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-rand/",
+  "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:fc711515946152e994d9d5404a98a58b962cb7697920613bd948d1cd8aeb6c83
+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 sha256:29da673c4905e7a45e667d98cc6dba2590666ebd0f7c106cd7dc47513b1ab207
+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:e08c52b589ff216ffb2d4ff9edd3004bcd5b5844be080057d21ea7caf220bc92
+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": 6.74153470993042,
+      "learning_rate": 2.325581395348837e-06,
+      "loss": 8.0682,
+      "step": 1
+    },
+    {
+      "epoch": 0.004694835680751174,
+      "grad_norm": 5.902026176452637,
+      "learning_rate": 4.651162790697674e-06,
+      "loss": 7.5085,
+      "step": 2
+    },
+    {
+      "epoch": 0.007042253521126761,
+      "grad_norm": 7.1663970947265625,
+      "learning_rate": 6.976744186046512e-06,
+      "loss": 8.4092,
+      "step": 3
+    },
+    {
+      "epoch": 0.009389671361502348,
+      "grad_norm": 4.929966449737549,
+      "learning_rate": 9.302325581395349e-06,
+      "loss": 6.8045,
+      "step": 4
+    },
+    {
+      "epoch": 0.011737089201877934,
+      "grad_norm": 5.453227996826172,
+      "learning_rate": 1.1627906976744187e-05,
+      "loss": 7.5007,
+      "step": 5
+    },
+    {
+      "epoch": 0.014084507042253521,
+      "grad_norm": 5.242559909820557,
+      "learning_rate": 1.3953488372093024e-05,
+      "loss": 7.1794,
+      "step": 6
+    },
+    {
+      "epoch": 0.01643192488262911,
+      "grad_norm": 5.786001682281494,
+      "learning_rate": 1.6279069767441862e-05,
+      "loss": 7.5557,
+      "step": 7
+    },
+    {
+      "epoch": 0.018779342723004695,
+      "grad_norm": 37.475582122802734,
+      "learning_rate": 1.8604651162790697e-05,
+      "loss": 6.7967,
+      "step": 8
+    },
+    {
+      "epoch": 0.02112676056338028,
+      "grad_norm": 8.487391471862793,
+      "learning_rate": 2.0930232558139536e-05,
+      "loss": 6.8769,
+      "step": 9
+    },
+    {
+      "epoch": 0.023474178403755867,
+      "grad_norm": 7.1259026527404785,
+      "learning_rate": 2.3255813953488374e-05,
+      "loss": 6.6679,
+      "step": 10
+    },
+    {
+      "epoch": 0.025821596244131457,
+      "grad_norm": 6.672307014465332,
+      "learning_rate": 2.5581395348837212e-05,
+      "loss": 6.381,
+      "step": 11
+    },
+    {
+      "epoch": 0.028169014084507043,
+      "grad_norm": 13.447758674621582,
+      "learning_rate": 2.7906976744186048e-05,
+      "loss": 6.1116,
+      "step": 12
+    },
+    {
+      "epoch": 0.03051643192488263,
+      "grad_norm": 5.074902534484863,
+      "learning_rate": 3.0232558139534883e-05,
+      "loss": 5.7065,
+      "step": 13
+    },
+    {
+      "epoch": 0.03286384976525822,
+      "grad_norm": 5.343297481536865,
+      "learning_rate": 3.2558139534883724e-05,
+      "loss": 6.07,
+      "step": 14
+    },
+    {
+      "epoch": 0.035211267605633804,
+      "grad_norm": 4.750032901763916,
+      "learning_rate": 3.488372093023256e-05,
+      "loss": 5.7021,
+      "step": 15
+    },
+    {
+      "epoch": 0.03755868544600939,
+      "grad_norm": 4.497730731964111,
+      "learning_rate": 3.7209302325581394e-05,
+      "loss": 5.2804,
+      "step": 16
+    },
+    {
+      "epoch": 0.03990610328638498,
+      "grad_norm": 6.505649566650391,
+      "learning_rate": 3.953488372093023e-05,
+      "loss": 5.7704,
+      "step": 17
+    },
+    {
+      "epoch": 0.04225352112676056,
+      "grad_norm": 3.8866803646087646,
+      "learning_rate": 4.186046511627907e-05,
+      "loss": 5.3131,
+      "step": 18
+    },
+    {
+      "epoch": 0.04460093896713615,
+      "grad_norm": 4.154186248779297,
+      "learning_rate": 4.418604651162791e-05,
+      "loss": 5.2177,
+      "step": 19
+    },
+    {
+      "epoch": 0.046948356807511735,
+      "grad_norm": 3.877696990966797,
+      "learning_rate": 4.651162790697675e-05,
+      "loss": 5.0364,
+      "step": 20
+    },
+    {
+      "epoch": 0.04929577464788732,
+      "grad_norm": 4.082431793212891,
+      "learning_rate": 4.883720930232558e-05,
+      "loss": 5.0723,
+      "step": 21
+    },
+    {
+      "epoch": 0.051643192488262914,
+      "grad_norm": 4.014376163482666,
+      "learning_rate": 5.1162790697674425e-05,
+      "loss": 5.0463,
+      "step": 22
+    },
+    {
+      "epoch": 0.0539906103286385,
+      "grad_norm": 3.7503790855407715,
+      "learning_rate": 5.348837209302326e-05,
+      "loss": 4.8761,
+      "step": 23
+    },
+    {
+      "epoch": 0.056338028169014086,
+      "grad_norm": 3.149980306625366,
+      "learning_rate": 5.5813953488372095e-05,
+      "loss": 4.5262,
+      "step": 24
+    },
+    {
+      "epoch": 0.05868544600938967,
+      "grad_norm": 3.164466619491577,
+      "learning_rate": 5.8139534883720933e-05,
+      "loss": 4.4134,
+      "step": 25
+    },
+    {
+      "epoch": 0.06103286384976526,
+      "grad_norm": 3.258887767791748,
+      "learning_rate": 6.0465116279069765e-05,
+      "loss": 4.4837,
+      "step": 26
+    },
+    {
+      "epoch": 0.06338028169014084,
+      "grad_norm": 2.9795877933502197,
+      "learning_rate": 6.27906976744186e-05,
+      "loss": 4.268,
+      "step": 27
+    },
+    {
+      "epoch": 0.06572769953051644,
+      "grad_norm": 3.418090581893921,
+      "learning_rate": 6.511627906976745e-05,
+      "loss": 4.1638,
+      "step": 28
+    },
+    {
+      "epoch": 0.06807511737089202,
+      "grad_norm": 3.0284407138824463,
+      "learning_rate": 6.744186046511628e-05,
+      "loss": 4.1912,
+      "step": 29
+    },
+    {
+      "epoch": 0.07042253521126761,
+      "grad_norm": 2.4764463901519775,
+      "learning_rate": 6.976744186046513e-05,
+      "loss": 3.7192,
+      "step": 30
+    },
+    {
+      "epoch": 0.07276995305164319,
+      "grad_norm": 3.433175563812256,
+      "learning_rate": 7.209302325581396e-05,
+      "loss": 4.0516,
+      "step": 31
+    },
+    {
+      "epoch": 0.07511737089201878,
+      "grad_norm": 3.6180222034454346,
+      "learning_rate": 7.441860465116279e-05,
+      "loss": 4.1871,
+      "step": 32
+    },
+    {
+      "epoch": 0.07746478873239436,
+      "grad_norm": 9.763015747070312,
+      "learning_rate": 7.674418604651163e-05,
+      "loss": 3.9995,
+      "step": 33
+    },
+    {
+      "epoch": 0.07981220657276995,
+      "grad_norm": 3.139212131500244,
+      "learning_rate": 7.906976744186047e-05,
+      "loss": 3.7272,
+      "step": 34
+    },
+    {
+      "epoch": 0.08215962441314555,
+      "grad_norm": 2.6827597618103027,
+      "learning_rate": 8.139534883720931e-05,
+      "loss": 3.751,
+      "step": 35
+    },
+    {
+      "epoch": 0.08450704225352113,
+      "grad_norm": 2.9310972690582275,
+      "learning_rate": 8.372093023255814e-05,
+      "loss": 4.0055,
+      "step": 36
+    },
+    {
+      "epoch": 0.08685446009389672,
+      "grad_norm": 2.362100839614868,
+      "learning_rate": 8.604651162790697e-05,
+      "loss": 3.6492,
+      "step": 37
+    },
+    {
+      "epoch": 0.0892018779342723,
+      "grad_norm": 2.7767703533172607,
+      "learning_rate": 8.837209302325582e-05,
+      "loss": 3.8801,
+      "step": 38
+    },
+    {
+      "epoch": 0.09154929577464789,
+      "grad_norm": 2.6525282859802246,
+      "learning_rate": 9.069767441860465e-05,
+      "loss": 3.5582,
+      "step": 39
+    },
+    {
+      "epoch": 0.09389671361502347,
+      "grad_norm": 2.6659605503082275,
+      "learning_rate": 9.30232558139535e-05,
+      "loss": 3.7146,
+      "step": 40
+    },
+    {
+      "epoch": 0.09624413145539906,
+      "grad_norm": 2.707975149154663,
+      "learning_rate": 9.534883720930233e-05,
+      "loss": 3.59,
+      "step": 41
+    },
+    {
+      "epoch": 0.09859154929577464,
+      "grad_norm": 2.599092721939087,
+      "learning_rate": 9.767441860465116e-05,
+      "loss": 3.4901,
+      "step": 42
+    },
+    {
+      "epoch": 0.10093896713615023,
+      "grad_norm": 2.611809492111206,
+      "learning_rate": 0.0001,
+      "loss": 3.7515,
+      "step": 43
+    },
+    {
+      "epoch": 0.10328638497652583,
+      "grad_norm": 2.4812469482421875,
+      "learning_rate": 9.999962299929265e-05,
+      "loss": 3.6521,
+      "step": 44
+    },
+    {
+      "epoch": 0.1056338028169014,
+      "grad_norm": 2.396311044692993,
+      "learning_rate": 9.999849200285573e-05,
+      "loss": 3.6281,
+      "step": 45
+    },
+    {
+      "epoch": 0.107981220657277,
+      "grad_norm": 2.4779536724090576,
+      "learning_rate": 9.999660702774474e-05,
+      "loss": 3.5034,
+      "step": 46
+    },
+    {
+      "epoch": 0.11032863849765258,
+      "grad_norm": 2.3948028087615967,
+      "learning_rate": 9.999396810238513e-05,
+      "loss": 3.5579,
+      "step": 47
+    },
+    {
+      "epoch": 0.11267605633802817,
+      "grad_norm": 2.401021957397461,
+      "learning_rate": 9.999057526657196e-05,
+      "loss": 3.4793,
+      "step": 48
+    },
+    {
+      "epoch": 0.11502347417840375,
+      "grad_norm": 2.6258273124694824,
+      "learning_rate": 9.998642857146934e-05,
+      "loss": 3.4992,
+      "step": 49
+    },
+    {
+      "epoch": 0.11737089201877934,
+      "grad_norm": 4.11713171005249,
+      "learning_rate": 9.99815280796095e-05,
+      "loss": 3.2179,
+      "step": 50
+    },
+    {
+      "epoch": 0.11971830985915492,
+      "grad_norm": 2.6189980506896973,
+      "learning_rate": 9.997587386489202e-05,
+      "loss": 3.4057,
+      "step": 51
+    },
+    {
+      "epoch": 0.12206572769953052,
+      "grad_norm": 3.7957706451416016,
+      "learning_rate": 9.996946601258261e-05,
+      "loss": 3.509,
+      "step": 52
+    },
