Training in progress, epoch 0, checkpoint

checkpoint-374/added_tokens.json

@@ -0,0 +1,5 @@
+{
+  "<|endofprompt|>": 100276,
+  "<|im_end|>": 100265,
+  "<|im_start|>": 100264
+}

checkpoint-374/chat_template.jinja

@@ -0,0 +1,5 @@
+{% for message in messages %}{{'<|im_start|>' + message['role'] + '
+' + message['content'] + '
+<|im_end|>' + '
+'}}{% endfor %}{{ '<|im_start|>assistant
+' }}

checkpoint-374/config.json

@@ -0,0 +1,29 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 100257,
+  "eos_token_id": 100257,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 8192,
+  "max_position_embeddings": 262144,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 128000000,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.53.1",
+  "use_cache": false,
+  "vocab_size": 102400
+}

checkpoint-374/generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "do_sample": true,
+  "eos_token_id": 2,
+  "transformers_version": "4.53.1"
+}

checkpoint-374/model-00001-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5fbe91c3ebf7f6e2fa87af0ec13a43a78401f748099afe223dc6187c03da9a0
+size 4983077832

checkpoint-374/model-00002-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60438aea893b13cb3d6d83f3fa59a0cbcdaa66b3076a6a1818f9374e8cccd7f3
+size 3842234168

checkpoint-374/model.safetensors.index.json

@@ -0,0 +1,262 @@
+{
+  "metadata": {
+    "total_parameters": 4412641280,
+    "total_size": 8825282560
+  },
+  "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-00002-of-00002.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-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-00002-of-00002.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-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-00002-of-00002.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-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"
+  }
+}

checkpoint-374/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c16b2c171bb7fcae63a9fce942c853cc491438cb188a396bef139ecfe058f80b
+size 11466963134

checkpoint-374/rng_state.pth

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

checkpoint-374/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e2430c244b4f62d796340abe2a913d27e65e5d0599d0a055a51542fe165d7db
+size 1064

checkpoint-374/special_tokens_map.json

@@ -0,0 +1,30 @@
+{
+  "bos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-374/tokenizer_config.json

@@ -0,0 +1,69 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "100257": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100258": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100259": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100260": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100264": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100265": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100276": {
+      "content": "<|endofprompt|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<|endoftext|>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "extra_special_tokens": {},
+  "model_max_length": 16384,
+  "pad_token": "<|endoftext|>",
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>"
+}

