.8472464084625244, + "learning_rate": 4.8058622002283975e-06, + "loss": 1.7258, + "step": 409 + }, + { + "epoch": 0.421160760143811, + "grad_norm": 3.0240273475646973, + "learning_rate": 4.803958888465931e-06, + "loss": 1.7789, + "step": 410 + }, + { + "epoch": 0.421160760143811, + "eval_loss": 1.9919871091842651, + "eval_runtime": 41.727, + "eval_samples_per_second": 23.965, + "eval_steps_per_second": 1.007, + "step": 410 + }, + { + "epoch": 0.4221879815100154, + "grad_norm": 2.6211893558502197, + "learning_rate": 4.802055576703464e-06, + "loss": 1.7315, + "step": 411 + }, + { + "epoch": 0.42321520287621983, + "grad_norm": 2.6072895526885986, + "learning_rate": 4.8001522649409976e-06, + "loss": 1.8176, + "step": 412 + }, + { + "epoch": 0.42424242424242425, + "grad_norm": 2.9197702407836914, + "learning_rate": 4.798248953178531e-06, + "loss": 1.6947, + "step": 413 + }, + { + "epoch": 0.4252696456086287, + "grad_norm": 2.4383556842803955, + "learning_rate": 4.796345641416064e-06, + "loss": 1.6728, + "step": 414 + }, + { + "epoch": 0.4262968669748331, + "grad_norm": 2.9951915740966797, + "learning_rate": 4.794442329653598e-06, + "loss": 1.7249, + "step": 415 + }, + { + "epoch": 0.42732408834103747, + "grad_norm": 2.6095519065856934, + "learning_rate": 4.792539017891131e-06, + "loss": 1.7735, + "step": 416 + }, + { + "epoch": 0.4283513097072419, + "grad_norm": 2.6429390907287598, + "learning_rate": 4.790635706128664e-06, + "loss": 1.704, + "step": 417 + }, + { + "epoch": 0.4293785310734463, + "grad_norm": 2.921812057495117, + "learning_rate": 4.788732394366197e-06, + "loss": 1.7355, + "step": 418 + }, + { + "epoch": 0.43040575243965074, + "grad_norm": 2.6050965785980225, + "learning_rate": 4.786829082603731e-06, + "loss": 1.6636, + "step": 419 + }, + { + "epoch": 0.43143297380585516, + "grad_norm": 2.672070264816284, + "learning_rate": 4.7849257708412645e-06, + "loss": 1.6627, + "step": 420 + }, + { + "epoch": 0.43143297380585516, + "eval_loss": 1.9990200996398926, + "eval_runtime": 45.0672, + "eval_samples_per_second": 22.189, + "eval_steps_per_second": 0.932, + "step": 420 + }, + { + "epoch": 0.4324601951720596, + "grad_norm": 2.6813488006591797, + "learning_rate": 4.783022459078798e-06, + "loss": 1.7453, + "step": 421 + }, + { + "epoch": 0.433487416538264, + "grad_norm": 2.6587321758270264, + "learning_rate": 4.781119147316331e-06, + "loss": 1.6795, + "step": 422 + }, + { + "epoch": 0.4345146379044684, + "grad_norm": 2.827758312225342, + "learning_rate": 4.7792158355538645e-06, + "loss": 1.7098, + "step": 423 + }, + { + "epoch": 0.43554185927067285, + "grad_norm": 2.8615901470184326, + "learning_rate": 4.777312523791398e-06, + "loss": 1.7171, + "step": 424 + }, + { + "epoch": 0.43656908063687727, + "grad_norm": 2.9509294033050537, + "learning_rate": 4.77540921202893e-06, + "loss": 1.7018, + "step": 425 + }, + { + "epoch": 0.43759630200308164, + "grad_norm": 2.7994418144226074, + "learning_rate": 4.773505900266464e-06, + "loss": 1.7328, + "step": 426 + }, + { + "epoch": 0.43862352336928606, + "grad_norm": 2.7476799488067627, + "learning_rate": 4.771602588503997e-06, + "loss": 1.7116, + "step": 427 + }, + { + "epoch": 0.4396507447354905, + "grad_norm": 2.7955048084259033, + "learning_rate": 4.7696992767415305e-06, + "loss": 1.7187, + "step": 428 + }, + { + "epoch": 0.4406779661016949, + "grad_norm": 2.4326071739196777, + "learning_rate": 4.767795964979064e-06, + "loss": 1.75, + "step": 429 + }, + { + "epoch": 0.44170518746789933, + "grad_norm": 2.7708163261413574, + "learning_rate": 4.765892653216597e-06, + "loss": 1.7562, + "step": 430 + }, + { + "epoch": 0.44170518746789933, + "eval_loss": 2.0262997150421143, + "eval_runtime": 46.1649, + "eval_samples_per_second": 21.661, + "eval_steps_per_second": 0.91, + "step": 430 + }, + { + "epoch": 0.44273240883410375, + "grad_norm": 2.872743606567383, + "learning_rate": 4.763989341454131e-06, + "loss": 1.7527, + "step": 431 + }, + { + "epoch": 0.4437596302003082, + "grad_norm": 2.6387035846710205, + "learning_rate": 4.762086029691664e-06, + "loss": 1.6742, + "step": 432 + }, + { + "epoch": 0.4447868515665126, + "grad_norm": 2.6522979736328125, + "learning_rate": 4.760182717929197e-06, + "loss": 1.6811, + "step": 433 + }, + { + "epoch": 0.445814072932717, + "grad_norm": 2.644155263900757, + "learning_rate": 4.758279406166731e-06, + "loss": 1.7083, + "step": 434 + }, + { + "epoch": 0.44684129429892144, + "grad_norm": 2.530224561691284, + "learning_rate": 4.756376094404263e-06, + "loss": 1.6318, + "step": 435 + }, + { + "epoch": 0.4478685156651258, + "grad_norm": 2.431049108505249, + "learning_rate": 4.7544727826417966e-06, + "loss": 1.7162, + "step": 436 + }, + { + "epoch": 0.44889573703133023, + "grad_norm": 2.8091282844543457, + "learning_rate": 4.75256947087933e-06, + "loss": 1.7717, + "step": 437 + }, + { + "epoch": 0.44992295839753466, + "grad_norm": 2.5520501136779785, + "learning_rate": 4.750666159116863e-06, + "loss": 1.7201, + "step": 438 + }, + { + "epoch": 0.4509501797637391, + "grad_norm": 2.596370220184326, + "learning_rate": 4.7487628473543975e-06, + "loss": 1.6882, + "step": 439 + }, + { + "epoch": 0.4519774011299435, + "grad_norm": 2.667034387588501, + "learning_rate": 4.746859535591931e-06, + "loss": 1.6499, + "step": 440 + }, + { + "epoch": 0.4519774011299435, + "eval_loss": 2.008521556854248, + "eval_runtime": 49.3626, + "eval_samples_per_second": 20.258, + "eval_steps_per_second": 0.851, + "step": 440 + }, + { + "epoch": 0.4530046224961479, + "grad_norm": 2.636491537094116, + "learning_rate": 4.744956223829464e-06, + "loss": 1.6987, + "step": 441 + }, + { + "epoch": 0.45403184386235235, + "grad_norm": 2.4653050899505615, + "learning_rate": 4.743052912066997e-06, + "loss": 1.6775, + "step": 442 + }, + { + "epoch": 0.45505906522855677, + "grad_norm": 2.3866467475891113, + "learning_rate": 4.74114960030453e-06, + "loss": 1.7234, + "step": 443 + }, + { + "epoch": 0.4560862865947612, + "grad_norm": 2.675262928009033, + "learning_rate": 4.7392462885420635e-06, + "loss": 1.7363, + "step": 444 + }, + { + "epoch": 0.4571135079609656, + "grad_norm": 2.6233296394348145, + "learning_rate": 4.737342976779597e-06, + "loss": 1.6236, + "step": 445 + }, + { + "epoch": 0.45814072932717, + "grad_norm": 2.5638058185577393, + "learning_rate": 4.73543966501713e-06, + "loss": 1.7097, + "step": 446 + }, + { + "epoch": 0.4591679506933744, + "grad_norm": 2.7206926345825195, + "learning_rate": 4.7335363532546636e-06, + "loss": 1.6909, + "step": 447 + }, + { + "epoch": 0.4601951720595788, + "grad_norm": 2.546241283416748, + "learning_rate": 4.731633041492197e-06, + "loss": 1.6951, + "step": 448 + }, + { + "epoch": 0.46122239342578325, + "grad_norm": 2.6549105644226074, + "learning_rate": 4.72972972972973e-06, + "loss": 1.7052, + "step": 449 + }, + { + "epoch": 0.4622496147919877, + "grad_norm": 2.6649065017700195, + "learning_rate": 4.727826417967264e-06, + "loss": 1.7406, + "step": 450 + }, + { + "epoch": 0.4622496147919877, + "eval_loss": 1.9886436462402344, + "eval_runtime": 47.48, + "eval_samples_per_second": 21.061, + "eval_steps_per_second": 0.885, + "step": 450 + }, + { + "epoch": 0.4632768361581921, + "grad_norm": 2.7138187885284424, + "learning_rate": 4.725923106204797e-06, + "loss": 1.7154, + "step": 451 + }, + { + "epoch": 0.4643040575243965, + "grad_norm": 2.6095476150512695, + "learning_rate": 4.72401979444233e-06, + "loss": 1.686, + "step": 452 + }, + { + "epoch": 0.46533127889060094, + "grad_norm": 2.866363048553467, + "learning_rate": 4.722116482679863e-06, + "loss": 1.721, + "step": 453 + }, + { + "epoch": 0.46635850025680536, + "grad_norm": 2.4947352409362793, + "learning_rate": 4.720213170917396e-06, + "loss": 1.708, + "step": 454 + }, + { + "epoch": 0.46738572162300973, + "grad_norm": 2.3888256549835205, + "learning_rate": 4.71830985915493e-06, + "loss": 1.7356, + "step": 455 + }, + { + "epoch": 0.46841294298921415, + "grad_norm": 2.7065343856811523, + "learning_rate": 4.716406547392463e-06, + "loss": 1.6943, + "step": 456 + }, + { + "epoch": 0.4694401643554186, + "grad_norm": 2.424285888671875, + "learning_rate": 4.714503235629996e-06, + "loss": 1.7385, + "step": 457 + }, + { + "epoch": 0.470467385721623, + "grad_norm": 2.4175925254821777, + "learning_rate": 4.71259992386753e-06, + "loss": 1.6654, + "step": 458 + }, + { + "epoch": 0.4714946070878274, + "grad_norm": 2.3203656673431396, + "learning_rate": 4.710696612105063e-06, + "loss": 1.6977, + "step": 459 + }, + { + "epoch": 0.47252182845403184, + "grad_norm": 2.6660172939300537, + "learning_rate": 4.708793300342596e-06, + "loss": 1.7055, + "step": 460 + }, + { + "epoch": 0.47252182845403184, + "eval_loss": 1.9864659309387207, + "eval_runtime": 47.6232, + "eval_samples_per_second": 20.998, + "eval_steps_per_second": 0.882, + "step": 460 + }, + { + "epoch": 0.47354904982023627, + "grad_norm": 2.515615701675415, + "learning_rate": 4.70688998858013e-06, + "loss": 1.689, + "step": 461 + }, + { + "epoch": 0.4745762711864407, + "grad_norm": 2.429264783859253, + "learning_rate": 4.704986676817663e-06, + "loss": 1.6796, + "step": 462 + }, + { + "epoch": 0.4756034925526451, + "grad_norm": 2.469543933868408, + "learning_rate": 4.7030833650551965e-06, + "loss": 1.7164, + "step": 463 + }, + { + "epoch": 0.47663071391884954, + "grad_norm": 2.5921683311462402, + "learning_rate": 4.70118005329273e-06, + "loss": 1.6746, + "step": 464 + }, + { + "epoch": 0.4776579352850539, + "grad_norm": 2.6612918376922607, + "learning_rate": 4.699276741530263e-06, + "loss": 1.7598, + "step": 465 + }, + { + "epoch": 0.4786851566512583, + "grad_norm": 2.4633305072784424, + "learning_rate": 4.697373429767797e-06, + "loss": 1.7241, + "step": 466 + }, + { + "epoch": 0.47971237801746275, + "grad_norm": 2.613023281097412, + "learning_rate": 4.69547011800533e-06, + "loss": 1.7088, + "step": 467 + }, + { + "epoch": 0.48073959938366717, + "grad_norm": 2.5329010486602783, + "learning_rate": 4.693566806242863e-06, + "loss": 1.7163, + "step": 468 + }, + { + "epoch": 0.4817668207498716, + "grad_norm": 2.7887041568756104, + "learning_rate": 4.691663494480397e-06, + "loss": 1.6969, + "step": 469 + }, + { + "epoch": 0.482794042116076, + "grad_norm": 2.72991943359375, + "learning_rate": 4.689760182717929e-06, + "loss": 1.7332, + "step": 470 + }, + { + "epoch": 0.482794042116076, + "eval_loss": 