Upload folder using huggingface_hub

checkpoint-437/config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-hne_Deva-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-437/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-437/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:679906230e987faec67c1c55504a78b57b6eabb435407c97b26e04c5c6714052
+size 4950070472

checkpoint-437/model-00002-of-00006.safetensors

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

checkpoint-437/model-00003-of-00006.safetensors

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

checkpoint-437/model-00004-of-00006.safetensors

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

checkpoint-437/model-00005-of-00006.safetensors

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

checkpoint-437/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23d2071f43c193018778cb6c395970f0737966d571a8977351421bd1770e5375
+size 4614509080

checkpoint-437/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-437/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:32bc8d04b8a4ec881fb99654aa7777a4ac93a437c51484fd8104f9f28aca6f49
+size 13055369358

checkpoint-437/rng_state.pth

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

checkpoint-437/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2fd5eb6b9346e93b07c8dd975df42b37279c222144e4a1e6f522f4c930275889
+size 1064

checkpoint-437/trainer_state.json

@@ -0,0 +1,3092 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 437,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002288329519450801,
+      "grad_norm": 18.883691787719727,
+      "learning_rate": 2.2727272727272728e-06,
+      "loss": 10.3613,
+      "step": 1
+    },
+    {
+      "epoch": 0.004576659038901602,
+      "grad_norm": 23.170867919921875,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 11.4465,
+      "step": 2
+    },
+    {
+      "epoch": 0.006864988558352402,
+      "grad_norm": 22.565134048461914,
+      "learning_rate": 6.818181818181818e-06,
+      "loss": 11.6681,
+      "step": 3
+    },
+    {
+      "epoch": 0.009153318077803204,
+      "grad_norm": 17.173921585083008,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 9.774,
+      "step": 4
+    },
+    {
+      "epoch": 0.011441647597254004,
+      "grad_norm": 20.478288650512695,
+      "learning_rate": 1.1363636363636365e-05,
+      "loss": 9.4622,
+      "step": 5
+    },
+    {
+      "epoch": 0.013729977116704805,
+      "grad_norm": 16.507568359375,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 9.8348,
+      "step": 6
+    },
+    {
+      "epoch": 0.016018306636155607,
+      "grad_norm": 19.91631317138672,
+      "learning_rate": 1.590909090909091e-05,
+      "loss": 8.3954,
+      "step": 7
+    },
+    {
+      "epoch": 0.018306636155606407,
+      "grad_norm": 18.219541549682617,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 8.8841,
+      "step": 8
+    },
+    {
+      "epoch": 0.020594965675057208,
+      "grad_norm": 18.8386173248291,
+      "learning_rate": 2.0454545454545457e-05,
+      "loss": 7.7361,
+      "step": 9
+    },
+    {
+      "epoch": 0.02288329519450801,
+      "grad_norm": 39.84943771362305,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 7.8007,
+      "step": 10
+    },
+    {
+      "epoch": 0.02517162471395881,
+      "grad_norm": 37.528114318847656,
+      "learning_rate": 2.5e-05,
+      "loss": 7.3203,
+      "step": 11
+    },
+    {
+      "epoch": 0.02745995423340961,
+      "grad_norm": 17.135168075561523,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 6.4416,
+      "step": 12
+    },
+    {
+      "epoch": 0.029748283752860413,
+      "grad_norm": 22.87267303466797,
+      "learning_rate": 2.954545454545455e-05,
+      "loss": 6.7089,
+      "step": 13
+    },
+    {
+      "epoch": 0.032036613272311214,
+      "grad_norm": 15.40108871459961,
+      "learning_rate": 3.181818181818182e-05,
+      "loss": 5.8722,
+      "step": 14
+    },
+    {
+      "epoch": 0.034324942791762014,
+      "grad_norm": 14.930537223815918,
+      "learning_rate": 3.409090909090909e-05,
+      "loss": 5.4715,
+      "step": 15
+    },
+    {
+      "epoch": 0.036613272311212815,
+      "grad_norm": 24.848709106445312,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 6.0102,
+      "step": 16
+    },
+    {
+      "epoch": 0.038901601830663615,
+      "grad_norm": 9.122722625732422,
+      "learning_rate": 3.8636363636363636e-05,
+      "loss": 5.1952,
+      "step": 17
+    },
+    {
+      "epoch": 0.041189931350114416,
+      "grad_norm": 19.82050895690918,
+      "learning_rate": 4.0909090909090915e-05,
+      "loss": 5.4025,
+      "step": 18
+    },
+    {
+      "epoch": 0.043478260869565216,
+      "grad_norm": 14.143157958984375,
+      "learning_rate": 4.318181818181819e-05,
+      "loss": 5.4865,
+      "step": 19
+    },
+    {
+      "epoch": 0.04576659038901602,
+      "grad_norm": 13.207295417785645,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 4.9825,
+      "step": 20
+    },
+    {
+      "epoch": 0.04805491990846682,
+      "grad_norm": 14.648661613464355,
+      "learning_rate": 4.772727272727273e-05,
+      "loss": 5.2058,
+      "step": 21
+    },
+    {
+      "epoch": 0.05034324942791762,
+      "grad_norm": 14.701571464538574,
+      "learning_rate": 5e-05,
+      "loss": 4.878,
+      "step": 22
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 11.080320358276367,
+      "learning_rate": 5.2272727272727274e-05,
+      "loss": 4.8911,
+      "step": 23
+    },
+    {
+      "epoch": 0.05491990846681922,
+      "grad_norm": 7.584497451782227,
+      "learning_rate": 5.4545454545454546e-05,
+      "loss": 4.755,
+      "step": 24
+    },
+    {
+      "epoch": 0.057208237986270026,
+      "grad_norm": 5.899540424346924,
+      "learning_rate": 5.6818181818181825e-05,
+      "loss": 4.7097,
+      "step": 25
+    },
+    {
+      "epoch": 0.059496567505720827,
+      "grad_norm": 25.04722785949707,
+      "learning_rate": 5.90909090909091e-05,
+      "loss": 4.2626,
+      "step": 26
+    },
+    {
+      "epoch": 0.06178489702517163,
+      "grad_norm": 22.6114501953125,
+      "learning_rate": 6.136363636363636e-05,
+      "loss": 4.7042,
+      "step": 27
+    },
+    {
+      "epoch": 0.06407322654462243,
+      "grad_norm": 9.285355567932129,
+      "learning_rate": 6.363636363636364e-05,
+      "loss": 4.4734,
+      "step": 28
+    },
+    {
+      "epoch": 0.06636155606407322,
+      "grad_norm": 6.6501874923706055,
+      "learning_rate": 6.59090909090909e-05,
+      "loss": 4.381,
+      "step": 29
+    },
+    {
+      "epoch": 0.06864988558352403,
+      "grad_norm": 5.746420860290527,
+      "learning_rate": 6.818181818181818e-05,
+      "loss": 4.2421,
+      "step": 30
+    },
+    {
+      "epoch": 0.07093821510297482,
+      "grad_norm": 3.9305100440979004,
+      "learning_rate": 7.045454545454546e-05,
+      "loss": 4.1085,
+      "step": 31
+    },
+    {
+      "epoch": 0.07322654462242563,
+      "grad_norm": 9.795059204101562,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 4.0933,
+      "step": 32
+    },
+    {
+      "epoch": 0.07551487414187644,
+      "grad_norm": 5.907273769378662,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 4.1663,
+      "step": 33
+    },
+    {
+      "epoch": 0.07780320366132723,
+      "grad_norm": 6.9800190925598145,
+      "learning_rate": 7.727272727272727e-05,
+      "loss": 4.1722,
+      "step": 34
+    },
+    {
+      "epoch": 0.08009153318077804,
+      "grad_norm": 7.007209777832031,
+      "learning_rate": 7.954545454545455e-05,
+      "loss": 4.1487,
+      "step": 35
+    },
+    {
+      "epoch": 0.08237986270022883,
+      "grad_norm": 7.624848365783691,
+      "learning_rate": 8.181818181818183e-05,
+      "loss": 4.2597,
+      "step": 36
+    },
+    {
+      "epoch": 0.08466819221967964,
+      "grad_norm": 6.604410171508789,
+      "learning_rate": 8.40909090909091e-05,
+      "loss": 4.1198,
+      "step": 37
+    },
+    {
+      "epoch": 0.08695652173913043,
+      "grad_norm": 5.09440803527832,
+      "learning_rate": 8.636363636363637e-05,
+      "loss": 3.8203,
+      "step": 38
+    },
+    {
+      "epoch": 0.08924485125858124,
+      "grad_norm": 5.450263500213623,
+      "learning_rate": 8.863636363636364e-05,
+      "loss": 3.9002,
+      "step": 39
+    },
+    {
+      "epoch": 0.09153318077803203,
+      "grad_norm": 3.4505696296691895,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 3.9174,
+      "step": 40
+    },
+    {
+      "epoch": 0.09382151029748284,
+      "grad_norm": 7.434864044189453,
+      "learning_rate": 9.318181818181818e-05,
+      "loss": 4.0441,
+      "step": 41
+    },
+    {
+      "epoch": 0.09610983981693363,
+      "grad_norm": 12.046749114990234,
+      "learning_rate": 9.545454545454546e-05,
+      "loss": 3.7604,
+      "step": 42
+    },
+    {
+      "epoch": 0.09839816933638444,
+      "grad_norm": 9.222173690795898,
+      "learning_rate": 9.772727272727274e-05,
+      "loss": 4.1223,
+      "step": 43
+    },
+    {
+      "epoch": 0.10068649885583524,
+      "grad_norm": 4.6726789474487305,
+      "learning_rate": 0.0001,
+      "loss": 3.9827,
+      "step": 44
+    },
+    {
+      "epoch": 0.10297482837528604,
+      "grad_norm": 4.8803791999816895,
+      "learning_rate": 9.999964183507702e-05,
+      "loss": 3.8722,
+      "step": 45
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 6.589260101318359,
+      "learning_rate": 9.999856734543933e-05,
+      "loss": 3.7152,
+      "step": 46
+    },
+    {
+      "epoch": 0.10755148741418764,
+      "grad_norm": 3.9649579524993896,
+      "learning_rate": 9.999677654648072e-05,
+      "loss": 3.606,
+      "step": 47
+    },
+    {
+      "epoch": 0.10983981693363844,
+      "grad_norm": 7.821356296539307,
+      "learning_rate": 9.999426946385727e-05,
+      "loss": 3.9462,
+      "step": 48
+    },
+    {
+      "epoch": 0.11212814645308924,
+      "grad_norm": 7.566446304321289,
+      "learning_rate": 9.999104613348688e-05,
+      "loss": 3.5093,
+      "step": 49
+    },
+    {
+      "epoch": 0.11441647597254005,
+      "grad_norm": 6.507759094238281,
+      "learning_rate": 9.998710660154898e-05,
+      "loss": 3.8733,
+      "step": 50
+    },
+    {
+      "epoch": 0.11670480549199085,
+      "grad_norm": 3.9552855491638184,
+      "learning_rate": 9.998245092448362e-05,
+      "loss": 3.6431,
+      "step": 51
+    },
+    {
+      "epoch": 0.11899313501144165,
+      "grad_norm": 6.334136486053467,
+      "learning_rate": 9.997707916899079e-05,
+      "loss": 3.602,
+      "step": 52
+    },
+    {
+      "epoch": 0.12128146453089245,
+      "grad_norm": 3.8587405681610107,
+      "learning_rate": 9.99709914120295e-05,
+      "loss": 3.5833,
+      "step": 53
+    },
+    {
+      "epoch": 0.12356979405034325,
+      "grad_norm": 8.523442268371582,
+      "learning_rate": 9.996418774081658e-05,
+      "loss": 3.7635,
+      "step": 54
+    },
+    {
+      "epoch": 0.12585812356979406,
+      "grad_norm": 6.120156288146973,
+      "learning_rate": 9.995666825282547e-05,
+      "loss": 3.5929,
+      "step": 55
+    },
+    {
+      "epoch": 0.12814645308924486,
+      "grad_norm": 6.545000076293945,
+      "learning_rate": 9.994843305578486e-05,
+      "loss": 3.5276,
+      "step": 56
+    },
+    {