+    {
+      "epoch": 0.12441314553990611,
+      "grad_norm": 2.797846555709839,
+      "learning_rate": 9.996230461931186e-05,
+      "loss": 3.1791,
+      "step": 53
+    },
+    {
+      "epoch": 0.1267605633802817,
+      "grad_norm": 3.2530624866485596,
+      "learning_rate": 9.995438979307381e-05,
+      "loss": 3.3071,
+      "step": 54
+    },
+    {
+      "epoch": 0.12910798122065728,
+      "grad_norm": 2.044215679168701,
+      "learning_rate": 9.994572165322422e-05,
+      "loss": 3.3685,
+      "step": 55
+    },
+    {
+      "epoch": 0.13145539906103287,
+      "grad_norm": 2.2636239528656006,
+      "learning_rate": 9.993630033047891e-05,
+      "loss": 3.4271,
+      "step": 56
+    },
+    {
+      "epoch": 0.13380281690140844,
+      "grad_norm": 4.127092361450195,
+      "learning_rate": 9.992612596691171e-05,
+      "loss": 3.3618,
+      "step": 57
+    },
+    {
+      "epoch": 0.13615023474178403,
+      "grad_norm": 2.1330583095550537,
+      "learning_rate": 9.991519871595227e-05,
+      "loss": 3.1341,
+      "step": 58
+    },
+    {
+      "epoch": 0.13849765258215962,
+      "grad_norm": 2.255793571472168,
+      "learning_rate": 9.990351874238388e-05,
+      "loss": 3.1382,
+      "step": 59
+    },
+    {
+      "epoch": 0.14084507042253522,
+      "grad_norm": 2.0289318561553955,
+      "learning_rate": 9.989108622234084e-05,
+      "loss": 3.0725,
+      "step": 60
+    },
+    {
+      "epoch": 0.1431924882629108,
+      "grad_norm": 2.0822529792785645,
+      "learning_rate": 9.987790134330593e-05,
+      "loss": 3.1886,
+      "step": 61
+    },
+    {
+      "epoch": 0.14553990610328638,
+      "grad_norm": 2.202343702316284,
+      "learning_rate": 9.986396430410749e-05,
+      "loss": 3.1895,
+      "step": 62
+    },
+    {
+      "epoch": 0.14788732394366197,
+      "grad_norm": 2.0780410766601562,
+      "learning_rate": 9.984927531491648e-05,
+      "loss": 3.2399,
+      "step": 63
+    },
+    {
+      "epoch": 0.15023474178403756,
+      "grad_norm": 1.9594764709472656,
+      "learning_rate": 9.983383459724322e-05,
+      "loss": 3.194,
+      "step": 64
+    },
+    {
+      "epoch": 0.15258215962441316,
+      "grad_norm": 2.6781210899353027,
+      "learning_rate": 9.981764238393424e-05,
+      "loss": 3.3434,
+      "step": 65
+    },
+    {
+      "epoch": 0.15492957746478872,
+      "grad_norm": 1.913921594619751,
+      "learning_rate": 9.980069891916854e-05,
+      "loss": 2.9046,
+      "step": 66
+    },
+    {
+      "epoch": 0.1572769953051643,
+      "grad_norm": 2.063204050064087,
+      "learning_rate": 9.978300445845404e-05,
+      "loss": 3.1134,
+      "step": 67
+    },
+    {
+      "epoch": 0.1596244131455399,
+      "grad_norm": 2.063624620437622,
+      "learning_rate": 9.976455926862371e-05,
+      "loss": 3.1757,
+      "step": 68
+    },
+    {
+      "epoch": 0.1619718309859155,
+      "grad_norm": 2.1056790351867676,
+      "learning_rate": 9.974536362783156e-05,
+      "loss": 3.1839,
+      "step": 69
+    },
+    {
+      "epoch": 0.1643192488262911,
+      "grad_norm": 2.1855363845825195,
+      "learning_rate": 9.972541782554836e-05,
+      "loss": 3.1744,
+      "step": 70
+    },
+    {
+      "epoch": 0.16666666666666666,
+      "grad_norm": 2.163306951522827,
+      "learning_rate": 9.970472216255741e-05,
+      "loss": 3.0721,
+      "step": 71
+    },
+    {
+      "epoch": 0.16901408450704225,
+      "grad_norm": 2.0321338176727295,
+      "learning_rate": 9.968327695094987e-05,
+      "loss": 3.2354,
+      "step": 72
+    },
+    {
+      "epoch": 0.17136150234741784,
+      "grad_norm": 2.473968505859375,
+      "learning_rate": 9.966108251412014e-05,
+      "loss": 3.265,
+      "step": 73
+    },
+    {
+      "epoch": 0.17370892018779344,
+      "grad_norm": 2.1940367221832275,
+      "learning_rate": 9.963813918676095e-05,
+      "loss": 3.2263,
+      "step": 74
+    },
+    {
+      "epoch": 0.176056338028169,
+      "grad_norm": 2.01611590385437,
+      "learning_rate": 9.961444731485836e-05,
+      "loss": 3.0679,
+      "step": 75
+    },
+    {
+      "epoch": 0.1784037558685446,
+      "grad_norm": 2.064678430557251,
+      "learning_rate": 9.959000725568641e-05,
+      "loss": 3.0248,
+      "step": 76
+    },
+    {
+      "epoch": 0.1807511737089202,
+      "grad_norm": 2.0593502521514893,
+      "learning_rate": 9.956481937780193e-05,
+      "loss": 3.1276,
+      "step": 77
+    },
+    {
+      "epoch": 0.18309859154929578,
+      "grad_norm": 2.0508029460906982,
+      "learning_rate": 9.953888406103883e-05,
+      "loss": 2.9601,
+      "step": 78
+    },
+    {
+      "epoch": 0.18544600938967137,
+      "grad_norm": 2.0320143699645996,
+      "learning_rate": 9.951220169650239e-05,
+      "loss": 3.2927,
+      "step": 79
+    },
+    {
+      "epoch": 0.18779342723004694,
+      "grad_norm": 2.048593759536743,
+      "learning_rate": 9.948477268656345e-05,
+      "loss": 2.9041,
+      "step": 80
+    },
+    {
+      "epoch": 0.19014084507042253,
+      "grad_norm": 1.9130347967147827,
+      "learning_rate": 9.945659744485225e-05,
+      "loss": 2.846,
+      "step": 81
+    },
+    {
+      "epoch": 0.19248826291079812,
+      "grad_norm": 1.9843111038208008,
+      "learning_rate": 9.942767639625224e-05,
+      "loss": 2.925,
+      "step": 82
+    },
+    {
+      "epoch": 0.19483568075117372,
+      "grad_norm": 1.976904034614563,
+      "learning_rate": 9.939800997689362e-05,
+      "loss": 3.3067,
+      "step": 83
+    },
+    {
+      "epoch": 0.19718309859154928,
+      "grad_norm": 1.9135102033615112,
+      "learning_rate": 9.936759863414685e-05,
+      "loss": 3.0173,
+      "step": 84
+    },
+    {
+      "epoch": 0.19953051643192488,
+      "grad_norm": 1.8903708457946777,
+      "learning_rate": 9.933644282661586e-05,
+      "loss": 3.0035,
+      "step": 85
+    },
+    {
+      "epoch": 0.20187793427230047,
+      "grad_norm": 2.104926109313965,
+      "learning_rate": 9.930454302413108e-05,
+      "loss": 3.1789,
+      "step": 86
+    },
+    {
+      "epoch": 0.20422535211267606,
+      "grad_norm": 2.187955379486084,
+      "learning_rate": 9.927189970774245e-05,
+      "loss": 3.0865,
+      "step": 87
+    },
+    {
+      "epoch": 0.20657276995305165,
+      "grad_norm": 1.9850472211837769,
+      "learning_rate": 9.923851336971208e-05,
+      "loss": 3.0692,
+      "step": 88
+    },
+    {
+      "epoch": 0.20892018779342722,
+      "grad_norm": 2.022155284881592,
+      "learning_rate": 9.920438451350694e-05,
+      "loss": 3.0953,
+      "step": 89
+    },
+    {
+      "epoch": 0.2112676056338028,
+      "grad_norm": 2.039721727371216,
+      "learning_rate": 9.916951365379112e-05,
+      "loss": 3.1641,
+      "step": 90
+    },
+    {
+      "epoch": 0.2136150234741784,
+      "grad_norm": 1.8413914442062378,
+      "learning_rate": 9.913390131641815e-05,
+      "loss": 2.9263,
+      "step": 91
+    },
+    {
+      "epoch": 0.215962441314554,
+      "grad_norm": 3.865889310836792,
+      "learning_rate": 9.909754803842313e-05,
+      "loss": 2.888,
+      "step": 92
+    },
+    {
+      "epoch": 0.21830985915492956,
+      "grad_norm": 2.0847761631011963,
+      "learning_rate": 9.906045436801448e-05,
+      "loss": 3.0763,
+      "step": 93
+    },
+    {
+      "epoch": 0.22065727699530516,
+      "grad_norm": 1.9045860767364502,
+      "learning_rate": 9.902262086456582e-05,
+      "loss": 2.9922,
+      "step": 94
+    },
+    {
+      "epoch": 0.22300469483568075,
+      "grad_norm": 2.059549331665039,
+      "learning_rate": 9.898404809860744e-05,
+      "loss": 2.9984,
+      "step": 95
+    },
+    {
+      "epoch": 0.22535211267605634,
+      "grad_norm": 1.974882960319519,
+      "learning_rate": 9.894473665181776e-05,
+      "loss": 3.0983,
+      "step": 96
+    },
+    {
+      "epoch": 0.22769953051643194,
+      "grad_norm": 1.9072291851043701,
+      "learning_rate": 9.89046871170145e-05,
+      "loss": 3.011,
+      "step": 97
+    },
+    {
+      "epoch": 0.2300469483568075,
+      "grad_norm": 2.395671844482422,
+      "learning_rate": 9.886390009814579e-05,
+      "loss": 2.8887,
+      "step": 98
+    },
+    {
+      "epoch": 0.2323943661971831,
+      "grad_norm": 1.9982821941375732,
+      "learning_rate": 9.882237621028101e-05,
+      "loss": 3.0971,
+      "step": 99
+    },
+    {
+      "epoch": 0.2347417840375587,
+      "grad_norm": 1.750945806503296,
+      "learning_rate": 9.878011607960156e-05,
+      "loss": 2.9084,
+      "step": 100
+    },
+    {
+      "epoch": 0.23708920187793428,
+      "grad_norm": 1.9544695615768433,
+      "learning_rate": 9.873712034339143e-05,
+      "loss": 3.0739,
+      "step": 101
+    },
+    {
+      "epoch": 0.23943661971830985,
+      "grad_norm": 1.9211771488189697,
+      "learning_rate": 9.869338965002752e-05,
+      "loss": 2.8189,
+      "step": 102
+    },
+    {
+      "epoch": 0.24178403755868544,
+      "grad_norm": 1.9226988554000854,
+      "learning_rate": 9.864892465896994e-05,
+      "loss": 2.8616,
+      "step": 103
+    },
+    {
+      "epoch": 0.24413145539906103,
+      "grad_norm": 1.9747202396392822,
+      "learning_rate": 9.860372604075199e-05,
+      "loss": 3.0266,
+      "step": 104
+    },
+    {
+      "epoch": 0.24647887323943662,
+      "grad_norm": 1.8386709690093994,
+      "learning_rate": 9.855779447697013e-05,
+      "loss": 2.7453,
+      "step": 105
+    },
+    {
+      "epoch": 0.24882629107981222,
+      "grad_norm": 1.9151238203048706,
+      "learning_rate": 9.851113066027364e-05,
+      "loss": 2.8376,
+      "step": 106
+    },
+    {
+      "epoch": 0.2511737089201878,
