Upload folder using huggingface_hub

checkpoint-426/config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-crh_Latn-coli-focus/",
+  "architectures": [
+    "Olmo2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "eos_token_id": 100257,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 4096,
+  "model_type": "olmo2",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 100277,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 500000,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.48.0.dev0",
+  "use_cache": true,
+  "vocab_size": 100384
+}

checkpoint-426/generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 100257,
+  "pad_token_id": 100277,
+  "transformers_version": "4.48.0.dev0"
+}

checkpoint-426/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34dc43b41f1b5be7c65f8f9ae0daa7949a1392367ffe28e6c3ec933394872a68
+size 4950070472

checkpoint-426/model-00002-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de0b6a357d03676c4986588aaf4dc96afc3f44a0498514961d2b2cfa1940655e
+size 4991622816

checkpoint-426/model-00003-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1038a7d346a681a8fb6949e054c2246f3f3c80e826855c1cec6af1155cc607d
+size 4924546896

checkpoint-426/model-00004-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33c0a21146ace02d36928787ae7f34be868f7c81b894dea98ec7b0e5a3ea01b2
+size 4857404928

checkpoint-426/model-00005-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e7f82a3014ccfc0f67208f63785ae4bc71f09aba86db48bf6139d18178abbfed
+size 4857404928

checkpoint-426/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3638691b7d5f88304e2ac0924f50ce4d4753f81bba744a611d7553a459daee07
+size 4614509080

checkpoint-426/model.safetensors.index.json

@@ -0,0 +1,362 @@
+{
+  "metadata": {
+    "total_size": 29195517952
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00006-of-00006.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.k_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.q_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.k_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.q_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.k_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.q_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.k_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.q_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.k_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.q_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.k_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.q_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.k_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.q_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.norm.weight": "model-00006-of-00006.safetensors"
+  }
+}

checkpoint-426/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1eac4b5cbe092ff8397c465243667b54db8bd6300afd717e41ba17a9c48c1d5
+size 13055369358

checkpoint-426/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:386fcc8cc1089aade9450d86fb239ea3483f455fd2d78d8378645feecfec9d69
+size 14244

checkpoint-426/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c2e469a95a5e22da498ff5e3c2d75190e44234622578023ca88508705c88838
+size 1064