+      "epoch": 0.13043478260869565,
+      "grad_norm": 6.0784759521484375,
+      "learning_rate": 9.99394822676771e-05,
+      "loss": 3.7363,
+      "step": 57
+    },
+    {
+      "epoch": 0.13272311212814644,
+      "grad_norm": 3.7013518810272217,
+      "learning_rate": 9.99298160167365e-05,
+      "loss": 3.4895,
+      "step": 58
+    },
+    {
+      "epoch": 0.13501144164759726,
+      "grad_norm": 7.4772419929504395,
+      "learning_rate": 9.991943444144757e-05,
+      "loss": 3.6833,
+      "step": 59
+    },
+    {
+      "epoch": 0.13729977116704806,
+      "grad_norm": 19.202077865600586,
+      "learning_rate": 9.990833769054293e-05,
+      "loss": 3.531,
+      "step": 60
+    },
+    {
+      "epoch": 0.13958810068649885,
+      "grad_norm": 5.234053611755371,
+      "learning_rate": 9.989652592300128e-05,
+      "loss": 3.6795,
+      "step": 61
+    },
+    {
+      "epoch": 0.14187643020594964,
+      "grad_norm": 4.528527736663818,
+      "learning_rate": 9.988399930804504e-05,
+      "loss": 3.6353,
+      "step": 62
+    },
+    {
+      "epoch": 0.14416475972540047,
+      "grad_norm": 4.144354820251465,
+      "learning_rate": 9.987075802513797e-05,
+      "loss": 3.7929,
+      "step": 63
+    },
+    {
+      "epoch": 0.14645308924485126,
+      "grad_norm": 7.227776050567627,
+      "learning_rate": 9.985680226398261e-05,
+      "loss": 3.6075,
+      "step": 64
+    },
+    {
+      "epoch": 0.14874141876430205,
+      "grad_norm": 3.7296457290649414,
+      "learning_rate": 9.98421322245175e-05,
+      "loss": 3.5842,
+      "step": 65
+    },
+    {
+      "epoch": 0.15102974828375287,
+      "grad_norm": 4.071051597595215,
+      "learning_rate": 9.98267481169144e-05,
+      "loss": 3.5031,
+      "step": 66
+    },
+    {
+      "epoch": 0.15331807780320367,
+      "grad_norm": 4.512979030609131,
+      "learning_rate": 9.981065016157522e-05,
+      "loss": 3.529,
+      "step": 67
+    },
+    {
+      "epoch": 0.15560640732265446,
+      "grad_norm": 4.703476905822754,
+      "learning_rate": 9.979383858912885e-05,
+      "loss": 3.4393,
+      "step": 68
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 3.3399972915649414,
+      "learning_rate": 9.977631364042795e-05,
+      "loss": 3.3321,
+      "step": 69
+    },
+    {
+      "epoch": 0.16018306636155608,
+      "grad_norm": 4.721877574920654,
+      "learning_rate": 9.975807556654537e-05,
+      "loss": 3.3929,
+      "step": 70
+    },
+    {
+      "epoch": 0.16247139588100687,
+      "grad_norm": 4.589676856994629,
+      "learning_rate": 9.973912462877066e-05,
+      "loss": 3.5313,
+      "step": 71
+    },
+    {
+      "epoch": 0.16475972540045766,
+      "grad_norm": 3.2385528087615967,
+      "learning_rate": 9.971946109860626e-05,
+      "loss": 3.3098,
+      "step": 72
+    },
+    {
+      "epoch": 0.16704805491990846,
+      "grad_norm": 10.516329765319824,
+      "learning_rate": 9.969908525776364e-05,
+      "loss": 3.2536,
+      "step": 73
+    },
+    {
+      "epoch": 0.16933638443935928,
+      "grad_norm": 4.250394821166992,
+      "learning_rate": 9.967799739815925e-05,
+      "loss": 3.5101,
+      "step": 74
+    },
+    {
+      "epoch": 0.17162471395881007,
+      "grad_norm": 6.82457160949707,
+      "learning_rate": 9.965619782191036e-05,
+      "loss": 3.3573,
+      "step": 75
+    },
+    {
+      "epoch": 0.17391304347826086,
+      "grad_norm": 3.3551783561706543,
+      "learning_rate": 9.963368684133072e-05,
+      "loss": 3.2689,
+      "step": 76
+    },
+    {
+      "epoch": 0.17620137299771166,
+      "grad_norm": 3.7537643909454346,
+      "learning_rate": 9.961046477892608e-05,
+      "loss": 3.4124,
+      "step": 77
+    },
+    {
+      "epoch": 0.17848970251716248,
+      "grad_norm": 4.550995826721191,
+      "learning_rate": 9.958653196738954e-05,
+      "loss": 3.2057,
+      "step": 78
+    },
+    {
+      "epoch": 0.18077803203661327,
+      "grad_norm": 3.5284054279327393,
+      "learning_rate": 9.956188874959687e-05,
+      "loss": 3.3048,
+      "step": 79
+    },
+    {
+      "epoch": 0.18306636155606407,
+      "grad_norm": 3.11651873588562,
+      "learning_rate": 9.953653547860151e-05,
+      "loss": 3.2424,
+      "step": 80
+    },
+    {
+      "epoch": 0.1853546910755149,
+      "grad_norm": 3.3010475635528564,
+      "learning_rate": 9.951047251762954e-05,
+      "loss": 3.3064,
+      "step": 81
+    },
+    {
+      "epoch": 0.18764302059496568,
+      "grad_norm": 3.0703697204589844,
+      "learning_rate": 9.948370024007454e-05,
+      "loss": 3.2671,
+      "step": 82
+    },
+    {
+      "epoch": 0.18993135011441648,
+      "grad_norm": 3.3035237789154053,
+      "learning_rate": 9.94562190294921e-05,
+      "loss": 3.2557,
+      "step": 83
+    },
+    {
+      "epoch": 0.19221967963386727,
+      "grad_norm": 4.406396865844727,
+      "learning_rate": 9.942802927959443e-05,
+      "loss": 3.2696,
+      "step": 84
+    },
+    {
+      "epoch": 0.1945080091533181,
+      "grad_norm": 3.550173759460449,
+      "learning_rate": 9.939913139424476e-05,
+      "loss": 3.195,
+      "step": 85
+    },
+    {
+      "epoch": 0.19679633867276888,
+      "grad_norm": 3.505958318710327,
+      "learning_rate": 9.936952578745142e-05,
+      "loss": 3.1276,
+      "step": 86
+    },
+    {
+      "epoch": 0.19908466819221968,
+      "grad_norm": 3.840190887451172,
+      "learning_rate": 9.933921288336201e-05,
+      "loss": 3.164,
+      "step": 87
+    },
+    {
+      "epoch": 0.20137299771167047,
+      "grad_norm": 3.551823139190674,
+      "learning_rate": 9.93081931162573e-05,
+      "loss": 3.1447,
+      "step": 88
+    },
+    {
+      "epoch": 0.2036613272311213,
+      "grad_norm": 2.611680269241333,
+      "learning_rate": 9.927646693054496e-05,
+      "loss": 2.9314,
+      "step": 89
+    },
+    {
+      "epoch": 0.20594965675057209,
+      "grad_norm": 3.5281317234039307,
+      "learning_rate": 9.92440347807533e-05,
+      "loss": 3.1775,
+      "step": 90
+    },
+    {
+      "epoch": 0.20823798627002288,
+      "grad_norm": 2.898061513900757,
+      "learning_rate": 9.921089713152462e-05,
+      "loss": 3.1434,
+      "step": 91
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 2.3443377017974854,
+      "learning_rate": 9.91770544576087e-05,
+      "loss": 3.1309,
+      "step": 92
+    },
+    {
+      "epoch": 0.2128146453089245,
+      "grad_norm": 3.889735460281372,
+      "learning_rate": 9.914250724385588e-05,
+      "loss": 3.413,
+      "step": 93
+    },
+    {
+      "epoch": 0.2151029748283753,
+      "grad_norm": 2.6670336723327637,
+      "learning_rate": 9.910725598521013e-05,
+      "loss": 3.3243,
+      "step": 94
+    },
+    {
+      "epoch": 0.21739130434782608,
+      "grad_norm": 3.1559407711029053,
+      "learning_rate": 9.907130118670207e-05,
+      "loss": 3.1358,
+      "step": 95
+    },
+    {
+      "epoch": 0.21967963386727687,
+      "grad_norm": 3.0841243267059326,
+      "learning_rate": 9.90346433634416e-05,
+      "loss": 3.0006,
+      "step": 96
+    },
+    {
+      "epoch": 0.2219679633867277,
+      "grad_norm": 3.750990867614746,
+      "learning_rate": 9.899728304061054e-05,
+      "loss": 2.9871,
+      "step": 97
+    },
+    {
+      "epoch": 0.2242562929061785,
+      "grad_norm": 3.793957471847534,
+      "learning_rate": 9.89592207534552e-05,
+      "loss": 3.1504,
+      "step": 98
+    },
+    {
+      "epoch": 0.22654462242562928,
+      "grad_norm": 2.926502227783203,
+      "learning_rate": 9.892045704727864e-05,
+      "loss": 3.0773,
+      "step": 99
+    },
+    {
+      "epoch": 0.2288329519450801,
+      "grad_norm": 3.8088338375091553,
+      "learning_rate": 9.888099247743283e-05,
+      "loss": 3.0159,
+      "step": 100
+    },
+    {
+      "epoch": 0.2311212814645309,
+      "grad_norm": 3.136817455291748,
+      "learning_rate": 9.884082760931078e-05,
+      "loss": 3.0799,
+      "step": 101
+    },
+    {
+      "epoch": 0.2334096109839817,
+      "grad_norm": 3.4826202392578125,
+      "learning_rate": 9.879996301833833e-05,
+      "loss": 2.9949,
+      "step": 102
+    },
+    {
+      "epoch": 0.23569794050343248,
+      "grad_norm": 2.8044424057006836,
+      "learning_rate": 9.875839928996605e-05,
+      "loss": 2.9263,
+      "step": 103
+    },
+    {
+      "epoch": 0.2379862700228833,
+      "grad_norm": 3.0698440074920654,
+      "learning_rate": 9.871613701966067e-05,
+      "loss": 2.901,
+      "step": 104
+    },
+    {
+      "epoch": 0.2402745995423341,
+      "grad_norm": 3.5167031288146973,
+      "learning_rate": 9.867317681289674e-05,
+      "loss": 2.8163,
+      "step": 105
+    },
+    {
+      "epoch": 0.2425629290617849,
+      "grad_norm": 4.114626407623291,
+      "learning_rate": 9.862951928514782e-05,
+      "loss": 2.8116,
+      "step": 106
+    },
+    {
+      "epoch": 0.2448512585812357,
+      "grad_norm": 3.7168688774108887,
+      "learning_rate": 9.858516506187769e-05,
+      "loss": 2.8926,
+      "step": 107
+    },
+    {
+      "epoch": 0.2471395881006865,
+      "grad_norm": 4.157980442047119,
+      "learning_rate": 9.854011477853146e-05,
+      "loss": 3.0417,
+      "step": 108
+    },
+    {
+      "epoch": 0.2494279176201373,
+      "grad_norm": 3.974653959274292,
+      "learning_rate": 9.849436908052636e-05,
+      "loss": 2.9427,
+      "step": 109
+    },
+    {
+      "epoch": 0.2517162471395881,
+      "grad_norm": 3.3569509983062744,
+      "learning_rate": 9.844792862324258e-05,
+      "loss": 3.0074,
+      "step": 110
+    },
+    {
+      "epoch": 0.2540045766590389,
+      "grad_norm": 3.318108081817627,
+      "learning_rate": 9.840079407201381e-05,
+      "loss": 2.9424,
+      "step": 111
+    },
+    {
+      "epoch": 0.2562929061784897,
+      "grad_norm": 2.5115249156951904,
+      "learning_rate": 9.835296610211779e-05,
+      "loss": 2.8719,
+      "step": 112
+    },
+    {
+      "epoch": 0.2585812356979405,
+      "grad_norm": 4.036254405975342,
+      "learning_rate": 9.830444539876655e-05,
+      "loss": 2.8323,
+      "step": 113
+    },
+    {
+      "epoch": 0.2608695652173913,
+      "grad_norm": 2.956651210784912,
+      "learning_rate": 9.825523265709666e-05,
+      "loss": 2.8217,
+      "step": 114
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 20.366439819335938,
+      "learning_rate": 9.820532858215924e-05,
+      "loss": 2.9269,
+      "step": 115
+    },
+    {
+      "epoch": 0.2654462242562929,
+      "grad_norm": 5.692528247833252,
+      "learning_rate": 9.815473388890983e-05,
+      "loss": 2.9319,
+      "step": 116
+    },
+    {
+      "epoch": 0.26773455377574373,
+      "grad_norm": 4.276037693023682,
+      "learning_rate": 9.810344930219824e-05,
+      "loss": 2.8439,
+      "step": 117
+    },
+    {
+      "epoch": 0.2700228832951945,
+      "grad_norm": 5.107546806335449,
+      "learning_rate": 9.805147555675805e-05,
+      "loss": 2.8193,
+      "step": 118
+    },
+    {
+      "epoch": 0.2723112128146453,
+      "grad_norm": 4.974969387054443,
+      "learning_rate": 9.799881339719615e-05,
+      "loss": 2.8969,
+      "step": 119
+    },
+    {
+      "epoch": 0.2745995423340961,
+      "grad_norm": 2.5996475219726562,
+      "learning_rate": 9.794546357798208e-05,
+      "loss": 2.7867,
+      "step": 120
