Upload folder using huggingface_hub

sft_output_simpleQA/checkpoint-2508/config.json

@@ -0,0 +1,55 @@
+{
+  "architectures": [
+    "MptForCausalLM"
+  ],
+  "attn_config": {
+    "alibi": true,
+    "alibi_bias_max": 8,
+    "attn_impl": "torch",
+    "attn_pdrop": 0.0,
+    "attn_type": "multihead_attention",
+    "attn_uses_sequence_id": false,
+    "clip_qkv": null,
+    "model_type": "",
+    "prefix_lm": false,
+    "qk_ln": false,
+    "softmax_scale": null,
+    "torch_dtype": "bfloat16"
+  },
+  "auto_map": {
+    "AutoConfig": "vinai/PhoGPT-4B-Chat--configuration_mpt.MPTConfig",
+    "AutoModelForCausalLM": "vinai/PhoGPT-4B-Chat--modeling_mpt.MPTForCausalLM"
+  },
+  "d_model": 3072,
+  "emb_pdrop": 0.0,
+  "embedding_fraction": 1.0,
+  "expansion_ratio": 4,
+  "init_config": {
+    "emb_init_std": null,
+    "emb_init_uniform_lim": null,
+    "fan_mode": "fan_in",
+    "init_div_is_residual": true,
+    "init_gain": 0.0,
+    "init_nonlinearity": "relu",
+    "init_std": null,
+    "name": "kaiming_normal_",
+    "verbose": 0
+  },
+  "init_device": "cpu",
+  "initializer_range": 0.02,
+  "layer_norm_epsilon": 1e-05,
+  "learned_pos_emb": true,
+  "logit_scale": null,
+  "max_seq_len": 8192,
+  "model_type": "mpt",
+  "n_heads": 24,
+  "n_layers": 32,
+  "no_bias": false,
+  "norm_type": "low_precision_layernorm",
+  "resid_pdrop": 0.0,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.51.3",
+  "use_cache": false,
+  "verbose": 0,
+  "vocab_size": 20480
+}

sft_output_simpleQA/checkpoint-2508/generation_config.json

@@ -0,0 +1,5 @@
+{
+  "_from_model_config": true,
+  "transformers_version": "4.51.3",
+  "use_cache": false
+}

sft_output_simpleQA/checkpoint-2508/model-00001-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:914f85135733d4c80639657e3889ebaa438af4bd2ddda22e0b4afde39183028a
+size 4957952800

sft_output_simpleQA/checkpoint-2508/model-00002-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aedaba4b108fa3668dfcc6736f0b71c640c3acab1674ef9ab79e84ed15b7c55b
+size 2416055432

sft_output_simpleQA/checkpoint-2508/model.safetensors.index.json

@@ -0,0 +1,201 @@
+{
+  "metadata": {
+    "total_size": 7373985792
+  },
+  "weight_map": {
+    "transformer.blocks.0.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.0.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.0.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.0.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.0.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.0.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.1.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.1.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.1.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.1.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.1.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.1.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.10.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.10.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.10.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.10.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.10.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.10.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.11.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.11.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.11.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.11.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.11.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.11.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.12.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.12.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.12.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.12.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.12.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.12.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.13.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.13.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.13.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.13.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.13.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.13.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.14.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.14.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.14.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.14.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.14.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.14.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.15.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.15.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.15.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.15.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.15.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.15.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.16.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.16.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.16.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.16.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.16.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.16.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.17.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.17.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.17.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.17.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.17.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.17.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.18.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.18.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.18.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.18.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.18.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.18.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.19.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.19.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.19.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.19.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.19.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.19.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.2.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.2.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.2.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.2.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.2.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.2.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.20.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.20.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.20.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.20.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.20.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.20.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.21.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.21.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.21.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.21.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.21.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.21.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.22.attn.Wqkv.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.22.attn.out_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.22.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.22.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.22.norm_1.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.22.norm_2.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.23.attn.Wqkv.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.23.attn.out_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.23.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.23.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.23.norm_1.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.23.norm_2.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.24.attn.Wqkv.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.24.attn.out_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.24.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.24.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.24.norm_1.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.24.norm_2.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.25.attn.Wqkv.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.25.attn.out_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.25.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.25.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.25.norm_1.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.25.norm_2.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.26.attn.Wqkv.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.26.attn.out_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.26.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.26.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.26.norm_1.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.26.norm_2.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.27.attn.Wqkv.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.27.attn.out_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.27.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.27.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.27.norm_1.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.27.norm_2.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.28.attn.Wqkv.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.28.attn.out_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.28.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.28.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.28.norm_1.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.28.norm_2.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.29.attn.Wqkv.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.29.attn.out_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.29.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.29.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.29.norm_1.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.29.norm_2.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.3.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.3.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.3.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.3.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.3.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.3.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.30.attn.Wqkv.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.30.attn.out_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.30.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.30.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.30.norm_1.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.30.norm_2.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.31.attn.Wqkv.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.31.attn.out_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.31.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.31.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.31.norm_1.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.31.norm_2.weight": "model-00002-of-00002.safetensors",
+    "transformer.blocks.4.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.4.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.4.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.4.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.4.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.4.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.5.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.5.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.5.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.5.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.5.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.5.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.6.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.6.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.6.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.6.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.6.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.6.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.7.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.7.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.7.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.7.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.7.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.7.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.8.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.8.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.8.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.8.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.8.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.8.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.9.attn.Wqkv.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.9.attn.out_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.9.ffn.down_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.9.ffn.up_proj.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.9.norm_1.weight": "model-00001-of-00002.safetensors",
+    "transformer.blocks.9.norm_2.weight": "model-00001-of-00002.safetensors",
+    "transformer.norm_f.weight": "model-00002-of-00002.safetensors",
+    "transformer.wte.weight": "model-00001-of-00002.safetensors"
+  }
+}

sft_output_simpleQA/checkpoint-2508/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2ba83ee07ad10e54499e2a884f0b1967a31dbbc0c9e5f43b5d41d1de6d93f9f
+size 14748140538

sft_output_simpleQA/checkpoint-2508/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95b6047bd8cc6f4cdf7c46dea47edb8e542435510070c6cd1e0a7d9ccf5fd7da
+size 14244

sft_output_simpleQA/checkpoint-2508/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb33588f8269547605dcf71e9d03456d8b26f2f7b5d096173f582d32bf3dd218
+size 1064

sft_output_simpleQA/checkpoint-2508/special_tokens_map.json

@@ -0,0 +1,30 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

sft_output_simpleQA/checkpoint-2508/tokenizer_config.json

@@ -0,0 +1,49 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<s>",
+  "chat_template": "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% for message in messages %}{% if message['role'] == 'user' and loop.first %}{{ '### Câu hỏi: ' + message['content'].strip() }}{% elif message['role'] == 'user' %}{{ '\n### Câu hỏi: ' + message['content'].strip() }}{% elif message['role'] == 'assistant' %}{{ '\n### Trả lời: ' + message['content'] + eos_token }}{% endif %}{% if loop.last %}{% if message['role'] == 'user' and add_generation_prompt %}{{ '\n### Trả lời:' }}{% endif %}{% endif %}{% endfor %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "extra_special_tokens": {},
+  "merges_file": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "tokenizer_class": "BloomTokenizer",
+  "unk_token": "<unk>",
+  "vocab_file": null
+}

