Upload folder using huggingface_hub

checkpoint-12208/config.json

@@ -0,0 +1,39 @@
+{
+  "_name_or_path": "meta-llama/Llama-3.1-8B-Instruct",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": [
+    128001,
+    128008,
+    128009
+  ],
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.44.0.dev0",
+  "use_cache": true,
+  "vocab_size": 128256
+}

checkpoint-12208/generation_config.json

@@ -0,0 +1,12 @@
+{
+  "bos_token_id": 128000,
+  "do_sample": true,
+  "eos_token_id": [
+    128001,
+    128008,
+    128009
+  ],
+  "temperature": 0.6,
+  "top_p": 0.9,
+  "transformers_version": "4.44.0.dev0"
+}

checkpoint-12208/model-00001-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d7aafe91b943225fdf31015a8c2d1e89599fba12c149feb5814336f5bf7297e6
+size 4886466168

checkpoint-12208/model-00002-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23fa5aaf8ac56c17c821980c9aee44fcbc2a617fc570e875e6fdb93d92886c72
+size 4832007448

checkpoint-12208/model-00003-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a223e87240e50dfb05634ea909cf1ec715d37a12b681b14a3a6e97f28688f2b
+size 4999813112

checkpoint-12208/model-00004-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3047d199c611c6d12b1cca780f033ad89df9c076f742674f7ac4a892ca9573b3
+size 4999813128

checkpoint-12208/model-00005-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a2f91d749d664ecd22c71f3d54488966fc6899a03924b8fea3613c8f83398a8
+size 4832007496

checkpoint-12208/model-00006-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d46422acb709980d248e047821977a961813fdeb1677dcbfad82e069c7eccbe6
+size 4999813120

checkpoint-12208/model-00007-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63be41870fa1c0cf75d08e1ccd55f8e568566e71220cf22b58113a7f1a3f241c
+size 2571158184

checkpoint-12208/model.safetensors.index.json

@@ -0,0 +1,298 @@
+{
+  "metadata": {
+    "total_size": 32121044992
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00007-of-00007.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00007-of-00007.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00007-of-00007.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00007-of-00007.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00007-of-00007.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.norm.weight": "model-00007-of-00007.safetensors"
+  }
+}

checkpoint-12208/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:012978646538badb26719dd07cb3108bffb6ba505cf96ce3b5e1998904027ee8
+size 15385036334

checkpoint-12208/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9d9cd6a0487226e5bd30d1846894c82af483733ab4381b75bae9c0745e05d405
+size 14244

checkpoint-12208/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abe644ed33a3c4139223f0857a985127f3e6fbaa8c89fa14b57671b49ca52c21
+size 1064