1.965472936630249, + "eval_runtime": 46.1059, + "eval_samples_per_second": 21.689, + "eval_steps_per_second": 0.911, + "step": 470 + }, + { + "epoch": 0.48382126348228044, + "grad_norm": 2.613785982131958, + "learning_rate": 4.6878568709554626e-06, + "loss": 1.6272, + "step": 471 + }, + { + "epoch": 0.48484848484848486, + "grad_norm": 2.605299472808838, + "learning_rate": 4.685953559192996e-06, + "loss": 1.6966, + "step": 472 + }, + { + "epoch": 0.4858757062146893, + "grad_norm": 2.672013998031616, + "learning_rate": 4.684050247430529e-06, + "loss": 1.7132, + "step": 473 + }, + { + "epoch": 0.4869029275808937, + "grad_norm": 2.576486825942993, + "learning_rate": 4.682146935668063e-06, + "loss": 1.6449, + "step": 474 + }, + { + "epoch": 0.4879301489470981, + "grad_norm": 2.626840114593506, + "learning_rate": 4.680243623905596e-06, + "loss": 1.6993, + "step": 475 + }, + { + "epoch": 0.4889573703133025, + "grad_norm": 2.974418878555298, + "learning_rate": 4.678340312143129e-06, + "loss": 1.7116, + "step": 476 + }, + { + "epoch": 0.4899845916795069, + "grad_norm": 2.541242837905884, + "learning_rate": 4.676437000380663e-06, + "loss": 1.6693, + "step": 477 + }, + { + "epoch": 0.49101181304571134, + "grad_norm": 2.5494985580444336, + "learning_rate": 4.674533688618196e-06, + "loss": 1.7819, + "step": 478 + }, + { + "epoch": 0.49203903441191577, + "grad_norm": 2.606137990951538, + "learning_rate": 4.6726303768557295e-06, + "loss": 1.7278, + "step": 479 + }, + { + "epoch": 0.4930662557781202, + "grad_norm": 2.5032176971435547, + "learning_rate": 4.670727065093263e-06, + "loss": 1.6318, + "step": 480 + }, + { + "epoch": 0.4930662557781202, + "eval_loss": 1.9851018190383911, + "eval_runtime": 46.0386, + "eval_samples_per_second": 21.721, + "eval_steps_per_second": 0.912, + "step": 480 + }, + { + "epoch": 0.4940934771443246, + "grad_norm": 2.703259229660034, + "learning_rate": 4.668823753330796e-06, + "loss": 1.6646, + "step": 481 + }, + { + "epoch": 0.49512069851052903, + "grad_norm": 2.681807041168213, + "learning_rate": 4.6669204415683295e-06, + "loss": 1.6759, + "step": 482 + }, + { + "epoch": 0.49614791987673346, + "grad_norm": 2.7280473709106445, + "learning_rate": 4.665017129805863e-06, + "loss": 1.7388, + "step": 483 + }, + { + "epoch": 0.4971751412429379, + "grad_norm": 2.420909881591797, + "learning_rate": 4.663113818043396e-06, + "loss": 1.729, + "step": 484 + }, + { + "epoch": 0.49820236260914225, + "grad_norm": 2.5764737129211426, + "learning_rate": 4.66121050628093e-06, + "loss": 1.6903, + "step": 485 + }, + { + "epoch": 0.49922958397534667, + "grad_norm": 2.389531135559082, + "learning_rate": 4.659307194518463e-06, + "loss": 1.7191, + "step": 486 + }, + { + "epoch": 0.5002568053415511, + "grad_norm": 2.5207972526550293, + "learning_rate": 4.6574038827559955e-06, + "loss": 1.6892, + "step": 487 + }, + { + "epoch": 0.5012840267077555, + "grad_norm": 2.5662360191345215, + "learning_rate": 4.655500570993529e-06, + "loss": 1.6978, + "step": 488 + }, + { + "epoch": 0.50231124807396, + "grad_norm": 2.50262713432312, + "learning_rate": 4.653597259231062e-06, + "loss": 1.7379, + "step": 489 + }, + { + "epoch": 0.5033384694401644, + "grad_norm": 2.56643009185791, + "learning_rate": 4.651693947468596e-06, + "loss": 1.6233, + "step": 490 + }, + { + "epoch": 0.5033384694401644, + "eval_loss": 1.9815036058425903, + "eval_runtime": 48.0257, + "eval_samples_per_second": 20.822, + "eval_steps_per_second": 0.875, + "step": 490 + }, + { + "epoch": 0.5043656908063687, + "grad_norm": 2.504117965698242, + "learning_rate": 4.649790635706129e-06, + "loss": 1.7664, + "step": 491 + }, + { + "epoch": 0.5053929121725732, + "grad_norm": 2.906843423843384, + "learning_rate": 4.647887323943662e-06, + "loss": 1.6731, + "step": 492 + }, + { + "epoch": 0.5064201335387776, + "grad_norm": 2.586747169494629, + "learning_rate": 4.645984012181196e-06, + "loss": 1.68, + "step": 493 + }, + { + "epoch": 0.507447354904982, + "grad_norm": 2.3300511837005615, + "learning_rate": 4.644080700418729e-06, + "loss": 1.5998, + "step": 494 + }, + { + "epoch": 0.5084745762711864, + "grad_norm": 2.8711049556732178, + "learning_rate": 4.642177388656262e-06, + "loss": 1.7198, + "step": 495 + }, + { + "epoch": 0.5095017976373909, + "grad_norm": 2.6463162899017334, + "learning_rate": 4.640274076893796e-06, + "loss": 1.6814, + "step": 496 + }, + { + "epoch": 0.5105290190035953, + "grad_norm": 2.650481700897217, + "learning_rate": 4.638370765131328e-06, + "loss": 1.7353, + "step": 497 + }, + { + "epoch": 0.5115562403697997, + "grad_norm": 2.4464986324310303, + "learning_rate": 4.636467453368862e-06, + "loss": 1.6579, + "step": 498 + }, + { + "epoch": 0.5125834617360041, + "grad_norm": 2.438021659851074, + "learning_rate": 4.634564141606396e-06, + "loss": 1.6485, + "step": 499 + }, + { + "epoch": 0.5136106831022085, + "grad_norm": 2.5354347229003906, + "learning_rate": 4.632660829843929e-06, + "loss": 1.7123, + "step": 500 + }, + { + "epoch": 0.5136106831022085, + "eval_loss": 1.9738385677337646, + "eval_runtime": 52.5306, + "eval_samples_per_second": 19.037, + "eval_steps_per_second": 0.8, + "step": 500 + }, + { + "epoch": 0.514637904468413, + "grad_norm": 2.3702070713043213, + "learning_rate": 4.630757518081463e-06, + "loss": 1.6423, + "step": 501 + }, + { + "epoch": 0.5156651258346173, + "grad_norm": 2.5336198806762695, + "learning_rate": 4.628854206318996e-06, + "loss": 1.771, + "step": 502 + }, + { + "epoch": 0.5166923472008218, + "grad_norm": 2.572751998901367, + "learning_rate": 4.626950894556529e-06, + "loss": 1.677, + "step": 503 + }, + { + "epoch": 0.5177195685670262, + "grad_norm": 2.3865134716033936, + "learning_rate": 4.625047582794062e-06, + "loss": 1.7111, + "step": 504 + }, + { + "epoch": 0.5187467899332306, + "grad_norm": 2.5152785778045654, + "learning_rate": 4.623144271031595e-06, + "loss": 1.6343, + "step": 505 + }, + { + "epoch": 0.519774011299435, + "grad_norm": 2.5121307373046875, + "learning_rate": 4.6212409592691286e-06, + "loss": 1.6357, + "step": 506 + }, + { + "epoch": 0.5208012326656395, + "grad_norm": 2.506908655166626, + "learning_rate": 4.619337647506662e-06, + "loss": 1.7295, + "step": 507 + }, + { + "epoch": 0.5218284540318439, + "grad_norm": 2.636082172393799, + "learning_rate": 4.617434335744195e-06, + "loss": 1.6916, + "step": 508 + }, + { + "epoch": 0.5228556753980482, + "grad_norm": 2.5987415313720703, + "learning_rate": 4.615531023981729e-06, + "loss": 1.659, + "step": 509 + }, + { + "epoch": 0.5238828967642527, + "grad_norm": 2.642770290374756, + "learning_rate": 4.613627712219262e-06, + "loss": 1.6789, + "step": 510 + }, + { + "epoch": 0.5238828967642527, + "eval_loss": 1.9659523963928223, + "eval_runtime": 42.9453, + "eval_samples_per_second": 23.285, + "eval_steps_per_second": 0.978, + "step": 510 + }, + { + "epoch": 0.5249101181304571, + "grad_norm": 2.438389301300049, + "learning_rate": 4.611724400456795e-06, + "loss": 1.68, + "step": 511 + }, + { + "epoch": 0.5259373394966615, + "grad_norm": 2.536546468734741, + "learning_rate": 4.609821088694329e-06, + "loss": 1.6847, + "step": 512 + }, + { + "epoch": 0.5269645608628659, + "grad_norm": 2.4056830406188965, + "learning_rate": 4.607917776931862e-06, + "loss": 1.6577, + "step": 513 + }, + { + "epoch": 0.5279917822290704, + "grad_norm": 2.81195330619812, + "learning_rate": 4.606014465169395e-06, + "loss": 1.7149, + "step": 514 + }, + { + "epoch": 0.5290190035952748, + "grad_norm": 2.684948205947876, + "learning_rate": 4.604111153406928e-06, + "loss": 1.717, + "step": 515 + }, + { + "epoch": 0.5300462249614792, + "grad_norm": 2.583092212677002, + "learning_rate": 4.602207841644461e-06, + "loss": 1.6238, + "step": 516 + }, + { + "epoch": 0.5310734463276836, + "grad_norm": 3.022528886795044, + "learning_rate": 4.600304529881995e-06, + "loss": 1.7262, + "step": 517 + }, + { + "epoch": 0.5321006676938881, + "grad_norm": 2.6941866874694824, + "learning_rate": 4.598401218119528e-06, + "loss": 1.7055, + "step": 518 + }, + { + "epoch": 0.5331278890600925, + "grad_norm": 2.5207839012145996, + "learning_rate": 4.5964979063570614e-06, + "loss": 1.7117, + "step": 519 + }, + { + "epoch": 0.5341551104262968, + "grad_norm": 2.8056678771972656, + "learning_rate": 4.594594594594596e-06, + "loss": 1.657, + "step": 520 + }, + { + "epoch": 0.5341551104262968, + "eval_loss": 1.9712116718292236, + "eval_runtime": 49.57, + "eval_samples_per_second": 20.173, + "eval_steps_per_second": 0.847, + "step": 520 + }, + { + "epoch": 0.5351823317925013, + "grad_norm": 2.6119930744171143, + "learning_rate": 4.592691282832128e-06, + "loss": 1.6897, + "step": 521 + }, + { + "epoch": 0.5362095531587057, + "grad_norm": 2.7165610790252686, + "learning_rate": 4.5907879710696615e-06, + "loss": 1.7038, + "step": 522 + }, + { + "epoch": 0.5372367745249101, + "grad_norm": 2.3613736629486084, + "learning_rate": 4.588884659307195e-06, + "loss": 1.6222, + "step": 523 + }, + { + "epoch": 0.5382639958911145, + "grad_norm": 2.647672176361084, + "learning_rate": 4.586981347544728e-06, + "loss": 1.702, + "step": 524 + }, + { + "epoch": 0.539291217257319, + "grad_norm": 2.704634428024292, + "learning_rate": 4.585078035782262e-06, + "loss": 1.6566, + "step": 525 + }, + { + "epoch": 0.5403184386235234, + "grad_norm": 2.6634597778320312, + "learning_rate": 4.583174724019795e-06, + "loss": 1.6903, + "step": 526 + }, + { + "epoch": 0.5413456599897278, + "grad_norm": 2.5745162963867188, + "learning_rate": 4.581271412257328e-06, + "loss": 1.6375, + "step": 527 + }, + { + "epoch": 0.5423728813559322, + "grad_norm": 2.5577778816223145, + "learning_rate": 4.579368100494862e-06, + "loss": 1.614, + "step": 528 + }, + { + "epoch": 0.5434001027221366, + "grad_norm": 2.7684454917907715, + "learning_rate": 4.577464788732395e-06, + "loss": 1.7168, + "step": 529 + }, + { + "epoch": 0.544427324088341, + "grad_norm": 2.582108736038208, + "learning_rate": 4.575561476969928e-06, + "loss": 1.6897, + "step": 530 + }, + { + "epoch": 0.544427324088341, + "eval_loss": 1.9821091890335083, + "eval_runtime": 50.9409, + "eval_samples_per_second": 19.631, + "eval_steps_per_second": 0.824, + "step": 530 + }, + { + "epoch": 0.5454545454545454, + "grad_norm": 2.516486406326294, + "learning_rate": 4.573658165207461e-06, + "loss": 1.7273, + "step": 531 + }, + { + "epoch": 0.5464817668207499, + "grad_norm": 2.786806344985962, + "learning_rate": 