+    },
+    {
+      "epoch": 0.2768878718535469,
+      "grad_norm": 5.79841947555542,
+      "learning_rate": 9.789142686343723e-05,
+      "loss": 2.9682,
+      "step": 121
+    },
+    {
+      "epoch": 0.2791762013729977,
+      "grad_norm": 3.4821016788482666,
+      "learning_rate": 9.783670402772379e-05,
+      "loss": 2.8194,
+      "step": 122
+    },
+    {
+      "epoch": 0.2814645308924485,
+      "grad_norm": 3.6274449825286865,
+      "learning_rate": 9.778129585483377e-05,
+      "loss": 2.9552,
+      "step": 123
+    },
+    {
+      "epoch": 0.2837528604118993,
+      "grad_norm": 5.462944507598877,
+      "learning_rate": 9.772520313857775e-05,
+      "loss": 2.82,
+      "step": 124
+    },
+    {
+      "epoch": 0.28604118993135014,
+      "grad_norm": 2.919630765914917,
+      "learning_rate": 9.766842668257348e-05,
+      "loss": 2.8958,
+      "step": 125
+    },
+    {
+      "epoch": 0.28832951945080093,
+      "grad_norm": 4.07204008102417,
+      "learning_rate": 9.761096730023432e-05,
+      "loss": 2.6548,
+      "step": 126
+    },
+    {
+      "epoch": 0.2906178489702517,
+      "grad_norm": 3.1327707767486572,
+      "learning_rate": 9.755282581475769e-05,
+      "loss": 2.7502,
+      "step": 127
+    },
+    {
+      "epoch": 0.2929061784897025,
+      "grad_norm": 3.4713220596313477,
+      "learning_rate": 9.749400305911322e-05,
+      "loss": 2.8989,
+      "step": 128
+    },
+    {
+      "epoch": 0.2951945080091533,
+      "grad_norm": 3.0974678993225098,
+      "learning_rate": 9.743449987603083e-05,
+      "loss": 2.7687,
+      "step": 129
+    },
+    {
+      "epoch": 0.2974828375286041,
+      "grad_norm": 2.840522527694702,
+      "learning_rate": 9.737431711798864e-05,
+      "loss": 2.6381,
+      "step": 130
+    },
+    {
+      "epoch": 0.2997711670480549,
+      "grad_norm": 3.248396635055542,
+      "learning_rate": 9.731345564720074e-05,
+      "loss": 2.803,
+      "step": 131
+    },
+    {
+      "epoch": 0.30205949656750575,
+      "grad_norm": 2.489511251449585,
+      "learning_rate": 9.725191633560491e-05,
+      "loss": 2.7353,
+      "step": 132
+    },
+    {
+      "epoch": 0.30434782608695654,
+      "grad_norm": 3.164222240447998,
+      "learning_rate": 9.718970006485006e-05,
+      "loss": 2.8043,
+      "step": 133
+    },
+    {
+      "epoch": 0.30663615560640733,
+      "grad_norm": 2.9290547370910645,
+      "learning_rate": 9.712680772628364e-05,
+      "loss": 2.8801,
+      "step": 134
+    },
+    {
+      "epoch": 0.30892448512585813,
+      "grad_norm": 2.126338243484497,
+      "learning_rate": 9.70632402209388e-05,
+      "loss": 2.6804,
+      "step": 135
+    },
+    {
+      "epoch": 0.3112128146453089,
+      "grad_norm": 3.28828501701355,
+      "learning_rate": 9.69989984595216e-05,
+      "loss": 2.7758,
+      "step": 136
+    },
+    {
+      "epoch": 0.3135011441647597,
+      "grad_norm": 2.5393142700195312,
+      "learning_rate": 9.693408336239783e-05,
+      "loss": 2.7694,
+      "step": 137
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 2.971083641052246,
+      "learning_rate": 9.686849585957994e-05,
+      "loss": 2.8087,
+      "step": 138
+    },
+    {
+      "epoch": 0.3180778032036613,
+      "grad_norm": 2.3476786613464355,
+      "learning_rate": 9.680223689071364e-05,
+      "loss": 2.6401,
+      "step": 139
+    },
+    {
+      "epoch": 0.32036613272311215,
+      "grad_norm": 3.1489346027374268,
+      "learning_rate": 9.673530740506447e-05,
+      "loss": 2.6452,
+      "step": 140
+    },
+    {
+      "epoch": 0.32265446224256294,
+      "grad_norm": 2.252417802810669,
+      "learning_rate": 9.666770836150421e-05,
+      "loss": 2.7393,
+      "step": 141
+    },
+    {
+      "epoch": 0.32494279176201374,
+      "grad_norm": 2.2149484157562256,
+      "learning_rate": 9.659944072849707e-05,
+      "loss": 2.6799,
+      "step": 142
+    },
+    {
+      "epoch": 0.32723112128146453,
+      "grad_norm": 2.29628849029541,
+      "learning_rate": 9.653050548408593e-05,
+      "loss": 2.7648,
+      "step": 143
+    },
+    {
+      "epoch": 0.3295194508009153,
+      "grad_norm": 2.487936019897461,
+      "learning_rate": 9.646090361587827e-05,
+      "loss": 2.7027,
+      "step": 144
+    },
+    {
+      "epoch": 0.3318077803203661,
+      "grad_norm": 2.9859704971313477,
+      "learning_rate": 9.639063612103198e-05,
+      "loss": 2.7749,
+      "step": 145
+    },
+    {
+      "epoch": 0.3340961098398169,
+      "grad_norm": 2.6007204055786133,
+      "learning_rate": 9.631970400624113e-05,
+      "loss": 2.7246,
+      "step": 146
+    },
+    {
+      "epoch": 0.33638443935926776,
+      "grad_norm": 33.11263656616211,
+      "learning_rate": 9.624810828772155e-05,
+      "loss": 2.7545,
+      "step": 147
+    },
+    {
+      "epoch": 0.33867276887871856,
+      "grad_norm": 3.515178918838501,
+      "learning_rate": 9.617584999119625e-05,
+      "loss": 2.732,
+      "step": 148
+    },
+    {
+      "epoch": 0.34096109839816935,
+      "grad_norm": 2.4641125202178955,
+      "learning_rate": 9.610293015188067e-05,
+      "loss": 2.5388,
+      "step": 149
+    },
+    {
+      "epoch": 0.34324942791762014,
+      "grad_norm": 3.40997314453125,
+      "learning_rate": 9.602934981446803e-05,
+      "loss": 2.7129,
+      "step": 150
+    },
+    {
+      "epoch": 0.34553775743707094,
+      "grad_norm": 2.843931198120117,
+      "learning_rate": 9.59551100331141e-05,
+      "loss": 2.5792,
+      "step": 151
+    },
+    {
+      "epoch": 0.34782608695652173,
+      "grad_norm": 3.463437795639038,
+      "learning_rate": 9.588021187142235e-05,
+      "loss": 2.7224,
+      "step": 152
+    },
+    {
+      "epoch": 0.3501144164759725,
+      "grad_norm": 3.4558653831481934,
+      "learning_rate": 9.580465640242851e-05,
+      "loss": 2.7847,
+      "step": 153
+    },
+    {
+      "epoch": 0.3524027459954233,
+      "grad_norm": 2.8027448654174805,
+      "learning_rate": 9.572844470858537e-05,
+      "loss": 2.636,
+      "step": 154
+    },
+    {
+      "epoch": 0.35469107551487417,
+      "grad_norm": 3.39350962638855,
+      "learning_rate": 9.565157788174712e-05,
+      "loss": 2.7577,
+      "step": 155
+    },
+    {
+      "epoch": 0.35697940503432496,
+      "grad_norm": 2.9632396697998047,
+      "learning_rate": 9.557405702315381e-05,
+      "loss": 2.6417,
+      "step": 156
+    },
+    {
+      "epoch": 0.35926773455377575,
+      "grad_norm": 2.825023889541626,
+      "learning_rate": 9.549588324341555e-05,
+      "loss": 2.5702,
+      "step": 157
+    },
+    {
+      "epoch": 0.36155606407322655,
+      "grad_norm": 3.1311004161834717,
+      "learning_rate": 9.541705766249655e-05,
+      "loss": 2.7682,
+      "step": 158
+    },
+    {
+      "epoch": 0.36384439359267734,
+      "grad_norm": 3.473098039627075,
+      "learning_rate": 9.533758140969912e-05,
+      "loss": 2.6545,
+      "step": 159
+    },
+    {
+      "epoch": 0.36613272311212813,
+      "grad_norm": 2.7823293209075928,
+      "learning_rate": 9.525745562364756e-05,
+      "loss": 2.6668,
+      "step": 160
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 2.742286205291748,
+      "learning_rate": 9.517668145227167e-05,
+      "loss": 2.7998,
+      "step": 161
+    },
+    {
+      "epoch": 0.3707093821510298,
+      "grad_norm": 4.77061653137207,
+      "learning_rate": 9.509526005279044e-05,
+      "loss": 2.789,
+      "step": 162
+    },
+    {
+      "epoch": 0.37299771167048057,
+      "grad_norm": 3.353771448135376,
+      "learning_rate": 9.501319259169543e-05,
+      "loss": 2.5644,
+      "step": 163
+    },
+    {
+      "epoch": 0.37528604118993136,
+      "grad_norm": 2.2853267192840576,
+      "learning_rate": 9.493048024473412e-05,
+      "loss": 2.6974,
+      "step": 164
+    },
+    {
+      "epoch": 0.37757437070938216,
+      "grad_norm": 3.146768569946289,
+      "learning_rate": 9.484712419689292e-05,
+      "loss": 2.568,
+      "step": 165
+    },
+    {
+      "epoch": 0.37986270022883295,
+      "grad_norm": 5.74379301071167,
+      "learning_rate": 9.476312564238034e-05,
+      "loss": 2.6581,
+      "step": 166
+    },
+    {
+      "epoch": 0.38215102974828374,
+      "grad_norm": 2.7834126949310303,
+      "learning_rate": 9.467848578460985e-05,
+      "loss": 2.6585,
+      "step": 167
+    },
+    {
+      "epoch": 0.38443935926773454,
+      "grad_norm": 4.0173139572143555,
+      "learning_rate": 9.459320583618252e-05,
+      "loss": 2.6218,
+      "step": 168
+    },
+    {
+      "epoch": 0.38672768878718533,
+      "grad_norm": 2.463062286376953,
+      "learning_rate": 9.450728701886983e-05,
+      "loss": 2.5853,
+      "step": 169
+    },
+    {
+      "epoch": 0.3890160183066362,
+      "grad_norm": 2.0359222888946533,
+      "learning_rate": 9.442073056359604e-05,
+      "loss": 2.4268,
+      "step": 170
+    },
+    {
+      "epoch": 0.391304347826087,
+      "grad_norm": 2.32696533203125,
+      "learning_rate": 9.433353771042059e-05,
+      "loss": 2.6057,
+      "step": 171
+    },
+    {
+      "epoch": 0.39359267734553777,
+      "grad_norm": 2.851733446121216,
+      "learning_rate": 9.424570970852034e-05,
+      "loss": 2.5317,
+      "step": 172
+    },
+    {
+      "epoch": 0.39588100686498856,
+      "grad_norm": 2.5211381912231445,
+      "learning_rate": 9.415724781617165e-05,
+      "loss": 2.577,
+      "step": 173
+    },
+    {
+      "epoch": 0.39816933638443935,
+      "grad_norm": 2.574113130569458,
+      "learning_rate": 9.406815330073244e-05,
+      "loss": 2.4746,
+      "step": 174
+    },
+    {
+      "epoch": 0.40045766590389015,
+      "grad_norm": 2.246934413909912,
+      "learning_rate": 9.397842743862391e-05,
+      "loss": 2.7172,
+      "step": 175
+    },
+    {
+      "epoch": 0.40274599542334094,
+      "grad_norm": 2.0411953926086426,
+      "learning_rate": 9.388807151531229e-05,
+      "loss": 2.3795,
+      "step": 176
+    },
+    {
+      "epoch": 0.40503432494279173,
+      "grad_norm": 2.0988056659698486,
+      "learning_rate": 9.37970868252905e-05,
+      "loss": 2.4858,
+      "step": 177
+    },
+    {
+      "epoch": 0.4073226544622426,
+      "grad_norm": 2.6199228763580322,
+      "learning_rate": 9.37054746720595e-05,
+      "loss": 2.5046,
+      "step": 178
+    },
+    {
+      "epoch": 0.4096109839816934,
+      "grad_norm": 2.879153251647949,
+      "learning_rate": 9.36132363681097e-05,
+      "loss": 2.5783,
+      "step": 179
+    },
+    {
+      "epoch": 0.41189931350114417,
+      "grad_norm": 2.188169002532959,
+      "learning_rate": 9.352037323490208e-05,
+      "loss": 2.5765,
+      "step": 180
+    },
+    {
+      "epoch": 0.41418764302059496,
+      "grad_norm": 2.8007380962371826,
+      "learning_rate": 9.342688660284935e-05,
+      "loss": 2.5559,
+      "step": 181
+    },
+    {
+      "epoch": 0.41647597254004576,
+      "grad_norm": 2.118467092514038,
+      "learning_rate": 9.333277781129678e-05,
+      "loss": 2.6657,
+      "step": 182
+    },
+    {
+      "epoch": 0.41876430205949655,
+      "grad_norm": 2.415208101272583,
+      "learning_rate": 9.32380482085031e-05,
+      "loss": 2.5692,
+      "step": 183
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 2.284472942352295,
+      "learning_rate": 9.314269915162114e-05,