+      "grad_norm": 2.1199045181274414,
+      "learning_rate": 9.846373529435418e-05,
+      "loss": 3.0903,
+      "step": 107
+    },
+    {
+      "epoch": 0.2535211267605634,
+      "grad_norm": 1.742322325706482,
+      "learning_rate": 9.841560909393523e-05,
+      "loss": 2.7989,
+      "step": 108
+    },
+    {
+      "epoch": 0.25586854460093894,
+      "grad_norm": 1.8751287460327148,
+      "learning_rate": 9.836675278476124e-05,
+      "loss": 2.9556,
+      "step": 109
+    },
+    {
+      "epoch": 0.25821596244131456,
+      "grad_norm": 1.8210289478302002,
+      "learning_rate": 9.831716710358673e-05,
+      "loss": 2.7869,
+      "step": 110
+    },
+    {
+      "epoch": 0.2605633802816901,
+      "grad_norm": 1.9199378490447998,
+      "learning_rate": 9.82668527981652e-05,
+      "loss": 2.8579,
+      "step": 111
+    },
+    {
+      "epoch": 0.26291079812206575,
+      "grad_norm": 2.19356107711792,
+      "learning_rate": 9.821581062723779e-05,
+      "loss": 2.8818,
+      "step": 112
+    },
+    {
+      "epoch": 0.2652582159624413,
+      "grad_norm": 1.755565881729126,
+      "learning_rate": 9.816404136052186e-05,
+      "loss": 2.7002,
+      "step": 113
+    },
+    {
+      "epoch": 0.2676056338028169,
+      "grad_norm": 2.039726972579956,
+      "learning_rate": 9.811154577869943e-05,
+      "loss": 2.9508,
+      "step": 114
+    },
+    {
+      "epoch": 0.2699530516431925,
+      "grad_norm": 1.824286699295044,
+      "learning_rate": 9.805832467340538e-05,
+      "loss": 2.8587,
+      "step": 115
+    },
+    {
+      "epoch": 0.27230046948356806,
+      "grad_norm": 1.9687395095825195,
+      "learning_rate": 9.800437884721545e-05,
+      "loss": 2.9892,
+      "step": 116
+    },
+    {
+      "epoch": 0.2746478873239437,
+      "grad_norm": 1.9100291728973389,
+      "learning_rate": 9.794970911363426e-05,
+      "loss": 2.9608,
+      "step": 117
+    },
+    {
+      "epoch": 0.27699530516431925,
+      "grad_norm": 1.8510876893997192,
+      "learning_rate": 9.78943162970829e-05,
+      "loss": 2.8168,
+      "step": 118
+    },
+    {
+      "epoch": 0.2793427230046948,
+      "grad_norm": 3.4659063816070557,
+      "learning_rate": 9.783820123288664e-05,
+      "loss": 2.8401,
+      "step": 119
+    },
+    {
+      "epoch": 0.28169014084507044,
+      "grad_norm": 1.894242286682129,
+      "learning_rate": 9.778136476726223e-05,
+      "loss": 2.7963,
+      "step": 120
+    },
+    {
+      "epoch": 0.284037558685446,
+      "grad_norm": 2.108842372894287,
+      "learning_rate": 9.772380775730516e-05,
+      "loss": 2.8951,
+      "step": 121
+    },
+    {
+      "epoch": 0.2863849765258216,
+      "grad_norm": 1.8829090595245361,
+      "learning_rate": 9.766553107097681e-05,
+      "loss": 3.0315,
+      "step": 122
+    },
+    {
+      "epoch": 0.2887323943661972,
+      "grad_norm": 1.8636614084243774,
+      "learning_rate": 9.760653558709122e-05,
+      "loss": 2.979,
+      "step": 123
+    },
+    {
+      "epoch": 0.29107981220657275,
+      "grad_norm": 1.8715219497680664,
+      "learning_rate": 9.754682219530199e-05,
+      "loss": 2.9428,
+      "step": 124
+    },
+    {
+      "epoch": 0.2934272300469484,
+      "grad_norm": 1.785217046737671,
+      "learning_rate": 9.748639179608872e-05,
+      "loss": 2.6377,
+      "step": 125
+    },
+    {
+      "epoch": 0.29577464788732394,
+      "grad_norm": 2.0240108966827393,
+      "learning_rate": 9.742524530074355e-05,
+      "loss": 2.7979,
+      "step": 126
+    },
+    {
+      "epoch": 0.2981220657276995,
+      "grad_norm": 2.643691062927246,
+      "learning_rate": 9.736338363135738e-05,
+      "loss": 2.9472,
+      "step": 127
+    },
+    {
+      "epoch": 0.3004694835680751,
+      "grad_norm": 1.8176496028900146,
+      "learning_rate": 9.730080772080592e-05,
+      "loss": 2.8626,
+      "step": 128
+    },
+    {
+      "epoch": 0.3028169014084507,
+      "grad_norm": 2.218360662460327,
+      "learning_rate": 9.723751851273568e-05,
+      "loss": 2.8979,
+      "step": 129
+    },
+    {
+      "epoch": 0.3051643192488263,
+      "grad_norm": 1.8391715288162231,
+      "learning_rate": 9.71735169615497e-05,
+      "loss": 2.8412,
+      "step": 130
+    },
+    {
+      "epoch": 0.3075117370892019,
+      "grad_norm": 2.497711420059204,
+      "learning_rate": 9.710880403239317e-05,
+      "loss": 2.7567,
+      "step": 131
+    },
+    {
+      "epoch": 0.30985915492957744,
+      "grad_norm": 1.859116554260254,
+      "learning_rate": 9.704338070113893e-05,
+      "loss": 2.8,
+      "step": 132
+    },
+    {
+      "epoch": 0.31220657276995306,
+      "grad_norm": 1.7876489162445068,
+      "learning_rate": 9.697724795437264e-05,
+      "loss": 2.7589,
+      "step": 133
+    },
+    {
+      "epoch": 0.3145539906103286,
+      "grad_norm": 1.6835530996322632,
+      "learning_rate": 9.6910406789378e-05,
+      "loss": 2.6982,
+      "step": 134
+    },
+    {
+      "epoch": 0.31690140845070425,
+      "grad_norm": 1.7438956499099731,
+      "learning_rate": 9.684285821412165e-05,
+      "loss": 2.7689,
+      "step": 135
+    },
+    {
+      "epoch": 0.3192488262910798,
+      "grad_norm": 1.9094929695129395,
+      "learning_rate": 9.677460324723806e-05,
+      "loss": 2.8394,
+      "step": 136
+    },
+    {
+      "epoch": 0.3215962441314554,
+      "grad_norm": 1.8689284324645996,
+      "learning_rate": 9.670564291801401e-05,
+      "loss": 2.7924,
+      "step": 137
+    },
+    {
+      "epoch": 0.323943661971831,
+      "grad_norm": 1.8969767093658447,
+      "learning_rate": 9.663597826637325e-05,
+      "loss": 2.8311,
+      "step": 138
+    },
+    {
+      "epoch": 0.32629107981220656,
+      "grad_norm": 1.758320689201355,
+      "learning_rate": 9.656561034286069e-05,
+      "loss": 2.769,
+      "step": 139
+    },
+    {
+      "epoch": 0.3286384976525822,
+      "grad_norm": 1.7114989757537842,
+      "learning_rate": 9.64945402086266e-05,
+      "loss": 2.7606,
+      "step": 140
+    },
+    {
+      "epoch": 0.33098591549295775,
+      "grad_norm": 1.708143711090088,
+      "learning_rate": 9.642276893541063e-05,
+      "loss": 2.6031,
+      "step": 141
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 1.8093912601470947,
+      "learning_rate": 9.63502976055256e-05,
+      "loss": 2.6895,
+      "step": 142
+    },
+    {
+      "epoch": 0.33568075117370894,
+      "grad_norm": 2.6389997005462646,
+      "learning_rate": 9.627712731184122e-05,
+      "loss": 2.7398,
+      "step": 143
+    },
+    {
+      "epoch": 0.3380281690140845,
+      "grad_norm": 2.378230094909668,
+      "learning_rate": 9.620325915776758e-05,
+      "loss": 2.6429,
+      "step": 144
+    },
+    {
+      "epoch": 0.3403755868544601,
+      "grad_norm": 1.9288902282714844,
+      "learning_rate": 9.612869425723854e-05,
+      "loss": 3.024,
+      "step": 145
+    },
+    {
+      "epoch": 0.3427230046948357,
+      "grad_norm": 1.8169496059417725,
+      "learning_rate": 9.605343373469491e-05,
+      "loss": 2.7254,
+      "step": 146
+    },
+    {
+      "epoch": 0.34507042253521125,
+      "grad_norm": 1.8046923875808716,
+      "learning_rate": 9.59774787250675e-05,
+      "loss": 2.7404,
+      "step": 147
+    },
+    {
+      "epoch": 0.3474178403755869,
+      "grad_norm": 1.9261105060577393,
+      "learning_rate": 9.590083037376001e-05,
+      "loss": 2.856,
+      "step": 148
+    },
+    {
+      "epoch": 0.34976525821596244,
+      "grad_norm": 1.845190405845642,
+      "learning_rate": 9.582348983663173e-05,
+      "loss": 2.8376,
+      "step": 149
+    },
+    {
+      "epoch": 0.352112676056338,
+      "grad_norm": 2.6601450443267822,
+      "learning_rate": 9.574545827998017e-05,
+      "loss": 2.7516,
+      "step": 150
+    },
+    {
+      "epoch": 0.3544600938967136,
+      "grad_norm": 1.7026342153549194,
+      "learning_rate": 9.566673688052339e-05,
+      "loss": 2.5917,
+      "step": 151
+    },
+    {
+      "epoch": 0.3568075117370892,
+      "grad_norm": 1.6950021982192993,
+      "learning_rate": 9.558732682538233e-05,
+      "loss": 2.6247,
+      "step": 152
+    },
+    {
+      "epoch": 0.3591549295774648,
+      "grad_norm": 2.047011613845825,
+      "learning_rate": 9.550722931206286e-05,
+      "loss": 2.8301,
+      "step": 153
+    },
+    {
+      "epoch": 0.3615023474178404,
+      "grad_norm": 1.7364177703857422,
+      "learning_rate": 9.542644554843778e-05,
+      "loss": 2.6797,
+      "step": 154
+    },
+    {
+      "epoch": 0.36384976525821594,
+      "grad_norm": 1.6910269260406494,
+      "learning_rate": 9.53449767527285e-05,
+      "loss": 2.6877,
+      "step": 155
+    },
+    {
+      "epoch": 0.36619718309859156,
+      "grad_norm": 1.8064078092575073,
+      "learning_rate": 9.526282415348677e-05,
+      "loss": 2.6628,
+      "step": 156
+    },
+    {
+      "epoch": 0.3685446009389671,
+      "grad_norm": 49.98039245605469,
+      "learning_rate": 9.517998898957611e-05,
+      "loss": 2.7553,
+      "step": 157
+    },
+    {
+      "epoch": 0.37089201877934275,
+      "grad_norm": 1.855872392654419,
+      "learning_rate": 9.509647251015314e-05,
+      "loss": 2.8635,
+      "step": 158
+    },
+    {
+      "epoch": 0.3732394366197183,
+      "grad_norm": 1.932474970817566,
+      "learning_rate": 9.501227597464875e-05,
+      "loss": 2.868,
+      "step": 159
+    },
+    {
+      "epoch": 0.3755868544600939,
+      "grad_norm": 1.7808077335357666,
+      "learning_rate": 9.492740065274904e-05,
+      "loss": 2.5917,
+      "step": 160
+    },
+    {
+      "epoch": 0.3779342723004695,
+      "grad_norm": 1.772858738899231,
+      "learning_rate": 9.484184782437628e-05,