4.571754853444994e-06, + "loss": 1.6898, + "step": 532 + }, + { + "epoch": 0.5475089881869543, + "grad_norm": 2.3137319087982178, + "learning_rate": 4.569851541682528e-06, + "loss": 1.676, + "step": 533 + }, + { + "epoch": 0.5485362095531587, + "grad_norm": 2.7268903255462646, + "learning_rate": 4.567948229920061e-06, + "loss": 1.6483, + "step": 534 + }, + { + "epoch": 0.5495634309193631, + "grad_norm": 2.4385902881622314, + "learning_rate": 4.566044918157594e-06, + "loss": 1.6172, + "step": 535 + }, + { + "epoch": 0.5505906522855676, + "grad_norm": 2.420576333999634, + "learning_rate": 4.564141606395128e-06, + "loss": 1.6569, + "step": 536 + }, + { + "epoch": 0.551617873651772, + "grad_norm": 2.592290163040161, + "learning_rate": 4.562238294632661e-06, + "loss": 1.7085, + "step": 537 + }, + { + "epoch": 0.5526450950179764, + "grad_norm": 2.485285520553589, + "learning_rate": 4.5603349828701945e-06, + "loss": 1.6699, + "step": 538 + }, + { + "epoch": 0.5536723163841808, + "grad_norm": 2.844710111618042, + "learning_rate": 4.558431671107728e-06, + "loss": 1.6873, + "step": 539 + }, + { + "epoch": 0.5546995377503852, + "grad_norm": 2.5279181003570557, + "learning_rate": 4.556528359345261e-06, + "loss": 1.7285, + "step": 540 + }, + { + "epoch": 0.5546995377503852, + "eval_loss": 1.9825981855392456, + "eval_runtime": 52.461, + "eval_samples_per_second": 19.062, + "eval_steps_per_second": 0.801, + "step": 540 + }, + { + "epoch": 0.5557267591165896, + "grad_norm": 2.673997163772583, + "learning_rate": 4.5546250475827946e-06, + "loss": 1.6658, + "step": 541 + }, + { + "epoch": 0.556753980482794, + "grad_norm": 2.6023194789886475, + "learning_rate": 4.552721735820328e-06, + "loss": 1.6262, + "step": 542 + }, + { + "epoch": 0.5577812018489985, + "grad_norm": 2.6327662467956543, + "learning_rate": 4.550818424057861e-06, + "loss": 1.6574, + "step": 543 + }, + { + "epoch": 0.5588084232152029, + "grad_norm": 2.5532140731811523, + "learning_rate": 4.548915112295395e-06, + "loss": 1.6607, + "step": 544 + }, + { + "epoch": 0.5598356445814073, + "grad_norm": 2.523966073989868, + "learning_rate": 4.547011800532928e-06, + "loss": 1.6752, + "step": 545 + }, + { + "epoch": 0.5608628659476117, + "grad_norm": 2.6338107585906982, + "learning_rate": 4.545108488770461e-06, + "loss": 1.6258, + "step": 546 + }, + { + "epoch": 0.5618900873138162, + "grad_norm": 2.500096559524536, + "learning_rate": 4.543205177007995e-06, + "loss": 1.7189, + "step": 547 + }, + { + "epoch": 0.5629173086800205, + "grad_norm": 2.644413471221924, + "learning_rate": 4.541301865245527e-06, + "loss": 1.6594, + "step": 548 + }, + { + "epoch": 0.5639445300462249, + "grad_norm": 2.451965093612671, + "learning_rate": 4.539398553483061e-06, + "loss": 1.6984, + "step": 549 + }, + { + "epoch": 0.5649717514124294, + "grad_norm": 2.7554423809051514, + "learning_rate": 4.537495241720594e-06, + "loss": 1.6626, + "step": 550 + }, + { + "epoch": 0.5649717514124294, + "eval_loss": 1.9897041320800781, + "eval_runtime": 45.9349, + "eval_samples_per_second": 21.77, + "eval_steps_per_second": 0.914, + "step": 550 + }, + { + "epoch": 0.5659989727786338, + "grad_norm": 2.5962674617767334, + "learning_rate": 4.535591929958127e-06, + "loss": 1.7015, + "step": 551 + }, + { + "epoch": 0.5670261941448382, + "grad_norm": 2.7917397022247314, + "learning_rate": 4.533688618195661e-06, + "loss": 1.6811, + "step": 552 + }, + { + "epoch": 0.5680534155110426, + "grad_norm": 2.4205498695373535, + "learning_rate": 4.531785306433194e-06, + "loss": 1.6659, + "step": 553 + }, + { + "epoch": 0.5690806368772471, + "grad_norm": 2.74898624420166, + "learning_rate": 4.529881994670727e-06, + "loss": 1.6521, + "step": 554 + }, + { + "epoch": 0.5701078582434514, + "grad_norm": 2.859898328781128, + "learning_rate": 4.527978682908261e-06, + "loss": 1.6445, + "step": 555 + }, + { + "epoch": 0.5711350796096559, + "grad_norm": 2.5526154041290283, + "learning_rate": 4.526075371145794e-06, + "loss": 1.6679, + "step": 556 + }, + { + "epoch": 0.5721623009758603, + "grad_norm": 2.96175217628479, + "learning_rate": 4.5241720593833275e-06, + "loss": 1.6784, + "step": 557 + }, + { + "epoch": 0.5731895223420647, + "grad_norm": 2.5037007331848145, + "learning_rate": 4.52226874762086e-06, + "loss": 1.6689, + "step": 558 + }, + { + "epoch": 0.5742167437082691, + "grad_norm": 2.6513259410858154, + "learning_rate": 4.520365435858393e-06, + "loss": 1.7038, + "step": 559 + }, + { + "epoch": 0.5752439650744735, + "grad_norm": 2.762770175933838, + "learning_rate": 4.518462124095928e-06, + "loss": 1.728, + "step": 560 + }, + { + "epoch": 0.5752439650744735, + "eval_loss": 1.972934603691101, + "eval_runtime": 44.2572, + "eval_samples_per_second": 22.595, + "eval_steps_per_second": 0.949, + "step": 560 + }, + { + "epoch": 0.576271186440678, + "grad_norm": 2.637406349182129, + "learning_rate": 4.516558812333461e-06, + "loss": 1.682, + "step": 561 + }, + { + "epoch": 0.5772984078068824, + "grad_norm": 2.545583724975586, + "learning_rate": 4.514655500570994e-06, + "loss": 1.6669, + "step": 562 + }, + { + "epoch": 0.5783256291730868, + "grad_norm": 2.5655033588409424, + "learning_rate": 4.512752188808528e-06, + "loss": 1.7279, + "step": 563 + }, + { + "epoch": 0.5793528505392912, + "grad_norm": 2.6856470108032227, + "learning_rate": 4.510848877046061e-06, + "loss": 1.7331, + "step": 564 + }, + { + "epoch": 0.5803800719054957, + "grad_norm": 2.605252981185913, + "learning_rate": 4.5089455652835936e-06, + "loss": 1.6904, + "step": 565 + }, + { + "epoch": 0.5814072932717, + "grad_norm": 2.610267400741577, + "learning_rate": 4.507042253521127e-06, + "loss": 1.6607, + "step": 566 + }, + { + "epoch": 0.5824345146379045, + "grad_norm": 2.6955432891845703, + "learning_rate": 4.50513894175866e-06, + "loss": 1.6166, + "step": 567 + }, + { + "epoch": 0.5834617360041089, + "grad_norm": 2.5661842823028564, + "learning_rate": 4.503235629996194e-06, + "loss": 1.6955, + "step": 568 + }, + { + "epoch": 0.5844889573703133, + "grad_norm": 2.6393115520477295, + "learning_rate": 4.501332318233727e-06, + "loss": 1.6933, + "step": 569 + }, + { + "epoch": 0.5855161787365177, + "grad_norm": 2.612645387649536, + "learning_rate": 4.49942900647126e-06, + "loss": 1.6585, + "step": 570 + }, + { + "epoch": 0.5855161787365177, + "eval_loss": 1.9599064588546753, + "eval_runtime": 46.0676, + "eval_samples_per_second": 21.707, + "eval_steps_per_second": 0.912, + "step": 570 + }, + { + "epoch": 0.5865434001027221, + "grad_norm": 2.4182891845703125, + "learning_rate": 4.497525694708794e-06, + "loss": 1.707, + "step": 571 + }, + { + "epoch": 0.5875706214689266, + "grad_norm": 2.7214412689208984, + "learning_rate": 4.495622382946327e-06, + "loss": 1.6966, + "step": 572 + }, + { + "epoch": 0.588597842835131, + "grad_norm": 2.5138018131256104, + "learning_rate": 4.4937190711838605e-06, + "loss": 1.6373, + "step": 573 + }, + { + "epoch": 0.5896250642013354, + "grad_norm": 2.5814292430877686, + "learning_rate": 4.491815759421394e-06, + "loss": 1.7313, + "step": 574 + }, + { + "epoch": 0.5906522855675398, + "grad_norm": 2.6236367225646973, + "learning_rate": 4.489912447658926e-06, + "loss": 1.6647, + "step": 575 + }, + { + "epoch": 0.5916795069337443, + "grad_norm": 2.468557357788086, + "learning_rate": 4.48800913589646e-06, + "loss": 1.6412, + "step": 576 + }, + { + "epoch": 0.5927067282999486, + "grad_norm": 2.454141855239868, + "learning_rate": 4.486105824133993e-06, + "loss": 1.6762, + "step": 577 + }, + { + "epoch": 0.593733949666153, + "grad_norm": 3.0009915828704834, + "learning_rate": 4.4842025123715264e-06, + "loss": 1.6847, + "step": 578 + }, + { + "epoch": 0.5947611710323575, + "grad_norm": 2.574394702911377, + "learning_rate": 4.48229920060906e-06, + "loss": 1.6421, + "step": 579 + }, + { + "epoch": 0.5957883923985619, + "grad_norm": 2.6978182792663574, + "learning_rate": 4.480395888846594e-06, + "loss": 1.6198, + "step": 580 + }, + { + "epoch": 0.5957883923985619, + "eval_loss": 1.9570144414901733, + "eval_runtime": 44.7661, + "eval_samples_per_second": 22.338, + "eval_steps_per_second": 0.938, + "step": 580 + }, + { + "epoch": 0.5968156137647663, + "grad_norm": 2.652461051940918, + "learning_rate": 4.478492577084127e-06, + "loss": 1.7334, + "step": 581 + }, + { + "epoch": 0.5978428351309707, + "grad_norm": 2.6292548179626465, + "learning_rate": 4.47658926532166e-06, + "loss": 1.6242, + "step": 582 + }, + { + "epoch": 0.5988700564971752, + "grad_norm": 2.474919080734253, + "learning_rate": 4.474685953559193e-06, + "loss": 1.6732, + "step": 583 + }, + { + "epoch": 0.5998972778633795, + "grad_norm": 2.7282516956329346, + "learning_rate": 4.472782641796727e-06, + "loss": 1.802, + "step": 584 + }, + { + "epoch": 0.600924499229584, + "grad_norm": 2.770307779312134, + "learning_rate": 4.47087933003426e-06, + "loss": 1.6156, + "step": 585 + }, + { + "epoch": 0.6019517205957884, + "grad_norm": 2.570681095123291, + "learning_rate": 4.468976018271793e-06, + "loss": 1.6309, + "step": 586 + }, + { + "epoch": 0.6029789419619929, + "grad_norm": 2.4075732231140137, + "learning_rate": 4.467072706509327e-06, + "loss": 1.6436, + "step": 587 + }, + { + "epoch": 0.6040061633281972, + "grad_norm": 2.4570937156677246, + "learning_rate": 4.46516939474686e-06, + "loss": 1.6737, + "step": 588 + }, + { + "epoch": 0.6050333846944016, + "grad_norm": 2.407274007797241, + "learning_rate": 4.463266082984393e-06, + "loss": 1.6179, + "step": 589 + }, + { + "epoch": 0.6060606060606061, + "grad_norm": 2.745486259460449, + "learning_rate": 4.461362771221927e-06, + "loss": 1.7458, + "step": 590 + }, + { + "epoch": 0.6060606060606061, + "eval_loss": 1.9700438976287842, + "eval_runtime": 45.3018, + "eval_samples_per_second": 22.074, + "eval_steps_per_second": 0.927, + "step": 590 + }, + { + "epoch": 0.6070878274268104, + "grad_norm": 2.4258809089660645, + "learning_rate": 4.45945945945946e-06, + "loss": 1.6372, + "step": 591 + }, + { + "epoch": 0.6081150487930149, + "grad_norm": 2.683662176132202, + "learning_rate": 4.457556147696993e-06, + "loss": 1.6649, + "step": 592 + }, + { + "epoch": 0.6091422701592193, + "grad_norm": 2.879727602005005, + "learning_rate": 4.455652835934526e-06, + "loss": 1.7376, + "step": 593 + }, + { + "epoch": 0.6101694915254238, + "grad_norm": 2.5620689392089844, + "learning_rate": 4.453749524172059e-06, + "loss": 