checkpoint-12208/trainer_state.json

@@ -0,0 +1,2784 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.1620640330818783,
+  "eval_steps": 500,
+  "global_step": 12208,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00041153219409716807,
+      "grad_norm": 4.147768020629883,
+      "learning_rate": 1.0157273918741808e-06,
+      "loss": 0.7978,
+      "step": 31
+    },
+    {
+      "epoch": 0.0008230643881943361,
+      "grad_norm": 3.117840051651001,
+      "learning_rate": 2.0314547837483616e-06,
+      "loss": 0.7723,
+      "step": 62
+    },
+    {
+      "epoch": 0.0012345965822915042,
+      "grad_norm": 2.83072829246521,
+      "learning_rate": 3.0471821756225426e-06,
+      "loss": 0.7493,
+      "step": 93
+    },
+    {
+      "epoch": 0.0016461287763886723,
+      "grad_norm": 2.9491968154907227,
+      "learning_rate": 4.062909567496723e-06,
+      "loss": 0.7352,
+      "step": 124
+    },
+    {
+      "epoch": 0.0020576609704858403,
+      "grad_norm": 2.798553228378296,
+      "learning_rate": 5.078636959370905e-06,
+      "loss": 0.7229,
+      "step": 155
+    },
+    {
+      "epoch": 0.0024691931645830084,
+      "grad_norm": 2.663148880004883,
+      "learning_rate": 6.094364351245085e-06,
+      "loss": 0.7253,
+      "step": 186
+    },
+    {
+      "epoch": 0.0028807253586801765,
+      "grad_norm": 2.564692974090576,
+      "learning_rate": 7.110091743119267e-06,
+      "loss": 0.7287,
+      "step": 217
+    },
+    {
+      "epoch": 0.0032922575527773446,
+      "grad_norm": 3.2043237686157227,
+      "learning_rate": 8.125819134993446e-06,
+      "loss": 0.7205,
+      "step": 248
+    },
+    {
+      "epoch": 0.0037037897468745126,
+      "grad_norm": 2.6705269813537598,
+      "learning_rate": 9.141546526867629e-06,
+      "loss": 0.714,
+      "step": 279
+    },
+    {
+      "epoch": 0.004115321940971681,
+      "grad_norm": 3.432569980621338,
+      "learning_rate": 1.015727391874181e-05,
+      "loss": 0.7217,
+      "step": 310
+    },
+    {
+      "epoch": 0.004526854135068849,
+      "grad_norm": 2.515808343887329,
+      "learning_rate": 1.117300131061599e-05,
+      "loss": 0.7173,
+      "step": 341
+    },
+    {
+      "epoch": 0.004938386329166017,
+      "grad_norm": 2.6708128452301025,
+      "learning_rate": 1.218872870249017e-05,
+      "loss": 0.7123,
+      "step": 372
+    },
+    {
+      "epoch": 0.005349918523263185,
+      "grad_norm": 2.472797155380249,
+      "learning_rate": 1.3204456094364351e-05,
+      "loss": 0.7079,
+      "step": 403
+    },
+    {
+      "epoch": 0.005761450717360353,
+      "grad_norm": 2.486147880554199,
+      "learning_rate": 1.4220183486238533e-05,
+      "loss": 0.7146,
+      "step": 434
+    },
+    {
+      "epoch": 0.006172982911457521,
+      "grad_norm": 2.310777187347412,
+      "learning_rate": 1.5235910878112714e-05,
+      "loss": 0.7,
+      "step": 465
+    },
+    {
+      "epoch": 0.006584515105554689,
+      "grad_norm": 3.1831858158111572,
+      "learning_rate": 1.6251638269986893e-05,
+      "loss": 0.7042,
+      "step": 496
+    },
+    {
+      "epoch": 0.006996047299651857,
+      "grad_norm": 2.505427360534668,
+      "learning_rate": 1.7267365661861077e-05,
+      "loss": 0.6991,
+      "step": 527
+    },
+    {
+      "epoch": 0.007407579493749025,
+      "grad_norm": 2.3627443313598633,
+      "learning_rate": 1.8283093053735257e-05,
+      "loss": 0.7045,
+      "step": 558
+    },
+    {
+      "epoch": 0.007819111687846193,
+      "grad_norm": 2.6250648498535156,
+      "learning_rate": 1.9298820445609438e-05,
+      "loss": 0.704,
+      "step": 589
+    },
+    {
+      "epoch": 0.008230643881943361,
+      "grad_norm": 2.3151464462280273,
+      "learning_rate": 2.031454783748362e-05,
+      "loss": 0.6959,
+      "step": 620
+    },
+    {
+      "epoch": 0.00864217607604053,
+      "grad_norm": 2.0312647819519043,
+      "learning_rate": 2.13302752293578e-05,
+      "loss": 0.689,
+      "step": 651
+    },
+    {
+      "epoch": 0.009053708270137698,
+      "grad_norm": 2.718970775604248,
+      "learning_rate": 2.234600262123198e-05,
+      "loss": 0.7031,
+      "step": 682
+    },
+    {
+      "epoch": 0.009465240464234866,
+      "grad_norm": 1.9124070405960083,
+      "learning_rate": 2.336173001310616e-05,
+      "loss": 0.6908,
+      "step": 713
+    },
+    {
+      "epoch": 0.009876772658332034,
+      "grad_norm": 2.368354082107544,
+      "learning_rate": 2.437745740498034e-05,
+      "loss": 0.6947,
+      "step": 744
+    },
+    {
+      "epoch": 0.010288304852429202,
+      "grad_norm": 1.8856632709503174,
+      "learning_rate": 2.5393184796854525e-05,
+      "loss": 0.6876,
+      "step": 775
+    },
+    {
+      "epoch": 0.01069983704652637,
+      "grad_norm": 2.3155925273895264,
+      "learning_rate": 2.6408912188728702e-05,
+      "loss": 0.7024,
+      "step": 806
+    },
+    {
+      "epoch": 0.011111369240623538,
+      "grad_norm": 2.2516837120056152,
+      "learning_rate": 2.7424639580602886e-05,
+      "loss": 0.695,
+      "step": 837
+    },
+    {
+      "epoch": 0.011522901434720706,
+      "grad_norm": 2.4444823265075684,
+      "learning_rate": 2.8440366972477066e-05,
+      "loss": 0.6892,
+      "step": 868
+    },
+    {
+      "epoch": 0.011934433628817874,
+      "grad_norm": 2.4233832359313965,
+      "learning_rate": 2.9456094364351244e-05,
+      "loss": 0.6871,
+      "step": 899
+    },
+    {
+      "epoch": 0.012345965822915042,
+      "grad_norm": 2.0262961387634277,
+      "learning_rate": 3.0471821756225428e-05,
+      "loss": 0.6888,
+      "step": 930
+    },
+    {
+      "epoch": 0.01275749801701221,
+      "grad_norm": 2.3908143043518066,
+      "learning_rate": 3.148754914809961e-05,
+      "loss": 0.6997,
+      "step": 961
+    },
+    {
+      "epoch": 0.013169030211109378,
+      "grad_norm": 1.839417576789856,
+      "learning_rate": 3.2503276539973785e-05,
+      "loss": 0.6826,
+      "step": 992
+    },
+    {
+      "epoch": 0.013580562405206546,
+      "grad_norm": 1.8175997734069824,
+      "learning_rate": 3.351900393184797e-05,
+      "loss": 0.6857,
+      "step": 1023
+    },
+    {
+      "epoch": 0.013992094599303714,
+      "grad_norm": 2.3080506324768066,
+      "learning_rate": 3.453473132372215e-05,
+      "loss": 0.6896,
+      "step": 1054
+    },
+    {
+      "epoch": 0.014403626793400882,
+      "grad_norm": 2.0574960708618164,
+      "learning_rate": 3.555045871559633e-05,
+      "loss": 0.6862,
+      "step": 1085
+    },
+    {
+      "epoch": 0.01481515898749805,
+      "grad_norm": 1.957221269607544,
+      "learning_rate": 3.6566186107470514e-05,
+      "loss": 0.6796,
+      "step": 1116
+    },
+    {
+      "epoch": 0.015226691181595219,
+      "grad_norm": 2.2740609645843506,
+      "learning_rate": 3.7581913499344695e-05,
+      "loss": 0.7082,
+      "step": 1147
+    },
+    {
+      "epoch": 0.015638223375692387,
+      "grad_norm": 2.1291615962982178,
+      "learning_rate": 3.8597640891218876e-05,
+      "loss": 0.6852,
+      "step": 1178
+    },
+    {
+      "epoch": 0.016049755569789555,
+      "grad_norm": 2.2918620109558105,
+      "learning_rate": 3.9613368283093056e-05,
+      "loss": 0.6821,
+      "step": 1209
+    },
+    {
+      "epoch": 0.016461287763886723,
+      "grad_norm": 1.9689708948135376,
+      "learning_rate": 4.062909567496724e-05,
+      "loss": 0.676,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01687281995798389,
+      "grad_norm": 2.4405298233032227,
+      "learning_rate": 4.164482306684142e-05,
+      "loss": 0.6927,
+      "step": 1271
+    },
+    {
+      "epoch": 0.01728435215208106,
+      "grad_norm": 2.0007450580596924,
+      "learning_rate": 4.26605504587156e-05,
+      "loss": 0.6864,
+      "step": 1302
+    },
+    {
+      "epoch": 0.017695884346178227,
+      "grad_norm": 1.7013226747512817,
+      "learning_rate": 4.367627785058978e-05,
+      "loss": 0.6827,
+      "step": 1333
+    },
+    {
+      "epoch": 0.018107416540275395,
+      "grad_norm": 2.2202930450439453,
+      "learning_rate": 4.469200524246396e-05,
+      "loss": 0.6722,
+      "step": 1364
+    },
+    {
+      "epoch": 0.018518948734372563,
+      "grad_norm": 1.690030813217163,
+      "learning_rate": 4.570773263433814e-05,
+      "loss": 0.6807,
+      "step": 1395
+    },
+    {
+      "epoch": 0.01893048092846973,
+      "grad_norm": 1.5886859893798828,
+      "learning_rate": 4.672346002621232e-05,
+      "loss": 0.6867,
+      "step": 1426
+    },
+    {
+      "epoch": 0.0193420131225669,
+      "grad_norm": 1.7371172904968262,
+      "learning_rate": 4.77391874180865e-05,
+      "loss": 0.6832,
+      "step": 1457
+    },
+    {
+      "epoch": 0.019753545316664067,
+      "grad_norm": 1.8376264572143555,
+      "learning_rate": 4.875491480996068e-05,
+      "loss": 0.681,
+      "step": 1488
+    },
+    {
+      "epoch": 0.020165077510761235,
+      "grad_norm": 1.6280311346054077,
+      "learning_rate": 4.977064220183487e-05,
+      "loss": 0.6795,
+      "step": 1519
+    },
+    {
+      "epoch": 0.020576609704858403,
+      "grad_norm": 1.8016823530197144,
+      "learning_rate": 4.9999915451558777e-05,
+      "loss": 0.6789,
+      "step": 1550
+    },
+    {
+      "epoch": 0.02098814189895557,
+      "grad_norm": 1.7355237007141113,
+      "learning_rate": 4.999955597496219e-05,
+      "loss": 0.6753,
+      "step": 1581
+    },
+    {
+      "epoch": 0.02139967409305274,
+      "grad_norm": 1.5687408447265625,
+      "learning_rate": 4.9998914381774255e-05,
+      "loss": 0.6777,
+      "step": 1612
+    },
+    {
+      "epoch": 0.021811206287149908,
+      "grad_norm": 1.790453314781189,
+      "learning_rate": 4.999799067923527e-05,
+      "loss": 0.684,
+      "step": 1643
+    },
+    {
+      "epoch": 0.022222738481247076,
+      "grad_norm": 1.9073406457901,
+      "learning_rate": 4.999678487776908e-05,
+      "loss": 0.6793,
+      "step": 1674
+    },
+    {
+      "epoch": 0.022634270675344244,
+      "grad_norm": 1.5575100183486938,
+      "learning_rate": 4.9995296990983006e-05,
+      "loss": 0.6794,
+      "step": 1705
+    },
+    {
+      "epoch": 0.023045802869441412,
+      "grad_norm": 2.100715160369873,
+      "learning_rate": 4.999352703566763e-05,
+      "loss": 0.6753,
+      "step": 1736
+    },
+    {
+      "epoch": 0.02345733506353858,
+      "grad_norm": 1.9425480365753174,
+      "learning_rate": 4.999147503179668e-05,
+      "loss": 0.6755,
+      "step": 1767
+    },
+    {
+      "epoch": 0.023868867257635748,
+      "grad_norm": 1.5791170597076416,
+      "learning_rate": 4.998914100252672e-05,
+      "loss": 0.6754,
+      "step": 1798
+    },
+    {
+      "epoch": 0.024280399451732916,
+      "grad_norm": 1.5692673921585083,
+      "learning_rate": 4.998652497419696e-05,
+      "loss": 0.6682,
+      "step": 1829
+    },
+    {
+      "epoch": 0.024691931645830084,
+      "grad_norm": 1.5757725238800049,
+      "learning_rate": 4.9983626976328927e-05,
+      "loss": 0.6723,
+      "step": 1860
+    },
+    {
+      "epoch": 0.025103463839927252,
+      "grad_norm": 1.8000696897506714,
+      "learning_rate": 4.998044704162613e-05,
+      "loss": 0.6647,
+      "step": 1891
+    },
+    {
+      "epoch": 0.02551499603402442,
+      "grad_norm": 1.682919979095459,
+      "learning_rate": 4.9976985205973705e-05,
+      "loss": 0.669,
+      "step": 1922
+    },
+    {
+      "epoch": 0.02592652822812159,
+      "grad_norm": 1.6138657331466675,
+      "learning_rate": 4.997324150843799e-05,
+      "loss": 0.6698,
+      "step": 1953
+    },
+    {
+      "epoch": 0.026338060422218756,
+      "grad_norm": 1.5016391277313232,
+      "learning_rate": 4.99692159912661e-05,
+      "loss": 0.6645,
+      "step": 1984
+    },
+    {
+      "epoch": 0.026749592616315924,
+      "grad_norm": 1.6850241422653198,
+      "learning_rate": 4.996490869988546e-05,
+      "loss": 0.6769,
+      "step": 2015
+    },
+    {
+      "epoch": 0.027161124810413093,
+      "grad_norm": 1.521161437034607,
+      "learning_rate": 4.996031968290326e-05,
+      "loss": 0.6599,
+      "step": 2046
+    },
+    {
+      "epoch": 0.02757265700451026,
+      "grad_norm": 1.8039467334747314,
+      "learning_rate": 4.995544899210594e-05,
+      "loss": 0.6693,
+      "step": 2077
+    },
+    {
+      "epoch": 0.02798418919860743,
+      "grad_norm": 1.688811182975769,
+      "learning_rate": 4.9950296682458583e-05,
+      "loss": 0.661,
+      "step": 2108
+    },
+    {
+      "epoch": 0.028395721392704597,
+      "grad_norm": 1.669871211051941,
+      "learning_rate": 4.994486281210429e-05,
+      "loss": 0.6611,
+      "step": 2139
+    },
+    {
+      "epoch": 0.028807253586801765,
+      "grad_norm": 1.3887263536453247,
+      "learning_rate": 4.9939147442363566e-05,
+      "loss": 0.6677,
+      "step": 2170
+    },
+    {
+      "epoch": 0.029218785780898933,
+      "grad_norm": 1.3242331743240356,
+      "learning_rate": 4.9933150637733574e-05,
+      "loss": 0.6537,
+      "step": 2201
+    },
+    {
+      "epoch": 0.0296303179749961,
+      "grad_norm": 1.6638622283935547,
+      "learning_rate": 4.992687246588743e-05,
+      "loss": 0.6653,
+      "step": 2232
+    },
+    {
+      "epoch": 0.03004185016909327,
+      "grad_norm": 1.5110372304916382,
+      "learning_rate": 4.992031299767347e-05,
+      "loss": 0.6615,
+      "step": 2263
+    },
+    {
+      "epoch": 0.030453382363190437,
+      "grad_norm": 1.5244065523147583,
+      "learning_rate": 4.9913472307114386e-05,
+      "loss": 0.6552,
+      "step": 2294
+    },
+    {
+      "epoch": 0.030864914557287605,
+      "grad_norm": 1.416624903678894,
+      "learning_rate": 4.9906350471406446e-05,
+      "loss": 0.6569,
+      "step": 2325
+    },
+    {
+      "epoch": 0.03127644675138477,
+      "grad_norm": 1.4467304944992065,
+      "learning_rate": 4.989894757091861e-05,
+      "loss": 0.6613,
+      "step": 2356
+    },
+    {
+      "epoch": 0.03168797894548194,
+      "grad_norm": 1.4649864435195923,
+      "learning_rate": 4.989126368919158e-05,
+      "loss": 0.6604,
+      "step": 2387
+    },
+    {
+      "epoch": 0.03209951113957911,
+      "grad_norm": 1.6256624460220337,
+      "learning_rate": 4.988329891293693e-05,
+      "loss": 0.655,
+      "step": 2418
+    },
+    {
+      "epoch": 0.03251104333367628,
+      "grad_norm": 1.3986462354660034,
+      "learning_rate": 4.987505333203608e-05,
+      "loss": 0.6514,
+      "step": 2449
+    },
+    {
+      "epoch": 0.032922575527773446,
+      "grad_norm": 1.4490609169006348,
+      "learning_rate": 4.9866527039539276e-05,
+      "loss": 0.65,
+      "step": 2480
+    },
+    {
+      "epoch": 0.033334107721870614,
+      "grad_norm": 1.4979615211486816,
+      "learning_rate": 4.9857720131664594e-05,
+      "loss": 0.6579,
+      "step": 2511
+    },
+    {
+      "epoch": 0.03374563991596778,
+      "grad_norm": 1.6314005851745605,
+      "learning_rate": 4.9848632707796773e-05,
+      "loss": 0.6604,
+      "step": 2542
+    },
+    {
+      "epoch": 0.03415717211006495,
+      "grad_norm": 1.310655951499939,
+      "learning_rate": 4.9839264870486155e-05,
+      "loss": 0.6518,
+      "step": 2573
+    },
+    {
+      "epoch": 0.03456870430416212,
+      "grad_norm": 1.4176305532455444,
+      "learning_rate": 4.9829616725447526e-05,
+      "loss": 0.6518,
+      "step": 2604
+    },
+    {
+      "epoch": 0.034980236498259286,
+      "grad_norm": 1.6330211162567139,
+      "learning_rate": 4.981968838155888e-05,
+      "loss": 0.6474,
+      "step": 2635
+    },
+    {
+      "epoch": 0.035391768692356454,
+      "grad_norm": 1.5854465961456299,
+      "learning_rate": 4.980947995086024e-05,
+      "loss": 0.6462,
+      "step": 2666
+    },
+    {
+      "epoch": 0.03580330088645362,
+      "grad_norm": 1.2944835424423218,
+      "learning_rate": 4.979899154855234e-05,
+      "loss": 0.6503,
+      "step": 2697
+    },
+    {
+      "epoch": 0.03621483308055079,
+      "grad_norm": 1.4056243896484375,
+      "learning_rate": 4.9788223292995386e-05,
+      "loss": 0.6414,
+      "step": 2728
+    },
+    {
+      "epoch": 0.03662636527464796,
+      "grad_norm": 1.5649598836898804,
+      "learning_rate": 4.977717530570768e-05,
+      "loss": 0.6449,
+      "step": 2759
+    },
+    {
+      "epoch": 0.037037897468745126,
+      "grad_norm": 1.5205659866333008,
+      "learning_rate": 4.976584771136425e-05,
+      "loss": 0.6334,
+      "step": 2790
+    },
+    {
+      "epoch": 0.037449429662842294,
+      "grad_norm": 1.3248411417007446,
+      "learning_rate": 4.975424063779547e-05,
+      "loss": 0.6448,
+      "step": 2821
+    },
+    {
+      "epoch": 0.03786096185693946,
+      "grad_norm": 1.2402842044830322,
+      "learning_rate": 4.974235421598557e-05,
+      "loss": 0.6476,
+      "step": 2852
+    },
+    {
+      "epoch": 0.03827249405103663,
+      "grad_norm": 1.7939503192901611,
+      "learning_rate": 4.973018858007122e-05,
+      "loss": 0.6472,
+      "step": 2883
+    },
+    {
+      "epoch": 0.0386840262451338,
+      "grad_norm": 1.2774665355682373,
+      "learning_rate": 4.9717743867339963e-05,
+      "loss": 0.6374,
+      "step": 2914
+    },
+    {
+      "epoch": 0.03909555843923097,
+      "grad_norm": 1.6384273767471313,
+      "learning_rate": 4.9705020218228695e-05,
+      "loss": 0.6623,
+      "step": 2945
+    },
+    {
+      "epoch": 0.039507090633328135,
+      "grad_norm": 1.559270977973938,
+      "learning_rate": 4.969201777632205e-05,
+      "loss": 0.6391,
+      "step": 2976
+    },
+    {
+      "epoch": 0.0399186228274253,
+      "grad_norm": 1.6405760049819946,
+      "learning_rate": 4.9678736688350846e-05,
+      "loss": 0.6437,
+      "step": 3007
+    },
+    {
+      "epoch": 0.04033015502152247,
+      "grad_norm": 1.3599917888641357,
+      "learning_rate": 4.966517710419033e-05,
+      "loss": 0.6436,
+      "step": 3038
+    },
+    {
+      "epoch": 0.04074168721561964,
+      "grad_norm": 1.3080962896347046,
+      "learning_rate": 4.965133917685858e-05,
+      "loss": 0.6331,
+      "step": 3069
+    },
+    {
+      "epoch": 0.04115321940971681,
+      "grad_norm": 1.435195803642273,
+      "learning_rate": 4.9637223062514714e-05,
+      "loss": 0.6417,
+      "step": 3100
+    },
+    {
+      "epoch": 0.041564751603813975,
+      "grad_norm": 1.342986822128296,
+      "learning_rate": 4.962282892045718e-05,
+      "loss": 0.6391,
+      "step": 3131
+    },
+    {
+      "epoch": 0.04197628379791114,
+      "grad_norm": 1.524418592453003,
+      "learning_rate": 4.9608156913121904e-05,
+      "loss": 0.6511,
+      "step": 3162
+    },
+    {
+      "epoch": 0.04238781599200831,
+      "grad_norm": 1.4617900848388672,
+      "learning_rate": 4.959320720608049e-05,
+      "loss": 0.6367,
+      "step": 3193
+    },
+    {
+      "epoch": 0.04279934818610548,
+      "grad_norm": 1.3968737125396729,
+      "learning_rate": 4.9577979968038354e-05,
+      "loss": 0.6387,
+      "step": 3224
+    },
+    {
+      "epoch": 0.04321088038020265,
+      "grad_norm": 1.258406400680542,
+      "learning_rate": 4.956247537083282e-05,
+      "loss": 0.6356,
+      "step": 3255
+    },
+    {
+      "epoch": 0.043622412574299815,
+      "grad_norm": 1.5494844913482666,
+      "learning_rate": 4.9546693589431145e-05,
+      "loss": 0.6433,
+      "step": 3286
+    },
+    {
+      "epoch": 0.04403394476839698,
+      "grad_norm": 1.2248284816741943,
+      "learning_rate": 4.9530634801928595e-05,
+      "loss": 0.6376,
+      "step": 3317
+    },
+    {
+      "epoch": 0.04444547696249415,
+      "grad_norm": 1.4243425130844116,
+      "learning_rate": 4.9514299189546395e-05,
+      "loss": 0.6341,
+      "step": 3348
+    },
+    {
+      "epoch": 0.04485700915659132,
+      "grad_norm": 1.379364252090454,
+      "learning_rate": 4.949768693662973e-05,
+      "loss": 0.6378,
+      "step": 3379
+    },
+    {
+      "epoch": 0.04526854135068849,
+      "grad_norm": 1.3460108041763306,
+      "learning_rate": 4.948079823064559e-05,
+      "loss": 0.639,
+      "step": 3410
+    },
+    {
+      "epoch": 0.045680073544785656,
+      "grad_norm": 1.382348656654358,
+      "learning_rate": 4.946363326218074e-05,
+      "loss": 0.6327,
+      "step": 3441
+    },
+    {
+      "epoch": 0.046091605738882824,
+      "grad_norm": 1.40754234790802,
+      "learning_rate": 4.9446192224939525e-05,
+      "loss": 0.6234,
+      "step": 3472
+    },
+    {
+      "epoch": 0.04650313793297999,
+      "grad_norm": 1.502922773361206,
+      "learning_rate": 4.942847531574167e-05,
+      "loss": 0.6404,
+      "step": 3503
+    },
+    {
+      "epoch": 0.04691467012707716,
+      "grad_norm": 1.2896559238433838,
+      "learning_rate": 4.941048273452008e-05,
+      "loss": 0.6378,
+      "step": 3534
+    },
+    {
+      "epoch": 0.04732620232117433,
+      "grad_norm": 1.3114830255508423,
+      "learning_rate": 4.9392214684318605e-05,
+      "loss": 0.6373,
+      "step": 3565
+    },
+    {
+      "epoch": 0.047737734515271496,
+      "grad_norm": 1.4122483730316162,
+      "learning_rate": 4.93736713712897e-05,
+      "loss": 0.6305,
+      "step": 3596
+    },
+    {
+      "epoch": 0.048149266709368664,
+      "grad_norm": 1.2669744491577148,
+      "learning_rate": 4.9354853004692124e-05,
+      "loss": 0.6413,
+      "step": 3627
+    },
+    {
+      "epoch": 0.04856079890346583,
+      "grad_norm": 1.2965583801269531,
+      "learning_rate": 4.93357597968886e-05,
+      "loss": 0.6364,
+      "step": 3658
+    },
+    {
+      "epoch": 0.048972331097563,
+      "grad_norm": 1.3328967094421387,
+      "learning_rate": 4.931639196334338e-05,
+      "loss": 0.6466,
+      "step": 3689
+    },
+    {
+      "epoch": 0.04938386329166017,
+      "grad_norm": 1.6291115283966064,
+      "learning_rate": 4.9296749722619826e-05,
+      "loss": 0.6348,
+      "step": 3720
+    },
+    {
+      "epoch": 0.049795395485757336,
+      "grad_norm": 1.3224235773086548,
+      "learning_rate": 4.9276833296377966e-05,
+      "loss": 0.6256,
+      "step": 3751
+    },
+    {
+      "epoch": 0.050206927679854504,
+      "grad_norm": 1.3253369331359863,
+      "learning_rate": 4.925664290937196e-05,
+      "loss": 0.6303,
+      "step": 3782
+    },
+    {
+      "epoch": 0.05061845987395167,
+      "grad_norm": 1.2806897163391113,
+      "learning_rate": 4.9236178789447576e-05,
+      "loss": 0.6403,
+      "step": 3813
+    },
+    {
+      "epoch": 0.05102999206804884,
+      "grad_norm": 1.3389384746551514,
+      "learning_rate": 4.921544116753962e-05,
+      "loss": 0.6288,
+      "step": 3844
+    },
+    {
+      "epoch": 0.05144152426214601,
+      "grad_norm": 1.251550555229187,
+      "learning_rate": 4.919443027766935e-05,
+      "loss": 0.637,
+      "step": 3875
+    },
+    {
+      "epoch": 0.05185305645624318,
+      "grad_norm": 1.258484959602356,
+      "learning_rate": 4.91731463569418e-05,
+      "loss": 0.629,
+      "step": 3906
+    },
+    {
+      "epoch": 0.052264588650340345,
+      "grad_norm": 1.3574656248092651,
+      "learning_rate": 4.915158964554312e-05,
+      "loss": 0.6262,
+      "step": 3937
+    },
+    {
+      "epoch": 0.05267612084443751,
+      "grad_norm": 1.327051043510437,
+      "learning_rate": 4.912976038673786e-05,
+      "loss": 0.629,
+      "step": 3968
+    },
+    {
+      "epoch": 0.05308765303853468,
+      "grad_norm": 1.3813196420669556,
+      "learning_rate": 4.9107658826866254e-05,
+      "loss": 0.6308,
+      "step": 3999
+    },
+    {
+      "epoch": 0.05349918523263185,
+      "grad_norm": 1.2064168453216553,
+      "learning_rate": 4.908528521534139e-05,
+      "loss": 0.6321,
+      "step": 4030
+    },
+    {
+      "epoch": 0.05391071742672902,
+      "grad_norm": 1.2771285772323608,
+      "learning_rate": 4.906263980464644e-05,
+      "loss": 0.622,
+      "step": 4061
+    },
+    {
+      "epoch": 0.054322249620826185,
+      "grad_norm": 1.1788181066513062,
+      "learning_rate": 4.903972285033178e-05,
+      "loss": 0.6268,
+      "step": 4092
+    },
+    {
+      "epoch": 0.05473378181492335,
+      "grad_norm": 1.3975950479507446,
+      "learning_rate": 4.901653461101213e-05,
+      "loss": 0.6251,
+      "step": 4123
+    },
+    {
+      "epoch": 0.05514531400902052,
+      "grad_norm": 1.2083688974380493,
+      "learning_rate": 4.8993075348363626e-05,
+      "loss": 0.6287,
+      "step": 4154
+    },
+    {
+      "epoch": 0.05555684620311769,
+      "grad_norm": 1.2795994281768799,
+      "learning_rate": 4.896934532712084e-05,
+      "loss": 0.6302,
+      "step": 4185
+    },
+    {
+      "epoch": 0.05596837839721486,
+      "grad_norm": 1.1056169271469116,
+      "learning_rate": 4.8945344815073846e-05,
+      "loss": 0.6256,
+      "step": 4216
+    },
+    {
+      "epoch": 0.056379910591312025,
+      "grad_norm": 1.415405511856079,
+      "learning_rate": 4.892107408306516e-05,
+      "loss": 0.6334,
+      "step": 4247
+    },
+    {
+      "epoch": 0.056791442785409194,
+      "grad_norm": 1.2858442068099976,
+      "learning_rate": 4.889653340498669e-05,
+      "loss": 0.6248,
+      "step": 4278
+    },
+    {
+      "epoch": 0.05720297497950636,
+      "grad_norm": 1.3914793729782104,
+      "learning_rate": 4.8871723057776664e-05,
+      "loss": 0.6229,
+      "step": 4309
+    },
+    {
+      "epoch": 0.05761450717360353,
+      "grad_norm": 1.1236270666122437,
+      "learning_rate": 4.8846643321416476e-05,
+      "loss": 0.6272,
+      "step": 4340
+    },
+    {
+      "epoch": 0.0580260393677007,
+      "grad_norm": 1.393020510673523,
+      "learning_rate": 4.882129447892753e-05,
+      "loss": 0.624,
+      "step": 4371
+    },
+    {
+      "epoch": 0.058437571561797866,
+      "grad_norm": 1.3223553895950317,
+      "learning_rate": 4.8795676816368076e-05,
+      "loss": 0.6267,
+      "step": 4402
+    },
+    {
+      "epoch": 0.058849103755895034,
+      "grad_norm": 1.2308725118637085,
+      "learning_rate": 4.876979062282995e-05,
+      "loss": 0.6255,
+      "step": 4433
+    },
+    {
+      "epoch": 0.0592606359499922,
+      "grad_norm": 1.3741412162780762,
+      "learning_rate": 4.8743636190435325e-05,
+      "loss": 0.6246,
+      "step": 4464
+    },
+    {
+      "epoch": 0.05967216814408937,
+      "grad_norm": 1.2487592697143555,
+      "learning_rate": 4.871721381433344e-05,
+      "loss": 0.6221,
+      "step": 4495
+    },
+    {
+      "epoch": 0.06008370033818654,
+      "grad_norm": 1.3239238262176514,
+      "learning_rate": 4.869052379269719e-05,
+      "loss": 0.621,
+      "step": 4526
+    },
+    {
+      "epoch": 0.060495232532283706,
+      "grad_norm": 1.1411608457565308,
+      "learning_rate": 4.866356642671985e-05,
+      "loss": 0.6211,
+      "step": 4557
+    },
+    {
+      "epoch": 0.060906764726380874,
+      "grad_norm": 1.1687365770339966,
+      "learning_rate": 4.8636342020611634e-05,
+      "loss": 0.6159,
+      "step": 4588
+    },
+    {
+      "epoch": 0.06131829692047804,
+      "grad_norm": 1.25679349899292,
+      "learning_rate": 4.860885088159626e-05,
+      "loss": 0.6254,
+      "step": 4619
+    },
+    {
+      "epoch": 0.06172982911457521,
+      "grad_norm": 1.2380110025405884,
+      "learning_rate": 4.858109331990751e-05,
+      "loss": 0.6305,
+      "step": 4650
+    },
+    {
+      "epoch": 0.06214136130867238,
+      "grad_norm": 1.3054466247558594,
+      "learning_rate": 4.855306964878567e-05,
+      "loss": 0.6202,
+      "step": 4681
+    },
+    {
+      "epoch": 0.06255289350276955,
+      "grad_norm": 1.3628545999526978,
+      "learning_rate": 4.8524780184474084e-05,
+      "loss": 0.6192,
+      "step": 4712
+    },
+    {
+      "epoch": 0.06296442569686671,
+      "grad_norm": 1.3576844930648804,
+      "learning_rate": 4.8496225246215496e-05,
+      "loss": 0.6213,
+      "step": 4743
+    },
+    {
+      "epoch": 0.06337595789096388,
+      "grad_norm": 1.3126753568649292,
+      "learning_rate": 4.8467405156248505e-05,
+      "loss": 0.6198,
+      "step": 4774
+    },
+    {
+      "epoch": 0.06378749008506104,
+      "grad_norm": 1.2266637086868286,
+      "learning_rate": 4.843832023980392e-05,
+      "loss": 0.6229,
+      "step": 4805
+    },
+    {
+      "epoch": 0.06419902227915822,
+      "grad_norm": 1.4964330196380615,
+      "learning_rate": 4.840897082510106e-05,
+      "loss": 0.6264,
+      "step": 4836
+    },
+    {
+      "epoch": 0.06461055447325538,
+      "grad_norm": 1.1138426065444946,
+      "learning_rate": 4.8379357243344084e-05,
+      "loss": 0.6258,
+      "step": 4867
+    },
+    {
+      "epoch": 0.06502208666735255,
+      "grad_norm": 1.2055357694625854,
+      "learning_rate": 4.8349479828718236e-05,
+      "loss": 0.6263,
+      "step": 4898
+    },
+    {
+      "epoch": 0.06543361886144972,
+      "grad_norm": 1.2684381008148193,
+      "learning_rate": 4.8319338918386075e-05,
+      "loss": 0.6212,
+      "step": 4929
+    },
+    {
+      "epoch": 0.06584515105554689,
+      "grad_norm": 1.2323859930038452,
+      "learning_rate": 4.828893485248369e-05,
+      "loss": 0.6253,
+      "step": 4960
+    },
+    {
+      "epoch": 0.06625668324964405,
+      "grad_norm": 1.091818928718567,
+      "learning_rate": 4.825826797411682e-05,
+      "loss": 0.6127,
+      "step": 4991
+    },
+    {
+      "epoch": 0.06666821544374123,
+      "grad_norm": 1.2208534479141235,
+      "learning_rate": 4.822733862935702e-05,
+      "loss": 0.6179,
+      "step": 5022
+    },
+    {
+      "epoch": 0.06707974763783839,
+      "grad_norm": 1.3426082134246826,
+      "learning_rate": 4.819614716723775e-05,
+      "loss": 0.6246,
+      "step": 5053
+    },
+    {
+      "epoch": 0.06749127983193556,
+      "grad_norm": 1.2961252927780151,
+      "learning_rate": 4.8164693939750425e-05,
+      "loss": 0.6149,
+      "step": 5084
+    },
+    {
+      "epoch": 0.06790281202603272,
+      "grad_norm": 1.2990977764129639,
+      "learning_rate": 4.813297930184042e-05,
+      "loss": 0.6303,
+      "step": 5115
+    },
+    {
+      "epoch": 0.0683143442201299,
+      "grad_norm": 1.2724366188049316,
+      "learning_rate": 4.810100361140314e-05,
+      "loss": 0.6234,
+      "step": 5146
+    },
+    {
+      "epoch": 0.06872587641422706,
+      "grad_norm": 1.2031605243682861,
+      "learning_rate": 4.8068767229279885e-05,
+      "loss": 0.6212,
+      "step": 5177
+    },
+    {
+      "epoch": 0.06913740860832424,
+      "grad_norm": 1.2580938339233398,
+      "learning_rate": 4.8036270519253854e-05,
+      "loss": 0.6144,
+      "step": 5208
+    },
+    {
+      "epoch": 0.0695489408024214,
+      "grad_norm": 1.203706979751587,
+      "learning_rate": 4.8003513848046e-05,
+      "loss": 0.6259,
+      "step": 5239
+    },
+    {
+      "epoch": 0.06996047299651857,
+      "grad_norm": 1.421762466430664,
+      "learning_rate": 4.79704975853109e-05,
+      "loss": 0.6221,
+      "step": 5270
+    },
+    {
+      "epoch": 0.07037200519061573,
+      "grad_norm": 1.326038122177124,
+      "learning_rate": 4.793722210363262e-05,
+      "loss": 0.6152,
+      "step": 5301
+    },
+    {
+      "epoch": 0.07078353738471291,
+      "grad_norm": 1.361939787864685,
+      "learning_rate": 4.7903687778520414e-05,
+      "loss": 0.6206,
+      "step": 5332
+    },
+    {
+      "epoch": 0.07119506957881007,
+      "grad_norm": 1.2153042554855347,
+      "learning_rate": 4.7869894988404593e-05,
+      "loss": 0.6209,
+      "step": 5363
+    },
+    {
+      "epoch": 0.07160660177290724,
+      "grad_norm": 1.3737229108810425,
+      "learning_rate": 4.783584411463221e-05,
+      "loss": 0.6121,
+      "step": 5394
+    },
+    {
+      "epoch": 0.0720181339670044,
+      "grad_norm": 1.1781859397888184,
+      "learning_rate": 4.780153554146274e-05,
+      "loss": 0.6183,
+      "step": 5425
+    },
+    {
+      "epoch": 0.07242966616110158,
+      "grad_norm": 1.069472312927246,
+      "learning_rate": 4.7766969656063766e-05,
+      "loss": 0.6117,
+      "step": 5456
+    },
+    {
+      "epoch": 0.07284119835519874,
+      "grad_norm": 1.380937099456787,
+      "learning_rate": 4.773214684850662e-05,
+      "loss": 0.6175,
+      "step": 5487
+    },
+    {
+      "epoch": 0.07325273054929592,
+      "grad_norm": 1.1551166772842407,
+      "learning_rate": 4.769706751176193e-05,
+      "loss": 0.6164,
+      "step": 5518
+    },
+    {
+      "epoch": 0.07366426274339308,
+      "grad_norm": 1.2816351652145386,
+      "learning_rate": 4.7661732041695264e-05,
+      "loss": 0.6168,
+      "step": 5549
+    },
+    {
+      "epoch": 0.07407579493749025,
+      "grad_norm": 1.0833333730697632,
+      "learning_rate": 4.762614083706258e-05,
+      "loss": 0.6159,
+      "step": 5580
+    },
+    {
+      "epoch": 0.07448732713158741,
+      "grad_norm": 1.2533819675445557,
+      "learning_rate": 4.759029429950581e-05,
+      "loss": 0.6141,
+      "step": 5611
+    },
+    {
+      "epoch": 0.07489885932568459,
+      "grad_norm": 1.1354937553405762,
+      "learning_rate": 4.7554192833548235e-05,
+      "loss": 0.612,
+      "step": 5642
+    },
+    {
+      "epoch": 0.07531039151978175,
+      "grad_norm": 1.2579649686813354,
+      "learning_rate": 4.751783684659e-05,
+      "loss": 0.6147,
+      "step": 5673
+    },
+    {
+      "epoch": 0.07572192371387892,
+      "grad_norm": 1.2830709218978882,
+      "learning_rate": 4.748122674890348e-05,
+      "loss": 0.62,
+      "step": 5704
+    },
+    {
+      "epoch": 0.07613345590797609,
+      "grad_norm": 1.2012654542922974,
+      "learning_rate": 4.7444362953628654e-05,
+      "loss": 0.6199,
+      "step": 5735
+    },
+    {
+      "epoch": 0.07654498810207326,
+      "grad_norm": 1.2926653623580933,
+      "learning_rate": 4.7407245876768424e-05,
+      "loss": 0.6104,
+      "step": 5766
+    },
+    {
+      "epoch": 0.07695652029617042,
+      "grad_norm": 1.2043049335479736,
+      "learning_rate": 4.736987593718397e-05,
+      "loss": 0.6261,
+      "step": 5797
+    },
+    {
+      "epoch": 0.0773680524902676,
+      "grad_norm": 1.1188119649887085,
+      "learning_rate": 4.733225355658999e-05,
+      "loss": 0.6056,
+      "step": 5828
+    },
+    {
+      "epoch": 0.07777958468436476,
+      "grad_norm": 1.1217443943023682,
+      "learning_rate": 4.7294379159549926e-05,
+      "loss": 0.6124,
+      "step": 5859
+    },
+    {
+      "epoch": 0.07819111687846193,
+      "grad_norm": 1.1623468399047852,
+      "learning_rate": 4.725625317347119e-05,
+      "loss": 0.6191,
+      "step": 5890
+    },
+    {
+      "epoch": 0.0786026490725591,
+      "grad_norm": 1.1147449016571045,
+      "learning_rate": 4.7217876028600374e-05,
+      "loss": 0.6091,
+      "step": 5921
+    },
+    {
+      "epoch": 0.07901418126665627,
+      "grad_norm": 1.0523818731307983,
+      "learning_rate": 4.717924815801832e-05,
+      "loss": 0.6151,
+      "step": 5952
+    },
+    {
+      "epoch": 0.07942571346075343,
+      "grad_norm": 1.3112642765045166,
+      "learning_rate": 4.714036999763532e-05,
+      "loss": 0.6106,
+      "step": 5983
+    },
+    {
+      "epoch": 0.0798372456548506,
+      "grad_norm": 1.1637545824050903,
+      "learning_rate": 4.7101241986186116e-05,
+      "loss": 0.6131,
+      "step": 6014
+    },
+    {
+      "epoch": 0.08024877784894777,
+      "grad_norm": 1.258862018585205,
+      "learning_rate": 4.7061864565225e-05,
+      "loss": 0.6119,
+      "step": 6045
+    },
+    {
+      "epoch": 0.08066031004304494,
+      "grad_norm": 1.0981336832046509,
+      "learning_rate": 4.702223817912081e-05,
+      "loss": 0.6149,
+      "step": 6076
+    },
+    {
+      "epoch": 0.0810718422371421,
+      "grad_norm": 1.1576476097106934,
+      "learning_rate": 4.698236327505195e-05,
+      "loss": 0.6076,
+      "step": 6107