checkpoint-374/trainer_state.json

@@ -0,0 +1,2652 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9986648865153538,
+  "eval_steps": 500,
+  "global_step": 374,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0026702269692923898,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0,
+      "loss": 2.534,
+      "step": 1
+    },
+    {
+      "epoch": 0.0053404539385847796,
+      "grad_norm": 0.55078125,
+      "learning_rate": 5.555555555555555e-07,
+      "loss": 2.5367,
+      "step": 2
+    },
+    {
+      "epoch": 0.00801068090787717,
+      "grad_norm": 0.5390625,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 2.5319,
+      "step": 3
+    },
+    {
+      "epoch": 0.010680907877169559,
+      "grad_norm": 0.55078125,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 2.5296,
+      "step": 4
+    },
+    {
+      "epoch": 0.01335113484646195,
+      "grad_norm": 0.55859375,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 2.5376,
+      "step": 5
+    },
+    {
+      "epoch": 0.01602136181575434,
+      "grad_norm": 0.57421875,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 2.5011,
+      "step": 6
+    },
+    {
+      "epoch": 0.018691588785046728,
+      "grad_norm": 0.51171875,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 2.501,
+      "step": 7
+    },
+    {
+      "epoch": 0.021361815754339118,
+      "grad_norm": 0.5390625,
+      "learning_rate": 3.88888888888889e-06,
+      "loss": 2.5337,
+      "step": 8
+    },
+    {
+      "epoch": 0.02403204272363151,
+      "grad_norm": 0.50390625,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 2.4829,
+      "step": 9
+    },
+    {
+      "epoch": 0.0267022696929239,
+      "grad_norm": 0.48828125,
+      "learning_rate": 5e-06,
+      "loss": 2.4611,
+      "step": 10
+    },
+    {
+      "epoch": 0.029372496662216287,
+      "grad_norm": 0.46484375,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 2.4401,
+      "step": 11
+    },
+    {
+      "epoch": 0.03204272363150868,
+      "grad_norm": 0.447265625,
+      "learning_rate": 6.111111111111112e-06,
+      "loss": 2.4879,
+      "step": 12
+    },
+    {
+      "epoch": 0.03471295060080107,
+      "grad_norm": 0.4296875,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 2.5059,
+      "step": 13
+    },
+    {
+      "epoch": 0.037383177570093455,
+      "grad_norm": 0.40234375,
+      "learning_rate": 7.222222222222223e-06,
+      "loss": 2.5041,
+      "step": 14
+    },
+    {
+      "epoch": 0.04005340453938585,
+      "grad_norm": 0.400390625,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 2.516,
+      "step": 15
+    },
+    {
+      "epoch": 0.042723631508678236,
+      "grad_norm": 0.3828125,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 2.504,
+      "step": 16
+    },
+    {
+      "epoch": 0.04539385847797063,
+      "grad_norm": 0.37109375,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 2.523,
+      "step": 17
+    },
+    {
+      "epoch": 0.04806408544726302,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.444444444444445e-06,
+      "loss": 2.5062,
+      "step": 18
+    },
+    {
+      "epoch": 0.050734312416555405,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1e-05,
+      "loss": 2.5219,
+      "step": 19
+    },
+    {
+      "epoch": 0.0534045393858478,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.999805313005946e-06,
+      "loss": 2.5032,
+      "step": 20
+    },
+    {
+      "epoch": 0.056074766355140186,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.999221267184993e-06,
+      "loss": 2.5027,
+      "step": 21
+    },
+    {
+      "epoch": 0.05874499332443257,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.998247908019594e-06,
+      "loss": 2.5135,
+      "step": 22
+    },
+    {
+      "epoch": 0.06141522029372497,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.996885311309892e-06,
+      "loss": 2.5042,
+      "step": 23
+    },
+    {
+      "epoch": 0.06408544726301736,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.995133583167833e-06,
+      "loss": 2.513,
+      "step": 24
+    },
+    {
+      "epoch": 0.06675567423230974,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.992992860008893e-06,
+      "loss": 2.5058,
+      "step": 25
+    },
+    {
+      "epoch": 0.06942590120160214,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.990463308541452e-06,
+      "loss": 2.5437,
+      "step": 26
+    },
+    {
+      "epoch": 0.07209612817089453,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.987545125753818e-06,
+      "loss": 2.4898,
+      "step": 27
+    },
+    {
+      "epoch": 0.07476635514018691,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.98423853889889e-06,
+      "loss": 2.4943,
+      "step": 28
+    },
+    {
+      "epoch": 0.0774365821094793,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.980543805476447e-06,
+      "loss": 2.5259,
+      "step": 29
+    },
+    {
+      "epoch": 0.0801068090787717,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.976461213213104e-06,
+      "loss": 2.5365,
+      "step": 30
+    },
+    {
+      "epoch": 0.08277703604806408,
+      "grad_norm": 0.296875,
+      "learning_rate": 9.971991080039912e-06,
+      "loss": 2.4535,
+      "step": 31
+    },
+    {
+      "epoch": 0.08544726301735647,
+      "grad_norm": 0.302734375,
+      "learning_rate": 9.967133754067581e-06,
+      "loss": 2.5043,
+      "step": 32
+    },
+    {
+      "epoch": 0.08811748998664887,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.961889613559396e-06,
+      "loss": 2.5127,
+      "step": 33
+    },
+    {
+      "epoch": 0.09078771695594126,
+      "grad_norm": 0.298828125,
+      "learning_rate": 9.956259066901733e-06,
+      "loss": 2.5135,
+      "step": 34
+    },
+    {
+      "epoch": 0.09345794392523364,
+      "grad_norm": 0.2890625,
+      "learning_rate": 9.950242552572272e-06,
+      "loss": 2.469,
+      "step": 35
+    },
+    {
+      "epoch": 0.09612817089452604,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.943840539105853e-06,
+      "loss": 2.5182,
+      "step": 36
+    },
+    {
+      "epoch": 0.09879839786381843,
+      "grad_norm": 0.29296875,
+      "learning_rate": 9.937053525057977e-06,
+      "loss": 2.5109,
+      "step": 37
+    },
+    {
+      "epoch": 0.10146862483311081,
+      "grad_norm": 0.291015625,
+      "learning_rate": 9.92988203896599e-06,
+      "loss": 2.5071,
+      "step": 38
+    },
+    {
+      "epoch": 0.1041388518024032,
+      "grad_norm": 0.287109375,
+      "learning_rate": 9.922326639307918e-06,
+      "loss": 2.473,
+      "step": 39
+    },
+    {
+      "epoch": 0.1068090787716956,
+      "grad_norm": 0.28125,
+      "learning_rate": 9.914387914458983e-06,
+      "loss": 2.4875,
+      "step": 40
+    },
+    {
+      "epoch": 0.10947930574098798,
+      "grad_norm": 0.28515625,
+      "learning_rate": 9.906066482645774e-06,
+      "loss": 2.5014,
+      "step": 41
+    },
+    {
+      "epoch": 0.11214953271028037,
+      "grad_norm": 0.283203125,
+      "learning_rate": 9.89736299189811e-06,
+      "loss": 2.5084,
+      "step": 42
+    },
+    {
+      "epoch": 0.11481975967957277,
+      "grad_norm": 0.2890625,
+      "learning_rate": 9.888278119998573e-06,
+      "loss": 2.4905,
+      "step": 43
+    },
+    {
+      "epoch": 0.11748998664886515,
+      "grad_norm": 0.279296875,
+      "learning_rate": 9.878812574429722e-06,
+      "loss": 2.4696,
+      "step": 44
+    },
+    {
+      "epoch": 0.12016021361815754,
+      "grad_norm": 0.298828125,
+      "learning_rate": 9.868967092319003e-06,
+      "loss": 2.4912,
+      "step": 45
+    },
+    {
+      "epoch": 0.12283044058744993,
+      "grad_norm": 0.287109375,
+      "learning_rate": 9.858742440381343e-06,
+      "loss": 2.4963,
+      "step": 46
+    },
+    {
+      "epoch": 0.12550066755674233,
+      "grad_norm": 0.279296875,
+      "learning_rate": 9.848139414859441e-06,
+      "loss": 2.5145,
+      "step": 47
+    },
+    {
+      "epoch": 0.12817089452603472,
+      "grad_norm": 0.279296875,
+      "learning_rate": 9.837158841461767e-06,
+      "loss": 2.4842,
+      "step": 48
+    },
+    {
+      "epoch": 0.1308411214953271,
+      "grad_norm": 0.283203125,
+      "learning_rate": 9.825801575298248e-06,
+      "loss": 2.5164,
+      "step": 49
+    },
+    {
+      "epoch": 0.13351134846461948,
+      "grad_norm": 0.26953125,
+      "learning_rate": 9.814068500813692e-06,
+      "loss": 2.5011,
+      "step": 50
+    },
+    {
+      "epoch": 0.13618157543391188,
+      "grad_norm": 0.275390625,
+      "learning_rate": 9.801960531718898e-06,
+      "loss": 2.5144,
+      "step": 51
+    },
+    {
+      "epoch": 0.13885180240320427,
+      "grad_norm": 0.279296875,
+      "learning_rate": 9.789478610919508e-06,
+      "loss": 2.4887,
+      "step": 52
+    },
+    {
+      "epoch": 0.14152202937249667,
+      "grad_norm": 0.275390625,
+      "learning_rate": 9.77662371044258e-06,
+      "loss": 2.508,
+      "step": 53
+    },
+    {
+      "epoch": 0.14419225634178906,
+      "grad_norm": 0.28125,
+      "learning_rate": 9.763396831360884e-06,
+      "loss": 2.506,