+      "loss": 2.6197,
+      "step": 161
+    },
+    {
+      "epoch": 0.38028169014084506,
+      "grad_norm": 2.6756508350372314,
+      "learning_rate": 9.475561877966956e-05,
+      "loss": 2.6527,
+      "step": 162
+    },
+    {
+      "epoch": 0.3826291079812207,
+      "grad_norm": 1.6639829874038696,
+      "learning_rate": 9.46687148189653e-05,
+      "loss": 2.633,
+      "step": 163
+    },
+    {
+      "epoch": 0.38497652582159625,
+      "grad_norm": 1.8140541315078735,
+      "learning_rate": 9.458113725277769e-05,
+      "loss": 2.7801,
+      "step": 164
+    },
+    {
+      "epoch": 0.3873239436619718,
+      "grad_norm": 11.151857376098633,
+      "learning_rate": 9.449288740177891e-05,
+      "loss": 2.7566,
+      "step": 165
+    },
+    {
+      "epoch": 0.38967136150234744,
+      "grad_norm": 2.033475637435913,
+      "learning_rate": 9.440396659677919e-05,
+      "loss": 2.6477,
+      "step": 166
+    },
+    {
+      "epoch": 0.392018779342723,
+      "grad_norm": 2.4777884483337402,
+      "learning_rate": 9.43143761787068e-05,
+      "loss": 3.0192,
+      "step": 167
+    },
+    {
+      "epoch": 0.39436619718309857,
+      "grad_norm": 1.8000035285949707,
+      "learning_rate": 9.422411749858779e-05,
+      "loss": 2.8657,
+      "step": 168
+    },
+    {
+      "epoch": 0.3967136150234742,
+      "grad_norm": 1.6467283964157104,
+      "learning_rate": 9.413319191752559e-05,
+      "loss": 2.6207,
+      "step": 169
+    },
+    {
+      "epoch": 0.39906103286384975,
+      "grad_norm": 1.7085107564926147,
+      "learning_rate": 9.404160080668055e-05,
+      "loss": 2.8527,
+      "step": 170
+    },
+    {
+      "epoch": 0.4014084507042254,
+      "grad_norm": 1.721285104751587,
+      "learning_rate": 9.394934554724921e-05,
+      "loss": 2.7701,
+      "step": 171
+    },
+    {
+      "epoch": 0.40375586854460094,
+      "grad_norm": 1.7729564905166626,
+      "learning_rate": 9.385642753044348e-05,
+      "loss": 2.738,
+      "step": 172
+    },
+    {
+      "epoch": 0.4061032863849765,
+      "grad_norm": 1.7724652290344238,
+      "learning_rate": 9.37628481574697e-05,
+      "loss": 2.8383,
+      "step": 173
+    },
+    {
+      "epoch": 0.4084507042253521,
+      "grad_norm": 1.6301813125610352,
+      "learning_rate": 9.366860883950745e-05,
+      "loss": 2.5906,
+      "step": 174
+    },
+    {
+      "epoch": 0.4107981220657277,
+      "grad_norm": 1.8605163097381592,
+      "learning_rate": 9.357371099768833e-05,
+      "loss": 2.7792,
+      "step": 175
+    },
+    {
+      "epoch": 0.4131455399061033,
+      "grad_norm": 1.8396178483963013,
+      "learning_rate": 9.347815606307445e-05,
+      "loss": 2.8461,
+      "step": 176
+    },
+    {
+      "epoch": 0.4154929577464789,
+      "grad_norm": 1.747826337814331,
+      "learning_rate": 9.338194547663694e-05,
+      "loss": 2.8471,
+      "step": 177
+    },
+    {
+      "epoch": 0.41784037558685444,
+      "grad_norm": 1.8125886917114258,
+      "learning_rate": 9.328508068923418e-05,
+      "loss": 2.6038,
+      "step": 178
+    },
+    {
+      "epoch": 0.42018779342723006,
+      "grad_norm": 2.1431937217712402,
+      "learning_rate": 9.31875631615899e-05,
+      "loss": 2.9192,
+      "step": 179
+    },
+    {
+      "epoch": 0.4225352112676056,
+      "grad_norm": 1.6201167106628418,
+      "learning_rate": 9.308939436427115e-05,
+      "loss": 2.7844,
+      "step": 180
+    },
+    {
+      "epoch": 0.42488262910798125,
+      "grad_norm": 1.9184616804122925,
+      "learning_rate": 9.299057577766622e-05,
+      "loss": 2.9886,
+      "step": 181
+    },
+    {
+      "epoch": 0.4272300469483568,
+      "grad_norm": 1.6579173803329468,
+      "learning_rate": 9.289110889196214e-05,
+      "loss": 2.7774,
+      "step": 182
+    },
+    {
+      "epoch": 0.4295774647887324,
+      "grad_norm": 1.7431479692459106,
+      "learning_rate": 9.27909952071224e-05,
+      "loss": 2.4447,
+      "step": 183
+    },
+    {
+      "epoch": 0.431924882629108,
+      "grad_norm": 1.8576884269714355,
+      "learning_rate": 9.269023623286417e-05,
+      "loss": 2.6607,
+      "step": 184
+    },
+    {
+      "epoch": 0.43427230046948356,
+      "grad_norm": 1.9461067914962769,
+      "learning_rate": 9.258883348863566e-05,
+      "loss": 2.6777,
+      "step": 185
+    },
+    {
+      "epoch": 0.43661971830985913,
+      "grad_norm": 2.585401773452759,
+      "learning_rate": 9.248678850359309e-05,
+      "loss": 2.6043,
+      "step": 186
+    },
+    {
+      "epoch": 0.43896713615023475,
+      "grad_norm": 1.724517822265625,
+      "learning_rate": 9.238410281657775e-05,
+      "loss": 2.8784,
+      "step": 187
+    },
+    {
+      "epoch": 0.4413145539906103,
+      "grad_norm": 1.6239253282546997,
+      "learning_rate": 9.228077797609269e-05,
+      "loss": 2.5965,
+      "step": 188
+    },
+    {
+      "epoch": 0.44366197183098594,
+      "grad_norm": 1.6929664611816406,
+      "learning_rate": 9.217681554027945e-05,
+      "loss": 2.7162,
+      "step": 189
+    },
+    {
+      "epoch": 0.4460093896713615,
+      "grad_norm": 5.029322147369385,
+      "learning_rate": 9.207221707689447e-05,
+      "loss": 2.5703,
+      "step": 190
+    },
+    {
+      "epoch": 0.44835680751173707,
+      "grad_norm": 1.8209108114242554,
+      "learning_rate": 9.196698416328557e-05,
+      "loss": 2.7178,
+      "step": 191
+    },
+    {
+      "epoch": 0.4507042253521127,
+      "grad_norm": 1.71205472946167,
+      "learning_rate": 9.186111838636804e-05,
+      "loss": 2.7824,
+      "step": 192
+    },
+    {
+      "epoch": 0.45305164319248825,
+      "grad_norm": 1.7310357093811035,
+      "learning_rate": 9.175462134260083e-05,
+      "loss": 2.535,
+      "step": 193
+    },
+    {
+      "epoch": 0.45539906103286387,
+      "grad_norm": 1.7333024740219116,
+      "learning_rate": 9.16474946379623e-05,
+      "loss": 2.5513,
+      "step": 194
+    },
+    {
+      "epoch": 0.45774647887323944,
+      "grad_norm": 1.8149805068969727,
+      "learning_rate": 9.153973988792626e-05,
+      "loss": 2.644,
+      "step": 195
+    },
+    {
+      "epoch": 0.460093896713615,
+      "grad_norm": 2.832623243331909,
+      "learning_rate": 9.143135871743736e-05,
+      "loss": 2.5307,
+      "step": 196
+    },
+    {
+      "epoch": 0.4624413145539906,
+      "grad_norm": 1.854354977607727,
+      "learning_rate": 9.132235276088671e-05,
+      "loss": 2.5903,
+      "step": 197
+    },
+    {
+      "epoch": 0.4647887323943662,
+      "grad_norm": 1.7942005395889282,
+      "learning_rate": 9.121272366208722e-05,
+      "loss": 2.609,
+      "step": 198
+    },
+    {
+      "epoch": 0.4671361502347418,
+      "grad_norm": 1.8697292804718018,
+      "learning_rate": 9.110247307424882e-05,
+      "loss": 2.8038,
+      "step": 199
+    },
+    {
+      "epoch": 0.4694835680751174,
+      "grad_norm": 1.690024495124817,
+      "learning_rate": 9.09916026599535e-05,
+      "loss": 2.54,
+      "step": 200
+    },
+    {
+      "epoch": 0.47183098591549294,
+      "grad_norm": 1.8546425104141235,
+      "learning_rate": 9.08801140911302e-05,
+      "loss": 2.7207,
+      "step": 201
+    },
+    {
+      "epoch": 0.47417840375586856,
+      "grad_norm": 1.7608038187026978,
+      "learning_rate": 9.076800904902975e-05,
+      "loss": 2.7212,
+      "step": 202
+    },
+    {
+      "epoch": 0.4765258215962441,
+      "grad_norm": 1.745438814163208,
+      "learning_rate": 9.06552892241993e-05,
+      "loss": 2.6066,
+      "step": 203
+    },
+    {
+      "epoch": 0.4788732394366197,
+      "grad_norm": 2.2760910987854004,
+      "learning_rate": 9.054195631645704e-05,
+      "loss": 2.7303,
+      "step": 204
+    },
+    {
+      "epoch": 0.4812206572769953,
+      "grad_norm": 1.6470048427581787,
+      "learning_rate": 9.042801203486641e-05,
+      "loss": 2.5848,
+      "step": 205
+    },
+    {
+      "epoch": 0.4835680751173709,
+      "grad_norm": 1.7573647499084473,
+      "learning_rate": 9.03134580977104e-05,
+      "loss": 2.6186,
+      "step": 206
+    },
+    {
+      "epoch": 0.4859154929577465,
+      "grad_norm": 1.668116569519043,
+      "learning_rate": 9.019829623246563e-05,
+      "loss": 2.5771,
+      "step": 207
+    },
+    {
+      "epoch": 0.48826291079812206,
+      "grad_norm": 1.744146466255188,
+      "learning_rate": 9.008252817577628e-05,
+      "loss": 2.6314,
+      "step": 208
+    },
+    {
+      "epoch": 0.49061032863849763,
+      "grad_norm": 1.6404486894607544,
+      "learning_rate": 8.99661556734279e-05,
+      "loss": 2.4092,
+      "step": 209
+    },
+    {
+      "epoch": 0.49295774647887325,
+      "grad_norm": 1.5752975940704346,
+      "learning_rate": 8.984918048032116e-05,
+      "loss": 2.4866,
+      "step": 210
+    },
+    {
+      "epoch": 0.4953051643192488,
+      "grad_norm": 1.5991661548614502,
+      "learning_rate": 8.973160436044526e-05,
+      "loss": 2.6372,
+      "step": 211
+    },
+    {
+      "epoch": 0.49765258215962443,
+      "grad_norm": 1.7967761754989624,
+      "learning_rate": 8.961342908685142e-05,
+      "loss": 2.8232,
+      "step": 212
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6597074270248413,
+      "learning_rate": 8.949465644162611e-05,
+      "loss": 2.6634,
+      "step": 213
+    },
+    {
+      "epoch": 0.5023474178403756,
+      "grad_norm": 1.6096906661987305,
+      "learning_rate": 8.937528821586416e-05,
+      "loss": 2.5347,
+      "step": 214
+    },
+    {
+      "epoch": 0.5046948356807511,
+      "grad_norm": 2.293700695037842,
+      "learning_rate": 8.92553262096418e-05,
+      "loss": 2.5662,
+      "step": 215