+    },
+    {
+      "epoch": 0.08148337443123928,
+      "grad_norm": 1.1076856851577759,
+      "learning_rate": 4.694224030300127e-05,
+      "loss": 0.6105,
+      "step": 6138
+    },
+    {
+      "epoch": 0.08189490662533644,
+      "grad_norm": 14.014957427978516,
+      "learning_rate": 4.690186971575107e-05,
+      "loss": 0.6189,
+      "step": 6169
+    },
+    {
+      "epoch": 0.08230643881943361,
+      "grad_norm": 1.1413066387176514,
+      "learning_rate": 4.6861251968877916e-05,
+      "loss": 0.6038,
+      "step": 6200
+    },
+    {
+      "epoch": 0.08271797101353077,
+      "grad_norm": 1.2316724061965942,
+      "learning_rate": 4.68203875207476e-05,
+      "loss": 0.6125,
+      "step": 6231
+    },
+    {
+      "epoch": 0.08312950320762795,
+      "grad_norm": 1.0919370651245117,
+      "learning_rate": 4.677927683250983e-05,
+      "loss": 0.6081,
+      "step": 6262
+    },
+    {
+      "epoch": 0.08354103540172511,
+      "grad_norm": 1.139345407485962,
+      "learning_rate": 4.6737920368093156e-05,
+      "loss": 0.6078,
+      "step": 6293
+    },
+    {
+      "epoch": 0.08395256759582229,
+      "grad_norm": 1.1303651332855225,
+      "learning_rate": 4.669631859419965e-05,
+      "loss": 0.6071,
+      "step": 6324
+    },
+    {
+      "epoch": 0.08436409978991945,
+      "grad_norm": 1.2546321153640747,
+      "learning_rate": 4.6654471980299676e-05,
+      "loss": 0.6092,
+      "step": 6355
+    },
+    {
+      "epoch": 0.08477563198401662,
+      "grad_norm": 1.195173740386963,
+      "learning_rate": 4.661238099862658e-05,
+      "loss": 0.5998,
+      "step": 6386
+    },
+    {
+      "epoch": 0.08518716417811378,
+      "grad_norm": 1.2116323709487915,
+      "learning_rate": 4.657004612417138e-05,
+      "loss": 0.6099,
+      "step": 6417
+    },
+    {
+      "epoch": 0.08559869637221096,
+      "grad_norm": 1.131951928138733,
+      "learning_rate": 4.6527467834677374e-05,
+      "loss": 0.6105,
+      "step": 6448
+    },
+    {
+      "epoch": 0.08601022856630812,
+      "grad_norm": 1.2176560163497925,
+      "learning_rate": 4.648464661063478e-05,
+      "loss": 0.6045,
+      "step": 6479
+    },
+    {
+      "epoch": 0.0864217607604053,
+      "grad_norm": 1.15361750125885,
+      "learning_rate": 4.6441582935275264e-05,
+      "loss": 0.6067,
+      "step": 6510
+    },
+    {
+      "epoch": 0.08683329295450246,
+      "grad_norm": 1.1422772407531738,
+      "learning_rate": 4.6398277294566586e-05,
+      "loss": 0.605,
+      "step": 6541
+    },
+    {
+      "epoch": 0.08724482514859963,
+      "grad_norm": 1.250319004058838,
+      "learning_rate": 4.6354730177207e-05,
+      "loss": 0.6069,
+      "step": 6572
+    },
+    {
+      "epoch": 0.08765635734269679,
+      "grad_norm": 1.0947929620742798,
+      "learning_rate": 4.6310942074619787e-05,
+      "loss": 0.6099,
+      "step": 6603
+    },
+    {
+      "epoch": 0.08806788953679397,
+      "grad_norm": 1.2815848588943481,
+      "learning_rate": 4.626691348094777e-05,
+      "loss": 0.6132,
+      "step": 6634
+    },
+    {
+      "epoch": 0.08847942173089113,
+      "grad_norm": 1.196083664894104,
+      "learning_rate": 4.622264489304762e-05,
+      "loss": 0.604,
+      "step": 6665
+    },
+    {
+      "epoch": 0.0888909539249883,
+      "grad_norm": 1.1473642587661743,
+      "learning_rate": 4.617813681048434e-05,
+      "loss": 0.6081,
+      "step": 6696
+    },
+    {
+      "epoch": 0.08930248611908546,
+      "grad_norm": 1.182445764541626,
+      "learning_rate": 4.61333897355256e-05,
+      "loss": 0.6108,
+      "step": 6727
+    },
+    {
+      "epoch": 0.08971401831318264,
+      "grad_norm": 1.2859700918197632,
+      "learning_rate": 4.608840417313604e-05,
+      "loss": 0.6074,
+      "step": 6758
+    },
+    {
+      "epoch": 0.0901255505072798,
+      "grad_norm": 1.1134952306747437,
+      "learning_rate": 4.6043180630971646e-05,
+      "loss": 0.6139,
+      "step": 6789
+    },
+    {
+      "epoch": 0.09053708270137698,
+      "grad_norm": 1.2447940111160278,
+      "learning_rate": 4.599771961937391e-05,
+      "loss": 0.6049,
+      "step": 6820
+    },
+    {
+      "epoch": 0.09094861489547414,
+      "grad_norm": 1.042465329170227,
+      "learning_rate": 4.5952021651364204e-05,
+      "loss": 0.6113,
+      "step": 6851
+    },
+    {
+      "epoch": 0.09136014708957131,
+      "grad_norm": 1.2242035865783691,
+      "learning_rate": 4.590608724263786e-05,
+      "loss": 0.6077,
+      "step": 6882
+    },
+    {
+      "epoch": 0.09177167928366847,
+      "grad_norm": 1.2531142234802246,
+      "learning_rate": 4.585991691155845e-05,
+      "loss": 0.6081,
+      "step": 6913
+    },
+    {
+      "epoch": 0.09218321147776565,
+      "grad_norm": 1.194645881652832,
+      "learning_rate": 4.581351117915188e-05,
+      "loss": 0.6158,
+      "step": 6944
+    },
+    {
+      "epoch": 0.09259474367186281,
+      "grad_norm": 1.1052149534225464,
+      "learning_rate": 4.5766870569100534e-05,
+      "loss": 0.6079,
+      "step": 6975
+    },
+    {
+      "epoch": 0.09300627586595998,
+      "grad_norm": 1.1275289058685303,
+      "learning_rate": 4.571999560773736e-05,
+      "loss": 0.6107,
+      "step": 7006
+    },
+    {
+      "epoch": 0.09341780806005714,
+      "grad_norm": 1.251237154006958,
+      "learning_rate": 4.5672886824039915e-05,
+      "loss": 0.6028,
+      "step": 7037
+    },
+    {
+      "epoch": 0.09382934025415432,
+      "grad_norm": 1.1206951141357422,
+      "learning_rate": 4.5625544749624435e-05,
+      "loss": 0.6035,
+      "step": 7068
+    },
+    {
+      "epoch": 0.09424087244825148,
+      "grad_norm": 1.2242387533187866,
+      "learning_rate": 4.5577969918739794e-05,
+      "loss": 0.5959,
+      "step": 7099
+    },
+    {
+      "epoch": 0.09465240464234866,
+      "grad_norm": 1.0749527215957642,
+      "learning_rate": 4.5530162868261486e-05,
+      "loss": 0.6005,
+      "step": 7130
+    },
+    {
+      "epoch": 0.09506393683644582,
+      "grad_norm": 1.183052659034729,
+      "learning_rate": 4.548212413768558e-05,
+      "loss": 0.6102,
+      "step": 7161
+    },
+    {
+      "epoch": 0.09547546903054299,
+      "grad_norm": 1.1045013666152954,
+      "learning_rate": 4.543385426912261e-05,
+      "loss": 0.6056,
+      "step": 7192
+    },
+    {
+      "epoch": 0.09588700122464015,
+      "grad_norm": 1.2442799806594849,
+      "learning_rate": 4.53853538072915e-05,
+      "loss": 0.6024,
+      "step": 7223
+    },
+    {
+      "epoch": 0.09629853341873733,
+      "grad_norm": 1.2083353996276855,
+      "learning_rate": 4.533662329951336e-05,
+      "loss": 0.6062,
+      "step": 7254
+    },
+    {
+      "epoch": 0.09671006561283449,
+      "grad_norm": 1.1391205787658691,
+      "learning_rate": 4.528766329570536e-05,
+      "loss": 0.5981,
+      "step": 7285
+    },
+    {
+      "epoch": 0.09712159780693166,
+      "grad_norm": 1.1213765144348145,
+      "learning_rate": 4.523847434837447e-05,
+      "loss": 0.5964,
+      "step": 7316
+    },
+    {
+      "epoch": 0.09753313000102883,
+      "grad_norm": 1.1574114561080933,
+      "learning_rate": 4.518905701261128e-05,
+      "loss": 0.6035,
+      "step": 7347
+    },
+    {
+      "epoch": 0.097944662195126,
+      "grad_norm": 1.0101516246795654,
+      "learning_rate": 4.5139411846083715e-05,
+      "loss": 0.6036,
+      "step": 7378
+    },
+    {
+      "epoch": 0.09835619438922316,
+      "grad_norm": 1.1439381837844849,
+      "learning_rate": 4.508953940903073e-05,
+      "loss": 0.6049,
+      "step": 7409
+    },
+    {
+      "epoch": 0.09876772658332034,
+      "grad_norm": 1.0668189525604248,
+      "learning_rate": 4.5039440264255994e-05,
+      "loss": 0.6035,
+      "step": 7440
+    },
+    {
+      "epoch": 0.0991792587774175,
+      "grad_norm": 1.2038415670394897,
+      "learning_rate": 4.498911497712155e-05,
+      "loss": 0.6062,
+      "step": 7471
+    },
+    {
+      "epoch": 0.09959079097151467,
+      "grad_norm": 1.1965882778167725,
+      "learning_rate": 4.493856411554142e-05,
+      "loss": 0.6041,
+      "step": 7502
+    },
+    {
+      "epoch": 0.10000232316561183,
+      "grad_norm": 1.5885149240493774,
+      "learning_rate": 4.4887788249975206e-05,
+      "loss": 0.6129,
+      "step": 7533
+    },
+    {
+      "epoch": 0.10041385535970901,
+      "grad_norm": 1.058237910270691,
+      "learning_rate": 4.4836787953421656e-05,
+      "loss": 0.5957,
+      "step": 7564
+    },
+    {
+      "epoch": 0.10082538755380617,
+      "grad_norm": 1.2043278217315674,
+      "learning_rate": 4.478556380141218e-05,
+      "loss": 0.5943,
+      "step": 7595
+    },
+    {
+      "epoch": 0.10123691974790335,
+      "grad_norm": 1.1602888107299805,
+      "learning_rate": 4.4734116372004375e-05,
+      "loss": 0.5971,
+      "step": 7626
+    },
+    {
+      "epoch": 0.1016484519420005,
+      "grad_norm": 1.1536785364151,
+      "learning_rate": 4.4682446245775477e-05,
+      "loss": 0.604,
+      "step": 7657
+    },
+    {
+      "epoch": 0.10205998413609768,
+      "grad_norm": 1.1859934329986572,
+      "learning_rate": 4.463055400581586e-05,
+      "loss": 0.6043,
+      "step": 7688
+    },
+    {
+      "epoch": 0.10247151633019484,
+      "grad_norm": 1.0551954507827759,
+      "learning_rate": 4.4578440237722374e-05,
+      "loss": 0.6008,
+      "step": 7719
+    },
+    {
+      "epoch": 0.10288304852429202,
+      "grad_norm": 1.015857458114624,
+      "learning_rate": 4.452610552959183e-05,
+      "loss": 0.5961,
+      "step": 7750
+    },
+    {
+      "epoch": 0.10329458071838918,
+      "grad_norm": 1.0759170055389404,
+      "learning_rate": 4.447355047201428e-05,
+      "loss": 0.6025,
+      "step": 7781
+    },
+    {
+      "epoch": 0.10370611291248635,
+      "grad_norm": 1.025030255317688,
+      "learning_rate": 4.4420775658066414e-05,
+      "loss": 0.602,
+      "step": 7812
+    },
+    {
+      "epoch": 0.10411764510658351,
+      "grad_norm": 1.0884742736816406,
+      "learning_rate": 4.436778168330484e-05,
+      "loss": 0.5989,
+      "step": 7843
+    },
+    {
+      "epoch": 0.10452917730068069,
+      "grad_norm": 1.1207304000854492,
+      "learning_rate": 4.4314569145759353e-05,
+      "loss": 0.6016,
+      "step": 7874
+    },
+    {
+      "epoch": 0.10494070949477785,
+      "grad_norm": 0.9768222570419312,
+      "learning_rate": 4.42611386459262e-05,
+      "loss": 0.605,
+      "step": 7905
+    },
+    {
+      "epoch": 0.10535224168887503,
+      "grad_norm": 1.0720235109329224,
+      "learning_rate": 4.420749078676133e-05,
+      "loss": 0.5971,
+      "step": 7936
+    },
+    {
+      "epoch": 0.10576377388297219,
+      "grad_norm": 1.1767033338546753,
+      "learning_rate": 4.4153626173673516e-05,
+      "loss": 0.5877,
+      "step": 7967
+    },
+    {
+      "epoch": 0.10617530607706936,
+      "grad_norm": 1.3390216827392578,
+      "learning_rate": 4.409954541451762e-05,
+      "loss": 0.5973,
+      "step": 7998
+    },
+    {
+      "epoch": 0.10658683827116652,
+      "grad_norm": 1.7948216199874878,
+      "learning_rate": 4.404524911958764e-05,
+      "loss": 0.5922,
+      "step": 8029
+    },
+    {
+      "epoch": 0.1069983704652637,
+      "grad_norm": 1.177847981452942,
+      "learning_rate": 4.399073790160989e-05,
+      "loss": 0.6002,
+      "step": 8060
+    },
+    {
+      "epoch": 0.10740990265936086,
+      "grad_norm": 1.1014128923416138,
+      "learning_rate": 4.393601237573607e-05,
+      "loss": 0.5945,
+      "step": 8091
+    },
+    {
+      "epoch": 0.10782143485345803,
+      "grad_norm": 1.067475438117981,
+      "learning_rate": 4.388107315953628e-05,
+      "loss": 0.5996,
+      "step": 8122
+    },
+    {
+      "epoch": 0.1082329670475552,
+      "grad_norm": 1.1241830587387085,
+      "learning_rate": 4.382592087299212e-05,
+      "loss": 0.5973,
+      "step": 8153
+    },
+    {
+      "epoch": 0.10864449924165237,
+      "grad_norm": 1.072426438331604,
+      "learning_rate": 4.377055613848964e-05,
+      "loss": 0.601,
+      "step": 8184
+    },
+    {
+      "epoch": 0.10905603143574953,
+      "grad_norm": 1.0138245820999146,
+      "learning_rate": 4.3714979580812355e-05,
+      "loss": 0.6034,
+      "step": 8215
+    },
+    {
+      "epoch": 0.1094675636298467,
+      "grad_norm": 0.9993209838867188,
+      "learning_rate": 4.365919182713416e-05,
+      "loss": 0.6024,
+      "step": 8246
+    },
+    {
+      "epoch": 0.10987909582394387,
+      "grad_norm": 1.157468318939209,
+      "learning_rate": 4.360319350701226e-05,
+      "loss": 0.5987,
+      "step": 8277
+    },
+    {
+      "epoch": 0.11029062801804104,
+      "grad_norm": 1.2073496580123901,
+      "learning_rate": 4.3546985252380115e-05,
+      "loss": 0.6023,
+      "step": 8308
+    },
+    {
+      "epoch": 0.1107021602121382,
+      "grad_norm": 1.0800622701644897,
+      "learning_rate": 4.349056769754021e-05,
+      "loss": 0.5993,
+      "step": 8339
+    },
+    {
+      "epoch": 0.11111369240623538,
+      "grad_norm": 1.1365702152252197,
+      "learning_rate": 4.3433941479156994e-05,
+      "loss": 0.595,
+      "step": 8370
+    },
+    {
+      "epoch": 0.11152522460033254,
+      "grad_norm": 1.1367030143737793,
+      "learning_rate": 4.3377107236249647e-05,
+      "loss": 0.5921,
+      "step": 8401
+    },
+    {
+      "epoch": 0.11193675679442971,
+      "grad_norm": 1.0323926210403442,
+      "learning_rate": 4.332006561018488e-05,
+      "loss": 0.5883,
+      "step": 8432
+    },
+    {
+      "epoch": 0.11234828898852688,
+      "grad_norm": 1.2352523803710938,
+      "learning_rate": 4.3262817244669683e-05,
+      "loss": 0.5957,
+      "step": 8463
+    },
+    {
+      "epoch": 0.11275982118262405,
+      "grad_norm": 1.1763888597488403,
+      "learning_rate": 4.3205362785744083e-05,
+      "loss": 0.5941,
+      "step": 8494
+    },
+    {
+      "epoch": 0.11317135337672121,
+      "grad_norm": 1.1331796646118164,
+      "learning_rate": 4.314770288177384e-05,
+      "loss": 0.5993,
+      "step": 8525
+    },
+    {
+      "epoch": 0.11358288557081839,
+      "grad_norm": 1.146270990371704,
+      "learning_rate": 4.308983818344313e-05,
+      "loss": 0.6041,
+      "step": 8556
+    },
+    {
+      "epoch": 0.11399441776491555,
+      "grad_norm": 1.0433791875839233,
+      "learning_rate": 4.3031769343747206e-05,
+      "loss": 0.5918,
+      "step": 8587
+    },
+    {
+      "epoch": 0.11440594995901272,
+      "grad_norm": 1.2013510465621948,
+      "learning_rate": 4.297349701798505e-05,
+      "loss": 0.6031,
+      "step": 8618
+    },
+    {
+      "epoch": 0.11481748215310988,
+      "grad_norm": 1.1263506412506104,
+      "learning_rate": 4.2915021863751916e-05,
+      "loss": 0.597,
+      "step": 8649
+    },
+    {
+      "epoch": 0.11522901434720706,
+      "grad_norm": 1.1360443830490112,
+      "learning_rate": 4.285634454093198e-05,
+      "loss": 0.5909,
+      "step": 8680
+    },
+    {
+      "epoch": 0.11564054654130422,
+      "grad_norm": 1.0095175504684448,
+      "learning_rate": 4.279746571169086e-05,
+      "loss": 0.59,
+      "step": 8711
+    },
+    {
+      "epoch": 0.1160520787354014,
+      "grad_norm": 1.0894793272018433,
+      "learning_rate": 4.2738386040468136e-05,
+      "loss": 0.5894,
+      "step": 8742
+    },
+    {
+      "epoch": 0.11646361092949856,
+      "grad_norm": 1.1603643894195557,
+      "learning_rate": 4.2679106193969866e-05,
+      "loss": 0.5924,
+      "step": 8773
+    },
+    {
+      "epoch": 0.11687514312359573,
+      "grad_norm": 1.177464485168457,
+      "learning_rate": 4.261962684116106e-05,
+      "loss": 0.5946,
+      "step": 8804
+    },
+    {
+      "epoch": 0.11728667531769289,
+      "grad_norm": 0.9632905125617981,
+      "learning_rate": 4.2559948653258145e-05,
+      "loss": 0.5944,
+      "step": 8835
+    },
+    {
+      "epoch": 0.11769820751179007,
+      "grad_norm": 1.0413568019866943,
+      "learning_rate": 4.250007230372134e-05,
+      "loss": 0.5844,
+      "step": 8866
+    },
+    {
+      "epoch": 0.11810973970588723,
+      "grad_norm": 1.0879931449890137,
+      "learning_rate": 4.2439998468247126e-05,
+      "loss": 0.588,
+      "step": 8897
+    },
+    {
+      "epoch": 0.1185212718999844,
+      "grad_norm": 1.0681732892990112,
+      "learning_rate": 4.2379727824760566e-05,
+      "loss": 0.5832,
+      "step": 8928
+    },
+    {
+      "epoch": 0.11893280409408157,
+      "grad_norm": 1.0609339475631714,
+      "learning_rate": 4.231926105340768e-05,
+      "loss": 0.5886,
+      "step": 8959
+    },
+    {
+      "epoch": 0.11934433628817874,
+      "grad_norm": 1.4936331510543823,
+      "learning_rate": 4.225859883654776e-05,
+      "loss": 0.5868,
+      "step": 8990
+    },
+    {
+      "epoch": 0.1197558684822759,
+      "grad_norm": 1.1860368251800537,
+      "learning_rate": 4.219774185874569e-05,
+      "loss": 0.5904,
+      "step": 9021
+    },
+    {
+      "epoch": 0.12016740067637308,
+      "grad_norm": 1.1657029390335083,
+      "learning_rate": 4.213669080676418e-05,
+      "loss": 0.5872,
+      "step": 9052
+    },
+    {
+      "epoch": 0.12057893287047024,
+      "grad_norm": 1.0721909999847412,
+      "learning_rate": 4.2075446369556056e-05,
+      "loss": 0.5977,
+      "step": 9083
+    },
+    {
+      "epoch": 0.12099046506456741,
+      "grad_norm": 1.1832354068756104,
+      "learning_rate": 4.201400923825648e-05,
+      "loss": 0.5902,
+      "step": 9114
+    },
+    {
+      "epoch": 0.12140199725866457,
+      "grad_norm": 1.1306614875793457,
+      "learning_rate": 4.195238010617511e-05,
+      "loss": 0.5931,
+      "step": 9145
+    },
+    {
+      "epoch": 0.12181352945276175,
+      "grad_norm": 0.9744802713394165,
+      "learning_rate": 4.1890559668788344e-05,
+      "loss": 0.6,
+      "step": 9176
+    },
+    {
+      "epoch": 0.12222506164685891,
+      "grad_norm": 1.0286784172058105,
+      "learning_rate": 4.1828548623731405e-05,
+      "loss": 0.5913,
+      "step": 9207
+    },
+    {
+      "epoch": 0.12263659384095608,
+      "grad_norm": 1.0559006929397583,
+      "learning_rate": 4.1766347670790506e-05,
+      "loss": 0.5999,
+      "step": 9238
+    },
+    {
+      "epoch": 0.12304812603505325,
+      "grad_norm": 1.0805801153182983,
+      "learning_rate": 4.170395751189495e-05,
+      "loss": 0.5929,
+      "step": 9269
+    },
+    {
+      "epoch": 0.12345965822915042,
+      "grad_norm": 1.088597059249878,
+      "learning_rate": 4.164137885110921e-05,
+      "loss": 0.5873,
+      "step": 9300
+    },
+    {
+      "epoch": 0.12387119042324758,
+      "grad_norm": 1.028403639793396,
+      "learning_rate": 4.157861239462495e-05,
+      "loss": 0.5939,
+      "step": 9331
+    },
+    {
+      "epoch": 0.12428272261734476,
+      "grad_norm": 0.9916087985038757,
+      "learning_rate": 4.1515658850753114e-05,
+      "loss": 0.5932,
+      "step": 9362
+    },
+    {
+      "epoch": 0.12469425481144192,
+      "grad_norm": 1.0268062353134155,
+      "learning_rate": 4.145251892991588e-05,
+      "loss": 0.5945,
+      "step": 9393
+    },
+    {
+      "epoch": 0.1251057870055391,
+      "grad_norm": 1.046953797340393,
+      "learning_rate": 4.138919334463868e-05,
+      "loss": 0.5923,
+      "step": 9424
+    },
+    {
+      "epoch": 0.12551731919963627,
+      "grad_norm": 1.1010463237762451,
+      "learning_rate": 4.1325682809542124e-05,
+      "loss": 0.5898,
+      "step": 9455
+    },
+    {
+      "epoch": 0.12592885139373342,
+      "grad_norm": 1.1851146221160889,
+      "learning_rate": 4.126198804133398e-05,
+      "loss": 0.591,
+      "step": 9486
+    },
+    {
+      "epoch": 0.1263403835878306,
+      "grad_norm": 1.106885313987732,
+      "learning_rate": 4.1198109758801055e-05,
+      "loss": 0.5893,
+      "step": 9517
+    },
+    {
+      "epoch": 0.12675191578192777,
+      "grad_norm": 1.0527287721633911,
+      "learning_rate": 4.113404868280107e-05,
+      "loss": 0.5937,
+      "step": 9548
+    },
+    {
+      "epoch": 0.12716344797602494,
+      "grad_norm": 0.9986059069633484,
+      "learning_rate": 4.106980553625457e-05,
+      "loss": 0.5961,
+      "step": 9579
+    },
+    {
+      "epoch": 0.1275749801701221,
+      "grad_norm": 1.025026798248291,
+      "learning_rate": 4.100538104413674e-05,
+      "loss": 0.5853,
+      "step": 9610
+    },
+    {
+      "epoch": 0.12798651236421926,
+      "grad_norm": 1.062345027923584,
+      "learning_rate": 4.09407759334692e-05,
+      "loss": 0.5916,
+      "step": 9641
+    },
+    {
+      "epoch": 0.12839804455831644,
+      "grad_norm": 1.0635534524917603,
+      "learning_rate": 4.087599093331186e-05,
+      "loss": 0.5919,
+      "step": 9672
+    },
+    {
+      "epoch": 0.1288095767524136,
+      "grad_norm": 1.1224849224090576,
+      "learning_rate": 4.081102677475462e-05,
+      "loss": 0.5948,
+      "step": 9703
+    },
+    {
+      "epoch": 0.12922110894651076,
+      "grad_norm": 1.113202452659607,
+      "learning_rate": 4.0745884190909194e-05,
+      "loss": 0.5807,
+      "step": 9734
+    },
+    {
+      "epoch": 0.12963264114060793,
+      "grad_norm": 1.0213284492492676,
+      "learning_rate": 4.0680563916900796e-05,
+      "loss": 0.5813,
+      "step": 9765
+    },
+    {
+      "epoch": 0.1300441733347051,
+      "grad_norm": 1.1083142757415771,
+      "learning_rate": 4.0615066689859815e-05,
+      "loss": 0.5856,
+      "step": 9796
+    },
+    {
+      "epoch": 0.13045570552880228,
+      "grad_norm": 1.1317702531814575,
+      "learning_rate": 4.0549393248913584e-05,
+      "loss": 0.586,
+      "step": 9827
+    },
+    {
+      "epoch": 0.13086723772289943,
+      "grad_norm": 0.9876610040664673,
+      "learning_rate": 4.048354433517794e-05,
+      "loss": 0.5921,
+      "step": 9858
+    },
+    {
+      "epoch": 0.1312787699169966,
+      "grad_norm": 1.085672378540039,
+      "learning_rate": 4.0417520691748916e-05,
+      "loss": 0.5791,
+      "step": 9889
+    },
+    {
+      "epoch": 0.13169030211109378,
+      "grad_norm": 1.014452338218689,
+      "learning_rate": 4.035132306369438e-05,
+      "loss": 0.5921,
+      "step": 9920
+    },
+    {
+      "epoch": 0.13210183430519096,
+      "grad_norm": 1.059309720993042,
+      "learning_rate": 4.028495219804555e-05,
+      "loss": 0.5834,
+      "step": 9951
+    },
+    {
+      "epoch": 0.1325133664992881,
+      "grad_norm": 1.1058080196380615,
+      "learning_rate": 4.021840884378864e-05,
+      "loss": 0.5826,
+      "step": 9982
+    },
+    {
+      "epoch": 0.13292489869338528,
+      "grad_norm": 1.0567057132720947,
+      "learning_rate": 4.015169375185633e-05,
+      "loss": 0.5924,
+      "step": 10013
+    },
+    {
+      "epoch": 0.13333643088748245,
+      "grad_norm": 1.0371674299240112,
+      "learning_rate": 4.0084807675119396e-05,
+      "loss": 0.589,
+      "step": 10044
+    },
+    {
+      "epoch": 0.13374796308157963,
+      "grad_norm": 0.909173309803009,
+      "learning_rate": 4.0017751368378106e-05,
+      "loss": 0.5978,
+      "step": 10075
+    },
+    {
+      "epoch": 0.13415949527567678,
+      "grad_norm": 1.0078933238983154,
+      "learning_rate": 3.995052558835377e-05,
+      "loss": 0.5814,
+      "step": 10106
+    },
+    {
+      "epoch": 0.13457102746977395,
+      "grad_norm": 0.9631708264350891,
+      "learning_rate": 3.988313109368017e-05,
+      "loss": 0.587,
+      "step": 10137
+    },
+    {
+      "epoch": 0.13498255966387113,
+      "grad_norm": 1.1884409189224243,
+      "learning_rate": 3.981556864489504e-05,
+      "loss": 0.5906,
+      "step": 10168
+    },
+    {
+      "epoch": 0.1353940918579683,
+      "grad_norm": 0.9554224014282227,
+      "learning_rate": 3.974783900443142e-05,
+      "loss": 0.5888,
+      "step": 10199
+    },
+    {
+      "epoch": 0.13580562405206545,
+      "grad_norm": 1.0510096549987793,
+      "learning_rate": 3.9679942936609095e-05,
+      "loss": 0.5828,
+      "step": 10230
+    },
+    {
+      "epoch": 0.13621715624616262,
+      "grad_norm": 0.9769448637962341,
+      "learning_rate": 3.961188120762596e-05,
+      "loss": 0.5878,
+      "step": 10261
+    },
+    {
+      "epoch": 0.1366286884402598,
+      "grad_norm": 0.9610607028007507,
+      "learning_rate": 3.954365458554938e-05,
+      "loss": 0.5863,
+      "step": 10292
+    },
+    {
+      "epoch": 0.13704022063435697,
+      "grad_norm": 1.0102349519729614,
+      "learning_rate": 3.947526384030751e-05,
+      "loss": 0.5812,
+      "step": 10323
+    },
+    {
+      "epoch": 0.13745175282845412,
+      "grad_norm": 1.138122320175171,
+      "learning_rate": 3.9406709743680624e-05,
+      "loss": 0.5841,
+      "step": 10354
+    },
+    {
+      "epoch": 0.1378632850225513,
+      "grad_norm": 1.0107802152633667,
+      "learning_rate": 3.9337993069292366e-05,
+      "loss": 0.5847,
+      "step": 10385
+    },
+    {
+      "epoch": 0.13827481721664847,
+      "grad_norm": 0.9923847317695618,
+      "learning_rate": 3.926911459260109e-05,
+      "loss": 0.5971,
+      "step": 10416
+    },
+    {
+      "epoch": 0.13868634941074565,
+      "grad_norm": 1.0700715780258179,
+      "learning_rate": 3.920007509089102e-05,
+      "loss": 0.5831,
+      "step": 10447
+    },
+    {
+      "epoch": 0.1390978816048428,
+      "grad_norm": 1.1638612747192383,
+      "learning_rate": 3.913087534326357e-05,
+      "loss": 0.5849,
+      "step": 10478
+    },
+    {
+      "epoch": 0.13950941379893997,
+      "grad_norm": 1.0560393333435059,
+      "learning_rate": 3.9061516130628475e-05,
+      "loss": 0.5897,
+      "step": 10509
+    },
+    {
+      "epoch": 0.13992094599303714,
+      "grad_norm": 0.9604248404502869,
+      "learning_rate": 3.8991998235695025e-05,
+      "loss": 0.5862,
+      "step": 10540
+    },
+    {
+      "epoch": 0.14033247818713432,
+      "grad_norm": 1.0003983974456787,
+      "learning_rate": 3.8922322442963224e-05,
+      "loss": 0.5905,
+      "step": 10571
+    },
+    {
+      "epoch": 0.14074401038123147,
+      "grad_norm": 1.0378546714782715,
+      "learning_rate": 3.885248953871491e-05,
+      "loss": 0.5938,
+      "step": 10602
+    },
+    {
+      "epoch": 0.14115554257532864,
+      "grad_norm": 0.9967820644378662,
+      "learning_rate": 3.8782500311004915e-05,
+      "loss": 0.5821,
+      "step": 10633
+    },
+    {
+      "epoch": 0.14156707476942582,
+      "grad_norm": 1.029188632965088,
+      "learning_rate": 3.871235554965218e-05,
+      "loss": 0.5882,
+      "step": 10664
+    },
+    {
+      "epoch": 0.141978606963523,
+      "grad_norm": 1.1272863149642944,
+      "learning_rate": 3.864205604623078e-05,
+      "loss": 0.5898,
+      "step": 10695
+    },
+    {
+      "epoch": 0.14239013915762014,
+      "grad_norm": 1.0054222345352173,
+      "learning_rate": 3.857160259406107e-05,
+      "loss": 0.5873,
+      "step": 10726
+    },
+    {
+      "epoch": 0.1428016713517173,
+      "grad_norm": 1.0327833890914917,
+      "learning_rate": 3.8500995988200674e-05,
+      "loss": 0.5842,
+      "step": 10757
+    },
+    {
+      "epoch": 0.1432132035458145,
+      "grad_norm": 1.0088285207748413,
+      "learning_rate": 3.843023702543556e-05,
+      "loss": 0.5903,
+      "step": 10788
+    },
+    {
+      "epoch": 0.14362473573991166,
+      "grad_norm": 0.9472872018814087,
+      "learning_rate": 3.8359326504270984e-05,
+      "loss": 0.5809,
+      "step": 10819
+    },
+    {
+      "epoch": 0.1440362679340088,
+      "grad_norm": 1.1008390188217163,
+      "learning_rate": 3.828826522492255e-05,
+      "loss": 0.5775,
+      "step": 10850
+    },
+    {
+      "epoch": 0.14444780012810599,
+      "grad_norm": 1.1251379251480103,
+      "learning_rate": 3.821705398930713e-05,
+      "loss": 0.5821,
+      "step": 10881
+    },
+    {
+      "epoch": 0.14485933232220316,
+      "grad_norm": 0.9825986623764038,
+      "learning_rate": 3.814569360103385e-05,
+      "loss": 0.5749,
+      "step": 10912
+    },
+    {
+      "epoch": 0.14527086451630034,
+      "grad_norm": 0.9303480982780457,
+      "learning_rate": 3.807418486539499e-05,
+      "loss": 0.5827,
+      "step": 10943
+    },
+    {
+      "epoch": 0.14568239671039748,
+      "grad_norm": 0.9948610663414001,
+      "learning_rate": 3.80025285893569e-05,
+      "loss": 0.598,
+      "step": 10974
+    },
+    {
+      "epoch": 0.14609392890449466,
+      "grad_norm": 1.096449851989746,
+      "learning_rate": 3.793072558155093e-05,
+      "loss": 0.5749,
+      "step": 11005
+    },
+    {
+      "epoch": 0.14650546109859183,
+      "grad_norm": 1.022290587425232,
+      "learning_rate": 3.785877665226426e-05,
+      "loss": 0.5824,
+      "step": 11036
+    },
+    {
+      "epoch": 0.146916993292689,
+      "grad_norm": 1.083853840827942,
+      "learning_rate": 3.778668261343079e-05,
+      "loss": 0.5923,
+      "step": 11067
+    },
+    {
+      "epoch": 0.14732852548678615,
+      "grad_norm": 0.919503390789032,
+      "learning_rate": 3.771444427862192e-05,
+      "loss": 0.5924,
+      "step": 11098
+    },
+    {
+      "epoch": 0.14774005768088333,
+      "grad_norm": 1.0682687759399414,
+      "learning_rate": 3.7642062463037465e-05,
+      "loss": 0.5824,
+      "step": 11129
+    },
+    {
+      "epoch": 0.1481515898749805,
+      "grad_norm": 0.9599080085754395,
+      "learning_rate": 3.7569537983496373e-05,
+      "loss": 0.5765,
+      "step": 11160
+    },
+    {
+      "epoch": 0.14856312206907768,
+      "grad_norm": 0.968349039554596,
+      "learning_rate": 3.749687165842753e-05,
+      "loss": 0.5932,
+      "step": 11191
+    },
+    {
+      "epoch": 0.14897465426317483,
+      "grad_norm": 0.9846999049186707,
+      "learning_rate": 3.7424064307860536e-05,
+      "loss": 0.5791,
+      "step": 11222
+    },
+    {
+      "epoch": 0.149386186457272,
+      "grad_norm": 1.0673292875289917,
+      "learning_rate": 3.735111675341645e-05,
+      "loss": 0.5803,
+      "step": 11253
+    },
+    {
+      "epoch": 0.14979771865136918,
+      "grad_norm": 1.0111202001571655,
+      "learning_rate": 3.7278029818298524e-05,
+      "loss": 0.5837,
+      "step": 11284
+    },
+    {
+      "epoch": 0.15020925084546635,
+      "grad_norm": 1.0955816507339478,
+      "learning_rate": 3.720480432728287e-05,
+      "loss": 0.5822,
+      "step": 11315
+    },
+    {
+      "epoch": 0.1506207830395635,
+      "grad_norm": 0.9147946238517761,
+      "learning_rate": 3.71314411067092e-05,
+      "loss": 0.5873,
+      "step": 11346
+    },
+    {
+      "epoch": 0.15103231523366067,
+      "grad_norm": 0.9630762934684753,
+      "learning_rate": 3.70579409844715e-05,
+      "loss": 0.5914,
+      "step": 11377
+    },
+    {
+      "epoch": 0.15144384742775785,
+      "grad_norm": 0.9123234748840332,
+      "learning_rate": 3.698430479000865e-05,
+      "loss": 0.5784,
+      "step": 11408
+    },
+    {
+      "epoch": 0.15185537962185502,
+      "grad_norm": 1.0654667615890503,
+      "learning_rate": 3.691053335429509e-05,
+      "loss": 0.5805,
+      "step": 11439
+    },
+    {
+      "epoch": 0.15226691181595217,
+      "grad_norm": 1.111958622932434,
+      "learning_rate": 3.683662750983147e-05,
+      "loss": 0.5868,
+      "step": 11470
+    },
+    {
+      "epoch": 0.15267844401004935,
+      "grad_norm": 1.024422526359558,
+      "learning_rate": 3.676258809063518e-05,
+      "loss": 0.5904,
+      "step": 11501
+    },
+    {
+      "epoch": 0.15308997620414652,
+      "grad_norm": 1.092668056488037,
+      "learning_rate": 3.6688415932231004e-05,
+      "loss": 0.5885,
+      "step": 11532
+    },
+    {
+      "epoch": 0.1535015083982437,
+      "grad_norm": 0.9739312529563904,
+      "learning_rate": 3.661411187164166e-05,
+      "loss": 0.5806,
+      "step": 11563
+    },
+    {
+      "epoch": 0.15391304059234084,
+      "grad_norm": 1.098699688911438,
+      "learning_rate": 3.65396767473784e-05,
+      "loss": 0.5755,
+      "step": 11594
+    },
+    {
+      "epoch": 0.15432457278643802,
+      "grad_norm": 0.9998940825462341,
+      "learning_rate": 3.6465111399431465e-05,
+      "loss": 0.5758,
+      "step": 11625
+    },
+    {
+      "epoch": 0.1547361049805352,
+      "grad_norm": 0.9077128767967224,
+      "learning_rate": 3.6390416669260674e-05,
+      "loss": 0.5742,
+      "step": 11656
+    },
+    {
+      "epoch": 0.15514763717463237,
+      "grad_norm": 0.9274038076400757,
+      "learning_rate": 3.63155933997859e-05,
+      "loss": 0.589,
+      "step": 11687
+    },
+    {
+      "epoch": 0.15555916936872952,
+      "grad_norm": 0.9404906034469604,
+      "learning_rate": 3.624064243537758e-05,
+      "loss": 0.5848,
+      "step": 11718
+    },
+    {
+      "epoch": 0.1559707015628267,
+      "grad_norm": 0.9896044731140137,
+      "learning_rate": 3.616556462184716e-05,
+      "loss": 0.5917,
+      "step": 11749
+    },
+    {
+      "epoch": 0.15638223375692387,
+      "grad_norm": 1.0280394554138184,
+      "learning_rate": 3.609036080643755e-05,
+      "loss": 0.5811,
+      "step": 11780
+    },
+    {
+      "epoch": 0.15679376595102104,
+      "grad_norm": 0.9462539553642273,
+      "learning_rate": 3.60150318378136e-05,
+      "loss": 0.5782,
+      "step": 11811
+    },
+    {
+      "epoch": 0.1572052981451182,
+      "grad_norm": 1.047635555267334,
+      "learning_rate": 3.5939578566052465e-05,
+      "loss": 0.5792,
+      "step": 11842
+    },
+    {
+      "epoch": 0.15761683033921536,
+      "grad_norm": 0.9613581299781799,
+      "learning_rate": 3.586400184263408e-05,
+      "loss": 0.5847,
+      "step": 11873
+    },
+    {
+      "epoch": 0.15802836253331254,
+      "grad_norm": 0.8944469094276428,
+      "learning_rate": 3.578830252043148e-05,
+      "loss": 0.5787,
+      "step": 11904
+    },
+    {
+      "epoch": 0.1584398947274097,
+      "grad_norm": 1.0031458139419556,
+      "learning_rate": 3.571248145370125e-05,
+      "loss": 0.5827,
+      "step": 11935
+    },
+    {
+      "epoch": 0.15885142692150686,
+      "grad_norm": 0.9425063729286194,
+      "learning_rate": 3.5636539498073794e-05,
+      "loss": 0.5899,
+      "step": 11966
+    },
+    {
+      "epoch": 0.15926295911560404,
+      "grad_norm": 1.0098518133163452,
+      "learning_rate": 3.556047751054378e-05,
+      "loss": 0.5849,
+      "step": 11997
+    },
+    {
+      "epoch": 0.1596744913097012,
+      "grad_norm": 0.8891544342041016,
+      "learning_rate": 3.548429634946039e-05,
+      "loss": 0.5772,
+      "step": 12028
+    },
+    {
+      "epoch": 0.16008602350379839,
+      "grad_norm": 1.1580991744995117,
+      "learning_rate": 3.540799687451768e-05,
+      "loss": 0.5859,
+      "step": 12059
+    },
+    {
+      "epoch": 0.16049755569789553,
+      "grad_norm": 1.046398639678955,
+      "learning_rate": 3.533157994674485e-05,
+      "loss": 0.582,
+      "step": 12090
+    },
+    {
+      "epoch": 0.1609090878919927,
+      "grad_norm": 1.0526766777038574,
+      "learning_rate": 3.5255046428496546e-05,
+      "loss": 0.5738,
+      "step": 12121
+    },
+    {
+      "epoch": 0.16132062008608988,
+      "grad_norm": 0.9915657043457031,
+      "learning_rate": 3.517839718344311e-05,
+      "loss": 0.5867,
+      "step": 12152
+    },
+    {
+      "epoch": 0.16173215228018706,
+      "grad_norm": 0.954210102558136,
+      "learning_rate": 3.510163307656086e-05,
+      "loss": 0.578,
+      "step": 12183
+    }
+  ],
+  "logging_steps": 31,
+  "max_steps": 30517,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 3052,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.006624315386364e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-12208/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1bf5225534ee746c36dba9f28281e0b858e6e9b4a308bdaf1de4f68e79e9ee83
+size 5304