1.7203, + "step": 594 + }, + { + "epoch": 0.6111967128916281, + "grad_norm": 2.662031888961792, + "learning_rate": 4.451846212409593e-06, + "loss": 1.6835, + "step": 595 + }, + { + "epoch": 0.6122239342578326, + "grad_norm": 2.6269922256469727, + "learning_rate": 4.449942900647126e-06, + "loss": 1.6726, + "step": 596 + }, + { + "epoch": 0.613251155624037, + "grad_norm": 2.515852212905884, + "learning_rate": 4.4480395888846595e-06, + "loss": 1.6045, + "step": 597 + }, + { + "epoch": 0.6142783769902413, + "grad_norm": 3.006873607635498, + "learning_rate": 4.446136277122193e-06, + "loss": 1.6827, + "step": 598 + }, + { + "epoch": 0.6153055983564458, + "grad_norm": 2.2904858589172363, + "learning_rate": 4.444232965359726e-06, + "loss": 1.6123, + "step": 599 + }, + { + "epoch": 0.6163328197226502, + "grad_norm": 2.7345471382141113, + "learning_rate": 4.4423296535972596e-06, + "loss": 1.6697, + "step": 600 + }, + { + "epoch": 0.6163328197226502, + "eval_loss": 1.9713997840881348, + "eval_runtime": 45.6715, + "eval_samples_per_second": 21.895, + "eval_steps_per_second": 0.92, + "step": 600 + }, + { + "epoch": 0.6173600410888547, + "grad_norm": 2.955853223800659, + "learning_rate": 4.440426341834793e-06, + "loss": 1.6195, + "step": 601 + }, + { + "epoch": 0.618387262455059, + "grad_norm": 2.4079766273498535, + "learning_rate": 4.438523030072326e-06, + "loss": 1.6589, + "step": 602 + }, + { + "epoch": 0.6194144838212635, + "grad_norm": 2.8158490657806396, + "learning_rate": 4.43661971830986e-06, + "loss": 1.6895, + "step": 603 + }, + { + "epoch": 0.6204417051874679, + "grad_norm": 2.491471529006958, + "learning_rate": 4.434716406547393e-06, + "loss": 1.6351, + "step": 604 + }, + { + "epoch": 0.6214689265536724, + "grad_norm": 2.4093260765075684, + "learning_rate": 4.432813094784926e-06, + "loss": 1.6402, + "step": 605 + }, + { + "epoch": 0.6224961479198767, + "grad_norm": 2.751018762588501, + "learning_rate": 4.43090978302246e-06, + "loss": 1.6962, + "step": 606 + }, + { + "epoch": 0.6235233692860811, + "grad_norm": 2.4410409927368164, + "learning_rate": 4.429006471259993e-06, + "loss": 1.6549, + "step": 607 + }, + { + "epoch": 0.6245505906522856, + "grad_norm": 2.360455274581909, + "learning_rate": 4.4271031594975265e-06, + "loss": 1.6449, + "step": 608 + }, + { + "epoch": 0.6255778120184899, + "grad_norm": 2.5058023929595947, + "learning_rate": 4.425199847735059e-06, + "loss": 1.617, + "step": 609 + }, + { + "epoch": 0.6266050333846944, + "grad_norm": 2.3429927825927734, + "learning_rate": 4.423296535972592e-06, + "loss": 1.6262, + "step": 610 + }, + { + "epoch": 0.6266050333846944, + "eval_loss": 1.9597365856170654, + "eval_runtime": 42.2034, + "eval_samples_per_second": 23.695, + "eval_steps_per_second": 0.995, + "step": 610 + }, + { + "epoch": 0.6276322547508988, + "grad_norm": 2.401761054992676, + "learning_rate": 4.421393224210126e-06, + "loss": 1.6497, + "step": 611 + }, + { + "epoch": 0.6286594761171033, + "grad_norm": 2.433439254760742, + "learning_rate": 4.419489912447659e-06, + "loss": 1.6117, + "step": 612 + }, + { + "epoch": 0.6296866974833076, + "grad_norm": 2.5147452354431152, + "learning_rate": 4.4175866006851924e-06, + "loss": 1.7041, + "step": 613 + }, + { + "epoch": 0.6307139188495121, + "grad_norm": 2.4174447059631348, + "learning_rate": 4.415683288922726e-06, + "loss": 1.6652, + "step": 614 + }, + { + "epoch": 0.6317411402157165, + "grad_norm": 2.424800157546997, + "learning_rate": 4.413779977160259e-06, + "loss": 1.6407, + "step": 615 + }, + { + "epoch": 0.632768361581921, + "grad_norm": 2.6760315895080566, + "learning_rate": 4.4118766653977925e-06, + "loss": 1.6566, + "step": 616 + }, + { + "epoch": 0.6337955829481253, + "grad_norm": 2.694810628890991, + "learning_rate": 4.409973353635326e-06, + "loss": 1.7125, + "step": 617 + }, + { + "epoch": 0.6348228043143297, + "grad_norm": 2.475126266479492, + "learning_rate": 4.408070041872859e-06, + "loss": 1.6663, + "step": 618 + }, + { + "epoch": 0.6358500256805342, + "grad_norm": 2.734738349914551, + "learning_rate": 4.406166730110393e-06, + "loss": 1.6115, + "step": 619 + }, + { + "epoch": 0.6368772470467385, + "grad_norm": 2.4914753437042236, + "learning_rate": 4.404263418347925e-06, + "loss": 1.6096, + "step": 620 + }, + { + "epoch": 0.6368772470467385, + "eval_loss": 1.952081322669983, + "eval_runtime": 45.4451, + "eval_samples_per_second": 22.005, + "eval_steps_per_second": 0.924, + "step": 620 + }, + { + "epoch": 0.637904468412943, + "grad_norm": 2.3727800846099854, + "learning_rate": 4.402360106585459e-06, + "loss": 1.6275, + "step": 621 + }, + { + "epoch": 0.6389316897791474, + "grad_norm": 2.576402187347412, + "learning_rate": 4.400456794822993e-06, + "loss": 1.6252, + "step": 622 + }, + { + "epoch": 0.6399589111453519, + "grad_norm": 2.560185432434082, + "learning_rate": 4.398553483060526e-06, + "loss": 1.6522, + "step": 623 + }, + { + "epoch": 0.6409861325115562, + "grad_norm": 2.462785005569458, + "learning_rate": 4.396650171298059e-06, + "loss": 1.6124, + "step": 624 + }, + { + "epoch": 0.6420133538777607, + "grad_norm": 2.3942930698394775, + "learning_rate": 4.394746859535593e-06, + "loss": 1.6231, + "step": 625 + }, + { + "epoch": 0.6430405752439651, + "grad_norm": 2.4870522022247314, + "learning_rate": 4.392843547773125e-06, + "loss": 1.6086, + "step": 626 + }, + { + "epoch": 0.6440677966101694, + "grad_norm": 2.5702786445617676, + "learning_rate": 4.390940236010659e-06, + "loss": 1.6228, + "step": 627 + }, + { + "epoch": 0.6450950179763739, + "grad_norm": 2.743983268737793, + "learning_rate": 4.389036924248192e-06, + "loss": 1.6394, + "step": 628 + }, + { + "epoch": 0.6461222393425783, + "grad_norm": 2.59942364692688, + "learning_rate": 4.387133612485725e-06, + "loss": 1.6721, + "step": 629 + }, + { + "epoch": 0.6471494607087828, + "grad_norm": 2.4473204612731934, + "learning_rate": 4.385230300723259e-06, + "loss": 1.6407, + "step": 630 + }, + { + "epoch": 0.6471494607087828, + "eval_loss": 1.9426974058151245, + "eval_runtime": 45.2318, + "eval_samples_per_second": 22.108, + "eval_steps_per_second": 0.929, + "step": 630 + }, + { + "epoch": 0.6481766820749871, + "grad_norm": 2.6036875247955322, + "learning_rate": 4.383326988960792e-06, + "loss": 1.6307, + "step": 631 + }, + { + "epoch": 0.6492039034411916, + "grad_norm": 2.691011428833008, + "learning_rate": 4.3814236771983255e-06, + "loss": 1.6059, + "step": 632 + }, + { + "epoch": 0.650231124807396, + "grad_norm": 2.585758924484253, + "learning_rate": 4.379520365435859e-06, + "loss": 1.6781, + "step": 633 + }, + { + "epoch": 0.6512583461736005, + "grad_norm": 2.9522950649261475, + "learning_rate": 4.377617053673392e-06, + "loss": 1.6559, + "step": 634 + }, + { + "epoch": 0.6522855675398048, + "grad_norm": 2.5368709564208984, + "learning_rate": 4.3757137419109256e-06, + "loss": 1.6445, + "step": 635 + }, + { + "epoch": 0.6533127889060092, + "grad_norm": 2.5002150535583496, + "learning_rate": 4.373810430148459e-06, + "loss": 1.7184, + "step": 636 + }, + { + "epoch": 0.6543400102722137, + "grad_norm": 2.7623229026794434, + "learning_rate": 4.3719071183859914e-06, + "loss": 1.6195, + "step": 637 + }, + { + "epoch": 0.655367231638418, + "grad_norm": 2.6188151836395264, + "learning_rate": 4.370003806623525e-06, + "loss": 1.6686, + "step": 638 + }, + { + "epoch": 0.6563944530046225, + "grad_norm": 2.685776948928833, + "learning_rate": 4.368100494861058e-06, + "loss": 1.6134, + "step": 639 + }, + { + "epoch": 0.6574216743708269, + "grad_norm": 2.5703907012939453, + "learning_rate": 4.3661971830985915e-06, + "loss": 1.665, + "step": 640 + }, + { + "epoch": 0.6574216743708269, + "eval_loss": 1.9504806995391846, + "eval_runtime": 47.1947, + "eval_samples_per_second": 21.189, + "eval_steps_per_second": 0.89, + "step": 640 + }, + { + "epoch": 0.6584488957370314, + "grad_norm": 2.511617660522461, + "learning_rate": 4.364293871336126e-06, + "loss": 1.6201, + "step": 641 + }, + { + "epoch": 0.6594761171032357, + "grad_norm": 2.549302339553833, + "learning_rate": 4.362390559573659e-06, + "loss": 1.647, + "step": 642 + }, + { + "epoch": 0.6605033384694402, + "grad_norm": 2.348747730255127, + "learning_rate": 4.3604872478111925e-06, + "loss": 1.5993, + "step": 643 + }, + { + "epoch": 0.6615305598356446, + "grad_norm": 2.7393345832824707, + "learning_rate": 4.358583936048725e-06, + "loss": 1.543, + "step": 644 + }, + { + "epoch": 0.662557781201849, + "grad_norm": 2.4242193698883057, + "learning_rate": 4.356680624286258e-06, + "loss": 1.6726, + "step": 645 + }, + { + "epoch": 0.6635850025680534, + "grad_norm": 2.6446714401245117, + "learning_rate": 4.354777312523792e-06, + "loss": 1.6312, + "step": 646 + }, + { + "epoch": 0.6646122239342578, + "grad_norm": 2.7407124042510986, + "learning_rate": 4.352874000761325e-06, + "loss": 1.6868, + "step": 647 + }, + { + "epoch": 0.6656394453004623, + "grad_norm": 2.3706371784210205, + "learning_rate": 4.3509706889988584e-06, + "loss": 1.6301, + "step": 648 + }, + { + "epoch": 0.6666666666666666, + "grad_norm": 2.5197198390960693, + "learning_rate": 4.349067377236392e-06, + "loss": 1.702, + "step": 649 + }, + { + "epoch": 0.6676938880328711, + "grad_norm": 2.4792377948760986, + "learning_rate": 4.347164065473925e-06, + "loss": 1.6891, + "step": 650 + }, + { + "epoch": 0.6676938880328711, + "eval_loss": 1.9589039087295532, + "eval_runtime": 47.1512, + "eval_samples_per_second": 21.208, + "eval_steps_per_second": 0.891, + "step": 650 + }, + { + "epoch": 0.6687211093990755, + "grad_norm": 2.4003701210021973, + "learning_rate": 4.3452607537114585e-06, + "loss": 1.7015, + "step": 651 + }, + { + "epoch": 0.66974833076528, + "grad_norm": 2.5363786220550537, + "learning_rate": 4.343357441948992e-06, + "loss": 1.6133, + "step": 652 + }, + { + "epoch": 0.6707755521314843, + "grad_norm": 2.5183844566345215, + "learning_rate": 4.341454130186525e-06, + "loss": 1.6574, + "step": 653 + }, + { + "epoch": 0.6718027734976888, + "grad_norm": 2.4668304920196533, + "learning_rate": 4.339550818424058e-06, + "loss": 1.6253, + "step": 654 + }, + { + "epoch": 0.6728299948638932, + "grad_norm": 2.4829814434051514, + "learning_rate": 4.337647506661591e-06, + "loss": 1.6099, + "step": 655 + }, + { + "epoch": 0.6738572162300975, + "grad_norm": 2.3683176040649414, + "learning_rate": 4.3357441948991245e-06, + "loss": 1.6385, + "step": 656 + }, + { + "epoch": 