+      "loss": 2.8171,
+      "step": 184
+    },
+    {
+      "epoch": 0.4233409610983982,
+      "grad_norm": 2.6300644874572754,
+      "learning_rate": 9.304673200667842e-05,
+      "loss": 2.5209,
+      "step": 185
+    },
+    {
+      "epoch": 0.425629290617849,
+      "grad_norm": 2.484619617462158,
+      "learning_rate": 9.295014814855753e-05,
+      "loss": 2.6103,
+      "step": 186
+    },
+    {
+      "epoch": 0.4279176201372998,
+      "grad_norm": 1.9404324293136597,
+      "learning_rate": 9.285294896097649e-05,
+      "loss": 2.6948,
+      "step": 187
+    },
+    {
+      "epoch": 0.4302059496567506,
+      "grad_norm": 5.251960754394531,
+      "learning_rate": 9.275513583646884e-05,
+      "loss": 2.5566,
+      "step": 188
+    },
+    {
+      "epoch": 0.43249427917620137,
+      "grad_norm": 3.4060168266296387,
+      "learning_rate": 9.265671017636383e-05,
+      "loss": 2.4573,
+      "step": 189
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 2.6057965755462646,
+      "learning_rate": 9.255767339076622e-05,
+      "loss": 2.578,
+      "step": 190
+    },
+    {
+      "epoch": 0.43707093821510296,
+      "grad_norm": 2.6781222820281982,
+      "learning_rate": 9.24580268985361e-05,
+      "loss": 2.6097,
+      "step": 191
+    },
+    {
+      "epoch": 0.43935926773455375,
+      "grad_norm": 2.630892038345337,
+      "learning_rate": 9.23577721272686e-05,
+      "loss": 2.4862,
+      "step": 192
+    },
+    {
+      "epoch": 0.4416475972540046,
+      "grad_norm": 2.559825897216797,
+      "learning_rate": 9.225691051327341e-05,
+      "loss": 2.4985,
+      "step": 193
+    },
+    {
+      "epoch": 0.4439359267734554,
+      "grad_norm": 2.574370861053467,
+      "learning_rate": 9.215544350155422e-05,
+      "loss": 2.5045,
+      "step": 194
+    },
+    {
+      "epoch": 0.4462242562929062,
+      "grad_norm": 2.611264944076538,
+      "learning_rate": 9.205337254578804e-05,
+      "loss": 2.5606,
+      "step": 195
+    },
+    {
+      "epoch": 0.448512585812357,
+      "grad_norm": 2.938662052154541,
+      "learning_rate": 9.195069910830427e-05,
+      "loss": 2.3961,
+      "step": 196
+    },
+    {
+      "epoch": 0.45080091533180777,
+      "grad_norm": 2.3292245864868164,
+      "learning_rate": 9.184742466006389e-05,
+      "loss": 2.3445,
+      "step": 197
+    },
+    {
+      "epoch": 0.45308924485125857,
+      "grad_norm": 3.314525604248047,
+      "learning_rate": 9.174355068063828e-05,
+      "loss": 2.3837,
+      "step": 198
+    },
+    {
+      "epoch": 0.45537757437070936,
+      "grad_norm": 2.9725730419158936,
+      "learning_rate": 9.163907865818806e-05,
+      "loss": 2.4219,
+      "step": 199
+    },
+    {
+      "epoch": 0.4576659038901602,
+      "grad_norm": 2.6355316638946533,
+      "learning_rate": 9.15340100894418e-05,
+      "loss": 2.5177,
+      "step": 200
+    },
+    {
+      "epoch": 0.459954233409611,
+      "grad_norm": 2.6601266860961914,
+      "learning_rate": 9.142834647967455e-05,
+      "loss": 2.5016,
+      "step": 201
+    },
+    {
+      "epoch": 0.4622425629290618,
+      "grad_norm": 2.8524606227874756,
+      "learning_rate": 9.132208934268622e-05,
+      "loss": 2.6128,
+      "step": 202
+    },
+    {
+      "epoch": 0.4645308924485126,
+      "grad_norm": 2.9725794792175293,
+      "learning_rate": 9.121524020078002e-05,
+      "loss": 2.4071,
+      "step": 203
+    },
+    {
+      "epoch": 0.4668192219679634,
+      "grad_norm": 2.737095832824707,
+      "learning_rate": 9.110780058474052e-05,
+      "loss": 2.5999,
+      "step": 204
+    },
+    {
+      "epoch": 0.4691075514874142,
+      "grad_norm": 2.458143711090088,
+      "learning_rate": 9.099977203381176e-05,
+      "loss": 2.4793,
+      "step": 205
+    },
+    {
+      "epoch": 0.47139588100686497,
+      "grad_norm": 2.4118494987487793,
+      "learning_rate": 9.08911560956753e-05,
+      "loss": 2.5604,
+      "step": 206
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 2.5340561866760254,
+      "learning_rate": 9.078195432642787e-05,
+      "loss": 2.3189,
+      "step": 207
+    },
+    {
+      "epoch": 0.4759725400457666,
+      "grad_norm": 15.165704727172852,
+      "learning_rate": 9.067216829055922e-05,
+      "loss": 2.3839,
+      "step": 208
+    },
+    {
+      "epoch": 0.4782608695652174,
+      "grad_norm": 3.408928394317627,
+      "learning_rate": 9.056179956092962e-05,
+      "loss": 2.4059,
+      "step": 209
+    },
+    {
+      "epoch": 0.4805491990846682,
+      "grad_norm": 2.3760733604431152,
+      "learning_rate": 9.045084971874738e-05,
+      "loss": 2.4665,
+      "step": 210
+    },
+    {
+      "epoch": 0.482837528604119,
+      "grad_norm": 2.2348382472991943,
+      "learning_rate": 9.033932035354616e-05,
+      "loss": 2.5414,
+      "step": 211
+    },
+    {
+      "epoch": 0.4851258581235698,
+      "grad_norm": 2.8291633129119873,
+      "learning_rate": 9.022721306316222e-05,
+      "loss": 2.4842,
+      "step": 212
+    },
+    {
+      "epoch": 0.4874141876430206,
+      "grad_norm": 2.2343130111694336,
+      "learning_rate": 9.011452945371153e-05,
+      "loss": 2.2786,
+      "step": 213
+    },
+    {
+      "epoch": 0.4897025171624714,
+      "grad_norm": 2.3048455715179443,
+      "learning_rate": 9.000127113956674e-05,
+      "loss": 2.6093,
+      "step": 214
+    },
+    {
+      "epoch": 0.4919908466819222,
+      "grad_norm": 2.0562024116516113,
+      "learning_rate": 8.988743974333405e-05,
+      "loss": 2.746,
+      "step": 215
+    },
+    {
+      "epoch": 0.494279176201373,
+      "grad_norm": 2.252657651901245,
+      "learning_rate": 8.977303689583e-05,
+      "loss": 2.3817,
+      "step": 216
+    },
+    {
+      "epoch": 0.4965675057208238,
+      "grad_norm": 2.080920696258545,
+      "learning_rate": 8.965806423605807e-05,
+      "loss": 2.3491,
+      "step": 217
+    },
+    {
+      "epoch": 0.4988558352402746,
+      "grad_norm": 2.2473278045654297,
+      "learning_rate": 8.954252341118523e-05,
+      "loss": 2.3824,
+      "step": 218
+    },
+    {
+      "epoch": 0.5011441647597255,
+      "grad_norm": 1.9687806367874146,
+      "learning_rate": 8.94264160765183e-05,
+      "loss": 2.3748,
+      "step": 219
+    },
+    {
+      "epoch": 0.5034324942791762,
+      "grad_norm": 6.516245365142822,
+      "learning_rate": 8.930974389548023e-05,
+      "loss": 2.3854,
+      "step": 220
+    },
+    {
+      "epoch": 0.505720823798627,
+      "grad_norm": 2.8898024559020996,
+      "learning_rate": 8.919250853958639e-05,
+      "loss": 2.4238,
+      "step": 221
+    },
+    {
+      "epoch": 0.5080091533180778,
+      "grad_norm": 2.3418991565704346,
+      "learning_rate": 8.90747116884204e-05,
+      "loss": 2.3012,
+      "step": 222
+    },
+    {
+      "epoch": 0.5102974828375286,
+      "grad_norm": 1.9986119270324707,
+      "learning_rate": 8.895635502961033e-05,
+      "loss": 2.3529,
+      "step": 223
+    },
+    {
+      "epoch": 0.5125858123569794,
+      "grad_norm": 2.4174418449401855,
+      "learning_rate": 8.883744025880428e-05,
+      "loss": 2.4234,
+      "step": 224
+    },
+    {
+      "epoch": 0.5148741418764302,
+      "grad_norm": 3.099142074584961,
+      "learning_rate": 8.871796907964625e-05,
+      "loss": 2.4744,
+      "step": 225
+    },
+    {
+      "epoch": 0.517162471395881,
+      "grad_norm": 2.573162317276001,
+      "learning_rate": 8.859794320375168e-05,
+      "loss": 2.4386,
+      "step": 226
+    },
+    {
+      "epoch": 0.5194508009153318,
+      "grad_norm": 2.158879518508911,
+      "learning_rate": 8.847736435068288e-05,
+      "loss": 2.453,
+      "step": 227
+    },
+    {
+      "epoch": 0.5217391304347826,
+      "grad_norm": 2.082505941390991,
+      "learning_rate": 8.835623424792452e-05,
+      "loss": 2.3861,
+      "step": 228
+    },
+    {
+      "epoch": 0.5240274599542334,
+      "grad_norm": 2.180189847946167,
+      "learning_rate": 8.823455463085873e-05,
+      "loss": 2.3747,
+      "step": 229
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 2.4317615032196045,
+      "learning_rate": 8.811232724274035e-05,
+      "loss": 2.4164,
+      "step": 230
+    },
+    {
+      "epoch": 0.528604118993135,
+      "grad_norm": 2.0874435901641846,
+      "learning_rate": 8.798955383467189e-05,
+      "loss": 2.3647,
+      "step": 231
+    },
+    {
+      "epoch": 0.5308924485125858,
+      "grad_norm": 2.1150388717651367,
+      "learning_rate": 8.786623616557847e-05,
+      "loss": 2.3115,
+      "step": 232
+    },
+    {
+      "epoch": 0.5331807780320366,
+      "grad_norm": 2.2048499584198,
+      "learning_rate": 8.774237600218266e-05,
+      "loss": 2.5097,
+      "step": 233
+    },
+    {
+      "epoch": 0.5354691075514875,
+      "grad_norm": 2.2788608074188232,
+      "learning_rate": 8.761797511897906e-05,
+      "loss": 2.2931,
+      "step": 234
+    },
+    {
+      "epoch": 0.5377574370709383,
+      "grad_norm": 2.0588808059692383,
+      "learning_rate": 8.749303529820903e-05,
+      "loss": 2.4971,
+      "step": 235
+    },
+    {
+      "epoch": 0.540045766590389,
+      "grad_norm": 2.104788303375244,
+      "learning_rate": 8.736755832983497e-05,
+      "loss": 2.3433,
+      "step": 236
+    },
+    {
+      "epoch": 0.5423340961098398,
+      "grad_norm": 1.8593131303787231,
+      "learning_rate": 8.724154601151484e-05,
+      "loss": 2.285,
+      "step": 237
+    },
+    {
+      "epoch": 0.5446224256292906,
+      "grad_norm": 1.9205878973007202,
+      "learning_rate": 8.711500014857634e-05,
+      "loss": 2.1847,
+      "step": 238
+    },
+    {
+      "epoch": 0.5469107551487414,
+      "grad_norm": 2.350421905517578,
+      "learning_rate": 8.698792255399104e-05,
+      "loss": 2.3161,
+      "step": 239
+    },
+    {
+      "epoch": 0.5491990846681922,
+      "grad_norm": 2.14412784576416,
+      "learning_rate": 8.686031504834843e-05,
+      "loss": 2.2517,
+      "step": 240
+    },
+    {
+      "epoch": 0.551487414187643,
+      "grad_norm": 1.8842304944992065,
+      "learning_rate": 8.673217945982978e-05,
+      "loss": 2.36,
+      "step": 241
+    },
+    {
+      "epoch": 0.5537757437070938,
+      "grad_norm": 1.9713618755340576,
+      "learning_rate": 8.660351762418203e-05,
+      "loss": 2.194,
+      "step": 242
+    },
+    {
+      "epoch": 0.5560640732265446,
+      "grad_norm": 1.9737403392791748,
+      "learning_rate": 8.647433138469144e-05,
+      "loss": 2.4538,
+      "step": 243
+    },
+    {
+      "epoch": 0.5583524027459954,
+      "grad_norm": 2.11395525932312,
+      "learning_rate": 8.634462259215719e-05,
+      "loss": 2.4534,
+      "step": 244
+    },
+    {
+      "epoch": 0.5606407322654462,
+      "grad_norm": 2.621248245239258,
+      "learning_rate": 8.621439310486486e-05,
+      "loss": 2.5009,
+      "step": 245
+    },
+    {
+      "epoch": 0.562929061784897,
+      "grad_norm": 2.1076266765594482,
+      "learning_rate": 8.608364478855983e-05,
+      "loss": 2.2991,
+      "step": 246
+    },
+    {
+      "epoch": 0.5652173913043478,
+      "grad_norm": 2.0468740463256836,
+      "learning_rate": 8.595237951642055e-05,
+      "loss": 2.4584,
+      "step": 247
+    },
+    {
+      "epoch": 0.5675057208237986,
+      "grad_norm": 2.757575750350952,
+      "learning_rate": 8.58205991690316e-05,