sft_output_simpleQA/checkpoint-2508/trainer_state.json

@@ -0,0 +1,2314 @@
+{
+  "best_global_step": 2508,
+  "best_metric": 7.677623271942139,
+  "best_model_checkpoint": "./sft_output_simpleQA/checkpoint-2508",
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 2508,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.011961722488038277,
+      "grad_norm": 82.0,
+      "learning_rate": 4.982057416267943e-05,
+      "loss": 35.9519,
+      "mean_token_accuracy": 0.00015143249183893203,
+      "num_tokens": 16296.0,
+      "step": 10
+    },
+    {
+      "epoch": 0.023923444976076555,
+      "grad_norm": 88.0,
+      "learning_rate": 4.962121212121213e-05,
+      "loss": 34.0394,
+      "mean_token_accuracy": 0.00014820520300418139,
+      "num_tokens": 30078.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.03588516746411483,
+      "grad_norm": 99.0,
+      "learning_rate": 4.9421850079744817e-05,
+      "loss": 32.9064,
+      "mean_token_accuracy": 0.0004058516991790384,
+      "num_tokens": 45602.0,
+      "step": 30
+    },
+    {
+      "epoch": 0.04784688995215311,
+      "grad_norm": 87.5,
+      "learning_rate": 4.922248803827751e-05,
+      "loss": 31.4831,
+      "mean_token_accuracy": 0.00040118659380823375,
+      "num_tokens": 58420.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.05980861244019139,
+      "grad_norm": 83.0,
+      "learning_rate": 4.902312599681021e-05,
+      "loss": 29.9708,
+      "mean_token_accuracy": 0.00044854874722659587,
+      "num_tokens": 75126.0,
+      "step": 50
+    },
+    {
+      "epoch": 0.07177033492822966,
+      "grad_norm": 78.0,
+      "learning_rate": 4.8823763955342905e-05,
+      "loss": 28.5442,
+      "mean_token_accuracy": 0.00010988024296239018,
+      "num_tokens": 92306.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.08373205741626795,
+      "grad_norm": 78.5,
+      "learning_rate": 4.86244019138756e-05,
+      "loss": 27.7258,
+      "mean_token_accuracy": 0.0008938259619753808,
+      "num_tokens": 108533.0,
+      "step": 70
+    },
+    {
+      "epoch": 0.09569377990430622,
+      "grad_norm": 75.5,
+      "learning_rate": 4.84250398724083e-05,
+      "loss": 26.4549,
+      "mean_token_accuracy": 0.0015082828351296485,
+      "num_tokens": 123975.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.1076555023923445,
+      "grad_norm": 92.5,
+      "learning_rate": 4.8225677830940993e-05,
+      "loss": 25.4496,
+      "mean_token_accuracy": 0.0011613189650233834,
+      "num_tokens": 138192.0,
+      "step": 90
+    },
+    {
+      "epoch": 0.11961722488038277,
+      "grad_norm": 69.0,
+      "learning_rate": 4.802631578947368e-05,
+      "loss": 24.3682,
+      "mean_token_accuracy": 0.0030715125263668596,
+      "num_tokens": 155705.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.13157894736842105,
+      "grad_norm": 68.5,
+      "learning_rate": 4.782695374800638e-05,
+      "loss": 22.8859,
+      "mean_token_accuracy": 0.0022452569246524945,
+      "num_tokens": 174596.0,
+      "step": 110
+    },
+    {
+      "epoch": 0.14354066985645933,
+      "grad_norm": 77.5,
+      "learning_rate": 4.7627591706539075e-05,
+      "loss": 22.18,
+      "mean_token_accuracy": 0.002278627915075049,
+      "num_tokens": 189101.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.15550239234449761,
+      "grad_norm": 61.75,
+      "learning_rate": 4.742822966507177e-05,
+      "loss": 21.2046,
+      "mean_token_accuracy": 0.0015775824780575932,
+      "num_tokens": 205871.0,
+      "step": 130
+    },
+    {
+      "epoch": 0.1674641148325359,
+      "grad_norm": 60.25,
+      "learning_rate": 4.722886762360447e-05,
+      "loss": 20.4473,
+      "mean_token_accuracy": 0.0028540739440359175,
+      "num_tokens": 221139.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.17942583732057416,
+      "grad_norm": 50.75,
+      "learning_rate": 4.7029505582137164e-05,
+      "loss": 19.1652,
+      "mean_token_accuracy": 0.0029712042480241506,
+      "num_tokens": 239226.0,
+      "step": 150
+    },
+    {
+      "epoch": 0.19138755980861244,
+      "grad_norm": 58.5,
+      "learning_rate": 4.683014354066986e-05,
+      "loss": 19.1199,
+      "mean_token_accuracy": 0.003232458623824641,
+      "num_tokens": 254905.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.20334928229665072,
+      "grad_norm": 44.75,
+      "learning_rate": 4.6630781499202556e-05,
+      "loss": 18.0852,
+      "mean_token_accuracy": 0.0037016855902038516,
+      "num_tokens": 269854.0,
+      "step": 170
+    },
+    {
+      "epoch": 0.215311004784689,
+      "grad_norm": 47.25,
+      "learning_rate": 4.6431419457735245e-05,
+      "loss": 17.1532,
+      "mean_token_accuracy": 0.0039575109840370715,
+      "num_tokens": 285467.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.22727272727272727,
+      "grad_norm": 46.25,
+      "learning_rate": 4.623205741626794e-05,
+      "loss": 16.3724,
+      "mean_token_accuracy": 0.005616716342046857,
+      "num_tokens": 300009.0,
+      "step": 190
+    },
+    {
+      "epoch": 0.23923444976076555,
+      "grad_norm": 36.25,
+      "learning_rate": 4.603269537480064e-05,
+      "loss": 15.551,
+      "mean_token_accuracy": 0.006238142051734031,
+      "num_tokens": 316750.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.2511961722488038,
+      "grad_norm": 41.75,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 14.3853,
+      "mean_token_accuracy": 0.007631447073072195,
+      "num_tokens": 332122.0,
+      "step": 210
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 35.25,
+      "learning_rate": 4.563397129186603e-05,
+      "loss": 14.1354,
+      "mean_token_accuracy": 0.011761691747233271,
+      "num_tokens": 345320.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.2751196172248804,
+      "grad_norm": 45.25,
+      "learning_rate": 4.5434609250398726e-05,
+      "loss": 13.9972,
+      "mean_token_accuracy": 0.011121365474537015,
+      "num_tokens": 360052.0,
+      "step": 230
+    },
+    {
+      "epoch": 0.28708133971291866,
+      "grad_norm": 26.875,
+      "learning_rate": 4.523524720893142e-05,
+      "loss": 13.4802,
+      "mean_token_accuracy": 0.014945084601640702,
+      "num_tokens": 375674.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.29904306220095694,
+      "grad_norm": 29.125,
+      "learning_rate": 4.503588516746411e-05,
+      "loss": 12.5142,
+      "mean_token_accuracy": 0.018064118921756744,
+      "num_tokens": 391484.0,
+      "step": 250
+    },
+    {
+      "epoch": 0.31100478468899523,
+      "grad_norm": 34.25,
+      "learning_rate": 4.483652312599681e-05,
+      "loss": 12.4949,
+      "mean_token_accuracy": 0.018949254974722864,
+      "num_tokens": 405586.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.3229665071770335,
+      "grad_norm": 41.75,
+      "learning_rate": 4.4637161084529504e-05,
+      "loss": 12.1934,
+      "mean_token_accuracy": 0.017779586557298898,
+      "num_tokens": 421161.0,
+      "step": 270
+    },
+    {
+      "epoch": 0.3349282296650718,
+      "grad_norm": 29.625,
+      "learning_rate": 4.44377990430622e-05,
+      "loss": 11.9731,
+      "mean_token_accuracy": 0.019403476919978856,
+      "num_tokens": 438046.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.34688995215311,
+      "grad_norm": 33.25,
+      "learning_rate": 4.4238437001594896e-05,
+      "loss": 11.2492,
+      "mean_token_accuracy": 0.021459076926112174,
+      "num_tokens": 454098.0,
+      "step": 290
+    },
+    {
+      "epoch": 0.3588516746411483,
+      "grad_norm": 25.875,
+      "learning_rate": 4.40390749601276e-05,
+      "loss": 11.2402,
+      "mean_token_accuracy": 0.0200654573738575,
+      "num_tokens": 471900.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.3708133971291866,
+      "grad_norm": 29.375,
+      "learning_rate": 4.383971291866029e-05,
+      "loss": 11.1068,
+      "mean_token_accuracy": 0.022695281356573106,
+      "num_tokens": 486795.0,
+      "step": 310
+    },
+    {
+      "epoch": 0.3827751196172249,
+      "grad_norm": 32.25,
+      "learning_rate": 4.3640350877192985e-05,
+      "loss": 10.9675,
+      "mean_token_accuracy": 0.021525637619197367,
+      "num_tokens": 503071.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.39473684210526316,
+      "grad_norm": 25.75,
+      "learning_rate": 4.344098883572568e-05,
+      "loss": 10.3584,
+      "mean_token_accuracy": 0.027966510131955148,
+      "num_tokens": 517177.0,
+      "step": 330
+    },
+    {
+      "epoch": 0.40669856459330145,
+      "grad_norm": 29.75,
+      "learning_rate": 4.324162679425838e-05,
+      "loss": 10.4161,
+      "mean_token_accuracy": 0.02753335628658533,
+      "num_tokens": 531444.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.41866028708133973,
+      "grad_norm": 25.5,
+      "learning_rate": 4.304226475279107e-05,
+      "loss": 10.2745,
+      "mean_token_accuracy": 0.027429930120706558,
+      "num_tokens": 547939.0,
+      "step": 350
+    },
+    {
+      "epoch": 0.430622009569378,
+      "grad_norm": 39.5,