checkpoint-426/trainer_state.json

@@ -0,0 +1,3015 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 426,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002347417840375587,
+      "grad_norm": 5.220906734466553,
+      "learning_rate": 2.325581395348837e-06,
+      "loss": 6.1043,
+      "step": 1
+    },
+    {
+      "epoch": 0.004694835680751174,
+      "grad_norm": 15.844380378723145,
+      "learning_rate": 4.651162790697674e-06,
+      "loss": 5.8695,
+      "step": 2
+    },
+    {
+      "epoch": 0.007042253521126761,
+      "grad_norm": 5.460394382476807,
+      "learning_rate": 6.976744186046512e-06,
+      "loss": 6.2482,
+      "step": 3
+    },
+    {
+      "epoch": 0.009389671361502348,
+      "grad_norm": 3.862333297729492,
+      "learning_rate": 9.302325581395349e-06,
+      "loss": 5.4233,
+      "step": 4
+    },
+    {
+      "epoch": 0.011737089201877934,
+      "grad_norm": 4.396307945251465,
+      "learning_rate": 1.1627906976744187e-05,
+      "loss": 5.9159,
+      "step": 5
+    },
+    {
+      "epoch": 0.014084507042253521,
+      "grad_norm": 4.15396785736084,
+      "learning_rate": 1.3953488372093024e-05,
+      "loss": 5.6399,
+      "step": 6
+    },
+    {
+      "epoch": 0.01643192488262911,
+      "grad_norm": 4.129617691040039,
+      "learning_rate": 1.6279069767441862e-05,
+      "loss": 5.9078,
+      "step": 7
+    },
+    {
+      "epoch": 0.018779342723004695,
+      "grad_norm": 3.6875030994415283,
+      "learning_rate": 1.8604651162790697e-05,
+      "loss": 5.4132,
+      "step": 8
+    },
+    {
+      "epoch": 0.02112676056338028,
+      "grad_norm": 3.7894229888916016,
+      "learning_rate": 2.0930232558139536e-05,
+      "loss": 5.5096,
+      "step": 9
+    },
+    {
+      "epoch": 0.023474178403755867,
+      "grad_norm": 3.776921510696411,
+      "learning_rate": 2.3255813953488374e-05,
+      "loss": 5.4388,
+      "step": 10
+    },
+    {
+      "epoch": 0.025821596244131457,
+      "grad_norm": 4.050227165222168,
+      "learning_rate": 2.5581395348837212e-05,
+      "loss": 5.2998,
+      "step": 11
+    },
+    {
+      "epoch": 0.028169014084507043,
+      "grad_norm": 3.6630780696868896,
+      "learning_rate": 2.7906976744186048e-05,
+      "loss": 5.1655,
+      "step": 12
+    },
+    {
+      "epoch": 0.03051643192488263,
+      "grad_norm": 3.518064022064209,
+      "learning_rate": 3.0232558139534883e-05,
+      "loss": 4.9769,
+      "step": 13
+    },
+    {
+      "epoch": 0.03286384976525822,
+      "grad_norm": 3.442387819290161,
+      "learning_rate": 3.2558139534883724e-05,
+      "loss": 5.1886,
+      "step": 14
+    },
+    {
+      "epoch": 0.035211267605633804,
+      "grad_norm": 3.6376004219055176,
+      "learning_rate": 3.488372093023256e-05,
+      "loss": 4.9641,
+      "step": 15
+    },
+    {
+      "epoch": 0.03755868544600939,
+      "grad_norm": 3.279391050338745,
+      "learning_rate": 3.7209302325581394e-05,
+      "loss": 4.6454,
+      "step": 16
+    },
+    {
+      "epoch": 0.03990610328638498,
+      "grad_norm": 3.5410876274108887,
+      "learning_rate": 3.953488372093023e-05,
+      "loss": 4.9914,
+      "step": 17
+    },
+    {
+      "epoch": 0.04225352112676056,
+      "grad_norm": 3.387657403945923,
+      "learning_rate": 4.186046511627907e-05,
+      "loss": 4.689,
+      "step": 18
+    },
+    {
+      "epoch": 0.04460093896713615,
+      "grad_norm": 3.4205541610717773,
+      "learning_rate": 4.418604651162791e-05,
+      "loss": 4.6573,
+      "step": 19
+    },
+    {
+      "epoch": 0.046948356807511735,
+      "grad_norm": 3.2225935459136963,
+      "learning_rate": 4.651162790697675e-05,
+      "loss": 4.5582,
+      "step": 20
+    },
+    {
+      "epoch": 0.04929577464788732,
+      "grad_norm": 3.5002527236938477,
+      "learning_rate": 4.883720930232558e-05,
+      "loss": 4.601,
+      "step": 21
+    },
+    {
+      "epoch": 0.051643192488262914,
+      "grad_norm": 3.211371421813965,
+      "learning_rate": 5.1162790697674425e-05,
+      "loss": 4.5814,
+      "step": 22
+    },
+    {
+      "epoch": 0.0539906103286385,
+      "grad_norm": 3.2980473041534424,
+      "learning_rate": 5.348837209302326e-05,
+      "loss": 4.4787,
+      "step": 23
+    },
+    {
+      "epoch": 0.056338028169014086,
+      "grad_norm": 3.2750470638275146,
+      "learning_rate": 5.5813953488372095e-05,
+      "loss": 4.1589,
+      "step": 24
+    },
+    {
+      "epoch": 0.05868544600938967,
+      "grad_norm": 2.6927506923675537,
+      "learning_rate": 5.8139534883720933e-05,
+      "loss": 4.122,
+      "step": 25
+    },
+    {
+      "epoch": 0.06103286384976526,
+      "grad_norm": 3.17693829536438,
+      "learning_rate": 6.0465116279069765e-05,
+      "loss": 4.1636,
+      "step": 26
+    },
+    {
+      "epoch": 0.06338028169014084,
+      "grad_norm": 2.782895803451538,
+      "learning_rate": 6.27906976744186e-05,
+      "loss": 3.992,
+      "step": 27
+    },
+    {
+      "epoch": 0.06572769953051644,
+      "grad_norm": 2.8504180908203125,
+      "learning_rate": 6.511627906976745e-05,
+      "loss": 3.8498,
+      "step": 28
+    },
+    {
+      "epoch": 0.06807511737089202,
+      "grad_norm": 2.6587069034576416,
+      "learning_rate": 6.744186046511628e-05,
+      "loss": 3.9163,
+      "step": 29
+    },
+    {
+      "epoch": 0.07042253521126761,
+      "grad_norm": 2.423595428466797,
+      "learning_rate": 6.976744186046513e-05,
+      "loss": 3.5204,
+      "step": 30
+    },
+    {
+      "epoch": 0.07276995305164319,
+      "grad_norm": 2.753427267074585,
+      "learning_rate": 7.209302325581396e-05,
+      "loss": 3.7614,
+      "step": 31
+    },
+    {
+      "epoch": 0.07511737089201878,
+      "grad_norm": 2.574305295944214,
+      "learning_rate": 7.441860465116279e-05,
+      "loss": 3.8774,
+      "step": 32
+    },
+    {
+      "epoch": 0.07746478873239436,
+      "grad_norm": 2.499316930770874,
+      "learning_rate": 7.674418604651163e-05,
+      "loss": 3.7411,
+      "step": 33
+    },
+    {
+      "epoch": 0.07981220657276995,
+      "grad_norm": 2.2817907333374023,
+      "learning_rate": 7.906976744186047e-05,
+      "loss": 3.4423,
+      "step": 34
+    },
+    {
+      "epoch": 0.08215962441314555,
+      "grad_norm": 2.522942304611206,
+      "learning_rate": 8.139534883720931e-05,
+      "loss": 3.4958,
+      "step": 35
+    },
+    {
+      "epoch": 0.08450704225352113,
+      "grad_norm": 2.4626259803771973,
+      "learning_rate": 8.372093023255814e-05,
+      "loss": 3.7215,
+      "step": 36
+    },
+    {
+      "epoch": 0.08685446009389672,
+      "grad_norm": 2.2035305500030518,
+      "learning_rate": 8.604651162790697e-05,
+      "loss": 3.4304,
+      "step": 37
+    },
+    {
+      "epoch": 0.0892018779342723,
+      "grad_norm": 2.4290783405303955,
+      "learning_rate": 8.837209302325582e-05,
+      "loss": 3.5983,
+      "step": 38
+    },
+    {
+      "epoch": 0.09154929577464789,
+      "grad_norm": 2.3165831565856934,
+      "learning_rate": 9.069767441860465e-05,
+      "loss": 3.332,
+      "step": 39
+    },
+    {
+      "epoch": 0.09389671361502347,
+      "grad_norm": 2.430633544921875,
+      "learning_rate": 9.30232558139535e-05,
+      "loss": 3.4888,
+      "step": 40
+    },
+    {
+      "epoch": 0.09624413145539906,
+      "grad_norm": 2.3381195068359375,
+      "learning_rate": 9.534883720930233e-05,
+      "loss": 3.4018,
+      "step": 41
+    },
+    {
+      "epoch": 0.09859154929577464,
+      "grad_norm": 2.3274364471435547,
+      "learning_rate": 9.767441860465116e-05,
+      "loss": 3.2374,
+      "step": 42
+    },
+    {
+      "epoch": 0.10093896713615023,
+      "grad_norm": 2.4454705715179443,
+      "learning_rate": 0.0001,
+      "loss": 3.5324,
+      "step": 43
+    },
+    {
+      "epoch": 0.10328638497652583,
+      "grad_norm": 2.446553945541382,
+      "learning_rate": 9.999962299929265e-05,
+      "loss": 3.497,
+      "step": 44
+    },
+    {
+      "epoch": 0.1056338028169014,
+      "grad_norm": 2.4503588676452637,
+      "learning_rate": 9.999849200285573e-05,
+      "loss": 3.4668,
+      "step": 45
+    },
+    {
+      "epoch": 0.107981220657277,
+      "grad_norm": 2.3425333499908447,
+      "learning_rate": 9.999660702774474e-05,
+      "loss": 3.3311,
+      "step": 46
+    },
+    {
+      "epoch": 0.11032863849765258,
+      "grad_norm": 2.1232123374938965,
+      "learning_rate": 9.999396810238513e-05,
+      "loss": 3.3883,
+      "step": 47
+    },
+    {
+      "epoch": 0.11267605633802817,
+      "grad_norm": 2.3176381587982178,
+      "learning_rate": 9.999057526657196e-05,
+      "loss": 3.3303,
+      "step": 48
+    },
+    {
+      "epoch": 0.11502347417840375,
+      "grad_norm": 2.322566509246826,
+      "learning_rate": 9.998642857146934e-05,
+      "loss": 3.3612,
+      "step": 49
+    },
+    {
+      "epoch": 0.11737089201877934,
+      "grad_norm": 2.373108148574829,
+      "learning_rate": 9.99815280796095e-05,
+      "loss": 3.0915,
+      "step": 50
+    },
+    {
+      "epoch": 0.11971830985915492,
+      "grad_norm": 2.4871981143951416,
+      "learning_rate": 9.997587386489202e-05,
+      "loss": 3.2645,
+      "step": 51
+    },
+    {
+      "epoch": 0.12206572769953052,
+      "grad_norm": 2.254300117492676,
+      "learning_rate": 9.996946601258261e-05,
+      "loss": 3.3558,
+      "step": 52
+    },
+    {
+      "epoch": 0.12441314553990611,
+      "grad_norm": 2.237212657928467,
+      "learning_rate": 9.996230461931186e-05,
+      "loss": 3.0436,
+      "step": 53
+    },
+    {
+      "epoch": 0.1267605633802817,
+      "grad_norm": 4.646354675292969,
+      "learning_rate": 9.995438979307381e-05,
+      "loss": 3.1667,
+      "step": 54
+    },
+    {
+      "epoch": 0.12910798122065728,
+      "grad_norm": 2.17333722114563,
+      "learning_rate": 9.994572165322422e-05,
+      "loss": 3.2507,
+      "step": 55
+    },
+    {
+      "epoch": 0.13145539906103287,
+      "grad_norm": 2.353606700897217,
+      "learning_rate": 9.993630033047891e-05,
+      "loss": 3.3378,
+      "step": 56
+    },
+    {
+      "epoch": 0.13380281690140844,
+      "grad_norm": 2.1471004486083984,
+      "learning_rate": 9.992612596691171e-05,
+      "loss": 3.2807,
+      "step": 57
+    },
+    {
+      "epoch": 0.13615023474178403,
+      "grad_norm": 2.406132936477661,
+      "learning_rate": 9.991519871595227e-05,
+      "loss": 3.0291,
+      "step": 58
+    },
+    {
+      "epoch": 0.13849765258215962,
+      "grad_norm": 2.1640617847442627,
+      "learning_rate": 9.990351874238388e-05,
+      "loss": 3.0177,
+      "step": 59
+    },
+    {
+      "epoch": 0.14084507042253522,
+      "grad_norm": 2.043344497680664,
+      "learning_rate": 9.989108622234084e-05,
+      "loss": 2.9692,
+      "step": 60
+    },
+    {
+      "epoch": 0.1431924882629108,
+      "grad_norm": 2.105788469314575,
+      "learning_rate": 9.987790134330593e-05,
+      "loss": 3.0785,
+      "step": 61
+    },
+    {
+      "epoch": 0.14553990610328638,
+      "grad_norm": 2.2852227687835693,
+      "learning_rate": 9.986396430410749e-05,
+      "loss": 3.1042,
+      "step": 62
+    },
+    {
+      "epoch": 0.14788732394366197,
+      "grad_norm": 2.2685129642486572,
+      "learning_rate": 9.984927531491648e-05,
+      "loss": 3.1491,
+      "step": 63
+    },
+    {
+      "epoch": 0.15023474178403756,
+      "grad_norm": 2.0371837615966797,
+      "learning_rate": 9.983383459724322e-05,
+      "loss": 3.0921,
+      "step": 64
+    },
+    {
+      "epoch": 0.15258215962441316,
+      "grad_norm": 2.1931519508361816,
+      "learning_rate": 9.981764238393424e-05,
+      "loss": 3.2441,
+      "step": 65
+    },
+    {
+      "epoch": 0.15492957746478872,
+      "grad_norm": 1.8535912036895752,
+      "learning_rate": 9.980069891916854e-05,
+      "loss": 2.8031,
+      "step": 66
+    },
+    {
+      "epoch": 0.1572769953051643,
+      "grad_norm": 2.0124950408935547,
+      "learning_rate": 9.978300445845404e-05,
+      "loss": 3.0444,
+      "step": 67
+    },
+    {
+      "epoch": 0.1596244131455399,
+      "grad_norm": 2.106775999069214,
+      "learning_rate": 9.976455926862371e-05,
+      "loss": 3.114,
+      "step": 68
+    },
+    {
+      "epoch": 0.1619718309859155,
+      "grad_norm": 2.230271339416504,
+      "learning_rate": 9.974536362783156e-05,
+      "loss": 3.0906,
+      "step": 69
+    },
+    {
+      "epoch": 0.1643192488262911,
+      "grad_norm": 2.040703773498535,
+      "learning_rate": 9.972541782554836e-05,
+      "loss": 3.0528,
+      "step": 70
+    },
+    {
+      "epoch": 0.16666666666666666,
+      "grad_norm": 2.1196584701538086,
+      "learning_rate": 9.970472216255741e-05,