checkpoint-15260/config.json

@@ -0,0 +1,39 @@
+{
+  "_name_or_path": "meta-llama/Llama-3.1-8B-Instruct",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": [
+    128001,
+    128008,
+    128009
+  ],
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.44.0.dev0",
+  "use_cache": true,
+  "vocab_size": 128256
+}

checkpoint-15260/generation_config.json

@@ -0,0 +1,12 @@
+{
+  "bos_token_id": 128000,
+  "do_sample": true,
+  "eos_token_id": [
+    128001,
+    128008,
+    128009
+  ],
+  "temperature": 0.6,
+  "top_p": 0.9,
+  "transformers_version": "4.44.0.dev0"
+}

checkpoint-15260/model-00001-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1041ad7bc3e2d487dbcb5abb5d2444828ba8b7af4227311ded8ea9c17d5f85e0
+size 4886466168

checkpoint-15260/model-00002-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23fa5aaf8ac56c17c821980c9aee44fcbc2a617fc570e875e6fdb93d92886c72
+size 4832007448

checkpoint-15260/model-00003-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a223e87240e50dfb05634ea909cf1ec715d37a12b681b14a3a6e97f28688f2b
+size 4999813112

checkpoint-15260/model-00004-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3047d199c611c6d12b1cca780f033ad89df9c076f742674f7ac4a892ca9573b3
+size 4999813128

checkpoint-15260/model-00005-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a2f91d749d664ecd22c71f3d54488966fc6899a03924b8fea3613c8f83398a8
+size 4832007496

checkpoint-15260/model-00006-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62e6de3c01ef59365296799084663b39e1a90e5d9635d63077f3baf2bdff970c
+size 4999813120

checkpoint-15260/model-00007-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b48b79e714fe0c8cd276f8bce7e9e22aed9445c9d775ae7ba28e76bfd00b4549
+size 2571158184

checkpoint-15260/model.safetensors.index.json

@@ -0,0 +1,298 @@
+{
+  "metadata": {
+    "total_size": 32121044992
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00007-of-00007.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00007-of-00007.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00007-of-00007.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00007-of-00007.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00007-of-00007.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.norm.weight": "model-00007-of-00007.safetensors"
+  }
+}

checkpoint-15260/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b189fc7ec92c48e8b419723e368a977a52262356ce9fad323467adc7754440d6
+size 15385036334

checkpoint-15260/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9d9cd6a0487226e5bd30d1846894c82af483733ab4381b75bae9c0745e05d405
+size 14244

checkpoint-15260/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ce335347efaebf038b8915bc0a8d2d587a4a1aa08ad0e015b4bc7cc4fba634e
+size 1064