+      "learning_rate": 4.284290271132377e-05,
+      "loss": 10.5799,
+      "mean_token_accuracy": 0.026776039227843284,
+      "num_tokens": 563428.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.44258373205741625,
+      "grad_norm": 21.375,
+      "learning_rate": 4.2643540669856466e-05,
+      "loss": 10.0957,
+      "mean_token_accuracy": 0.02906850315630436,
+      "num_tokens": 578486.0,
+      "step": 370
+    },
+    {
+      "epoch": 0.45454545454545453,
+      "grad_norm": 20.75,
+      "learning_rate": 4.244417862838916e-05,
+      "loss": 9.7364,
+      "mean_token_accuracy": 0.025704856589436532,
+      "num_tokens": 592561.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.4665071770334928,
+      "grad_norm": 19.625,
+      "learning_rate": 4.224481658692185e-05,
+      "loss": 9.6673,
+      "mean_token_accuracy": 0.028244724869728087,
+      "num_tokens": 609128.0,
+      "step": 390
+    },
+    {
+      "epoch": 0.4784688995215311,
+      "grad_norm": 20.0,
+      "learning_rate": 4.204545454545455e-05,
+      "loss": 9.4029,
+      "mean_token_accuracy": 0.029186627827584743,
+      "num_tokens": 623498.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.4904306220095694,
+      "grad_norm": 19.625,
+      "learning_rate": 4.1846092503987243e-05,
+      "loss": 9.6126,
+      "mean_token_accuracy": 0.02799250017851591,
+      "num_tokens": 639552.0,
+      "step": 410
+    },
+    {
+      "epoch": 0.5023923444976076,
+      "grad_norm": 18.5,
+      "learning_rate": 4.164673046251994e-05,
+      "loss": 9.4314,
+      "mean_token_accuracy": 0.029687324352562428,
+      "num_tokens": 654069.0,
+      "step": 420
+    },
+    {
+      "epoch": 0.5143540669856459,
+      "grad_norm": 21.875,
+      "learning_rate": 4.1447368421052636e-05,
+      "loss": 9.111,
+      "mean_token_accuracy": 0.028119230084121226,
+      "num_tokens": 669050.0,
+      "step": 430
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 16.5,
+      "learning_rate": 4.124800637958533e-05,
+      "loss": 9.6644,
+      "mean_token_accuracy": 0.03025283953174949,
+      "num_tokens": 684231.0,
+      "step": 440
+    },
+    {
+      "epoch": 0.5382775119617225,
+      "grad_norm": 16.5,
+      "learning_rate": 4.104864433811803e-05,
+      "loss": 9.3751,
+      "mean_token_accuracy": 0.02908811327069998,
+      "num_tokens": 702272.0,
+      "step": 450
+    },
+    {
+      "epoch": 0.5502392344497608,
+      "grad_norm": 17.375,
+      "learning_rate": 4.084928229665072e-05,
+      "loss": 9.1689,
+      "mean_token_accuracy": 0.030976119451224804,
+      "num_tokens": 720268.0,
+      "step": 460
+    },
+    {
+      "epoch": 0.562200956937799,
+      "grad_norm": 26.875,
+      "learning_rate": 4.0649920255183414e-05,
+      "loss": 9.0435,
+      "mean_token_accuracy": 0.03120351955294609,
+      "num_tokens": 738391.0,
+      "step": 470
+    },
+    {
+      "epoch": 0.5741626794258373,
+      "grad_norm": 16.125,
+      "learning_rate": 4.045055821371611e-05,
+      "loss": 8.8265,
+      "mean_token_accuracy": 0.03519163243472576,
+      "num_tokens": 752428.0,
+      "step": 480
+    },
+    {
+      "epoch": 0.5861244019138756,
+      "grad_norm": 16.75,
+      "learning_rate": 4.0251196172248806e-05,
+      "loss": 9.4865,
+      "mean_token_accuracy": 0.03259586580097675,
+      "num_tokens": 770127.0,
+      "step": 490
+    },
+    {
+      "epoch": 0.5980861244019139,
+      "grad_norm": 22.0,
+      "learning_rate": 4.00518341307815e-05,
+      "loss": 8.697,
+      "mean_token_accuracy": 0.03240090515464544,
+      "num_tokens": 786336.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.6100478468899522,
+      "grad_norm": 21.5,
+      "learning_rate": 3.98524720893142e-05,
+      "loss": 8.8026,
+      "mean_token_accuracy": 0.028156513161957265,
+      "num_tokens": 802572.0,
+      "step": 510
+    },
+    {
+      "epoch": 0.6220095693779905,
+      "grad_norm": 14.5625,
+      "learning_rate": 3.9653110047846894e-05,
+      "loss": 8.5123,
+      "mean_token_accuracy": 0.03563346900045872,
+      "num_tokens": 817380.0,
+      "step": 520
+    },
+    {
+      "epoch": 0.6339712918660287,
+      "grad_norm": 19.125,
+      "learning_rate": 3.945374800637959e-05,
+      "loss": 8.5729,
+      "mean_token_accuracy": 0.03419111575931311,
+      "num_tokens": 834232.0,
+      "step": 530
+    },
+    {
+      "epoch": 0.645933014354067,
+      "grad_norm": 15.1875,
+      "learning_rate": 3.925438596491228e-05,
+      "loss": 8.9406,
+      "mean_token_accuracy": 0.03414075216278434,
+      "num_tokens": 850055.0,
+      "step": 540
+    },
+    {
+      "epoch": 0.6578947368421053,
+      "grad_norm": 25.875,
+      "learning_rate": 3.9055023923444976e-05,
+      "loss": 8.4975,
+      "mean_token_accuracy": 0.03595046941190958,
+      "num_tokens": 866135.0,
+      "step": 550
+    },
+    {
+      "epoch": 0.6698564593301436,
+      "grad_norm": 12.75,
+      "learning_rate": 3.885566188197767e-05,
+      "loss": 8.2385,
+      "mean_token_accuracy": 0.03583654277026653,
+      "num_tokens": 884639.0,
+      "step": 560
+    },
+    {
+      "epoch": 0.6818181818181818,
+      "grad_norm": 13.75,
+      "learning_rate": 3.865629984051037e-05,
+      "loss": 8.4336,
+      "mean_token_accuracy": 0.032993039861321447,
+      "num_tokens": 901515.0,
+      "step": 570
+    },
+    {
+      "epoch": 0.69377990430622,
+      "grad_norm": 21.0,
+      "learning_rate": 3.8456937799043065e-05,
+      "loss": 8.6048,
+      "mean_token_accuracy": 0.0326311569660902,
+      "num_tokens": 918500.0,
+      "step": 580
+    },
+    {
+      "epoch": 0.7057416267942583,
+      "grad_norm": 16.625,
+      "learning_rate": 3.825757575757576e-05,
+      "loss": 8.3489,
+      "mean_token_accuracy": 0.037175539322197436,
+      "num_tokens": 934654.0,
+      "step": 590
+    },
+    {
+      "epoch": 0.7177033492822966,
+      "grad_norm": 12.875,
+      "learning_rate": 3.805821371610846e-05,
+      "loss": 7.919,
+      "mean_token_accuracy": 0.040041149407625196,
+      "num_tokens": 951137.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.7296650717703349,
+      "grad_norm": 20.0,
+      "learning_rate": 3.7858851674641146e-05,
+      "loss": 8.0379,
+      "mean_token_accuracy": 0.034196126647293566,
+      "num_tokens": 966104.0,
+      "step": 610
+    },
+    {
+      "epoch": 0.7416267942583732,
+      "grad_norm": 14.3125,
+      "learning_rate": 3.765948963317384e-05,
+      "loss": 8.185,
+      "mean_token_accuracy": 0.037530180625617506,
+      "num_tokens": 978723.0,
+      "step": 620
+    },
+    {
+      "epoch": 0.7535885167464115,
+      "grad_norm": 19.125,
+      "learning_rate": 3.746012759170654e-05,
+      "loss": 7.9819,
+      "mean_token_accuracy": 0.04008628278970718,
+      "num_tokens": 996286.0,
+      "step": 630
+    },
+    {
+      "epoch": 0.7655502392344498,
+      "grad_norm": 13.625,
+      "learning_rate": 3.7260765550239235e-05,
+      "loss": 7.7242,
+      "mean_token_accuracy": 0.03889909721910954,
+      "num_tokens": 1012265.0,
+      "step": 640
+    },
+    {
+      "epoch": 0.777511961722488,
+      "grad_norm": 19.375,
+      "learning_rate": 3.706140350877193e-05,
+      "loss": 7.9709,
+      "mean_token_accuracy": 0.04032406695187092,
+      "num_tokens": 1028334.0,
+      "step": 650
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 13.25,
+      "learning_rate": 3.686204146730463e-05,
+      "loss": 8.0423,
+      "mean_token_accuracy": 0.038657473400235176,
+      "num_tokens": 1043429.0,
+      "step": 660
+    },
+    {
+      "epoch": 0.8014354066985646,
+      "grad_norm": 14.375,
+      "learning_rate": 3.666267942583732e-05,
+      "loss": 7.9946,
+      "mean_token_accuracy": 0.03983959686011076,
+      "num_tokens": 1057274.0,
+      "step": 670
+    },
+    {
+      "epoch": 0.8133971291866029,
+      "grad_norm": 16.375,
+      "learning_rate": 3.646331738437001e-05,
+      "loss": 8.3528,
+      "mean_token_accuracy": 0.037851406075060365,
+      "num_tokens": 1071259.0,
+      "step": 680
+    },
+    {
+      "epoch": 0.8253588516746412,
+      "grad_norm": 15.3125,
+      "learning_rate": 3.626395534290271e-05,
+      "loss": 8.3224,
+      "mean_token_accuracy": 0.03589223362505436,
+      "num_tokens": 1089053.0,
+      "step": 690
+    },
+    {
+      "epoch": 0.8373205741626795,
+      "grad_norm": 20.375,
+      "learning_rate": 3.6064593301435405e-05,
+      "loss": 7.845,
+      "mean_token_accuracy": 0.03959129601716995,
+      "num_tokens": 1102285.0,
+      "step": 700
+    },
+    {
+      "epoch": 0.8492822966507177,
+      "grad_norm": 13.8125,
+      "learning_rate": 3.58652312599681e-05,
+      "loss": 7.8353,
+      "mean_token_accuracy": 0.046173612400889394,
+      "num_tokens": 1118501.0,
+      "step": 710
+    },
+    {
+      "epoch": 0.861244019138756,
+      "grad_norm": 14.625,
+      "learning_rate": 3.56658692185008e-05,
+      "loss": 8.0074,
+      "mean_token_accuracy": 0.03960017450153828,
+      "num_tokens": 1135070.0,