+      "step": 54
+    },
+    {
+      "epoch": 0.14686248331108145,
+      "grad_norm": 0.28125,
+      "learning_rate": 9.749799003714954e-06,
+      "loss": 2.5483,
+      "step": 55
+    },
+    {
+      "epoch": 0.14953271028037382,
+      "grad_norm": 0.279296875,
+      "learning_rate": 9.735831286432869e-06,
+      "loss": 2.5136,
+      "step": 56
+    },
+    {
+      "epoch": 0.15220293724966621,
+      "grad_norm": 0.2734375,
+      "learning_rate": 9.721494767247779e-06,
+      "loss": 2.487,
+      "step": 57
+    },
+    {
+      "epoch": 0.1548731642189586,
+      "grad_norm": 0.28515625,
+      "learning_rate": 9.70679056261322e-06,
+      "loss": 2.5026,
+      "step": 58
+    },
+    {
+      "epoch": 0.157543391188251,
+      "grad_norm": 0.27734375,
+      "learning_rate": 9.691719817616148e-06,
+      "loss": 2.5142,
+      "step": 59
+    },
+    {
+      "epoch": 0.1602136181575434,
+      "grad_norm": 0.275390625,
+      "learning_rate": 9.676283705887783e-06,
+      "loss": 2.5027,
+      "step": 60
+    },
+    {
+      "epoch": 0.1628838451268358,
+      "grad_norm": 0.2734375,
+      "learning_rate": 9.660483429512198e-06,
+      "loss": 2.5344,
+      "step": 61
+    },
+    {
+      "epoch": 0.16555407209612816,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.644320218932723e-06,
+      "loss": 2.5107,
+      "step": 62
+    },
+    {
+      "epoch": 0.16822429906542055,
+      "grad_norm": 0.267578125,
+      "learning_rate": 9.627795332856107e-06,
+      "loss": 2.4842,
+      "step": 63
+    },
+    {
+      "epoch": 0.17089452603471295,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.61091005815451e-06,
+      "loss": 2.4635,
+      "step": 64
+    },
+    {
+      "epoch": 0.17356475300400534,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.59366570976528e-06,
+      "loss": 2.5131,
+      "step": 65
+    },
+    {
+      "epoch": 0.17623497997329773,
+      "grad_norm": 0.2734375,
+      "learning_rate": 9.576063630588563e-06,
+      "loss": 2.5098,
+      "step": 66
+    },
+    {
+      "epoch": 0.17890520694259013,
+      "grad_norm": 0.275390625,
+      "learning_rate": 9.55810519138271e-06,
+      "loss": 2.5209,
+      "step": 67
+    },
+    {
+      "epoch": 0.18157543391188252,
+      "grad_norm": 0.28125,
+      "learning_rate": 9.53979179065754e-06,
+      "loss": 2.5272,
+      "step": 68
+    },
+    {
+      "epoch": 0.1842456608811749,
+      "grad_norm": 0.26953125,
+      "learning_rate": 9.521124854565425e-06,
+      "loss": 2.5067,
+      "step": 69
+    },
+    {
+      "epoch": 0.18691588785046728,
+      "grad_norm": 0.26953125,
+      "learning_rate": 9.50210583679024e-06,
+      "loss": 2.5268,
+      "step": 70
+    },
+    {
+      "epoch": 0.18958611481975968,
+      "grad_norm": 0.2734375,
+      "learning_rate": 9.482736218434144e-06,
+      "loss": 2.5179,
+      "step": 71
+    },
+    {
+      "epoch": 0.19225634178905207,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.463017507902245e-06,
+      "loss": 2.4965,
+      "step": 72
+    },
+    {
+      "epoch": 0.19492656875834447,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.442951240785135e-06,
+      "loss": 2.5109,
+      "step": 73
+    },
+    {
+      "epoch": 0.19759679572763686,
+      "grad_norm": 0.265625,
+      "learning_rate": 9.422538979739307e-06,
+      "loss": 2.4834,
+      "step": 74
+    },
+    {
+      "epoch": 0.20026702269692923,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.401782314365458e-06,
+      "loss": 2.4986,
+      "step": 75
+    },
+    {
+      "epoch": 0.20293724966622162,
+      "grad_norm": 0.26953125,
+      "learning_rate": 9.380682861084703e-06,
+      "loss": 2.4779,
+      "step": 76
+    },
+    {
+      "epoch": 0.205607476635514,
+      "grad_norm": 0.26953125,
+      "learning_rate": 9.359242263012693e-06,
+      "loss": 2.4841,
+      "step": 77
+    },
+    {
+      "epoch": 0.2082777036048064,
+      "grad_norm": 0.2734375,
+      "learning_rate": 9.33746218983167e-06,
+      "loss": 2.4902,
+      "step": 78
+    },
+    {
+      "epoch": 0.2109479305740988,
+      "grad_norm": 0.30078125,
+      "learning_rate": 9.315344337660422e-06,
+      "loss": 2.4984,
+      "step": 79
+    },
+    {
+      "epoch": 0.2136181575433912,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.29289042892221e-06,
+      "loss": 2.4941,
+      "step": 80
+    },
+    {
+      "epoch": 0.2162883845126836,
+      "grad_norm": 0.279296875,
+      "learning_rate": 9.270102212210632e-06,
+      "loss": 2.5192,
+      "step": 81
+    },
+    {
+      "epoch": 0.21895861148197596,
+      "grad_norm": 0.26953125,
+      "learning_rate": 9.246981462153456e-06,
+      "loss": 2.4991,
+      "step": 82
+    },
+    {
+      "epoch": 0.22162883845126835,
+      "grad_norm": 0.26953125,
+      "learning_rate": 9.223529979274411e-06,
+      "loss": 2.483,
+      "step": 83
+    },
+    {
+      "epoch": 0.22429906542056074,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.19974958985298e-06,
+      "loss": 2.5091,
+      "step": 84
+    },
+    {
+      "epoch": 0.22696929238985314,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.175642145782179e-06,
+      "loss": 2.5119,
+      "step": 85
+    },
+    {
+      "epoch": 0.22963951935914553,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.151209524424333e-06,
+      "loss": 2.5313,
+      "step": 86
+    },
+    {
+      "epoch": 0.23230974632843793,
+      "grad_norm": 0.28125,
+      "learning_rate": 9.126453628464889e-06,
+      "loss": 2.4813,
+      "step": 87
+    },
+    {
+      "epoch": 0.2349799732977303,
+      "grad_norm": 0.267578125,
+      "learning_rate": 9.10137638576423e-06,
+      "loss": 2.489,
+      "step": 88
+    },
+    {
+      "epoch": 0.2376502002670227,
+      "grad_norm": 0.2734375,
+      "learning_rate": 9.07597974920756e-06,
+      "loss": 2.5331,
+      "step": 89
+    },
+    {
+      "epoch": 0.24032042723631508,
+      "grad_norm": 0.265625,
+      "learning_rate": 9.05026569655281e-06,
+      "loss": 2.4919,
+      "step": 90
+    },
+    {
+      "epoch": 0.24299065420560748,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.02423623027663e-06,
+      "loss": 2.561,
+      "step": 91
+    },
+    {
+      "epoch": 0.24566088117489987,
+      "grad_norm": 0.26953125,
+      "learning_rate": 8.997893377418432e-06,
+      "loss": 2.5266,
+      "step": 92
+    },
+    {
+      "epoch": 0.24833110814419226,
+      "grad_norm": 0.265625,
+      "learning_rate": 8.971239189422555e-06,
+      "loss": 2.4969,
+      "step": 93
+    },
+    {
+      "epoch": 0.25100133511348466,
+      "grad_norm": 0.2734375,
+      "learning_rate": 8.944275741978495e-06,
+      "loss": 2.4977,
+      "step": 94
+    },
+    {
+      "epoch": 0.253671562082777,
+      "grad_norm": 0.271484375,
+      "learning_rate": 8.917005134859263e-06,
+      "loss": 2.4885,
+      "step": 95
+    },
+    {
+      "epoch": 0.25634178905206945,
+      "grad_norm": 0.26953125,
+      "learning_rate": 8.889429491757872e-06,
+      "loss": 2.4995,
+      "step": 96
+    },
+    {
+      "epoch": 0.2590120160213618,
+      "grad_norm": 0.271484375,
+      "learning_rate": 8.861550960121946e-06,
+      "loss": 2.5081,
+      "step": 97
+    },
+    {
+      "epoch": 0.2616822429906542,
+      "grad_norm": 0.265625,
+      "learning_rate": 8.833371710986493e-06,
+      "loss": 2.4995,
+      "step": 98
+    },
+    {
+      "epoch": 0.2643524699599466,
+      "grad_norm": 0.26171875,
+      "learning_rate": 8.804893938804839e-06,
+      "loss": 2.4881,
+      "step": 99
+    },
+    {
+      "epoch": 0.26702269692923897,
+      "grad_norm": 0.271484375,
+      "learning_rate": 8.77611986127773e-06,
+      "loss": 2.5327,
+      "step": 100
+    },
+    {
+      "epoch": 0.2696929238985314,
+      "grad_norm": 0.267578125,
+      "learning_rate": 8.747051719180626e-06,
+      "loss": 2.526,
+      "step": 101
+    },
+    {
+      "epoch": 0.27236315086782376,
+      "grad_norm": 0.26953125,
+      "learning_rate": 8.717691776189214e-06,
+      "loss": 2.5366,
+      "step": 102
+    },
+    {
+      "epoch": 0.2750333778371162,
+      "grad_norm": 0.259765625,
+      "learning_rate": 8.688042318703111e-06,
+      "loss": 2.4877,
+      "step": 103
+    },
+    {
+      "epoch": 0.27770360480640854,
+      "grad_norm": 0.263671875,
+      "learning_rate": 8.65810565566782e-06,
+      "loss": 2.4742,
+      "step": 104
+    },
+    {
+      "epoch": 0.2803738317757009,
+      "grad_norm": 0.263671875,
+      "learning_rate": 8.627884118394913e-06,
+      "loss": 2.4916,
+      "step": 105
+    },
+    {
+      "epoch": 0.28304405874499333,
+      "grad_norm": 0.263671875,
+      "learning_rate": 8.597380060380493e-06,
+      "loss": 2.536,
+      "step": 106
+    },
+    {
+      "epoch": 0.2857142857142857,
+      "grad_norm": 0.271484375,
+      "learning_rate": 8.566595857121902e-06,
+      "loss": 2.4748,
+      "step": 107
+    },
+    {
+      "epoch": 0.2883845126835781,
+      "grad_norm": 0.275390625,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 2.4833,
+      "step": 108
+    },
+    {