checkpoint-15260/trainer_state.json

@@ -0,0 +1,3477 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.2025800413523479,
+  "eval_steps": 500,
+  "global_step": 15260,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00041153219409716807,
+      "grad_norm": 4.147768020629883,
+      "learning_rate": 1.0157273918741808e-06,
+      "loss": 0.7978,
+      "step": 31
+    },
+    {
+      "epoch": 0.0008230643881943361,
+      "grad_norm": 3.117840051651001,
+      "learning_rate": 2.0314547837483616e-06,
+      "loss": 0.7723,
+      "step": 62
+    },
+    {
+      "epoch": 0.0012345965822915042,
+      "grad_norm": 2.83072829246521,
+      "learning_rate": 3.0471821756225426e-06,
+      "loss": 0.7493,
+      "step": 93
+    },
+    {
+      "epoch": 0.0016461287763886723,
+      "grad_norm": 2.9491968154907227,
+      "learning_rate": 4.062909567496723e-06,
+      "loss": 0.7352,
+      "step": 124
+    },
+    {
+      "epoch": 0.0020576609704858403,
+      "grad_norm": 2.798553228378296,
+      "learning_rate": 5.078636959370905e-06,
+      "loss": 0.7229,
+      "step": 155
+    },
+    {
+      "epoch": 0.0024691931645830084,
+      "grad_norm": 2.663148880004883,
+      "learning_rate": 6.094364351245085e-06,
+      "loss": 0.7253,
+      "step": 186
+    },
+    {
+      "epoch": 0.0028807253586801765,
+      "grad_norm": 2.564692974090576,
+      "learning_rate": 7.110091743119267e-06,
+      "loss": 0.7287,
+      "step": 217
+    },
+    {
+      "epoch": 0.0032922575527773446,
+      "grad_norm": 3.2043237686157227,
+      "learning_rate": 8.125819134993446e-06,
+      "loss": 0.7205,
+      "step": 248
+    },
+    {
+      "epoch": 0.0037037897468745126,
+      "grad_norm": 2.6705269813537598,
+      "learning_rate": 9.141546526867629e-06,
+      "loss": 0.714,
+      "step": 279
+    },
+    {
+      "epoch": 0.004115321940971681,
+      "grad_norm": 3.432569980621338,
+      "learning_rate": 1.015727391874181e-05,
+      "loss": 0.7217,
+      "step": 310
+    },
+    {
+      "epoch": 0.004526854135068849,
+      "grad_norm": 2.515808343887329,
+      "learning_rate": 1.117300131061599e-05,
+      "loss": 0.7173,
+      "step": 341
+    },
+    {
+      "epoch": 0.004938386329166017,
+      "grad_norm": 2.6708128452301025,
+      "learning_rate": 1.218872870249017e-05,
+      "loss": 0.7123,
+      "step": 372
+    },
+    {
+      "epoch": 0.005349918523263185,
+      "grad_norm": 2.472797155380249,
+      "learning_rate": 1.3204456094364351e-05,
+      "loss": 0.7079,
+      "step": 403
+    },
+    {
+      "epoch": 0.005761450717360353,
+      "grad_norm": 2.486147880554199,
+      "learning_rate": 1.4220183486238533e-05,
+      "loss": 0.7146,
+      "step": 434
+    },
+    {
+      "epoch": 0.006172982911457521,
+      "grad_norm": 2.310777187347412,
+      "learning_rate": 1.5235910878112714e-05,
+      "loss": 0.7,
+      "step": 465
+    },
+    {
+      "epoch": 0.006584515105554689,
+      "grad_norm": 3.1831858158111572,
+      "learning_rate": 1.6251638269986893e-05,
+      "loss": 0.7042,
+      "step": 496
+    },
+    {
+      "epoch": 0.006996047299651857,
+      "grad_norm": 2.505427360534668,
+      "learning_rate": 1.7267365661861077e-05,
+      "loss": 0.6991,
+      "step": 527
+    },
+    {
+      "epoch": 0.007407579493749025,
+      "grad_norm": 2.3627443313598633,
+      "learning_rate": 1.8283093053735257e-05,
+      "loss": 0.7045,
+      "step": 558
+    },
+    {
+      "epoch": 0.007819111687846193,
+      "grad_norm": 2.6250648498535156,
+      "learning_rate": 1.9298820445609438e-05,
+      "loss": 0.704,
+      "step": 589
+    },
+    {
+      "epoch": 0.008230643881943361,
+      "grad_norm": 2.3151464462280273,
+      "learning_rate": 2.031454783748362e-05,
+      "loss": 0.6959,
+      "step": 620
+    },
+    {
+      "epoch": 0.00864217607604053,
+      "grad_norm": 2.0312647819519043,
+      "learning_rate": 2.13302752293578e-05,
+      "loss": 0.689,
+      "step": 651
+    },
+    {
+      "epoch": 0.009053708270137698,
+      "grad_norm": 2.718970775604248,
+      "learning_rate": 2.234600262123198e-05,
+      "loss": 0.7031,
+      "step": 682
+    },
+    {
+      "epoch": 0.009465240464234866,
+      "grad_norm": 1.9124070405960083,
+      "learning_rate": 2.336173001310616e-05,
+      "loss": 0.6908,
+      "step": 713
+    },
+    {
+      "epoch": 0.009876772658332034,
+      "grad_norm": 2.368354082107544,
+      "learning_rate": 2.437745740498034e-05,
+      "loss": 0.6947,
+      "step": 744
+    },
+    {
+      "epoch": 0.010288304852429202,
+      "grad_norm": 1.8856632709503174,
+      "learning_rate": 2.5393184796854525e-05,
+      "loss": 0.6876,
+      "step": 775
+    },
+    {
+      "epoch": 0.01069983704652637,
+      "grad_norm": 2.3155925273895264,
+      "learning_rate": 2.6408912188728702e-05,
+      "loss": 0.7024,
+      "step": 806
+    },
+    {
+      "epoch": 0.011111369240623538,
+      "grad_norm": 2.2516837120056152,
+      "learning_rate": 2.7424639580602886e-05,
+      "loss": 0.695,
+      "step": 837
+    },
+    {
+      "epoch": 0.011522901434720706,
+      "grad_norm": 2.4444823265075684,
+      "learning_rate": 2.8440366972477066e-05,
+      "loss": 0.6892,
+      "step": 868
+    },
+    {
+      "epoch": 0.011934433628817874,
+      "grad_norm": 2.4233832359313965,
+      "learning_rate": 2.9456094364351244e-05,
+      "loss": 0.6871,
+      "step": 899
+    },
+    {
+      "epoch": 0.012345965822915042,
+      "grad_norm": 2.0262961387634277,
+      "learning_rate": 3.0471821756225428e-05,
+      "loss": 0.6888,
+      "step": 930
+    },
+    {
+      "epoch": 0.01275749801701221,
+      "grad_norm": 2.3908143043518066,
+      "learning_rate": 3.148754914809961e-05,
+      "loss": 0.6997,
+      "step": 961
+    },
+    {
+      "epoch": 0.013169030211109378,
+      "grad_norm": 1.839417576789856,
+      "learning_rate": 3.2503276539973785e-05,
+      "loss": 0.6826,
+      "step": 992
+    },
+    {
+      "epoch": 0.013580562405206546,
+      "grad_norm": 1.8175997734069824,
+      "learning_rate": 3.351900393184797e-05,
+      "loss": 0.6857,
+      "step": 1023
+    },
+    {
+      "epoch": 0.013992094599303714,
+      "grad_norm": 2.3080506324768066,
+      "learning_rate": 3.453473132372215e-05,
+      "loss": 0.6896,
+      "step": 1054
+    },
+    {
+      "epoch": 0.014403626793400882,
+      "grad_norm": 2.0574960708618164,
+      "learning_rate": 3.555045871559633e-05,
+      "loss": 0.6862,
+      "step": 1085
+    },
+    {
+      "epoch": 0.01481515898749805,
+      "grad_norm": 1.957221269607544,
+      "learning_rate": 3.6566186107470514e-05,
+      "loss": 0.6796,
+      "step": 1116
+    },
+    {
+      "epoch": 0.015226691181595219,
+      "grad_norm": 2.2740609645843506,
+      "learning_rate": 3.7581913499344695e-05,
+      "loss": 0.7082,
+      "step": 1147
+    },
+    {
+      "epoch": 0.015638223375692387,
+      "grad_norm": 2.1291615962982178,
+      "learning_rate": 3.8597640891218876e-05,
+      "loss": 0.6852,
+      "step": 1178
+    },
+    {
+      "epoch": 0.016049755569789555,
+      "grad_norm": 2.2918620109558105,
+      "learning_rate": 3.9613368283093056e-05,
+      "loss": 0.6821,
+      "step": 1209
+    },
+    {
+      "epoch": 0.016461287763886723,
+      "grad_norm": 1.9689708948135376,
+      "learning_rate": 4.062909567496724e-05,
+      "loss": 0.676,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01687281995798389,
+      "grad_norm": 2.4405298233032227,
+      "learning_rate": 4.164482306684142e-05,
+      "loss": 0.6927,
+      "step": 1271
+    },
+    {
+      "epoch": 0.01728435215208106,
+      "grad_norm": 2.0007450580596924,
+      "learning_rate": 4.26605504587156e-05,
+      "loss": 0.6864,
+      "step": 1302
+    },
+    {
+      "epoch": 0.017695884346178227,
+      "grad_norm": 1.7013226747512817,
+      "learning_rate": 4.367627785058978e-05,
+      "loss": 0.6827,
+      "step": 1333
+    },
+    {
+      "epoch": 0.018107416540275395,
+      "grad_norm": 2.2202930450439453,
+      "learning_rate": 4.469200524246396e-05,
+      "loss": 0.6722,
+      "step": 1364
+    },
+    {
+      "epoch": 0.018518948734372563,
+      "grad_norm": 1.690030813217163,
+      "learning_rate": 4.570773263433814e-05,
+      "loss": 0.6807,
+      "step": 1395
+    },
+    {
+      "epoch": 0.01893048092846973,
+      "grad_norm": 1.5886859893798828,
+      "learning_rate": 4.672346002621232e-05,
+      "loss": 0.6867,
+      "step": 1426
+    },
+    {
+      "epoch": 0.0193420131225669,
+      "grad_norm": 1.7371172904968262,
+      "learning_rate": 4.77391874180865e-05,
+      "loss": 0.6832,
+      "step": 1457
+    },
+    {
+      "epoch": 0.019753545316664067,
+      "grad_norm": 1.8376264572143555,
+      "learning_rate": 4.875491480996068e-05,
+      "loss": 0.681,
+      "step": 1488
+    },
+    {
+      "epoch": 0.020165077510761235,
+      "grad_norm": 1.6280311346054077,
+      "learning_rate": 4.977064220183487e-05,
+      "loss": 0.6795,
+      "step": 1519
+    },
+    {
+      "epoch": 0.020576609704858403,
+      "grad_norm": 1.8016823530197144,
+      "learning_rate": 4.9999915451558777e-05,
+      "loss": 0.6789,
+      "step": 1550
+    },
+    {
+      "epoch": 0.02098814189895557,
+      "grad_norm": 1.7355237007141113,
+      "learning_rate": 4.999955597496219e-05,
+      "loss": 0.6753,
+      "step": 1581
+    },
+    {
+      "epoch": 0.02139967409305274,
+      "grad_norm": 1.5687408447265625,
+      "learning_rate": 4.9998914381774255e-05,
+      "loss": 0.6777,
+      "step": 1612
+    },
+    {
+      "epoch": 0.021811206287149908,
+      "grad_norm": 1.790453314781189,
+      "learning_rate": 4.999799067923527e-05,
+      "loss": 0.684,
+      "step": 1643
+    },
+    {
+      "epoch": 0.022222738481247076,
+      "grad_norm": 1.9073406457901,
+      "learning_rate": 4.999678487776908e-05,
+      "loss": 0.6793,
+      "step": 1674
+    },
+    {
+      "epoch": 0.022634270675344244,
+      "grad_norm": 1.5575100183486938,
+      "learning_rate": 4.9995296990983006e-05,
+      "loss": 0.6794,
+      "step": 1705
+    },
+    {
+      "epoch": 0.023045802869441412,
+      "grad_norm": 2.100715160369873,
+      "learning_rate": 4.999352703566763e-05,
+      "loss": 0.6753,
+      "step": 1736
+    },
+    {
+      "epoch": 0.02345733506353858,
+      "grad_norm": 1.9425480365753174,
+      "learning_rate": 4.999147503179668e-05,
+      "loss": 0.6755,
+      "step": 1767
+    },
+    {
+      "epoch": 0.023868867257635748,
+      "grad_norm": 1.5791170597076416,
+      "learning_rate": 4.998914100252672e-05,
+      "loss": 0.6754,
+      "step": 1798
+    },
+    {
+      "epoch": 0.024280399451732916,
+      "grad_norm": 1.5692673921585083,
+      "learning_rate": 4.998652497419696e-05,
+      "loss": 0.6682,
+      "step": 1829
+    },
+    {
+      "epoch": 0.024691931645830084,
+      "grad_norm": 1.5757725238800049,
+      "learning_rate": 4.9983626976328927e-05,
+      "loss": 0.6723,
+      "step": 1860
+    },
+    {
+      "epoch": 0.025103463839927252,
+      "grad_norm": 1.8000696897506714,
+      "learning_rate": 4.998044704162613e-05,
+      "loss": 0.6647,
+      "step": 1891
+    },
+    {
+      "epoch": 0.02551499603402442,
+      "grad_norm": 1.682919979095459,
+      "learning_rate": 4.9976985205973705e-05,
+      "loss": 0.669,
+      "step": 1922
+    },
+    {
+      "epoch": 0.02592652822812159,
+      "grad_norm": 1.6138657331466675,
+      "learning_rate": 4.997324150843799e-05,
+      "loss": 0.6698,
+      "step": 1953
+    },
+    {
+      "epoch": 0.026338060422218756,
+      "grad_norm": 1.5016391277313232,
+      "learning_rate": 4.99692159912661e-05,
+      "loss": 0.6645,
+      "step": 1984
+    },
+    {
+      "epoch": 0.026749592616315924,
+      "grad_norm": 1.6850241422653198,
+      "learning_rate": 4.996490869988546e-05,
+      "loss": 0.6769,
+      "step": 2015
+    },
+    {
+      "epoch": 0.027161124810413093,
+      "grad_norm": 1.521161437034607,
+      "learning_rate": 4.996031968290326e-05,
+      "loss": 0.6599,
+      "step": 2046
+    },
+    {
+      "epoch": 0.02757265700451026,
+      "grad_norm": 1.8039467334747314,
+      "learning_rate": 4.995544899210594e-05,
+      "loss": 0.6693,
+      "step": 2077
+    },
+    {
+      "epoch": 0.02798418919860743,
+      "grad_norm": 1.688811182975769,
+      "learning_rate": 4.9950296682458583e-05,
+      "loss": 0.661,
+      "step": 2108
+    },
+    {
+      "epoch": 0.028395721392704597,
+      "grad_norm": 1.669871211051941,
+      "learning_rate": 4.994486281210429e-05,
+      "loss": 0.6611,
+      "step": 2139
+    },
+    {
+      "epoch": 0.028807253586801765,
+      "grad_norm": 1.3887263536453247,
+      "learning_rate": 4.9939147442363566e-05,
+      "loss": 0.6677,
+      "step": 2170
+    },
+    {
+      "epoch": 0.029218785780898933,
+      "grad_norm": 1.3242331743240356,
+      "learning_rate": 4.9933150637733574e-05,
+      "loss": 0.6537,
+      "step": 2201
+    },
+    {
+      "epoch": 0.0296303179749961,
+      "grad_norm": 1.6638622283935547,
+      "learning_rate": 4.992687246588743e-05,
+      "loss": 0.6653,
+      "step": 2232
+    },
+    {
+      "epoch": 0.03004185016909327,
+      "grad_norm": 1.5110372304916382,
+      "learning_rate": 4.992031299767347e-05,
+      "loss": 0.6615,
+      "step": 2263
+    },
+    {
+      "epoch": 0.030453382363190437,
+      "grad_norm": 1.5244065523147583,
+      "learning_rate": 4.9913472307114386e-05,
+      "loss": 0.6552,
+      "step": 2294
+    },
+    {
+      "epoch": 0.030864914557287605,
+      "grad_norm": 1.416624903678894,
+      "learning_rate": 4.9906350471406446e-05,
+      "loss": 0.6569,
+      "step": 2325
+    },
+    {
+      "epoch": 0.03127644675138477,
+      "grad_norm": 1.4467304944992065,
+      "learning_rate": 4.989894757091861e-05,
+      "loss": 0.6613,
+      "step": 2356
+    },
+    {
+      "epoch": 0.03168797894548194,
+      "grad_norm": 1.4649864435195923,
+      "learning_rate": 4.989126368919158e-05,
+      "loss": 0.6604,
+      "step": 2387
+    },
+    {
+      "epoch": 0.03209951113957911,
+      "grad_norm": 1.6256624460220337,
+      "learning_rate": 4.988329891293693e-05,
+      "loss": 0.655,
+      "step": 2418
+    },
+    {
+      "epoch": 0.03251104333367628,
+      "grad_norm": 1.3986462354660034,
+      "learning_rate": 4.987505333203608e-05,
+      "loss": 0.6514,
+      "step": 2449
+    },
+    {
+      "epoch": 0.032922575527773446,
+      "grad_norm": 1.4490609169006348,
+      "learning_rate": 4.9866527039539276e-05,
+      "loss": 0.65,
+      "step": 2480
+    },
+    {
+      "epoch": 0.033334107721870614,
+      "grad_norm": 1.4979615211486816,
+      "learning_rate": 4.9857720131664594e-05,
+      "loss": 0.6579,
+      "step": 2511
+    },
+    {
+      "epoch": 0.03374563991596778,
+      "grad_norm": 1.6314005851745605,
+      "learning_rate": 4.9848632707796773e-05,
+      "loss": 0.6604,
+      "step": 2542
+    },
+    {
+      "epoch": 0.03415717211006495,
+      "grad_norm": 1.310655951499939,
+      "learning_rate": 4.9839264870486155e-05,
+      "loss": 0.6518,
+      "step": 2573
+    },
+    {
+      "epoch": 0.03456870430416212,
+      "grad_norm": 1.4176305532455444,
+      "learning_rate": 4.9829616725447526e-05,
+      "loss": 0.6518,
+      "step": 2604
+    },
+    {
+      "epoch": 0.034980236498259286,
+      "grad_norm": 1.6330211162567139,
+      "learning_rate": 4.981968838155888e-05,
+      "loss": 0.6474,
+      "step": 2635
+    },
+    {
+      "epoch": 0.035391768692356454,
+      "grad_norm": 1.5854465961456299,
+      "learning_rate": 4.980947995086024e-05,
+      "loss": 0.6462,
+      "step": 2666
+    },
+    {
+      "epoch": 0.03580330088645362,
+      "grad_norm": 1.2944835424423218,
+      "learning_rate": 4.979899154855234e-05,
+      "loss": 0.6503,
+      "step": 2697
+    },
+    {
+      "epoch": 0.03621483308055079,
+      "grad_norm": 1.4056243896484375,
+      "learning_rate": 4.9788223292995386e-05,
+      "loss": 0.6414,
+      "step": 2728
+    },
+    {
+      "epoch": 0.03662636527464796,
+      "grad_norm": 1.5649598836898804,
+      "learning_rate": 4.977717530570768e-05,
+      "loss": 0.6449,
+      "step": 2759
+    },
+    {
+      "epoch": 0.037037897468745126,
+      "grad_norm": 1.5205659866333008,
+      "learning_rate": 4.976584771136425e-05,
+      "loss": 0.6334,
+      "step": 2790
+    },
+    {
+      "epoch": 0.037449429662842294,
+      "grad_norm": 1.3248411417007446,
+      "learning_rate": 4.975424063779547e-05,
+      "loss": 0.6448,
+      "step": 2821
+    },
+    {
+      "epoch": 0.03786096185693946,
+      "grad_norm": 1.2402842044830322,
+      "learning_rate": 4.974235421598557e-05,
+      "loss": 0.6476,
+      "step": 2852
+    },
+    {
+      "epoch": 0.03827249405103663,
+      "grad_norm": 1.7939503192901611,
+      "learning_rate": 4.973018858007122e-05,
+      "loss": 0.6472,
+      "step": 2883
+    },
+    {
+      "epoch": 0.0386840262451338,
+      "grad_norm": 1.2774665355682373,
+      "learning_rate": 4.9717743867339963e-05,
+      "loss": 0.6374,
+      "step": 2914
+    },
+    {
+      "epoch": 0.03909555843923097,
+      "grad_norm": 1.6384273767471313,
+      "learning_rate": 4.9705020218228695e-05,
+      "loss": 0.6623,
+      "step": 2945
+    },
+    {
+      "epoch": 0.039507090633328135,
+      "grad_norm": 1.559270977973938,
+      "learning_rate": 4.969201777632205e-05,
+      "loss": 0.6391,
+      "step": 2976
+    },
+    {
+      "epoch": 0.0399186228274253,
+      "grad_norm": 1.6405760049819946,
+      "learning_rate": 4.9678736688350846e-05,
+      "loss": 0.6437,
+      "step": 3007
+    },
+    {
+      "epoch": 0.04033015502152247,
+      "grad_norm": 1.3599917888641357,
+      "learning_rate": 4.966517710419033e-05,
+      "loss": 0.6436,
+      "step": 3038
+    },
+    {
+      "epoch": 0.04074168721561964,
+      "grad_norm": 1.3080962896347046,
+      "learning_rate": 4.965133917685858e-05,
+      "loss": 0.6331,
+      "step": 3069
+    },
+    {
+      "epoch": 0.04115321940971681,
+      "grad_norm": 1.435195803642273,
+      "learning_rate": 4.9637223062514714e-05,
+      "loss": 0.6417,
+      "step": 3100
+    },
+    {
+      "epoch": 0.041564751603813975,
+      "grad_norm": 1.342986822128296,
+      "learning_rate": 4.962282892045718e-05,
+      "loss": 0.6391,
+      "step": 3131
+    },
+    {
+      "epoch": 0.04197628379791114,
+      "grad_norm": 1.524418592453003,
+      "learning_rate": 4.9608156913121904e-05,
+      "loss": 0.6511,
+      "step": 3162
+    },
+    {
+      "epoch": 0.04238781599200831,
+      "grad_norm": 1.4617900848388672,
+      "learning_rate": 4.959320720608049e-05,
+      "loss": 0.6367,
+      "step": 3193
+    },
+    {
+      "epoch": 0.04279934818610548,
+      "grad_norm": 1.3968737125396729,
+      "learning_rate": 4.9577979968038354e-05,
+      "loss": 0.6387,
+      "step": 3224
+    },
+    {
+      "epoch": 0.04321088038020265,
+      "grad_norm": 1.258406400680542,
+      "learning_rate": 4.956247537083282e-05,
+      "loss": 0.6356,
+      "step": 3255
+    },
+    {
+      "epoch": 0.043622412574299815,
+      "grad_norm": 1.5494844913482666,
+      "learning_rate": 4.9546693589431145e-05,
+      "loss": 0.6433,
+      "step": 3286
+    },
+    {
+      "epoch": 0.04403394476839698,
+      "grad_norm": 1.2248284816741943,
+      "learning_rate": 4.9530634801928595e-05,
+      "loss": 0.6376,
+      "step": 3317
+    },
+    {
+      "epoch": 0.04444547696249415,
+      "grad_norm": 1.4243425130844116,
+      "learning_rate": 4.9514299189546395e-05,
+      "loss": 0.6341,
+      "step": 3348
+    },
+    {
+      "epoch": 0.04485700915659132,
+      "grad_norm": 1.379364252090454,
+      "learning_rate": 4.949768693662973e-05,
+      "loss": 0.6378,
+      "step": 3379
+    },
+    {
+      "epoch": 0.04526854135068849,
+      "grad_norm": 1.3460108041763306,
+      "learning_rate": 4.948079823064559e-05,
+      "loss": 0.639,
+      "step": 3410
+    },
+    {
+      "epoch": 0.045680073544785656,
+      "grad_norm": 1.382348656654358,
+      "learning_rate": 4.946363326218074e-05,
+      "loss": 0.6327,
+      "step": 3441
+    },
+    {
+      "epoch": 0.046091605738882824,
+      "grad_norm": 1.40754234790802,
+      "learning_rate": 4.9446192224939525e-05,
+      "loss": 0.6234,
+      "step": 3472
+    },
+    {
+      "epoch": 0.04650313793297999,
+      "grad_norm": 1.502922773361206,
+      "learning_rate": 4.942847531574167e-05,
+      "loss": 0.6404,
+      "step": 3503
+    },
+    {
+      "epoch": 0.04691467012707716,
+      "grad_norm": 1.2896559238433838,
+      "learning_rate": 4.941048273452008e-05,
+      "loss": 0.6378,
+      "step": 3534
+    },
+    {
+      "epoch": 0.04732620232117433,
+      "grad_norm": 1.3114830255508423,
+      "learning_rate": 4.9392214684318605e-05,
+      "loss": 0.6373,
+      "step": 3565
+    },
+    {
+      "epoch": 0.047737734515271496,
+      "grad_norm": 1.4122483730316162,
+      "learning_rate": 4.93736713712897e-05,
+      "loss": 0.6305,
+      "step": 3596
+    },
+    {
+      "epoch": 0.048149266709368664,
+      "grad_norm": 1.2669744491577148,
+      "learning_rate": 4.9354853004692124e-05,
+      "loss": 0.6413,
+      "step": 3627
+    },
+    {
+      "epoch": 0.04856079890346583,
+      "grad_norm": 1.2965583801269531,
+      "learning_rate": 4.93357597968886e-05,
+      "loss": 0.6364,
+      "step": 3658
+    },
+    {
+      "epoch": 0.048972331097563,
+      "grad_norm": 1.3328967094421387,
+      "learning_rate": 4.931639196334338e-05,
+      "loss": 0.6466,
+      "step": 3689
+    },
+    {
+      "epoch": 0.04938386329166017,
+      "grad_norm": 1.6291115283966064,
+      "learning_rate": 4.9296749722619826e-05,
+      "loss": 0.6348,
+      "step": 3720
+    },
+    {
+      "epoch": 0.049795395485757336,
+      "grad_norm": 1.3224235773086548,
+      "learning_rate": 4.9276833296377966e-05,
+      "loss": 0.6256,
+      "step": 3751
+    },
+    {
+      "epoch": 0.050206927679854504,
+      "grad_norm": 1.3253369331359863,
+      "learning_rate": 4.925664290937196e-05,
+      "loss": 0.6303,
+      "step": 3782
+    },
+    {
+      "epoch": 0.05061845987395167,
+      "grad_norm": 1.2806897163391113,
+      "learning_rate": 4.9236178789447576e-05,
+      "loss": 0.6403,
+      "step": 3813
+    },
+    {
+      "epoch": 0.05102999206804884,
+      "grad_norm": 1.3389384746551514,
+      "learning_rate": 4.921544116753962e-05,
+      "loss": 0.6288,
+      "step": 3844
+    },
+    {
+      "epoch": 0.05144152426214601,
+      "grad_norm": 1.251550555229187,
+      "learning_rate": 4.919443027766935e-05,
+      "loss": 0.637,
+      "step": 3875
+    },
+    {
+      "epoch": 0.05185305645624318,
+      "grad_norm": 1.258484959602356,
+      "learning_rate": 4.91731463569418e-05,
+      "loss": 0.629,
+      "step": 3906
+    },
+    {
+      "epoch": 0.052264588650340345,
+      "grad_norm": 1.3574656248092651,
+      "learning_rate": 4.915158964554312e-05,
+      "loss": 0.6262,
+      "step": 3937
+    },
+    {
+      "epoch": 0.05267612084443751,
+      "grad_norm": 1.327051043510437,
+      "learning_rate": 4.912976038673786e-05,
+      "loss": 0.629,
+      "step": 3968
+    },
+    {
+      "epoch": 0.05308765303853468,
+      "grad_norm": 1.3813196420669556,
+      "learning_rate": 4.9107658826866254e-05,
+      "loss": 0.6308,
+      "step": 3999
+    },
+    {
+      "epoch": 0.05349918523263185,
+      "grad_norm": 1.2064168453216553,
+      "learning_rate": 4.908528521534139e-05,
+      "loss": 0.6321,
+      "step": 4030
+    },
+    {
+      "epoch": 0.05391071742672902,
+      "grad_norm": 1.2771285772323608,
+      "learning_rate": 4.906263980464644e-05,
+      "loss": 0.622,
+      "step": 4061
+    },
+    {
+      "epoch": 0.054322249620826185,
+      "grad_norm": 1.1788181066513062,
+      "learning_rate": 4.903972285033178e-05,
+      "loss": 0.6268,
+      "step": 4092
+    },
+    {
+      "epoch": 0.05473378181492335,
+      "grad_norm": 1.3975950479507446,
+      "learning_rate": 4.901653461101213e-05,
+      "loss": 0.6251,
+      "step": 4123
+    },
+    {
+      "epoch": 0.05514531400902052,
+      "grad_norm": 1.2083688974380493,
+      "learning_rate": 4.8993075348363626e-05,
+      "loss": 0.6287,
+      "step": 4154
+    },
+    {
+      "epoch": 0.05555684620311769,
+      "grad_norm": 1.2795994281768799,
+      "learning_rate": 4.896934532712084e-05,
+      "loss": 0.6302,
+      "step": 4185
+    },
+    {
+      "epoch": 0.05596837839721486,
+      "grad_norm": 1.1056169271469116,
+      "learning_rate": 4.8945344815073846e-05,
+      "loss": 0.6256,
+      "step": 4216
+    },
+    {
+      "epoch": 0.056379910591312025,
+      "grad_norm": 1.415405511856079,
+      "learning_rate": 4.892107408306516e-05,
+      "loss": 0.6334,
+      "step": 4247
+    },
+    {
+      "epoch": 0.056791442785409194,
+      "grad_norm": 1.2858442068099976,
+      "learning_rate": 4.889653340498669e-05,
+      "loss": 0.6248,
+      "step": 4278
+    },
+    {
+      "epoch": 0.05720297497950636,
+      "grad_norm": 1.3914793729782104,
+      "learning_rate": 4.8871723057776664e-05,
+      "loss": 0.6229,
+      "step": 4309
+    },
+    {
+      "epoch": 0.05761450717360353,
+      "grad_norm": 1.1236270666122437,
+      "learning_rate": 4.8846643321416476e-05,
+      "loss": 0.6272,
+      "step": 4340
+    },
+    {
+      "epoch": 0.0580260393677007,
+      "grad_norm": 1.393020510673523,
+      "learning_rate": 4.882129447892753e-05,
+      "loss": 0.624,
+      "step": 4371
+    },
+    {
+      "epoch": 0.058437571561797866,
+      "grad_norm": 1.3223553895950317,
+      "learning_rate": 4.8795676816368076e-05,
+      "loss": 0.6267,
+      "step": 4402
+    },
+    {
+      "epoch": 0.058849103755895034,
+      "grad_norm": 1.2308725118637085,
+      "learning_rate": 4.876979062282995e-05,
+      "loss": 0.6255,
+      "step": 4433
+    },
+    {
+      "epoch": 0.0592606359499922,
+      "grad_norm": 1.3741412162780762,
+      "learning_rate": 4.8743636190435325e-05,
+      "loss": 0.6246,
+      "step": 4464
+    },
+    {
+      "epoch": 0.05967216814408937,
+      "grad_norm": 1.2487592697143555,
+      "learning_rate": 4.871721381433344e-05,
+      "loss": 0.6221,
+      "step": 4495
+    },
+    {
+      "epoch": 0.06008370033818654,
+      "grad_norm": 1.3239238262176514,
+      "learning_rate": 4.869052379269719e-05,
+      "loss": 0.621,
+      "step": 4526
+    },
+    {
+      "epoch": 0.060495232532283706,
+      "grad_norm": 1.1411608457565308,
+      "learning_rate": 4.866356642671985e-05,
+      "loss": 0.6211,
+      "step": 4557
+    },
+    {
+      "epoch": 0.060906764726380874,
+      "grad_norm": 1.1687365770339966,
+      "learning_rate": 4.8636342020611634e-05,
+      "loss": 0.6159,
+      "step": 4588
+    },
+    {
+      "epoch": 0.06131829692047804,
+      "grad_norm": 1.25679349899292,
+      "learning_rate": 4.860885088159626e-05,
+      "loss": 0.6254,
+      "step": 4619
+    },
+    {
+      "epoch": 0.06172982911457521,
+      "grad_norm": 1.2380110025405884,
+      "learning_rate": 4.858109331990751e-05,
+      "loss": 0.6305,
+      "step": 4650
+    },
+    {
+      "epoch": 0.06214136130867238,
+      "grad_norm": 1.3054466247558594,
+      "learning_rate": 4.855306964878567e-05,
+      "loss": 0.6202,
+      "step": 4681
+    },
+    {
+      "epoch": 0.06255289350276955,
+      "grad_norm": 1.3628545999526978,
+      "learning_rate": 4.8524780184474084e-05,
+      "loss": 0.6192,
+      "step": 4712
+    },
+    {
+      "epoch": 0.06296442569686671,
+      "grad_norm": 1.3576844930648804,
+      "learning_rate": 4.8496225246215496e-05,
+      "loss": 0.6213,
+      "step": 4743
+    },
+    {
+      "epoch": 0.06337595789096388,
+      "grad_norm": 1.3126753568649292,
+      "learning_rate": 4.8467405156248505e-05,
+      "loss": 0.6198,
+      "step": 4774
+    },
+    {
+      "epoch": 0.06378749008506104,
+      "grad_norm": 1.2266637086868286,
+      "learning_rate": 4.843832023980392e-05,
+      "loss": 0.6229,
+      "step": 4805
+    },
+    {
+      "epoch": 0.06419902227915822,
+      "grad_norm": 1.4964330196380615,
+      "learning_rate": 4.840897082510106e-05,
+      "loss": 0.6264,
+      "step": 4836
+    },
+    {
+      "epoch": 0.06461055447325538,
+      "grad_norm": 1.1138426065444946,
+      "learning_rate": 4.8379357243344084e-05,
+      "loss": 0.6258,
+      "step": 4867
+    },
+    {
+      "epoch": 0.06502208666735255,
+      "grad_norm": 1.2055357694625854,
+      "learning_rate": 4.8349479828718236e-05,
+      "loss": 0.6263,
+      "step": 4898
+    },
+    {
+      "epoch": 0.06543361886144972,
+      "grad_norm": 1.2684381008148193,
+      "learning_rate": 4.8319338918386075e-05,
+      "loss": 0.6212,
+      "step": 4929
+    },
+    {
+      "epoch": 0.06584515105554689,
+      "grad_norm": 1.2323859930038452,
+      "learning_rate": 4.828893485248369e-05,
+      "loss": 0.6253,
+      "step": 4960
+    },
+    {
+      "epoch": 0.06625668324964405,
+      "grad_norm": 1.091818928718567,
+      "learning_rate": 4.825826797411682e-05,
+      "loss": 0.6127,
+      "step": 4991
+    },
+    {
+      "epoch": 0.06666821544374123,
+      "grad_norm": 1.2208534479141235,
+      "learning_rate": 4.822733862935702e-05,
+      "loss": 0.6179,
+      "step": 5022
+    },
+    {
+      "epoch": 0.06707974763783839,
+      "grad_norm": 1.3426082134246826,
+      "learning_rate": 4.819614716723775e-05,
+      "loss": 0.6246,
+      "step": 5053
+    },
+    {
+      "epoch": 0.06749127983193556,
+      "grad_norm": 1.2961252927780151,
+      "learning_rate": 4.8164693939750425e-05,
+      "loss": 0.6149,
+      "step": 5084