+      "step": 720
+    },
+    {
+      "epoch": 0.8732057416267942,
+      "grad_norm": 14.3125,
+      "learning_rate": 3.5466507177033493e-05,
+      "loss": 7.8224,
+      "mean_token_accuracy": 0.04144163206219673,
+      "num_tokens": 1150879.0,
+      "step": 730
+    },
+    {
+      "epoch": 0.8851674641148325,
+      "grad_norm": 15.25,
+      "learning_rate": 3.526714513556619e-05,
+      "loss": 7.8332,
+      "mean_token_accuracy": 0.04649510532617569,
+      "num_tokens": 1166334.0,
+      "step": 740
+    },
+    {
+      "epoch": 0.8971291866028708,
+      "grad_norm": 29.0,
+      "learning_rate": 3.5067783094098886e-05,
+      "loss": 8.3253,
+      "mean_token_accuracy": 0.03786750454455614,
+      "num_tokens": 1183891.0,
+      "step": 750
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 14.9375,
+      "learning_rate": 3.4868421052631575e-05,
+      "loss": 7.776,
+      "mean_token_accuracy": 0.04886236339807511,
+      "num_tokens": 1199155.0,
+      "step": 760
+    },
+    {
+      "epoch": 0.9210526315789473,
+      "grad_norm": 15.5,
+      "learning_rate": 3.466905901116427e-05,
+      "loss": 8.1109,
+      "mean_token_accuracy": 0.04119991734623909,
+      "num_tokens": 1212791.0,
+      "step": 770
+    },
+    {
+      "epoch": 0.9330143540669856,
+      "grad_norm": 13.75,
+      "learning_rate": 3.4469696969696974e-05,
+      "loss": 7.8783,
+      "mean_token_accuracy": 0.05077510289847851,
+      "num_tokens": 1227455.0,
+      "step": 780
+    },
+    {
+      "epoch": 0.9449760765550239,
+      "grad_norm": 23.875,
+      "learning_rate": 3.427033492822967e-05,
+      "loss": 8.0403,
+      "mean_token_accuracy": 0.041047198325395585,
+      "num_tokens": 1245814.0,
+      "step": 790
+    },
+    {
+      "epoch": 0.9569377990430622,
+      "grad_norm": 10.875,
+      "learning_rate": 3.4070972886762367e-05,
+      "loss": 7.9784,
+      "mean_token_accuracy": 0.044691753759980204,
+      "num_tokens": 1261146.0,
+      "step": 800
+    },
+    {
+      "epoch": 0.9688995215311005,
+      "grad_norm": 13.875,
+      "learning_rate": 3.387161084529506e-05,
+      "loss": 7.8961,
+      "mean_token_accuracy": 0.04515785053372383,
+      "num_tokens": 1279584.0,
+      "step": 810
+    },
+    {
+      "epoch": 0.9808612440191388,
+      "grad_norm": 9.25,
+      "learning_rate": 3.367224880382775e-05,
+      "loss": 7.6902,
+      "mean_token_accuracy": 0.04966730885207653,
+      "num_tokens": 1298622.0,
+      "step": 820
+    },
+    {
+      "epoch": 0.992822966507177,
+      "grad_norm": 10.625,
+      "learning_rate": 3.347288676236045e-05,
+      "loss": 7.6308,
+      "mean_token_accuracy": 0.051312201842665675,
+      "num_tokens": 1313079.0,
+      "step": 830
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 7.889608860015869,
+      "eval_mean_token_accuracy": 0.04921758498223323,
+      "eval_num_tokens": 1321280.0,
+      "eval_runtime": 18.7031,
+      "eval_samples_per_second": 22.349,
+      "eval_steps_per_second": 2.834,
+      "step": 836
+    },
+    {
+      "epoch": 1.0047846889952152,
+      "grad_norm": 10.625,
+      "learning_rate": 3.3273524720893144e-05,
+      "loss": 7.6017,
+      "mean_token_accuracy": 0.04911007285118103,
+      "num_tokens": 1327489.0,
+      "step": 840
+    },
+    {
+      "epoch": 1.0167464114832536,
+      "grad_norm": 13.625,
+      "learning_rate": 3.307416267942584e-05,
+      "loss": 7.7674,
+      "mean_token_accuracy": 0.04897945895791054,
+      "num_tokens": 1343987.0,
+      "step": 850
+    },
+    {
+      "epoch": 1.0287081339712918,
+      "grad_norm": 15.625,
+      "learning_rate": 3.287480063795854e-05,
+      "loss": 7.6781,
+      "mean_token_accuracy": 0.04750379957258701,
+      "num_tokens": 1360126.0,
+      "step": 860
+    },
+    {
+      "epoch": 1.0406698564593302,
+      "grad_norm": 9.9375,
+      "learning_rate": 3.267543859649123e-05,
+      "loss": 7.4758,
+      "mean_token_accuracy": 0.04961836487054825,
+      "num_tokens": 1376926.0,
+      "step": 870
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 10.25,
+      "learning_rate": 3.247607655502393e-05,
+      "loss": 7.6265,
+      "mean_token_accuracy": 0.049552620574831965,
+      "num_tokens": 1392486.0,
+      "step": 880
+    },
+    {
+      "epoch": 1.0645933014354068,
+      "grad_norm": 11.5625,
+      "learning_rate": 3.227671451355662e-05,
+      "loss": 8.0101,
+      "mean_token_accuracy": 0.04870691932737827,
+      "num_tokens": 1409932.0,
+      "step": 890
+    },
+    {
+      "epoch": 1.076555023923445,
+      "grad_norm": 14.375,
+      "learning_rate": 3.2077352472089315e-05,
+      "loss": 7.6739,
+      "mean_token_accuracy": 0.048780468106269834,
+      "num_tokens": 1427292.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.0885167464114833,
+      "grad_norm": 10.625,
+      "learning_rate": 3.187799043062201e-05,
+      "loss": 7.7819,
+      "mean_token_accuracy": 0.0481975007802248,
+      "num_tokens": 1443541.0,
+      "step": 910
+    },
+    {
+      "epoch": 1.1004784688995215,
+      "grad_norm": 40.75,
+      "learning_rate": 3.167862838915471e-05,
+      "loss": 7.7404,
+      "mean_token_accuracy": 0.048559962399303915,
+      "num_tokens": 1457736.0,
+      "step": 920
+    },
+    {
+      "epoch": 1.11244019138756,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.14792663476874e-05,
+      "loss": 7.8908,
+      "mean_token_accuracy": 0.04830890707671642,
+      "num_tokens": 1472783.0,
+      "step": 930
+    },
+    {
+      "epoch": 1.124401913875598,
+      "grad_norm": 13.0,
+      "learning_rate": 3.12799043062201e-05,
+      "loss": 7.562,
+      "mean_token_accuracy": 0.052664054557681084,
+      "num_tokens": 1489101.0,
+      "step": 940
+    },
+    {
+      "epoch": 1.1363636363636362,
+      "grad_norm": 9.625,
+      "learning_rate": 3.1080542264752795e-05,
+      "loss": 7.5301,
+      "mean_token_accuracy": 0.05596833899617195,
+      "num_tokens": 1505652.0,
+      "step": 950
+    },
+    {
+      "epoch": 1.1483253588516746,
+      "grad_norm": 12.3125,
+      "learning_rate": 3.088118022328549e-05,
+      "loss": 7.9597,
+      "mean_token_accuracy": 0.051548514328897,
+      "num_tokens": 1520909.0,
+      "step": 960
+    },
+    {
+      "epoch": 1.160287081339713,
+      "grad_norm": 13.3125,
+      "learning_rate": 3.068181818181818e-05,
+      "loss": 7.649,
+      "mean_token_accuracy": 0.052403326518833636,
+      "num_tokens": 1535652.0,
+      "step": 970
+    },
+    {
+      "epoch": 1.1722488038277512,
+      "grad_norm": 13.375,
+      "learning_rate": 3.048245614035088e-05,
+      "loss": 7.6625,
+      "mean_token_accuracy": 0.048475722596049306,
+      "num_tokens": 1552205.0,
+      "step": 980
+    },
+    {
+      "epoch": 1.1842105263157894,
+      "grad_norm": 10.3125,
+      "learning_rate": 3.0283094098883573e-05,
+      "loss": 7.3846,
+      "mean_token_accuracy": 0.053659194707870485,
+      "num_tokens": 1567310.0,
+      "step": 990
+    },
+    {
+      "epoch": 1.1961722488038278,
+      "grad_norm": 13.125,
+      "learning_rate": 3.008373205741627e-05,
+      "loss": 7.6728,
+      "mean_token_accuracy": 0.062439772114157674,
+      "num_tokens": 1581582.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.208133971291866,
+      "grad_norm": 13.25,
+      "learning_rate": 2.9884370015948966e-05,
+      "loss": 7.619,
+      "mean_token_accuracy": 0.04972767271101475,
+      "num_tokens": 1598082.0,
+      "step": 1010
+    },
+    {
+      "epoch": 1.2200956937799043,
+      "grad_norm": 11.25,
+      "learning_rate": 2.968500797448166e-05,
+      "loss": 7.681,
+      "mean_token_accuracy": 0.04762520119547844,
+      "num_tokens": 1613460.0,
+      "step": 1020
+    },
+    {
+      "epoch": 1.2320574162679425,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.9485645933014355e-05,
+      "loss": 7.9789,
+      "mean_token_accuracy": 0.053393319994211194,
+      "num_tokens": 1631405.0,
+      "step": 1030
+    },
+    {
+      "epoch": 1.244019138755981,
+      "grad_norm": 13.875,
+      "learning_rate": 2.928628389154705e-05,
+      "loss": 7.5258,
+      "mean_token_accuracy": 0.0497327771037817,
+      "num_tokens": 1646315.0,
+      "step": 1040
+    },
+    {
+      "epoch": 1.255980861244019,
+      "grad_norm": 13.625,
+      "learning_rate": 2.9086921850079747e-05,
+      "loss": 7.5095,
+      "mean_token_accuracy": 0.04922787994146347,
+      "num_tokens": 1663451.0,
+      "step": 1050
+    },
+    {
+      "epoch": 1.2679425837320575,
+      "grad_norm": 15.5,
+      "learning_rate": 2.888755980861244e-05,
+      "loss": 7.7784,
+      "mean_token_accuracy": 0.055617674812674524,
+      "num_tokens": 1677447.0,
+      "step": 1060
+    },
+    {
+      "epoch": 1.2799043062200957,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.8688197767145136e-05,
+      "loss": 7.5099,
+      "mean_token_accuracy": 0.04883741177618504,
+      "num_tokens": 1695963.0,
+      "step": 1070
+    },
+    {
+      "epoch": 1.291866028708134,
+      "grad_norm": 11.0,