+    },
+    {
+      "epoch": 0.5070422535211268,
+      "grad_norm": 1.690674066543579,
+      "learning_rate": 8.913477223198951e-05,
+      "loss": 2.6845,
+      "step": 216
+    },
+    {
+      "epoch": 0.5093896713615024,
+      "grad_norm": 1.8968983888626099,
+      "learning_rate": 8.901362810086464e-05,
+      "loss": 2.7832,
+      "step": 217
+    },
+    {
+      "epoch": 0.5117370892018779,
+      "grad_norm": 1.60163414478302,
+      "learning_rate": 8.889189564312417e-05,
+      "loss": 2.4938,
+      "step": 218
+    },
+    {
+      "epoch": 0.5140845070422535,
+      "grad_norm": 1.7134038209915161,
+      "learning_rate": 8.876957669449694e-05,
+      "loss": 2.6695,
+      "step": 219
+    },
+    {
+      "epoch": 0.5164319248826291,
+      "grad_norm": 1.6584218740463257,
+      "learning_rate": 8.864667309955619e-05,
+      "loss": 2.7791,
+      "step": 220
+    },
+    {
+      "epoch": 0.5187793427230047,
+      "grad_norm": 1.782516598701477,
+      "learning_rate": 8.852318671169162e-05,
+      "loss": 2.5737,
+      "step": 221
+    },
+    {
+      "epoch": 0.5211267605633803,
+      "grad_norm": 1.6085395812988281,
+      "learning_rate": 8.839911939308143e-05,
+      "loss": 2.3568,
+      "step": 222
+    },
+    {
+      "epoch": 0.5234741784037559,
+      "grad_norm": 1.5431314706802368,
+      "learning_rate": 8.827447301466432e-05,
+      "loss": 2.4808,
+      "step": 223
+    },
+    {
+      "epoch": 0.5258215962441315,
+      "grad_norm": 1.6283578872680664,
+      "learning_rate": 8.814924945611118e-05,
+      "loss": 2.5186,
+      "step": 224
+    },
+    {
+      "epoch": 0.528169014084507,
+      "grad_norm": 1.6893575191497803,
+      "learning_rate": 8.802345060579684e-05,
+      "loss": 2.7792,
+      "step": 225
+    },
+    {
+      "epoch": 0.5305164319248826,
+      "grad_norm": 1.9037388563156128,
+      "learning_rate": 8.789707836077149e-05,
+      "loss": 2.5587,
+      "step": 226
+    },
+    {
+      "epoch": 0.5328638497652582,
+      "grad_norm": 1.57706618309021,
+      "learning_rate": 8.777013462673217e-05,
+      "loss": 2.4527,
+      "step": 227
+    },
+    {
+      "epoch": 0.5352112676056338,
+      "grad_norm": 2.0377097129821777,
+      "learning_rate": 8.764262131799401e-05,
+      "loss": 2.6479,
+      "step": 228
+    },
+    {
+      "epoch": 0.5375586854460094,
+      "grad_norm": 1.6766222715377808,
+      "learning_rate": 8.75145403574613e-05,
+      "loss": 2.5871,
+      "step": 229
+    },
+    {
+      "epoch": 0.539906103286385,
+      "grad_norm": 1.616554617881775,
+      "learning_rate": 8.738589367659852e-05,
+      "loss": 2.6863,
+      "step": 230
+    },
+    {
+      "epoch": 0.5422535211267606,
+      "grad_norm": 1.6433231830596924,
+      "learning_rate": 8.725668321540128e-05,
+      "loss": 2.5667,
+      "step": 231
+    },
+    {
+      "epoch": 0.5446009389671361,
+      "grad_norm": 1.7055845260620117,
+      "learning_rate": 8.7126910922367e-05,
+      "loss": 2.6972,
+      "step": 232
+    },
+    {
+      "epoch": 0.5469483568075117,
+      "grad_norm": 1.6490628719329834,
+      "learning_rate": 8.699657875446551e-05,
+      "loss": 2.663,
+      "step": 233
+    },
+    {
+      "epoch": 0.5492957746478874,
+      "grad_norm": 1.7133729457855225,
+      "learning_rate": 8.686568867710962e-05,
+      "loss": 2.707,
+      "step": 234
+    },
+    {
+      "epoch": 0.5516431924882629,
+      "grad_norm": 4.825928211212158,
+      "learning_rate": 8.673424266412538e-05,
+      "loss": 2.9167,
+      "step": 235
+    },
+    {
+      "epoch": 0.5539906103286385,
+      "grad_norm": 3.086928367614746,
+      "learning_rate": 8.660224269772237e-05,
+      "loss": 2.5891,
+      "step": 236
+    },
+    {
+      "epoch": 0.5563380281690141,
+      "grad_norm": 4.451052665710449,
+      "learning_rate": 8.646969076846383e-05,
+      "loss": 2.5727,
+      "step": 237
+    },
+    {
+      "epoch": 0.5586854460093896,
+      "grad_norm": 1.7714531421661377,
+      "learning_rate": 8.633658887523664e-05,
+      "loss": 2.9154,
+      "step": 238
+    },
+    {
+      "epoch": 0.5610328638497653,
+      "grad_norm": 1.6821494102478027,
+      "learning_rate": 8.620293902522105e-05,
+      "loss": 2.4986,
+      "step": 239
+    },
+    {
+      "epoch": 0.5633802816901409,
+      "grad_norm": 1.5702682733535767,
+      "learning_rate": 8.606874323386062e-05,
+      "loss": 2.6296,
+      "step": 240
+    },
+    {
+      "epoch": 0.5657276995305164,
+      "grad_norm": 1.569319486618042,
+      "learning_rate": 8.593400352483168e-05,
+      "loss": 2.4851,
+      "step": 241
+    },
+    {
+      "epoch": 0.568075117370892,
+      "grad_norm": 1.9096845388412476,
+      "learning_rate": 8.579872193001285e-05,
+      "loss": 2.6623,
+      "step": 242
+    },
+    {
+      "epoch": 0.5704225352112676,
+      "grad_norm": 1.598663091659546,
+      "learning_rate": 8.56629004894544e-05,
+      "loss": 2.6143,
+      "step": 243
+    },
+    {
+      "epoch": 0.5727699530516432,
+      "grad_norm": 1.6170557737350464,
+      "learning_rate": 8.552654125134751e-05,
+      "loss": 2.7028,
+      "step": 244
+    },
+    {
+      "epoch": 0.5751173708920188,
+      "grad_norm": 1.655263066291809,
+      "learning_rate": 8.538964627199332e-05,
+      "loss": 2.6627,
+      "step": 245
+    },
+    {
+      "epoch": 0.5774647887323944,
+      "grad_norm": 1.723892092704773,
+      "learning_rate": 8.525221761577204e-05,
+      "loss": 2.8072,
+      "step": 246
+    },
+    {
+      "epoch": 0.57981220657277,
+      "grad_norm": 3.896911859512329,
+      "learning_rate": 8.511425735511167e-05,
+      "loss": 2.6754,
+      "step": 247
+    },
+    {
+      "epoch": 0.5821596244131455,
+      "grad_norm": 1.606334924697876,
+      "learning_rate": 8.497576757045683e-05,
+      "loss": 2.2084,
+      "step": 248
+    },
+    {
+      "epoch": 0.5845070422535211,
+      "grad_norm": 1.7658907175064087,
+      "learning_rate": 8.483675035023739e-05,
+      "loss": 2.6787,
+      "step": 249
+    },
+    {
+      "epoch": 0.5868544600938967,
+      "grad_norm": 1.6326708793640137,
+      "learning_rate": 8.4697207790837e-05,
+      "loss": 2.5509,
+      "step": 250
+    },
+    {
+      "epoch": 0.5892018779342723,
+      "grad_norm": 1.6483526229858398,
+      "learning_rate": 8.455714199656138e-05,
+      "loss": 2.468,
+      "step": 251
+    },
+    {
+      "epoch": 0.5915492957746479,
+      "grad_norm": 1.5847233533859253,
+      "learning_rate": 8.441655507960667e-05,
+      "loss": 2.5259,
+      "step": 252
+    },
+    {
+      "epoch": 0.5938967136150235,
+      "grad_norm": 1.744722604751587,
+      "learning_rate": 8.427544916002756e-05,
+      "loss": 2.8526,
+      "step": 253
+    },
+    {
+      "epoch": 0.596244131455399,
+      "grad_norm": 1.7232311964035034,
+      "learning_rate": 8.41338263657053e-05,
+      "loss": 2.9207,
+      "step": 254
+    },
+    {
+      "epoch": 0.5985915492957746,
+      "grad_norm": 1.6628774404525757,
+      "learning_rate": 8.399168883231564e-05,
+      "loss": 2.5764,
+      "step": 255
+    },
+    {
+      "epoch": 0.6009389671361502,
+      "grad_norm": 1.5911755561828613,
+      "learning_rate": 8.38490387032966e-05,
+      "loss": 2.5154,
+      "step": 256
+    },
+    {
+      "epoch": 0.6032863849765259,
+      "grad_norm": 1.588707685470581,
+      "learning_rate": 8.37058781298162e-05,
+      "loss": 2.7086,
+      "step": 257
+    },
+    {
+      "epoch": 0.6056338028169014,
+      "grad_norm": 1.6533771753311157,
+      "learning_rate": 8.35622092707399e-05,
+      "loss": 2.7257,
+      "step": 258
+    },
+    {
+      "epoch": 0.607981220657277,
+      "grad_norm": 1.5928781032562256,
+      "learning_rate": 8.341803429259817e-05,
+      "loss": 2.4508,
+      "step": 259
+    },
+    {
+      "epoch": 0.6103286384976526,
+      "grad_norm": 1.610063910484314,
+      "learning_rate": 8.327335536955376e-05,
+      "loss": 2.5847,
+      "step": 260
+    },
+    {
+      "epoch": 0.6126760563380281,
+      "grad_norm": 1.5072062015533447,
+      "learning_rate": 8.31281746833689e-05,
+      "loss": 2.4901,
+      "step": 261
+    },
+    {
+      "epoch": 0.6150234741784038,
+      "grad_norm": 1.5797420740127563,
+      "learning_rate": 8.29824944233725e-05,
+      "loss": 2.6541,
+      "step": 262
+    },
+    {
+      "epoch": 0.6173708920187794,
+      "grad_norm": 1.695048213005066,
+      "learning_rate": 8.283631678642694e-05,
+      "loss": 2.6204,
+      "step": 263
+    },
+    {
+      "epoch": 0.6197183098591549,
+      "grad_norm": 1.665381669998169,
+      "learning_rate": 8.268964397689516e-05,
+      "loss": 2.3731,
+      "step": 264
+    },
+    {
+      "epoch": 0.6220657276995305,
+      "grad_norm": 1.600093960762024,
+      "learning_rate": 8.254247820660727e-05,
+      "loss": 2.5218,
+      "step": 265
+    },
+    {
+      "epoch": 0.6244131455399061,
+      "grad_norm": 1.901977300643921,
+      "learning_rate": 8.239482169482726e-05,
+      "loss": 2.7333,
+      "step": 266
+    },
+    {
+      "epoch": 0.6267605633802817,
+      "grad_norm": 1.5469740629196167,
+      "learning_rate": 8.22466766682195e-05,
+      "loss": 2.3281,
+      "step": 267
+    },
+    {
+      "epoch": 0.6291079812206573,
+      "grad_norm": 1.6751726865768433,
+      "learning_rate": 8.209804536081516e-05,
+      "loss": 2.8722,
+      "step": 268
+    },
+    {
+      "epoch": 0.6314553990610329,
+      "grad_norm": 1.8046952486038208,
+      "learning_rate": 8.194893001397858e-05,
+      "loss": 2.7165,
+      "step": 269
+    },
+    {
+      "epoch": 0.6338028169014085,