0.674884437596302, + "grad_norm": 2.4906842708587646, + "learning_rate": 4.333840883136658e-06, + "loss": 1.6755, + "step": 657 + }, + { + "epoch": 0.6759116589625064, + "grad_norm": 2.558776617050171, + "learning_rate": 4.331937571374191e-06, + "loss": 1.6152, + "step": 658 + }, + { + "epoch": 0.6769388803287109, + "grad_norm": 2.4081523418426514, + "learning_rate": 4.3300342596117246e-06, + "loss": 1.6224, + "step": 659 + }, + { + "epoch": 0.6779661016949152, + "grad_norm": 2.488370180130005, + "learning_rate": 4.328130947849258e-06, + "loss": 1.6298, + "step": 660 + }, + { + "epoch": 0.6779661016949152, + "eval_loss": 1.9484621286392212, + "eval_runtime": 52.549, + "eval_samples_per_second": 19.03, + "eval_steps_per_second": 0.799, + "step": 660 + }, + { + "epoch": 0.6789933230611197, + "grad_norm": 2.495748281478882, + "learning_rate": 4.326227636086791e-06, + "loss": 1.5852, + "step": 661 + }, + { + "epoch": 0.6800205444273241, + "grad_norm": 2.525031566619873, + "learning_rate": 4.324324324324325e-06, + "loss": 1.6348, + "step": 662 + }, + { + "epoch": 0.6810477657935285, + "grad_norm": 2.427615165710449, + "learning_rate": 4.322421012561858e-06, + "loss": 1.6352, + "step": 663 + }, + { + "epoch": 0.6820749871597329, + "grad_norm": 2.5084927082061768, + "learning_rate": 4.320517700799391e-06, + "loss": 1.6334, + "step": 664 + }, + { + "epoch": 0.6831022085259374, + "grad_norm": 2.6018898487091064, + "learning_rate": 4.318614389036925e-06, + "loss": 1.7212, + "step": 665 + }, + { + "epoch": 0.6841294298921418, + "grad_norm": 2.44184947013855, + "learning_rate": 4.316711077274458e-06, + "loss": 1.6802, + "step": 666 + }, + { + "epoch": 0.6851566512583461, + "grad_norm": 2.542123317718506, + "learning_rate": 4.3148077655119915e-06, + "loss": 1.5912, + "step": 667 + }, + { + "epoch": 0.6861838726245506, + "grad_norm": 2.5208137035369873, + "learning_rate": 4.312904453749525e-06, + "loss": 1.6965, + "step": 668 + }, + { + "epoch": 0.687211093990755, + "grad_norm": 2.391904592514038, + "learning_rate": 4.311001141987058e-06, + "loss": 1.6388, + "step": 669 + }, + { + "epoch": 0.6882383153569595, + "grad_norm": 2.629565477371216, + "learning_rate": 4.3090978302245916e-06, + "loss": 1.6475, + "step": 670 + }, + { + "epoch": 0.6882383153569595, + "eval_loss": 1.9572818279266357, + "eval_runtime": 48.5354, + "eval_samples_per_second": 20.604, + "eval_steps_per_second": 0.865, + "step": 670 + }, + { + "epoch": 0.6892655367231638, + "grad_norm": 2.3132376670837402, + "learning_rate": 4.307194518462124e-06, + "loss": 1.6399, + "step": 671 + }, + { + "epoch": 0.6902927580893683, + "grad_norm": 2.5003411769866943, + "learning_rate": 4.3052912066996574e-06, + "loss": 1.7136, + "step": 672 + }, + { + "epoch": 0.6913199794555727, + "grad_norm": 2.545541286468506, + "learning_rate": 4.303387894937191e-06, + "loss": 1.6825, + "step": 673 + }, + { + "epoch": 0.6923472008217771, + "grad_norm": 2.342423439025879, + "learning_rate": 4.301484583174724e-06, + "loss": 1.6088, + "step": 674 + }, + { + "epoch": 0.6933744221879815, + "grad_norm": 2.5516135692596436, + "learning_rate": 4.2995812714122575e-06, + "loss": 1.5934, + "step": 675 + }, + { + "epoch": 0.6944016435541859, + "grad_norm": 2.438675880432129, + "learning_rate": 4.297677959649791e-06, + "loss": 1.6236, + "step": 676 + }, + { + "epoch": 0.6954288649203904, + "grad_norm": 2.767305612564087, + "learning_rate": 4.295774647887324e-06, + "loss": 1.6514, + "step": 677 + }, + { + "epoch": 0.6964560862865947, + "grad_norm": 2.591994047164917, + "learning_rate": 4.293871336124858e-06, + "loss": 1.6643, + "step": 678 + }, + { + "epoch": 0.6974833076527992, + "grad_norm": 2.4734244346618652, + "learning_rate": 4.291968024362391e-06, + "loss": 1.6362, + "step": 679 + }, + { + "epoch": 0.6985105290190036, + "grad_norm": 2.6181890964508057, + "learning_rate": 4.290064712599924e-06, + "loss": 1.6666, + "step": 680 + }, + { + "epoch": 0.6985105290190036, + "eval_loss": 1.9585388898849487, + "eval_runtime": 46.8094, + "eval_samples_per_second": 21.363, + "eval_steps_per_second": 0.897, + "step": 680 + }, + { + "epoch": 0.699537750385208, + "grad_norm": 2.465571403503418, + "learning_rate": 4.288161400837458e-06, + "loss": 1.6588, + "step": 681 + }, + { + "epoch": 0.7005649717514124, + "grad_norm": 2.664262294769287, + "learning_rate": 4.286258089074991e-06, + "loss": 1.5862, + "step": 682 + }, + { + "epoch": 0.7015921931176169, + "grad_norm": 2.677140474319458, + "learning_rate": 4.284354777312524e-06, + "loss": 1.6549, + "step": 683 + }, + { + "epoch": 0.7026194144838213, + "grad_norm": 2.559051275253296, + "learning_rate": 4.282451465550058e-06, + "loss": 1.609, + "step": 684 + }, + { + "epoch": 0.7036466358500256, + "grad_norm": 2.428424119949341, + "learning_rate": 4.280548153787591e-06, + "loss": 1.6243, + "step": 685 + }, + { + "epoch": 0.7046738572162301, + "grad_norm": 2.54136061668396, + "learning_rate": 4.2786448420251245e-06, + "loss": 1.6301, + "step": 686 + }, + { + "epoch": 0.7057010785824345, + "grad_norm": 2.377699136734009, + "learning_rate": 4.276741530262658e-06, + "loss": 1.5903, + "step": 687 + }, + { + "epoch": 0.706728299948639, + "grad_norm": 2.5548322200775146, + "learning_rate": 4.27483821850019e-06, + "loss": 1.6161, + "step": 688 + }, + { + "epoch": 0.7077555213148433, + "grad_norm": 2.5993409156799316, + "learning_rate": 4.272934906737724e-06, + "loss": 1.6882, + "step": 689 + }, + { + "epoch": 0.7087827426810478, + "grad_norm": 2.4249117374420166, + "learning_rate": 4.271031594975257e-06, + "loss": 1.6723, + "step": 690 + }, + { + "epoch": 0.7087827426810478, + "eval_loss": 1.9526821374893188, + "eval_runtime": 48.2168, + "eval_samples_per_second": 20.74, + "eval_steps_per_second": 0.871, + "step": 690 + }, + { + "epoch": 0.7098099640472522, + "grad_norm": 2.6965081691741943, + "learning_rate": 4.2691282832127905e-06, + "loss": 1.6636, + "step": 691 + }, + { + "epoch": 0.7108371854134566, + "grad_norm": 2.6205356121063232, + "learning_rate": 4.267224971450324e-06, + "loss": 1.6231, + "step": 692 + }, + { + "epoch": 0.711864406779661, + "grad_norm": 2.5924129486083984, + "learning_rate": 4.265321659687857e-06, + "loss": 1.5634, + "step": 693 + }, + { + "epoch": 0.7128916281458655, + "grad_norm": 2.9710943698883057, + "learning_rate": 4.2634183479253906e-06, + "loss": 1.6134, + "step": 694 + }, + { + "epoch": 0.7139188495120699, + "grad_norm": 2.6252782344818115, + "learning_rate": 4.261515036162924e-06, + "loss": 1.5997, + "step": 695 + }, + { + "epoch": 0.7149460708782742, + "grad_norm": 2.6178832054138184, + "learning_rate": 4.259611724400457e-06, + "loss": 1.6255, + "step": 696 + }, + { + "epoch": 0.7159732922444787, + "grad_norm": 2.82548189163208, + "learning_rate": 4.257708412637991e-06, + "loss": 1.6167, + "step": 697 + }, + { + "epoch": 0.7170005136106831, + "grad_norm": 2.6430351734161377, + "learning_rate": 4.255805100875523e-06, + "loss": 1.5585, + "step": 698 + }, + { + "epoch": 0.7180277349768875, + "grad_norm": 2.85046124458313, + "learning_rate": 4.2539017891130565e-06, + "loss": 1.5873, + "step": 699 + }, + { + "epoch": 0.7190549563430919, + "grad_norm": 2.5643832683563232, + "learning_rate": 4.25199847735059e-06, + "loss": 1.6705, + "step": 700 + }, + { + "epoch": 0.7190549563430919, + "eval_loss": 1.945617437362671, + "eval_runtime": 54.2778, + "eval_samples_per_second": 18.424, + "eval_steps_per_second": 0.774, + "step": 700 + }, + { + "epoch": 0.7200821777092964, + "grad_norm": 2.7856431007385254, + "learning_rate": 4.250095165588123e-06, + "loss": 1.5833, + "step": 701 + }, + { + "epoch": 0.7211093990755008, + "grad_norm": 2.9881107807159424, + "learning_rate": 4.2481918538256575e-06, + "loss": 1.6438, + "step": 702 + }, + { + "epoch": 0.7221366204417052, + "grad_norm": 2.602938652038574, + "learning_rate": 4.246288542063191e-06, + "loss": 1.6135, + "step": 703 + }, + { + "epoch": 0.7231638418079096, + "grad_norm": 2.997450590133667, + "learning_rate": 4.244385230300724e-06, + "loss": 1.6419, + "step": 704 + }, + { + "epoch": 0.724191063174114, + "grad_norm": 2.5592496395111084, + "learning_rate": 4.242481918538257e-06, + "loss": 1.5886, + "step": 705 + }, + { + "epoch": 0.7252182845403184, + "grad_norm": 2.4073948860168457, + "learning_rate": 4.24057860677579e-06, + "loss": 1.6219, + "step": 706 + }, + { + "epoch": 0.7262455059065228, + "grad_norm": 2.545997381210327, + "learning_rate": 4.2386752950133234e-06, + "loss": 1.6684, + "step": 707 + }, + { + "epoch": 0.7272727272727273, + "grad_norm": 2.468264579772949, + "learning_rate": 4.236771983250857e-06, + "loss": 1.6155, + "step": 708 + }, + { + "epoch": 0.7282999486389317, + "grad_norm": 2.6635043621063232, + "learning_rate": 4.23486867148839e-06, + "loss": 1.5956, + "step": 709 + }, + { + "epoch": 0.7293271700051361, + "grad_norm": 2.8751156330108643, + "learning_rate": 4.2329653597259235e-06, + "loss": 1.6211, + "step": 710 + }, + { + "epoch": 0.7293271700051361, + "eval_loss": 1.9319628477096558, + "eval_runtime": 43.9701, + "eval_samples_per_second": 22.743, + "eval_steps_per_second": 0.955, + "step": 710 + }, + { + "epoch": 0.7303543913713405, + "grad_norm": 2.5999035835266113, + "learning_rate": 4.231062047963457e-06, + "loss": 1.6442, + "step": 711 + }, + { + "epoch": 0.731381612737545, + "grad_norm": 2.611095666885376, + "learning_rate": 4.22915873620099e-06, + "loss": 1.6745, + "step": 712 + }, + { + "epoch": 0.7324088341037494, + "grad_norm": 2.7235333919525146, + "learning_rate": 4.227255424438524e-06, + "loss": 1.6353, + "step": 713 + }, + { + "epoch": 0.7334360554699538, + "grad_norm": 2.453639507293701, + "learning_rate": 4.225352112676057e-06, + "loss": 1.6657, + "step": 714 + }, + { + "epoch": 0.7344632768361582, + "grad_norm": 2.4031429290771484, + "learning_rate": 4.2234488009135895e-06, + "loss": 1.603, + "step": 715 + }, + { + "epoch": 0.7354904982023626, + "grad_norm": 2.631052017211914, + "learning_rate": 4.221545489151123e-06, + "loss": 1.5681, + "step": 716 + }, + { + "epoch": 0.736517719568567, + "grad_norm": 2.518758773803711, + "learning_rate": 4.219642177388656e-06, + "loss": 1.5941, + "step": 717 + }, + { + "epoch": 0.7375449409347714, + "grad_norm": 2.699491500854492, + "learning_rate": 4.2177388656261896e-06, + "loss": 1.6327, + "step": 718 + }, + { + "epoch": 0.7385721623009759, + "grad_norm": 