+      "learning_rate": 2.8488835725677832e-05,
+      "loss": 7.8607,
+      "mean_token_accuracy": 0.04243286289274693,
+      "num_tokens": 1713184.0,
+      "step": 1080
+    },
+    {
+      "epoch": 1.3038277511961722,
+      "grad_norm": 11.875,
+      "learning_rate": 2.8289473684210528e-05,
+      "loss": 7.6724,
+      "mean_token_accuracy": 0.05059143900871277,
+      "num_tokens": 1727104.0,
+      "step": 1090
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.809011164274322e-05,
+      "loss": 7.7283,
+      "mean_token_accuracy": 0.05400256849825382,
+      "num_tokens": 1739981.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.3277511961722488,
+      "grad_norm": 10.625,
+      "learning_rate": 2.7890749601275917e-05,
+      "loss": 7.6728,
+      "mean_token_accuracy": 0.05072182305157184,
+      "num_tokens": 1756367.0,
+      "step": 1110
+    },
+    {
+      "epoch": 1.339712918660287,
+      "grad_norm": 15.25,
+      "learning_rate": 2.7691387559808613e-05,
+      "loss": 7.5879,
+      "mean_token_accuracy": 0.0517768245190382,
+      "num_tokens": 1770201.0,
+      "step": 1120
+    },
+    {
+      "epoch": 1.3516746411483254,
+      "grad_norm": 28.25,
+      "learning_rate": 2.749202551834131e-05,
+      "loss": 8.4438,
+      "mean_token_accuracy": 0.054370812699198724,
+      "num_tokens": 1784812.0,
+      "step": 1130
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.7292663476874002e-05,
+      "loss": 7.7018,
+      "mean_token_accuracy": 0.05111252702772617,
+      "num_tokens": 1800658.0,
+      "step": 1140
+    },
+    {
+      "epoch": 1.375598086124402,
+      "grad_norm": 20.125,
+      "learning_rate": 2.7093301435406698e-05,
+      "loss": 7.7442,
+      "mean_token_accuracy": 0.050131586566567424,
+      "num_tokens": 1817111.0,
+      "step": 1150
+    },
+    {
+      "epoch": 1.38755980861244,
+      "grad_norm": 15.375,
+      "learning_rate": 2.6893939393939394e-05,
+      "loss": 7.8474,
+      "mean_token_accuracy": 0.051092632487416266,
+      "num_tokens": 1832422.0,
+      "step": 1160
+    },
+    {
+      "epoch": 1.3995215311004785,
+      "grad_norm": 13.625,
+      "learning_rate": 2.6694577352472087e-05,
+      "loss": 7.8175,
+      "mean_token_accuracy": 0.04797025993466377,
+      "num_tokens": 1849302.0,
+      "step": 1170
+    },
+    {
+      "epoch": 1.4114832535885167,
+      "grad_norm": 9.9375,
+      "learning_rate": 2.6495215311004783e-05,
+      "loss": 7.7398,
+      "mean_token_accuracy": 0.05692324116826057,
+      "num_tokens": 1863943.0,
+      "step": 1180
+    },
+    {
+      "epoch": 1.423444976076555,
+      "grad_norm": 10.875,
+      "learning_rate": 2.629585326953748e-05,
+      "loss": 7.5425,
+      "mean_token_accuracy": 0.055062346160411835,
+      "num_tokens": 1878423.0,
+      "step": 1190
+    },
+    {
+      "epoch": 1.4354066985645932,
+      "grad_norm": 13.125,
+      "learning_rate": 2.6096491228070176e-05,
+      "loss": 7.7075,
+      "mean_token_accuracy": 0.05213572233915329,
+      "num_tokens": 1893221.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.4473684210526316,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.589712918660287e-05,
+      "loss": 7.6832,
+      "mean_token_accuracy": 0.05459029637277126,
+      "num_tokens": 1910194.0,
+      "step": 1210
+    },
+    {
+      "epoch": 1.4593301435406698,
+      "grad_norm": 16.25,
+      "learning_rate": 2.5697767145135565e-05,
+      "loss": 7.5755,
+      "mean_token_accuracy": 0.05433895252645016,
+      "num_tokens": 1927848.0,
+      "step": 1220
+    },
+    {
+      "epoch": 1.4712918660287082,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.549840510366826e-05,
+      "loss": 7.61,
+      "mean_token_accuracy": 0.05629041828215122,
+      "num_tokens": 1947519.0,
+      "step": 1230
+    },
+    {
+      "epoch": 1.4832535885167464,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.5299043062200957e-05,
+      "loss": 7.6037,
+      "mean_token_accuracy": 0.057362372800707814,
+      "num_tokens": 1962606.0,
+      "step": 1240
+    },
+    {
+      "epoch": 1.4952153110047846,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.509968102073365e-05,
+      "loss": 7.9281,
+      "mean_token_accuracy": 0.05096495188772678,
+      "num_tokens": 1979220.0,
+      "step": 1250
+    },
+    {
+      "epoch": 1.507177033492823,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.490031897926635e-05,
+      "loss": 7.5714,
+      "mean_token_accuracy": 0.051651550084352495,
+      "num_tokens": 1995455.0,
+      "step": 1260
+    },
+    {
+      "epoch": 1.5191387559808613,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4700956937799045e-05,
+      "loss": 7.8689,
+      "mean_token_accuracy": 0.04886260032653809,
+      "num_tokens": 2012448.0,
+      "step": 1270
+    },
+    {
+      "epoch": 1.5311004784688995,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.4501594896331738e-05,
+      "loss": 7.5645,
+      "mean_token_accuracy": 0.05832184366881847,
+      "num_tokens": 2029242.0,
+      "step": 1280
+    },
+    {
+      "epoch": 1.5430622009569377,
+      "grad_norm": 9.0625,
+      "learning_rate": 2.4302232854864434e-05,
+      "loss": 7.6672,
+      "mean_token_accuracy": 0.05073030404746533,
+      "num_tokens": 2045026.0,
+      "step": 1290
+    },
+    {
+      "epoch": 1.555023923444976,
+      "grad_norm": 16.125,
+      "learning_rate": 2.410287081339713e-05,
+      "loss": 8.02,
+      "mean_token_accuracy": 0.05017628595232963,
+      "num_tokens": 2059452.0,
+      "step": 1300
+    },
+    {
+      "epoch": 1.5669856459330145,
+      "grad_norm": 13.375,
+      "learning_rate": 2.3903508771929827e-05,
+      "loss": 7.5553,
+      "mean_token_accuracy": 0.05098568722605705,
+      "num_tokens": 2076536.0,
+      "step": 1310
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 14.875,
+      "learning_rate": 2.370414673046252e-05,
+      "loss": 7.6946,
+      "mean_token_accuracy": 0.053875621780753134,
+      "num_tokens": 2092058.0,
+      "step": 1320
+    },
+    {
+      "epoch": 1.5909090909090908,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.3504784688995216e-05,
+      "loss": 7.815,
+      "mean_token_accuracy": 0.05061427466571331,
+      "num_tokens": 2108006.0,
+      "step": 1330
+    },
+    {
+      "epoch": 1.6028708133971292,
+      "grad_norm": 16.0,
+      "learning_rate": 2.3305422647527912e-05,
+      "loss": 7.8098,
+      "mean_token_accuracy": 0.048553376272320745,
+      "num_tokens": 2125780.0,
+      "step": 1340
+    },
+    {
+      "epoch": 1.6148325358851676,
+      "grad_norm": 13.375,
+      "learning_rate": 2.3106060606060605e-05,
+      "loss": 7.4606,
+      "mean_token_accuracy": 0.06270428001880646,
+      "num_tokens": 2140186.0,
+      "step": 1350
+    },
+    {
+      "epoch": 1.6267942583732058,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.29066985645933e-05,
+      "loss": 7.5675,
+      "mean_token_accuracy": 0.05657584816217422,
+      "num_tokens": 2157025.0,
+      "step": 1360
+    },
+    {
+      "epoch": 1.638755980861244,
+      "grad_norm": 12.5,
+      "learning_rate": 2.2707336523125997e-05,
+      "loss": 7.4467,
+      "mean_token_accuracy": 0.053974258899688723,
+      "num_tokens": 2171122.0,
+      "step": 1370
+    },
+    {
+      "epoch": 1.6507177033492821,
+      "grad_norm": 10.875,
+      "learning_rate": 2.2507974481658693e-05,
+      "loss": 7.4626,
+      "mean_token_accuracy": 0.052587350085377696,
+      "num_tokens": 2186298.0,
+      "step": 1380
+    },
+    {
+      "epoch": 1.6626794258373205,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.230861244019139e-05,
+      "loss": 7.5117,
+      "mean_token_accuracy": 0.055986383184790614,
+      "num_tokens": 2201725.0,
+      "step": 1390
+    },
+    {
+      "epoch": 1.674641148325359,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.2109250398724085e-05,
+      "loss": 7.7233,
+      "mean_token_accuracy": 0.05318257175385952,
+      "num_tokens": 2216843.0,
+      "step": 1400
+    },
+    {
+      "epoch": 1.686602870813397,
+      "grad_norm": 16.75,
+      "learning_rate": 2.190988835725678e-05,
+      "loss": 7.6867,
+      "mean_token_accuracy": 0.052845988795161246,
+      "num_tokens": 2234658.0,
+      "step": 1410
+    },
+    {
+      "epoch": 1.6985645933014353,
+      "grad_norm": 9.8125,
+      "learning_rate": 2.1710526315789474e-05,
+      "loss": 7.5536,
+      "mean_token_accuracy": 0.05676522888243198,
+      "num_tokens": 2249972.0,
+      "step": 1420
+    },
+    {
+      "epoch": 1.7105263157894737,
+      "grad_norm": 15.25,
+      "learning_rate": 2.151116427432217e-05,
+      "loss": 7.8057,
+      "mean_token_accuracy": 0.052781854569911954,
+      "num_tokens": 2265621.0,
+      "step": 1430
+    },
+    {
+      "epoch": 1.722488038277512,
+      "grad_norm": 12.125,
+      "learning_rate": 2.1311802232854867e-05,
+      "loss": 7.7877,
+      "mean_token_accuracy": 0.05187459997832775,
+      "num_tokens": 2281490.0,
+      "step": 1440
+    },
+    {