+      "grad_norm": 1.6492973566055298,
+      "learning_rate": 8.179933287637342e-05,
+      "loss": 2.7958,
+      "step": 270
+    },
+    {
+      "epoch": 0.636150234741784,
+      "grad_norm": 1.8264124393463135,
+      "learning_rate": 8.164925620392871e-05,
+      "loss": 2.6456,
+      "step": 271
+    },
+    {
+      "epoch": 0.6384976525821596,
+      "grad_norm": 1.7997009754180908,
+      "learning_rate": 8.149870225980498e-05,
+      "loss": 2.7984,
+      "step": 272
+    },
+    {
+      "epoch": 0.6408450704225352,
+      "grad_norm": 1.5842161178588867,
+      "learning_rate": 8.134767331435991e-05,
+      "loss": 2.4858,
+      "step": 273
+    },
+    {
+      "epoch": 0.6431924882629108,
+      "grad_norm": 1.5902608633041382,
+      "learning_rate": 8.11961716451143e-05,
+      "loss": 2.4969,
+      "step": 274
+    },
+    {
+      "epoch": 0.6455399061032864,
+      "grad_norm": 1.6638398170471191,
+      "learning_rate": 8.104419953671759e-05,
+      "loss": 2.5576,
+      "step": 275
+    },
+    {
+      "epoch": 0.647887323943662,
+      "grad_norm": 1.6883394718170166,
+      "learning_rate": 8.089175928091349e-05,
+      "loss": 2.7457,
+      "step": 276
+    },
+    {
+      "epoch": 0.6502347417840375,
+      "grad_norm": 1.686235785484314,
+      "learning_rate": 8.073885317650534e-05,
+      "loss": 2.8693,
+      "step": 277
+    },
+    {
+      "epoch": 0.6525821596244131,
+      "grad_norm": 1.7170850038528442,
+      "learning_rate": 8.058548352932158e-05,
+      "loss": 2.6033,
+      "step": 278
+    },
+    {
+      "epoch": 0.6549295774647887,
+      "grad_norm": 1.674912691116333,
+      "learning_rate": 8.043165265218078e-05,
+      "loss": 2.5084,
+      "step": 279
+    },
+    {
+      "epoch": 0.6572769953051644,
+      "grad_norm": 1.605450987815857,
+      "learning_rate": 8.027736286485694e-05,
+      "loss": 2.6496,
+      "step": 280
+    },
+    {
+      "epoch": 0.6596244131455399,
+      "grad_norm": 1.6739922761917114,
+      "learning_rate": 8.01226164940444e-05,
+      "loss": 2.6467,
+      "step": 281
+    },
+    {
+      "epoch": 0.6619718309859155,
+      "grad_norm": 2.5874874591827393,
+      "learning_rate": 7.996741587332284e-05,
+      "loss": 2.7073,
+      "step": 282
+    },
+    {
+      "epoch": 0.6643192488262911,
+      "grad_norm": 1.6068830490112305,
+      "learning_rate": 7.981176334312199e-05,
+      "loss": 2.6242,
+      "step": 283
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 1.5618075132369995,
+      "learning_rate": 7.965566125068642e-05,
+      "loss": 2.4312,
+      "step": 284
+    },
+    {
+      "epoch": 0.6690140845070423,
+      "grad_norm": 1.6623550653457642,
+      "learning_rate": 7.949911195004008e-05,
+      "loss": 2.7836,
+      "step": 285
+    },
+    {
+      "epoch": 0.6713615023474179,
+      "grad_norm": 1.6733129024505615,
+      "learning_rate": 7.934211780195089e-05,
+      "loss": 2.6002,
+      "step": 286
+    },
+    {
+      "epoch": 0.6737089201877934,
+      "grad_norm": 1.675602674484253,
+      "learning_rate": 7.9184681173895e-05,
+      "loss": 2.536,
+      "step": 287
+    },
+    {
+      "epoch": 0.676056338028169,
+      "grad_norm": 1.6826605796813965,
+      "learning_rate": 7.902680444002127e-05,
+      "loss": 2.7009,
+      "step": 288
+    },
+    {
+      "epoch": 0.6784037558685446,
+      "grad_norm": 1.4956185817718506,
+      "learning_rate": 7.886848998111526e-05,
+      "loss": 2.3685,
+      "step": 289
+    },
+    {
+      "epoch": 0.6807511737089202,
+      "grad_norm": 1.5748240947723389,
+      "learning_rate": 7.870974018456352e-05,
+      "loss": 2.4995,
+      "step": 290
+    },
+    {
+      "epoch": 0.6830985915492958,
+      "grad_norm": 1.536579966545105,
+      "learning_rate": 7.855055744431747e-05,
+      "loss": 2.5763,
+      "step": 291
+    },
+    {
+      "epoch": 0.6854460093896714,
+      "grad_norm": 1.5213676691055298,
+      "learning_rate": 7.839094416085733e-05,
+      "loss": 2.4565,
+      "step": 292
+    },
+    {
+      "epoch": 0.687793427230047,
+      "grad_norm": 1.668575406074524,
+      "learning_rate": 7.823090274115592e-05,
+      "loss": 2.7248,
+      "step": 293
+    },
+    {
+      "epoch": 0.6901408450704225,
+      "grad_norm": 1.6413085460662842,
+      "learning_rate": 7.807043559864242e-05,
+      "loss": 2.4725,
+      "step": 294
+    },
+    {
+      "epoch": 0.6924882629107981,
+      "grad_norm": 1.682806372642517,
+      "learning_rate": 7.790954515316584e-05,
+      "loss": 2.5853,
+      "step": 295
+    },
+    {
+      "epoch": 0.6948356807511737,
+      "grad_norm": 1.5448178052902222,
+      "learning_rate": 7.774823383095867e-05,
+      "loss": 2.326,
+      "step": 296
+    },
+    {
+      "epoch": 0.6971830985915493,
+      "grad_norm": 1.6233508586883545,
+      "learning_rate": 7.75865040646002e-05,
+      "loss": 2.4861,
+      "step": 297
+    },
+    {
+      "epoch": 0.6995305164319249,
+      "grad_norm": 3.1035714149475098,
+      "learning_rate": 7.742435829297988e-05,
+      "loss": 2.7022,
+      "step": 298
+    },
+    {
+      "epoch": 0.7018779342723005,
+      "grad_norm": 1.6291568279266357,
+      "learning_rate": 7.726179896126055e-05,
+      "loss": 2.4584,
+      "step": 299
+    },
+    {
+      "epoch": 0.704225352112676,
+      "grad_norm": 1.5778498649597168,
+      "learning_rate": 7.709882852084152e-05,
+      "loss": 2.4744,
+      "step": 300
+    },
+    {
+      "epoch": 0.7065727699530516,
+      "grad_norm": 1.5400241613388062,
+      "learning_rate": 7.693544942932162e-05,
+      "loss": 2.3768,
+      "step": 301
+    },
+    {
+      "epoch": 0.7089201877934272,
+      "grad_norm": 1.661958932876587,
+      "learning_rate": 7.677166415046225e-05,
+      "loss": 2.6589,
+      "step": 302
+    },
+    {
+      "epoch": 0.7112676056338029,
+      "grad_norm": 1.5739407539367676,
+      "learning_rate": 7.660747515414996e-05,
+      "loss": 2.4508,
+      "step": 303
+    },
+    {
+      "epoch": 0.7136150234741784,
+      "grad_norm": 1.5613700151443481,
+      "learning_rate": 7.644288491635952e-05,
+      "loss": 2.4655,
+      "step": 304
+    },
+    {
+      "epoch": 0.715962441314554,
+      "grad_norm": 1.6637332439422607,
+      "learning_rate": 7.627789591911634e-05,
+      "loss": 2.4657,
+      "step": 305
+    },
+    {
+      "epoch": 0.7183098591549296,
+      "grad_norm": 1.6113313436508179,
+      "learning_rate": 7.611251065045918e-05,
+      "loss": 2.5994,
+      "step": 306
+    },
+    {
+      "epoch": 0.7206572769953051,
+      "grad_norm": 1.656441569328308,
+      "learning_rate": 7.594673160440258e-05,
+      "loss": 2.5098,
+      "step": 307
+    },
+    {
+      "epoch": 0.7230046948356808,
+      "grad_norm": 1.5881431102752686,
+      "learning_rate": 7.578056128089921e-05,
+      "loss": 2.5514,
+      "step": 308
+    },
+    {
+      "epoch": 0.7253521126760564,
+      "grad_norm": 1.5691207647323608,
+      "learning_rate": 7.56140021858023e-05,
+      "loss": 2.3184,
+      "step": 309
+    },
+    {
+      "epoch": 0.7276995305164319,
+      "grad_norm": 1.5510538816452026,
+      "learning_rate": 7.544705683082768e-05,
+      "loss": 2.6,
+      "step": 310
+    },
+    {
+      "epoch": 0.7300469483568075,
+      "grad_norm": 1.492972731590271,
+      "learning_rate": 7.527972773351604e-05,
+      "loss": 2.5153,
+      "step": 311
+    },
+    {
+      "epoch": 0.7323943661971831,
+      "grad_norm": 1.6028672456741333,
+      "learning_rate": 7.511201741719489e-05,
+      "loss": 2.6437,
+      "step": 312
+    },
+    {
+      "epoch": 0.7347417840375586,
+      "grad_norm": 1.5389173030853271,
+      "learning_rate": 7.494392841094058e-05,
+      "loss": 2.3661,
+      "step": 313
+    },
+    {
+      "epoch": 0.7370892018779343,
+      "grad_norm": 1.5952788591384888,
+      "learning_rate": 7.477546324954002e-05,
+      "loss": 2.5836,
+      "step": 314
+    },
+    {
+      "epoch": 0.7394366197183099,
+      "grad_norm": 1.7208667993545532,
+      "learning_rate": 7.460662447345265e-05,
+      "loss": 2.6799,
+      "step": 315
+    },
+    {
+      "epoch": 0.7417840375586855,
+      "grad_norm": 1.5389221906661987,
+      "learning_rate": 7.443741462877202e-05,
+      "loss": 2.4238,
+      "step": 316
+    },
+    {
+      "epoch": 0.744131455399061,
+      "grad_norm": 2.6154470443725586,
+      "learning_rate": 7.426783626718732e-05,
+      "loss": 2.4256,
+      "step": 317
+    },
+    {
+      "epoch": 0.7464788732394366,
+      "grad_norm": 1.661379337310791,
+      "learning_rate": 7.409789194594507e-05,
+      "loss": 2.5471,
+      "step": 318
+    },
+    {
+      "epoch": 0.7488262910798122,
+      "grad_norm": 1.676131248474121,
+      "learning_rate": 7.392758422781044e-05,
+      "loss": 2.4158,
+      "step": 319
+    },
+    {
+      "epoch": 0.7511737089201878,
+      "grad_norm": 1.7112098932266235,
+      "learning_rate": 7.375691568102863e-05,
+      "loss": 2.7728,
+      "step": 320
+    },
+    {
+      "epoch": 0.7535211267605634,
+      "grad_norm": 1.705582857131958,
+      "learning_rate": 7.358588887928614e-05,
+      "loss": 2.6519,
+      "step": 321
+    },
+    {
+      "epoch": 0.755868544600939,
+      "grad_norm": 1.5166276693344116,
+      "learning_rate": 7.341450640167202e-05,
+      "loss": 2.5022,
+      "step": 322
+    },
+    {
+      "epoch": 0.7582159624413145,
+      "grad_norm": 1.5449557304382324,
+      "learning_rate": 7.324277083263885e-05,
+      "loss": 2.4206,
+      "step": 323
+    },
+    {
+      "epoch": 0.7605633802816901,
+      "grad_norm": 1.4922847747802734,
+      "learning_rate": 7.307068476196388e-05,