+    },
+    {
+      "epoch": 0.06790281202603272,
+      "grad_norm": 1.2990977764129639,
+      "learning_rate": 4.813297930184042e-05,
+      "loss": 0.6303,
+      "step": 5115
+    },
+    {
+      "epoch": 0.0683143442201299,
+      "grad_norm": 1.2724366188049316,
+      "learning_rate": 4.810100361140314e-05,
+      "loss": 0.6234,
+      "step": 5146
+    },
+    {
+      "epoch": 0.06872587641422706,
+      "grad_norm": 1.2031605243682861,
+      "learning_rate": 4.8068767229279885e-05,
+      "loss": 0.6212,
+      "step": 5177
+    },
+    {
+      "epoch": 0.06913740860832424,
+      "grad_norm": 1.2580938339233398,
+      "learning_rate": 4.8036270519253854e-05,
+      "loss": 0.6144,
+      "step": 5208
+    },
+    {
+      "epoch": 0.0695489408024214,
+      "grad_norm": 1.203706979751587,
+      "learning_rate": 4.8003513848046e-05,
+      "loss": 0.6259,
+      "step": 5239
+    },
+    {
+      "epoch": 0.06996047299651857,
+      "grad_norm": 1.421762466430664,
+      "learning_rate": 4.79704975853109e-05,
+      "loss": 0.6221,
+      "step": 5270
+    },
+    {
+      "epoch": 0.07037200519061573,
+      "grad_norm": 1.326038122177124,
+      "learning_rate": 4.793722210363262e-05,
+      "loss": 0.6152,
+      "step": 5301
+    },
+    {
+      "epoch": 0.07078353738471291,
+      "grad_norm": 1.361939787864685,
+      "learning_rate": 4.7903687778520414e-05,
+      "loss": 0.6206,
+      "step": 5332
+    },
+    {
+      "epoch": 0.07119506957881007,
+      "grad_norm": 1.2153042554855347,
+      "learning_rate": 4.7869894988404593e-05,
+      "loss": 0.6209,
+      "step": 5363
+    },
+    {
+      "epoch": 0.07160660177290724,
+      "grad_norm": 1.3737229108810425,
+      "learning_rate": 4.783584411463221e-05,
+      "loss": 0.6121,
+      "step": 5394
+    },
+    {
+      "epoch": 0.0720181339670044,
+      "grad_norm": 1.1781859397888184,
+      "learning_rate": 4.780153554146274e-05,
+      "loss": 0.6183,
+      "step": 5425
+    },
+    {
+      "epoch": 0.07242966616110158,
+      "grad_norm": 1.069472312927246,
+      "learning_rate": 4.7766969656063766e-05,
+      "loss": 0.6117,
+      "step": 5456
+    },
+    {
+      "epoch": 0.07284119835519874,
+      "grad_norm": 1.380937099456787,
+      "learning_rate": 4.773214684850662e-05,
+      "loss": 0.6175,
+      "step": 5487
+    },
+    {
+      "epoch": 0.07325273054929592,
+      "grad_norm": 1.1551166772842407,
+      "learning_rate": 4.769706751176193e-05,
+      "loss": 0.6164,
+      "step": 5518
+    },
+    {
+      "epoch": 0.07366426274339308,
+      "grad_norm": 1.2816351652145386,
+      "learning_rate": 4.7661732041695264e-05,
+      "loss": 0.6168,
+      "step": 5549
+    },
+    {
+      "epoch": 0.07407579493749025,
+      "grad_norm": 1.0833333730697632,
+      "learning_rate": 4.762614083706258e-05,
+      "loss": 0.6159,
+      "step": 5580
+    },
+    {
+      "epoch": 0.07448732713158741,
+      "grad_norm": 1.2533819675445557,
+      "learning_rate": 4.759029429950581e-05,
+      "loss": 0.6141,
+      "step": 5611
+    },
+    {
+      "epoch": 0.07489885932568459,
+      "grad_norm": 1.1354937553405762,
+      "learning_rate": 4.7554192833548235e-05,
+      "loss": 0.612,
+      "step": 5642
+    },
+    {
+      "epoch": 0.07531039151978175,
+      "grad_norm": 1.2579649686813354,
+      "learning_rate": 4.751783684659e-05,
+      "loss": 0.6147,
+      "step": 5673
+    },
+    {
+      "epoch": 0.07572192371387892,
+      "grad_norm": 1.2830709218978882,
+      "learning_rate": 4.748122674890348e-05,
+      "loss": 0.62,
+      "step": 5704
+    },
+    {
+      "epoch": 0.07613345590797609,
+      "grad_norm": 1.2012654542922974,
+      "learning_rate": 4.7444362953628654e-05,
+      "loss": 0.6199,
+      "step": 5735
+    },
+    {
+      "epoch": 0.07654498810207326,
+      "grad_norm": 1.2926653623580933,
+      "learning_rate": 4.7407245876768424e-05,
+      "loss": 0.6104,
+      "step": 5766
+    },
+    {
+      "epoch": 0.07695652029617042,
+      "grad_norm": 1.2043049335479736,
+      "learning_rate": 4.736987593718397e-05,
+      "loss": 0.6261,
+      "step": 5797
+    },
+    {
+      "epoch": 0.0773680524902676,
+      "grad_norm": 1.1188119649887085,
+      "learning_rate": 4.733225355658999e-05,
+      "loss": 0.6056,
+      "step": 5828
+    },
+    {
+      "epoch": 0.07777958468436476,
+      "grad_norm": 1.1217443943023682,
+      "learning_rate": 4.7294379159549926e-05,
+      "loss": 0.6124,
+      "step": 5859
+    },
+    {
+      "epoch": 0.07819111687846193,
+      "grad_norm": 1.1623468399047852,
+      "learning_rate": 4.725625317347119e-05,
+      "loss": 0.6191,
+      "step": 5890
+    },
+    {
+      "epoch": 0.0786026490725591,
+      "grad_norm": 1.1147449016571045,
+      "learning_rate": 4.7217876028600374e-05,
+      "loss": 0.6091,
+      "step": 5921
+    },
+    {
+      "epoch": 0.07901418126665627,
+      "grad_norm": 1.0523818731307983,
+      "learning_rate": 4.717924815801832e-05,
+      "loss": 0.6151,
+      "step": 5952
+    },
+    {
+      "epoch": 0.07942571346075343,
+      "grad_norm": 1.3112642765045166,
+      "learning_rate": 4.714036999763532e-05,
+      "loss": 0.6106,
+      "step": 5983
+    },
+    {
+      "epoch": 0.0798372456548506,
+      "grad_norm": 1.1637545824050903,
+      "learning_rate": 4.7101241986186116e-05,
+      "loss": 0.6131,
+      "step": 6014
+    },
+    {
+      "epoch": 0.08024877784894777,
+      "grad_norm": 1.258862018585205,
+      "learning_rate": 4.7061864565225e-05,
+      "loss": 0.6119,
+      "step": 6045
+    },
+    {
+      "epoch": 0.08066031004304494,
+      "grad_norm": 1.0981336832046509,
+      "learning_rate": 4.702223817912081e-05,
+      "loss": 0.6149,
+      "step": 6076
+    },
+    {
+      "epoch": 0.0810718422371421,
+      "grad_norm": 1.1576476097106934,
+      "learning_rate": 4.698236327505195e-05,
+      "loss": 0.6076,
+      "step": 6107
+    },
+    {
+      "epoch": 0.08148337443123928,
+      "grad_norm": 1.1076856851577759,
+      "learning_rate": 4.694224030300127e-05,
+      "loss": 0.6105,
+      "step": 6138
+    },
+    {
+      "epoch": 0.08189490662533644,
+      "grad_norm": 14.014957427978516,
+      "learning_rate": 4.690186971575107e-05,
+      "loss": 0.6189,
+      "step": 6169
+    },
+    {
+      "epoch": 0.08230643881943361,
+      "grad_norm": 1.1413066387176514,
+      "learning_rate": 4.6861251968877916e-05,
+      "loss": 0.6038,
+      "step": 6200
+    },
+    {
+      "epoch": 0.08271797101353077,
+      "grad_norm": 1.2316724061965942,
+      "learning_rate": 4.68203875207476e-05,
+      "loss": 0.6125,
+      "step": 6231
+    },
+    {
+      "epoch": 0.08312950320762795,
+      "grad_norm": 1.0919370651245117,
+      "learning_rate": 4.677927683250983e-05,
+      "loss": 0.6081,
+      "step": 6262
+    },
+    {
+      "epoch": 0.08354103540172511,
+      "grad_norm": 1.139345407485962,
+      "learning_rate": 4.6737920368093156e-05,
+      "loss": 0.6078,
+      "step": 6293
+    },
+    {
+      "epoch": 0.08395256759582229,
+      "grad_norm": 1.1303651332855225,
+      "learning_rate": 4.669631859419965e-05,
+      "loss": 0.6071,
+      "step": 6324
+    },
+    {
+      "epoch": 0.08436409978991945,
+      "grad_norm": 1.2546321153640747,
+      "learning_rate": 4.6654471980299676e-05,
+      "loss": 0.6092,
+      "step": 6355
+    },
+    {
+      "epoch": 0.08477563198401662,
+      "grad_norm": 1.195173740386963,
+      "learning_rate": 4.661238099862658e-05,
+      "loss": 0.5998,
+      "step": 6386
+    },
+    {
+      "epoch": 0.08518716417811378,
+      "grad_norm": 1.2116323709487915,
+      "learning_rate": 4.657004612417138e-05,
+      "loss": 0.6099,
+      "step": 6417
+    },
+    {
+      "epoch": 0.08559869637221096,
+      "grad_norm": 1.131951928138733,
+      "learning_rate": 4.6527467834677374e-05,
+      "loss": 0.6105,
+      "step": 6448
+    },
+    {
+      "epoch": 0.08601022856630812,
+      "grad_norm": 1.2176560163497925,
+      "learning_rate": 4.648464661063478e-05,
+      "loss": 0.6045,
+      "step": 6479
+    },
+    {
+      "epoch": 0.0864217607604053,
+      "grad_norm": 1.15361750125885,
+      "learning_rate": 4.6441582935275264e-05,
+      "loss": 0.6067,
+      "step": 6510
+    },
+    {
+      "epoch": 0.08683329295450246,
+      "grad_norm": 1.1422772407531738,
+      "learning_rate": 4.6398277294566586e-05,
+      "loss": 0.605,
+      "step": 6541
+    },
+    {
+      "epoch": 0.08724482514859963,
+      "grad_norm": 1.250319004058838,
+      "learning_rate": 4.6354730177207e-05,
+      "loss": 0.6069,
+      "step": 6572
+    },
+    {
+      "epoch": 0.08765635734269679,
+      "grad_norm": 1.0947929620742798,
+      "learning_rate": 4.6310942074619787e-05,
+      "loss": 0.6099,
+      "step": 6603
+    },
+    {
+      "epoch": 0.08806788953679397,
+      "grad_norm": 1.2815848588943481,
+      "learning_rate": 4.626691348094777e-05,
+      "loss": 0.6132,
+      "step": 6634
+    },
+    {
+      "epoch": 0.08847942173089113,
+      "grad_norm": 1.196083664894104,
+      "learning_rate": 4.622264489304762e-05,
+      "loss": 0.604,
+      "step": 6665
+    },
+    {
+      "epoch": 0.0888909539249883,
+      "grad_norm": 1.1473642587661743,
+      "learning_rate": 4.617813681048434e-05,
+      "loss": 0.6081,
+      "step": 6696
+    },
+    {
+      "epoch": 0.08930248611908546,
+      "grad_norm": 1.182445764541626,
+      "learning_rate": 4.61333897355256e-05,
+      "loss": 0.6108,
+      "step": 6727
+    },
+    {
+      "epoch": 0.08971401831318264,
+      "grad_norm": 1.2859700918197632,
+      "learning_rate": 4.608840417313604e-05,
+      "loss": 0.6074,
+      "step": 6758
+    },
+    {
+      "epoch": 0.0901255505072798,
+      "grad_norm": 1.1134952306747437,
+      "learning_rate": 4.6043180630971646e-05,
+      "loss": 0.6139,
+      "step": 6789
+    },
+    {
+      "epoch": 0.09053708270137698,
+      "grad_norm": 1.2447940111160278,
+      "learning_rate": 4.599771961937391e-05,
+      "loss": 0.6049,
+      "step": 6820
+    },
+    {
+      "epoch": 0.09094861489547414,
+      "grad_norm": 1.042465329170227,
+      "learning_rate": 4.5952021651364204e-05,
+      "loss": 0.6113,
+      "step": 6851
+    },
+    {
+      "epoch": 0.09136014708957131,
+      "grad_norm": 1.2242035865783691,
+      "learning_rate": 4.590608724263786e-05,
+      "loss": 0.6077,
+      "step": 6882
+    },
+    {
+      "epoch": 0.09177167928366847,
+      "grad_norm": 1.2531142234802246,
+      "learning_rate": 4.585991691155845e-05,
+      "loss": 0.6081,
+      "step": 6913
+    },
+    {
+      "epoch": 0.09218321147776565,
+      "grad_norm": 1.194645881652832,
+      "learning_rate": 4.581351117915188e-05,
+      "loss": 0.6158,
+      "step": 6944
+    },
+    {
+      "epoch": 0.09259474367186281,
+      "grad_norm": 1.1052149534225464,
+      "learning_rate": 4.5766870569100534e-05,
+      "loss": 0.6079,
+      "step": 6975
+    },
+    {
+      "epoch": 0.09300627586595998,
+      "grad_norm": 1.1275289058685303,
+      "learning_rate": 4.571999560773736e-05,
+      "loss": 0.6107,
+      "step": 7006
+    },
+    {
+      "epoch": 0.09341780806005714,
+      "grad_norm": 1.251237154006958,
+      "learning_rate": 4.5672886824039915e-05,
+      "loss": 0.6028,
+      "step": 7037
+    },
+    {
+      "epoch": 0.09382934025415432,
+      "grad_norm": 1.1206951141357422,
+      "learning_rate": 4.5625544749624435e-05,
+      "loss": 0.6035,
+      "step": 7068
+    },
+    {
+      "epoch": 0.09424087244825148,
+      "grad_norm": 1.2242387533187866,
+      "learning_rate": 4.5577969918739794e-05,
+      "loss": 0.5959,
+      "step": 7099
+    },
+    {
+      "epoch": 0.09465240464234866,
+      "grad_norm": 1.0749527215957642,
+      "learning_rate": 4.5530162868261486e-05,
+      "loss": 0.6005,
+      "step": 7130
+    },
+    {
+      "epoch": 0.09506393683644582,
+      "grad_norm": 1.183052659034729,
+      "learning_rate": 4.548212413768558e-05,
+      "loss": 0.6102,
+      "step": 7161
+    },
+    {
+      "epoch": 0.09547546903054299,
+      "grad_norm": 1.1045013666152954,
+      "learning_rate": 4.543385426912261e-05,
+      "loss": 0.6056,
+      "step": 7192
+    },
+    {
+      "epoch": 0.09588700122464015,
+      "grad_norm": 1.2442799806594849,
+      "learning_rate": 4.53853538072915e-05,
+      "loss": 0.6024,
+      "step": 7223
+    },
+    {
+      "epoch": 0.09629853341873733,
+      "grad_norm": 1.2083353996276855,
+      "learning_rate": 4.533662329951336e-05,
+      "loss": 0.6062,
+      "step": 7254
+    },
+    {
+      "epoch": 0.09671006561283449,
+      "grad_norm": 1.1391205787658691,
+      "learning_rate": 4.528766329570536e-05,
+      "loss": 0.5981,
+      "step": 7285
+    },
+    {
+      "epoch": 0.09712159780693166,
+      "grad_norm": 1.1213765144348145,
+      "learning_rate": 4.523847434837447e-05,
+      "loss": 0.5964,
+      "step": 7316
+    },
+    {
+      "epoch": 0.09753313000102883,
+      "grad_norm": 1.1574114561080933,
+      "learning_rate": 4.518905701261128e-05,
+      "loss": 0.6035,
+      "step": 7347
+    },
+    {
+      "epoch": 0.097944662195126,
+      "grad_norm": 1.0101516246795654,
+      "learning_rate": 4.5139411846083715e-05,
+      "loss": 0.6036,
+      "step": 7378
+    },
+    {
+      "epoch": 0.09835619438922316,
+      "grad_norm": 1.1439381837844849,
+      "learning_rate": 4.508953940903073e-05,
+      "loss": 0.6049,
+      "step": 7409
+    },
+    {
+      "epoch": 0.09876772658332034,
+      "grad_norm": 1.0668189525604248,
+      "learning_rate": 4.5039440264255994e-05,
+      "loss": 0.6035,
+      "step": 7440
+    },
+    {
+      "epoch": 0.0991792587774175,
+      "grad_norm": 1.2038415670394897,
+      "learning_rate": 4.498911497712155e-05,
+      "loss": 0.6062,
+      "step": 7471
+    },
+    {
+      "epoch": 0.09959079097151467,
+      "grad_norm": 1.1965882778167725,
+      "learning_rate": 4.493856411554142e-05,
+      "loss": 0.6041,
+      "step": 7502
+    },
+    {
+      "epoch": 0.10000232316561183,
+      "grad_norm": 1.5885149240493774,
+      "learning_rate": 4.4887788249975206e-05,
+      "loss": 0.6129,
+      "step": 7533
+    },
+    {
+      "epoch": 0.10041385535970901,
+      "grad_norm": 1.058237910270691,
+      "learning_rate": 4.4836787953421656e-05,
+      "loss": 0.5957,
+      "step": 7564
+    },
+    {
+      "epoch": 0.10082538755380617,
+      "grad_norm": 1.2043278217315674,
+      "learning_rate": 4.478556380141218e-05,
+      "loss": 0.5943,
+      "step": 7595
+    },
+    {
+      "epoch": 0.10123691974790335,
+      "grad_norm": 1.1602888107299805,
+      "learning_rate": 4.4734116372004375e-05,
+      "loss": 0.5971,
+      "step": 7626
+    },
+    {
+      "epoch": 0.1016484519420005,
+      "grad_norm": 1.1536785364151,
+      "learning_rate": 4.4682446245775477e-05,
+      "loss": 0.604,
+      "step": 7657
+    },
+    {
+      "epoch": 0.10205998413609768,
+      "grad_norm": 1.1859934329986572,
+      "learning_rate": 4.463055400581586e-05,
+      "loss": 0.6043,
+      "step": 7688
+    },
+    {
+      "epoch": 0.10247151633019484,
+      "grad_norm": 1.0551954507827759,
+      "learning_rate": 4.4578440237722374e-05,
+      "loss": 0.6008,
+      "step": 7719
+    },
+    {
+      "epoch": 0.10288304852429202,
+      "grad_norm": 1.015857458114624,
+      "learning_rate": 4.452610552959183e-05,
+      "loss": 0.5961,
+      "step": 7750
+    },
+    {
+      "epoch": 0.10329458071838918,
+      "grad_norm": 1.0759170055389404,
+      "learning_rate": 4.447355047201428e-05,
+      "loss": 0.6025,
+      "step": 7781
+    },
+    {
+      "epoch": 0.10370611291248635,
+      "grad_norm": 1.025030255317688,
+      "learning_rate": 4.4420775658066414e-05,
+      "loss": 0.602,
+      "step": 7812
+    },
+    {
+      "epoch": 0.10411764510658351,
+      "grad_norm": 1.0884742736816406,
+      "learning_rate": 4.436778168330484e-05,
+      "loss": 0.5989,
+      "step": 7843
+    },
+    {
+      "epoch": 0.10452917730068069,
+      "grad_norm": 1.1207304000854492,
+      "learning_rate": 4.4314569145759353e-05,
+      "loss": 0.6016,
+      "step": 7874
+    },
+    {
+      "epoch": 0.10494070949477785,
+      "grad_norm": 0.9768222570419312,
+      "learning_rate": 4.42611386459262e-05,
+      "loss": 0.605,
+      "step": 7905
+    },
+    {
+      "epoch": 0.10535224168887503,
+      "grad_norm": 1.0720235109329224,
+      "learning_rate": 4.420749078676133e-05,
+      "loss": 0.5971,
+      "step": 7936
+    },
+    {
+      "epoch": 0.10576377388297219,
+      "grad_norm": 1.1767033338546753,
+      "learning_rate": 4.4153626173673516e-05,
+      "loss": 0.5877,
+      "step": 7967
+    },
+    {
+      "epoch": 0.10617530607706936,
+      "grad_norm": 1.3390216827392578,
+      "learning_rate": 4.409954541451762e-05,
+      "loss": 0.5973,
+      "step": 7998
+    },
+    {
+      "epoch": 0.10658683827116652,
+      "grad_norm": 1.7948216199874878,
+      "learning_rate": 4.404524911958764e-05,
+      "loss": 0.5922,
+      "step": 8029
+    },
+    {
+      "epoch": 0.1069983704652637,
+      "grad_norm": 1.177847981452942,
+      "learning_rate": 4.399073790160989e-05,
+      "loss": 0.6002,
+      "step": 8060
+    },
+    {
+      "epoch": 0.10740990265936086,
+      "grad_norm": 1.1014128923416138,
+      "learning_rate": 4.393601237573607e-05,
+      "loss": 0.5945,
+      "step": 8091
+    },
+    {
+      "epoch": 0.10782143485345803,
+      "grad_norm": 1.067475438117981,
+      "learning_rate": 4.388107315953628e-05,
+      "loss": 0.5996,
+      "step": 8122
+    },
+    {
+      "epoch": 0.1082329670475552,
+      "grad_norm": 1.1241830587387085,
+      "learning_rate": 4.382592087299212e-05,
+      "loss": 0.5973,
+      "step": 8153
+    },
+    {
+      "epoch": 0.10864449924165237,
+      "grad_norm": 1.072426438331604,
+      "learning_rate": 4.377055613848964e-05,
+      "loss": 0.601,
+      "step": 8184
+    },
+    {
+      "epoch": 0.10905603143574953,
+      "grad_norm": 1.0138245820999146,
+      "learning_rate": 4.3714979580812355e-05,
+      "loss": 0.6034,
+      "step": 8215
+    },
+    {
+      "epoch": 0.1094675636298467,
+      "grad_norm": 0.9993209838867188,
+      "learning_rate": 4.365919182713416e-05,
+      "loss": 0.6024,
+      "step": 8246
+    },
+    {
+      "epoch": 0.10987909582394387,
+      "grad_norm": 1.157468318939209,
+      "learning_rate": 4.360319350701226e-05,
+      "loss": 0.5987,
+      "step": 8277
+    },
+    {
+      "epoch": 0.11029062801804104,
+      "grad_norm": 1.2073496580123901,
+      "learning_rate": 4.3546985252380115e-05,
+      "loss": 0.6023,
+      "step": 8308
+    },
+    {
+      "epoch": 0.1107021602121382,
+      "grad_norm": 1.0800622701644897,
+      "learning_rate": 4.349056769754021e-05,
+      "loss": 0.5993,
+      "step": 8339
+    },
+    {
+      "epoch": 0.11111369240623538,
+      "grad_norm": 1.1365702152252197,
+      "learning_rate": 4.3433941479156994e-05,
+      "loss": 0.595,
+      "step": 8370
+    },
+    {
+      "epoch": 0.11152522460033254,
+      "grad_norm": 1.1367030143737793,
+      "learning_rate": 4.3377107236249647e-05,
+      "loss": 0.5921,
+      "step": 8401
+    },
+    {
+      "epoch": 0.11193675679442971,
+      "grad_norm": 1.0323926210403442,
+      "learning_rate": 4.332006561018488e-05,
+      "loss": 0.5883,
+      "step": 8432
+    },
+    {
+      "epoch": 0.11234828898852688,
+      "grad_norm": 1.2352523803710938,
+      "learning_rate": 4.3262817244669683e-05,
+      "loss": 0.5957,
+      "step": 8463
+    },
+    {
+      "epoch": 0.11275982118262405,
+      "grad_norm": 1.1763888597488403,
+      "learning_rate": 4.3205362785744083e-05,
+      "loss": 0.5941,
+      "step": 8494
+    },
+    {
+      "epoch": 0.11317135337672121,
+      "grad_norm": 1.1331796646118164,
+      "learning_rate": 4.314770288177384e-05,
+      "loss": 0.5993,
+      "step": 8525
+    },
+    {
+      "epoch": 0.11358288557081839,
+      "grad_norm": 1.146270990371704,
+      "learning_rate": 4.308983818344313e-05,
+      "loss": 0.6041,
+      "step": 8556
+    },
+    {
+      "epoch": 0.11399441776491555,
+      "grad_norm": 1.0433791875839233,
+      "learning_rate": 4.3031769343747206e-05,
+      "loss": 0.5918,
+      "step": 8587
+    },
+    {
+      "epoch": 0.11440594995901272,
+      "grad_norm": 1.2013510465621948,
+      "learning_rate": 4.297349701798505e-05,
+      "loss": 0.6031,
+      "step": 8618
+    },
+    {
+      "epoch": 0.11481748215310988,
+      "grad_norm": 1.1263506412506104,
+      "learning_rate": 4.2915021863751916e-05,
+      "loss": 0.597,
+      "step": 8649
+    },
+    {
+      "epoch": 0.11522901434720706,
+      "grad_norm": 1.1360443830490112,
+      "learning_rate": 4.285634454093198e-05,
+      "loss": 0.5909,
+      "step": 8680
+    },
+    {
+      "epoch": 0.11564054654130422,
+      "grad_norm": 1.0095175504684448,
+      "learning_rate": 4.279746571169086e-05,
+      "loss": 0.59,
+      "step": 8711
+    },
+    {
+      "epoch": 0.1160520787354014,
+      "grad_norm": 1.0894793272018433,
+      "learning_rate": 4.2738386040468136e-05,
+      "loss": 0.5894,
+      "step": 8742
+    },
+    {
+      "epoch": 0.11646361092949856,
+      "grad_norm": 1.1603643894195557,
+      "learning_rate": 4.2679106193969866e-05,
+      "loss": 0.5924,
+      "step": 8773
+    },
+    {
+      "epoch": 0.11687514312359573,
+      "grad_norm": 1.177464485168457,
+      "learning_rate": 4.261962684116106e-05,
+      "loss": 0.5946,
+      "step": 8804
+    },
+    {
+      "epoch": 0.11728667531769289,
+      "grad_norm": 0.9632905125617981,
+      "learning_rate": 4.2559948653258145e-05,
+      "loss": 0.5944,
+      "step": 8835
+    },
+    {
+      "epoch": 0.11769820751179007,
+      "grad_norm": 1.0413568019866943,
+      "learning_rate": 4.250007230372134e-05,
+      "loss": 0.5844,
+      "step": 8866
+    },
+    {
+      "epoch": 0.11810973970588723,
+      "grad_norm": 1.0879931449890137,
+      "learning_rate": 4.2439998468247126e-05,
+      "loss": 0.588,
+      "step": 8897
+    },
+    {
+      "epoch": 0.1185212718999844,
+      "grad_norm": 1.0681732892990112,
+      "learning_rate": 4.2379727824760566e-05,
+      "loss": 0.5832,
+      "step": 8928
+    },
+    {
+      "epoch": 0.11893280409408157,
+      "grad_norm": 1.0609339475631714,
+      "learning_rate": 4.231926105340768e-05,
+      "loss": 0.5886,
+      "step": 8959
+    },
+    {
+      "epoch": 0.11934433628817874,
+      "grad_norm": 1.4936331510543823,
+      "learning_rate": 4.225859883654776e-05,
+      "loss": 0.5868,
+      "step": 8990
+    },
+    {
+      "epoch": 0.1197558684822759,
+      "grad_norm": 1.1860368251800537,
+      "learning_rate": 4.219774185874569e-05,
+      "loss": 0.5904,
+      "step": 9021
+    },
+    {
+      "epoch": 0.12016740067637308,
+      "grad_norm": 1.1657029390335083,
+      "learning_rate": 4.213669080676418e-05,
+      "loss": 0.5872,
+      "step": 9052
+    },
+    {
+      "epoch": 0.12057893287047024,
+      "grad_norm": 1.0721909999847412,
+      "learning_rate": 4.2075446369556056e-05,
+      "loss": 0.5977,
+      "step": 9083
+    },
+    {
+      "epoch": 0.12099046506456741,
+      "grad_norm": 1.1832354068756104,
+      "learning_rate": 4.201400923825648e-05,
+      "loss": 0.5902,
+      "step": 9114
+    },
+    {
+      "epoch": 0.12140199725866457,
+      "grad_norm": 1.1306614875793457,
+      "learning_rate": 4.195238010617511e-05,
+      "loss": 0.5931,
+      "step": 9145
+    },
+    {
+      "epoch": 0.12181352945276175,
+      "grad_norm": 0.9744802713394165,
+      "learning_rate": 4.1890559668788344e-05,
+      "loss": 0.6,
+      "step": 9176
+    },
+    {
+      "epoch": 0.12222506164685891,
+      "grad_norm": 1.0286784172058105,
+      "learning_rate": 4.1828548623731405e-05,
+      "loss": 0.5913,
+      "step": 9207
+    },
+    {
+      "epoch": 0.12263659384095608,
+      "grad_norm": 1.0559006929397583,
+      "learning_rate": 4.1766347670790506e-05,
+      "loss": 0.5999,
+      "step": 9238
+    },
+    {
+      "epoch": 0.12304812603505325,
+      "grad_norm": 1.0805801153182983,
+      "learning_rate": 4.170395751189495e-05,
+      "loss": 0.5929,
+      "step": 9269
+    },
+    {
+      "epoch": 0.12345965822915042,
+      "grad_norm": 1.088597059249878,
+      "learning_rate": 4.164137885110921e-05,
+      "loss": 0.5873,
+      "step": 9300
+    },
+    {
+      "epoch": 0.12387119042324758,
+      "grad_norm": 1.028403639793396,
+      "learning_rate": 4.157861239462495e-05,
+      "loss": 0.5939,
+      "step": 9331
+    },
+    {
+      "epoch": 0.12428272261734476,
+      "grad_norm": 0.9916087985038757,
+      "learning_rate": 4.1515658850753114e-05,
+      "loss": 0.5932,
+      "step": 9362
+    },
+    {
+      "epoch": 0.12469425481144192,
+      "grad_norm": 1.0268062353134155,
+      "learning_rate": 4.145251892991588e-05,
+      "loss": 0.5945,
+      "step": 9393
+    },
+    {
+      "epoch": 0.1251057870055391,
+      "grad_norm": 1.046953797340393,
+      "learning_rate": 4.138919334463868e-05,
+      "loss": 0.5923,
+      "step": 9424
+    },
+    {
+      "epoch": 0.12551731919963627,
+      "grad_norm": 1.1010463237762451,
+      "learning_rate": 4.1325682809542124e-05,
+      "loss": 0.5898,
+      "step": 9455
+    },
+    {
+      "epoch": 0.12592885139373342,
+      "grad_norm": 1.1851146221160889,
+      "learning_rate": 4.126198804133398e-05,
+      "loss": 0.591,
+      "step": 9486
+    },
+    {
+      "epoch": 0.1263403835878306,
+      "grad_norm": 1.106885313987732,
+      "learning_rate": 4.1198109758801055e-05,
+      "loss": 0.5893,
+      "step": 9517
+    },
+    {
+      "epoch": 0.12675191578192777,
+      "grad_norm": 1.0527287721633911,
+      "learning_rate": 4.113404868280107e-05,
+      "loss": 0.5937,
+      "step": 9548
+    },
+    {
+      "epoch": 0.12716344797602494,
+      "grad_norm": 0.9986059069633484,
+      "learning_rate": 4.106980553625457e-05,
+      "loss": 0.5961,
+      "step": 9579
+    },
+    {
+      "epoch": 0.1275749801701221,
+      "grad_norm": 1.025026798248291,
+      "learning_rate": 4.100538104413674e-05,
+      "loss": 0.5853,
+      "step": 9610
+    },
+    {
+      "epoch": 0.12798651236421926,
+      "grad_norm": 1.062345027923584,
+      "learning_rate": 4.09407759334692e-05,
+      "loss": 0.5916,
+      "step": 9641
+    },
+    {
+      "epoch": 0.12839804455831644,
+      "grad_norm": 1.0635534524917603,
+      "learning_rate": 4.087599093331186e-05,
+      "loss": 0.5919,
+      "step": 9672
+    },
+    {
+      "epoch": 0.1288095767524136,
+      "grad_norm": 1.1224849224090576,
+      "learning_rate": 4.081102677475462e-05,
+      "loss": 0.5948,
+      "step": 9703
+    },
+    {
+      "epoch": 0.12922110894651076,
+      "grad_norm": 1.113202452659607,
+      "learning_rate": 4.0745884190909194e-05,
+      "loss": 0.5807,
+      "step": 9734
+    },
+    {
+      "epoch": 0.12963264114060793,
+      "grad_norm": 1.0213284492492676,
+      "learning_rate": 4.0680563916900796e-05,
+      "loss": 0.5813,
+      "step": 9765
+    },
+    {
+      "epoch": 0.1300441733347051,
+      "grad_norm": 1.1083142757415771,
+      "learning_rate": 4.0615066689859815e-05,
+      "loss": 0.5856,
+      "step": 9796
+    },
+    {
+      "epoch": 0.13045570552880228,
+      "grad_norm": 1.1317702531814575,
+      "learning_rate": 4.0549393248913584e-05,
+      "loss": 0.586,
+      "step": 9827
+    },
+    {
+      "epoch": 0.13086723772289943,
+      "grad_norm": 0.9876610040664673,
+      "learning_rate": 4.048354433517794e-05,
+      "loss": 0.5921,
+      "step": 9858
+    },
+    {
+      "epoch": 0.1312787699169966,
+      "grad_norm": 1.085672378540039,
+      "learning_rate": 4.0417520691748916e-05,
+      "loss": 0.5791,
+      "step": 9889
+    },
+    {
+      "epoch": 0.13169030211109378,
+      "grad_norm": 1.014452338218689,
+      "learning_rate": 4.035132306369438e-05,
+      "loss": 0.5921,
+      "step": 9920
+    },
+    {
+      "epoch": 0.13210183430519096,
+      "grad_norm": 1.059309720993042,
+      "learning_rate": 4.028495219804555e-05,
+      "loss": 0.5834,
+      "step": 9951
+    },
+    {
+      "epoch": 0.1325133664992881,
+      "grad_norm": 1.1058080196380615,
+      "learning_rate": 4.021840884378864e-05,
+      "loss": 0.5826,
+      "step": 9982
+    },
+    {
+      "epoch": 0.13292489869338528,
+      "grad_norm": 1.0567057132720947,
+      "learning_rate": 4.015169375185633e-05,
+      "loss": 0.5924,
+      "step": 10013
+    },
+    {
+      "epoch": 0.13333643088748245,
+      "grad_norm": 1.0371674299240112,
+      "learning_rate": 4.0084807675119396e-05,
+      "loss": 0.589,
+      "step": 10044
+    },
+    {
+      "epoch": 0.13374796308157963,
+      "grad_norm": 0.909173309803009,
+      "learning_rate": 4.0017751368378106e-05,
+      "loss": 0.5978,
+      "step": 10075
+    },
+    {
+      "epoch": 0.13415949527567678,
+      "grad_norm": 1.0078933238983154,
+      "learning_rate": 3.995052558835377e-05,
+      "loss": 0.5814,
+      "step": 10106
+    },
+    {
+      "epoch": 0.13457102746977395,
+      "grad_norm": 0.9631708264350891,
+      "learning_rate": 3.988313109368017e-05,
+      "loss": 0.587,
+      "step": 10137
+    },
+    {
+      "epoch": 0.13498255966387113,
+      "grad_norm": 1.1884409189224243,
+      "learning_rate": 3.981556864489504e-05,
+      "loss": 0.5906,
+      "step": 10168
+    },
+    {
+      "epoch": 0.1353940918579683,
+      "grad_norm": 0.9554224014282227,
+      "learning_rate": 3.974783900443142e-05,
+      "loss": 0.5888,
+      "step": 10199
+    },
+    {
+      "epoch": 0.13580562405206545,
+      "grad_norm": 1.0510096549987793,