+      "loss": 2.9784,
+      "step": 71
+    },
+    {
+      "epoch": 0.16901408450704225,
+      "grad_norm": 2.0533766746520996,
+      "learning_rate": 9.968327695094987e-05,
+      "loss": 3.1503,
+      "step": 72
+    },
+    {
+      "epoch": 0.17136150234741784,
+      "grad_norm": 2.0557050704956055,
+      "learning_rate": 9.966108251412014e-05,
+      "loss": 3.1607,
+      "step": 73
+    },
+    {
+      "epoch": 0.17370892018779344,
+      "grad_norm": 2.1403017044067383,
+      "learning_rate": 9.963813918676095e-05,
+      "loss": 3.1178,
+      "step": 74
+    },
+    {
+      "epoch": 0.176056338028169,
+      "grad_norm": 2.0884039402008057,
+      "learning_rate": 9.961444731485836e-05,
+      "loss": 2.9815,
+      "step": 75
+    },
+    {
+      "epoch": 0.1784037558685446,
+      "grad_norm": 2.026855945587158,
+      "learning_rate": 9.959000725568641e-05,
+      "loss": 2.939,
+      "step": 76
+    },
+    {
+      "epoch": 0.1807511737089202,
+      "grad_norm": 2.05558180809021,
+      "learning_rate": 9.956481937780193e-05,
+      "loss": 3.0461,
+      "step": 77
+    },
+    {
+      "epoch": 0.18309859154929578,
+      "grad_norm": 2.1393980979919434,
+      "learning_rate": 9.953888406103883e-05,
+      "loss": 2.9019,
+      "step": 78
+    },
+    {
+      "epoch": 0.18544600938967137,
+      "grad_norm": 2.192837953567505,
+      "learning_rate": 9.951220169650239e-05,
+      "loss": 3.198,
+      "step": 79
+    },
+    {
+      "epoch": 0.18779342723004694,
+      "grad_norm": 2.011807441711426,
+      "learning_rate": 9.948477268656345e-05,
+      "loss": 2.8485,
+      "step": 80
+    },
+    {
+      "epoch": 0.19014084507042253,
+      "grad_norm": 1.94229257106781,
+      "learning_rate": 9.945659744485225e-05,
+      "loss": 2.7786,
+      "step": 81
+    },
+    {
+      "epoch": 0.19248826291079812,
+      "grad_norm": 2.1388649940490723,
+      "learning_rate": 9.942767639625224e-05,
+      "loss": 2.8766,
+      "step": 82
+    },
+    {
+      "epoch": 0.19483568075117372,
+      "grad_norm": 2.0566341876983643,
+      "learning_rate": 9.939800997689362e-05,
+      "loss": 3.2437,
+      "step": 83
+    },
+    {
+      "epoch": 0.19718309859154928,
+      "grad_norm": 1.9980113506317139,
+      "learning_rate": 9.936759863414685e-05,
+      "loss": 2.9535,
+      "step": 84
+    },
+    {
+      "epoch": 0.19953051643192488,
+      "grad_norm": 1.9093738794326782,
+      "learning_rate": 9.933644282661586e-05,
+      "loss": 2.9371,
+      "step": 85
+    },
+    {
+      "epoch": 0.20187793427230047,
+      "grad_norm": 2.054807186126709,
+      "learning_rate": 9.930454302413108e-05,
+      "loss": 3.0883,
+      "step": 86
+    },
+    {
+      "epoch": 0.20422535211267606,
+      "grad_norm": 2.235306978225708,
+      "learning_rate": 9.927189970774245e-05,
+      "loss": 3.0301,
+      "step": 87
+    },
+    {
+      "epoch": 0.20657276995305165,
+      "grad_norm": 2.04367995262146,
+      "learning_rate": 9.923851336971208e-05,
+      "loss": 2.9837,
+      "step": 88
+    },
+    {
+      "epoch": 0.20892018779342722,
+      "grad_norm": 2.107414722442627,
+      "learning_rate": 9.920438451350694e-05,
+      "loss": 3.0341,
+      "step": 89
+    },
+    {
+      "epoch": 0.2112676056338028,
+      "grad_norm": 2.0880820751190186,
+      "learning_rate": 9.916951365379112e-05,
+      "loss": 3.0869,
+      "step": 90
+    },
+    {
+      "epoch": 0.2136150234741784,
+      "grad_norm": 1.8780890703201294,
+      "learning_rate": 9.913390131641815e-05,
+      "loss": 2.8691,
+      "step": 91
+    },
+    {
+      "epoch": 0.215962441314554,
+      "grad_norm": 1.9201158285140991,
+      "learning_rate": 9.909754803842313e-05,
+      "loss": 2.8319,
+      "step": 92
+    },
+    {
+      "epoch": 0.21830985915492956,
+      "grad_norm": 2.0277490615844727,
+      "learning_rate": 9.906045436801448e-05,
+      "loss": 2.9869,
+      "step": 93
+    },
+    {
+      "epoch": 0.22065727699530516,
+      "grad_norm": 1.8977261781692505,
+      "learning_rate": 9.902262086456582e-05,
+      "loss": 2.9073,
+      "step": 94
+    },
+    {
+      "epoch": 0.22300469483568075,
+      "grad_norm": 2.1085357666015625,
+      "learning_rate": 9.898404809860744e-05,
+      "loss": 2.9194,
+      "step": 95
+    },
+    {
+      "epoch": 0.22535211267605634,
+      "grad_norm": 1.9869375228881836,
+      "learning_rate": 9.894473665181776e-05,
+      "loss": 3.0151,
+      "step": 96
+    },
+    {
+      "epoch": 0.22769953051643194,
+      "grad_norm": 1.9094806909561157,
+      "learning_rate": 9.89046871170145e-05,
+      "loss": 2.9288,
+      "step": 97
+    },
+    {
+      "epoch": 0.2300469483568075,
+      "grad_norm": 2.171858549118042,
+      "learning_rate": 9.886390009814579e-05,
+      "loss": 2.812,
+      "step": 98
+    },
+    {
+      "epoch": 0.2323943661971831,
+      "grad_norm": 2.082744836807251,
+      "learning_rate": 9.882237621028101e-05,
+      "loss": 3.0519,
+      "step": 99
+    },
+    {
+      "epoch": 0.2347417840375587,
+      "grad_norm": 1.8079471588134766,
+      "learning_rate": 9.878011607960156e-05,
+      "loss": 2.8567,
+      "step": 100
+    },
+    {
+      "epoch": 0.23708920187793428,
+      "grad_norm": 2.0076165199279785,
+      "learning_rate": 9.873712034339143e-05,
+      "loss": 2.9788,
+      "step": 101
+    },
+    {
+      "epoch": 0.23943661971830985,
+      "grad_norm": 2.177436113357544,
+      "learning_rate": 9.869338965002752e-05,
+      "loss": 2.778,
+      "step": 102
+    },
+    {
+      "epoch": 0.24178403755868544,
+      "grad_norm": 1.8783988952636719,
+      "learning_rate": 9.864892465896994e-05,
+      "loss": 2.7731,
+      "step": 103
+    },
+    {
+      "epoch": 0.24413145539906103,
+      "grad_norm": 1.989122748374939,
+      "learning_rate": 9.860372604075199e-05,
+      "loss": 2.9812,
+      "step": 104
+    },
+    {
+      "epoch": 0.24647887323943662,
+      "grad_norm": 1.842214822769165,
+      "learning_rate": 9.855779447697013e-05,
+      "loss": 2.6943,
+      "step": 105
+    },
+    {
+      "epoch": 0.24882629107981222,
+      "grad_norm": 1.8644382953643799,
+      "learning_rate": 9.851113066027364e-05,
+      "loss": 2.7616,
+      "step": 106
+    },
+    {
+      "epoch": 0.2511737089201878,
+      "grad_norm": 2.2355384826660156,
+      "learning_rate": 9.846373529435418e-05,
+      "loss": 3.0248,
+      "step": 107
+    },
+    {
+      "epoch": 0.2535211267605634,
+      "grad_norm": 1.8173452615737915,
+      "learning_rate": 9.841560909393523e-05,
+      "loss": 2.7491,
+      "step": 108
+    },
+    {
+      "epoch": 0.25586854460093894,
+      "grad_norm": 3.4129257202148438,
+      "learning_rate": 9.836675278476124e-05,
+      "loss": 2.8953,
+      "step": 109
+    },
+    {
+      "epoch": 0.25821596244131456,
+      "grad_norm": 1.8477997779846191,
+      "learning_rate": 9.831716710358673e-05,
+      "loss": 2.7385,
+      "step": 110
+    },
+    {
+      "epoch": 0.2605633802816901,
+      "grad_norm": 1.8806322813034058,
+      "learning_rate": 9.82668527981652e-05,
+      "loss": 2.7953,
+      "step": 111
+    },
+    {
+      "epoch": 0.26291079812206575,
+      "grad_norm": 2.0917327404022217,
+      "learning_rate": 9.821581062723779e-05,
+      "loss": 2.8176,
+      "step": 112
+    },
+    {
+      "epoch": 0.2652582159624413,
+      "grad_norm": 1.8491564989089966,
+      "learning_rate": 9.816404136052186e-05,
+      "loss": 2.6668,
+      "step": 113
+    },
+    {
+      "epoch": 0.2676056338028169,
+      "grad_norm": 2.1390769481658936,
+      "learning_rate": 9.811154577869943e-05,
+      "loss": 2.9057,
+      "step": 114
+    },
+    {
+      "epoch": 0.2699530516431925,
+      "grad_norm": 1.824035406112671,
+      "learning_rate": 9.805832467340538e-05,
+      "loss": 2.8375,
+      "step": 115
+    },
+    {
+      "epoch": 0.27230046948356806,
+      "grad_norm": 2.00138521194458,
+      "learning_rate": 9.800437884721545e-05,
+      "loss": 2.9452,
+      "step": 116
+    },
+    {
+      "epoch": 0.2746478873239437,
+      "grad_norm": 1.9364036321640015,
+      "learning_rate": 9.794970911363426e-05,
+      "loss": 2.9226,
+      "step": 117
+    },
+    {
+      "epoch": 0.27699530516431925,
+      "grad_norm": 1.9031555652618408,
+      "learning_rate": 9.78943162970829e-05,
+      "loss": 2.7785,
+      "step": 118
+    },
+    {
+      "epoch": 0.2793427230046948,
+      "grad_norm": 1.8698092699050903,
+      "learning_rate": 9.783820123288664e-05,
+      "loss": 2.7981,
+      "step": 119
+    },
+    {
+      "epoch": 0.28169014084507044,
+      "grad_norm": 1.9158368110656738,
+      "learning_rate": 9.778136476726223e-05,
+      "loss": 2.7539,
+      "step": 120
+    },
+    {
+      "epoch": 0.284037558685446,
+      "grad_norm": 2.0701334476470947,
+      "learning_rate": 9.772380775730516e-05,
+      "loss": 2.8241,
+      "step": 121
+    },
+    {
+      "epoch": 0.2863849765258216,
+      "grad_norm": 1.950182557106018,
+      "learning_rate": 9.766553107097681e-05,
+      "loss": 2.9841,
+      "step": 122
+    },
+    {
+      "epoch": 0.2887323943661972,
+      "grad_norm": 1.9728721380233765,
+      "learning_rate": 9.760653558709122e-05,
+      "loss": 2.9186,
+      "step": 123
+    },
+    {
+      "epoch": 0.29107981220657275,
+      "grad_norm": 1.9013452529907227,
+      "learning_rate": 9.754682219530199e-05,
+      "loss": 2.9011,
+      "step": 124
+    },
+    {
+      "epoch": 0.2934272300469484,
+      "grad_norm": 1.8539355993270874,
+      "learning_rate": 9.748639179608872e-05,
+      "loss": 2.6117,
+      "step": 125
+    },
+    {
+      "epoch": 0.29577464788732394,
+      "grad_norm": 1.9760916233062744,
+      "learning_rate": 9.742524530074355e-05,
+      "loss": 2.7754,
+      "step": 126
+    },
+    {
+      "epoch": 0.2981220657276995,
+      "grad_norm": 1.9756038188934326,
+      "learning_rate": 9.736338363135738e-05,
+      "loss": 2.8952,
+      "step": 127
+    },
+    {
+      "epoch": 0.3004694835680751,
+      "grad_norm": 1.8712913990020752,
+      "learning_rate": 9.730080772080592e-05,
+      "loss": 2.8044,
+      "step": 128
+    },
+    {
+      "epoch": 0.3028169014084507,
+      "grad_norm": 1.926211953163147,
+      "learning_rate": 9.723751851273568e-05,
+      "loss": 2.8259,
+      "step": 129
+    },
+    {
+      "epoch": 0.3051643192488263,
+      "grad_norm": 2.0510101318359375,
+      "learning_rate": 9.71735169615497e-05,
+      "loss": 2.8108,
+      "step": 130
+    },
+    {
+      "epoch": 0.3075117370892019,
+      "grad_norm": 2.0798439979553223,
+      "learning_rate": 9.710880403239317e-05,
+      "loss": 2.6996,
+      "step": 131
+    },
+    {
+      "epoch": 0.30985915492957744,
+      "grad_norm": 1.8246979713439941,
+      "learning_rate": 9.704338070113893e-05,
+      "loss": 2.7598,
+      "step": 132
+    },
+    {
+      "epoch": 0.31220657276995306,
+      "grad_norm": 1.8421847820281982,
+      "learning_rate": 9.697724795437264e-05,
+      "loss": 2.6863,
+      "step": 133
+    },
+    {
+      "epoch": 0.3145539906103286,
+      "grad_norm": 1.75267493724823,
+      "learning_rate": 9.6910406789378e-05,
+      "loss": 2.6435,
+      "step": 134
+    },
+    {
+      "epoch": 0.31690140845070425,
+      "grad_norm": 1.816770076751709,
+      "learning_rate": 9.684285821412165e-05,
+      "loss": 2.7311,
+      "step": 135
+    },
+    {
+      "epoch": 0.3192488262910798,
+      "grad_norm": 1.9300565719604492,
+      "learning_rate": 9.677460324723806e-05,
+      "loss": 2.8216,
+      "step": 136
+    },
+    {
+      "epoch": 0.3215962441314554,
+      "grad_norm": 2.002140760421753,
+      "learning_rate": 9.670564291801401e-05,
+      "loss": 2.7712,
+      "step": 137
+    },
+    {
+      "epoch": 0.323943661971831,
+      "grad_norm": 1.8995237350463867,
+      "learning_rate": 9.663597826637325e-05,
+      "loss": 2.7779,
+      "step": 138
+    },
+    {
+      "epoch": 0.32629107981220656,
+      "grad_norm": 1.8259207010269165,
+      "learning_rate": 9.656561034286069e-05,
+      "loss": 2.719,
+      "step": 139
+    },
+    {
+      "epoch": 0.3286384976525822,
+      "grad_norm": 1.828887701034546,
+      "learning_rate": 9.64945402086266e-05,
+      "loss": 2.7256,
+      "step": 140
+    },
+    {
+      "epoch": 0.33098591549295775,
+      "grad_norm": 1.8349099159240723,
+      "learning_rate": 9.642276893541063e-05,
+      "loss": 2.5764,
+      "step": 141
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 1.7998813390731812,
+      "learning_rate": 9.63502976055256e-05,
+      "loss": 2.654,
+      "step": 142
+    },
+    {