+      "loss": 2.3874,
+      "step": 248
+    },
+    {
+      "epoch": 0.5697940503432495,
+      "grad_norm": 2.700618267059326,
+      "learning_rate": 8.568830563435694e-05,
+      "loss": 2.2581,
+      "step": 249
+    },
+    {
+      "epoch": 0.5720823798627003,
+      "grad_norm": 2.136777400970459,
+      "learning_rate": 8.555550080771273e-05,
+      "loss": 2.3313,
+      "step": 250
+    },
+    {
+      "epoch": 0.5743707093821511,
+      "grad_norm": 2.2393014430999756,
+      "learning_rate": 8.542218659174017e-05,
+      "loss": 2.3745,
+      "step": 251
+    },
+    {
+      "epoch": 0.5766590389016019,
+      "grad_norm": 1.9977840185165405,
+      "learning_rate": 8.528836489637828e-05,
+      "loss": 2.4189,
+      "step": 252
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 1.8844605684280396,
+      "learning_rate": 8.515403763883659e-05,
+      "loss": 2.2503,
+      "step": 253
+    },
+    {
+      "epoch": 0.5812356979405034,
+      "grad_norm": 2.225182056427002,
+      "learning_rate": 8.501920674356754e-05,
+      "loss": 2.3373,
+      "step": 254
+    },
+    {
+      "epoch": 0.5835240274599542,
+      "grad_norm": 3.804654598236084,
+      "learning_rate": 8.488387414223904e-05,
+      "loss": 2.4731,
+      "step": 255
+    },
+    {
+      "epoch": 0.585812356979405,
+      "grad_norm": 2.7391419410705566,
+      "learning_rate": 8.47480417737067e-05,
+      "loss": 2.3385,
+      "step": 256
+    },
+    {
+      "epoch": 0.5881006864988558,
+      "grad_norm": 1.8760191202163696,
+      "learning_rate": 8.461171158398612e-05,
+      "loss": 2.2315,
+      "step": 257
+    },
+    {
+      "epoch": 0.5903890160183066,
+      "grad_norm": 2.5383033752441406,
+      "learning_rate": 8.447488552622498e-05,
+      "loss": 2.2305,
+      "step": 258
+    },
+    {
+      "epoch": 0.5926773455377574,
+      "grad_norm": 2.4613940715789795,
+      "learning_rate": 8.433756556067506e-05,
+      "loss": 2.3998,
+      "step": 259
+    },
+    {
+      "epoch": 0.5949656750572082,
+      "grad_norm": 2.1040103435516357,
+      "learning_rate": 8.419975365466415e-05,
+      "loss": 2.4298,
+      "step": 260
+    },
+    {
+      "epoch": 0.597254004576659,
+      "grad_norm": 2.3999195098876953,
+      "learning_rate": 8.406145178256788e-05,
+      "loss": 2.3064,
+      "step": 261
+    },
+    {
+      "epoch": 0.5995423340961098,
+      "grad_norm": 2.633197069168091,
+      "learning_rate": 8.392266192578143e-05,
+      "loss": 2.4663,
+      "step": 262
+    },
+    {
+      "epoch": 0.6018306636155606,
+      "grad_norm": 2.4251699447631836,
+      "learning_rate": 8.37833860726911e-05,
+      "loss": 2.3587,
+      "step": 263
+    },
+    {
+      "epoch": 0.6041189931350115,
+      "grad_norm": 1.8132442235946655,
+      "learning_rate": 8.364362621864595e-05,
+      "loss": 2.3555,
+      "step": 264
+    },
+    {
+      "epoch": 0.6064073226544623,
+      "grad_norm": 2.6563608646392822,
+      "learning_rate": 8.350338436592905e-05,
+      "loss": 2.2702,
+      "step": 265
+    },
+    {
+      "epoch": 0.6086956521739131,
+      "grad_norm": 2.4801723957061768,
+      "learning_rate": 8.336266252372889e-05,
+      "loss": 2.266,
+      "step": 266
+    },
+    {
+      "epoch": 0.6109839816933639,
+      "grad_norm": 2.3440706729888916,
+      "learning_rate": 8.322146270811059e-05,
+      "loss": 2.3181,
+      "step": 267
+    },
+    {
+      "epoch": 0.6132723112128147,
+      "grad_norm": 2.256977081298828,
+      "learning_rate": 8.307978694198699e-05,
+      "loss": 2.2594,
+      "step": 268
+    },
+    {
+      "epoch": 0.6155606407322655,
+      "grad_norm": 2.4283649921417236,
+      "learning_rate": 8.293763725508969e-05,
+      "loss": 2.171,
+      "step": 269
+    },
+    {
+      "epoch": 0.6178489702517163,
+      "grad_norm": 1.9902546405792236,
+      "learning_rate": 8.279501568393994e-05,
+      "loss": 2.3791,
+      "step": 270
+    },
+    {
+      "epoch": 0.620137299771167,
+      "grad_norm": 2.388878583908081,
+      "learning_rate": 8.265192427181954e-05,
+      "loss": 2.2571,
+      "step": 271
+    },
+    {
+      "epoch": 0.6224256292906178,
+      "grad_norm": 2.6152727603912354,
+      "learning_rate": 8.250836506874142e-05,
+      "loss": 2.3477,
+      "step": 272
+    },
+    {
+      "epoch": 0.6247139588100686,
+      "grad_norm": 1.8113371133804321,
+      "learning_rate": 8.236434013142045e-05,
+      "loss": 2.1395,
+      "step": 273
+    },
+    {
+      "epoch": 0.6270022883295194,
+      "grad_norm": 9.44183349609375,
+      "learning_rate": 8.221985152324385e-05,
+      "loss": 2.4124,
+      "step": 274
+    },
+    {
+      "epoch": 0.6292906178489702,
+      "grad_norm": 2.5722267627716064,
+      "learning_rate": 8.207490131424167e-05,
+      "loss": 2.3806,
+      "step": 275
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 2.5060784816741943,
+      "learning_rate": 8.192949158105713e-05,
+      "loss": 2.4453,
+      "step": 276
+    },
+    {
+      "epoch": 0.6338672768878718,
+      "grad_norm": 1.9694451093673706,
+      "learning_rate": 8.178362440691685e-05,
+      "loss": 2.2709,
+      "step": 277
+    },
+    {
+      "epoch": 0.6361556064073226,
+      "grad_norm": 2.1653637886047363,
+      "learning_rate": 8.163730188160105e-05,
+      "loss": 2.2756,
+      "step": 278
+    },
+    {
+      "epoch": 0.6384439359267735,
+      "grad_norm": 2.000991106033325,
+      "learning_rate": 8.149052610141357e-05,
+      "loss": 2.3256,
+      "step": 279
+    },
+    {
+      "epoch": 0.6407322654462243,
+      "grad_norm": 1.9745527505874634,
+      "learning_rate": 8.134329916915184e-05,
+      "loss": 2.1619,
+      "step": 280
+    },
+    {
+      "epoch": 0.6430205949656751,
+      "grad_norm": 1.7634015083312988,
+      "learning_rate": 8.119562319407679e-05,
+      "loss": 2.3815,
+      "step": 281
+    },
+    {
+      "epoch": 0.6453089244851259,
+      "grad_norm": 2.338313341140747,
+      "learning_rate": 8.104750029188257e-05,
+      "loss": 2.2965,
+      "step": 282
+    },
+    {
+      "epoch": 0.6475972540045767,
+      "grad_norm": 5.942075729370117,
+      "learning_rate": 8.089893258466632e-05,
+      "loss": 2.2091,
+      "step": 283
+    },
+    {
+      "epoch": 0.6498855835240275,
+      "grad_norm": 2.5060672760009766,
+      "learning_rate": 8.074992220089769e-05,
+      "loss": 2.4674,
+      "step": 284
+    },
+    {
+      "epoch": 0.6521739130434783,
+      "grad_norm": 2.6582159996032715,
+      "learning_rate": 8.060047127538835e-05,
+      "loss": 2.3614,
+      "step": 285
+    },
+    {
+      "epoch": 0.6544622425629291,
+      "grad_norm": 2.005200147628784,
+      "learning_rate": 8.045058194926153e-05,
+      "loss": 2.2056,
+      "step": 286
+    },
+    {
+      "epoch": 0.6567505720823799,
+      "grad_norm": 2.410688877105713,
+      "learning_rate": 8.030025636992113e-05,
+      "loss": 2.2247,
+      "step": 287
+    },
+    {
+      "epoch": 0.6590389016018307,
+      "grad_norm": 1.9305633306503296,
+      "learning_rate": 8.014949669102117e-05,
+      "loss": 2.0635,
+      "step": 288
+    },
+    {
+      "epoch": 0.6613272311212814,
+      "grad_norm": 2.0201923847198486,
+      "learning_rate": 7.999830507243478e-05,
+      "loss": 2.1299,
+      "step": 289
+    },
+    {
+      "epoch": 0.6636155606407322,
+      "grad_norm": 1.8325549364089966,
+      "learning_rate": 7.984668368022335e-05,
+      "loss": 2.1106,
+      "step": 290
+    },
+    {
+      "epoch": 0.665903890160183,
+      "grad_norm": 2.2241806983947754,
+      "learning_rate": 7.969463468660545e-05,
+      "loss": 2.2576,
+      "step": 291
+    },
+    {
+      "epoch": 0.6681922196796338,
+      "grad_norm": 2.101597547531128,
+      "learning_rate": 7.954216026992571e-05,
+      "loss": 2.3561,
+      "step": 292
+    },
+    {
+      "epoch": 0.6704805491990846,
+      "grad_norm": 2.0303845405578613,
+      "learning_rate": 7.938926261462366e-05,
+      "loss": 2.1977,
+      "step": 293
+    },
+    {
+      "epoch": 0.6727688787185355,
+      "grad_norm": 2.016955852508545,
+      "learning_rate": 7.923594391120236e-05,
+      "loss": 2.3022,
+      "step": 294
+    },
+    {
+      "epoch": 0.6750572082379863,
+      "grad_norm": 1.99684476852417,
+      "learning_rate": 7.908220635619708e-05,
+      "loss": 2.1994,
+      "step": 295
+    },
+    {
+      "epoch": 0.6773455377574371,
+      "grad_norm": 2.2181754112243652,
+      "learning_rate": 7.892805215214381e-05,
+      "loss": 2.2126,
+      "step": 296
+    },
+    {
+      "epoch": 0.6796338672768879,
+      "grad_norm": 2.0173442363739014,
+      "learning_rate": 7.877348350754767e-05,
+      "loss": 2.2134,
+      "step": 297
+    },
+    {
+      "epoch": 0.6819221967963387,
+      "grad_norm": 1.8784480094909668,
+      "learning_rate": 7.861850263685134e-05,
+      "loss": 2.3381,
+      "step": 298
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 2.2438318729400635,
+      "learning_rate": 7.846311176040331e-05,
+      "loss": 2.3408,
+      "step": 299
+    },
+    {
+      "epoch": 0.6864988558352403,
+      "grad_norm": 2.977372884750366,
+      "learning_rate": 7.830731310442599e-05,
+      "loss": 2.2358,
+      "step": 300
+    },
+    {
+      "epoch": 0.6887871853546911,
+      "grad_norm": 2.1407394409179688,
+      "learning_rate": 7.815110890098397e-05,
+      "loss": 2.2125,
+      "step": 301
+    },
+    {
+      "epoch": 0.6910755148741419,
+      "grad_norm": 1.6845601797103882,
+      "learning_rate": 7.799450138795185e-05,
+      "loss": 2.0579,
+      "step": 302
+    },
+    {
+      "epoch": 0.6933638443935927,
+      "grad_norm": 1.834087610244751,
+      "learning_rate": 7.78374928089824e-05,
+      "loss": 2.3022,
+      "step": 303
+    },
+    {
+      "epoch": 0.6956521739130435,
+      "grad_norm": 1.8871235847473145,
+      "learning_rate": 7.768008541347423e-05,
+      "loss": 2.2783,
+      "step": 304
+    },
+    {
+      "epoch": 0.6979405034324943,
+      "grad_norm": 1.9850207567214966,
+      "learning_rate": 7.752228145653964e-05,
+      "loss": 2.3613,
+      "step": 305
+    },
+    {
+      "epoch": 0.700228832951945,
+      "grad_norm": 1.9508417844772339,
+      "learning_rate": 7.73640831989723e-05,
+      "loss": 2.3295,
+      "step": 306
+    },
+    {
+      "epoch": 0.7025171624713958,
+      "grad_norm": 2.0530943870544434,
+      "learning_rate": 7.72054929072149e-05,
+      "loss": 2.2485,
+      "step": 307
+    },
+    {
+      "epoch": 0.7048054919908466,
+      "grad_norm": 2.0443074703216553,
+      "learning_rate": 7.704651285332663e-05,
+      "loss": 2.1668,
+      "step": 308
+    },
+    {
+      "epoch": 0.7070938215102975,
+      "grad_norm": 2.0727176666259766,
+      "learning_rate": 7.688714531495061e-05,
+      "loss": 2.1517,
+      "step": 309
+    },
+    {
+      "epoch": 0.7093821510297483,
+      "grad_norm": 2.033459186553955,
+      "learning_rate": 7.672739257528134e-05,
+      "loss": 2.2783,
+      "step": 310
+    },
+    {
+      "epoch": 0.7116704805491991,
+      "grad_norm": 2.2704668045043945,
+      "learning_rate": 7.656725692303195e-05,
+      "loss": 2.3378,
+      "step": 311
+    },
+    {
+      "epoch": 0.7139588100686499,
+      "grad_norm": 2.021702766418457,