+      "epoch": 1.7344497607655502,
+      "grad_norm": 14.25,
+      "learning_rate": 2.1112440191387563e-05,
+      "loss": 7.5293,
+      "mean_token_accuracy": 0.05277816876769066,
+      "num_tokens": 2296372.0,
+      "step": 1450
+    },
+    {
+      "epoch": 1.7464114832535884,
+      "grad_norm": 10.75,
+      "learning_rate": 2.0913078149920256e-05,
+      "loss": 7.5161,
+      "mean_token_accuracy": 0.059847253188490865,
+      "num_tokens": 2311710.0,
+      "step": 1460
+    },
+    {
+      "epoch": 1.7583732057416268,
+      "grad_norm": 17.125,
+      "learning_rate": 2.071371610845295e-05,
+      "loss": 7.7205,
+      "mean_token_accuracy": 0.05046251565217972,
+      "num_tokens": 2328125.0,
+      "step": 1470
+    },
+    {
+      "epoch": 1.7703349282296652,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.0514354066985648e-05,
+      "loss": 7.4838,
+      "mean_token_accuracy": 0.05662818215787411,
+      "num_tokens": 2342723.0,
+      "step": 1480
+    },
+    {
+      "epoch": 1.7822966507177034,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.0314992025518344e-05,
+      "loss": 7.5797,
+      "mean_token_accuracy": 0.05622794292867184,
+      "num_tokens": 2357907.0,
+      "step": 1490
+    },
+    {
+      "epoch": 1.7942583732057416,
+      "grad_norm": 19.75,
+      "learning_rate": 2.0115629984051037e-05,
+      "loss": 7.8304,
+      "mean_token_accuracy": 0.05124356746673584,
+      "num_tokens": 2374672.0,
+      "step": 1500
+    },
+    {
+      "epoch": 1.80622009569378,
+      "grad_norm": 10.75,
+      "learning_rate": 1.9916267942583733e-05,
+      "loss": 7.6402,
+      "mean_token_accuracy": 0.05051091276109219,
+      "num_tokens": 2391528.0,
+      "step": 1510
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 15.125,
+      "learning_rate": 1.971690590111643e-05,
+      "loss": 7.4673,
+      "mean_token_accuracy": 0.05444490201771259,
+      "num_tokens": 2407044.0,
+      "step": 1520
+    },
+    {
+      "epoch": 1.8301435406698565,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.9517543859649122e-05,
+      "loss": 7.647,
+      "mean_token_accuracy": 0.048004547134041786,
+      "num_tokens": 2422247.0,
+      "step": 1530
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 17.375,
+      "learning_rate": 1.9318181818181818e-05,
+      "loss": 7.582,
+      "mean_token_accuracy": 0.05293577797710895,
+      "num_tokens": 2438387.0,
+      "step": 1540
+    },
+    {
+      "epoch": 1.8540669856459329,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.9118819776714514e-05,
+      "loss": 7.6128,
+      "mean_token_accuracy": 0.05296882390975952,
+      "num_tokens": 2454353.0,
+      "step": 1550
+    },
+    {
+      "epoch": 1.8660287081339713,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.891945773524721e-05,
+      "loss": 7.5914,
+      "mean_token_accuracy": 0.05177001543343067,
+      "num_tokens": 2469893.0,
+      "step": 1560
+    },
+    {
+      "epoch": 1.8779904306220097,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.8720095693779903e-05,
+      "loss": 7.8388,
+      "mean_token_accuracy": 0.05344080198556185,
+      "num_tokens": 2485080.0,
+      "step": 1570
+    },
+    {
+      "epoch": 1.8899521531100478,
+      "grad_norm": 13.375,
+      "learning_rate": 1.85207336523126e-05,
+      "loss": 7.4122,
+      "mean_token_accuracy": 0.056378521770238874,
+      "num_tokens": 2499300.0,
+      "step": 1580
+    },
+    {
+      "epoch": 1.901913875598086,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.8321371610845295e-05,
+      "loss": 7.4838,
+      "mean_token_accuracy": 0.052063405886292456,
+      "num_tokens": 2515672.0,
+      "step": 1590
+    },
+    {
+      "epoch": 1.9138755980861244,
+      "grad_norm": 10.25,
+      "learning_rate": 1.812200956937799e-05,
+      "loss": 7.6252,
+      "mean_token_accuracy": 0.052615062519907954,
+      "num_tokens": 2532131.0,
+      "step": 1600
+    },
+    {
+      "epoch": 1.9258373205741628,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.7922647527910684e-05,
+      "loss": 7.4704,
+      "mean_token_accuracy": 0.05482649989426136,
+      "num_tokens": 2548166.0,
+      "step": 1610
+    },
+    {
+      "epoch": 1.937799043062201,
+      "grad_norm": 11.125,
+      "learning_rate": 1.7723285486443384e-05,
+      "loss": 7.5689,
+      "mean_token_accuracy": 0.05435803644359112,
+      "num_tokens": 2562836.0,
+      "step": 1620
+    },
+    {
+      "epoch": 1.9497607655502391,
+      "grad_norm": 12.0,
+      "learning_rate": 1.752392344497608e-05,
+      "loss": 7.652,
+      "mean_token_accuracy": 0.05930311307311058,
+      "num_tokens": 2576424.0,
+      "step": 1630
+    },
+    {
+      "epoch": 1.9617224880382775,
+      "grad_norm": 12.0,
+      "learning_rate": 1.7324561403508773e-05,
+      "loss": 7.5536,
+      "mean_token_accuracy": 0.05557992160320282,
+      "num_tokens": 2592898.0,
+      "step": 1640
+    },
+    {
+      "epoch": 1.973684210526316,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.712519936204147e-05,
+      "loss": 8.0379,
+      "mean_token_accuracy": 0.05598388984799385,
+      "num_tokens": 2608168.0,
+      "step": 1650
+    },
+    {
+      "epoch": 1.985645933014354,
+      "grad_norm": 12.75,
+      "learning_rate": 1.6925837320574165e-05,
+      "loss": 7.5948,
+      "mean_token_accuracy": 0.0535994254052639,
+      "num_tokens": 2624729.0,
+      "step": 1660
+    },
+    {
+      "epoch": 1.9976076555023923,
+      "grad_norm": 17.875,
+      "learning_rate": 1.6726475279106858e-05,
+      "loss": 7.5617,
+      "mean_token_accuracy": 0.055696948617696765,
+      "num_tokens": 2640184.0,
+      "step": 1670
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 7.681546211242676,
+      "eval_mean_token_accuracy": 0.05555312321433481,
+      "eval_num_tokens": 2642560.0,
+      "eval_runtime": 18.6343,
+      "eval_samples_per_second": 22.432,
+      "eval_steps_per_second": 2.844,
+      "step": 1672
+    },
+    {
+      "epoch": 2.0095693779904304,
+      "grad_norm": 14.0,
+      "learning_rate": 1.6527113237639554e-05,
+      "loss": 7.7797,
+      "mean_token_accuracy": 0.06362233757972717,
+      "num_tokens": 2654711.0,
+      "step": 1680
+    },
+    {
+      "epoch": 2.021531100478469,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.632775119617225e-05,
+      "loss": 7.9745,
+      "mean_token_accuracy": 0.057430309057235715,
+      "num_tokens": 2669744.0,
+      "step": 1690
+    },
+    {
+      "epoch": 2.0334928229665072,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.6128389154704946e-05,
+      "loss": 7.5017,
+      "mean_token_accuracy": 0.05466706827282906,
+      "num_tokens": 2687783.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.0454545454545454,
+      "grad_norm": 12.75,
+      "learning_rate": 1.592902711323764e-05,
+      "loss": 7.5026,
+      "mean_token_accuracy": 0.053967179358005525,
+      "num_tokens": 2705217.0,
+      "step": 1710
+    },
+    {
+      "epoch": 2.0574162679425836,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.5729665071770335e-05,
+      "loss": 7.6858,
+      "mean_token_accuracy": 0.04923243746161461,
+      "num_tokens": 2719654.0,
+      "step": 1720
+    },
+    {
+      "epoch": 2.069377990430622,
+      "grad_norm": 17.5,
+      "learning_rate": 1.553030303030303e-05,
+      "loss": 7.6658,
+      "mean_token_accuracy": 0.05302671529352665,
+      "num_tokens": 2736205.0,
+      "step": 1730
+    },
+    {
+      "epoch": 2.0813397129186604,
+      "grad_norm": 10.75,
+      "learning_rate": 1.5330940988835728e-05,
+      "loss": 8.0743,
+      "mean_token_accuracy": 0.05337436944246292,
+      "num_tokens": 2750160.0,
+      "step": 1740
+    },
+    {
+      "epoch": 2.0933014354066986,
+      "grad_norm": 18.125,
+      "learning_rate": 1.5131578947368422e-05,
+      "loss": 7.9227,
+      "mean_token_accuracy": 0.052077919244766235,
+      "num_tokens": 2765861.0,
+      "step": 1750
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.4932216905901117e-05,
+      "loss": 7.6672,
+      "mean_token_accuracy": 0.05186783894896507,
+      "num_tokens": 2781742.0,
+      "step": 1760
+    },
+    {
+      "epoch": 2.117224880382775,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.4732854864433813e-05,
+      "loss": 7.7002,
+      "mean_token_accuracy": 0.05188237279653549,
+      "num_tokens": 2799667.0,
+      "step": 1770
+    },
+    {
+      "epoch": 2.1291866028708135,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.4533492822966507e-05,
+      "loss": 7.4252,
+      "mean_token_accuracy": 0.053877873718738555,
+      "num_tokens": 2816009.0,
+      "step": 1780
+    },
+    {
+      "epoch": 2.1411483253588517,
+      "grad_norm": 17.125,
+      "learning_rate": 1.4334130781499203e-05,
+      "loss": 7.535,
+      "mean_token_accuracy": 0.05251451525837183,
+      "num_tokens": 2833484.0,
+      "step": 1790
+    },
+    {
+      "epoch": 2.15311004784689,
+      "grad_norm": 11.875,
+      "learning_rate": 1.4134768740031898e-05,