+      "epoch": 0.33568075117370894,
+      "grad_norm": 1.9068450927734375,
+      "learning_rate": 9.627712731184122e-05,
+      "loss": 2.6458,
+      "step": 143
+    },
+    {
+      "epoch": 0.3380281690140845,
+      "grad_norm": 1.7458933591842651,
+      "learning_rate": 9.620325915776758e-05,
+      "loss": 2.5668,
+      "step": 144
+    },
+    {
+      "epoch": 0.3403755868544601,
+      "grad_norm": 1.9490630626678467,
+      "learning_rate": 9.612869425723854e-05,
+      "loss": 2.9618,
+      "step": 145
+    },
+    {
+      "epoch": 0.3427230046948357,
+      "grad_norm": 1.7958039045333862,
+      "learning_rate": 9.605343373469491e-05,
+      "loss": 2.66,
+      "step": 146
+    },
+    {
+      "epoch": 0.34507042253521125,
+      "grad_norm": 1.9553214311599731,
+      "learning_rate": 9.59774787250675e-05,
+      "loss": 2.697,
+      "step": 147
+    },
+    {
+      "epoch": 0.3474178403755869,
+      "grad_norm": 1.8985018730163574,
+      "learning_rate": 9.590083037376001e-05,
+      "loss": 2.7875,
+      "step": 148
+    },
+    {
+      "epoch": 0.34976525821596244,
+      "grad_norm": 1.8991280794143677,
+      "learning_rate": 9.582348983663173e-05,
+      "loss": 2.7981,
+      "step": 149
+    },
+    {
+      "epoch": 0.352112676056338,
+      "grad_norm": 1.8634614944458008,
+      "learning_rate": 9.574545827998017e-05,
+      "loss": 2.7077,
+      "step": 150
+    },
+    {
+      "epoch": 0.3544600938967136,
+      "grad_norm": 1.8225376605987549,
+      "learning_rate": 9.566673688052339e-05,
+      "loss": 2.549,
+      "step": 151
+    },
+    {
+      "epoch": 0.3568075117370892,
+      "grad_norm": 1.90485680103302,
+      "learning_rate": 9.558732682538233e-05,
+      "loss": 2.5873,
+      "step": 152
+    },
+    {
+      "epoch": 0.3591549295774648,
+      "grad_norm": 1.9928096532821655,
+      "learning_rate": 9.550722931206286e-05,
+      "loss": 2.7677,
+      "step": 153
+    },
+    {
+      "epoch": 0.3615023474178404,
+      "grad_norm": 1.7967183589935303,
+      "learning_rate": 9.542644554843778e-05,
+      "loss": 2.6462,
+      "step": 154
+    },
+    {
+      "epoch": 0.36384976525821594,
+      "grad_norm": 2.4245903491973877,
+      "learning_rate": 9.53449767527285e-05,
+      "loss": 2.6752,
+      "step": 155
+    },
+    {
+      "epoch": 0.36619718309859156,
+      "grad_norm": 1.9551819562911987,
+      "learning_rate": 9.526282415348677e-05,
+      "loss": 2.6328,
+      "step": 156
+    },
+    {
+      "epoch": 0.3685446009389671,
+      "grad_norm": 1.7784936428070068,
+      "learning_rate": 9.517998898957611e-05,
+      "loss": 2.7277,
+      "step": 157
+    },
+    {
+      "epoch": 0.37089201877934275,
+      "grad_norm": 1.885576605796814,
+      "learning_rate": 9.509647251015314e-05,
+      "loss": 2.8092,
+      "step": 158
+    },
+    {
+      "epoch": 0.3732394366197183,
+      "grad_norm": 1.983398199081421,
+      "learning_rate": 9.501227597464875e-05,
+      "loss": 2.8169,
+      "step": 159
+    },
+    {
+      "epoch": 0.3755868544600939,
+      "grad_norm": 1.7774951457977295,
+      "learning_rate": 9.492740065274904e-05,
+      "loss": 2.5589,
+      "step": 160
+    },
+    {
+      "epoch": 0.3779342723004695,
+      "grad_norm": 1.8690886497497559,
+      "learning_rate": 9.484184782437628e-05,
+      "loss": 2.5976,
+      "step": 161
+    },
+    {
+      "epoch": 0.38028169014084506,
+      "grad_norm": 1.8335201740264893,
+      "learning_rate": 9.475561877966956e-05,
+      "loss": 2.602,
+      "step": 162
+    },
+    {
+      "epoch": 0.3826291079812207,
+      "grad_norm": 1.7333793640136719,
+      "learning_rate": 9.46687148189653e-05,
+      "loss": 2.592,
+      "step": 163
+    },
+    {
+      "epoch": 0.38497652582159625,
+      "grad_norm": 1.879469871520996,
+      "learning_rate": 9.458113725277769e-05,
+      "loss": 2.7365,
+      "step": 164
+    },
+    {
+      "epoch": 0.3873239436619718,
+      "grad_norm": 1.7681939601898193,
+      "learning_rate": 9.449288740177891e-05,
+      "loss": 2.6856,
+      "step": 165
+    },
+    {
+      "epoch": 0.38967136150234744,
+      "grad_norm": 1.9350696802139282,
+      "learning_rate": 9.440396659677919e-05,
+      "loss": 2.586,
+      "step": 166
+    },
+    {
+      "epoch": 0.392018779342723,
+      "grad_norm": 1.889666199684143,
+      "learning_rate": 9.43143761787068e-05,
+      "loss": 2.9714,
+      "step": 167
+    },
+    {
+      "epoch": 0.39436619718309857,
+      "grad_norm": 2.181389808654785,
+      "learning_rate": 9.422411749858779e-05,
+      "loss": 2.808,
+      "step": 168
+    },
+    {
+      "epoch": 0.3967136150234742,
+      "grad_norm": 1.6735141277313232,
+      "learning_rate": 9.413319191752559e-05,
+      "loss": 2.5881,
+      "step": 169
+    },
+    {
+      "epoch": 0.39906103286384975,
+      "grad_norm": 1.7800458669662476,
+      "learning_rate": 9.404160080668055e-05,
+      "loss": 2.7968,
+      "step": 170
+    },
+    {
+      "epoch": 0.4014084507042254,
+      "grad_norm": 1.8180594444274902,
+      "learning_rate": 9.394934554724921e-05,
+      "loss": 2.72,
+      "step": 171
+    },
+    {
+      "epoch": 0.40375586854460094,
+      "grad_norm": 1.8427355289459229,
+      "learning_rate": 9.385642753044348e-05,
+      "loss": 2.6987,
+      "step": 172
+    },
+    {
+      "epoch": 0.4061032863849765,
+      "grad_norm": 1.8793630599975586,
+      "learning_rate": 9.37628481574697e-05,
+      "loss": 2.8138,
+      "step": 173
+    },
+    {
+      "epoch": 0.4084507042253521,
+      "grad_norm": 1.6694236993789673,
+      "learning_rate": 9.366860883950745e-05,
+      "loss": 2.5372,
+      "step": 174
+    },
+    {
+      "epoch": 0.4107981220657277,
+      "grad_norm": 1.8845412731170654,
+      "learning_rate": 9.357371099768833e-05,
+      "loss": 2.7629,
+      "step": 175
+    },
+    {
+      "epoch": 0.4131455399061033,
+      "grad_norm": 1.8960492610931396,
+      "learning_rate": 9.347815606307445e-05,
+      "loss": 2.8028,
+      "step": 176
+    },
+    {
+      "epoch": 0.4154929577464789,
+      "grad_norm": 1.8638942241668701,
+      "learning_rate": 9.338194547663694e-05,
+      "loss": 2.8008,
+      "step": 177
+    },
+    {
+      "epoch": 0.41784037558685444,
+      "grad_norm": 1.8362078666687012,
+      "learning_rate": 9.328508068923418e-05,
+      "loss": 2.5764,
+      "step": 178
+    },
+    {
+      "epoch": 0.42018779342723006,
+      "grad_norm": 2.151021957397461,
+      "learning_rate": 9.31875631615899e-05,
+      "loss": 2.8898,
+      "step": 179
+    },
+    {
+      "epoch": 0.4225352112676056,
+      "grad_norm": 1.6628094911575317,
+      "learning_rate": 9.308939436427115e-05,
+      "loss": 2.744,
+      "step": 180
+    },
+    {
+      "epoch": 0.42488262910798125,
+      "grad_norm": 1.8483365774154663,
+      "learning_rate": 9.299057577766622e-05,
+      "loss": 2.9555,
+      "step": 181
+    },
+    {
+      "epoch": 0.4272300469483568,
+      "grad_norm": 1.710507869720459,
+      "learning_rate": 9.289110889196214e-05,
+      "loss": 2.7485,
+      "step": 182
+    },
+    {
+      "epoch": 0.4295774647887324,
+      "grad_norm": 1.7637556791305542,
+      "learning_rate": 9.27909952071224e-05,
+      "loss": 2.3881,
+      "step": 183
+    },
+    {
+      "epoch": 0.431924882629108,
+      "grad_norm": 2.0040581226348877,
+      "learning_rate": 9.269023623286417e-05,
+      "loss": 2.636,
+      "step": 184
+    },
+    {
+      "epoch": 0.43427230046948356,
+      "grad_norm": 1.805621862411499,
+      "learning_rate": 9.258883348863566e-05,
+      "loss": 2.6267,
+      "step": 185
+    },
+    {
+      "epoch": 0.43661971830985913,
+      "grad_norm": 1.877990961074829,
+      "learning_rate": 9.248678850359309e-05,
+      "loss": 2.5549,
+      "step": 186
+    },
+    {
+      "epoch": 0.43896713615023475,
+      "grad_norm": 1.8069863319396973,
+      "learning_rate": 9.238410281657775e-05,
+      "loss": 2.8426,
+      "step": 187
+    },
+    {
+      "epoch": 0.4413145539906103,
+      "grad_norm": 1.6978660821914673,
+      "learning_rate": 9.228077797609269e-05,
+      "loss": 2.5319,
+      "step": 188
+    },
+    {
+      "epoch": 0.44366197183098594,
+      "grad_norm": 1.751092791557312,
+      "learning_rate": 9.217681554027945e-05,
+      "loss": 2.6608,
+      "step": 189
+    },
+    {
+      "epoch": 0.4460093896713615,
+      "grad_norm": 1.84478759765625,
+      "learning_rate": 9.207221707689447e-05,
+      "loss": 2.5484,
+      "step": 190
+    },
+    {
+      "epoch": 0.44835680751173707,
+      "grad_norm": 1.8681899309158325,
+      "learning_rate": 9.196698416328557e-05,
+      "loss": 2.6899,
+      "step": 191
+    },
+    {
+      "epoch": 0.4507042253521127,
+      "grad_norm": 1.7344541549682617,
+      "learning_rate": 9.186111838636804e-05,
+      "loss": 2.7196,
+      "step": 192
+    },
+    {
+      "epoch": 0.45305164319248825,
+      "grad_norm": 1.7881977558135986,
+      "learning_rate": 9.175462134260083e-05,
+      "loss": 2.4755,
+      "step": 193
+    },
+    {
+      "epoch": 0.45539906103286387,
+      "grad_norm": 1.8259541988372803,
+      "learning_rate": 9.16474946379623e-05,
+      "loss": 2.519,
+      "step": 194
+    },
+    {
+      "epoch": 0.45774647887323944,
+      "grad_norm": 1.7830413579940796,
+      "learning_rate": 9.153973988792626e-05,
+      "loss": 2.6012,
+      "step": 195
+    },
+    {
+      "epoch": 0.460093896713615,
+      "grad_norm": 1.7015130519866943,
+      "learning_rate": 9.143135871743736e-05,
+      "loss": 2.479,
+      "step": 196
+    },
+    {
+      "epoch": 0.4624413145539906,
+      "grad_norm": 1.860366940498352,
+      "learning_rate": 9.132235276088671e-05,
+      "loss": 2.5597,
+      "step": 197
+    },
+    {
+      "epoch": 0.4647887323943662,
+      "grad_norm": 1.7631980180740356,
+      "learning_rate": 9.121272366208722e-05,
+      "loss": 2.5369,
+      "step": 198
+    },
+    {
+      "epoch": 0.4671361502347418,
+      "grad_norm": 1.9521044492721558,
+      "learning_rate": 9.110247307424882e-05,
+      "loss": 2.7421,
+      "step": 199
+    },
+    {
+      "epoch": 0.4694835680751174,
+      "grad_norm": 1.8334943056106567,
+      "learning_rate": 9.09916026599535e-05,
+      "loss": 2.4926,
+      "step": 200
+    },
+    {
+      "epoch": 0.47183098591549294,
+      "grad_norm": 1.7488991022109985,
+      "learning_rate": 9.08801140911302e-05,
+      "loss": 2.6996,
+      "step": 201
+    },
+    {
+      "epoch": 0.47417840375586856,
+      "grad_norm": 1.7841949462890625,
+      "learning_rate": 9.076800904902975e-05,
+      "loss": 2.6786,
+      "step": 202
+    },
+    {
+      "epoch": 0.4765258215962441,
+      "grad_norm": 1.7553770542144775,
+      "learning_rate": 9.06552892241993e-05,
+      "loss": 2.5512,
+      "step": 203
+    },
+    {
+      "epoch": 0.4788732394366197,
+      "grad_norm": 1.783362627029419,
+      "learning_rate": 9.054195631645704e-05,
+      "loss": 2.6995,
+      "step": 204
+    },
+    {
+      "epoch": 0.4812206572769953,
+      "grad_norm": 1.6964733600616455,
+      "learning_rate": 9.042801203486641e-05,
+      "loss": 2.5438,
+      "step": 205
+    },
+    {
+      "epoch": 0.4835680751173709,
+      "grad_norm": 1.8186368942260742,
+      "learning_rate": 9.03134580977104e-05,
+      "loss": 2.5825,
+      "step": 206
+    },
+    {
+      "epoch": 0.4859154929577465,
+      "grad_norm": 1.7369734048843384,
+      "learning_rate": 9.019829623246563e-05,
+      "loss": 2.5467,
+      "step": 207
+    },
+    {
+      "epoch": 0.48826291079812206,
+      "grad_norm": 1.798734426498413,
+      "learning_rate": 9.008252817577628e-05,
+      "loss": 2.5932,
+      "step": 208
+    },
+    {
+      "epoch": 0.49061032863849763,
+      "grad_norm": 1.7268017530441284,
+      "learning_rate": 8.99661556734279e-05,
+      "loss": 2.3792,
+      "step": 209
+    },
+    {
+      "epoch": 0.49295774647887325,
+      "grad_norm": 1.737761378288269,
+      "learning_rate": 8.984918048032116e-05,
+      "loss": 2.4749,
+      "step": 210
+    },
+    {
+      "epoch": 0.4953051643192488,
+      "grad_norm": 1.6726443767547607,
+      "learning_rate": 8.973160436044526e-05,
+      "loss": 2.5939,
+      "step": 211
+    },
+    {
+      "epoch": 0.49765258215962443,
+      "grad_norm": 1.9009385108947754,
+      "learning_rate": 8.961342908685142e-05,
+      "loss": 2.7815,
+      "step": 212
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7680021524429321,
+      "learning_rate": 8.949465644162611e-05,
+      "loss": 2.6297,
+      "step": 213
+    },
+    {
+      "epoch": 0.5023474178403756,
+      "grad_norm": 1.7540963888168335,