+      "learning_rate": 7.640674065240136e-05,
+      "loss": 2.1894,
+      "step": 312
+    },
+    {
+      "epoch": 0.7162471395881007,
+      "grad_norm": 2.0539019107818604,
+      "learning_rate": 7.624584606304147e-05,
+      "loss": 2.2705,
+      "step": 313
+    },
+    {
+      "epoch": 0.7185354691075515,
+      "grad_norm": 1.6023483276367188,
+      "learning_rate": 7.608457546002424e-05,
+      "loss": 2.1957,
+      "step": 314
+    },
+    {
+      "epoch": 0.7208237986270023,
+      "grad_norm": 1.9021177291870117,
+      "learning_rate": 7.592293115380855e-05,
+      "loss": 2.1939,
+      "step": 315
+    },
+    {
+      "epoch": 0.7231121281464531,
+      "grad_norm": 1.7106837034225464,
+      "learning_rate": 7.576091546020725e-05,
+      "loss": 2.1383,
+      "step": 316
+    },
+    {
+      "epoch": 0.7254004576659039,
+      "grad_norm": 2.3384652137756348,
+      "learning_rate": 7.559853070035389e-05,
+      "loss": 2.1954,
+      "step": 317
+    },
+    {
+      "epoch": 0.7276887871853547,
+      "grad_norm": 2.0455210208892822,
+      "learning_rate": 7.543577920066944e-05,
+      "loss": 2.2671,
+      "step": 318
+    },
+    {
+      "epoch": 0.7299771167048055,
+      "grad_norm": 2.6084513664245605,
+      "learning_rate": 7.527266329282905e-05,
+      "loss": 2.1935,
+      "step": 319
+    },
+    {
+      "epoch": 0.7322654462242563,
+      "grad_norm": 1.9567663669586182,
+      "learning_rate": 7.510918531372857e-05,
+      "loss": 2.3201,
+      "step": 320
+    },
+    {
+      "epoch": 0.7345537757437071,
+      "grad_norm": 2.3495213985443115,
+      "learning_rate": 7.494534760545113e-05,
+      "loss": 2.2716,
+      "step": 321
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 2.139066696166992,
+      "learning_rate": 7.478115251523352e-05,
+      "loss": 2.3472,
+      "step": 322
+    },
+    {
+      "epoch": 0.7391304347826086,
+      "grad_norm": 1.9608685970306396,
+      "learning_rate": 7.461660239543261e-05,
+      "loss": 2.2516,
+      "step": 323
+    },
+    {
+      "epoch": 0.7414187643020596,
+      "grad_norm": 1.9583091735839844,
+      "learning_rate": 7.445169960349167e-05,
+      "loss": 2.0278,
+      "step": 324
+    },
+    {
+      "epoch": 0.7437070938215103,
+      "grad_norm": 1.9751293659210205,
+      "learning_rate": 7.42864465019065e-05,
+      "loss": 2.111,
+      "step": 325
+    },
+    {
+      "epoch": 0.7459954233409611,
+      "grad_norm": 1.8538637161254883,
+      "learning_rate": 7.412084545819168e-05,
+      "loss": 2.3414,
+      "step": 326
+    },
+    {
+      "epoch": 0.7482837528604119,
+      "grad_norm": 2.0090718269348145,
+      "learning_rate": 7.395489884484665e-05,
+      "loss": 2.3156,
+      "step": 327
+    },
+    {
+      "epoch": 0.7505720823798627,
+      "grad_norm": 1.8915284872055054,
+      "learning_rate": 7.378860903932159e-05,
+      "loss": 2.2654,
+      "step": 328
+    },
+    {
+      "epoch": 0.7528604118993135,
+      "grad_norm": 1.6930265426635742,
+      "learning_rate": 7.362197842398355e-05,
+      "loss": 2.0445,
+      "step": 329
+    },
+    {
+      "epoch": 0.7551487414187643,
+      "grad_norm": 1.9732853174209595,
+      "learning_rate": 7.34550093860822e-05,
+      "loss": 2.2913,
+      "step": 330
+    },
+    {
+      "epoch": 0.7574370709382151,
+      "grad_norm": 1.865723967552185,
+      "learning_rate": 7.32877043177156e-05,
+      "loss": 2.0363,
+      "step": 331
+    },
+    {
+      "epoch": 0.7597254004576659,
+      "grad_norm": 2.0198216438293457,
+      "learning_rate": 7.31200656157961e-05,
+      "loss": 2.1827,
+      "step": 332
+    },
+    {
+      "epoch": 0.7620137299771167,
+      "grad_norm": 1.8572312593460083,
+      "learning_rate": 7.295209568201574e-05,
+      "loss": 2.3125,
+      "step": 333
+    },
+    {
+      "epoch": 0.7643020594965675,
+      "grad_norm": 1.9783533811569214,
+      "learning_rate": 7.278379692281208e-05,
+      "loss": 2.0752,
+      "step": 334
+    },
+    {
+      "epoch": 0.7665903890160183,
+      "grad_norm": 1.9235711097717285,
+      "learning_rate": 7.261517174933362e-05,
+      "loss": 2.2604,
+      "step": 335
+    },
+    {
+      "epoch": 0.7688787185354691,
+      "grad_norm": 1.8910268545150757,
+      "learning_rate": 7.244622257740523e-05,
+      "loss": 2.5354,
+      "step": 336
+    },
+    {
+      "epoch": 0.7711670480549199,
+      "grad_norm": 1.8306984901428223,
+      "learning_rate": 7.227695182749361e-05,
+      "loss": 2.3317,
+      "step": 337
+    },
+    {
+      "epoch": 0.7734553775743707,
+      "grad_norm": 1.7641491889953613,
+      "learning_rate": 7.210736192467256e-05,
+      "loss": 2.1461,
+      "step": 338
+    },
+    {
+      "epoch": 0.7757437070938215,
+      "grad_norm": 2.0800931453704834,
+      "learning_rate": 7.193745529858826e-05,
+      "loss": 2.3707,
+      "step": 339
+    },
+    {
+      "epoch": 0.7780320366132724,
+      "grad_norm": 2.155435800552368,
+      "learning_rate": 7.176723438342446e-05,
+      "loss": 2.3194,
+      "step": 340
+    },
+    {
+      "epoch": 0.7803203661327232,
+      "grad_norm": 1.8292967081069946,
+      "learning_rate": 7.159670161786759e-05,
+      "loss": 2.2795,
+      "step": 341
+    },
+    {
+      "epoch": 0.782608695652174,
+      "grad_norm": 2.0984854698181152,
+      "learning_rate": 7.142585944507185e-05,
+      "loss": 2.2426,
+      "step": 342
+    },
+    {
+      "epoch": 0.7848970251716247,
+      "grad_norm": 2.123730421066284,
+      "learning_rate": 7.125471031262417e-05,
+      "loss": 2.0311,
+      "step": 343
+    },
+    {
+      "epoch": 0.7871853546910755,
+      "grad_norm": 2.2667717933654785,
+      "learning_rate": 7.10832566725092e-05,
+      "loss": 2.1289,
+      "step": 344
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 1.7086976766586304,
+      "learning_rate": 7.091150098107414e-05,
+      "loss": 2.4501,
+      "step": 345
+    },
+    {
+      "epoch": 0.7917620137299771,
+      "grad_norm": 2.096912384033203,
+      "learning_rate": 7.073944569899354e-05,
+      "loss": 2.2119,
+      "step": 346
+    },
+    {
+      "epoch": 0.7940503432494279,
+      "grad_norm": 1.7574291229248047,
+      "learning_rate": 7.056709329123408e-05,
+      "loss": 2.2372,
+      "step": 347
+    },
+    {
+      "epoch": 0.7963386727688787,
+      "grad_norm": 1.718612790107727,
+      "learning_rate": 7.039444622701922e-05,
+      "loss": 2.2722,
+      "step": 348
+    },
+    {
+      "epoch": 0.7986270022883295,
+      "grad_norm": 2.0621182918548584,
+      "learning_rate": 7.022150697979384e-05,
+      "loss": 1.9942,
+      "step": 349
+    },
+    {
+      "epoch": 0.8009153318077803,
+      "grad_norm": 2.8785934448242188,
+      "learning_rate": 7.00482780271889e-05,
+      "loss": 2.2271,
+      "step": 350
+    },
+    {
+      "epoch": 0.8032036613272311,
+      "grad_norm": 1.8720818758010864,
+      "learning_rate": 6.98747618509857e-05,
+      "loss": 2.1045,
+      "step": 351
+    },
+    {
+      "epoch": 0.8054919908466819,
+      "grad_norm": 2.2784407138824463,
+      "learning_rate": 6.97009609370806e-05,
+      "loss": 2.22,
+      "step": 352
+    },
+    {
+      "epoch": 0.8077803203661327,
+      "grad_norm": 1.8243550062179565,
+      "learning_rate": 6.952687777544922e-05,
+      "loss": 2.1709,
+      "step": 353
+    },
+    {
+      "epoch": 0.8100686498855835,
+      "grad_norm": 1.965592384338379,
+      "learning_rate": 6.935251486011087e-05,
+      "loss": 2.2081,
+      "step": 354
+    },
+    {
+      "epoch": 0.8123569794050344,
+      "grad_norm": 1.8083536624908447,
+      "learning_rate": 6.917787468909271e-05,
+      "loss": 2.0548,
+      "step": 355
+    },
+    {
+      "epoch": 0.8146453089244852,
+      "grad_norm": 2.135918140411377,
+      "learning_rate": 6.900295976439413e-05,
+      "loss": 2.1584,
+      "step": 356
+    },
+    {
+      "epoch": 0.816933638443936,
+      "grad_norm": 1.90550696849823,
+      "learning_rate": 6.882777259195071e-05,
+      "loss": 2.0983,
+      "step": 357
+    },
+    {
+      "epoch": 0.8192219679633868,
+      "grad_norm": 2.0657262802124023,
+      "learning_rate": 6.865231568159846e-05,
+      "loss": 2.2419,
+      "step": 358
+    },
+    {
+      "epoch": 0.8215102974828375,
+      "grad_norm": 1.8104684352874756,
+      "learning_rate": 6.847659154703785e-05,
+      "loss": 2.0536,
+      "step": 359
+    },
+    {
+      "epoch": 0.8237986270022883,
+      "grad_norm": 1.855124831199646,
+      "learning_rate": 6.830060270579768e-05,
+      "loss": 2.2918,
+      "step": 360
+    },
+    {
+      "epoch": 0.8260869565217391,
+      "grad_norm": 1.8047131299972534,
+      "learning_rate": 6.812435167919918e-05,
+      "loss": 2.2204,
+      "step": 361
+    },
+    {
+      "epoch": 0.8283752860411899,
+      "grad_norm": 1.9419913291931152,
+      "learning_rate": 6.794784099231972e-05,
+      "loss": 2.1344,
+      "step": 362
+    },
+    {
+      "epoch": 0.8306636155606407,
+      "grad_norm": 2.067949056625366,
+      "learning_rate": 6.777107317395679e-05,
+      "loss": 2.1645,
+      "step": 363
+    },
+    {
+      "epoch": 0.8329519450800915,
+      "grad_norm": 1.7515795230865479,
+      "learning_rate": 6.759405075659166e-05,
+      "loss": 2.585,
+      "step": 364
+    },
+    {
+      "epoch": 0.8352402745995423,
+      "grad_norm": 1.828723430633545,
+      "learning_rate": 6.741677627635317e-05,
+      "loss": 2.1944,
+      "step": 365
+    },
+    {
+      "epoch": 0.8375286041189931,
+      "grad_norm": 2.222910165786743,
+      "learning_rate": 6.723925227298132e-05,
+      "loss": 2.1099,
+      "step": 366
+    },
+    {
+      "epoch": 0.8398169336384439,
+      "grad_norm": 1.6402628421783447,
+      "learning_rate": 6.706148128979095e-05,
+      "loss": 2.3331,
+      "step": 367
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 2.122103214263916,
+      "learning_rate": 6.688346587363533e-05,
+      "loss": 2.167,
+      "step": 368
+    },
+    {
+      "epoch": 0.8443935926773455,
+      "grad_norm": 1.705405592918396,
+      "learning_rate": 6.67052085748695e-05,
+      "loss": 2.1447,
+      "step": 369
+    },
+    {
+      "epoch": 0.8466819221967964,
+      "grad_norm": 1.9684945344924927,
+      "learning_rate": 6.652671194731396e-05,
+      "loss": 2.1688,
+      "step": 370
+    },
+    {
+      "epoch": 0.8489702517162472,
+      "grad_norm": 1.9052337408065796,
+      "learning_rate": 6.634797854821795e-05,
+      "loss": 2.1049,
+      "step": 371
+    },
+    {
+      "epoch": 0.851258581235698,
+      "grad_norm": 1.8532124757766724,
+      "learning_rate": 6.616901093822283e-05,
+      "loss": 2.2438,
+      "step": 372
+    },
+    {
+      "epoch": 0.8535469107551488,
+      "grad_norm": 1.8264321088790894,
+      "learning_rate": 6.598981168132539e-05,
+      "loss": 2.1135,
+      "step": 373
+    },
+    {
+      "epoch": 0.8558352402745996,
+      "grad_norm": 1.6481093168258667,
+      "learning_rate": 6.58103833448412e-05,
+      "loss": 2.1073,
+      "step": 374
+    },
+    {
+      "epoch": 0.8581235697940504,
+      "grad_norm": 1.8912767171859741,
+      "learning_rate": 6.563072849936766e-05,
+      "loss": 2.0309,
+      "step": 375
+    },
+    {
+      "epoch": 0.8604118993135011,