+      "loss": 7.4562,
+      "mean_token_accuracy": 0.054729873687028883,
+      "num_tokens": 2848560.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.165071770334928,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.3935406698564592e-05,
+      "loss": 7.3528,
+      "mean_token_accuracy": 0.05922457054257393,
+      "num_tokens": 2861737.0,
+      "step": 1810
+    },
+    {
+      "epoch": 2.1770334928229667,
+      "grad_norm": 13.75,
+      "learning_rate": 1.3736044657097288e-05,
+      "loss": 7.515,
+      "mean_token_accuracy": 0.056258665025234224,
+      "num_tokens": 2876906.0,
+      "step": 1820
+    },
+    {
+      "epoch": 2.188995215311005,
+      "grad_norm": 13.375,
+      "learning_rate": 1.3536682615629983e-05,
+      "loss": 7.662,
+      "mean_token_accuracy": 0.05119372643530369,
+      "num_tokens": 2893621.0,
+      "step": 1830
+    },
+    {
+      "epoch": 2.200956937799043,
+      "grad_norm": 10.5,
+      "learning_rate": 1.3337320574162679e-05,
+      "loss": 7.5924,
+      "mean_token_accuracy": 0.0524879340082407,
+      "num_tokens": 2910153.0,
+      "step": 1840
+    },
+    {
+      "epoch": 2.212918660287081,
+      "grad_norm": 9.375,
+      "learning_rate": 1.3137958532695374e-05,
+      "loss": 7.5488,
+      "mean_token_accuracy": 0.05508296974003315,
+      "num_tokens": 2925796.0,
+      "step": 1850
+    },
+    {
+      "epoch": 2.22488038277512,
+      "grad_norm": 15.75,
+      "learning_rate": 1.2938596491228071e-05,
+      "loss": 7.6105,
+      "mean_token_accuracy": 0.05919160135090351,
+      "num_tokens": 2944005.0,
+      "step": 1860
+    },
+    {
+      "epoch": 2.236842105263158,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.2739234449760768e-05,
+      "loss": 7.3525,
+      "mean_token_accuracy": 0.057943170145154,
+      "num_tokens": 2959480.0,
+      "step": 1870
+    },
+    {
+      "epoch": 2.248803827751196,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.2539872408293462e-05,
+      "loss": 7.7431,
+      "mean_token_accuracy": 0.06011910997331142,
+      "num_tokens": 2975510.0,
+      "step": 1880
+    },
+    {
+      "epoch": 2.2607655502392343,
+      "grad_norm": 9.375,
+      "learning_rate": 1.2340510366826157e-05,
+      "loss": 7.4682,
+      "mean_token_accuracy": 0.05141804702579975,
+      "num_tokens": 2993262.0,
+      "step": 1890
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 23.125,
+      "learning_rate": 1.2141148325358851e-05,
+      "loss": 7.7091,
+      "mean_token_accuracy": 0.05523250661790371,
+      "num_tokens": 3008482.0,
+      "step": 1900
+    },
+    {
+      "epoch": 2.284688995215311,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.1941786283891547e-05,
+      "loss": 7.6048,
+      "mean_token_accuracy": 0.05504187680780888,
+      "num_tokens": 3023533.0,
+      "step": 1910
+    },
+    {
+      "epoch": 2.2966507177033493,
+      "grad_norm": 11.875,
+      "learning_rate": 1.1742424242424243e-05,
+      "loss": 7.4626,
+      "mean_token_accuracy": 0.05014841184020043,
+      "num_tokens": 3039230.0,
+      "step": 1920
+    },
+    {
+      "epoch": 2.3086124401913874,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.154306220095694e-05,
+      "loss": 7.4728,
+      "mean_token_accuracy": 0.051592012494802476,
+      "num_tokens": 3055324.0,
+      "step": 1930
+    },
+    {
+      "epoch": 2.320574162679426,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.1343700159489634e-05,
+      "loss": 7.6298,
+      "mean_token_accuracy": 0.051040302589535715,
+      "num_tokens": 3071645.0,
+      "step": 1940
+    },
+    {
+      "epoch": 2.3325358851674642,
+      "grad_norm": 16.625,
+      "learning_rate": 1.114433811802233e-05,
+      "loss": 7.6678,
+      "mean_token_accuracy": 0.05561281181871891,
+      "num_tokens": 3085756.0,
+      "step": 1950
+    },
+    {
+      "epoch": 2.3444976076555024,
+      "grad_norm": 14.625,
+      "learning_rate": 1.0944976076555025e-05,
+      "loss": 7.5681,
+      "mean_token_accuracy": 0.05392424911260605,
+      "num_tokens": 3100796.0,
+      "step": 1960
+    },
+    {
+      "epoch": 2.3564593301435406,
+      "grad_norm": 11.625,
+      "learning_rate": 1.074561403508772e-05,
+      "loss": 7.5295,
+      "mean_token_accuracy": 0.05229526199400425,
+      "num_tokens": 3115388.0,
+      "step": 1970
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 11.625,
+      "learning_rate": 1.0546251993620415e-05,
+      "loss": 7.4864,
+      "mean_token_accuracy": 0.055051923543214795,
+      "num_tokens": 3130295.0,
+      "step": 1980
+    },
+    {
+      "epoch": 2.3803827751196174,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.034688995215311e-05,
+      "loss": 7.8901,
+      "mean_token_accuracy": 0.057028749212622645,
+      "num_tokens": 3145780.0,
+      "step": 1990
+    },
+    {
+      "epoch": 2.3923444976076556,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.0147527910685806e-05,
+      "loss": 7.3899,
+      "mean_token_accuracy": 0.05296322256326676,
+      "num_tokens": 3162641.0,
+      "step": 2000
+    },
+    {
+      "epoch": 2.4043062200956937,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.9481658692185e-06,
+      "loss": 7.4156,
+      "mean_token_accuracy": 0.05216502845287323,
+      "num_tokens": 3180101.0,
+      "step": 2010
+    },
+    {
+      "epoch": 2.416267942583732,
+      "grad_norm": 9.4375,
+      "learning_rate": 9.748803827751196e-06,
+      "loss": 7.6813,
+      "mean_token_accuracy": 0.05489225313067436,
+      "num_tokens": 3195441.0,
+      "step": 2020
+    },
+    {
+      "epoch": 2.4282296650717705,
+      "grad_norm": 14.125,
+      "learning_rate": 9.549441786283891e-06,
+      "loss": 7.5118,
+      "mean_token_accuracy": 0.05190498977899551,
+      "num_tokens": 3209632.0,
+      "step": 2030
+    },
+    {
+      "epoch": 2.4401913875598087,
+      "grad_norm": 12.0625,
+      "learning_rate": 9.350079744816589e-06,
+      "loss": 7.4348,
+      "mean_token_accuracy": 0.05217030458152294,
+      "num_tokens": 3227006.0,
+      "step": 2040
+    },
+    {
+      "epoch": 2.452153110047847,
+      "grad_norm": 19.75,
+      "learning_rate": 9.150717703349283e-06,
+      "loss": 7.7842,
+      "mean_token_accuracy": 0.05061689857393503,
+      "num_tokens": 3243617.0,
+      "step": 2050
+    },
+    {
+      "epoch": 2.464114832535885,
+      "grad_norm": 16.5,
+      "learning_rate": 8.95135566188198e-06,
+      "loss": 7.8997,
+      "mean_token_accuracy": 0.05135177038609982,
+      "num_tokens": 3262345.0,
+      "step": 2060
+    },
+    {
+      "epoch": 2.4760765550239237,
+      "grad_norm": 12.4375,
+      "learning_rate": 8.751993620414674e-06,
+      "loss": 7.5425,
+      "mean_token_accuracy": 0.05693710744380951,
+      "num_tokens": 3280049.0,
+      "step": 2070
+    },
+    {
+      "epoch": 2.488038277511962,
+      "grad_norm": 12.125,
+      "learning_rate": 8.552631578947368e-06,
+      "loss": 7.7683,
+      "mean_token_accuracy": 0.052999266609549525,
+      "num_tokens": 3293627.0,
+      "step": 2080
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 18.5,
+      "learning_rate": 8.353269537480064e-06,
+      "loss": 7.5021,
+      "mean_token_accuracy": 0.0573310736566782,
+      "num_tokens": 3308963.0,
+      "step": 2090
+    },
+    {
+      "epoch": 2.511961722488038,
+      "grad_norm": 11.1875,
+      "learning_rate": 8.153907496012759e-06,
+      "loss": 7.5266,
+      "mean_token_accuracy": 0.055393803864717484,
+      "num_tokens": 3327211.0,
+      "step": 2100
+    },
+    {
+      "epoch": 2.5239234449760763,
+      "grad_norm": 12.4375,
+      "learning_rate": 7.954545454545455e-06,
+      "loss": 8.0202,
+      "mean_token_accuracy": 0.0516157153993845,
+      "num_tokens": 3342704.0,
+      "step": 2110
+    },
+    {
+      "epoch": 2.535885167464115,
+      "grad_norm": 11.75,
+      "learning_rate": 7.75518341307815e-06,
+      "loss": 7.6579,
+      "mean_token_accuracy": 0.053000685200095175,
+      "num_tokens": 3356957.0,
+      "step": 2120
+    },
+    {
+      "epoch": 2.547846889952153,
+      "grad_norm": 12.0,
+      "learning_rate": 7.555821371610846e-06,
+      "loss": 7.5,
+      "mean_token_accuracy": 0.05583856776356697,
+      "num_tokens": 3371814.0,
+      "step": 2130
+    },
+    {
+      "epoch": 2.5598086124401913,
+      "grad_norm": 11.25,
+      "learning_rate": 7.35645933014354e-06,
+      "loss": 7.4175,
+      "mean_token_accuracy": 0.056200267374515535,
+      "num_tokens": 3387835.0,
+      "step": 2140
+    },
+    {
+      "epoch": 2.57177033492823,
+      "grad_norm": 14.75,
+      "learning_rate": 7.1570972886762355e-06,
+      "loss": 7.601,
+      "mean_token_accuracy": 0.058812175691127774,
+      "num_tokens": 3402724.0,
+      "step": 2150
+    },
+    {
+      "epoch": 2.583732057416268,
+      "grad_norm": 15.125,
+      "learning_rate": 6.9577352472089325e-06,
+      "loss": 7.6731,
+      "mean_token_accuracy": 0.05726858489215374,
+      "num_tokens": 3415782.0,
+      "step": 2160
+    },
+    {