+      "epoch": 0.2910547396528705,
+      "grad_norm": 0.265625,
+      "learning_rate": 8.504196625756166e-06,
+      "loss": 2.5434,
+      "step": 109
+    },
+    {
+      "epoch": 0.2937249666221629,
+      "grad_norm": 0.26953125,
+      "learning_rate": 8.472586456976534e-06,
+      "loss": 2.4924,
+      "step": 110
+    },
+    {
+      "epoch": 0.2963951935914553,
+      "grad_norm": 0.267578125,
+      "learning_rate": 8.440705861229344e-06,
+      "loss": 2.5269,
+      "step": 111
+    },
+    {
+      "epoch": 0.29906542056074764,
+      "grad_norm": 0.26171875,
+      "learning_rate": 8.408557321209534e-06,
+      "loss": 2.4687,
+      "step": 112
+    },
+    {
+      "epoch": 0.30173564753004006,
+      "grad_norm": 0.263671875,
+      "learning_rate": 8.376143340478153e-06,
+      "loss": 2.4686,
+      "step": 113
+    },
+    {
+      "epoch": 0.30440587449933243,
+      "grad_norm": 0.27734375,
+      "learning_rate": 8.34346644326739e-06,
+      "loss": 2.4978,
+      "step": 114
+    },
+    {
+      "epoch": 0.30707610146862485,
+      "grad_norm": 0.263671875,
+      "learning_rate": 8.310529174284004e-06,
+      "loss": 2.5137,
+      "step": 115
+    },
+    {
+      "epoch": 0.3097463284379172,
+      "grad_norm": 0.27734375,
+      "learning_rate": 8.277334098511147e-06,
+      "loss": 2.5235,
+      "step": 116
+    },
+    {
+      "epoch": 0.31241655540720964,
+      "grad_norm": 0.275390625,
+      "learning_rate": 8.243883801008632e-06,
+      "loss": 2.4973,
+      "step": 117
+    },
+    {
+      "epoch": 0.315086782376502,
+      "grad_norm": 0.263671875,
+      "learning_rate": 8.210180886711603e-06,
+      "loss": 2.5124,
+      "step": 118
+    },
+    {
+      "epoch": 0.3177570093457944,
+      "grad_norm": 0.271484375,
+      "learning_rate": 8.176227980227693e-06,
+      "loss": 2.5167,
+      "step": 119
+    },
+    {
+      "epoch": 0.3204272363150868,
+      "grad_norm": 0.267578125,
+      "learning_rate": 8.142027725632622e-06,
+      "loss": 2.454,
+      "step": 120
+    },
+    {
+      "epoch": 0.32309746328437916,
+      "grad_norm": 0.267578125,
+      "learning_rate": 8.107582786264299e-06,
+      "loss": 2.5076,
+      "step": 121
+    },
+    {
+      "epoch": 0.3257676902536716,
+      "grad_norm": 0.265625,
+      "learning_rate": 8.072895844515398e-06,
+      "loss": 2.5122,
+      "step": 122
+    },
+    {
+      "epoch": 0.32843791722296395,
+      "grad_norm": 0.275390625,
+      "learning_rate": 8.037969601624495e-06,
+      "loss": 2.5294,
+      "step": 123
+    },
+    {
+      "epoch": 0.3311081441922563,
+      "grad_norm": 0.271484375,
+      "learning_rate": 8.002806777465685e-06,
+      "loss": 2.537,
+      "step": 124
+    },
+    {
+      "epoch": 0.33377837116154874,
+      "grad_norm": 0.263671875,
+      "learning_rate": 7.967410110336782e-06,
+      "loss": 2.4899,
+      "step": 125
+    },
+    {
+      "epoch": 0.3364485981308411,
+      "grad_norm": 0.267578125,
+      "learning_rate": 7.931782356746076e-06,
+      "loss": 2.5312,
+      "step": 126
+    },
+    {
+      "epoch": 0.3391188251001335,
+      "grad_norm": 0.267578125,
+      "learning_rate": 7.895926291197667e-06,
+      "loss": 2.4986,
+      "step": 127
+    },
+    {
+      "epoch": 0.3417890520694259,
+      "grad_norm": 0.267578125,
+      "learning_rate": 7.859844705975405e-06,
+      "loss": 2.5286,
+      "step": 128
+    },
+    {
+      "epoch": 0.3444592790387183,
+      "grad_norm": 0.26953125,
+      "learning_rate": 7.823540410925434e-06,
+      "loss": 2.4856,
+      "step": 129
+    },
+    {
+      "epoch": 0.3471295060080107,
+      "grad_norm": 0.263671875,
+      "learning_rate": 7.787016233237387e-06,
+      "loss": 2.5236,
+      "step": 130
+    },
+    {
+      "epoch": 0.34979973297730305,
+      "grad_norm": 0.2734375,
+      "learning_rate": 7.750275017224208e-06,
+      "loss": 2.5115,
+      "step": 131
+    },
+    {
+      "epoch": 0.35246995994659547,
+      "grad_norm": 0.267578125,
+      "learning_rate": 7.713319624100657e-06,
+      "loss": 2.5059,
+      "step": 132
+    },
+    {
+      "epoch": 0.35514018691588783,
+      "grad_norm": 0.26953125,
+      "learning_rate": 7.676152931760496e-06,
+      "loss": 2.5122,
+      "step": 133
+    },
+    {
+      "epoch": 0.35781041388518026,
+      "grad_norm": 0.267578125,
+      "learning_rate": 7.638777834552372e-06,
+      "loss": 2.5179,
+      "step": 134
+    },
+    {
+      "epoch": 0.3604806408544726,
+      "grad_norm": 0.271484375,
+      "learning_rate": 7.601197243054411e-06,
+      "loss": 2.4895,
+      "step": 135
+    },
+    {
+      "epoch": 0.36315086782376504,
+      "grad_norm": 0.26171875,
+      "learning_rate": 7.563414083847573e-06,
+      "loss": 2.4843,
+      "step": 136
+    },
+    {
+      "epoch": 0.3658210947930574,
+      "grad_norm": 0.271484375,
+      "learning_rate": 7.525431299287737e-06,
+      "loss": 2.5075,
+      "step": 137
+    },
+    {
+      "epoch": 0.3684913217623498,
+      "grad_norm": 0.265625,
+      "learning_rate": 7.4872518472765594e-06,
+      "loss": 2.5155,
+      "step": 138
+    },
+    {
+      "epoch": 0.3711615487316422,
+      "grad_norm": 0.263671875,
+      "learning_rate": 7.4488787010311425e-06,
+      "loss": 2.4949,
+      "step": 139
+    },
+    {
+      "epoch": 0.37383177570093457,
+      "grad_norm": 0.263671875,
+      "learning_rate": 7.4103148488524824e-06,
+      "loss": 2.5134,
+      "step": 140
+    },
+    {
+      "epoch": 0.376502002670227,
+      "grad_norm": 0.267578125,
+      "learning_rate": 7.371563293892761e-06,
+      "loss": 2.516,
+      "step": 141
+    },
+    {
+      "epoch": 0.37917222963951935,
+      "grad_norm": 0.26953125,
+      "learning_rate": 7.3326270539214826e-06,
+      "loss": 2.4839,
+      "step": 142
+    },
+    {
+      "epoch": 0.3818424566088118,
+      "grad_norm": 0.265625,
+      "learning_rate": 7.293509161090453e-06,
+      "loss": 2.4952,
+      "step": 143
+    },
+    {
+      "epoch": 0.38451268357810414,
+      "grad_norm": 0.263671875,
+      "learning_rate": 7.2542126616976596e-06,
+      "loss": 2.4828,
+      "step": 144
+    },
+    {
+      "epoch": 0.3871829105473965,
+      "grad_norm": 0.271484375,
+      "learning_rate": 7.214740615950041e-06,
+      "loss": 2.5365,
+      "step": 145
+    },
+    {
+      "epoch": 0.38985313751668893,
+      "grad_norm": 0.267578125,
+      "learning_rate": 7.175096097725169e-06,
+      "loss": 2.5147,
+      "step": 146
+    },
+    {
+      "epoch": 0.3925233644859813,
+      "grad_norm": 0.26171875,
+      "learning_rate": 7.135282194331881e-06,
+      "loss": 2.4745,
+      "step": 147
+    },
+    {
+      "epoch": 0.3951935914552737,
+      "grad_norm": 0.26171875,
+      "learning_rate": 7.095302006269842e-06,
+      "loss": 2.4743,
+      "step": 148
+    },
+    {
+      "epoch": 0.3978638184245661,
+      "grad_norm": 0.265625,
+      "learning_rate": 7.05515864698811e-06,
+      "loss": 2.4727,
+      "step": 149
+    },
+    {
+      "epoch": 0.40053404539385845,
+      "grad_norm": 0.267578125,
+      "learning_rate": 7.014855242642662e-06,
+      "loss": 2.4808,
+      "step": 150
+    },
+    {
+      "epoch": 0.4032042723631509,
+      "grad_norm": 0.265625,
+      "learning_rate": 6.974394931852957e-06,
+      "loss": 2.4861,
+      "step": 151
+    },
+    {
+      "epoch": 0.40587449933244324,
+      "grad_norm": 0.26953125,
+      "learning_rate": 6.933780865457508e-06,
+      "loss": 2.5099,
+      "step": 152
+    },
+    {
+      "epoch": 0.40854472630173566,
+      "grad_norm": 0.26171875,
+      "learning_rate": 6.893016206268518e-06,
+      "loss": 2.5029,
+      "step": 153
+    },
+    {
+      "epoch": 0.411214953271028,
+      "grad_norm": 0.267578125,
+      "learning_rate": 6.85210412882557e-06,
+      "loss": 2.496,
+      "step": 154
+    },
+    {
+      "epoch": 0.41388518024032045,
+      "grad_norm": 0.265625,
+      "learning_rate": 6.811047819148413e-06,
+      "loss": 2.4808,
+      "step": 155
+    },
+    {
+      "epoch": 0.4165554072096128,
+      "grad_norm": 0.267578125,
+      "learning_rate": 6.769850474488859e-06,
+      "loss": 2.5062,
+      "step": 156
+    },
+    {
+      "epoch": 0.4192256341789052,
+      "grad_norm": 0.263671875,
+      "learning_rate": 6.728515303081782e-06,
+      "loss": 2.4872,
+      "step": 157
+    },
+    {
+      "epoch": 0.4218958611481976,
+      "grad_norm": 0.263671875,
+      "learning_rate": 6.687045523895292e-06,
+      "loss": 2.5125,
+      "step": 158
+    },
+    {
+      "epoch": 0.42456608811748997,
+      "grad_norm": 0.2578125,
+      "learning_rate": 6.64544436638005e-06,
+      "loss": 2.4739,
+      "step": 159
+    },
+    {
+      "epoch": 0.4272363150867824,
+      "grad_norm": 0.26171875,
+      "learning_rate": 6.603715070217779e-06,
+      "loss": 2.4561,
+      "step": 160
+    },
+    {
+      "epoch": 0.42990654205607476,
+      "grad_norm": 0.267578125,
+      "learning_rate": 6.561860885068972e-06,
+      "loss": 2.5166,
+      "step": 161
+    },
+    {
+      "epoch": 0.4325767690253672,
+      "grad_norm": 0.26171875,