+      "grad_norm": 1.7808010578155518,
+      "learning_rate": 6.545084971874738e-05,
+      "loss": 2.3258,
+      "step": 376
+    },
+    {
+      "epoch": 0.8627002288329519,
+      "grad_norm": 1.9393279552459717,
+      "learning_rate": 6.527074958003109e-05,
+      "loss": 2.1877,
+      "step": 377
+    },
+    {
+      "epoch": 0.8649885583524027,
+      "grad_norm": 1.8000222444534302,
+      "learning_rate": 6.509043066344092e-05,
+      "loss": 2.172,
+      "step": 378
+    },
+    {
+      "epoch": 0.8672768878718535,
+      "grad_norm": 1.6351038217544556,
+      "learning_rate": 6.490989555233327e-05,
+      "loss": 1.9942,
+      "step": 379
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 1.7084710597991943,
+      "learning_rate": 6.472914683316195e-05,
+      "loss": 2.0928,
+      "step": 380
+    },
+    {
+      "epoch": 0.8718535469107551,
+      "grad_norm": 2.094414234161377,
+      "learning_rate": 6.454818709544097e-05,
+      "loss": 2.1815,
+      "step": 381
+    },
+    {
+      "epoch": 0.8741418764302059,
+      "grad_norm": 2.299128293991089,
+      "learning_rate": 6.436701893170756e-05,
+      "loss": 2.5142,
+      "step": 382
+    },
+    {
+      "epoch": 0.8764302059496567,
+      "grad_norm": 2.0357043743133545,
+      "learning_rate": 6.4185644937485e-05,
+      "loss": 2.2609,
+      "step": 383
+    },
+    {
+      "epoch": 0.8787185354691075,
+      "grad_norm": 1.9757893085479736,
+      "learning_rate": 6.400406771124536e-05,
+      "loss": 2.0867,
+      "step": 384
+    },
+    {
+      "epoch": 0.8810068649885584,
+      "grad_norm": 1.849135160446167,
+      "learning_rate": 6.382228985437237e-05,
+      "loss": 2.2756,
+      "step": 385
+    },
+    {
+      "epoch": 0.8832951945080092,
+      "grad_norm": 2.372392177581787,
+      "learning_rate": 6.364031397112416e-05,
+      "loss": 2.1671,
+      "step": 386
+    },
+    {
+      "epoch": 0.88558352402746,
+      "grad_norm": 1.765740156173706,
+      "learning_rate": 6.345814266859581e-05,
+      "loss": 1.9694,
+      "step": 387
+    },
+    {
+      "epoch": 0.8878718535469108,
+      "grad_norm": 2.085186719894409,
+      "learning_rate": 6.327577855668216e-05,
+      "loss": 2.1867,
+      "step": 388
+    },
+    {
+      "epoch": 0.8901601830663616,
+      "grad_norm": 2.0762360095977783,
+      "learning_rate": 6.309322424804034e-05,
+      "loss": 2.3435,
+      "step": 389
+    },
+    {
+      "epoch": 0.8924485125858124,
+      "grad_norm": 1.8886313438415527,
+      "learning_rate": 6.291048235805234e-05,
+      "loss": 1.9982,
+      "step": 390
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 1.963932752609253,
+      "learning_rate": 6.272755550478757e-05,
+      "loss": 2.1239,
+      "step": 391
+    },
+    {
+      "epoch": 0.897025171624714,
+      "grad_norm": 1.6394884586334229,
+      "learning_rate": 6.254444630896529e-05,
+      "loss": 2.2411,
+      "step": 392
+    },
+    {
+      "epoch": 0.8993135011441648,
+      "grad_norm": 2.1238386631011963,
+      "learning_rate": 6.236115739391716e-05,
+      "loss": 2.1889,
+      "step": 393
+    },
+    {
+      "epoch": 0.9016018306636155,
+      "grad_norm": 1.8659511804580688,
+      "learning_rate": 6.21776913855496e-05,
+      "loss": 1.9483,
+      "step": 394
+    },
+    {
+      "epoch": 0.9038901601830663,
+      "grad_norm": 1.7186646461486816,
+      "learning_rate": 6.199405091230614e-05,
+      "loss": 1.8723,
+      "step": 395
+    },
+    {
+      "epoch": 0.9061784897025171,
+      "grad_norm": 2.3829565048217773,
+      "learning_rate": 6.181023860512984e-05,
+      "loss": 2.188,
+      "step": 396
+    },
+    {
+      "epoch": 0.9084668192219679,
+      "grad_norm": 1.926705002784729,
+      "learning_rate": 6.162625709742551e-05,
+      "loss": 2.1078,
+      "step": 397
+    },
+    {
+      "epoch": 0.9107551487414187,
+      "grad_norm": 1.8216861486434937,
+      "learning_rate": 6.144210902502207e-05,
+      "loss": 2.0508,
+      "step": 398
+    },
+    {
+      "epoch": 0.9130434782608695,
+      "grad_norm": 1.8305362462997437,
+      "learning_rate": 6.125779702613471e-05,
+      "loss": 2.0667,
+      "step": 399
+    },
+    {
+      "epoch": 0.9153318077803204,
+      "grad_norm": 1.6279104948043823,
+      "learning_rate": 6.107332374132715e-05,
+      "loss": 2.0211,
+      "step": 400
+    },
+    {
+      "epoch": 0.9176201372997712,
+      "grad_norm": 1.9483661651611328,
+      "learning_rate": 6.088869181347379e-05,
+      "loss": 1.9574,
+      "step": 401
+    },
+    {
+      "epoch": 0.919908466819222,
+      "grad_norm": 1.8065564632415771,
+      "learning_rate": 6.0703903887721837e-05,
+      "loss": 2.1856,
+      "step": 402
+    },
+    {
+      "epoch": 0.9221967963386728,
+      "grad_norm": 1.7617079019546509,
+      "learning_rate": 6.051896261145341e-05,
+      "loss": 2.0909,
+      "step": 403
+    },
+    {
+      "epoch": 0.9244851258581236,
+      "grad_norm": 2.038517951965332,
+      "learning_rate": 6.0333870634247645e-05,
+      "loss": 2.0191,
+      "step": 404
+    },
+    {
+      "epoch": 0.9267734553775744,
+      "grad_norm": 1.9872255325317383,
+      "learning_rate": 6.0148630607842706e-05,
+      "loss": 2.0042,
+      "step": 405
+    },
+    {
+      "epoch": 0.9290617848970252,
+      "grad_norm": 1.7905998229980469,
+      "learning_rate": 5.9963245186097725e-05,
+      "loss": 2.5172,
+      "step": 406
+    },
+    {
+      "epoch": 0.931350114416476,
+      "grad_norm": 1.7833902835845947,
+      "learning_rate": 5.977771702495497e-05,
+      "loss": 2.2731,
+      "step": 407
+    },
+    {
+      "epoch": 0.9336384439359268,
+      "grad_norm": 1.748382806777954,
+      "learning_rate": 5.95920487824016e-05,
+      "loss": 2.2315,
+      "step": 408
+    },
+    {
+      "epoch": 0.9359267734553776,
+      "grad_norm": 1.7928653955459595,
+      "learning_rate": 5.940624311843169e-05,
+      "loss": 1.9035,
+      "step": 409
+    },
+    {
+      "epoch": 0.9382151029748284,
+      "grad_norm": 1.7036387920379639,
+      "learning_rate": 5.922030269500809e-05,
+      "loss": 1.9166,
+      "step": 410
+    },
+    {
+      "epoch": 0.9405034324942791,
+      "grad_norm": 2.600257396697998,
+      "learning_rate": 5.9034230176024316e-05,
+      "loss": 1.9559,
+      "step": 411
+    },
+    {
+      "epoch": 0.9427917620137299,
+      "grad_norm": 1.8736798763275146,
+      "learning_rate": 5.8848028227266325e-05,
+      "loss": 1.9929,
+      "step": 412
+    },
+    {
+      "epoch": 0.9450800915331807,
+      "grad_norm": 1.917090892791748,
+      "learning_rate": 5.866169951637439e-05,
+      "loss": 2.1785,
+      "step": 413
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 2.0305333137512207,
+      "learning_rate": 5.847524671280484e-05,
+      "loss": 2.1782,
+      "step": 414
+    },
+    {
+      "epoch": 0.9496567505720824,
+      "grad_norm": 1.9978338479995728,
+      "learning_rate": 5.8288672487791854e-05,
+      "loss": 2.1131,
+      "step": 415
+    },
+    {
+      "epoch": 0.9519450800915332,
+      "grad_norm": 1.714215874671936,
+      "learning_rate": 5.810197951430911e-05,
+      "loss": 2.0684,
+      "step": 416
+    },
+    {
+      "epoch": 0.954233409610984,
+      "grad_norm": 1.8952488899230957,
+      "learning_rate": 5.7915170467031635e-05,
+      "loss": 2.2021,
+      "step": 417
+    },
+    {
+      "epoch": 0.9565217391304348,
+      "grad_norm": 2.1692068576812744,
+      "learning_rate": 5.772824802229733e-05,
+      "loss": 2.1982,
+      "step": 418
+    },
+    {
+      "epoch": 0.9588100686498856,
+      "grad_norm": 1.9838628768920898,
+      "learning_rate": 5.7541214858068705e-05,
+      "loss": 1.9893,
+      "step": 419
+    },
+    {
+      "epoch": 0.9610983981693364,
+      "grad_norm": 1.965591311454773,
+      "learning_rate": 5.735407365389453e-05,
+      "loss": 2.2526,
+      "step": 420
+    },
+    {
+      "epoch": 0.9633867276887872,
+      "grad_norm": 1.8901221752166748,
+      "learning_rate": 5.716682709087139e-05,
+      "loss": 2.1193,
+      "step": 421
+    },
+    {
+      "epoch": 0.965675057208238,
+      "grad_norm": 1.8736552000045776,
+      "learning_rate": 5.697947785160532e-05,
+      "loss": 1.9734,
+      "step": 422
+    },
+    {
+      "epoch": 0.9679633867276888,
+      "grad_norm": 1.5453659296035767,
+      "learning_rate": 5.679202862017338e-05,
+      "loss": 2.1037,
+      "step": 423
+    },
+    {
+      "epoch": 0.9702517162471396,
+      "grad_norm": 2.2312018871307373,
+      "learning_rate": 5.660448208208513e-05,
+      "loss": 2.2064,
+      "step": 424
+    },
+    {
+      "epoch": 0.9725400457665904,
+      "grad_norm": 2.2896294593811035,
+      "learning_rate": 5.641684092424421e-05,
+      "loss": 2.1111,
+      "step": 425
+    },
+    {
+      "epoch": 0.9748283752860412,
+      "grad_norm": 1.7300447225570679,
+      "learning_rate": 5.622910783490988e-05,
+      "loss": 2.1577,
+      "step": 426
+    },
+    {
+      "epoch": 0.977116704805492,
+      "grad_norm": 1.988135576248169,
+      "learning_rate": 5.604128550365845e-05,
+      "loss": 2.0322,
+      "step": 427
+    },
+    {
+      "epoch": 0.9794050343249427,
+      "grad_norm": 1.9037761688232422,
+      "learning_rate": 5.585337662134471e-05,
+      "loss": 2.351,
+      "step": 428
+    },
+    {
+      "epoch": 0.9816933638443935,
+      "grad_norm": 1.5671565532684326,
+      "learning_rate": 5.56653838800635e-05,
+      "loss": 1.8901,
+      "step": 429
+    },
+    {
+      "epoch": 0.9839816933638444,
+      "grad_norm": 1.687756896018982,
+      "learning_rate": 5.5477309973111046e-05,
+      "loss": 2.2379,
+      "step": 430
+    },
+    {
+      "epoch": 0.9862700228832952,
+      "grad_norm": 1.8213844299316406,
+      "learning_rate": 5.52891575949464e-05,
+      "loss": 2.1647,
+      "step": 431
+    },
+    {
+      "epoch": 0.988558352402746,
+      "grad_norm": 2.0748496055603027,
+      "learning_rate": 5.510092944115286e-05,
+      "loss": 2.0475,
+      "step": 432
+    },
+    {
+      "epoch": 0.9908466819221968,
+      "grad_norm": 1.5618698596954346,
+      "learning_rate": 5.4912628208399294e-05,
+      "loss": 1.9637,
+      "step": 433
+    },
+    {
+      "epoch": 0.9931350114416476,
+      "grad_norm": 2.596681594848633,
+      "learning_rate": 5.472425659440157e-05,
+      "loss": 2.408,
+      "step": 434
+    },
+    {
+      "epoch": 0.9954233409610984,
+      "grad_norm": 1.877379059791565,
+      "learning_rate": 5.4535817297883876e-05,
+      "loss": 2.009,
+      "step": 435
+    },
+    {
+      "epoch": 0.9977116704805492,
+      "grad_norm": 2.1219913959503174,
+      "learning_rate": 5.4347313018540056e-05,
+      "loss": 2.0394,
+      "step": 436
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.9108868837356567,
+      "learning_rate": 5.415874645699492e-05,
+      "loss": 2.2033,
+      "step": 437
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 874,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 437,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.397198548225229e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-437/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:407aae785290aa0626a98c1bf487720496944cfff2631a5dbe89e4ab47993f7d
+size 5432