+      "epoch": 2.5956937799043063,
+      "grad_norm": 16.25,
+      "learning_rate": 6.758373205741628e-06,
+      "loss": 7.9197,
+      "mean_token_accuracy": 0.04841956198215484,
+      "num_tokens": 3430907.0,
+      "step": 2170
+    },
+    {
+      "epoch": 2.6076555023923444,
+      "grad_norm": 15.125,
+      "learning_rate": 6.559011164274323e-06,
+      "loss": 7.4308,
+      "mean_token_accuracy": 0.05858968757092953,
+      "num_tokens": 3445725.0,
+      "step": 2180
+    },
+    {
+      "epoch": 2.6196172248803826,
+      "grad_norm": 13.5,
+      "learning_rate": 6.3596491228070184e-06,
+      "loss": 7.7449,
+      "mean_token_accuracy": 0.05409410893917084,
+      "num_tokens": 3463327.0,
+      "step": 2190
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 12.125,
+      "learning_rate": 6.160287081339714e-06,
+      "loss": 7.5525,
+      "mean_token_accuracy": 0.0532541248947382,
+      "num_tokens": 3477759.0,
+      "step": 2200
+    },
+    {
+      "epoch": 2.6435406698564594,
+      "grad_norm": 9.6875,
+      "learning_rate": 5.960925039872409e-06,
+      "loss": 7.408,
+      "mean_token_accuracy": 0.06432338766753673,
+      "num_tokens": 3493604.0,
+      "step": 2210
+    },
+    {
+      "epoch": 2.6555023923444976,
+      "grad_norm": 18.25,
+      "learning_rate": 5.7615629984051035e-06,
+      "loss": 8.0218,
+      "mean_token_accuracy": 0.053415299765765666,
+      "num_tokens": 3508326.0,
+      "step": 2220
+    },
+    {
+      "epoch": 2.6674641148325358,
+      "grad_norm": 11.25,
+      "learning_rate": 5.562200956937799e-06,
+      "loss": 7.7336,
+      "mean_token_accuracy": 0.053274786472320555,
+      "num_tokens": 3523788.0,
+      "step": 2230
+    },
+    {
+      "epoch": 2.679425837320574,
+      "grad_norm": 9.75,
+      "learning_rate": 5.362838915470494e-06,
+      "loss": 7.5168,
+      "mean_token_accuracy": 0.06089800409972668,
+      "num_tokens": 3537455.0,
+      "step": 2240
+    },
+    {
+      "epoch": 2.6913875598086126,
+      "grad_norm": 9.0,
+      "learning_rate": 5.16347687400319e-06,
+      "loss": 7.9244,
+      "mean_token_accuracy": 0.05054541490972042,
+      "num_tokens": 3554477.0,
+      "step": 2250
+    },
+    {
+      "epoch": 2.7033492822966507,
+      "grad_norm": 14.0625,
+      "learning_rate": 4.964114832535886e-06,
+      "loss": 7.493,
+      "mean_token_accuracy": 0.05583830550312996,
+      "num_tokens": 3568763.0,
+      "step": 2260
+    },
+    {
+      "epoch": 2.715311004784689,
+      "grad_norm": 10.6875,
+      "learning_rate": 4.764752791068581e-06,
+      "loss": 7.6837,
+      "mean_token_accuracy": 0.056372621655464174,
+      "num_tokens": 3584199.0,
+      "step": 2270
+    },
+    {
+      "epoch": 2.7272727272727275,
+      "grad_norm": 8.875,
+      "learning_rate": 4.565390749601276e-06,
+      "loss": 7.5824,
+      "mean_token_accuracy": 0.05312733016908169,
+      "num_tokens": 3601486.0,
+      "step": 2280
+    },
+    {
+      "epoch": 2.7392344497607657,
+      "grad_norm": 13.0,
+      "learning_rate": 4.3660287081339716e-06,
+      "loss": 7.5518,
+      "mean_token_accuracy": 0.05006753616034985,
+      "num_tokens": 3620786.0,
+      "step": 2290
+    },
+    {
+      "epoch": 2.751196172248804,
+      "grad_norm": 13.875,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 7.3837,
+      "mean_token_accuracy": 0.059043745324015615,
+      "num_tokens": 3636570.0,
+      "step": 2300
+    },
+    {
+      "epoch": 2.763157894736842,
+      "grad_norm": 12.5,
+      "learning_rate": 3.967304625199362e-06,
+      "loss": 7.523,
+      "mean_token_accuracy": 0.05304129309952259,
+      "num_tokens": 3651913.0,
+      "step": 2310
+    },
+    {
+      "epoch": 2.77511961722488,
+      "grad_norm": 11.875,
+      "learning_rate": 3.767942583732058e-06,
+      "loss": 7.5162,
+      "mean_token_accuracy": 0.055253866314888,
+      "num_tokens": 3668785.0,
+      "step": 2320
+    },
+    {
+      "epoch": 2.787081339712919,
+      "grad_norm": 16.625,
+      "learning_rate": 3.5685805422647532e-06,
+      "loss": 7.3978,
+      "mean_token_accuracy": 0.05370482616126537,
+      "num_tokens": 3685088.0,
+      "step": 2330
+    },
+    {
+      "epoch": 2.799043062200957,
+      "grad_norm": 49.25,
+      "learning_rate": 3.3692185007974485e-06,
+      "loss": 7.9911,
+      "mean_token_accuracy": 0.05654057189822197,
+      "num_tokens": 3699717.0,
+      "step": 2340
+    },
+    {
+      "epoch": 2.811004784688995,
+      "grad_norm": 9.1875,
+      "learning_rate": 3.1698564593301434e-06,
+      "loss": 7.3809,
+      "mean_token_accuracy": 0.05578604489564896,
+      "num_tokens": 3712729.0,
+      "step": 2350
+    },
+    {
+      "epoch": 2.8229665071770333,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.970494417862839e-06,
+      "loss": 7.4346,
+      "mean_token_accuracy": 0.054243154078722,
+      "num_tokens": 3727180.0,
+      "step": 2360
+    },
+    {
+      "epoch": 2.8349282296650715,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.7711323763955345e-06,
+      "loss": 7.4536,
+      "mean_token_accuracy": 0.05370322801172733,
+      "num_tokens": 3743937.0,
+      "step": 2370
+    },
+    {
+      "epoch": 2.84688995215311,
+      "grad_norm": 27.875,
+      "learning_rate": 2.57177033492823e-06,
+      "loss": 7.5259,
+      "mean_token_accuracy": 0.05500344783067703,
+      "num_tokens": 3760306.0,
+      "step": 2380
+    },
+    {
+      "epoch": 2.8588516746411483,
+      "grad_norm": 11.5,
+      "learning_rate": 2.372408293460925e-06,
+      "loss": 7.7337,
+      "mean_token_accuracy": 0.05269717015326023,
+      "num_tokens": 3776435.0,
+      "step": 2390
+    },
+    {
+      "epoch": 2.8708133971291865,
+      "grad_norm": 11.75,
+      "learning_rate": 2.1730462519936204e-06,
+      "loss": 7.6882,
+      "mean_token_accuracy": 0.06056210286915302,
+      "num_tokens": 3792698.0,
+      "step": 2400
+    },
+    {
+      "epoch": 2.882775119617225,
+      "grad_norm": 14.375,
+      "learning_rate": 1.9736842105263157e-06,
+      "loss": 7.3884,
+      "mean_token_accuracy": 0.058717917650938034,
+      "num_tokens": 3808389.0,
+      "step": 2410
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 18.5,
+      "learning_rate": 1.7743221690590115e-06,
+      "loss": 7.7266,
+      "mean_token_accuracy": 0.05015675462782383,
+      "num_tokens": 3826096.0,
+      "step": 2420
+    },
+    {
+      "epoch": 2.9066985645933014,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.5749601275917066e-06,
+      "loss": 7.5727,
+      "mean_token_accuracy": 0.055614912137389186,
+      "num_tokens": 3841264.0,
+      "step": 2430
+    },
+    {
+      "epoch": 2.9186602870813396,
+      "grad_norm": 15.75,
+      "learning_rate": 1.375598086124402e-06,
+      "loss": 7.7109,
+      "mean_token_accuracy": 0.054452039673924445,
+      "num_tokens": 3858262.0,
+      "step": 2440
+    },
+    {
+      "epoch": 2.930622009569378,
+      "grad_norm": 13.5,
+      "learning_rate": 1.1762360446570972e-06,
+      "loss": 7.4082,
+      "mean_token_accuracy": 0.05675496719777584,
+      "num_tokens": 3874012.0,
+      "step": 2450
+    },
+    {
+      "epoch": 2.9425837320574164,
+      "grad_norm": 11.9375,
+      "learning_rate": 9.768740031897927e-07,
+      "loss": 7.4929,
+      "mean_token_accuracy": 0.05380352921783924,
+      "num_tokens": 3889829.0,
+      "step": 2460
+    },
+    {
+      "epoch": 2.9545454545454546,
+      "grad_norm": 13.4375,
+      "learning_rate": 7.77511961722488e-07,
+      "loss": 7.4973,
+      "mean_token_accuracy": 0.057183431833982466,
+      "num_tokens": 3905985.0,
+      "step": 2470
+    },
+    {
+      "epoch": 2.9665071770334928,
+      "grad_norm": 23.875,
+      "learning_rate": 5.781499202551834e-07,
+      "loss": 7.7017,
+      "mean_token_accuracy": 0.05155903100967407,
+      "num_tokens": 3921607.0,
+      "step": 2480
+    },
+    {
+      "epoch": 2.9784688995215314,
+      "grad_norm": 11.8125,
+      "learning_rate": 3.787878787878788e-07,
+      "loss": 7.7229,
+      "mean_token_accuracy": 0.05097230635583401,
+      "num_tokens": 3937739.0,
+      "step": 2490
+    },
+    {
+      "epoch": 2.990430622009569,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.7942583732057418e-07,
+      "loss": 7.5145,
+      "mean_token_accuracy": 0.058688531070947646,
+      "num_tokens": 3953646.0,
+      "step": 2500
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 7.677623271942139,
+      "eval_mean_token_accuracy": 0.05568885817280356,
+      "eval_num_tokens": 3963840.0,
+      "eval_runtime": 18.6561,
+      "eval_samples_per_second": 22.406,
+      "eval_steps_per_second": 2.841,
+      "step": 2508
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 2508,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.363266938316718e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

sft_output_simpleQA/checkpoint-2508/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:010b0827320093053654035df7d8f073bcd42c7bac7c3dea5c34279a14edc2af
+size 5688