+      "learning_rate": 8.937528821586416e-05,
+      "loss": 2.5078,
+      "step": 214
+    },
+    {
+      "epoch": 0.5046948356807511,
+      "grad_norm": 1.8365845680236816,
+      "learning_rate": 8.92553262096418e-05,
+      "loss": 2.5226,
+      "step": 215
+    },
+    {
+      "epoch": 0.5070422535211268,
+      "grad_norm": 1.7208716869354248,
+      "learning_rate": 8.913477223198951e-05,
+      "loss": 2.635,
+      "step": 216
+    },
+    {
+      "epoch": 0.5093896713615024,
+      "grad_norm": 1.9662084579467773,
+      "learning_rate": 8.901362810086464e-05,
+      "loss": 2.7483,
+      "step": 217
+    },
+    {
+      "epoch": 0.5117370892018779,
+      "grad_norm": 1.7092418670654297,
+      "learning_rate": 8.889189564312417e-05,
+      "loss": 2.4455,
+      "step": 218
+    },
+    {
+      "epoch": 0.5140845070422535,
+      "grad_norm": 1.9196945428848267,
+      "learning_rate": 8.876957669449694e-05,
+      "loss": 2.6385,
+      "step": 219
+    },
+    {
+      "epoch": 0.5164319248826291,
+      "grad_norm": 1.7106877565383911,
+      "learning_rate": 8.864667309955619e-05,
+      "loss": 2.7406,
+      "step": 220
+    },
+    {
+      "epoch": 0.5187793427230047,
+      "grad_norm": 1.9064276218414307,
+      "learning_rate": 8.852318671169162e-05,
+      "loss": 2.5328,
+      "step": 221
+    },
+    {
+      "epoch": 0.5211267605633803,
+      "grad_norm": 1.6696475744247437,
+      "learning_rate": 8.839911939308143e-05,
+      "loss": 2.3245,
+      "step": 222
+    },
+    {
+      "epoch": 0.5234741784037559,
+      "grad_norm": 1.7798056602478027,
+      "learning_rate": 8.827447301466432e-05,
+      "loss": 2.4523,
+      "step": 223
+    },
+    {
+      "epoch": 0.5258215962441315,
+      "grad_norm": 1.6665363311767578,
+      "learning_rate": 8.814924945611118e-05,
+      "loss": 2.4794,
+      "step": 224
+    },
+    {
+      "epoch": 0.528169014084507,
+      "grad_norm": 1.6816179752349854,
+      "learning_rate": 8.802345060579684e-05,
+      "loss": 2.7276,
+      "step": 225
+    },
+    {
+      "epoch": 0.5305164319248826,
+      "grad_norm": 1.61789870262146,
+      "learning_rate": 8.789707836077149e-05,
+      "loss": 2.4726,
+      "step": 226
+    },
+    {
+      "epoch": 0.5328638497652582,
+      "grad_norm": 1.699591040611267,
+      "learning_rate": 8.777013462673217e-05,
+      "loss": 2.4201,
+      "step": 227
+    },
+    {
+      "epoch": 0.5352112676056338,
+      "grad_norm": 1.639424443244934,
+      "learning_rate": 8.764262131799401e-05,
+      "loss": 2.6096,
+      "step": 228
+    },
+    {
+      "epoch": 0.5375586854460094,
+      "grad_norm": 1.7515225410461426,
+      "learning_rate": 8.75145403574613e-05,
+      "loss": 2.5429,
+      "step": 229
+    },
+    {
+      "epoch": 0.539906103286385,
+      "grad_norm": 1.6951862573623657,
+      "learning_rate": 8.738589367659852e-05,
+      "loss": 2.6377,
+      "step": 230
+    },
+    {
+      "epoch": 0.5422535211267606,
+      "grad_norm": 1.7144036293029785,
+      "learning_rate": 8.725668321540128e-05,
+      "loss": 2.5363,
+      "step": 231
+    },
+    {
+      "epoch": 0.5446009389671361,
+      "grad_norm": 1.840879201889038,
+      "learning_rate": 8.7126910922367e-05,
+      "loss": 2.6542,
+      "step": 232
+    },
+    {
+      "epoch": 0.5469483568075117,
+      "grad_norm": 1.7334511280059814,
+      "learning_rate": 8.699657875446551e-05,
+      "loss": 2.6133,
+      "step": 233
+    },
+    {
+      "epoch": 0.5492957746478874,
+      "grad_norm": 1.787042498588562,
+      "learning_rate": 8.686568867710962e-05,
+      "loss": 2.6379,
+      "step": 234
+    },
+    {
+      "epoch": 0.5516431924882629,
+      "grad_norm": 1.8859440088272095,
+      "learning_rate": 8.673424266412538e-05,
+      "loss": 2.7889,
+      "step": 235
+    },
+    {
+      "epoch": 0.5539906103286385,
+      "grad_norm": 1.7652978897094727,
+      "learning_rate": 8.660224269772237e-05,
+      "loss": 2.4923,
+      "step": 236
+    },
+    {
+      "epoch": 0.5563380281690141,
+      "grad_norm": 1.67996346950531,
+      "learning_rate": 8.646969076846383e-05,
+      "loss": 2.494,
+      "step": 237
+    },
+    {
+      "epoch": 0.5586854460093896,
+      "grad_norm": 1.906628966331482,
+      "learning_rate": 8.633658887523664e-05,
+      "loss": 2.8697,
+      "step": 238
+    },
+    {
+      "epoch": 0.5610328638497653,
+      "grad_norm": 1.7415281534194946,
+      "learning_rate": 8.620293902522105e-05,
+      "loss": 2.4383,
+      "step": 239
+    },
+    {
+      "epoch": 0.5633802816901409,
+      "grad_norm": 1.6793745756149292,
+      "learning_rate": 8.606874323386062e-05,
+      "loss": 2.5955,
+      "step": 240
+    },
+    {
+      "epoch": 0.5657276995305164,
+      "grad_norm": 1.6611347198486328,
+      "learning_rate": 8.593400352483168e-05,
+      "loss": 2.4733,
+      "step": 241
+    },
+    {
+      "epoch": 0.568075117370892,
+      "grad_norm": 1.715775966644287,
+      "learning_rate": 8.579872193001285e-05,
+      "loss": 2.593,
+      "step": 242
+    },
+    {
+      "epoch": 0.5704225352112676,
+      "grad_norm": 1.7070114612579346,
+      "learning_rate": 8.56629004894544e-05,
+      "loss": 2.5713,
+      "step": 243
+    },
+    {
+      "epoch": 0.5727699530516432,
+      "grad_norm": 1.6215184926986694,
+      "learning_rate": 8.552654125134751e-05,
+      "loss": 2.6497,
+      "step": 244
+    },
+    {
+      "epoch": 0.5751173708920188,
+      "grad_norm": 1.7327271699905396,
+      "learning_rate": 8.538964627199332e-05,
+      "loss": 2.6146,
+      "step": 245
+    },
+    {
+      "epoch": 0.5774647887323944,
+      "grad_norm": 1.832642912864685,
+      "learning_rate": 8.525221761577204e-05,
+      "loss": 2.7755,
+      "step": 246
+    },
+    {
+      "epoch": 0.57981220657277,
+      "grad_norm": 1.660343050956726,
+      "learning_rate": 8.511425735511167e-05,
+      "loss": 2.6576,
+      "step": 247
+    },
+    {
+      "epoch": 0.5821596244131455,
+      "grad_norm": 1.6340296268463135,
+      "learning_rate": 8.497576757045683e-05,
+      "loss": 2.1778,
+      "step": 248
+    },
+    {
+      "epoch": 0.5845070422535211,
+      "grad_norm": 1.8316805362701416,
+      "learning_rate": 8.483675035023739e-05,
+      "loss": 2.6375,
+      "step": 249
+    },
+    {
+      "epoch": 0.5868544600938967,
+      "grad_norm": 1.6770654916763306,
+      "learning_rate": 8.4697207790837e-05,
+      "loss": 2.497,
+      "step": 250
+    },
+    {
+      "epoch": 0.5892018779342723,
+      "grad_norm": 1.7220332622528076,
+      "learning_rate": 8.455714199656138e-05,
+      "loss": 2.4157,
+      "step": 251
+    },
+    {
+      "epoch": 0.5915492957746479,
+      "grad_norm": 1.6152470111846924,
+      "learning_rate": 8.441655507960667e-05,
+      "loss": 2.4673,
+      "step": 252
+    },
+    {
+      "epoch": 0.5938967136150235,
+      "grad_norm": 2.3333070278167725,
+      "learning_rate": 8.427544916002756e-05,
+      "loss": 2.8198,
+      "step": 253
+    },
+    {
+      "epoch": 0.596244131455399,
+      "grad_norm": 1.7391387224197388,
+      "learning_rate": 8.41338263657053e-05,
+      "loss": 2.8779,
+      "step": 254
+    },
+    {
+      "epoch": 0.5985915492957746,
+      "grad_norm": 1.75752592086792,
+      "learning_rate": 8.399168883231564e-05,
+      "loss": 2.5419,
+      "step": 255
+    },
+    {
+      "epoch": 0.6009389671361502,
+      "grad_norm": 1.707080602645874,
+      "learning_rate": 8.38490387032966e-05,
+      "loss": 2.4909,
+      "step": 256
+    },
+    {
+      "epoch": 0.6032863849765259,
+      "grad_norm": 1.6730165481567383,
+      "learning_rate": 8.37058781298162e-05,
+      "loss": 2.677,
+      "step": 257
+    },
+    {
+      "epoch": 0.6056338028169014,
+      "grad_norm": 1.7518807649612427,
+      "learning_rate": 8.35622092707399e-05,
+      "loss": 2.6762,
+      "step": 258
+    },
+    {
+      "epoch": 0.607981220657277,
+      "grad_norm": 1.681357502937317,
+      "learning_rate": 8.341803429259817e-05,
+      "loss": 2.4185,
+      "step": 259
+    },
+    {
+      "epoch": 0.6103286384976526,
+      "grad_norm": 1.6799571514129639,
+      "learning_rate": 8.327335536955376e-05,
+      "loss": 2.5536,
+      "step": 260
+    },
+    {
+      "epoch": 0.6126760563380281,
+      "grad_norm": 1.6633497476577759,
+      "learning_rate": 8.31281746833689e-05,
+      "loss": 2.462,
+      "step": 261
+    },
+    {
+      "epoch": 0.6150234741784038,
+      "grad_norm": 1.6778583526611328,
+      "learning_rate": 8.29824944233725e-05,
+      "loss": 2.612,
+      "step": 262
+    },
+    {
+      "epoch": 0.6173708920187794,
+      "grad_norm": 1.8236974477767944,
+      "learning_rate": 8.283631678642694e-05,
+      "loss": 2.5978,
+      "step": 263
+    },
+    {
+      "epoch": 0.6197183098591549,
+      "grad_norm": 1.7540674209594727,
+      "learning_rate": 8.268964397689516e-05,
+      "loss": 2.3425,
+      "step": 264
+    },
+    {
+      "epoch": 0.6220657276995305,
+      "grad_norm": 1.6756924390792847,
+      "learning_rate": 8.254247820660727e-05,
+      "loss": 2.4991,
+      "step": 265
+    },
+    {
+      "epoch": 0.6244131455399061,
+      "grad_norm": 1.7354806661605835,
+      "learning_rate": 8.239482169482726e-05,
+      "loss": 2.6997,
+      "step": 266
+    },
+    {
+      "epoch": 0.6267605633802817,
+      "grad_norm": 1.5830391645431519,
+      "learning_rate": 8.22466766682195e-05,
+      "loss": 2.2771,
+      "step": 267
+    },
+    {
+      "epoch": 0.6291079812206573,
+      "grad_norm": 1.716260313987732,
+      "learning_rate": 8.209804536081516e-05,
+      "loss": 2.8291,
+      "step": 268
+    },
+    {
+      "epoch": 0.6314553990610329,
+      "grad_norm": 1.8733055591583252,
+      "learning_rate": 8.194893001397858e-05,
+      "loss": 2.658,
+      "step": 269
+    },
+    {
+      "epoch": 0.6338028169014085,
+      "grad_norm": 1.7264173030853271,
+      "learning_rate": 8.179933287637342e-05,
+      "loss": 2.7606,
+      "step": 270
+    },
+    {
+      "epoch": 0.636150234741784,
+      "grad_norm": 1.7096571922302246,
+      "learning_rate": 8.164925620392871e-05,
+      "loss": 2.6032,
+      "step": 271
+    },
+    {
+      "epoch": 0.6384976525821596,
+      "grad_norm": 1.757120132446289,
+      "learning_rate": 8.149870225980498e-05,
+      "loss": 2.743,
+      "step": 272
+    },
+    {
+      "epoch": 0.6408450704225352,
+      "grad_norm": 1.8931641578674316,
+      "learning_rate": 8.134767331435991e-05,
+      "loss": 2.452,
+      "step": 273
+    },
+    {
+      "epoch": 0.6431924882629108,
+      "grad_norm": 1.86612069606781,
+      "learning_rate": 8.11961716451143e-05,
+      "loss": 2.4612,
+      "step": 274
+    },
+    {
+      "epoch": 0.6455399061032864,
+      "grad_norm": 1.8044425249099731,
+      "learning_rate": 8.104419953671759e-05,
+      "loss": 2.521,
+      "step": 275
+    },
+    {
+      "epoch": 0.647887323943662,
+      "grad_norm": 1.7253297567367554,
+      "learning_rate": 8.089175928091349e-05,
+      "loss": 2.7162,
+      "step": 276
+    },
+    {
+      "epoch": 0.6502347417840375,
+      "grad_norm": 1.8354557752609253,
+      "learning_rate": 8.073885317650534e-05,
+      "loss": 2.8202,
+      "step": 277
+    },
+    {
+      "epoch": 0.6525821596244131,
+      "grad_norm": 1.8952256441116333,
+      "learning_rate": 8.058548352932158e-05,
+      "loss": 2.5724,
+      "step": 278
+    },
+    {
+      "epoch": 0.6549295774647887,
+      "grad_norm": 1.7274194955825806,
+      "learning_rate": 8.043165265218078e-05,
+      "loss": 2.4746,
+      "step": 279
+    },
+    {
+      "epoch": 0.6572769953051644,
+      "grad_norm": 1.65982186794281,
+      "learning_rate": 8.027736286485694e-05,
+      "loss": 2.6111,
+      "step": 280
+    },
+    {
+      "epoch": 0.6596244131455399,
+      "grad_norm": 1.7060494422912598,
+      "learning_rate": 8.01226164940444e-05,
+      "loss": 2.6162,
+      "step": 281
+    },
+    {
+      "epoch": 0.6619718309859155,
+      "grad_norm": 1.6539195775985718,
+      "learning_rate": 7.996741587332284e-05,
+      "loss": 2.6578,
+      "step": 282
+    },
+    {
+      "epoch": 0.6643192488262911,
+      "grad_norm": 1.7097866535186768,
+      "learning_rate": 7.981176334312199e-05,
+      "loss": 2.5812,
+      "step": 283
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 1.577872633934021,
+      "learning_rate": 7.965566125068642e-05,
+      "loss": 2.3894,
+      "step": 284
+    },
+    {
+      "epoch": 0.6690140845070423,
+      "grad_norm": 1.8474918603897095,
+      "learning_rate": 7.949911195004008e-05,
+      "loss": 2.7251,
+      "step": 285
+    },
+    {