+      "learning_rate": 6.519885070319827e-06,
+      "loss": 2.4901,
+      "step": 162
+    },
+    {
+      "epoch": 0.43524699599465955,
+      "grad_norm": 0.2734375,
+      "learning_rate": 6.477790894828422e-06,
+      "loss": 2.4906,
+      "step": 163
+    },
+    {
+      "epoch": 0.4379172229639519,
+      "grad_norm": 0.265625,
+      "learning_rate": 6.435581636670154e-06,
+      "loss": 2.4935,
+      "step": 164
+    },
+    {
+      "epoch": 0.44058744993324434,
+      "grad_norm": 0.265625,
+      "learning_rate": 6.393260582882462e-06,
+      "loss": 2.4686,
+      "step": 165
+    },
+    {
+      "epoch": 0.4432576769025367,
+      "grad_norm": 0.26171875,
+      "learning_rate": 6.350831029208844e-06,
+      "loss": 2.4516,
+      "step": 166
+    },
+    {
+      "epoch": 0.4459279038718291,
+      "grad_norm": 0.26171875,
+      "learning_rate": 6.308296279842204e-06,
+      "loss": 2.479,
+      "step": 167
+    },
+    {
+      "epoch": 0.4485981308411215,
+      "grad_norm": 0.26171875,
+      "learning_rate": 6.265659647167542e-06,
+      "loss": 2.5068,
+      "step": 168
+    },
+    {
+      "epoch": 0.4512683578104139,
+      "grad_norm": 0.263671875,
+      "learning_rate": 6.222924451504001e-06,
+      "loss": 2.4598,
+      "step": 169
+    },
+    {
+      "epoch": 0.4539385847797063,
+      "grad_norm": 0.271484375,
+      "learning_rate": 6.180094020846291e-06,
+      "loss": 2.5116,
+      "step": 170
+    },
+    {
+      "epoch": 0.45660881174899864,
+      "grad_norm": 0.26953125,
+      "learning_rate": 6.1371716906055336e-06,
+      "loss": 2.4649,
+      "step": 171
+    },
+    {
+      "epoch": 0.45927903871829107,
+      "grad_norm": 0.26171875,
+      "learning_rate": 6.094160803349508e-06,
+      "loss": 2.4785,
+      "step": 172
+    },
+    {
+      "epoch": 0.46194926568758343,
+      "grad_norm": 0.26171875,
+      "learning_rate": 6.051064708542357e-06,
+      "loss": 2.5259,
+      "step": 173
+    },
+    {
+      "epoch": 0.46461949265687585,
+      "grad_norm": 0.271484375,
+      "learning_rate": 6.00788676228374e-06,
+      "loss": 2.5047,
+      "step": 174
+    },
+    {
+      "epoch": 0.4672897196261682,
+      "grad_norm": 0.267578125,
+      "learning_rate": 5.964630327047485e-06,
+      "loss": 2.5377,
+      "step": 175
+    },
+    {
+      "epoch": 0.4699599465954606,
+      "grad_norm": 0.267578125,
+      "learning_rate": 5.921298771419731e-06,
+      "loss": 2.5288,
+      "step": 176
+    },
+    {
+      "epoch": 0.472630173564753,
+      "grad_norm": 0.267578125,
+      "learning_rate": 5.877895469836604e-06,
+      "loss": 2.5045,
+      "step": 177
+    },
+    {
+      "epoch": 0.4753004005340454,
+      "grad_norm": 0.265625,
+      "learning_rate": 5.8344238023214305e-06,
+      "loss": 2.5275,
+      "step": 178
+    },
+    {
+      "epoch": 0.4779706275033378,
+      "grad_norm": 0.267578125,
+      "learning_rate": 5.790887154221521e-06,
+      "loss": 2.5307,
+      "step": 179
+    },
+    {
+      "epoch": 0.48064085447263016,
+      "grad_norm": 0.279296875,
+      "learning_rate": 5.747288915944533e-06,
+      "loss": 2.4982,
+      "step": 180
+    },
+    {
+      "epoch": 0.4833110814419226,
+      "grad_norm": 0.263671875,
+      "learning_rate": 5.703632482694453e-06,
+      "loss": 2.472,
+      "step": 181
+    },
+    {
+      "epoch": 0.48598130841121495,
+      "grad_norm": 0.271484375,
+      "learning_rate": 5.659921254207183e-06,
+      "loss": 2.5524,
+      "step": 182
+    },
+    {
+      "epoch": 0.4886515353805073,
+      "grad_norm": 0.267578125,
+      "learning_rate": 5.616158634485793e-06,
+      "loss": 2.4878,
+      "step": 183
+    },
+    {
+      "epoch": 0.49132176234979974,
+      "grad_norm": 0.2578125,
+      "learning_rate": 5.572348031535442e-06,
+      "loss": 2.5004,
+      "step": 184
+    },
+    {
+      "epoch": 0.4939919893190921,
+      "grad_norm": 0.255859375,
+      "learning_rate": 5.528492857097966e-06,
+      "loss": 2.4946,
+      "step": 185
+    },
+    {
+      "epoch": 0.49666221628838453,
+      "grad_norm": 0.263671875,
+      "learning_rate": 5.484596526386198e-06,
+      "loss": 2.4705,
+      "step": 186
+    },
+    {
+      "epoch": 0.4993324432576769,
+      "grad_norm": 0.263671875,
+      "learning_rate": 5.44066245781801e-06,
+      "loss": 2.5064,
+      "step": 187
+    },
+    {
+      "epoch": 0.5020026702269693,
+      "grad_norm": 0.267578125,
+      "learning_rate": 5.396694072750099e-06,
+      "loss": 2.4749,
+      "step": 188
+    },
+    {
+      "epoch": 0.5046728971962616,
+      "grad_norm": 0.263671875,
+      "learning_rate": 5.352694795211555e-06,
+      "loss": 2.5226,
+      "step": 189
+    },
+    {
+      "epoch": 0.507343124165554,
+      "grad_norm": 0.265625,
+      "learning_rate": 5.308668051637213e-06,
+      "loss": 2.5068,
+      "step": 190
+    },
+    {
+      "epoch": 0.5100133511348465,
+      "grad_norm": 0.2578125,
+      "learning_rate": 5.2646172706008154e-06,
+      "loss": 2.4547,
+      "step": 191
+    },
+    {
+      "epoch": 0.5126835781041389,
+      "grad_norm": 0.265625,
+      "learning_rate": 5.220545882548024e-06,
+      "loss": 2.4742,
+      "step": 192
+    },
+    {
+      "epoch": 0.5153538050734312,
+      "grad_norm": 0.26171875,
+      "learning_rate": 5.176457319529264e-06,
+      "loss": 2.5493,
+      "step": 193
+    },
+    {
+      "epoch": 0.5180240320427236,
+      "grad_norm": 0.263671875,
+      "learning_rate": 5.132355014932455e-06,
+      "loss": 2.5024,
+      "step": 194
+    },
+    {
+      "epoch": 0.520694259012016,
+      "grad_norm": 0.263671875,
+      "learning_rate": 5.088242403215644e-06,
+      "loss": 2.5104,
+      "step": 195
+    },
+    {
+      "epoch": 0.5233644859813084,
+      "grad_norm": 0.265625,
+      "learning_rate": 5.0441229196395416e-06,
+      "loss": 2.4975,
+      "step": 196
+    },
+    {
+      "epoch": 0.5260347129506008,
+      "grad_norm": 0.265625,
+      "learning_rate": 5e-06,
+      "loss": 2.4902,
+      "step": 197
+    },
+    {
+      "epoch": 0.5287049399198932,
+      "grad_norm": 0.2734375,
+      "learning_rate": 4.955877080360462e-06,
+      "loss": 2.5088,
+      "step": 198
+    },
+    {
+      "epoch": 0.5313751668891856,
+      "grad_norm": 0.267578125,
+      "learning_rate": 4.911757596784358e-06,
+      "loss": 2.5015,
+      "step": 199
+    },
+    {
+      "epoch": 0.5340453938584779,
+      "grad_norm": 0.26953125,
+      "learning_rate": 4.867644985067548e-06,
+      "loss": 2.5029,
+      "step": 200
+    },
+    {
+      "epoch": 0.5367156208277704,
+      "grad_norm": 0.26171875,
+      "learning_rate": 4.823542680470738e-06,
+      "loss": 2.5171,
+      "step": 201
+    },
+    {
+      "epoch": 0.5393858477970628,
+      "grad_norm": 0.259765625,
+      "learning_rate": 4.779454117451978e-06,
+      "loss": 2.5242,
+      "step": 202
+    },
+    {
+      "epoch": 0.5420560747663551,
+      "grad_norm": 0.263671875,
+      "learning_rate": 4.7353827293991845e-06,
+      "loss": 2.5121,
+      "step": 203
+    },
+    {
+      "epoch": 0.5447263017356475,
+      "grad_norm": 0.26171875,
+      "learning_rate": 4.691331948362789e-06,
+      "loss": 2.4662,
+      "step": 204
+    },
+    {
+      "epoch": 0.5473965287049399,
+      "grad_norm": 0.265625,
+      "learning_rate": 4.647305204788445e-06,
+      "loss": 2.5188,
+      "step": 205
+    },
+    {
+      "epoch": 0.5500667556742324,
+      "grad_norm": 0.26171875,
+      "learning_rate": 4.603305927249902e-06,
+      "loss": 2.4847,
+      "step": 206
+    },
+    {
+      "epoch": 0.5527369826435247,
+      "grad_norm": 0.2578125,
+      "learning_rate": 4.559337542181993e-06,
+      "loss": 2.4865,
+      "step": 207
+    },
+    {
+      "epoch": 0.5554072096128171,
+      "grad_norm": 0.26171875,
+      "learning_rate": 4.5154034736138035e-06,
+      "loss": 2.4858,
+      "step": 208
+    },
+    {
+      "epoch": 0.5580774365821095,
+      "grad_norm": 0.2578125,
+      "learning_rate": 4.471507142902036e-06,
+      "loss": 2.4824,
+      "step": 209
+    },
+    {
+      "epoch": 0.5607476635514018,
+      "grad_norm": 0.265625,
+      "learning_rate": 4.427651968464559e-06,
+      "loss": 2.515,
+      "step": 210
+    },
+    {
+      "epoch": 0.5634178905206942,
+      "grad_norm": 0.271484375,
+      "learning_rate": 4.383841365514208e-06,
+      "loss": 2.5034,
+      "step": 211
+    },
+    {
+      "epoch": 0.5660881174899867,
+      "grad_norm": 0.267578125,
+      "learning_rate": 4.340078745792818e-06,
+      "loss": 2.4928,
+      "step": 212
+    },
+    {
+      "epoch": 0.5687583444592791,
+      "grad_norm": 0.263671875,
+      "learning_rate": 4.296367517305548e-06,
+      "loss": 2.51,
+      "step": 213
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "grad_norm": 0.265625,
+      "learning_rate": 4.252711084055468e-06,
+      "loss": 2.4997,
+      "step": 214
+    },
+    {
+      "epoch": 0.5740987983978638,
+      "grad_norm": 0.265625,
+      "learning_rate": 4.209112845778481e-06,
+      "loss": 2.4813,
+      "step": 215
+    },
+    {