+      "learning_rate": 3.9679942936609095e-05,
+      "loss": 0.5828,
+      "step": 10230
+    },
+    {
+      "epoch": 0.13621715624616262,
+      "grad_norm": 0.9769448637962341,
+      "learning_rate": 3.961188120762596e-05,
+      "loss": 0.5878,
+      "step": 10261
+    },
+    {
+      "epoch": 0.1366286884402598,
+      "grad_norm": 0.9610607028007507,
+      "learning_rate": 3.954365458554938e-05,
+      "loss": 0.5863,
+      "step": 10292
+    },
+    {
+      "epoch": 0.13704022063435697,
+      "grad_norm": 1.0102349519729614,
+      "learning_rate": 3.947526384030751e-05,
+      "loss": 0.5812,
+      "step": 10323
+    },
+    {
+      "epoch": 0.13745175282845412,
+      "grad_norm": 1.138122320175171,
+      "learning_rate": 3.9406709743680624e-05,
+      "loss": 0.5841,
+      "step": 10354
+    },
+    {
+      "epoch": 0.1378632850225513,
+      "grad_norm": 1.0107802152633667,
+      "learning_rate": 3.9337993069292366e-05,
+      "loss": 0.5847,
+      "step": 10385
+    },
+    {
+      "epoch": 0.13827481721664847,
+      "grad_norm": 0.9923847317695618,
+      "learning_rate": 3.926911459260109e-05,
+      "loss": 0.5971,
+      "step": 10416
+    },
+    {
+      "epoch": 0.13868634941074565,
+      "grad_norm": 1.0700715780258179,
+      "learning_rate": 3.920007509089102e-05,
+      "loss": 0.5831,
+      "step": 10447
+    },
+    {
+      "epoch": 0.1390978816048428,
+      "grad_norm": 1.1638612747192383,
+      "learning_rate": 3.913087534326357e-05,
+      "loss": 0.5849,
+      "step": 10478
+    },
+    {
+      "epoch": 0.13950941379893997,
+      "grad_norm": 1.0560393333435059,
+      "learning_rate": 3.9061516130628475e-05,
+      "loss": 0.5897,
+      "step": 10509
+    },
+    {
+      "epoch": 0.13992094599303714,
+      "grad_norm": 0.9604248404502869,
+      "learning_rate": 3.8991998235695025e-05,
+      "loss": 0.5862,
+      "step": 10540
+    },
+    {
+      "epoch": 0.14033247818713432,
+      "grad_norm": 1.0003983974456787,
+      "learning_rate": 3.8922322442963224e-05,
+      "loss": 0.5905,
+      "step": 10571
+    },
+    {
+      "epoch": 0.14074401038123147,
+      "grad_norm": 1.0378546714782715,
+      "learning_rate": 3.885248953871491e-05,
+      "loss": 0.5938,
+      "step": 10602
+    },
+    {
+      "epoch": 0.14115554257532864,
+      "grad_norm": 0.9967820644378662,
+      "learning_rate": 3.8782500311004915e-05,
+      "loss": 0.5821,
+      "step": 10633
+    },
+    {
+      "epoch": 0.14156707476942582,
+      "grad_norm": 1.029188632965088,
+      "learning_rate": 3.871235554965218e-05,
+      "loss": 0.5882,
+      "step": 10664
+    },
+    {
+      "epoch": 0.141978606963523,
+      "grad_norm": 1.1272863149642944,
+      "learning_rate": 3.864205604623078e-05,
+      "loss": 0.5898,
+      "step": 10695
+    },
+    {
+      "epoch": 0.14239013915762014,
+      "grad_norm": 1.0054222345352173,
+      "learning_rate": 3.857160259406107e-05,
+      "loss": 0.5873,
+      "step": 10726
+    },
+    {
+      "epoch": 0.1428016713517173,
+      "grad_norm": 1.0327833890914917,
+      "learning_rate": 3.8500995988200674e-05,
+      "loss": 0.5842,
+      "step": 10757
+    },
+    {
+      "epoch": 0.1432132035458145,
+      "grad_norm": 1.0088285207748413,
+      "learning_rate": 3.843023702543556e-05,
+      "loss": 0.5903,
+      "step": 10788
+    },
+    {
+      "epoch": 0.14362473573991166,
+      "grad_norm": 0.9472872018814087,
+      "learning_rate": 3.8359326504270984e-05,
+      "loss": 0.5809,
+      "step": 10819
+    },
+    {
+      "epoch": 0.1440362679340088,
+      "grad_norm": 1.1008390188217163,
+      "learning_rate": 3.828826522492255e-05,
+      "loss": 0.5775,
+      "step": 10850
+    },
+    {
+      "epoch": 0.14444780012810599,
+      "grad_norm": 1.1251379251480103,
+      "learning_rate": 3.821705398930713e-05,
+      "loss": 0.5821,
+      "step": 10881
+    },
+    {
+      "epoch": 0.14485933232220316,
+      "grad_norm": 0.9825986623764038,
+      "learning_rate": 3.814569360103385e-05,
+      "loss": 0.5749,
+      "step": 10912
+    },
+    {
+      "epoch": 0.14527086451630034,
+      "grad_norm": 0.9303480982780457,
+      "learning_rate": 3.807418486539499e-05,
+      "loss": 0.5827,
+      "step": 10943
+    },
+    {
+      "epoch": 0.14568239671039748,
+      "grad_norm": 0.9948610663414001,
+      "learning_rate": 3.80025285893569e-05,
+      "loss": 0.598,
+      "step": 10974
+    },
+    {
+      "epoch": 0.14609392890449466,
+      "grad_norm": 1.096449851989746,
+      "learning_rate": 3.793072558155093e-05,
+      "loss": 0.5749,
+      "step": 11005
+    },
+    {
+      "epoch": 0.14650546109859183,
+      "grad_norm": 1.022290587425232,
+      "learning_rate": 3.785877665226426e-05,
+      "loss": 0.5824,
+      "step": 11036
+    },
+    {
+      "epoch": 0.146916993292689,
+      "grad_norm": 1.083853840827942,
+      "learning_rate": 3.778668261343079e-05,
+      "loss": 0.5923,
+      "step": 11067
+    },
+    {
+      "epoch": 0.14732852548678615,
+      "grad_norm": 0.919503390789032,
+      "learning_rate": 3.771444427862192e-05,
+      "loss": 0.5924,
+      "step": 11098
+    },
+    {
+      "epoch": 0.14774005768088333,
+      "grad_norm": 1.0682687759399414,
+      "learning_rate": 3.7642062463037465e-05,
+      "loss": 0.5824,
+      "step": 11129
+    },
+    {
+      "epoch": 0.1481515898749805,
+      "grad_norm": 0.9599080085754395,
+      "learning_rate": 3.7569537983496373e-05,
+      "loss": 0.5765,
+      "step": 11160
+    },
+    {
+      "epoch": 0.14856312206907768,
+      "grad_norm": 0.968349039554596,
+      "learning_rate": 3.749687165842753e-05,
+      "loss": 0.5932,
+      "step": 11191
+    },
+    {
+      "epoch": 0.14897465426317483,
+      "grad_norm": 0.9846999049186707,
+      "learning_rate": 3.7424064307860536e-05,
+      "loss": 0.5791,
+      "step": 11222
+    },
+    {
+      "epoch": 0.149386186457272,
+      "grad_norm": 1.0673292875289917,
+      "learning_rate": 3.735111675341645e-05,
+      "loss": 0.5803,
+      "step": 11253
+    },
+    {
+      "epoch": 0.14979771865136918,
+      "grad_norm": 1.0111202001571655,
+      "learning_rate": 3.7278029818298524e-05,
+      "loss": 0.5837,
+      "step": 11284
+    },
+    {
+      "epoch": 0.15020925084546635,
+      "grad_norm": 1.0955816507339478,
+      "learning_rate": 3.720480432728287e-05,
+      "loss": 0.5822,
+      "step": 11315
+    },
+    {
+      "epoch": 0.1506207830395635,
+      "grad_norm": 0.9147946238517761,
+      "learning_rate": 3.71314411067092e-05,
+      "loss": 0.5873,
+      "step": 11346
+    },
+    {
+      "epoch": 0.15103231523366067,
+      "grad_norm": 0.9630762934684753,
+      "learning_rate": 3.70579409844715e-05,
+      "loss": 0.5914,
+      "step": 11377
+    },
+    {
+      "epoch": 0.15144384742775785,
+      "grad_norm": 0.9123234748840332,
+      "learning_rate": 3.698430479000865e-05,
+      "loss": 0.5784,
+      "step": 11408
+    },
+    {
+      "epoch": 0.15185537962185502,
+      "grad_norm": 1.0654667615890503,
+      "learning_rate": 3.691053335429509e-05,
+      "loss": 0.5805,
+      "step": 11439
+    },
+    {
+      "epoch": 0.15226691181595217,
+      "grad_norm": 1.111958622932434,
+      "learning_rate": 3.683662750983147e-05,
+      "loss": 0.5868,
+      "step": 11470
+    },
+    {
+      "epoch": 0.15267844401004935,
+      "grad_norm": 1.024422526359558,
+      "learning_rate": 3.676258809063518e-05,
+      "loss": 0.5904,
+      "step": 11501
+    },
+    {
+      "epoch": 0.15308997620414652,
+      "grad_norm": 1.092668056488037,
+      "learning_rate": 3.6688415932231004e-05,
+      "loss": 0.5885,
+      "step": 11532
+    },
+    {
+      "epoch": 0.1535015083982437,
+      "grad_norm": 0.9739312529563904,
+      "learning_rate": 3.661411187164166e-05,
+      "loss": 0.5806,
+      "step": 11563
+    },
+    {
+      "epoch": 0.15391304059234084,
+      "grad_norm": 1.098699688911438,
+      "learning_rate": 3.65396767473784e-05,
+      "loss": 0.5755,
+      "step": 11594
+    },
+    {
+      "epoch": 0.15432457278643802,
+      "grad_norm": 0.9998940825462341,
+      "learning_rate": 3.6465111399431465e-05,
+      "loss": 0.5758,
+      "step": 11625
+    },
+    {
+      "epoch": 0.1547361049805352,
+      "grad_norm": 0.9077128767967224,
+      "learning_rate": 3.6390416669260674e-05,
+      "loss": 0.5742,
+      "step": 11656
+    },
+    {
+      "epoch": 0.15514763717463237,
+      "grad_norm": 0.9274038076400757,
+      "learning_rate": 3.63155933997859e-05,
+      "loss": 0.589,
+      "step": 11687
+    },
+    {
+      "epoch": 0.15555916936872952,
+      "grad_norm": 0.9404906034469604,
+      "learning_rate": 3.624064243537758e-05,
+      "loss": 0.5848,
+      "step": 11718
+    },
+    {
+      "epoch": 0.1559707015628267,
+      "grad_norm": 0.9896044731140137,
+      "learning_rate": 3.616556462184716e-05,
+      "loss": 0.5917,
+      "step": 11749
+    },
+    {
+      "epoch": 0.15638223375692387,
+      "grad_norm": 1.0280394554138184,
+      "learning_rate": 3.609036080643755e-05,
+      "loss": 0.5811,
+      "step": 11780
+    },
+    {
+      "epoch": 0.15679376595102104,
+      "grad_norm": 0.9462539553642273,
+      "learning_rate": 3.60150318378136e-05,
+      "loss": 0.5782,
+      "step": 11811
+    },
+    {
+      "epoch": 0.1572052981451182,
+      "grad_norm": 1.047635555267334,
+      "learning_rate": 3.5939578566052465e-05,
+      "loss": 0.5792,
+      "step": 11842
+    },
+    {
+      "epoch": 0.15761683033921536,
+      "grad_norm": 0.9613581299781799,
+      "learning_rate": 3.586400184263408e-05,
+      "loss": 0.5847,
+      "step": 11873
+    },
+    {
+      "epoch": 0.15802836253331254,
+      "grad_norm": 0.8944469094276428,
+      "learning_rate": 3.578830252043148e-05,
+      "loss": 0.5787,
+      "step": 11904
+    },
+    {
+      "epoch": 0.1584398947274097,
+      "grad_norm": 1.0031458139419556,
+      "learning_rate": 3.571248145370125e-05,
+      "loss": 0.5827,
+      "step": 11935
+    },
+    {
+      "epoch": 0.15885142692150686,
+      "grad_norm": 0.9425063729286194,
+      "learning_rate": 3.5636539498073794e-05,
+      "loss": 0.5899,
+      "step": 11966
+    },
+    {
+      "epoch": 0.15926295911560404,
+      "grad_norm": 1.0098518133163452,
+      "learning_rate": 3.556047751054378e-05,
+      "loss": 0.5849,
+      "step": 11997
+    },
+    {
+      "epoch": 0.1596744913097012,
+      "grad_norm": 0.8891544342041016,
+      "learning_rate": 3.548429634946039e-05,
+      "loss": 0.5772,
+      "step": 12028
+    },
+    {
+      "epoch": 0.16008602350379839,
+      "grad_norm": 1.1580991744995117,
+      "learning_rate": 3.540799687451768e-05,
+      "loss": 0.5859,
+      "step": 12059
+    },
+    {
+      "epoch": 0.16049755569789553,
+      "grad_norm": 1.046398639678955,
+      "learning_rate": 3.533157994674485e-05,
+      "loss": 0.582,
+      "step": 12090
+    },
+    {
+      "epoch": 0.1609090878919927,
+      "grad_norm": 1.0526766777038574,
+      "learning_rate": 3.5255046428496546e-05,
+      "loss": 0.5738,
+      "step": 12121
+    },
+    {
+      "epoch": 0.16132062008608988,
+      "grad_norm": 0.9915657043457031,
+      "learning_rate": 3.517839718344311e-05,
+      "loss": 0.5867,
+      "step": 12152
+    },
+    {
+      "epoch": 0.16173215228018706,
+      "grad_norm": 0.954210102558136,
+      "learning_rate": 3.510163307656086e-05,
+      "loss": 0.578,
+      "step": 12183
+    },
+    {
+      "epoch": 0.1621436844742842,
+      "grad_norm": 0.952920138835907,
+      "learning_rate": 3.5024754974122324e-05,
+      "loss": 0.5843,
+      "step": 12214
+    },
+    {
+      "epoch": 0.16255521666838138,
+      "grad_norm": 0.9184344410896301,
+      "learning_rate": 3.494776374368643e-05,
+      "loss": 0.5748,
+      "step": 12245
+    },
+    {
+      "epoch": 0.16296674886247856,
+      "grad_norm": 0.9797667264938354,
+      "learning_rate": 3.4870660254088724e-05,
+      "loss": 0.5788,
+      "step": 12276
+    },
+    {
+      "epoch": 0.16337828105657573,
+      "grad_norm": 0.968781054019928,
+      "learning_rate": 3.479344537543164e-05,
+      "loss": 0.5835,
+      "step": 12307
+    },
+    {
+      "epoch": 0.16378981325067288,
+      "grad_norm": 1.0632243156433105,
+      "learning_rate": 3.4716119979074565e-05,
+      "loss": 0.5742,
+      "step": 12338
+    },
+    {
+      "epoch": 0.16420134544477005,
+      "grad_norm": 0.9538365006446838,
+      "learning_rate": 3.463868493762412e-05,
+      "loss": 0.5737,
+      "step": 12369
+    },
+    {
+      "epoch": 0.16461287763886723,
+      "grad_norm": 0.9336950778961182,
+      "learning_rate": 3.456114112492418e-05,
+      "loss": 0.5821,
+      "step": 12400
+    },
+    {
+      "epoch": 0.1650244098329644,
+      "grad_norm": 0.9611359238624573,
+      "learning_rate": 3.4483489416046164e-05,
+      "loss": 0.5847,
+      "step": 12431
+    },
+    {
+      "epoch": 0.16543594202706155,
+      "grad_norm": 1.0813709497451782,
+      "learning_rate": 3.440573068727905e-05,
+      "loss": 0.5805,
+      "step": 12462
+    },
+    {
+      "epoch": 0.16584747422115872,
+      "grad_norm": 0.9780453443527222,
+      "learning_rate": 3.4327865816119495e-05,
+      "loss": 0.5761,
+      "step": 12493
+    },
+    {
+      "epoch": 0.1662590064152559,
+      "grad_norm": 0.9629074335098267,
+      "learning_rate": 3.4249895681262025e-05,
+      "loss": 0.58,
+      "step": 12524
+    },
+    {
+      "epoch": 0.16667053860935307,
+      "grad_norm": 0.9241612553596497,
+      "learning_rate": 3.417182116258899e-05,
+      "loss": 0.5732,
+      "step": 12555
+    },
+    {
+      "epoch": 0.16708207080345022,
+      "grad_norm": 1.0314536094665527,
+      "learning_rate": 3.409364314116074e-05,
+      "loss": 0.578,
+      "step": 12586
+    },
+    {
+      "epoch": 0.1674936029975474,
+      "grad_norm": 0.9907404184341431,
+      "learning_rate": 3.401536249920559e-05,
+      "loss": 0.591,
+      "step": 12617
+    },
+    {
+      "epoch": 0.16790513519164457,
+      "grad_norm": 0.9442338943481445,
+      "learning_rate": 3.393698012010998e-05,
+      "loss": 0.5754,
+      "step": 12648
+    },
+    {
+      "epoch": 0.16831666738574175,
+      "grad_norm": 0.9101460576057434,
+      "learning_rate": 3.385849688840839e-05,
+      "loss": 0.5815,
+      "step": 12679
+    },
+    {
+      "epoch": 0.1687281995798389,
+      "grad_norm": 1.029105544090271,
+      "learning_rate": 3.3779913689773414e-05,
+      "loss": 0.5766,
+      "step": 12710
+    },
+    {
+      "epoch": 0.16913973177393607,
+      "grad_norm": 1.028542160987854,
+      "learning_rate": 3.370123141100578e-05,
+      "loss": 0.5769,
+      "step": 12741
+    },
+    {
+      "epoch": 0.16955126396803324,
+      "grad_norm": 0.9411474466323853,
+      "learning_rate": 3.3622450940024305e-05,
+      "loss": 0.5808,
+      "step": 12772
+    },
+    {
+      "epoch": 0.16996279616213042,
+      "grad_norm": 0.9260859489440918,
+      "learning_rate": 3.35435731658559e-05,
+      "loss": 0.5683,
+      "step": 12803
+    },
+    {
+      "epoch": 0.17037432835622757,
+      "grad_norm": 0.9867289066314697,
+      "learning_rate": 3.346459897862552e-05,
+      "loss": 0.5781,
+      "step": 12834
+    },
+    {
+      "epoch": 0.17078586055032474,
+      "grad_norm": 0.9310885071754456,
+      "learning_rate": 3.338552926954613e-05,
+      "loss": 0.5876,
+      "step": 12865
+    },
+    {
+      "epoch": 0.17119739274442192,
+      "grad_norm": 1.1955841779708862,
+      "learning_rate": 3.330636493090868e-05,
+      "loss": 0.577,
+      "step": 12896
+    },
+    {
+      "epoch": 0.1716089249385191,
+      "grad_norm": 1.0715487003326416,
+      "learning_rate": 3.322710685607193e-05,
+      "loss": 0.5835,
+      "step": 12927
+    },
+    {
+      "epoch": 0.17202045713261624,
+      "grad_norm": 0.9181262850761414,
+      "learning_rate": 3.314775593945251e-05,
+      "loss": 0.5733,
+      "step": 12958
+    },
+    {
+      "epoch": 0.17243198932671341,
+      "grad_norm": 0.9595162868499756,
+      "learning_rate": 3.3068313076514714e-05,
+      "loss": 0.5685,
+      "step": 12989
+    },
+    {
+      "epoch": 0.1728435215208106,
+      "grad_norm": 1.057614803314209,
+      "learning_rate": 3.298877916376047e-05,
+      "loss": 0.5736,
+      "step": 13020
+    },
+    {
+      "epoch": 0.17325505371490776,
+      "grad_norm": 0.9654355049133301,
+      "learning_rate": 3.290915509871915e-05,
+      "loss": 0.5782,
+      "step": 13051
+    },
+    {
+      "epoch": 0.1736665859090049,
+      "grad_norm": 1.0919837951660156,
+      "learning_rate": 3.282944177993753e-05,
+      "loss": 0.5771,
+      "step": 13082
+    },
+    {
+      "epoch": 0.1740781181031021,
+      "grad_norm": 1.064331293106079,
+      "learning_rate": 3.274964010696957e-05,
+      "loss": 0.5731,
+      "step": 13113
+    },
+    {
+      "epoch": 0.17448965029719926,
+      "grad_norm": 0.8971393704414368,
+      "learning_rate": 3.266975098036629e-05,
+      "loss": 0.5855,
+      "step": 13144
+    },
+    {
+      "epoch": 0.17490118249129644,
+      "grad_norm": 0.9515472054481506,
+      "learning_rate": 3.258977530166562e-05,
+      "loss": 0.5776,
+      "step": 13175
+    },
+    {
+      "epoch": 0.17531271468539358,
+      "grad_norm": 0.9811455607414246,
+      "learning_rate": 3.250971397338227e-05,
+      "loss": 0.5817,
+      "step": 13206
+    },
+    {
+      "epoch": 0.17572424687949076,
+      "grad_norm": 0.9187535643577576,
+      "learning_rate": 3.2429567898997404e-05,
+      "loss": 0.5751,
+      "step": 13237
+    },
+    {
+      "epoch": 0.17613577907358793,
+      "grad_norm": 0.9396876096725464,
+      "learning_rate": 3.234933798294859e-05,
+      "loss": 0.5744,
+      "step": 13268
+    },
+    {
+      "epoch": 0.1765473112676851,
+      "grad_norm": 1.064921259880066,
+      "learning_rate": 3.2269025130619535e-05,
+      "loss": 0.5731,
+      "step": 13299
+    },
+    {
+      "epoch": 0.17695884346178226,
+      "grad_norm": 0.946753978729248,
+      "learning_rate": 3.218863024832985e-05,
+      "loss": 0.5824,
+      "step": 13330
+    },
+    {
+      "epoch": 0.17737037565587943,
+      "grad_norm": 0.8623449206352234,
+      "learning_rate": 3.2108154243324864e-05,
+      "loss": 0.5764,
+      "step": 13361
+    },
+    {
+      "epoch": 0.1777819078499766,
+      "grad_norm": 0.9526584148406982,
+      "learning_rate": 3.2027598023765345e-05,
+      "loss": 0.5696,
+      "step": 13392
+    },
+    {
+      "epoch": 0.17819344004407378,
+      "grad_norm": 0.9413353800773621,
+      "learning_rate": 3.194696249871729e-05,
+      "loss": 0.5688,
+      "step": 13423
+    },
+    {
+      "epoch": 0.17860497223817093,
+      "grad_norm": 0.9612322449684143,
+      "learning_rate": 3.186624857814164e-05,
+      "loss": 0.5694,
+      "step": 13454
+    },
+    {
+      "epoch": 0.1790165044322681,
+      "grad_norm": 1.0289851427078247,
+      "learning_rate": 3.178545717288401e-05,
+      "loss": 0.5799,
+      "step": 13485
+    },
+    {
+      "epoch": 0.17942803662636528,
+      "grad_norm": 0.9124505519866943,
+      "learning_rate": 3.170458919466444e-05,
+      "loss": 0.5729,
+      "step": 13516
+    },
+    {
+      "epoch": 0.17983956882046245,
+      "grad_norm": 0.9874391555786133,
+      "learning_rate": 3.1623645556067063e-05,
+      "loss": 0.5761,
+      "step": 13547
+    },
+    {
+      "epoch": 0.1802511010145596,
+      "grad_norm": 1.0058220624923706,
+      "learning_rate": 3.154262717052985e-05,
+      "loss": 0.58,
+      "step": 13578
+    },
+    {
+      "epoch": 0.18066263320865678,
+      "grad_norm": 0.9753491282463074,
+      "learning_rate": 3.146153495233426e-05,
+      "loss": 0.5749,
+      "step": 13609
+    },
+    {
+      "epoch": 0.18107416540275395,
+      "grad_norm": 1.0167535543441772,
+      "learning_rate": 3.1380369816594944e-05,
+      "loss": 0.5848,
+      "step": 13640
+    },
+    {
+      "epoch": 0.18148569759685113,
+      "grad_norm": 0.9970638155937195,
+      "learning_rate": 3.129913267924946e-05,
+      "loss": 0.5743,
+      "step": 13671
+    },
+    {
+      "epoch": 0.18189722979094827,
+      "grad_norm": 1.0033198595046997,
+      "learning_rate": 3.121782445704782e-05,
+      "loss": 0.5747,
+      "step": 13702
+    },
+    {
+      "epoch": 0.18230876198504545,
+      "grad_norm": 1.0107535123825073,
+      "learning_rate": 3.11364460675423e-05,
+      "loss": 0.5725,
+      "step": 13733
+    },
+    {
+      "epoch": 0.18272029417914262,
+      "grad_norm": 0.9341425895690918,
+      "learning_rate": 3.1054998429076934e-05,
+      "loss": 0.5726,
+      "step": 13764
+    },
+    {
+      "epoch": 0.1831318263732398,
+      "grad_norm": 1.0088342428207397,
+      "learning_rate": 3.097348246077728e-05,
+      "loss": 0.5737,
+      "step": 13795
+    },
+    {
+      "epoch": 0.18354335856733694,
+      "grad_norm": 0.8871366381645203,
+      "learning_rate": 3.0891899082539924e-05,
+      "loss": 0.576,
+      "step": 13826
+    },
+    {
+      "epoch": 0.18395489076143412,
+      "grad_norm": 0.971552848815918,
+      "learning_rate": 3.0810249215022233e-05,
+      "loss": 0.57,
+      "step": 13857
+    },
+    {
+      "epoch": 0.1843664229555313,
+      "grad_norm": 1.1167266368865967,
+      "learning_rate": 3.0728533779631865e-05,
+      "loss": 0.5652,
+      "step": 13888
+    },
+    {
+      "epoch": 0.18477795514962847,
+      "grad_norm": 0.9768183827400208,
+      "learning_rate": 3.064675369851637e-05,
+      "loss": 0.5671,
+      "step": 13919
+    },
+    {
+      "epoch": 0.18518948734372562,
+      "grad_norm": 1.0097460746765137,
+      "learning_rate": 3.056490989455289e-05,
+      "loss": 0.5771,
+      "step": 13950
+    },
+    {
+      "epoch": 0.1856010195378228,
+      "grad_norm": 1.010986089706421,
+      "learning_rate": 3.0483003291337596e-05,
+      "loss": 0.5723,
+      "step": 13981
+    },
+    {
+      "epoch": 0.18601255173191997,
+      "grad_norm": 0.8861789107322693,
+      "learning_rate": 3.040103481317539e-05,
+      "loss": 0.5798,
+      "step": 14012
+    },
+    {
+      "epoch": 0.18642408392601714,
+      "grad_norm": 0.9811322689056396,
+      "learning_rate": 3.03190053850694e-05,
+      "loss": 0.5703,
+      "step": 14043
+    },
+    {
+      "epoch": 0.1868356161201143,
+      "grad_norm": 0.9183774590492249,
+      "learning_rate": 3.0236915932710573e-05,
+      "loss": 0.5655,
+      "step": 14074
+    },
+    {
+      "epoch": 0.18724714831421146,
+      "grad_norm": 0.9655874371528625,
+      "learning_rate": 3.0154767382467232e-05,
+      "loss": 0.57,
+      "step": 14105
+    },
+    {
+      "epoch": 0.18765868050830864,
+      "grad_norm": 0.9735666513442993,
+      "learning_rate": 3.0072560661374582e-05,
+      "loss": 0.5748,
+      "step": 14136
+    },
+    {
+      "epoch": 0.18807021270240581,
+      "grad_norm": 0.9084784388542175,
+      "learning_rate": 2.999029669712431e-05,
+      "loss": 0.568,
+      "step": 14167
+    },
+    {
+      "epoch": 0.18848174489650296,
+      "grad_norm": 0.8866302371025085,
+      "learning_rate": 2.990797641805408e-05,
+      "loss": 0.5768,
+      "step": 14198
+    },
+    {
+      "epoch": 0.18889327709060014,
+      "grad_norm": 0.9936773180961609,
+      "learning_rate": 2.982560075313704e-05,
+      "loss": 0.5701,
+      "step": 14229
+    },
+    {
+      "epoch": 0.1893048092846973,
+      "grad_norm": 0.9721146821975708,
+      "learning_rate": 2.9743170631971368e-05,
+      "loss": 0.5638,
+      "step": 14260
+    },
+    {
+      "epoch": 0.1897163414787945,
+      "grad_norm": 0.921503484249115,
+      "learning_rate": 2.9660686984769792e-05,
+      "loss": 0.5645,
+      "step": 14291
+    },
+    {
+      "epoch": 0.19012787367289163,
+      "grad_norm": 0.9194900989532471,
+      "learning_rate": 2.9578150742349047e-05,
+      "loss": 0.5724,
+      "step": 14322
+    },
+    {
+      "epoch": 0.1905394058669888,
+      "grad_norm": 0.9940860271453857,
+      "learning_rate": 2.949556283611942e-05,
+      "loss": 0.5731,
+      "step": 14353
+    },
+    {
+      "epoch": 0.19095093806108598,
+      "grad_norm": 0.9163900017738342,
+      "learning_rate": 2.9412924198074206e-05,
+      "loss": 0.5761,
+      "step": 14384
+    },
+    {
+      "epoch": 0.19136247025518316,
+      "grad_norm": 0.960118293762207,
+      "learning_rate": 2.9330235760779208e-05,
+      "loss": 0.572,
+      "step": 14415
+    },
+    {
+      "epoch": 0.1917740024492803,
+      "grad_norm": 0.9942934513092041,
+      "learning_rate": 2.9247498457362188e-05,
+      "loss": 0.5606,
+      "step": 14446
+    },
+    {
+      "epoch": 0.19218553464337748,
+      "grad_norm": 0.9903921484947205,
+      "learning_rate": 2.9164713221502373e-05,
+      "loss": 0.5704,
+      "step": 14477
+    },
+    {
+      "epoch": 0.19259706683747466,
+      "grad_norm": 0.9734652638435364,
+      "learning_rate": 2.9081880987419912e-05,
+      "loss": 0.5768,
+      "step": 14508
+    },
+    {
+      "epoch": 0.19300859903157183,
+      "grad_norm": 0.9251719117164612,
+      "learning_rate": 2.8999002689865296e-05,
+      "loss": 0.565,
+      "step": 14539
+    },
+    {
+      "epoch": 0.19342013122566898,
+      "grad_norm": 0.9734142422676086,
+      "learning_rate": 2.8916079264108852e-05,
+      "loss": 0.5702,
+      "step": 14570
+    },
+    {
+      "epoch": 0.19383166341976615,
+      "grad_norm": 0.9692136645317078,
+      "learning_rate": 2.883311164593017e-05,
+      "loss": 0.5581,
+      "step": 14601
+    },
+    {
+      "epoch": 0.19424319561386333,
+      "grad_norm": 0.9654501676559448,
+      "learning_rate": 2.875010077160754e-05,
+      "loss": 0.5667,
+      "step": 14632
+    },
+    {
+      "epoch": 0.1946547278079605,
+      "grad_norm": 1.0234603881835938,
+      "learning_rate": 2.866704757790741e-05,
+      "loss": 0.5688,
+      "step": 14663
+    },
+    {
+      "epoch": 0.19506626000205765,
+      "grad_norm": 0.9619661569595337,
+      "learning_rate": 2.858395300207376e-05,
+      "loss": 0.5713,
+      "step": 14694
+    },
+    {
+      "epoch": 0.19547779219615483,
+      "grad_norm": 1.0468534231185913,
+      "learning_rate": 2.8500817981817607e-05,
+      "loss": 0.5702,
+      "step": 14725
+    },
+    {
+      "epoch": 0.195889324390252,
+      "grad_norm": 0.9849717617034912,
+      "learning_rate": 2.8417643455306336e-05,
+      "loss": 0.5695,
+      "step": 14756
+    },
+    {
+      "epoch": 0.19630085658434918,
+      "grad_norm": 0.8887643218040466,
+      "learning_rate": 2.8334430361153185e-05,
+      "loss": 0.5688,
+      "step": 14787
+    },
+    {
+      "epoch": 0.19671238877844632,
+      "grad_norm": 0.9583494067192078,
+      "learning_rate": 2.8251179638406612e-05,
+      "loss": 0.5775,
+      "step": 14818
+    },
+    {
+      "epoch": 0.1971239209725435,
+      "grad_norm": 0.9766852259635925,
+      "learning_rate": 2.8167892226539704e-05,
+      "loss": 0.5688,
+      "step": 14849
+    },
+    {
+      "epoch": 0.19753545316664067,
+      "grad_norm": 0.9450570940971375,
+      "learning_rate": 2.8084569065439588e-05,
+      "loss": 0.5773,
+      "step": 14880
+    },
+    {
+      "epoch": 0.19794698536073785,
+      "grad_norm": 0.9476996064186096,
+      "learning_rate": 2.8001211095396807e-05,
+      "loss": 0.5688,
+      "step": 14911
+    },
+    {
+      "epoch": 0.198358517554835,
+      "grad_norm": 1.0239835977554321,
+      "learning_rate": 2.791781925709473e-05,
+      "loss": 0.5717,
+      "step": 14942
+    },
+    {
+      "epoch": 0.19877004974893217,
+      "grad_norm": 0.9848045110702515,
+      "learning_rate": 2.7834394491598908e-05,
+      "loss": 0.5691,
+      "step": 14973
+    },
+    {
+      "epoch": 0.19918158194302935,
+      "grad_norm": 1.0350494384765625,
+      "learning_rate": 2.7750937740346485e-05,
+      "loss": 0.5808,
+      "step": 15004
+    },
+    {
+      "epoch": 0.19959311413712652,
+      "grad_norm": 0.8743448853492737,
+      "learning_rate": 2.7667449945135564e-05,
+      "loss": 0.5694,
+      "step": 15035
+    },
+    {
+      "epoch": 0.20000464633122367,
+      "grad_norm": 0.9410389065742493,
+      "learning_rate": 2.7583932048114557e-05,
+      "loss": 0.5739,
+      "step": 15066
+    },
+    {
+      "epoch": 0.20041617852532084,
+      "grad_norm": 1.0113860368728638,
+      "learning_rate": 2.7500384991771587e-05,
+      "loss": 0.5689,
+      "step": 15097
+    },
+    {
+      "epoch": 0.20082771071941802,
+      "grad_norm": 0.998852550983429,
+      "learning_rate": 2.7416809718923825e-05,
+      "loss": 0.5619,
+      "step": 15128
+    },
+    {
+      "epoch": 0.2012392429135152,
+      "grad_norm": 0.9299794435501099,
+      "learning_rate": 2.7333207172706864e-05,
+      "loss": 0.5672,
+      "step": 15159
+    },
+    {
+      "epoch": 0.20165077510761234,
+      "grad_norm": 1.0109282732009888,
+      "learning_rate": 2.7249578296564088e-05,
+      "loss": 0.5691,
+      "step": 15190
+    },
+    {
+      "epoch": 0.20206230730170952,
+      "grad_norm": 0.8805022835731506,
+      "learning_rate": 2.7165924034235973e-05,
+      "loss": 0.5601,
+      "step": 15221
+    },
+    {
+      "epoch": 0.2024738394958067,
+      "grad_norm": 0.8976924419403076,
+      "learning_rate": 2.708224532974953e-05,
+      "loss": 0.5669,
+      "step": 15252
+    }
+  ],
+  "logging_steps": 31,
+  "max_steps": 30517,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 3052,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.1258280394232955e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-15260/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1bf5225534ee746c36dba9f28281e0b858e6e9b4a308bdaf1de4f68e79e9ee83
+size 5304