+      "loss": 2.4094,
+      "step": 324
+    },
+    {
+      "epoch": 0.7629107981220657,
+      "grad_norm": 1.6599231958389282,
+      "learning_rate": 7.289825078470993e-05,
+      "loss": 2.6286,
+      "step": 325
+    },
+    {
+      "epoch": 0.7652582159624414,
+      "grad_norm": 1.625157356262207,
+      "learning_rate": 7.272547150118624e-05,
+      "loss": 2.3925,
+      "step": 326
+    },
+    {
+      "epoch": 0.7676056338028169,
+      "grad_norm": 1.5863850116729736,
+      "learning_rate": 7.255234951690932e-05,
+      "loss": 2.5999,
+      "step": 327
+    },
+    {
+      "epoch": 0.7699530516431925,
+      "grad_norm": 1.731698751449585,
+      "learning_rate": 7.237888744256357e-05,
+      "loss": 2.6319,
+      "step": 328
+    },
+    {
+      "epoch": 0.7723004694835681,
+      "grad_norm": 1.7433608770370483,
+      "learning_rate": 7.220508789396197e-05,
+      "loss": 2.6355,
+      "step": 329
+    },
+    {
+      "epoch": 0.7746478873239436,
+      "grad_norm": 1.8249096870422363,
+      "learning_rate": 7.203095349200666e-05,
+      "loss": 2.6562,
+      "step": 330
+    },
+    {
+      "epoch": 0.7769953051643192,
+      "grad_norm": 1.5074961185455322,
+      "learning_rate": 7.185648686264934e-05,
+      "loss": 2.4015,
+      "step": 331
+    },
+    {
+      "epoch": 0.7793427230046949,
+      "grad_norm": 1.6418836116790771,
+      "learning_rate": 7.168169063685171e-05,
+      "loss": 2.5743,
+      "step": 332
+    },
+    {
+      "epoch": 0.7816901408450704,
+      "grad_norm": 1.518595814704895,
+      "learning_rate": 7.15065674505458e-05,
+      "loss": 2.4607,
+      "step": 333
+    },
+    {
+      "epoch": 0.784037558685446,
+      "grad_norm": 1.5618414878845215,
+      "learning_rate": 7.13311199445942e-05,
+      "loss": 2.6172,
+      "step": 334
+    },
+    {
+      "epoch": 0.7863849765258216,
+      "grad_norm": 1.5513619184494019,
+      "learning_rate": 7.115535076475031e-05,
+      "loss": 2.4506,
+      "step": 335
+    },
+    {
+      "epoch": 0.7887323943661971,
+      "grad_norm": 1.492394208908081,
+      "learning_rate": 7.09792625616183e-05,
+      "loss": 2.4202,
+      "step": 336
+    },
+    {
+      "epoch": 0.7910798122065728,
+      "grad_norm": 1.5783830881118774,
+      "learning_rate": 7.080285799061324e-05,
+      "loss": 2.5364,
+      "step": 337
+    },
+    {
+      "epoch": 0.7934272300469484,
+      "grad_norm": 1.5449110269546509,
+      "learning_rate": 7.06261397119211e-05,
+      "loss": 2.5843,
+      "step": 338
+    },
+    {
+      "epoch": 0.795774647887324,
+      "grad_norm": 1.5108921527862549,
+      "learning_rate": 7.044911039045847e-05,
+      "loss": 2.474,
+      "step": 339
+    },
+    {
+      "epoch": 0.7981220657276995,
+      "grad_norm": 1.5703331232070923,
+      "learning_rate": 7.027177269583256e-05,
+      "loss": 2.6182,
+      "step": 340
+    },
+    {
+      "epoch": 0.8004694835680751,
+      "grad_norm": 1.5418925285339355,
+      "learning_rate": 7.009412930230084e-05,
+      "loss": 2.486,
+      "step": 341
+    },
+    {
+      "epoch": 0.8028169014084507,
+      "grad_norm": 1.617924690246582,
+      "learning_rate": 6.991618288873066e-05,
+      "loss": 2.5069,
+      "step": 342
+    },
+    {
+      "epoch": 0.8051643192488263,
+      "grad_norm": 1.559870719909668,
+      "learning_rate": 6.973793613855902e-05,
+      "loss": 2.3108,
+      "step": 343
+    },
+    {
+      "epoch": 0.8075117370892019,
+      "grad_norm": 1.569182276725769,
+      "learning_rate": 6.955939173975191e-05,
+      "loss": 2.4741,
+      "step": 344
+    },
+    {
+      "epoch": 0.8098591549295775,
+      "grad_norm": 1.5738292932510376,
+      "learning_rate": 6.938055238476395e-05,
+      "loss": 2.5839,
+      "step": 345
+    },
+    {
+      "epoch": 0.812206572769953,
+      "grad_norm": 1.6121405363082886,
+      "learning_rate": 6.920142077049766e-05,
+      "loss": 2.4468,
+      "step": 346
+    },
+    {
+      "epoch": 0.8145539906103286,
+      "grad_norm": 1.474192500114441,
+      "learning_rate": 6.902199959826286e-05,
+      "loss": 2.3262,
+      "step": 347
+    },
+    {
+      "epoch": 0.8169014084507042,
+      "grad_norm": 1.590667486190796,
+      "learning_rate": 6.88422915737359e-05,
+      "loss": 2.348,
+      "step": 348
+    },
+    {
+      "epoch": 0.8192488262910798,
+      "grad_norm": 1.623644232749939,
+      "learning_rate": 6.866229940691888e-05,
+      "loss": 2.6121,
+      "step": 349
+    },
+    {
+      "epoch": 0.8215962441314554,
+      "grad_norm": 1.6322968006134033,
+      "learning_rate": 6.848202581209875e-05,
+      "loss": 2.4334,
+      "step": 350
+    },
+    {
+      "epoch": 0.823943661971831,
+      "grad_norm": 1.60962975025177,
+      "learning_rate": 6.830147350780645e-05,
+      "loss": 2.4688,
+      "step": 351
+    },
+    {
+      "epoch": 0.8262910798122066,
+      "grad_norm": 1.553849458694458,
+      "learning_rate": 6.812064521677579e-05,
+      "loss": 2.558,
+      "step": 352
+    },
+    {
+      "epoch": 0.8286384976525821,
+      "grad_norm": 1.4718174934387207,
+      "learning_rate": 6.793954366590257e-05,
+      "loss": 2.3784,
+      "step": 353
+    },
+    {
+      "epoch": 0.8309859154929577,
+      "grad_norm": 1.4675418138504028,
+      "learning_rate": 6.775817158620328e-05,
+      "loss": 2.3752,
+      "step": 354
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 1.591253399848938,
+      "learning_rate": 6.7576531712774e-05,
+      "loss": 2.6141,
+      "step": 355
+    },
+    {
+      "epoch": 0.8356807511737089,
+      "grad_norm": 1.6021631956100464,
+      "learning_rate": 6.739462678474917e-05,
+      "loss": 2.3948,
+      "step": 356
+    },
+    {
+      "epoch": 0.8380281690140845,
+      "grad_norm": 1.6207449436187744,
+      "learning_rate": 6.721245954526025e-05,
+      "loss": 2.4396,
+      "step": 357
+    },
+    {
+      "epoch": 0.8403755868544601,
+      "grad_norm": 1.5401954650878906,
+      "learning_rate": 6.703003274139438e-05,
+      "loss": 2.7093,
+      "step": 358
+    },
+    {
+      "epoch": 0.8427230046948356,
+      "grad_norm": 1.6794401407241821,
+      "learning_rate": 6.684734912415289e-05,
+      "loss": 2.5922,
+      "step": 359
+    },
+    {
+      "epoch": 0.8450704225352113,
+      "grad_norm": 1.5993467569351196,
+      "learning_rate": 6.666441144840994e-05,
+      "loss": 2.5032,
+      "step": 360
+    },
+    {
+      "epoch": 0.8474178403755869,
+      "grad_norm": 1.5863590240478516,
+      "learning_rate": 6.648122247287083e-05,
+      "loss": 2.4734,
+      "step": 361
+    },
+    {
+      "epoch": 0.8497652582159625,
+      "grad_norm": 1.555893063545227,
+      "learning_rate": 6.62977849600305e-05,
+      "loss": 2.4898,
+      "step": 362
+    },
+    {
+      "epoch": 0.852112676056338,
+      "grad_norm": 1.5755188465118408,
+      "learning_rate": 6.611410167613184e-05,
+      "loss": 2.4916,
+      "step": 363
+    },
+    {
+      "epoch": 0.8544600938967136,
+      "grad_norm": 1.7223633527755737,
+      "learning_rate": 6.593017539112396e-05,
+      "loss": 2.5071,
+      "step": 364
+    },
+    {
+      "epoch": 0.8568075117370892,
+      "grad_norm": 1.5736289024353027,
+      "learning_rate": 6.574600887862043e-05,
+      "loss": 2.5697,
+      "step": 365
+    },
+    {
+      "epoch": 0.8591549295774648,
+      "grad_norm": 1.4946801662445068,
+      "learning_rate": 6.55616049158575e-05,
+      "loss": 2.4197,
+      "step": 366
+    },
+    {
+      "epoch": 0.8615023474178404,
+      "grad_norm": 1.5540156364440918,
+      "learning_rate": 6.537696628365212e-05,
+      "loss": 2.5144,
+      "step": 367
+    },
+    {
+      "epoch": 0.863849765258216,
+      "grad_norm": 1.4268841743469238,
+      "learning_rate": 6.51920957663601e-05,
+      "loss": 2.2671,
+      "step": 368
+    },
+    {
+      "epoch": 0.8661971830985915,
+      "grad_norm": 1.5012619495391846,
+      "learning_rate": 6.500699615183408e-05,
+      "loss": 2.3217,
+      "step": 369
+    },
+    {
+      "epoch": 0.8685446009389671,
+      "grad_norm": 1.5565698146820068,
+      "learning_rate": 6.482167023138147e-05,
+      "loss": 2.3991,
+      "step": 370
+    },
+    {
+      "epoch": 0.8708920187793427,
+      "grad_norm": 1.6378949880599976,
+      "learning_rate": 6.46361207997224e-05,
+      "loss": 2.6809,
+      "step": 371
+    },
+    {
+      "epoch": 0.8732394366197183,
+      "grad_norm": 1.472183346748352,
+      "learning_rate": 6.445035065494754e-05,
+      "loss": 2.4983,
+      "step": 372
+    },
+    {
+      "epoch": 0.8755868544600939,
+      "grad_norm": 1.5875253677368164,
+      "learning_rate": 6.426436259847595e-05,
+      "loss": 2.5813,
+      "step": 373
+    },
+    {
+      "epoch": 0.8779342723004695,
+      "grad_norm": 1.8145997524261475,
+      "learning_rate": 6.407815943501274e-05,
+      "loss": 2.681,
+      "step": 374
+    },
+    {
+      "epoch": 0.8802816901408451,
+      "grad_norm": 1.617915391921997,
+      "learning_rate": 6.389174397250693e-05,
+      "loss": 2.8412,
+      "step": 375
+    },
+    {
+      "epoch": 0.8826291079812206,
+      "grad_norm": 1.451400876045227,
+      "learning_rate": 6.370511902210897e-05,
+      "loss": 2.3213,
+      "step": 376
+    },
+    {
+      "epoch": 0.8849765258215962,
+      "grad_norm": 1.6040557622909546,
+      "learning_rate": 6.351828739812836e-05,
+      "loss": 2.5939,
+      "step": 377
+    },
+    {
+      "epoch": 0.8873239436619719,
+      "grad_norm": 1.62812340259552,
+      "learning_rate": 6.33312519179913e-05,
+      "loss": 2.483,
+      "step": 378
+    },
+    {