+      "epoch": 0.5767690253671562,
+      "grad_norm": 0.263671875,
+      "learning_rate": 4.165576197678571e-06,
+      "loss": 2.481,
+      "step": 216
+    },
+    {
+      "epoch": 0.5794392523364486,
+      "grad_norm": 0.259765625,
+      "learning_rate": 4.122104530163397e-06,
+      "loss": 2.4856,
+      "step": 217
+    },
+    {
+      "epoch": 0.582109479305741,
+      "grad_norm": 0.26171875,
+      "learning_rate": 4.0787012285802695e-06,
+      "loss": 2.5178,
+      "step": 218
+    },
+    {
+      "epoch": 0.5847797062750334,
+      "grad_norm": 0.265625,
+      "learning_rate": 4.035369672952516e-06,
+      "loss": 2.5174,
+      "step": 219
+    },
+    {
+      "epoch": 0.5874499332443258,
+      "grad_norm": 0.263671875,
+      "learning_rate": 3.992113237716261e-06,
+      "loss": 2.495,
+      "step": 220
+    },
+    {
+      "epoch": 0.5901201602136181,
+      "grad_norm": 0.267578125,
+      "learning_rate": 3.948935291457645e-06,
+      "loss": 2.5052,
+      "step": 221
+    },
+    {
+      "epoch": 0.5927903871829105,
+      "grad_norm": 0.26953125,
+      "learning_rate": 3.905839196650494e-06,
+      "loss": 2.5198,
+      "step": 222
+    },
+    {
+      "epoch": 0.595460614152203,
+      "grad_norm": 0.267578125,
+      "learning_rate": 3.862828309394469e-06,
+      "loss": 2.4906,
+      "step": 223
+    },
+    {
+      "epoch": 0.5981308411214953,
+      "grad_norm": 0.26171875,
+      "learning_rate": 3.8199059791537105e-06,
+      "loss": 2.5027,
+      "step": 224
+    },
+    {
+      "epoch": 0.6008010680907877,
+      "grad_norm": 0.267578125,
+      "learning_rate": 3.777075548496001e-06,
+      "loss": 2.4962,
+      "step": 225
+    },
+    {
+      "epoch": 0.6034712950600801,
+      "grad_norm": 0.271484375,
+      "learning_rate": 3.7343403528324574e-06,
+      "loss": 2.5218,
+      "step": 226
+    },
+    {
+      "epoch": 0.6061415220293725,
+      "grad_norm": 0.263671875,
+      "learning_rate": 3.6917037201577977e-06,
+      "loss": 2.5101,
+      "step": 227
+    },
+    {
+      "epoch": 0.6088117489986649,
+      "grad_norm": 0.265625,
+      "learning_rate": 3.649168970791157e-06,
+      "loss": 2.5235,
+      "step": 228
+    },
+    {
+      "epoch": 0.6114819759679573,
+      "grad_norm": 0.259765625,
+      "learning_rate": 3.6067394171175397e-06,
+      "loss": 2.4863,
+      "step": 229
+    },
+    {
+      "epoch": 0.6141522029372497,
+      "grad_norm": 0.267578125,
+      "learning_rate": 3.564418363329848e-06,
+      "loss": 2.5164,
+      "step": 230
+    },
+    {
+      "epoch": 0.616822429906542,
+      "grad_norm": 0.26171875,
+      "learning_rate": 3.5222091051715803e-06,
+      "loss": 2.5003,
+      "step": 231
+    },
+    {
+      "epoch": 0.6194926568758344,
+      "grad_norm": 0.263671875,
+      "learning_rate": 3.480114929680176e-06,
+      "loss": 2.5213,
+      "step": 232
+    },
+    {
+      "epoch": 0.6221628838451269,
+      "grad_norm": 0.265625,
+      "learning_rate": 3.4381391149310294e-06,
+      "loss": 2.5545,
+      "step": 233
+    },
+    {
+      "epoch": 0.6248331108144193,
+      "grad_norm": 0.27734375,
+      "learning_rate": 3.3962849297822225e-06,
+      "loss": 2.4738,
+      "step": 234
+    },
+    {
+      "epoch": 0.6275033377837116,
+      "grad_norm": 0.26171875,
+      "learning_rate": 3.35455563361995e-06,
+      "loss": 2.4812,
+      "step": 235
+    },
+    {
+      "epoch": 0.630173564753004,
+      "grad_norm": 0.26171875,
+      "learning_rate": 3.3129544761047093e-06,
+      "loss": 2.4738,
+      "step": 236
+    },
+    {
+      "epoch": 0.6328437917222964,
+      "grad_norm": 0.263671875,
+      "learning_rate": 3.271484696918218e-06,
+      "loss": 2.5361,
+      "step": 237
+    },
+    {
+      "epoch": 0.6355140186915887,
+      "grad_norm": 0.2578125,
+      "learning_rate": 3.2301495255111426e-06,
+      "loss": 2.4995,
+      "step": 238
+    },
+    {
+      "epoch": 0.6381842456608812,
+      "grad_norm": 0.267578125,
+      "learning_rate": 3.1889521808515888e-06,
+      "loss": 2.5055,
+      "step": 239
+    },
+    {
+      "epoch": 0.6408544726301736,
+      "grad_norm": 0.26171875,
+      "learning_rate": 3.1478958711744324e-06,
+      "loss": 2.4533,
+      "step": 240
+    },
+    {
+      "epoch": 0.6435246995994659,
+      "grad_norm": 0.26171875,
+      "learning_rate": 3.1069837937314846e-06,
+      "loss": 2.4603,
+      "step": 241
+    },
+    {
+      "epoch": 0.6461949265687583,
+      "grad_norm": 0.265625,
+      "learning_rate": 3.0662191345424925e-06,
+      "loss": 2.5259,
+      "step": 242
+    },
+    {
+      "epoch": 0.6488651535380507,
+      "grad_norm": 0.26171875,
+      "learning_rate": 3.0256050681470446e-06,
+      "loss": 2.5498,
+      "step": 243
+    },
+    {
+      "epoch": 0.6515353805073432,
+      "grad_norm": 0.259765625,
+      "learning_rate": 2.9851447573573383e-06,
+      "loss": 2.4606,
+      "step": 244
+    },
+    {
+      "epoch": 0.6542056074766355,
+      "grad_norm": 0.26953125,
+      "learning_rate": 2.9448413530118912e-06,
+      "loss": 2.5477,
+      "step": 245
+    },
+    {
+      "epoch": 0.6568758344459279,
+      "grad_norm": 0.265625,
+      "learning_rate": 2.904697993730159e-06,
+      "loss": 2.5476,
+      "step": 246
+    },
+    {
+      "epoch": 0.6595460614152203,
+      "grad_norm": 0.263671875,
+      "learning_rate": 2.8647178056681197e-06,
+      "loss": 2.5033,
+      "step": 247
+    },
+    {
+      "epoch": 0.6622162883845126,
+      "grad_norm": 0.2578125,
+      "learning_rate": 2.8249039022748315e-06,
+      "loss": 2.4971,
+      "step": 248
+    },
+    {
+      "epoch": 0.664886515353805,
+      "grad_norm": 0.267578125,
+      "learning_rate": 2.785259384049959e-06,
+      "loss": 2.4596,
+      "step": 249
+    },
+    {
+      "epoch": 0.6675567423230975,
+      "grad_norm": 0.263671875,
+      "learning_rate": 2.745787338302341e-06,
+      "loss": 2.4948,
+      "step": 250
+    },
+    {
+      "epoch": 0.6702269692923899,
+      "grad_norm": 0.267578125,
+      "learning_rate": 2.706490838909547e-06,
+      "loss": 2.4986,
+      "step": 251
+    },
+    {
+      "epoch": 0.6728971962616822,
+      "grad_norm": 0.263671875,
+      "learning_rate": 2.6673729460785174e-06,
+      "loss": 2.4885,
+      "step": 252
+    },
+    {
+      "epoch": 0.6755674232309746,
+      "grad_norm": 0.26953125,
+      "learning_rate": 2.628436706107238e-06,
+      "loss": 2.5042,
+      "step": 253
+    },
+    {
+      "epoch": 0.678237650200267,
+      "grad_norm": 0.26171875,
+      "learning_rate": 2.5896851511475184e-06,
+      "loss": 2.5229,
+      "step": 254
+    },
+    {
+      "epoch": 0.6809078771695594,
+      "grad_norm": 0.2578125,
+      "learning_rate": 2.5511212989688587e-06,
+      "loss": 2.4748,
+      "step": 255
+    },
+    {
+      "epoch": 0.6835781041388518,
+      "grad_norm": 0.26171875,
+      "learning_rate": 2.5127481527234397e-06,
+      "loss": 2.4837,
+      "step": 256
+    },
+    {
+      "epoch": 0.6862483311081442,
+      "grad_norm": 0.26171875,
+      "learning_rate": 2.4745687007122636e-06,
+      "loss": 2.5272,
+      "step": 257
+    },
+    {
+      "epoch": 0.6889185580774366,
+      "grad_norm": 0.26171875,
+      "learning_rate": 2.436585916152426e-06,
+      "loss": 2.4953,
+      "step": 258
+    },
+    {
+      "epoch": 0.6915887850467289,
+      "grad_norm": 0.255859375,
+      "learning_rate": 2.3988027569455895e-06,
+      "loss": 2.4866,
+      "step": 259
+    },
+    {
+      "epoch": 0.6942590120160214,
+      "grad_norm": 0.26171875,
+      "learning_rate": 2.361222165447628e-06,
+      "loss": 2.5,
+      "step": 260
+    },
+    {
+      "epoch": 0.6969292389853138,
+      "grad_norm": 0.259765625,
+      "learning_rate": 2.323847068239504e-06,
+      "loss": 2.5211,
+      "step": 261
+    },
+    {
+      "epoch": 0.6995994659546061,
+      "grad_norm": 0.271484375,
+      "learning_rate": 2.2866803758993446e-06,
+      "loss": 2.5103,
+      "step": 262
+    },
+    {
+      "epoch": 0.7022696929238985,
+      "grad_norm": 0.265625,
+      "learning_rate": 2.2497249827757933e-06,
+      "loss": 2.5008,
+      "step": 263
+    },
+    {
+      "epoch": 0.7049399198931909,
+      "grad_norm": 0.2578125,
+      "learning_rate": 2.2129837667626147e-06,
+      "loss": 2.4844,
+      "step": 264
+    },
+    {
+      "epoch": 0.7076101468624834,
+      "grad_norm": 0.26953125,
+      "learning_rate": 2.176459589074566e-06,
+      "loss": 2.5163,
+      "step": 265
+    },
+    {
+      "epoch": 0.7102803738317757,
+      "grad_norm": 0.265625,
+      "learning_rate": 2.1401552940245962e-06,
+      "loss": 2.5074,
+      "step": 266
+    },
+    {
+      "epoch": 0.7129506008010681,
+      "grad_norm": 0.265625,
+      "learning_rate": 2.1040737088023323e-06,
+      "loss": 2.4936,
+      "step": 267
+    },
+    {
+      "epoch": 0.7156208277703605,
+      "grad_norm": 0.265625,
+      "learning_rate": 2.068217643253925e-06,
+      "loss": 2.491,
+      "step": 268
+    },
+    {
+      "epoch": 0.7182910547396528,
+      "grad_norm": 0.265625,