checkpoint-874/config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-hne_Deva-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-874/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-874/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abca4ffe0c71a6cba24465f1eafe4a459b33d4e183a6d6b7b280739ae81eb5e9
+size 4950070472

checkpoint-874/model-00002-of-00006.safetensors

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

checkpoint-874/model-00003-of-00006.safetensors

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

checkpoint-874/model-00004-of-00006.safetensors

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

checkpoint-874/model-00005-of-00006.safetensors

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

checkpoint-874/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:847a2fdd02fbc0c023b36c23cb4f1d5eba52470b08e017c482c23e1b5372631c
+size 4614509080

checkpoint-874/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-874/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cacee80fba84856a5452998cd2a56057921ab37afa4b3732b20b672bd6b0e092
+size 13055369358

checkpoint-874/rng_state.pth

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

checkpoint-874/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:921ef81d5cd742cea225b9dd7afbd661ef4841a6f5ca87bde1ad41497b498ff5
+size 1064

checkpoint-874/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:407aae785290aa0626a98c1bf487720496944cfff2631a5dbe89e4ab47993f7d
+size 5432

config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-hne_Deva-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:abca4ffe0c71a6cba24465f1eafe4a459b33d4e183a6d6b7b280739ae81eb5e9
+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 sha256:847a2fdd02fbc0c023b36c23cb4f1d5eba52470b08e017c482c23e1b5372631c
+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 sha256:407aae785290aa0626a98c1bf487720496944cfff2631a5dbe89e4ab47993f7d
+size 5432