+      "epoch": 0.8896713615023474,
+      "grad_norm": 1.653408169746399,
+      "learning_rate": 6.31440154021981e-05,
+      "loss": 2.4626,
+      "step": 379
+    },
+    {
+      "epoch": 0.892018779342723,
+      "grad_norm": 1.5635944604873657,
+      "learning_rate": 6.295658067428077e-05,
+      "loss": 2.5795,
+      "step": 380
+    },
+    {
+      "epoch": 0.8943661971830986,
+      "grad_norm": 1.5737652778625488,
+      "learning_rate": 6.276895056076022e-05,
+      "loss": 2.6681,
+      "step": 381
+    },
+    {
+      "epoch": 0.8967136150234741,
+      "grad_norm": 1.5483229160308838,
+      "learning_rate": 6.258112789110395e-05,
+      "loss": 2.3571,
+      "step": 382
+    },
+    {
+      "epoch": 0.8990610328638498,
+      "grad_norm": 1.458562970161438,
+      "learning_rate": 6.239311549768311e-05,
+      "loss": 2.3653,
+      "step": 383
+    },
+    {
+      "epoch": 0.9014084507042254,
+      "grad_norm": 1.5463008880615234,
+      "learning_rate": 6.220491621572989e-05,
+      "loss": 2.3952,
+      "step": 384
+    },
+    {
+      "epoch": 0.903755868544601,
+      "grad_norm": 1.5711524486541748,
+      "learning_rate": 6.20165328832948e-05,
+      "loss": 2.4119,
+      "step": 385
+    },
+    {
+      "epoch": 0.9061032863849765,
+      "grad_norm": 1.5307199954986572,
+      "learning_rate": 6.182796834120386e-05,
+      "loss": 2.4746,
+      "step": 386
+    },
+    {
+      "epoch": 0.9084507042253521,
+      "grad_norm": 1.694445013999939,
+      "learning_rate": 6.163922543301565e-05,
+      "loss": 2.5904,
+      "step": 387
+    },
+    {
+      "epoch": 0.9107981220657277,
+      "grad_norm": 1.5599974393844604,
+      "learning_rate": 6.145030700497857e-05,
+      "loss": 2.4916,
+      "step": 388
+    },
+    {
+      "epoch": 0.9131455399061033,
+      "grad_norm": 1.569372534751892,
+      "learning_rate": 6.126121590598788e-05,
+      "loss": 2.4512,
+      "step": 389
+    },
+    {
+      "epoch": 0.9154929577464789,
+      "grad_norm": 1.6844699382781982,
+      "learning_rate": 6.10719549875427e-05,
+      "loss": 2.5407,
+      "step": 390
+    },
+    {
+      "epoch": 0.9178403755868545,
+      "grad_norm": 1.5658828020095825,
+      "learning_rate": 6.088252710370302e-05,
+      "loss": 2.4303,
+      "step": 391
+    },
+    {
+      "epoch": 0.92018779342723,
+      "grad_norm": 1.615698218345642,
+      "learning_rate": 6.069293511104672e-05,
+      "loss": 2.3797,
+      "step": 392
+    },
+    {
+      "epoch": 0.9225352112676056,
+      "grad_norm": 1.5572597980499268,
+      "learning_rate": 6.0503181868626394e-05,
+      "loss": 2.43,
+      "step": 393
+    },
+    {
+      "epoch": 0.9248826291079812,
+      "grad_norm": 1.5271580219268799,
+      "learning_rate": 6.031327023792629e-05,
+      "loss": 2.5207,
+      "step": 394
+    },
+    {
+      "epoch": 0.9272300469483568,
+      "grad_norm": 1.5841909646987915,
+      "learning_rate": 6.012320308281919e-05,
+      "loss": 2.5794,
+      "step": 395
+    },
+    {
+      "epoch": 0.9295774647887324,
+      "grad_norm": 1.4326632022857666,
+      "learning_rate": 5.993298326952318e-05,
+      "loss": 2.2667,
+      "step": 396
+    },
+    {
+      "epoch": 0.931924882629108,
+      "grad_norm": 1.5157475471496582,
+      "learning_rate": 5.974261366655841e-05,
+      "loss": 2.363,
+      "step": 397
+    },
+    {
+      "epoch": 0.9342723004694836,
+      "grad_norm": 1.5970475673675537,
+      "learning_rate": 5.955209714470388e-05,
+      "loss": 2.519,
+      "step": 398
+    },
+    {
+      "epoch": 0.9366197183098591,
+      "grad_norm": 1.5131961107254028,
+      "learning_rate": 5.9361436576954157e-05,
+      "loss": 2.5372,
+      "step": 399
+    },
+    {
+      "epoch": 0.9389671361502347,
+      "grad_norm": 1.5415359735488892,
+      "learning_rate": 5.9170634838475955e-05,
+      "loss": 2.2527,
+      "step": 400
+    },
+    {
+      "epoch": 0.9413145539906104,
+      "grad_norm": 1.4601541757583618,
+      "learning_rate": 5.897969480656491e-05,
+      "loss": 2.2968,
+      "step": 401
+    },
+    {
+      "epoch": 0.9436619718309859,
+      "grad_norm": 1.6026557683944702,
+      "learning_rate": 5.8788619360602126e-05,
+      "loss": 2.4206,
+      "step": 402
+    },
+    {
+      "epoch": 0.9460093896713615,
+      "grad_norm": 1.6263443231582642,
+      "learning_rate": 5.859741138201068e-05,
+      "loss": 2.4799,
+      "step": 403
+    },
+    {
+      "epoch": 0.9483568075117371,
+      "grad_norm": 1.715550184249878,
+      "learning_rate": 5.8406073754212355e-05,
+      "loss": 2.685,
+      "step": 404
+    },
+    {
+      "epoch": 0.9507042253521126,
+      "grad_norm": 1.536697268486023,
+      "learning_rate": 5.8214609362583974e-05,
+      "loss": 2.4646,
+      "step": 405
+    },
+    {
+      "epoch": 0.9530516431924883,
+      "grad_norm": 1.559288501739502,
+      "learning_rate": 5.8023021094413966e-05,
+      "loss": 2.4852,
+      "step": 406
+    },
+    {
+      "epoch": 0.9553990610328639,
+      "grad_norm": 1.523455262184143,
+      "learning_rate": 5.783131183885885e-05,
+      "loss": 2.5411,
+      "step": 407
+    },
+    {
+      "epoch": 0.9577464788732394,
+      "grad_norm": 1.5540595054626465,
+      "learning_rate": 5.763948448689963e-05,
+      "loss": 2.4401,
+      "step": 408
+    },
+    {
+      "epoch": 0.960093896713615,
+      "grad_norm": 1.5767078399658203,
+      "learning_rate": 5.744754193129818e-05,
+      "loss": 2.5617,
+      "step": 409
+    },
+    {
+      "epoch": 0.9624413145539906,
+      "grad_norm": 1.4613933563232422,
+      "learning_rate": 5.725548706655368e-05,
+      "loss": 2.3601,
+      "step": 410
+    },
+    {
+      "epoch": 0.9647887323943662,
+      "grad_norm": 1.5543547868728638,
+      "learning_rate": 5.706332278885893e-05,
+      "loss": 2.4387,
+      "step": 411
+    },
+    {
+      "epoch": 0.9671361502347418,
+      "grad_norm": 1.5075656175613403,
+      "learning_rate": 5.687105199605667e-05,
+      "loss": 2.5223,
+      "step": 412
+    },
+    {
+      "epoch": 0.9694835680751174,
+      "grad_norm": 1.468519926071167,
+      "learning_rate": 5.6678677587595884e-05,
+      "loss": 2.3275,
+      "step": 413
+    },
+    {
+      "epoch": 0.971830985915493,
+      "grad_norm": 1.6226619482040405,
+      "learning_rate": 5.648620246448813e-05,
+      "loss": 2.5644,
+      "step": 414
+    },
+    {
+      "epoch": 0.9741784037558685,
+      "grad_norm": 1.442681908607483,
+      "learning_rate": 5.629362952926367e-05,
+      "loss": 2.2933,
+      "step": 415
+    },
+    {
+      "epoch": 0.9765258215962441,
+      "grad_norm": 1.4851113557815552,
+      "learning_rate": 5.610096168592785e-05,
+      "loss": 2.3604,
+      "step": 416
+    },
+    {
+      "epoch": 0.9788732394366197,
+      "grad_norm": 1.491042971611023,
+      "learning_rate": 5.590820183991716e-05,
+      "loss": 2.4396,
+      "step": 417
+    },
+    {
+      "epoch": 0.9812206572769953,
+      "grad_norm": 1.5098199844360352,
+      "learning_rate": 5.571535289805556e-05,
+      "loss": 2.3233,
+      "step": 418
+    },
+    {
+      "epoch": 0.9835680751173709,
+      "grad_norm": 1.4578444957733154,
+      "learning_rate": 5.552241776851055e-05,
+      "loss": 2.254,
+      "step": 419
+    },
+    {
+      "epoch": 0.9859154929577465,
+      "grad_norm": 1.5896917581558228,
+      "learning_rate": 5.5329399360749336e-05,
+      "loss": 2.8669,
+      "step": 420
+    },
+    {
+      "epoch": 0.9882629107981221,
+      "grad_norm": 1.4958128929138184,
+      "learning_rate": 5.513630058549497e-05,
+      "loss": 2.5112,
+      "step": 421
+    },
+    {
+      "epoch": 0.9906103286384976,
+      "grad_norm": 1.4808344841003418,
+      "learning_rate": 5.494312435468244e-05,
+      "loss": 2.2581,
+      "step": 422
+    },
+    {
+      "epoch": 0.9929577464788732,
+      "grad_norm": 1.4487098455429077,
+      "learning_rate": 5.474987358141478e-05,
+      "loss": 2.3525,
+      "step": 423
+    },
+    {
+      "epoch": 0.9953051643192489,
+      "grad_norm": 1.4762784242630005,
+      "learning_rate": 5.4556551179919134e-05,
+      "loss": 2.3316,
+      "step": 424
+    },
+    {
+      "epoch": 0.9976525821596244,
+      "grad_norm": 1.417960286140442,
+      "learning_rate": 5.436316006550275e-05,
+      "loss": 2.3885,
+      "step": 425
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.407947540283203,
+      "learning_rate": 5.416970315450911e-05,
+      "loss": 2.2996,
+      "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 sha256:2d16d7e5a1b8ebc246a136672a8198a86c78590e65cd7079e16ff28a7c35b806
+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-rand/",
+  "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:aed389e7bb4419dc0f14dd0267a040dcae6db60659553c6013a0628643283b29
+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 sha256:5ca1e341cbba0b2cf0b378b6b832beaba815fcdd3e495b1a4cf32769cc4d6160
+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 sha256:86858c5089fb6f904c3052abc848a4de6a39a3fe1cf34a72d033bb76a837751a
+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 sha256:2d16d7e5a1b8ebc246a136672a8198a86c78590e65cd7079e16ff28a7c35b806
+size 5432

config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-crh_Latn-coli-rand/",
+  "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:aed389e7bb4419dc0f14dd0267a040dcae6db60659553c6013a0628643283b29
+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 sha256:5ca1e341cbba0b2cf0b378b6b832beaba815fcdd3e495b1a4cf32769cc4d6160
+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 sha256:2d16d7e5a1b8ebc246a136672a8198a86c78590e65cd7079e16ff28a7c35b806
+size 5432