+      "learning_rate": 2.0325898896632178e-06,
+      "loss": 2.5246,
+      "step": 269
+    },
+    {
+      "epoch": 0.7209612817089452,
+      "grad_norm": 0.267578125,
+      "learning_rate": 1.997193222534316e-06,
+      "loss": 2.5156,
+      "step": 270
+    },
+    {
+      "epoch": 0.7236315086782377,
+      "grad_norm": 0.267578125,
+      "learning_rate": 1.962030398375506e-06,
+      "loss": 2.4926,
+      "step": 271
+    },
+    {
+      "epoch": 0.7263017356475301,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.927104155484602e-06,
+      "loss": 2.5143,
+      "step": 272
+    },
+    {
+      "epoch": 0.7289719626168224,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.8924172137357038e-06,
+      "loss": 2.4877,
+      "step": 273
+    },
+    {
+      "epoch": 0.7316421895861148,
+      "grad_norm": 0.2734375,
+      "learning_rate": 1.8579722743673773e-06,
+      "loss": 2.513,
+      "step": 274
+    },
+    {
+      "epoch": 0.7343124165554072,
+      "grad_norm": 0.263671875,
+      "learning_rate": 1.8237720197723075e-06,
+      "loss": 2.5352,
+      "step": 275
+    },
+    {
+      "epoch": 0.7369826435246996,
+      "grad_norm": 0.263671875,
+      "learning_rate": 1.789819113288397e-06,
+      "loss": 2.4835,
+      "step": 276
+    },
+    {
+      "epoch": 0.739652870493992,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.75611619899137e-06,
+      "loss": 2.4761,
+      "step": 277
+    },
+    {
+      "epoch": 0.7423230974632844,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.7226659014888548e-06,
+      "loss": 2.4956,
+      "step": 278
+    },
+    {
+      "epoch": 0.7449933244325768,
+      "grad_norm": 0.265625,
+      "learning_rate": 1.689470825715998e-06,
+      "loss": 2.5053,
+      "step": 279
+    },
+    {
+      "epoch": 0.7476635514018691,
+      "grad_norm": 0.263671875,
+      "learning_rate": 1.6565335567326112e-06,
+      "loss": 2.4997,
+      "step": 280
+    },
+    {
+      "epoch": 0.7503337783711616,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.6238566595218475e-06,
+      "loss": 2.5431,
+      "step": 281
+    },
+    {
+      "epoch": 0.753004005340454,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.591442678790467e-06,
+      "loss": 2.4902,
+      "step": 282
+    },
+    {
+      "epoch": 0.7556742323097463,
+      "grad_norm": 0.2734375,
+      "learning_rate": 1.5592941387706562e-06,
+      "loss": 2.5166,
+      "step": 283
+    },
+    {
+      "epoch": 0.7583444592790387,
+      "grad_norm": 0.271484375,
+      "learning_rate": 1.5274135430234654e-06,
+      "loss": 2.501,
+      "step": 284
+    },
+    {
+      "epoch": 0.7610146862483311,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.4958033742438348e-06,
+      "loss": 2.5154,
+      "step": 285
+    },
+    {
+      "epoch": 0.7636849132176236,
+      "grad_norm": 0.294921875,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 2.5048,
+      "step": 286
+    },
+    {
+      "epoch": 0.7663551401869159,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.4334041428781003e-06,
+      "loss": 2.4991,
+      "step": 287
+    },
+    {
+      "epoch": 0.7690253671562083,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.4026199396195078e-06,
+      "loss": 2.488,
+      "step": 288
+    },
+    {
+      "epoch": 0.7716955941255007,
+      "grad_norm": 0.263671875,
+      "learning_rate": 1.3721158816050872e-06,
+      "loss": 2.4893,
+      "step": 289
+    },
+    {
+      "epoch": 0.774365821094793,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.3418943443321807e-06,
+      "loss": 2.5037,
+      "step": 290
+    },
+    {
+      "epoch": 0.7770360480640854,
+      "grad_norm": 0.265625,
+      "learning_rate": 1.3119576812968893e-06,
+      "loss": 2.4731,
+      "step": 291
+    },
+    {
+      "epoch": 0.7797062750333779,
+      "grad_norm": 0.267578125,
+      "learning_rate": 1.282308223810786e-06,
+      "loss": 2.503,
+      "step": 292
+    },
+    {
+      "epoch": 0.7823765020026703,
+      "grad_norm": 0.265625,
+      "learning_rate": 1.252948280819375e-06,
+      "loss": 2.5442,
+      "step": 293
+    },
+    {
+      "epoch": 0.7850467289719626,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.2238801387222716e-06,
+      "loss": 2.4877,
+      "step": 294
+    },
+    {
+      "epoch": 0.787716955941255,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.1951060611951615e-06,
+      "loss": 2.5067,
+      "step": 295
+    },
+    {
+      "epoch": 0.7903871829105474,
+      "grad_norm": 0.26953125,
+      "learning_rate": 1.1666282890135083e-06,
+      "loss": 2.5179,
+      "step": 296
+    },
+    {
+      "epoch": 0.7930574098798397,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.1384490398780563e-06,
+      "loss": 2.4758,
+      "step": 297
+    },
+    {
+      "epoch": 0.7957276368491322,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.1105705082421303e-06,
+      "loss": 2.4833,
+      "step": 298
+    },
+    {
+      "epoch": 0.7983978638184246,
+      "grad_norm": 0.263671875,
+      "learning_rate": 1.0829948651407374e-06,
+      "loss": 2.4751,
+      "step": 299
+    },
+    {
+      "epoch": 0.8010680907877169,
+      "grad_norm": 0.255859375,
+      "learning_rate": 1.0557242580215066e-06,
+      "loss": 2.4916,
+      "step": 300
+    },
+    {
+      "epoch": 0.8037383177570093,
+      "grad_norm": 0.263671875,
+      "learning_rate": 1.0287608105774456e-06,
+      "loss": 2.5196,
+      "step": 301
+    },
+    {
+      "epoch": 0.8064085447263017,
+      "grad_norm": 0.255859375,
+      "learning_rate": 1.002106622581569e-06,
+      "loss": 2.5008,
+      "step": 302
+    },
+    {
+      "epoch": 0.8090787716955942,
+      "grad_norm": 0.28515625,
+      "learning_rate": 9.757637697233723e-07,
+      "loss": 2.4333,
+      "step": 303
+    },
+    {
+      "epoch": 0.8117489986648865,
+      "grad_norm": 0.267578125,
+      "learning_rate": 9.497343034471896e-07,
+      "loss": 2.5224,
+      "step": 304
+    },
+    {
+      "epoch": 0.8144192256341789,
+      "grad_norm": 0.263671875,
+      "learning_rate": 9.240202507924412e-07,
+      "loss": 2.5119,
+      "step": 305
+    },
+    {
+      "epoch": 0.8170894526034713,
+      "grad_norm": 0.259765625,
+      "learning_rate": 8.986236142357707e-07,
+      "loss": 2.5334,
+      "step": 306
+    },
+    {
+      "epoch": 0.8197596795727636,
+      "grad_norm": 0.2578125,
+      "learning_rate": 8.735463715351139e-07,
+      "loss": 2.4918,
+      "step": 307
+    },
+    {
+      "epoch": 0.822429906542056,
+      "grad_norm": 0.263671875,
+      "learning_rate": 8.487904755756676e-07,
+      "loss": 2.4838,
+      "step": 308
+    },
+    {
+      "epoch": 0.8251001335113485,
+      "grad_norm": 0.265625,
+      "learning_rate": 8.243578542178227e-07,
+      "loss": 2.5301,
+      "step": 309
+    },
+    {
+      "epoch": 0.8277703604806409,
+      "grad_norm": 0.259765625,
+      "learning_rate": 8.002504101470204e-07,
+      "loss": 2.5125,
+      "step": 310
+    },
+    {
+      "epoch": 0.8304405874499332,
+      "grad_norm": 0.25390625,
+      "learning_rate": 7.764700207255904e-07,
+      "loss": 2.5022,
+      "step": 311
+    },
+    {
+      "epoch": 0.8331108144192256,
+      "grad_norm": 0.263671875,
+      "learning_rate": 7.530185378465459e-07,
+      "loss": 2.527,
+      "step": 312
+    },
+    {
+      "epoch": 0.835781041388518,
+      "grad_norm": 0.26171875,
+      "learning_rate": 7.298977877893688e-07,
+      "loss": 2.5099,
+      "step": 313
+    },
+    {
+      "epoch": 0.8384512683578104,
+      "grad_norm": 0.26171875,
+      "learning_rate": 7.071095710777925e-07,
+      "loss": 2.4672,
+      "step": 314
+    },
+    {
+      "epoch": 0.8411214953271028,
+      "grad_norm": 0.259765625,
+      "learning_rate": 6.846556623395795e-07,
+      "loss": 2.4596,
+      "step": 315
+    },
+    {
+      "epoch": 0.8437917222963952,
+      "grad_norm": 0.265625,
+      "learning_rate": 6.625378101683317e-07,
+      "loss": 2.5339,
+      "step": 316
+    },
+    {
+      "epoch": 0.8464619492656876,
+      "grad_norm": 0.259765625,
+      "learning_rate": 6.40757736987307e-07,
+      "loss": 2.4655,
+      "step": 317
+    },
+    {
+      "epoch": 0.8491321762349799,
+      "grad_norm": 0.263671875,
+      "learning_rate": 6.193171389152996e-07,
+      "loss": 2.5052,
+      "step": 318
+    },
+    {
+      "epoch": 0.8518024032042724,
+      "grad_norm": 0.267578125,
+      "learning_rate": 5.982176856345445e-07,
+      "loss": 2.5025,
+      "step": 319
+    },
+    {
+      "epoch": 0.8544726301735648,
+      "grad_norm": 0.26953125,
+      "learning_rate": 5.774610202606939e-07,
+      "loss": 2.4683,
+      "step": 320
+    },
+    {
+      "epoch": 0.8571428571428571,
+      "grad_norm": 0.26171875,
+      "learning_rate": 5.570487592148666e-07,
+      "loss": 2.4735,
+      "step": 321
+    },
+    {
+      "epoch": 0.8598130841121495,
+      "grad_norm": 0.265625,
+      "learning_rate": 5.369824920977567e-07,
+      "loss": 2.5021,
+      "step": 322
+    },
+    {
+      "epoch": 0.8624833110814419,
+      "grad_norm": 0.263671875,