+      "epoch": 0.6713615023474179,
+      "grad_norm": 1.7533137798309326,
+      "learning_rate": 7.934211780195089e-05,
+      "loss": 2.5598,
+      "step": 286
+    },
+    {
+      "epoch": 0.6737089201877934,
+      "grad_norm": 1.8090660572052002,
+      "learning_rate": 7.9184681173895e-05,
+      "loss": 2.4811,
+      "step": 287
+    },
+    {
+      "epoch": 0.676056338028169,
+      "grad_norm": 1.7923107147216797,
+      "learning_rate": 7.902680444002127e-05,
+      "loss": 2.6725,
+      "step": 288
+    },
+    {
+      "epoch": 0.6784037558685446,
+      "grad_norm": 1.5660858154296875,
+      "learning_rate": 7.886848998111526e-05,
+      "loss": 2.3519,
+      "step": 289
+    },
+    {
+      "epoch": 0.6807511737089202,
+      "grad_norm": 1.6969128847122192,
+      "learning_rate": 7.870974018456352e-05,
+      "loss": 2.4519,
+      "step": 290
+    },
+    {
+      "epoch": 0.6830985915492958,
+      "grad_norm": 1.591829776763916,
+      "learning_rate": 7.855055744431747e-05,
+      "loss": 2.5443,
+      "step": 291
+    },
+    {
+      "epoch": 0.6854460093896714,
+      "grad_norm": 3.7944729328155518,
+      "learning_rate": 7.839094416085733e-05,
+      "loss": 2.4167,
+      "step": 292
+    },
+    {
+      "epoch": 0.687793427230047,
+      "grad_norm": 2.0311732292175293,
+      "learning_rate": 7.823090274115592e-05,
+      "loss": 2.7071,
+      "step": 293
+    },
+    {
+      "epoch": 0.6901408450704225,
+      "grad_norm": 1.7271850109100342,
+      "learning_rate": 7.807043559864242e-05,
+      "loss": 2.4388,
+      "step": 294
+    },
+    {
+      "epoch": 0.6924882629107981,
+      "grad_norm": 1.7520400285720825,
+      "learning_rate": 7.790954515316584e-05,
+      "loss": 2.5614,
+      "step": 295
+    },
+    {
+      "epoch": 0.6948356807511737,
+      "grad_norm": 1.612693428993225,
+      "learning_rate": 7.774823383095867e-05,
+      "loss": 2.3016,
+      "step": 296
+    },
+    {
+      "epoch": 0.6971830985915493,
+      "grad_norm": 1.7074772119522095,
+      "learning_rate": 7.75865040646002e-05,
+      "loss": 2.4481,
+      "step": 297
+    },
+    {
+      "epoch": 0.6995305164319249,
+      "grad_norm": 1.695130467414856,
+      "learning_rate": 7.742435829297988e-05,
+      "loss": 2.6779,
+      "step": 298
+    },
+    {
+      "epoch": 0.7018779342723005,
+      "grad_norm": 1.6866894960403442,
+      "learning_rate": 7.726179896126055e-05,
+      "loss": 2.435,
+      "step": 299
+    },
+    {
+      "epoch": 0.704225352112676,
+      "grad_norm": 1.6453986167907715,
+      "learning_rate": 7.709882852084152e-05,
+      "loss": 2.4564,
+      "step": 300
+    },
+    {
+      "epoch": 0.7065727699530516,
+      "grad_norm": 1.6527467966079712,
+      "learning_rate": 7.693544942932162e-05,
+      "loss": 2.3505,
+      "step": 301
+    },
+    {
+      "epoch": 0.7089201877934272,
+      "grad_norm": 1.831211805343628,
+      "learning_rate": 7.677166415046225e-05,
+      "loss": 2.617,
+      "step": 302
+    },
+    {
+      "epoch": 0.7112676056338029,
+      "grad_norm": 1.693515419960022,
+      "learning_rate": 7.660747515414996e-05,
+      "loss": 2.4229,
+      "step": 303
+    },
+    {
+      "epoch": 0.7136150234741784,
+      "grad_norm": 1.7261874675750732,
+      "learning_rate": 7.644288491635952e-05,
+      "loss": 2.4348,
+      "step": 304
+    },
+    {
+      "epoch": 0.715962441314554,
+      "grad_norm": 1.6822891235351562,
+      "learning_rate": 7.627789591911634e-05,
+      "loss": 2.4353,
+      "step": 305
+    },
+    {
+      "epoch": 0.7183098591549296,
+      "grad_norm": 1.7033827304840088,
+      "learning_rate": 7.611251065045918e-05,
+      "loss": 2.554,
+      "step": 306
+    },
+    {
+      "epoch": 0.7206572769953051,
+      "grad_norm": 1.7407300472259521,
+      "learning_rate": 7.594673160440258e-05,
+      "loss": 2.4696,
+      "step": 307
+    },
+    {
+      "epoch": 0.7230046948356808,
+      "grad_norm": 1.7128525972366333,
+      "learning_rate": 7.578056128089921e-05,
+      "loss": 2.534,
+      "step": 308
+    },
+    {
+      "epoch": 0.7253521126760564,
+      "grad_norm": 1.6343917846679688,
+      "learning_rate": 7.56140021858023e-05,
+      "loss": 2.2916,
+      "step": 309
+    },
+    {
+      "epoch": 0.7276995305164319,
+      "grad_norm": 1.6431267261505127,
+      "learning_rate": 7.544705683082768e-05,
+      "loss": 2.5699,
+      "step": 310
+    },
+    {
+      "epoch": 0.7300469483568075,
+      "grad_norm": 1.5325164794921875,
+      "learning_rate": 7.527972773351604e-05,
+      "loss": 2.4804,
+      "step": 311
+    },
+    {
+      "epoch": 0.7323943661971831,
+      "grad_norm": 1.787161111831665,
+      "learning_rate": 7.511201741719489e-05,
+      "loss": 2.6212,
+      "step": 312
+    },
+    {
+      "epoch": 0.7347417840375586,
+      "grad_norm": 1.818595051765442,
+      "learning_rate": 7.494392841094058e-05,
+      "loss": 2.3339,
+      "step": 313
+    },
+    {
+      "epoch": 0.7370892018779343,
+      "grad_norm": 1.7084897756576538,
+      "learning_rate": 7.477546324954002e-05,
+      "loss": 2.5509,
+      "step": 314
+    },
+    {
+      "epoch": 0.7394366197183099,
+      "grad_norm": 1.7448676824569702,
+      "learning_rate": 7.460662447345265e-05,
+      "loss": 2.6375,
+      "step": 315
+    },
+    {
+      "epoch": 0.7417840375586855,
+      "grad_norm": 1.6353211402893066,
+      "learning_rate": 7.443741462877202e-05,
+      "loss": 2.3909,
+      "step": 316
+    },
+    {
+      "epoch": 0.744131455399061,
+      "grad_norm": 1.6458829641342163,
+      "learning_rate": 7.426783626718732e-05,
+      "loss": 2.3643,
+      "step": 317
+    },
+    {
+      "epoch": 0.7464788732394366,
+      "grad_norm": 1.7494622468948364,
+      "learning_rate": 7.409789194594507e-05,
+      "loss": 2.495,
+      "step": 318
+    },
+    {
+      "epoch": 0.7488262910798122,
+      "grad_norm": 1.7642425298690796,
+      "learning_rate": 7.392758422781044e-05,
+      "loss": 2.3718,
+      "step": 319
+    },
+    {
+      "epoch": 0.7511737089201878,
+      "grad_norm": 1.7770228385925293,
+      "learning_rate": 7.375691568102863e-05,
+      "loss": 2.7252,
+      "step": 320
+    },
+    {
+      "epoch": 0.7535211267605634,
+      "grad_norm": 1.855891227722168,
+      "learning_rate": 7.358588887928614e-05,
+      "loss": 2.6202,
+      "step": 321
+    },
+    {
+      "epoch": 0.755868544600939,
+      "grad_norm": 1.6031696796417236,
+      "learning_rate": 7.341450640167202e-05,
+      "loss": 2.462,
+      "step": 322
+    },
+    {
+      "epoch": 0.7582159624413145,
+      "grad_norm": 1.6653591394424438,
+      "learning_rate": 7.324277083263885e-05,
+      "loss": 2.3727,
+      "step": 323
+    },
+    {
+      "epoch": 0.7605633802816901,
+      "grad_norm": 1.6277172565460205,
+      "learning_rate": 7.307068476196388e-05,
+      "loss": 2.4018,
+      "step": 324
+    },
+    {
+      "epoch": 0.7629107981220657,
+      "grad_norm": 2.210129737854004,
+      "learning_rate": 7.289825078470993e-05,
+      "loss": 2.6157,
+      "step": 325
+    },
+    {
+      "epoch": 0.7652582159624414,
+      "grad_norm": 1.7617254257202148,
+      "learning_rate": 7.272547150118624e-05,
+      "loss": 2.3518,
+      "step": 326
+    },
+    {
+      "epoch": 0.7676056338028169,
+      "grad_norm": 1.6546744108200073,
+      "learning_rate": 7.255234951690932e-05,
+      "loss": 2.5607,
+      "step": 327
+    },
+    {
+      "epoch": 0.7699530516431925,
+      "grad_norm": 1.8078160285949707,
+      "learning_rate": 7.237888744256357e-05,
+      "loss": 2.5907,
+      "step": 328
+    },
+    {
+      "epoch": 0.7723004694835681,
+      "grad_norm": 1.6897518634796143,
+      "learning_rate": 7.220508789396197e-05,
+      "loss": 2.6059,
+      "step": 329
+    },
+    {
+      "epoch": 0.7746478873239436,
+      "grad_norm": 1.715383529663086,
+      "learning_rate": 7.203095349200666e-05,
+      "loss": 2.6292,
+      "step": 330
+    },
+    {
+      "epoch": 0.7769953051643192,
+      "grad_norm": 1.796817660331726,
+      "learning_rate": 7.185648686264934e-05,
+      "loss": 2.3609,
+      "step": 331
+    },
+    {
+      "epoch": 0.7793427230046949,
+      "grad_norm": 1.8774749040603638,
+      "learning_rate": 7.168169063685171e-05,
+      "loss": 2.5167,
+      "step": 332
+    },
+    {
+      "epoch": 0.7816901408450704,
+      "grad_norm": 1.6688238382339478,
+      "learning_rate": 7.15065674505458e-05,
+      "loss": 2.422,
+      "step": 333
+    },
+    {
+      "epoch": 0.784037558685446,
+      "grad_norm": 1.6255617141723633,
+      "learning_rate": 7.13311199445942e-05,
+      "loss": 2.587,
+      "step": 334
+    },
+    {
+      "epoch": 0.7863849765258216,
+      "grad_norm": 1.6177692413330078,
+      "learning_rate": 7.115535076475031e-05,
+      "loss": 2.4323,
+      "step": 335
+    },
+    {
+      "epoch": 0.7887323943661971,
+      "grad_norm": 1.6124238967895508,
+      "learning_rate": 7.09792625616183e-05,
+      "loss": 2.3881,
+      "step": 336
+    },
+    {
+      "epoch": 0.7910798122065728,
+      "grad_norm": 1.635842204093933,
+      "learning_rate": 7.080285799061324e-05,
+      "loss": 2.51,
+      "step": 337
+    },
+    {
+      "epoch": 0.7934272300469484,
+      "grad_norm": 1.5977977514266968,
+      "learning_rate": 7.06261397119211e-05,
+      "loss": 2.5427,
+      "step": 338
+    },
+    {
+      "epoch": 0.795774647887324,
+      "grad_norm": 1.6340340375900269,
+      "learning_rate": 7.044911039045847e-05,
+      "loss": 2.4539,
+      "step": 339
+    },
+    {
+      "epoch": 0.7981220657276995,
+      "grad_norm": 1.6684269905090332,
+      "learning_rate": 7.027177269583256e-05,
+      "loss": 2.5914,
+      "step": 340
+    },
+    {
+      "epoch": 0.8004694835680751,
+      "grad_norm": 1.658740758895874,
+      "learning_rate": 7.009412930230084e-05,
+      "loss": 2.4745,
+      "step": 341
+    },
+    {
+      "epoch": 0.8028169014084507,
+      "grad_norm": 1.762073040008545,
+      "learning_rate": 6.991618288873066e-05,
+      "loss": 2.4735,
+      "step": 342
+    },
+    {
+      "epoch": 0.8051643192488263,
+      "grad_norm": 1.6312123537063599,
+      "learning_rate": 6.973793613855902e-05,
+      "loss": 2.2814,
+      "step": 343
+    },
+    {
+      "epoch": 0.8075117370892019,
+      "grad_norm": 1.651703953742981,
+      "learning_rate": 6.955939173975191e-05,
+      "loss": 2.4313,
+      "step": 344
+    },
+    {
+      "epoch": 0.8098591549295775,
+      "grad_norm": 1.6794708967208862,
+      "learning_rate": 6.938055238476395e-05,
+      "loss": 2.5693,
+      "step": 345
+    },
+    {
+      "epoch": 0.812206572769953,
+      "grad_norm": 1.622802972793579,
+      "learning_rate": 6.920142077049766e-05,
+      "loss": 2.3973,
+      "step": 346
+    },
+    {
+      "epoch": 0.8145539906103286,
+      "grad_norm": 1.5529680252075195,
+      "learning_rate": 6.902199959826286e-05,
+      "loss": 2.3006,
+      "step": 347
+    },
+    {
+      "epoch": 0.8169014084507042,
+      "grad_norm": 1.6821764707565308,
+      "learning_rate": 6.88422915737359e-05,
+      "loss": 2.3309,
+      "step": 348
+    },
+    {
+      "epoch": 0.8192488262910798,
+      "grad_norm": 1.6601792573928833,
+      "learning_rate": 6.866229940691888e-05,
+      "loss": 2.5736,
+      "step": 349
+    },
+    {
+      "epoch": 0.8215962441314554,
+      "grad_norm": 1.6162347793579102,
+      "learning_rate": 6.848202581209875e-05,
+      "loss": 2.3876,
+      "step": 350
+    },
+    {
+      "epoch": 0.823943661971831,
+      "grad_norm": 1.687666654586792,
+      "learning_rate": 6.830147350780645e-05,
+      "loss": 2.4175,
+      "step": 351
+    },
+    {
+      "epoch": 0.8262910798122066,
+      "grad_norm": 1.7783905267715454,
+      "learning_rate": 6.812064521677579e-05,
+      "loss": 2.5448,
+      "step": 352
+    },
+    {
+      "epoch": 0.8286384976525821,
+      "grad_norm": 1.5999135971069336,
+      "learning_rate": 6.793954366590257e-05,
+      "loss": 2.3465,
+      "step": 353
+    },
+    {
+      "epoch": 0.8309859154929577,
+      "grad_norm": 1.5779527425765991,
+      "learning_rate": 6.775817158620328e-05,
+      "loss": 2.336,
+      "step": 354
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 1.6518279314041138,
+      "learning_rate": 6.7576531712774e-05,
+      "loss": 2.5975,
+      "step": 355
+    },
+    {
+      "epoch": 0.8356807511737089,
+      "grad_norm": 1.6484471559524536,
+      "learning_rate": 6.739462678474917e-05,
+      "loss": 2.3712,
+      "step": 356
+    },
+    {
+      "epoch": 0.8380281690140845,
+      "grad_norm": 1.6995164155960083,