2.749356985092163, + "learning_rate": 4.215835553863723e-06, + "loss": 1.5864, + "step": 719 + }, + { + "epoch": 0.7395993836671803, + "grad_norm": 2.5699031352996826, + "learning_rate": 4.213932242101256e-06, + "loss": 1.5444, + "step": 720 + }, + { + "epoch": 0.7395993836671803, + "eval_loss": 1.9521623849868774, + "eval_runtime": 46.5749, + "eval_samples_per_second": 21.471, + "eval_steps_per_second": 0.902, + "step": 720 + }, + { + "epoch": 0.7406266050333847, + "grad_norm": 3.197425127029419, + "learning_rate": 4.21202893033879e-06, + "loss": 1.6158, + "step": 721 + }, + { + "epoch": 0.7416538263995891, + "grad_norm": 2.619459390640259, + "learning_rate": 4.210125618576323e-06, + "loss": 1.6328, + "step": 722 + }, + { + "epoch": 0.7426810477657936, + "grad_norm": 2.792086601257324, + "learning_rate": 4.208222306813856e-06, + "loss": 1.6333, + "step": 723 + }, + { + "epoch": 0.743708269131998, + "grad_norm": 2.7023141384124756, + "learning_rate": 4.20631899505139e-06, + "loss": 1.6787, + "step": 724 + }, + { + "epoch": 0.7447354904982023, + "grad_norm": 2.6319735050201416, + "learning_rate": 4.204415683288923e-06, + "loss": 1.54, + "step": 725 + }, + { + "epoch": 0.7457627118644068, + "grad_norm": 2.7032439708709717, + "learning_rate": 4.2025123715264565e-06, + "loss": 1.641, + "step": 726 + }, + { + "epoch": 0.7467899332306112, + "grad_norm": 2.9133267402648926, + "learning_rate": 4.20060905976399e-06, + "loss": 1.5901, + "step": 727 + }, + { + "epoch": 0.7478171545968156, + "grad_norm": 2.492403745651245, + "learning_rate": 4.198705748001523e-06, + "loss": 1.5516, + "step": 728 + }, + { + "epoch": 0.74884437596302, + "grad_norm": 2.604520082473755, + "learning_rate": 4.1968024362390566e-06, + "loss": 1.6574, + "step": 729 + }, + { + "epoch": 0.7498715973292245, + "grad_norm": 2.521969795227051, + "learning_rate": 4.19489912447659e-06, + "loss": 1.6099, + "step": 730 + }, + { + "epoch": 0.7498715973292245, + "eval_loss": 1.9449840784072876, + "eval_runtime": 44.9334, + "eval_samples_per_second": 22.255, + "eval_steps_per_second": 0.935, + "step": 730 + }, + { + "epoch": 0.7508988186954288, + "grad_norm": 2.4645657539367676, + "learning_rate": 4.192995812714123e-06, + "loss": 1.6752, + "step": 731 + }, + { + "epoch": 0.7519260400616333, + "grad_norm": 2.7852981090545654, + "learning_rate": 4.191092500951656e-06, + "loss": 1.6176, + "step": 732 + }, + { + "epoch": 0.7529532614278377, + "grad_norm": 2.2817575931549072, + "learning_rate": 4.189189189189189e-06, + "loss": 1.5934, + "step": 733 + }, + { + "epoch": 0.7539804827940421, + "grad_norm": 2.6218173503875732, + "learning_rate": 4.1872858774267225e-06, + "loss": 1.5902, + "step": 734 + }, + { + "epoch": 0.7550077041602465, + "grad_norm": 2.6671013832092285, + "learning_rate": 4.185382565664256e-06, + "loss": 1.657, + "step": 735 + }, + { + "epoch": 0.7560349255264509, + "grad_norm": 2.669757843017578, + "learning_rate": 4.183479253901789e-06, + "loss": 1.5884, + "step": 736 + }, + { + "epoch": 0.7570621468926554, + "grad_norm": 2.7210471630096436, + "learning_rate": 4.181575942139323e-06, + "loss": 1.5558, + "step": 737 + }, + { + "epoch": 0.7580893682588598, + "grad_norm": 2.414276123046875, + "learning_rate": 4.179672630376856e-06, + "loss": 1.6124, + "step": 738 + }, + { + "epoch": 0.7591165896250642, + "grad_norm": 2.4643094539642334, + "learning_rate": 4.177769318614389e-06, + "loss": 1.5904, + "step": 739 + }, + { + "epoch": 0.7601438109912686, + "grad_norm": 2.5447018146514893, + "learning_rate": 4.175866006851923e-06, + "loss": 1.6705, + "step": 740 + }, + { + "epoch": 0.7601438109912686, + "eval_loss": 1.9437954425811768, + "eval_runtime": 49.2702, + "eval_samples_per_second": 20.296, + "eval_steps_per_second": 0.852, + "step": 740 + }, + { + "epoch": 0.7611710323574731, + "grad_norm": 2.41048002243042, + "learning_rate": 4.173962695089456e-06, + "loss": 1.6398, + "step": 741 + }, + { + "epoch": 0.7621982537236774, + "grad_norm": 2.3949906826019287, + "learning_rate": 4.1720593833269894e-06, + "loss": 1.6198, + "step": 742 + }, + { + "epoch": 0.7632254750898819, + "grad_norm": 2.36966609954834, + "learning_rate": 4.170156071564523e-06, + "loss": 1.6275, + "step": 743 + }, + { + "epoch": 0.7642526964560863, + "grad_norm": 2.623347043991089, + "learning_rate": 4.168252759802056e-06, + "loss": 1.7059, + "step": 744 + }, + { + "epoch": 0.7652799178222907, + "grad_norm": 2.384788751602173, + "learning_rate": 4.1663494480395895e-06, + "loss": 1.6015, + "step": 745 + }, + { + "epoch": 0.7663071391884951, + "grad_norm": 2.586009979248047, + "learning_rate": 4.164446136277123e-06, + "loss": 1.6445, + "step": 746 + }, + { + "epoch": 0.7673343605546995, + "grad_norm": 2.396684169769287, + "learning_rate": 4.162542824514656e-06, + "loss": 1.6539, + "step": 747 + }, + { + "epoch": 0.768361581920904, + "grad_norm": 2.416243553161621, + "learning_rate": 4.16063951275219e-06, + "loss": 1.6142, + "step": 748 + }, + { + "epoch": 0.7693888032871083, + "grad_norm": 2.480912208557129, + "learning_rate": 4.158736200989722e-06, + "loss": 1.5768, + "step": 749 + }, + { + "epoch": 0.7704160246533128, + "grad_norm": 2.5516929626464844, + "learning_rate": 4.1568328892272555e-06, + "loss": 1.6074, + "step": 750 + }, + { + "epoch": 0.7704160246533128, + "eval_loss": 1.9409116506576538, + "eval_runtime": 45.3465, + "eval_samples_per_second": 22.052, + "eval_steps_per_second": 0.926, + "step": 750 + }, + { + "epoch": 0.7714432460195172, + "grad_norm": 2.4248714447021484, + "learning_rate": 4.154929577464789e-06, + "loss": 1.6308, + "step": 751 + }, + { + "epoch": 0.7724704673857217, + "grad_norm": 2.520810604095459, + "learning_rate": 4.153026265702322e-06, + "loss": 1.6227, + "step": 752 + }, + { + "epoch": 0.773497688751926, + "grad_norm": 2.5938422679901123, + "learning_rate": 4.1511229539398556e-06, + "loss": 1.5601, + "step": 753 + }, + { + "epoch": 0.7745249101181304, + "grad_norm": 2.4538512229919434, + "learning_rate": 4.149219642177389e-06, + "loss": 1.6006, + "step": 754 + }, + { + "epoch": 0.7755521314843349, + "grad_norm": 2.4936258792877197, + "learning_rate": 4.147316330414922e-06, + "loss": 1.6281, + "step": 755 + }, + { + "epoch": 0.7765793528505393, + "grad_norm": 2.5774738788604736, + "learning_rate": 4.145413018652456e-06, + "loss": 1.5328, + "step": 756 + }, + { + "epoch": 0.7776065742167437, + "grad_norm": 2.3616905212402344, + "learning_rate": 4.143509706889989e-06, + "loss": 1.629, + "step": 757 + }, + { + "epoch": 0.7786337955829481, + "grad_norm": 2.4983086585998535, + "learning_rate": 4.141606395127522e-06, + "loss": 1.6298, + "step": 758 + }, + { + "epoch": 0.7796610169491526, + "grad_norm": 2.4313876628875732, + "learning_rate": 4.139703083365056e-06, + "loss": 1.6113, + "step": 759 + }, + { + "epoch": 0.7806882383153569, + "grad_norm": 2.6290934085845947, + "learning_rate": 4.137799771602588e-06, + "loss": 1.6632, + "step": 760 + }, + { + "epoch": 0.7806882383153569, + "eval_loss": 1.950036644935608, + "eval_runtime": 47.4523, + "eval_samples_per_second": 21.074, + "eval_steps_per_second": 0.885, + "step": 760 + }, + { + "epoch": 0.7817154596815614, + "grad_norm": 2.3637750148773193, + "learning_rate": 4.135896459840122e-06, + "loss": 1.6268, + "step": 761 + }, + { + "epoch": 0.7827426810477658, + "grad_norm": 2.5537843704223633, + "learning_rate": 4.133993148077656e-06, + "loss": 1.5736, + "step": 762 + }, + { + "epoch": 0.7837699024139703, + "grad_norm": 2.4945273399353027, + "learning_rate": 4.132089836315189e-06, + "loss": 1.612, + "step": 763 + }, + { + "epoch": 0.7847971237801746, + "grad_norm": 2.506983757019043, + "learning_rate": 4.1301865245527226e-06, + "loss": 1.6112, + "step": 764 + }, + { + "epoch": 0.785824345146379, + "grad_norm": 2.599721908569336, + "learning_rate": 4.128283212790256e-06, + "loss": 1.5698, + "step": 765 + }, + { + "epoch": 0.7868515665125835, + "grad_norm": 2.4561266899108887, + "learning_rate": 4.1263799010277884e-06, + "loss": 1.5354, + "step": 766 + }, + { + "epoch": 0.7878787878787878, + "grad_norm": 2.445420026779175, + "learning_rate": 4.124476589265322e-06, + "loss": 1.6555, + "step": 767 + }, + { + "epoch": 0.7889060092449923, + "grad_norm": 2.576533079147339, + "learning_rate": 4.122573277502855e-06, + "loss": 1.5915, + "step": 768 + }, + { + "epoch": 0.7899332306111967, + "grad_norm": 2.503164291381836, + "learning_rate": 4.1206699657403885e-06, + "loss": 1.5893, + "step": 769 + }, + { + "epoch": 0.7909604519774012, + "grad_norm": 2.9463603496551514, + "learning_rate": 4.118766653977922e-06, + "loss": 1.6178, + "step": 770 + }, + { + "epoch": 0.7909604519774012, + "eval_loss": 1.9529738426208496, + "eval_runtime": 45.0922, + "eval_samples_per_second": 22.177, + "eval_steps_per_second": 0.931, + "step": 770 + }, + { + "epoch": 0.7919876733436055, + "grad_norm": 2.47148060798645, + "learning_rate": 4.116863342215455e-06, + "loss": 1.5991, + "step": 771 + }, + { + "epoch": 0.79301489470981, + "grad_norm": 2.515717029571533, + "learning_rate": 4.114960030452989e-06, + "loss": 1.5725, + "step": 772 + }, + { + "epoch": 0.7940421160760144, + "grad_norm": 2.5231993198394775, + "learning_rate": 4.113056718690522e-06, + "loss": 1.6145, + "step": 773 + }, + { + "epoch": 0.7950693374422187, + "grad_norm": 2.479978084564209, + "learning_rate": 4.111153406928055e-06, + "loss": 1.5537, + "step": 774 + }, + { + "epoch": 0.7960965588084232, + "grad_norm": 2.5341250896453857, + "learning_rate": 4.109250095165589e-06, + "loss": 1.6419, + "step": 775 + }, + { + "epoch": 0.7971237801746276, + "grad_norm": 2.6277401447296143, + "learning_rate": 4.107346783403122e-06, + "loss": 1.6505, + "step": 776 + }, + { + "epoch": 0.7981510015408321, + "grad_norm": 2.4688351154327393, + "learning_rate": 4.105443471640655e-06, + "loss": 1.5896, + "step": 777 + }, + { + "epoch": 0.7991782229070364, + "grad_norm": 2.4134581089019775, + "learning_rate": 4.103540159878188e-06, + "loss": 1.5788, + "step": 778 + }, + { + "epoch": 0.8002054442732409, + "grad_norm": 2.5090012550354004, + "learning_rate": 4.101636848115721e-06, + "loss": 1.6162, + "step": 779 + }, + { + "epoch": 0.8012326656394453, + "grad_norm": 2.5123302936553955, + "learning_rate": 4.099733536353255e-06, + "loss": 1.5868, + "step": 780 + }, + { + "epoch": 0.8012326656394453, + "eval_loss": 1.9545625448226929, + "eval_runtime": 