+      "learning_rate": 5.172637815658583e-07,
+      "loss": 2.5157,
+      "step": 323
+    },
+    {
+      "epoch": 0.8651535380507344,
+      "grad_norm": 0.2578125,
+      "learning_rate": 4.978941632097612e-07,
+      "loss": 2.4827,
+      "step": 324
+    },
+    {
+      "epoch": 0.8678237650200267,
+      "grad_norm": 0.259765625,
+      "learning_rate": 4.788751454345763e-07,
+      "loss": 2.4453,
+      "step": 325
+    },
+    {
+      "epoch": 0.8704939919893191,
+      "grad_norm": 0.259765625,
+      "learning_rate": 4.60208209342462e-07,
+      "loss": 2.4767,
+      "step": 326
+    },
+    {
+      "epoch": 0.8731642189586115,
+      "grad_norm": 0.267578125,
+      "learning_rate": 4.4189480861729137e-07,
+      "loss": 2.5088,
+      "step": 327
+    },
+    {
+      "epoch": 0.8758344459279038,
+      "grad_norm": 0.263671875,
+      "learning_rate": 4.239363694114368e-07,
+      "loss": 2.4893,
+      "step": 328
+    },
+    {
+      "epoch": 0.8785046728971962,
+      "grad_norm": 0.26171875,
+      "learning_rate": 4.0633429023472004e-07,
+      "loss": 2.5054,
+      "step": 329
+    },
+    {
+      "epoch": 0.8811748998664887,
+      "grad_norm": 0.263671875,
+      "learning_rate": 3.890899418454913e-07,
+      "loss": 2.4951,
+      "step": 330
+    },
+    {
+      "epoch": 0.8838451268357811,
+      "grad_norm": 0.26171875,
+      "learning_rate": 3.72204667143895e-07,
+      "loss": 2.4751,
+      "step": 331
+    },
+    {
+      "epoch": 0.8865153538050734,
+      "grad_norm": 0.259765625,
+      "learning_rate": 3.556797810672785e-07,
+      "loss": 2.488,
+      "step": 332
+    },
+    {
+      "epoch": 0.8891855807743658,
+      "grad_norm": 0.26171875,
+      "learning_rate": 3.395165704878023e-07,
+      "loss": 2.5096,
+      "step": 333
+    },
+    {
+      "epoch": 0.8918558077436582,
+      "grad_norm": 0.26171875,
+      "learning_rate": 3.237162941122185e-07,
+      "loss": 2.4998,
+      "step": 334
+    },
+    {
+      "epoch": 0.8945260347129506,
+      "grad_norm": 0.255859375,
+      "learning_rate": 3.082801823838527e-07,
+      "loss": 2.4736,
+      "step": 335
+    },
+    {
+      "epoch": 0.897196261682243,
+      "grad_norm": 0.265625,
+      "learning_rate": 2.932094373867811e-07,
+      "loss": 2.5098,
+      "step": 336
+    },
+    {
+      "epoch": 0.8998664886515354,
+      "grad_norm": 0.259765625,
+      "learning_rate": 2.785052327522214e-07,
+      "loss": 2.4961,
+      "step": 337
+    },
+    {
+      "epoch": 0.9025367156208278,
+      "grad_norm": 0.26171875,
+      "learning_rate": 2.6416871356713224e-07,
+      "loss": 2.5133,
+      "step": 338
+    },
+    {
+      "epoch": 0.9052069425901201,
+      "grad_norm": 0.267578125,
+      "learning_rate": 2.5020099628504603e-07,
+      "loss": 2.4595,
+      "step": 339
+    },
+    {
+      "epoch": 0.9078771695594126,
+      "grad_norm": 0.263671875,
+      "learning_rate": 2.3660316863911682e-07,
+      "loss": 2.5258,
+      "step": 340
+    },
+    {
+      "epoch": 0.910547396528705,
+      "grad_norm": 0.265625,
+      "learning_rate": 2.2337628955742263e-07,
+      "loss": 2.4918,
+      "step": 341
+    },
+    {
+      "epoch": 0.9132176234979973,
+      "grad_norm": 0.259765625,
+      "learning_rate": 2.1052138908049303e-07,
+      "loss": 2.5225,
+      "step": 342
+    },
+    {
+      "epoch": 0.9158878504672897,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.9803946828110376e-07,
+      "loss": 2.4688,
+      "step": 343
+    },
+    {
+      "epoch": 0.9185580774365821,
+      "grad_norm": 0.255859375,
+      "learning_rate": 1.8593149918630927e-07,
+      "loss": 2.4878,
+      "step": 344
+    },
+    {
+      "epoch": 0.9212283044058746,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.7419842470175196e-07,
+      "loss": 2.5185,
+      "step": 345
+    },
+    {
+      "epoch": 0.9238985313751669,
+      "grad_norm": 0.265625,
+      "learning_rate": 1.6284115853823445e-07,
+      "loss": 2.5176,
+      "step": 346
+    },
+    {
+      "epoch": 0.9265687583444593,
+      "grad_norm": 0.2578125,
+      "learning_rate": 1.5186058514055912e-07,
+      "loss": 2.4453,
+      "step": 347
+    },
+    {
+      "epoch": 0.9292389853137517,
+      "grad_norm": 0.265625,
+      "learning_rate": 1.4125755961865827e-07,
+      "loss": 2.5187,
+      "step": 348
+    },
+    {
+      "epoch": 0.931909212283044,
+      "grad_norm": 0.271484375,
+      "learning_rate": 1.3103290768099796e-07,
+      "loss": 2.5128,
+      "step": 349
+    },
+    {
+      "epoch": 0.9345794392523364,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.2118742557027885e-07,
+      "loss": 2.5018,
+      "step": 350
+    },
+    {
+      "epoch": 0.9372496662216289,
+      "grad_norm": 0.26953125,
+      "learning_rate": 1.1172188000142803e-07,
+      "loss": 2.5313,
+      "step": 351
+    },
+    {
+      "epoch": 0.9399198931909212,
+      "grad_norm": 0.263671875,
+      "learning_rate": 1.026370081018907e-07,
+      "loss": 2.5207,
+      "step": 352
+    },
+    {
+      "epoch": 0.9425901201602136,
+      "grad_norm": 0.2578125,
+      "learning_rate": 9.393351735422773e-08,
+      "loss": 2.5082,
+      "step": 353
+    },
+    {
+      "epoch": 0.945260347129506,
+      "grad_norm": 0.2578125,
+      "learning_rate": 8.561208554101863e-08,
+      "loss": 2.4809,
+      "step": 354
+    },
+    {
+      "epoch": 0.9479305740987984,
+      "grad_norm": 0.267578125,
+      "learning_rate": 7.76733606920832e-08,
+      "loss": 2.5182,
+      "step": 355
+    },
+    {
+      "epoch": 0.9506008010680908,
+      "grad_norm": 0.259765625,
+      "learning_rate": 7.011796103401192e-08,
+      "loss": 2.4804,
+      "step": 356
+    },
+    {
+      "epoch": 0.9532710280373832,
+      "grad_norm": 0.263671875,
+      "learning_rate": 6.294647494202444e-08,
+      "loss": 2.4949,
+      "step": 357
+    },
+    {
+      "epoch": 0.9559412550066756,
+      "grad_norm": 0.26171875,
+      "learning_rate": 5.615946089414737e-08,
+      "loss": 2.5131,
+      "step": 358
+    },
+    {
+      "epoch": 0.9586114819759679,
+      "grad_norm": 0.259765625,
+      "learning_rate": 4.975744742772848e-08,
+      "loss": 2.4944,
+      "step": 359
+    },
+    {
+      "epoch": 0.9612817089452603,
+      "grad_norm": 0.259765625,
+      "learning_rate": 4.37409330982691e-08,
+      "loss": 2.5229,
+      "step": 360
+    },
+    {
+      "epoch": 0.9639519359145527,
+      "grad_norm": 0.265625,
+      "learning_rate": 3.8110386440605164e-08,
+      "loss": 2.5288,
+      "step": 361
+    },
+    {
+      "epoch": 0.9666221628838452,
+      "grad_norm": 0.265625,
+      "learning_rate": 3.2866245932418606e-08,
+      "loss": 2.4581,
+      "step": 362
+    },
+    {
+      "epoch": 0.9692923898531375,
+      "grad_norm": 0.267578125,
+      "learning_rate": 2.8008919960090253e-08,
+      "loss": 2.5052,
+      "step": 363
+    },
+    {
+      "epoch": 0.9719626168224299,
+      "grad_norm": 0.265625,
+      "learning_rate": 2.3538786786896918e-08,
+      "loss": 2.5043,
+      "step": 364
+    },
+    {
+      "epoch": 0.9746328437917223,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.9456194523554404e-08,
+      "loss": 2.4863,
+      "step": 365
+    },
+    {
+      "epoch": 0.9773030707610146,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.576146110111032e-08,
+      "loss": 2.5012,
+      "step": 366
+    },
+    {
+      "epoch": 0.9799732977303071,
+      "grad_norm": 0.2578125,
+      "learning_rate": 1.2454874246181081e-08,
+      "loss": 2.516,
+      "step": 367
+    },
+    {
+      "epoch": 0.9826435246995995,
+      "grad_norm": 0.267578125,
+      "learning_rate": 9.536691458548741e-09,
+      "loss": 2.5053,
+      "step": 368
+    },
+    {
+      "epoch": 0.9853137516688919,
+      "grad_norm": 0.26171875,
+      "learning_rate": 7.007139991108136e-09,
+      "loss": 2.4899,
+      "step": 369
+    },
+    {
+      "epoch": 0.9879839786381842,
+      "grad_norm": 0.26171875,
+      "learning_rate": 4.866416832167153e-09,
+      "loss": 2.4932,
+      "step": 370
+    },
+    {
+      "epoch": 0.9906542056074766,
+      "grad_norm": 0.259765625,
+      "learning_rate": 3.1146886901090024e-09,
+      "loss": 2.5244,
+      "step": 371
+    },
+    {
+      "epoch": 0.9933244325767691,
+      "grad_norm": 0.259765625,
+      "learning_rate": 1.7520919804075997e-09,
+      "loss": 2.494,
+      "step": 372
+    },
+    {
+      "epoch": 0.9959946595460614,
+      "grad_norm": 0.26171875,
+      "learning_rate": 7.787328150071771e-10,
+      "loss": 2.4716,
+      "step": 373
+    },
+    {
+      "epoch": 0.9986648865153538,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.9468699405444936e-10,
+      "loss": 2.4657,
+      "step": 374
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 374,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "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": 4.698021672895119e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-374/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59d09c40ae667149d4baa83fd68385fe4aa5cfecd5f3c620712e3f588661904a
+size 6712