+      "learning_rate": 6.721245954526025e-05,
+      "loss": 2.3961,
+      "step": 357
+    },
+    {
+      "epoch": 0.8403755868544601,
+      "grad_norm": 1.620734691619873,
+      "learning_rate": 6.703003274139438e-05,
+      "loss": 2.6909,
+      "step": 358
+    },
+    {
+      "epoch": 0.8427230046948356,
+      "grad_norm": 1.750163197517395,
+      "learning_rate": 6.684734912415289e-05,
+      "loss": 2.554,
+      "step": 359
+    },
+    {
+      "epoch": 0.8450704225352113,
+      "grad_norm": 1.6951026916503906,
+      "learning_rate": 6.666441144840994e-05,
+      "loss": 2.4729,
+      "step": 360
+    },
+    {
+      "epoch": 0.8474178403755869,
+      "grad_norm": 1.6454946994781494,
+      "learning_rate": 6.648122247287083e-05,
+      "loss": 2.441,
+      "step": 361
+    },
+    {
+      "epoch": 0.8497652582159625,
+      "grad_norm": 1.6438918113708496,
+      "learning_rate": 6.62977849600305e-05,
+      "loss": 2.4567,
+      "step": 362
+    },
+    {
+      "epoch": 0.852112676056338,
+      "grad_norm": 1.6957948207855225,
+      "learning_rate": 6.611410167613184e-05,
+      "loss": 2.4575,
+      "step": 363
+    },
+    {
+      "epoch": 0.8544600938967136,
+      "grad_norm": 1.7275480031967163,
+      "learning_rate": 6.593017539112396e-05,
+      "loss": 2.4869,
+      "step": 364
+    },
+    {
+      "epoch": 0.8568075117370892,
+      "grad_norm": 1.6698849201202393,
+      "learning_rate": 6.574600887862043e-05,
+      "loss": 2.539,
+      "step": 365
+    },
+    {
+      "epoch": 0.8591549295774648,
+      "grad_norm": 1.5995240211486816,
+      "learning_rate": 6.55616049158575e-05,
+      "loss": 2.3934,
+      "step": 366
+    },
+    {
+      "epoch": 0.8615023474178404,
+      "grad_norm": 1.645573377609253,
+      "learning_rate": 6.537696628365212e-05,
+      "loss": 2.4846,
+      "step": 367
+    },
+    {
+      "epoch": 0.863849765258216,
+      "grad_norm": 1.4631596803665161,
+      "learning_rate": 6.51920957663601e-05,
+      "loss": 2.2284,
+      "step": 368
+    },
+    {
+      "epoch": 0.8661971830985915,
+      "grad_norm": 1.6029276847839355,
+      "learning_rate": 6.500699615183408e-05,
+      "loss": 2.2829,
+      "step": 369
+    },
+    {
+      "epoch": 0.8685446009389671,
+      "grad_norm": 1.6571834087371826,
+      "learning_rate": 6.482167023138147e-05,
+      "loss": 2.3553,
+      "step": 370
+    },
+    {
+      "epoch": 0.8708920187793427,
+      "grad_norm": 1.7219182252883911,
+      "learning_rate": 6.46361207997224e-05,
+      "loss": 2.6597,
+      "step": 371
+    },
+    {
+      "epoch": 0.8732394366197183,
+      "grad_norm": 1.5665297508239746,
+      "learning_rate": 6.445035065494754e-05,
+      "loss": 2.4517,
+      "step": 372
+    },
+    {
+      "epoch": 0.8755868544600939,
+      "grad_norm": 1.6524207592010498,
+      "learning_rate": 6.426436259847595e-05,
+      "loss": 2.5584,
+      "step": 373
+    },
+    {
+      "epoch": 0.8779342723004695,
+      "grad_norm": 1.702607274055481,
+      "learning_rate": 6.407815943501274e-05,
+      "loss": 2.6585,
+      "step": 374
+    },
+    {
+      "epoch": 0.8802816901408451,
+      "grad_norm": 1.6633670330047607,
+      "learning_rate": 6.389174397250693e-05,
+      "loss": 2.8056,
+      "step": 375
+    },
+    {
+      "epoch": 0.8826291079812206,
+      "grad_norm": 1.5368313789367676,
+      "learning_rate": 6.370511902210897e-05,
+      "loss": 2.3103,
+      "step": 376
+    },
+    {
+      "epoch": 0.8849765258215962,
+      "grad_norm": 1.7158435583114624,
+      "learning_rate": 6.351828739812836e-05,
+      "loss": 2.569,
+      "step": 377
+    },
+    {
+      "epoch": 0.8873239436619719,
+      "grad_norm": 1.6899515390396118,
+      "learning_rate": 6.33312519179913e-05,
+      "loss": 2.4678,
+      "step": 378
+    },
+    {
+      "epoch": 0.8896713615023474,
+      "grad_norm": 1.768155574798584,
+      "learning_rate": 6.31440154021981e-05,
+      "loss": 2.4346,
+      "step": 379
+    },
+    {
+      "epoch": 0.892018779342723,
+      "grad_norm": 1.5976459980010986,
+      "learning_rate": 6.295658067428077e-05,
+      "loss": 2.5652,
+      "step": 380
+    },
+    {
+      "epoch": 0.8943661971830986,
+      "grad_norm": 1.6183762550354004,
+      "learning_rate": 6.276895056076022e-05,
+      "loss": 2.6315,
+      "step": 381
+    },
+    {
+      "epoch": 0.8967136150234741,
+      "grad_norm": 1.6042938232421875,
+      "learning_rate": 6.258112789110395e-05,
+      "loss": 2.342,
+      "step": 382
+    },
+    {
+      "epoch": 0.8990610328638498,
+      "grad_norm": 1.5818735361099243,
+      "learning_rate": 6.239311549768311e-05,
+      "loss": 2.3361,
+      "step": 383
+    },
+    {
+      "epoch": 0.9014084507042254,
+      "grad_norm": 1.6001895666122437,
+      "learning_rate": 6.220491621572989e-05,
+      "loss": 2.3697,
+      "step": 384
+    },
+    {
+      "epoch": 0.903755868544601,
+      "grad_norm": 1.6264625787734985,
+      "learning_rate": 6.20165328832948e-05,
+      "loss": 2.404,
+      "step": 385
+    },
+    {
+      "epoch": 0.9061032863849765,
+      "grad_norm": 1.625516653060913,
+      "learning_rate": 6.182796834120386e-05,
+      "loss": 2.4486,
+      "step": 386
+    },
+    {
+      "epoch": 0.9084507042253521,
+      "grad_norm": 1.7253628969192505,
+      "learning_rate": 6.163922543301565e-05,
+      "loss": 2.5588,
+      "step": 387
+    },
+    {
+      "epoch": 0.9107981220657277,
+      "grad_norm": 1.6850779056549072,
+      "learning_rate": 6.145030700497857e-05,
+      "loss": 2.4637,
+      "step": 388
+    },
+    {
+      "epoch": 0.9131455399061033,
+      "grad_norm": 1.6117552518844604,
+      "learning_rate": 6.126121590598788e-05,
+      "loss": 2.4271,
+      "step": 389
+    },
+    {
+      "epoch": 0.9154929577464789,
+      "grad_norm": 1.672454595565796,
+      "learning_rate": 6.10719549875427e-05,
+      "loss": 2.5023,
+      "step": 390
+    },
+    {
+      "epoch": 0.9178403755868545,
+      "grad_norm": 1.688650369644165,
+      "learning_rate": 6.088252710370302e-05,
+      "loss": 2.4107,
+      "step": 391
+    },
+    {
+      "epoch": 0.92018779342723,
+      "grad_norm": 1.7508164644241333,
+      "learning_rate": 6.069293511104672e-05,
+      "loss": 2.3532,
+      "step": 392
+    },
+    {
+      "epoch": 0.9225352112676056,
+      "grad_norm": 1.6823396682739258,
+      "learning_rate": 6.0503181868626394e-05,
+      "loss": 2.4022,
+      "step": 393
+    },
+    {
+      "epoch": 0.9248826291079812,
+      "grad_norm": 1.5990012884140015,
+      "learning_rate": 6.031327023792629e-05,
+      "loss": 2.4995,
+      "step": 394
+    },
+    {
+      "epoch": 0.9272300469483568,
+      "grad_norm": 1.6524854898452759,
+      "learning_rate": 6.012320308281919e-05,
+      "loss": 2.5553,
+      "step": 395
+    },
+    {
+      "epoch": 0.9295774647887324,
+      "grad_norm": 1.564630150794983,
+      "learning_rate": 5.993298326952318e-05,
+      "loss": 2.2477,
+      "step": 396
+    },
+    {
+      "epoch": 0.931924882629108,
+      "grad_norm": 1.6353230476379395,
+      "learning_rate": 5.974261366655841e-05,
+      "loss": 2.3449,
+      "step": 397
+    },
+    {
+      "epoch": 0.9342723004694836,
+      "grad_norm": 1.6802188158035278,
+      "learning_rate": 5.955209714470388e-05,
+      "loss": 2.4921,
+      "step": 398
+    },
+    {
+      "epoch": 0.9366197183098591,
+      "grad_norm": 1.585205316543579,
+      "learning_rate": 5.9361436576954157e-05,
+      "loss": 2.5039,
+      "step": 399
+    },
+    {
+      "epoch": 0.9389671361502347,
+      "grad_norm": 1.6235421895980835,
+      "learning_rate": 5.9170634838475955e-05,
+      "loss": 2.2194,
+      "step": 400
+    },
+    {
+      "epoch": 0.9413145539906104,
+      "grad_norm": 1.5911391973495483,
+      "learning_rate": 5.897969480656491e-05,
+      "loss": 2.2783,
+      "step": 401
+    },
+    {
+      "epoch": 0.9436619718309859,
+      "grad_norm": 1.5980224609375,
+      "learning_rate": 5.8788619360602126e-05,
+      "loss": 2.4084,
+      "step": 402
+    },
+    {
+      "epoch": 0.9460093896713615,
+      "grad_norm": 1.7125698328018188,
+      "learning_rate": 5.859741138201068e-05,
+      "loss": 2.4377,
+      "step": 403
+    },
+    {
+      "epoch": 0.9483568075117371,
+      "grad_norm": 1.7280176877975464,
+      "learning_rate": 5.8406073754212355e-05,
+      "loss": 2.6577,
+      "step": 404
+    },
+    {
+      "epoch": 0.9507042253521126,
+      "grad_norm": 1.5822030305862427,
+      "learning_rate": 5.8214609362583974e-05,
+      "loss": 2.4425,
+      "step": 405
+    },
+    {
+      "epoch": 0.9530516431924883,
+      "grad_norm": 1.6493258476257324,
+      "learning_rate": 5.8023021094413966e-05,
+      "loss": 2.463,
+      "step": 406
+    },
+    {
+      "epoch": 0.9553990610328639,
+      "grad_norm": 1.613416075706482,
+      "learning_rate": 5.783131183885885e-05,
+      "loss": 2.5065,
+      "step": 407
+    },
+    {
+      "epoch": 0.9577464788732394,
+      "grad_norm": 1.654819130897522,
+      "learning_rate": 5.763948448689963e-05,
+      "loss": 2.4145,
+      "step": 408
+    },
+    {
+      "epoch": 0.960093896713615,
+      "grad_norm": 1.8096052408218384,
+      "learning_rate": 5.744754193129818e-05,
+      "loss": 2.5207,
+      "step": 409
+    },
+    {
+      "epoch": 0.9624413145539906,
+      "grad_norm": 1.52267587184906,
+      "learning_rate": 5.725548706655368e-05,
+      "loss": 2.3283,
+      "step": 410
+    },
+    {
+      "epoch": 0.9647887323943662,
+      "grad_norm": 1.817744255065918,
+      "learning_rate": 5.706332278885893e-05,
+      "loss": 2.408,
+      "step": 411
+    },
+    {
+      "epoch": 0.9671361502347418,
+      "grad_norm": 1.618360161781311,
+      "learning_rate": 5.687105199605667e-05,
+      "loss": 2.5073,
+      "step": 412
+    },
+    {
+      "epoch": 0.9694835680751174,
+      "grad_norm": 1.5666344165802002,
+      "learning_rate": 5.6678677587595884e-05,
+      "loss": 2.2989,
+      "step": 413
+    },
+    {
+      "epoch": 0.971830985915493,
+      "grad_norm": 1.7902641296386719,
+      "learning_rate": 5.648620246448813e-05,
+      "loss": 2.5295,
+      "step": 414
+    },
+    {
+      "epoch": 0.9741784037558685,
+      "grad_norm": 1.5790274143218994,
+      "learning_rate": 5.629362952926367e-05,
+      "loss": 2.2716,
+      "step": 415
+    },
+    {
+      "epoch": 0.9765258215962441,
+      "grad_norm": 1.5773351192474365,
+      "learning_rate": 5.610096168592785e-05,
+      "loss": 2.3217,
+      "step": 416
+    },
+    {
+      "epoch": 0.9788732394366197,
+      "grad_norm": 1.6175018548965454,
+      "learning_rate": 5.590820183991716e-05,
+      "loss": 2.4215,
+      "step": 417
+    },
+    {
+      "epoch": 0.9812206572769953,
+      "grad_norm": 1.585377812385559,
+      "learning_rate": 5.571535289805556e-05,
+      "loss": 2.2939,
+      "step": 418
+    },
+    {
+      "epoch": 0.9835680751173709,
+      "grad_norm": 1.6029826402664185,
+      "learning_rate": 5.552241776851055e-05,
+      "loss": 2.2161,
+      "step": 419
+    },
+    {
+      "epoch": 0.9859154929577465,
+      "grad_norm": 1.702242136001587,
+      "learning_rate": 5.5329399360749336e-05,
+      "loss": 2.8479,
+      "step": 420
+    },
+    {
+      "epoch": 0.9882629107981221,
+      "grad_norm": 1.665329098701477,
+      "learning_rate": 5.513630058549497e-05,
+      "loss": 2.4748,
+      "step": 421
+    },
+    {
+      "epoch": 0.9906103286384976,
+      "grad_norm": 1.5718384981155396,
+      "learning_rate": 5.494312435468244e-05,
+      "loss": 2.2289,
+      "step": 422
+    },
+    {
+      "epoch": 0.9929577464788732,
+      "grad_norm": 1.5208048820495605,
+      "learning_rate": 5.474987358141478e-05,
+      "loss": 2.3309,
+      "step": 423
+    },
+    {
+      "epoch": 0.9953051643192489,
+      "grad_norm": 1.6006495952606201,
+      "learning_rate": 5.4556551179919134e-05,
+      "loss": 2.3086,
+      "step": 424
+    },
+    {
+      "epoch": 0.9976525821596244,
+      "grad_norm": 1.530830979347229,
+      "learning_rate": 5.436316006550275e-05,
+      "loss": 2.3678,
+      "step": 425
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.515878677368164,
+      "learning_rate": 5.416970315450911e-05,
+      "loss": 2.2671,
+      "step": 426
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 852,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 426,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.196187718110413e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-426/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:944bede9af88780161cfb3481d76d2d8d340944f9c23201e7523cf705bc0a218
+size 5432