45.9316, + "eval_samples_per_second": 21.772, + "eval_steps_per_second": 0.914, + "step": 780 + }, + { + "epoch": 0.8022598870056498, + "grad_norm": 2.787181854248047, + "learning_rate": 4.097830224590788e-06, + "loss": 1.612, + "step": 781 + }, + { + "epoch": 0.8032871083718541, + "grad_norm": 2.414426326751709, + "learning_rate": 4.095926912828322e-06, + "loss": 1.6908, + "step": 782 + }, + { + "epoch": 0.8043143297380585, + "grad_norm": 2.452697992324829, + "learning_rate": 4.094023601065855e-06, + "loss": 1.6575, + "step": 783 + }, + { + "epoch": 0.805341551104263, + "grad_norm": 2.4377169609069824, + "learning_rate": 4.092120289303388e-06, + "loss": 1.6016, + "step": 784 + }, + { + "epoch": 0.8063687724704673, + "grad_norm": 2.369154214859009, + "learning_rate": 4.0902169775409215e-06, + "loss": 1.6133, + "step": 785 + }, + { + "epoch": 0.8073959938366718, + "grad_norm": 2.261869430541992, + "learning_rate": 4.088313665778455e-06, + "loss": 1.5555, + "step": 786 + }, + { + "epoch": 0.8084232152028762, + "grad_norm": 2.7058632373809814, + "learning_rate": 4.086410354015988e-06, + "loss": 1.5957, + "step": 787 + }, + { + "epoch": 0.8094504365690807, + "grad_norm": 2.3907508850097656, + "learning_rate": 4.0845070422535216e-06, + "loss": 1.5676, + "step": 788 + }, + { + "epoch": 0.810477657935285, + "grad_norm": 2.510284900665283, + "learning_rate": 4.082603730491055e-06, + "loss": 1.5872, + "step": 789 + }, + { + "epoch": 0.8115048793014895, + "grad_norm": 2.590810775756836, + "learning_rate": 4.080700418728588e-06, + "loss": 1.6315, + "step": 790 + }, + { + "epoch": 0.8115048793014895, + "eval_loss": 1.9573780298233032, + "eval_runtime": 44.5677, + "eval_samples_per_second": 22.438, + "eval_steps_per_second": 0.942, + "step": 790 + }, + { + "epoch": 0.8125321006676939, + "grad_norm": 2.466134786605835, + "learning_rate": 4.078797106966122e-06, + "loss": 1.6018, + "step": 791 + }, + { + "epoch": 0.8135593220338984, + "grad_norm": 2.474653482437134, + "learning_rate": 4.076893795203655e-06, + "loss": 1.5101, + "step": 792 + }, + { + "epoch": 0.8145865434001027, + "grad_norm": 2.4463517665863037, + "learning_rate": 4.074990483441188e-06, + "loss": 1.5686, + "step": 793 + }, + { + "epoch": 0.8156137647663071, + "grad_norm": 2.5148985385894775, + "learning_rate": 4.073087171678721e-06, + "loss": 1.6544, + "step": 794 + }, + { + "epoch": 0.8166409861325116, + "grad_norm": 2.6183927059173584, + "learning_rate": 4.071183859916254e-06, + "loss": 1.6277, + "step": 795 + }, + { + "epoch": 0.8176682074987159, + "grad_norm": 2.438427686691284, + "learning_rate": 4.069280548153788e-06, + "loss": 1.6002, + "step": 796 + }, + { + "epoch": 0.8186954288649204, + "grad_norm": 2.662855863571167, + "learning_rate": 4.067377236391321e-06, + "loss": 1.5594, + "step": 797 + }, + { + "epoch": 0.8197226502311248, + "grad_norm": 2.64789080619812, + "learning_rate": 4.065473924628854e-06, + "loss": 1.6007, + "step": 798 + }, + { + "epoch": 0.8207498715973293, + "grad_norm": 2.5152997970581055, + "learning_rate": 4.063570612866388e-06, + "loss": 1.6366, + "step": 799 + }, + { + "epoch": 0.8217770929635336, + "grad_norm": 2.468672037124634, + "learning_rate": 4.061667301103921e-06, + "loss": 1.6313, + "step": 800 + }, + { + "epoch": 0.8217770929635336, + "eval_loss": 1.95821213722229, + "eval_runtime": 45.9362, + "eval_samples_per_second": 21.769, + "eval_steps_per_second": 0.914, + "step": 800 + }, + { + "epoch": 0.8228043143297381, + "grad_norm": 2.440307855606079, + "learning_rate": 4.0597639893414544e-06, + "loss": 1.5993, + "step": 801 + }, + { + "epoch": 0.8238315356959425, + "grad_norm": 2.6519880294799805, + "learning_rate": 4.057860677578988e-06, + "loss": 1.6106, + "step": 802 + }, + { + "epoch": 0.8248587570621468, + "grad_norm": 2.4224767684936523, + "learning_rate": 4.055957365816521e-06, + "loss": 1.6247, + "step": 803 + }, + { + "epoch": 0.8258859784283513, + "grad_norm": 2.346858263015747, + "learning_rate": 4.0540540540540545e-06, + "loss": 1.5699, + "step": 804 + }, + { + "epoch": 0.8269131997945557, + "grad_norm": 2.577711582183838, + "learning_rate": 4.052150742291588e-06, + "loss": 1.5665, + "step": 805 + }, + { + "epoch": 0.8279404211607602, + "grad_norm": 2.604501724243164, + "learning_rate": 4.050247430529121e-06, + "loss": 1.561, + "step": 806 + }, + { + "epoch": 0.8289676425269645, + "grad_norm": 2.534820795059204, + "learning_rate": 4.048344118766655e-06, + "loss": 1.6391, + "step": 807 + }, + { + "epoch": 0.829994863893169, + "grad_norm": 2.496765613555908, + "learning_rate": 4.046440807004188e-06, + "loss": 1.6338, + "step": 808 + }, + { + "epoch": 0.8310220852593734, + "grad_norm": 2.587185859680176, + "learning_rate": 4.044537495241721e-06, + "loss": 1.6009, + "step": 809 + }, + { + "epoch": 0.8320493066255779, + "grad_norm": 2.7336575984954834, + "learning_rate": 4.042634183479255e-06, + "loss": 1.5911, + "step": 810 + }, + { + "epoch": 0.8320493066255779, + "eval_loss": 1.951606273651123, + "eval_runtime": 42.4435, + "eval_samples_per_second": 23.561, + "eval_steps_per_second": 0.99, + "step": 810 + }, + { + "epoch": 0.8330765279917822, + "grad_norm": 2.5524845123291016, + "learning_rate": 4.040730871716787e-06, + "loss": 1.6406, + "step": 811 + }, + { + "epoch": 0.8341037493579867, + "grad_norm": 2.4861226081848145, + "learning_rate": 4.038827559954321e-06, + "loss": 1.6092, + "step": 812 + }, + { + "epoch": 0.8351309707241911, + "grad_norm": 2.668187141418457, + "learning_rate": 4.036924248191854e-06, + "loss": 1.6529, + "step": 813 + }, + { + "epoch": 0.8361581920903954, + "grad_norm": 2.5616471767425537, + "learning_rate": 4.035020936429387e-06, + "loss": 1.6253, + "step": 814 + }, + { + "epoch": 0.8371854134565999, + "grad_norm": 2.4261975288391113, + "learning_rate": 4.033117624666921e-06, + "loss": 1.5955, + "step": 815 + }, + { + "epoch": 0.8382126348228043, + "grad_norm": 2.4947080612182617, + "learning_rate": 4.031214312904454e-06, + "loss": 1.5686, + "step": 816 + }, + { + "epoch": 0.8392398561890088, + "grad_norm": 2.747781276702881, + "learning_rate": 4.029311001141987e-06, + "loss": 1.5642, + "step": 817 + }, + { + "epoch": 0.8402670775552131, + "grad_norm": 2.4762001037597656, + "learning_rate": 4.027407689379521e-06, + "loss": 1.5301, + "step": 818 + }, + { + "epoch": 0.8412942989214176, + "grad_norm": 2.6253550052642822, + "learning_rate": 4.025504377617054e-06, + "loss": 1.5885, + "step": 819 + }, + { + "epoch": 0.842321520287622, + "grad_norm": 2.6974875926971436, + "learning_rate": 4.0236010658545875e-06, + "loss": 1.5656, + "step": 820 + }, + { + "epoch": 0.842321520287622, + "eval_loss": 1.9420732259750366, + "eval_runtime": 46.9525, + "eval_samples_per_second": 21.298, + "eval_steps_per_second": 0.895, + "step": 820 + }, + { + "epoch": 0.8433487416538265, + "grad_norm": 2.638587474822998, + "learning_rate": 4.02169775409212e-06, + "loss": 1.6081, + "step": 821 + }, + { + "epoch": 0.8443759630200308, + "grad_norm": 2.963974714279175, + "learning_rate": 4.019794442329653e-06, + "loss": 1.6851, + "step": 822 + }, + { + "epoch": 0.8454031843862352, + "grad_norm": 2.4657533168792725, + "learning_rate": 4.0178911305671876e-06, + "loss": 1.5396, + "step": 823 + }, + { + "epoch": 0.8464304057524397, + "grad_norm": 2.5373876094818115, + "learning_rate": 4.015987818804721e-06, + "loss": 1.6186, + "step": 824 + }, + { + "epoch": 0.847457627118644, + "grad_norm": 2.5873284339904785, + "learning_rate": 4.014084507042254e-06, + "loss": 1.5618, + "step": 825 + }, + { + "epoch": 0.8484848484848485, + "grad_norm": 2.4308598041534424, + "learning_rate": 4.012181195279788e-06, + "loss": 1.5399, + "step": 826 + }, + { + "epoch": 0.8495120698510529, + "grad_norm": 2.5274946689605713, + "learning_rate": 4.010277883517321e-06, + "loss": 1.6375, + "step": 827 + }, + { + "epoch": 0.8505392912172574, + "grad_norm": 2.4370181560516357, + "learning_rate": 4.0083745717548535e-06, + "loss": 1.6313, + "step": 828 + }, + { + "epoch": 0.8515665125834617, + "grad_norm": 2.4524898529052734, + "learning_rate": 4.006471259992387e-06, + "loss": 1.5865, + "step": 829 + }, + { + "epoch": 0.8525937339496662, + "grad_norm": 2.5587961673736572, + "learning_rate": 4.00456794822992e-06, + "loss": 1.608, + "step": 830 + }, + { + "epoch": 0.8525937339496662, + "eval_loss": 1.94539475440979, + "eval_runtime": 49.4543, + "eval_samples_per_second": 20.221, + "eval_steps_per_second": 0.849, + "step": 830 + }, + { + "epoch": 0.8536209553158706, + "grad_norm": 2.512253761291504, + "learning_rate": 4.002664636467454e-06, + "loss": 1.6607, + "step": 831 + }, + { + "epoch": 0.8546481766820749, + "grad_norm": 2.438451051712036, + "learning_rate": 4.000761324704987e-06, + "loss": 1.6153, + "step": 832 + }, + { + "epoch": 0.8556753980482794, + "grad_norm": 2.5550496578216553, + "learning_rate": 3.99885801294252e-06, + "loss": 1.5489, + "step": 833 + }, + { + "epoch": 0.8567026194144838, + "grad_norm": 2.559732437133789, + "learning_rate": 3.996954701180054e-06, + "loss": 1.6045, + "step": 834 + }, + { + "epoch": 0.8577298407806883, + "grad_norm": 2.317692995071411, + "learning_rate": 3.995051389417587e-06, + "loss": 1.5316, + "step": 835 + }, + { + "epoch": 0.8587570621468926, + "grad_norm": 2.3742363452911377, + "learning_rate": 3.9931480776551204e-06, + "loss": 1.553, + "step": 836 + }, + { + "epoch": 0.8597842835130971, + "grad_norm": 2.5645551681518555, + "learning_rate": 3.991244765892654e-06, + "loss": 1.6558, + "step": 837 + }, + { + "epoch": 0.8608115048793015, + "grad_norm": 2.494830846786499, + "learning_rate": 3.989341454130186e-06, + "loss": 1.5713, + "step": 838 + }, + { + "epoch": 0.861838726245506, + "grad_norm": 2.416020393371582, + "learning_rate": 3.98743814236772e-06, + "loss": 1.5383, + "step": 839 + }, + { + "epoch": 0.8628659476117103, + "grad_norm": 2.6063015460968018, + "learning_rate": 3.985534830605253e-06, + "loss": 1.5831, + "step": 840 + }, + { + "epoch": 0.8628659476117103, + "eval_loss": 1.9441509246826172, + "eval_runtime": 46.0083, + "eval_samples_per_second": 21.735, + "eval_steps_per_second": 0.913, + "step": 840 + }, + { + "epoch": 0.8638931689779148, + "grad_norm": 2.5771663188934326, + "learning_rate": 3.983631518842786e-06, + "loss": 1.6077, + "step": 841 + }, + { + "epoch": 0.8649203903441192, + "grad_norm": 2.573119878768921, + "learning_rate": 