checkpoint-18312/config.json

@@ -0,0 +1,39 @@
+{
+  "_name_or_path": "meta-llama/Llama-3.1-8B-Instruct",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": [
+    128001,
+    128008,
+    128009
+  ],
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.44.0.dev0",
+  "use_cache": true,
+  "vocab_size": 128256
+}

checkpoint-18312/generation_config.json

@@ -0,0 +1,12 @@
+{
+  "bos_token_id": 128000,
+  "do_sample": true,
+  "eos_token_id": [
+    128001,
+    128008,
+    128009
+  ],
+  "temperature": 0.6,
+  "top_p": 0.9,
+  "transformers_version": "4.44.0.dev0"
+}

checkpoint-18312/model-00001-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9da583efd4b108a0cf4092795df246eab9009dc721e00caa77c8a7b89382c481
+size 4886466168

checkpoint-18312/model-00002-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23fa5aaf8ac56c17c821980c9aee44fcbc2a617fc570e875e6fdb93d92886c72
+size 4832007448

checkpoint-18312/model-00003-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a223e87240e50dfb05634ea909cf1ec715d37a12b681b14a3a6e97f28688f2b
+size 4999813112

checkpoint-18312/model-00004-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3047d199c611c6d12b1cca780f033ad89df9c076f742674f7ac4a892ca9573b3
+size 4999813128

checkpoint-18312/model-00005-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a2f91d749d664ecd22c71f3d54488966fc6899a03924b8fea3613c8f83398a8
+size 4832007496

checkpoint-18312/model-00006-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f261f76e4ce233bc88ed14286ede45f5659ba8e4998ff9d64dc54aec6c834254
+size 4999813120

checkpoint-18312/model-00007-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3ce2de917797dc9fc88acfd0dbf2eb74d9a30720ea31edf94ae68250f8fc973
+size 2571158184

checkpoint-18312/model.safetensors.index.json

@@ -0,0 +1,298 @@
+{
+  "metadata": {
+    "total_size": 32121044992
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00007-of-00007.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00004-of-00007.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00005-of-00007.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00007.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00007-of-00007.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00007-of-00007.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00007-of-00007.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00007-of-00007.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00006-of-00007.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00007.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00003-of-00007.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00003-of-00007.safetensors",
+    "model.norm.weight": "model-00007-of-00007.safetensors"
+  }
+}

checkpoint-18312/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c60452df8a312f7d6249784451c9d60a09cda5da56d0637a615ec6e2c069545
+size 15385036334

checkpoint-18312/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9d9cd6a0487226e5bd30d1846894c82af483733ab4381b75bae9c0745e05d405
+size 14244

checkpoint-18312/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0e65c3d6f29e706fd941a38280ce5628189a6998eac6d29abbeab00ad838d00
+size 1064

checkpoint-18312/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1bf5225534ee746c36dba9f28281e0b858e6e9b4a308bdaf1de4f68e79e9ee83
+size 5304

checkpoint-21364/config.json

@@ -0,0 +1,39 @@
+{
+  "_name_or_path": "meta-llama/Llama-3.1-8B-Instruct",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": [
+    128001,
+    128008,
+    128009
+  ],
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.44.0.dev0",
+  "use_cache": true,
+  "vocab_size": 128256
+}

checkpoint-21364/generation_config.json

@@ -0,0 +1,12 @@
+{
+  "bos_token_id": 128000,
+  "do_sample": true,
+  "eos_token_id": [
+    128001,
+    128008,
+    128009
+  ],
+  "temperature": 0.6,
+  "top_p": 0.9,
+  "transformers_version": "4.44.0.dev0"
+}

checkpoint-21364/model-00001-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b23c1daf15a450da6427a6c1c9869383ab8e27cc46ad11eed8fb2a3f2c16d06
+size 4886466168

checkpoint-21364/model-00002-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23fa5aaf8ac56c17c821980c9aee44fcbc2a617fc570e875e6fdb93d92886c72
+size 4832007448

checkpoint-21364/model-00003-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a223e87240e50dfb05634ea909cf1ec715d37a12b681b14a3a6e97f28688f2b
+size 4999813112