checkpoint-852/config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-crh_Latn-coli-focus/",
+  "architectures": [
+    "Olmo2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "eos_token_id": 100257,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 4096,
+  "model_type": "olmo2",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 100277,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 500000,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.48.0.dev0",
+  "use_cache": true,
+  "vocab_size": 100384
+}

checkpoint-852/generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 100257,
+  "pad_token_id": 100277,
+  "transformers_version": "4.48.0.dev0"
+}

checkpoint-852/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bcbd595872bbb202af539c7ecf11f1f2305034418bd7108a853d75db6701c5e0
+size 4950070472

checkpoint-852/model-00002-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de0b6a357d03676c4986588aaf4dc96afc3f44a0498514961d2b2cfa1940655e
+size 4991622816

checkpoint-852/model-00003-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1038a7d346a681a8fb6949e054c2246f3f3c80e826855c1cec6af1155cc607d
+size 4924546896

checkpoint-852/model-00004-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33c0a21146ace02d36928787ae7f34be868f7c81b894dea98ec7b0e5a3ea01b2
+size 4857404928

checkpoint-852/model-00005-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e7f82a3014ccfc0f67208f63785ae4bc71f09aba86db48bf6139d18178abbfed
+size 4857404928

checkpoint-852/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:298913408be33fd839412a7946b1ab6299505f21e620ebfe610d4f91bd89f620
+size 4614509080

checkpoint-852/model.safetensors.index.json

@@ -0,0 +1,362 @@
+{
+  "metadata": {
+    "total_size": 29195517952
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00006-of-00006.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.k_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.q_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.k_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.q_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.k_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.q_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.k_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.q_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.k_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.q_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.k_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.q_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.k_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.q_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.norm.weight": "model-00006-of-00006.safetensors"
+  }
+}

checkpoint-852/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4f3e8182227a861feefd1f5db625edc3e95cbc3268593e5af433b605e0597f0
+size 13055369358

checkpoint-852/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ff264f99d31b522cc7e2a4eac9d38606d0c58a34c0adc74d71e0ca8b371dc36
+size 14244

checkpoint-852/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a602b0b1e5b019b056d033f0ec99aa1437e1ef95082d76692a2f275f004f383c
+size 1064

checkpoint-852/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:944bede9af88780161cfb3481d76d2d8d340944f9c23201e7523cf705bc0a218
+size 5432

config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-crh_Latn-coli-focus/",
+  "architectures": [
+    "Olmo2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "eos_token_id": 100257,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 4096,
+  "model_type": "olmo2",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 100277,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 500000,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.48.0.dev0",
+  "use_cache": true,
+  "vocab_size": 100384
+}

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 100257,
+  "pad_token_id": 100277,
+  "transformers_version": "4.48.0.dev0"
+}

model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bcbd595872bbb202af539c7ecf11f1f2305034418bd7108a853d75db6701c5e0
+size 4950070472

model-00002-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de0b6a357d03676c4986588aaf4dc96afc3f44a0498514961d2b2cfa1940655e
+size 4991622816

model-00003-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1038a7d346a681a8fb6949e054c2246f3f3c80e826855c1cec6af1155cc607d
+size 4924546896

model-00004-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33c0a21146ace02d36928787ae7f34be868f7c81b894dea98ec7b0e5a3ea01b2
+size 4857404928

model-00005-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e7f82a3014ccfc0f67208f63785ae4bc71f09aba86db48bf6139d18178abbfed
+size 4857404928

model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:298913408be33fd839412a7946b1ab6299505f21e620ebfe610d4f91bd89f620
+size 4614509080

model.safetensors.index.json

@@ -0,0 +1,362 @@
+{
+  "metadata": {
+    "total_size": 29195517952
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00006-of-00006.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.k_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.q_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.k_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.q_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.16.self_attn.k_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.q_norm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.k_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.q_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.22.self_attn.k_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.q_norm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.28.self_attn.k_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.q_norm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.k_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.q_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.k_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.q_norm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.k_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.q_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.k_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.q_norm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.k_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.q_norm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.norm.weight": "model-00006-of-00006.safetensors"
+  }
+}

special_tokens_map.json

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

tokenizer_config.json

@@ -0,0 +1,189 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "100256": {
+      "content": "<|extra_id_0|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100257": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100258": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100259": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100260": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100261": {
+      "content": "|||PHONE_NUMBER|||",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100262": {
+      "content": "|||EMAIL_ADDRESS|||",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100263": {
+      "content": "|||IP_ADDRESS|||",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100264": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100265": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100266": {
+      "content": "<|extra_id_1|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100267": {
+      "content": "<|extra_id_2|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100268": {
+      "content": "<|extra_id_3|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100269": {
+      "content": "<|extra_id_4|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100270": {
+      "content": "<|extra_id_5|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100271": {
+      "content": "<|extra_id_6|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100272": {
+      "content": "<|extra_id_7|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100273": {
+      "content": "<|extra_id_8|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100274": {
+      "content": "<|extra_id_9|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100275": {
+      "content": "<|extra_id_10|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100276": {
+      "content": "<|endofprompt|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100277": {
+      "content": "<|pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<|endoftext|>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<|pad|>",
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>"
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:944bede9af88780161cfb3481d76d2d8d340944f9c23201e7523cf705bc0a218
+size 5432