3.98172820708032e-06, + "loss": 1.6301, + "step": 842 + }, + { + "epoch": 0.8659476117103235, + "grad_norm": 2.41396164894104, + "learning_rate": 3.979824895317854e-06, + "loss": 1.5758, + "step": 843 + }, + { + "epoch": 0.866974833076528, + "grad_norm": 2.532883882522583, + "learning_rate": 3.977921583555387e-06, + "loss": 1.5525, + "step": 844 + }, + { + "epoch": 0.8680020544427324, + "grad_norm": 2.669311285018921, + "learning_rate": 3.97601827179292e-06, + "loss": 1.6183, + "step": 845 + }, + { + "epoch": 0.8690292758089369, + "grad_norm": 2.366861343383789, + "learning_rate": 3.974114960030453e-06, + "loss": 1.5551, + "step": 846 + }, + { + "epoch": 0.8700564971751412, + "grad_norm": 2.3896048069000244, + "learning_rate": 3.9722116482679866e-06, + "loss": 1.5894, + "step": 847 + }, + { + "epoch": 0.8710837185413457, + "grad_norm": 2.6509673595428467, + "learning_rate": 3.97030833650552e-06, + "loss": 1.5813, + "step": 848 + }, + { + "epoch": 0.8721109399075501, + "grad_norm": 2.531726121902466, + "learning_rate": 3.968405024743053e-06, + "loss": 1.5899, + "step": 849 + }, + { + "epoch": 0.8731381612737545, + "grad_norm": 2.4117019176483154, + "learning_rate": 3.966501712980587e-06, + "loss": 1.5117, + "step": 850 + }, + { + "epoch": 0.8731381612737545, + "eval_loss": 1.9224250316619873, + "eval_runtime": 47.7854, + "eval_samples_per_second": 20.927, + "eval_steps_per_second": 0.879, + "step": 850 + }, + { + "epoch": 0.8741653826399589, + "grad_norm": 2.3789749145507812, + "learning_rate": 3.96459840121812e-06, + "loss": 1.5708, + "step": 851 + }, + { + "epoch": 0.8751926040061633, + "grad_norm": 2.386664867401123, + "learning_rate": 3.962695089455653e-06, + "loss": 1.5734, + "step": 852 + }, + { + "epoch": 0.8762198253723678, + "grad_norm": 2.4788997173309326, + "learning_rate": 3.960791777693187e-06, + "loss": 1.6315, + "step": 853 + }, + { + "epoch": 0.8772470467385721, + "grad_norm": 2.630183219909668, + "learning_rate": 3.95888846593072e-06, + "loss": 1.614, + "step": 854 + }, + { + "epoch": 0.8782742681047766, + "grad_norm": 2.3957736492156982, + "learning_rate": 3.956985154168253e-06, + "loss": 1.61, + "step": 855 + }, + { + "epoch": 0.879301489470981, + "grad_norm": 2.652036428451538, + "learning_rate": 3.955081842405786e-06, + "loss": 1.6463, + "step": 856 + }, + { + "epoch": 0.8803287108371854, + "grad_norm": 2.579867124557495, + "learning_rate": 3.953178530643319e-06, + "loss": 1.6624, + "step": 857 + }, + { + "epoch": 0.8813559322033898, + "grad_norm": 2.5118343830108643, + "learning_rate": 3.951275218880853e-06, + "loss": 1.6816, + "step": 858 + }, + { + "epoch": 0.8823831535695943, + "grad_norm": 2.5963029861450195, + "learning_rate": 3.949371907118386e-06, + "loss": 1.5296, + "step": 859 + }, + { + "epoch": 0.8834103749357987, + "grad_norm": 2.571241855621338, + "learning_rate": 3.9474685953559194e-06, + "loss": 1.6362, + "step": 860 + }, + { + "epoch": 0.8834103749357987, + "eval_loss": 1.931492567062378, + "eval_runtime": 52.2417, + "eval_samples_per_second": 19.142, + "eval_steps_per_second": 0.804, + "step": 860 + }, + { + "epoch": 0.884437596302003, + "grad_norm": 2.518036365509033, + "learning_rate": 3.945565283593453e-06, + "loss": 1.5865, + "step": 861 + }, + { + "epoch": 0.8854648176682075, + "grad_norm": 3.14587664604187, + "learning_rate": 3.943661971830986e-06, + "loss": 1.586, + "step": 862 + }, + { + "epoch": 0.8864920390344119, + "grad_norm": 2.686350107192993, + "learning_rate": 3.9417586600685195e-06, + "loss": 1.6385, + "step": 863 + }, + { + "epoch": 0.8875192604006163, + "grad_norm": 3.0324997901916504, + "learning_rate": 3.939855348306053e-06, + "loss": 1.6199, + "step": 864 + }, + { + "epoch": 0.8885464817668207, + "grad_norm": 2.6129438877105713, + "learning_rate": 3.937952036543586e-06, + "loss": 1.6049, + "step": 865 + }, + { + "epoch": 0.8895737031330252, + "grad_norm": 2.795729160308838, + "learning_rate": 3.93604872478112e-06, + "loss": 1.6027, + "step": 866 + }, + { + "epoch": 0.8906009244992296, + "grad_norm": 3.2352302074432373, + "learning_rate": 3.934145413018653e-06, + "loss": 1.6063, + "step": 867 + }, + { + "epoch": 0.891628145865434, + "grad_norm": 2.5050623416900635, + "learning_rate": 3.932242101256186e-06, + "loss": 1.5622, + "step": 868 + }, + { + "epoch": 0.8926553672316384, + "grad_norm": 2.8998777866363525, + "learning_rate": 3.93033878949372e-06, + "loss": 1.5729, + "step": 869 + }, + { + "epoch": 0.8936825885978429, + "grad_norm": 2.60239577293396, + "learning_rate": 3.928435477731253e-06, + "loss": 1.5377, + "step": 870 + }, + { + "epoch": 0.8936825885978429, + "eval_loss": 1.9311563968658447, + "eval_runtime": 46.8098, + "eval_samples_per_second": 21.363, + "eval_steps_per_second": 0.897, + "step": 870 + }, + { + "epoch": 0.8947098099640473, + "grad_norm": 2.5061111450195312, + "learning_rate": 3.9265321659687864e-06, + "loss": 1.6216, + "step": 871 + }, + { + "epoch": 0.8957370313302516, + "grad_norm": 2.7476489543914795, + "learning_rate": 3.924628854206319e-06, + "loss": 1.6357, + "step": 872 + }, + { + "epoch": 0.8967642526964561, + "grad_norm": 2.605536937713623, + "learning_rate": 3.922725542443852e-06, + "loss": 1.5983, + "step": 873 + }, + { + "epoch": 0.8977914740626605, + "grad_norm": 2.6024532318115234, + "learning_rate": 3.920822230681386e-06, + "loss": 1.5571, + "step": 874 + }, + { + "epoch": 0.8988186954288649, + "grad_norm": 2.7357826232910156, + "learning_rate": 3.918918918918919e-06, + "loss": 1.5802, + "step": 875 + }, + { + "epoch": 0.8998459167950693, + "grad_norm": 2.6406939029693604, + "learning_rate": 3.917015607156452e-06, + "loss": 1.6046, + "step": 876 + }, + { + "epoch": 0.9008731381612738, + "grad_norm": 2.4611780643463135, + "learning_rate": 3.915112295393986e-06, + "loss": 1.5803, + "step": 877 + }, + { + "epoch": 0.9019003595274782, + "grad_norm": 3.0608181953430176, + "learning_rate": 3.913208983631519e-06, + "loss": 1.6655, + "step": 878 + }, + { + "epoch": 0.9029275808936826, + "grad_norm": 2.482142448425293, + "learning_rate": 3.9113056718690525e-06, + "loss": 1.6444, + "step": 879 + }, + { + "epoch": 0.903954802259887, + "grad_norm": 2.4936728477478027, + "learning_rate": 3.909402360106586e-06, + "loss": 1.5596, + "step": 880 + }, + { + "epoch": 0.903954802259887, + "eval_loss": 1.933276653289795, + "eval_runtime": 48.0836, + "eval_samples_per_second": 20.797, + "eval_steps_per_second": 0.873, + "step": 880 + }, + { + "epoch": 0.9049820236260914, + "grad_norm": 2.7802248001098633, + "learning_rate": 3.907499048344119e-06, + "loss": 1.6133, + "step": 881 + }, + { + "epoch": 0.9060092449922958, + "grad_norm": 2.486586093902588, + "learning_rate": 3.905595736581652e-06, + "loss": 1.623, + "step": 882 + }, + { + "epoch": 0.9070364663585002, + "grad_norm": 2.477555751800537, + "learning_rate": 3.903692424819186e-06, + "loss": 1.6219, + "step": 883 + }, + { + "epoch": 0.9080636877247047, + "grad_norm": 2.5033347606658936, + "learning_rate": 3.901789113056719e-06, + "loss": 1.5612, + "step": 884 + }, + { + "epoch": 0.9090909090909091, + "grad_norm": 2.410464286804199, + "learning_rate": 3.899885801294253e-06, + "loss": 1.6243, + "step": 885 + }, + { + "epoch": 0.9101181304571135, + "grad_norm": 2.451402425765991, + "learning_rate": 3.897982489531786e-06, + "loss": 1.5961, + "step": 886 + }, + { + "epoch": 0.9111453518233179, + "grad_norm": 2.439730167388916, + "learning_rate": 3.896079177769319e-06, + "loss": 1.6471, + "step": 887 + }, + { + "epoch": 0.9121725731895224, + "grad_norm": 2.527719259262085, + "learning_rate": 3.894175866006853e-06, + "loss": 1.6262, + "step": 888 + }, + { + "epoch": 0.9131997945557268, + "grad_norm": 2.612992763519287, + "learning_rate": 3.892272554244385e-06, + "loss": 1.6533, + "step": 889 + }, + { + "epoch": 0.9142270159219312, + "grad_norm": 2.509989023208618, + "learning_rate": 3.890369242481919e-06, + "loss": 1.5596, + "step": 890 + }, + { + "epoch": 0.9142270159219312, + "eval_loss": 1.9224061965942383, + "eval_runtime": 45.5462, + "eval_samples_per_second": 21.956, + "eval_steps_per_second": 0.922, + "step": 890 + }, + { + "epoch": 0.9152542372881356, + "grad_norm": 2.3849823474884033, + "learning_rate": 3.888465930719452e-06, + "loss": 1.6336, + "step": 891 + }, + { + "epoch": 0.91628145865434, + "grad_norm": 2.4989709854125977, + "learning_rate": 3.886562618956985e-06, + "loss": 1.5733, + "step": 892 + }, + { + "epoch": 0.9173086800205444, + "grad_norm": 2.4675447940826416, + "learning_rate": 3.884659307194519e-06, + "loss": 1.5818, + "step": 893 + }, + { + "epoch": 0.9183359013867488, + "grad_norm": 2.3105645179748535, + "learning_rate": 3.882755995432052e-06, + "loss": 1.5251, + "step": 894 + }, + { + "epoch": 0.9193631227529533, + "grad_norm": 2.5410287380218506, + "learning_rate": 3.8808526836695854e-06, + "loss": 1.6089, + "step": 895 + }, + { + "epoch": 0.9203903441191577, + "grad_norm": 2.617091655731201, + "learning_rate": 3.878949371907119e-06, + "loss": 1.5631, + "step": 896 + }, + { + "epoch": 0.9214175654853621, + "grad_norm": 2.4979820251464844, + "learning_rate": 3.877046060144652e-06, + "loss": 1.5577, + "step": 897 + }, + { + "epoch": 0.9224447868515665, + "grad_norm": 2.4857399463653564, + "learning_rate": 3.8751427483821855e-06, + "loss": 1.56, + "step": 898 + }, + { + "epoch": 0.923472008217771, + "grad_norm": 2.456709146499634, + "learning_rate": 3.873239436619718e-06, + "loss": 1.594, + "step": 899 + }, + { + "epoch": 0.9244992295839753, + "grad_norm": 2.355168342590332, + "learning_rate": 3.871336124857251e-06, + "loss": 1.5371, + "step": 900 + }, + { + "epoch": 0.9244992295839753, + "eval_loss": 1.9357926845550537, + "eval_runtime": 44.4044, + "eval_samples_per_second": 22.52, + "eval_steps_per_second": 0.946, + "step": 900 + } + ], + "logging_steps": 1, + "max_steps": 2919, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 100, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 2.805578468425728e+17, + "train_batch_size": 3, + "trial_name": null, + "trial_params": null +} diff --git a/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/zero_to_fp32.py b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/zero_to_fp32.py new file mode 100644 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/babylm_reverse_partial_10M_seed0/runs/checkpoint-900/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters)