Upload folder using huggingface_hub

checkpoint-1257/config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-shn_Mymr-coli-mean/",
+  "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-1257/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-1257/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6e2acabc4196172af01028dec9ae21376feaabf6c5a581edec561f443debdf5
+size 4950070472

checkpoint-1257/model-00002-of-00006.safetensors

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

checkpoint-1257/model-00003-of-00006.safetensors

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

checkpoint-1257/model-00004-of-00006.safetensors

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

checkpoint-1257/model-00005-of-00006.safetensors

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

checkpoint-1257/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:43554e957dcd93c11630fedd0813f9ba86fb9b0ca314c898c845d85b5fca443f
+size 4614509080

checkpoint-1257/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-1257/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b2292952881fab2c2602716f943fc07500ca21368513cc9a6e16e2e457a7155
+size 13055369358

checkpoint-1257/rng_state.pth

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

checkpoint-1257/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bcb9e7faab64cd29d028952960038e2b4190024df9968b31414a2af32e5951e
+size 1064

checkpoint-1257/trainer_state.json

@@ -0,0 +1,2966 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 1257,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002386634844868735,
+      "grad_norm": 79.25410461425781,
+      "learning_rate": 2.3809523809523808e-06,
+      "loss": 9.018,
+      "step": 3
+    },
+    {
+      "epoch": 0.00477326968973747,
+      "grad_norm": 69.27806854248047,
+      "learning_rate": 4.7619047619047615e-06,
+      "loss": 8.1397,
+      "step": 6
+    },
+    {
+      "epoch": 0.007159904534606206,
+      "grad_norm": 44.50214767456055,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 6.5055,
+      "step": 9
+    },
+    {
+      "epoch": 0.00954653937947494,
+      "grad_norm": 22.070268630981445,
+      "learning_rate": 9.523809523809523e-06,
+      "loss": 5.3274,
+      "step": 12
+    },
+    {
+      "epoch": 0.011933174224343675,
+      "grad_norm": 17.811525344848633,
+      "learning_rate": 1.1904761904761905e-05,
+      "loss": 5.0629,
+      "step": 15
+    },
+    {
+      "epoch": 0.014319809069212411,
+      "grad_norm": 13.21521282196045,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 4.8887,
+      "step": 18
+    },
+    {
+      "epoch": 0.016706443914081145,
+      "grad_norm": 85.0751724243164,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 4.6865,
+      "step": 21
+    },
+    {
+      "epoch": 0.01909307875894988,
+      "grad_norm": 11.005032539367676,
+      "learning_rate": 1.9047619047619046e-05,
+      "loss": 4.4609,
+      "step": 24
+    },
+    {
+      "epoch": 0.021479713603818614,
+      "grad_norm": 9.891936302185059,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 4.4208,
+      "step": 27
+    },
+    {
+      "epoch": 0.02386634844868735,
+      "grad_norm": 6.92100715637207,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 4.3251,
+      "step": 30
+    },
+    {
+      "epoch": 0.026252983293556086,
+      "grad_norm": 4.971158504486084,
+      "learning_rate": 2.6190476190476192e-05,
+      "loss": 3.9813,
+      "step": 33
+    },
+    {
+      "epoch": 0.028639618138424822,
+      "grad_norm": 4.140079021453857,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 4.0783,
+      "step": 36
+    },
+    {
+      "epoch": 0.031026252983293555,
+      "grad_norm": 4.761113166809082,
+      "learning_rate": 3.095238095238095e-05,
+      "loss": 4.1418,
+      "step": 39
+    },
+    {
+      "epoch": 0.03341288782816229,
+      "grad_norm": 5.746503829956055,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 3.8944,
+      "step": 42
+    },
+    {
+      "epoch": 0.03579952267303103,
+      "grad_norm": 4.554972171783447,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 3.7807,
+      "step": 45
+    },
+    {
+      "epoch": 0.03818615751789976,
+      "grad_norm": 3.873955726623535,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 3.607,
+      "step": 48
+    },
+    {
+      "epoch": 0.0405727923627685,
+      "grad_norm": 3.376633644104004,
+      "learning_rate": 4.047619047619048e-05,
+      "loss": 3.4685,
+      "step": 51
+    },
+    {
+      "epoch": 0.04295942720763723,
+      "grad_norm": 5.888181686401367,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 3.4832,
+      "step": 54
+    },
+    {
+      "epoch": 0.045346062052505964,
+      "grad_norm": 5.094118595123291,
+      "learning_rate": 4.523809523809524e-05,
+      "loss": 3.3647,
+      "step": 57
+    },
+    {
+      "epoch": 0.0477326968973747,
+      "grad_norm": 4.940067291259766,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 3.2268,
+      "step": 60
+    },
+    {
+      "epoch": 0.050119331742243436,
+      "grad_norm": 3.514155149459839,
+      "learning_rate": 5e-05,
+      "loss": 3.1229,
+      "step": 63
+    },
+    {
+      "epoch": 0.05250596658711217,
+      "grad_norm": 3.922811985015869,
+      "learning_rate": 5.2380952380952384e-05,
+      "loss": 3.0612,
+      "step": 66
+    },
+    {
+      "epoch": 0.05489260143198091,
+      "grad_norm": 3.203274726867676,
+      "learning_rate": 5.4761904761904766e-05,
+      "loss": 2.8537,
+      "step": 69
+    },
+    {
+      "epoch": 0.057279236276849645,
+      "grad_norm": 2.8945529460906982,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 2.9128,
+      "step": 72
+    },
+    {
+      "epoch": 0.059665871121718374,
+      "grad_norm": 3.287409543991089,
+      "learning_rate": 5.9523809523809524e-05,
+      "loss": 2.8958,
+      "step": 75
+    },
+    {
+      "epoch": 0.06205250596658711,
+      "grad_norm": 3.5249319076538086,
+      "learning_rate": 6.19047619047619e-05,
+      "loss": 2.8925,
+      "step": 78
+    },
+    {
+      "epoch": 0.06443914081145585,
+      "grad_norm": 3.4028635025024414,
+      "learning_rate": 6.428571428571429e-05,
+      "loss": 2.8153,
+      "step": 81
+    },
+    {
+      "epoch": 0.06682577565632458,
+      "grad_norm": 3.1509742736816406,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 2.7562,
+      "step": 84
+    },
+    {
+      "epoch": 0.06921241050119331,
+      "grad_norm": 2.745244264602661,
+      "learning_rate": 6.904761904761905e-05,
+      "loss": 2.6552,
+      "step": 87
+    },
+    {
+      "epoch": 0.07159904534606205,
+      "grad_norm": 2.562229633331299,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 2.6686,
+      "step": 90
+    },
+    {
+      "epoch": 0.07398568019093078,
+      "grad_norm": 3.0172386169433594,
+      "learning_rate": 7.380952380952382e-05,
+      "loss": 2.782,
+      "step": 93
+    },
+    {
+      "epoch": 0.07637231503579953,
+      "grad_norm": 3.038167953491211,
+      "learning_rate": 7.619047619047618e-05,
+      "loss": 2.6845,
+      "step": 96
+    },
+    {
+      "epoch": 0.07875894988066826,
+      "grad_norm": 3.3530476093292236,
+      "learning_rate": 7.857142857142858e-05,
+      "loss": 2.5982,
+      "step": 99
+    },
+    {
+      "epoch": 0.081145584725537,
+      "grad_norm": 2.9540939331054688,
+      "learning_rate": 8.095238095238096e-05,
+      "loss": 2.5005,
+      "step": 102
+    },
+    {
+      "epoch": 0.08353221957040573,
+      "grad_norm": 2.9407269954681396,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 2.5421,
+      "step": 105
+    },
+    {
+      "epoch": 0.08591885441527446,
+      "grad_norm": 2.8846545219421387,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 2.5134,
+      "step": 108
+    },
+    {
+      "epoch": 0.0883054892601432,
+      "grad_norm": 3.2429230213165283,
+      "learning_rate": 8.80952380952381e-05,
+      "loss": 2.4542,
+      "step": 111
+    },
+    {
+      "epoch": 0.09069212410501193,
+      "grad_norm": 2.5776901245117188,
+      "learning_rate": 9.047619047619048e-05,
+      "loss": 2.5908,
+      "step": 114
+    },
+    {
+      "epoch": 0.09307875894988067,
+      "grad_norm": 2.4036340713500977,
+      "learning_rate": 9.285714285714286e-05,
+      "loss": 2.4376,
+      "step": 117
+    },
+    {
+      "epoch": 0.0954653937947494,
+      "grad_norm": 3.9573254585266113,
+      "learning_rate": 9.523809523809524e-05,
+      "loss": 2.4042,
+      "step": 120
+    },
+    {
+      "epoch": 0.09785202863961814,
+      "grad_norm": 3.6064600944519043,
+      "learning_rate": 9.761904761904762e-05,
+      "loss": 2.5074,
+      "step": 123
+    },
+    {
+      "epoch": 0.10023866348448687,
+      "grad_norm": 3.3975508213043213,
+      "learning_rate": 0.0001,
+      "loss": 2.2322,
+      "step": 126
+    },
+    {
+      "epoch": 0.1026252983293556,
+      "grad_norm": 2.3529391288757324,
+      "learning_rate": 9.999961058466053e-05,
+      "loss": 2.4629,
+      "step": 129
+    },
+    {
+      "epoch": 0.10501193317422435,
+      "grad_norm": 2.434084892272949,
+      "learning_rate": 9.999844234470782e-05,
+      "loss": 2.3296,
+      "step": 132
+    },
+    {
+      "epoch": 0.10739856801909307,
+      "grad_norm": 2.450005054473877,
+      "learning_rate": 9.999649529833915e-05,
+      "loss": 2.3464,
+      "step": 135
+    },
+    {
+      "epoch": 0.10978520286396182,
+      "grad_norm": 3.4513914585113525,
+      "learning_rate": 9.999376947588288e-05,
+      "loss": 2.4644,
+      "step": 138
+    },
+    {
+      "epoch": 0.11217183770883055,
+      "grad_norm": 3.2405099868774414,
+      "learning_rate": 9.999026491979808e-05,
+      "loss": 2.3977,
+      "step": 141
+    },
+    {
+      "epoch": 0.11455847255369929,
+      "grad_norm": 1.8948777914047241,
+      "learning_rate": 9.99859816846739e-05,
+      "loss": 2.3372,
+      "step": 144
+    },
+    {
+      "epoch": 0.11694510739856802,
+      "grad_norm": 2.3878233432769775,
+      "learning_rate": 9.998091983722863e-05,
+      "loss": 2.3408,
+      "step": 147
+    },
+    {
+      "epoch": 0.11933174224343675,
+      "grad_norm": 2.373782157897949,
+      "learning_rate": 9.99750794563087e-05,
+      "loss": 2.1927,
+      "step": 150
+    },
+    {
+      "epoch": 0.12171837708830549,
+      "grad_norm": 2.473146677017212,
+      "learning_rate": 9.996846063288747e-05,
+      "loss": 2.2674,
+      "step": 153
+    },
+    {
+      "epoch": 0.12410501193317422,
+      "grad_norm": 2.17854905128479,
+      "learning_rate": 9.996106347006379e-05,
+      "loss": 2.4093,
+      "step": 156
+    },
+    {
+      "epoch": 0.12649164677804295,
+      "grad_norm": 2.668506622314453,
+      "learning_rate": 9.99528880830604e-05,
+      "loss": 2.234,
+      "step": 159
+    },
+    {
+      "epoch": 0.1288782816229117,
+      "grad_norm": 2.30295991897583,
+      "learning_rate": 9.994393459922218e-05,
+      "loss": 2.174,
+      "step": 162
+    },
+    {
+      "epoch": 0.13126491646778043,
+      "grad_norm": 2.299704074859619,
+      "learning_rate": 9.993420315801406e-05,
+      "loss": 2.1369,
+      "step": 165
+    },
+    {
+      "epoch": 0.13365155131264916,
+      "grad_norm": 4.072019100189209,
+      "learning_rate": 9.992369391101895e-05,
+      "loss": 2.2051,
+      "step": 168
+    },
+    {
+      "epoch": 0.1360381861575179,
+      "grad_norm": 2.203756809234619,
+      "learning_rate": 9.991240702193532e-05,
+      "loss": 2.3608,
+      "step": 171
+    },
+    {
+      "epoch": 0.13842482100238662,
+      "grad_norm": 2.2594192028045654,
+      "learning_rate": 9.990034266657467e-05,
+      "loss": 2.2503,
+      "step": 174
+    },
+    {
+      "epoch": 0.14081145584725538,
+      "grad_norm": 2.214170217514038,
+      "learning_rate": 9.988750103285883e-05,
+      "loss": 2.1698,
+      "step": 177
+    },
+    {
+      "epoch": 0.1431980906921241,
+      "grad_norm": 2.1277706623077393,
+      "learning_rate": 9.987388232081694e-05,
+      "loss": 2.2199,
+      "step": 180
+    },
+    {
+      "epoch": 0.14558472553699284,
+      "grad_norm": 2.1861696243286133,
+      "learning_rate": 9.985948674258243e-05,
+      "loss": 2.1487,
+      "step": 183
+    },
+    {
+      "epoch": 0.14797136038186157,
+      "grad_norm": 2.062450647354126,
+      "learning_rate": 9.984431452238967e-05,
+      "loss": 2.2716,
+      "step": 186
+    },
+    {
+      "epoch": 0.15035799522673032,
+      "grad_norm": 2.335073471069336,
+      "learning_rate": 9.982836589657043e-05,
+      "loss": 2.1853,
+      "step": 189
+    },
+    {
+      "epoch": 0.15274463007159905,
+      "grad_norm": 1.7519389390945435,
+      "learning_rate": 9.981164111355035e-05,
+      "loss": 2.2452,
+      "step": 192
+    },
+    {
+      "epoch": 0.15513126491646778,
+      "grad_norm": 4.789851665496826,
+      "learning_rate": 9.979414043384485e-05,
+      "loss": 2.224,
+      "step": 195
+    },
+    {
+      "epoch": 0.1575178997613365,
+      "grad_norm": 2.454585552215576,
+      "learning_rate": 9.977586413005531e-05,
+      "loss": 2.2943,
+      "step": 198
+    },
+    {
+      "epoch": 0.15990453460620524,
+      "grad_norm": 2.2278664112091064,
+      "learning_rate": 9.975681248686461e-05,
+      "loss": 2.2947,
+      "step": 201
+    },
+    {
+      "epoch": 0.162291169451074,
+      "grad_norm": 2.118945837020874,
+      "learning_rate": 9.973698580103285e-05,
+      "loss": 2.2039,
+      "step": 204
+    },
+    {
+      "epoch": 0.16467780429594273,
+      "grad_norm": 2.1315999031066895,
+      "learning_rate": 9.971638438139266e-05,
+      "loss": 2.3493,
+      "step": 207
+    },
+    {
+      "epoch": 0.16706443914081145,
+      "grad_norm": 2.902245283126831,
+      "learning_rate": 9.96950085488444e-05,
+      "loss": 2.0854,
+      "step": 210
+    },
+    {
+      "epoch": 0.16945107398568018,
+      "grad_norm": 2.160647392272949,
+      "learning_rate": 9.967285863635112e-05,
+      "loss": 2.3386,
+      "step": 213
+    },
+    {
+      "epoch": 0.1718377088305489,
+      "grad_norm": 3.592740774154663,
+      "learning_rate": 9.964993498893349e-05,
+      "loss": 2.2387,
+      "step": 216
+    },
+    {
+      "epoch": 0.17422434367541767,
+      "grad_norm": 3.9422338008880615,
+      "learning_rate": 9.962623796366429e-05,
+      "loss": 2.2304,
+      "step": 219
+    },
+    {
+      "epoch": 0.1766109785202864,
+      "grad_norm": 3.213841438293457,
+      "learning_rate": 9.960176792966289e-05,
+      "loss": 2.1835,
+      "step": 222
+    },
+    {
+      "epoch": 0.17899761336515513,
+      "grad_norm": 2.4797472953796387,
+      "learning_rate": 9.95765252680896e-05,
+      "loss": 2.2491,
+      "step": 225
+    },
+    {
+      "epoch": 0.18138424821002386,
+      "grad_norm": 2.0377020835876465,
+      "learning_rate": 9.95505103721396e-05,
+      "loss": 2.1921,
+      "step": 228
+    },
+    {
+      "epoch": 0.18377088305489261,
+      "grad_norm": 2.2581214904785156,
+      "learning_rate": 9.952372364703687e-05,
+      "loss": 2.2655,
+      "step": 231
+    },
+    {
+      "epoch": 0.18615751789976134,
+      "grad_norm": 2.89260196685791,
+      "learning_rate": 9.949616551002787e-05,
+      "loss": 2.1804,
+      "step": 234
+    },
+    {
+      "epoch": 0.18854415274463007,
+      "grad_norm": 2.4222378730773926,
+      "learning_rate": 9.946783639037504e-05,
+      "loss": 2.1525,
+      "step": 237
+    },
+    {
+      "epoch": 0.1909307875894988,
+      "grad_norm": 2.1491341590881348,
+      "learning_rate": 9.943873672935014e-05,
+      "loss": 2.1605,
+      "step": 240
+    },
+    {
+      "epoch": 0.19331742243436753,
+      "grad_norm": 1.9782716035842896,
+      "learning_rate": 9.940886698022734e-05,
+      "loss": 2.0402,
+      "step": 243
+    },
+    {
+      "epoch": 0.1957040572792363,
+      "grad_norm": 1.9801812171936035,
+      "learning_rate": 9.93782276082762e-05,
+      "loss": 2.1493,
+      "step": 246
+    },
+    {
+      "epoch": 0.19809069212410502,
+      "grad_norm": 1.7495440244674683,
+      "learning_rate": 9.934681909075434e-05,
+      "loss": 2.1087,
+      "step": 249
+    },
+    {
+      "epoch": 0.20047732696897375,
+      "grad_norm": 1.9998902082443237,
+      "learning_rate": 9.931464191690015e-05,
+      "loss": 2.1841,
+      "step": 252
+    },
+    {
+      "epoch": 0.20286396181384247,
+      "grad_norm": 2.2117197513580322,
+      "learning_rate": 9.928169658792498e-05,
+      "loss": 2.1848,
+      "step": 255
+    },
+    {
+      "epoch": 0.2052505966587112,
+      "grad_norm": 2.210057497024536,
+      "learning_rate": 9.924798361700553e-05,
+      "loss": 2.1324,
+      "step": 258
+    },
+    {
+      "epoch": 0.20763723150357996,
+      "grad_norm": 1.9545291662216187,
+      "learning_rate": 9.92135035292757e-05,
+      "loss": 2.1454,
+      "step": 261
+    },
+    {
+      "epoch": 0.2100238663484487,
+      "grad_norm": 2.3102903366088867,
+      "learning_rate": 9.91782568618185e-05,
+      "loss": 2.183,
+      "step": 264
+    },
+    {
+      "epoch": 0.21241050119331742,
+      "grad_norm": 1.937774419784546,
+      "learning_rate": 9.914224416365764e-05,
+      "loss": 2.2033,
+      "step": 267
+    },
+    {
+      "epoch": 0.21479713603818615,
+      "grad_norm": 1.8689621686935425,
+      "learning_rate": 9.910546599574902e-05,
+      "loss": 2.3252,
+      "step": 270
+    },
+    {
+      "epoch": 0.2171837708830549,
+      "grad_norm": 2.4020369052886963,
+      "learning_rate": 9.906792293097194e-05,
+      "loss": 2.1485,
+      "step": 273
+    },
+    {
+      "epoch": 0.21957040572792363,
+      "grad_norm": 4.170408248901367,
+      "learning_rate": 9.90296155541202e-05,
+      "loss": 2.1413,
+      "step": 276
+    },
+    {
+      "epoch": 0.22195704057279236,
+      "grad_norm": 1.9683239459991455,
+      "learning_rate": 9.899054446189304e-05,
+      "loss": 1.9998,
+      "step": 279
+    },
+    {
+      "epoch": 0.2243436754176611,
+      "grad_norm": 1.9885108470916748,
+      "learning_rate": 9.895071026288574e-05,
+      "loss": 2.0373,
+      "step": 282
+    },
+    {
+      "epoch": 0.22673031026252982,
+      "grad_norm": 2.297595977783203,
+      "learning_rate": 9.891011357758022e-05,
+      "loss": 2.0459,
+      "step": 285
+    },
+    {
+      "epoch": 0.22911694510739858,
+      "grad_norm": 2.0850205421447754,
+      "learning_rate": 9.886875503833536e-05,
+      "loss": 2.2041,
+      "step": 288
+    },
+    {
+      "epoch": 0.2315035799522673,
+      "grad_norm": 2.3734419345855713,
+      "learning_rate": 9.882663528937717e-05,
+      "loss": 2.1135,
+      "step": 291
+    },
+    {
+      "epoch": 0.23389021479713604,
+      "grad_norm": 2.0763649940490723,
+      "learning_rate": 9.87837549867887e-05,
+      "loss": 1.955,
+      "step": 294
+    },
+    {
+      "epoch": 0.23627684964200477,
+      "grad_norm": 2.171724557876587,
+      "learning_rate": 9.87401147984998e-05,
+      "loss": 2.2409,
+      "step": 297
+    },
+    {
+      "epoch": 0.2386634844868735,
+      "grad_norm": 2.6933743953704834,
+      "learning_rate": 9.869571540427689e-05,
+      "loss": 2.1241,
+      "step": 300
+    },
+    {
+      "epoch": 0.24105011933174225,
+      "grad_norm": 1.917708396911621,
+      "learning_rate": 9.865055749571215e-05,
+      "loss": 2.1745,
+      "step": 303
+    },
+    {
+      "epoch": 0.24343675417661098,
+      "grad_norm": 2.3014838695526123,
+      "learning_rate": 9.860464177621284e-05,
+      "loss": 1.9313,
+      "step": 306
+    },
+    {
+      "epoch": 0.2458233890214797,
+      "grad_norm": 2.2263333797454834,
+      "learning_rate": 9.855796896099045e-05,
+      "loss": 2.1881,
+      "step": 309
+    },
+    {
+      "epoch": 0.24821002386634844,
+      "grad_norm": 2.161557197570801,
+      "learning_rate": 9.851053977704931e-05,
+      "loss": 1.9702,
+      "step": 312
+    },
+    {
+      "epoch": 0.25059665871121717,
+      "grad_norm": 2.1806318759918213,
+      "learning_rate": 9.846235496317555e-05,
+      "loss": 1.9893,
+      "step": 315
+    },
+    {
+      "epoch": 0.2529832935560859,
+      "grad_norm": 2.226500988006592,
+      "learning_rate": 9.841341526992536e-05,
+      "loss": 2.0987,
+      "step": 318
+    },
+    {
+      "epoch": 0.2553699284009546,
+      "grad_norm": 1.8919053077697754,
+      "learning_rate": 9.836372145961345e-05,
+      "loss": 2.1568,
+      "step": 321
+    },
+    {
+      "epoch": 0.2577565632458234,
+      "grad_norm": 1.9861871004104614,
+      "learning_rate": 9.83132743063011e-05,
+      "loss": 2.1504,
+      "step": 324
+    },
+    {
+      "epoch": 0.26014319809069214,
+      "grad_norm": 2.438837766647339,
+      "learning_rate": 9.826207459578411e-05,
+      "loss": 1.9779,
+      "step": 327
+    },
+    {
+      "epoch": 0.26252983293556087,
+      "grad_norm": 2.6724672317504883,
+      "learning_rate": 9.821012312558058e-05,
+      "loss": 1.9427,
+      "step": 330
+    },
+    {
+      "epoch": 0.2649164677804296,
+      "grad_norm": 1.813621163368225,
+      "learning_rate": 9.815742070491852e-05,
+      "loss": 2.1153,
+      "step": 333
+    },
+    {
+      "epoch": 0.26730310262529833,
+      "grad_norm": 2.0258703231811523,
+      "learning_rate": 9.810396815472314e-05,
+      "loss": 2.02,
+      "step": 336
+    },
+    {
+      "epoch": 0.26968973747016706,
+      "grad_norm": 1.9743555784225464,
+      "learning_rate": 9.804976630760419e-05,
+      "loss": 2.0681,
+      "step": 339
+    },
+    {
+      "epoch": 0.2720763723150358,
+      "grad_norm": 1.8202836513519287,
+      "learning_rate": 9.799481600784286e-05,
+      "loss": 1.9733,
+      "step": 342
+    },
+    {
+      "epoch": 0.2744630071599045,
+      "grad_norm": 2.296698808670044,
+      "learning_rate": 9.793911811137875e-05,
+      "loss": 2.0407,
+      "step": 345
+    },
+    {
+      "epoch": 0.27684964200477324,
+      "grad_norm": 2.002511501312256,
+      "learning_rate": 9.788267348579648e-05,
+      "loss": 2.0666,
+      "step": 348
+    },
+    {
+      "epoch": 0.27923627684964203,
+      "grad_norm": 2.08190655708313,
+      "learning_rate": 9.782548301031217e-05,
+      "loss": 2.0803,
+      "step": 351
+    },
+    {
+      "epoch": 0.28162291169451076,
+      "grad_norm": 2.150238513946533,
+      "learning_rate": 9.776754757575975e-05,
+      "loss": 2.0132,
+      "step": 354
+    },
+    {
+      "epoch": 0.2840095465393795,
+      "grad_norm": 2.2527501583099365,
+      "learning_rate": 9.770886808457709e-05,
+      "loss": 2.0905,
+      "step": 357
+    },
+    {
+      "epoch": 0.2863961813842482,
+      "grad_norm": 2.284032106399536,
+      "learning_rate": 9.764944545079196e-05,
+      "loss": 2.1845,
+      "step": 360
+    },
+    {
+      "epoch": 0.28878281622911695,
+      "grad_norm": 1.89547598361969,
+      "learning_rate": 9.758928060000778e-05,
+      "loss": 2.1135,
+      "step": 363
+    },
+    {
+      "epoch": 0.2911694510739857,
+      "grad_norm": 2.5930140018463135,
+      "learning_rate": 9.752837446938915e-05,
+      "loss": 1.9517,
+      "step": 366
+    },
+    {
+      "epoch": 0.2935560859188544,
+      "grad_norm": 1.86557137966156,
+      "learning_rate": 9.746672800764735e-05,
+      "loss": 1.9579,
+      "step": 369
+    },
+    {
+      "epoch": 0.29594272076372313,
+      "grad_norm": 1.8821232318878174,
+      "learning_rate": 9.740434217502547e-05,
+      "loss": 1.9665,
+      "step": 372
+    },
+    {
+      "epoch": 0.29832935560859186,
+      "grad_norm": 2.0216808319091797,
+      "learning_rate": 9.734121794328357e-05,
+      "loss": 2.0612,
+      "step": 375
+    },
+    {
+      "epoch": 0.30071599045346065,
+      "grad_norm": 1.8245267868041992,
+      "learning_rate": 9.727735629568336e-05,
+      "loss": 2.0857,
+      "step": 378
+    },
+    {
+      "epoch": 0.3031026252983294,
+      "grad_norm": 1.9001407623291016,
+      "learning_rate": 9.721275822697306e-05,
+      "loss": 1.9954,
+      "step": 381
+    },
+    {
+      "epoch": 0.3054892601431981,
+      "grad_norm": 1.7628698348999023,
+      "learning_rate": 9.714742474337186e-05,
+      "loss": 2.1095,
+      "step": 384
+    },
+    {
+      "epoch": 0.30787589498806683,
+      "grad_norm": 1.687436819076538,
+      "learning_rate": 9.708135686255416e-05,
+      "loss": 2.1919,
+      "step": 387
+    },
+    {
+      "epoch": 0.31026252983293556,
+      "grad_norm": 2.219071388244629,
+      "learning_rate": 9.701455561363379e-05,
+      "loss": 1.9392,
+      "step": 390
+    },
+    {
+      "epoch": 0.3126491646778043,
+      "grad_norm": 1.9545831680297852,
+      "learning_rate": 9.6947022037148e-05,
+      "loss": 1.9879,
+      "step": 393
+    },
+    {
+      "epoch": 0.315035799522673,
+      "grad_norm": 2.1863789558410645,
+      "learning_rate": 9.687875718504126e-05,
+      "loss": 1.9631,
+      "step": 396
+    },
+    {
+      "epoch": 0.31742243436754175,
+      "grad_norm": 3.1137101650238037,
+      "learning_rate": 9.680976212064874e-05,
+      "loss": 2.0387,
+      "step": 399
+    },
+    {
+      "epoch": 0.3198090692124105,
+      "grad_norm": 1.9021557569503784,
+      "learning_rate": 9.674003791867991e-05,
+      "loss": 2.0447,
+      "step": 402
+    },
+    {
+      "epoch": 0.3221957040572792,
+      "grad_norm": 1.8783704042434692,
+      "learning_rate": 9.666958566520174e-05,
+      "loss": 2.0777,
+      "step": 405
+    },
+    {
+      "epoch": 0.324582338902148,
+      "grad_norm": 1.910521388053894,
+      "learning_rate": 9.659840645762175e-05,
+      "loss": 2.1084,
+      "step": 408
+    },
+    {
+      "epoch": 0.3269689737470167,
+      "grad_norm": 1.9456645250320435,
+      "learning_rate": 9.652650140467093e-05,
+      "loss": 2.0317,
+      "step": 411
+    },
+    {
+      "epoch": 0.32935560859188545,
+      "grad_norm": 1.8211055994033813,
+      "learning_rate": 9.645387162638652e-05,
+      "loss": 2.0386,
+      "step": 414
+    },
+    {
+      "epoch": 0.3317422434367542,
+      "grad_norm": 2.032345771789551,
+      "learning_rate": 9.638051825409453e-05,
+      "loss": 2.2154,
+      "step": 417
+    },
+    {
+      "epoch": 0.3341288782816229,
+      "grad_norm": 1.869388461112976,
+      "learning_rate": 9.630644243039207e-05,
+      "loss": 1.9595,
+      "step": 420
+    },
+    {
+      "epoch": 0.33651551312649164,
+      "grad_norm": 1.9663021564483643,
+      "learning_rate": 9.623164530912963e-05,
+      "loss": 2.1678,
+      "step": 423
+    },
+    {
+      "epoch": 0.33890214797136037,
+      "grad_norm": 1.684556484222412,
+      "learning_rate": 9.615612805539305e-05,
+      "loss": 1.9458,
+      "step": 426
+    },
+    {
+      "epoch": 0.3412887828162291,
+      "grad_norm": 2.050321102142334,
+      "learning_rate": 9.607989184548543e-05,
+      "loss": 2.0412,
+      "step": 429
+    },
+    {
+      "epoch": 0.3436754176610978,
+      "grad_norm": 1.753670334815979,
+      "learning_rate": 9.600293786690872e-05,
+      "loss": 2.037,
+      "step": 432
+    },
+    {
+      "epoch": 0.3460620525059666,
+      "grad_norm": 1.8368828296661377,
+      "learning_rate": 9.592526731834537e-05,
+      "loss": 2.1845,
+      "step": 435
+    },
+    {
+      "epoch": 0.34844868735083534,
+      "grad_norm": 1.8000643253326416,
+      "learning_rate": 9.584688140963944e-05,
+      "loss": 1.8592,
+      "step": 438
+    },
+    {
+      "epoch": 0.35083532219570407,
+      "grad_norm": 1.7274373769760132,
+      "learning_rate": 9.576778136177798e-05,
+      "loss": 2.0366,
+      "step": 441
+    },
+    {
+      "epoch": 0.3532219570405728,
+      "grad_norm": 1.8421188592910767,
+      "learning_rate": 9.568796840687184e-05,
+      "loss": 1.9281,
+      "step": 444
+    },
+    {
+      "epoch": 0.3556085918854415,
+      "grad_norm": 1.8605895042419434,
+      "learning_rate": 9.560744378813659e-05,
+      "loss": 2.0214,
+      "step": 447
+    },
+    {
+      "epoch": 0.35799522673031026,
+      "grad_norm": 2.4788737297058105,
+      "learning_rate": 9.552620875987311e-05,
+      "loss": 1.9457,
+      "step": 450
+    },
+    {
+      "epoch": 0.360381861575179,
+      "grad_norm": 3.01055645942688,
+      "learning_rate": 9.544426458744804e-05,
+      "loss": 1.879,
+      "step": 453
+    },
+    {
+      "epoch": 0.3627684964200477,
+      "grad_norm": 2.6994807720184326,
+      "learning_rate": 9.536161254727408e-05,
+      "loss": 1.9495,
+      "step": 456
+    },
+    {
+      "epoch": 0.36515513126491644,
+      "grad_norm": 1.736722469329834,
+      "learning_rate": 9.527825392679012e-05,
+      "loss": 1.942,
+      "step": 459
+    },
+    {
+      "epoch": 0.36754176610978523,
+      "grad_norm": 2.2707650661468506,
+      "learning_rate": 9.51941900244412e-05,
+      "loss": 1.9525,
+      "step": 462
+    },
+    {
+      "epoch": 0.36992840095465396,
+      "grad_norm": 1.7479629516601562,
+      "learning_rate": 9.51094221496582e-05,
+      "loss": 1.9385,
+      "step": 465
+    },
+    {
+      "epoch": 0.3723150357995227,
+      "grad_norm": 2.1110448837280273,
+      "learning_rate": 9.502395162283759e-05,
+      "loss": 1.8335,
+      "step": 468
+    },
+    {
+      "epoch": 0.3747016706443914,
+      "grad_norm": 1.8206532001495361,
+      "learning_rate": 9.493777977532072e-05,
+      "loss": 1.9642,
+      "step": 471
+    },
+    {
+      "epoch": 0.37708830548926014,
+      "grad_norm": 2.622044801712036,
+      "learning_rate": 9.485090794937319e-05,
+      "loss": 1.9383,
+      "step": 474
+    },
+    {
+      "epoch": 0.3794749403341289,
+      "grad_norm": 2.016352653503418,
+      "learning_rate": 9.476333749816382e-05,
+      "loss": 1.8639,
+      "step": 477
+    },
+    {
+      "epoch": 0.3818615751789976,
+      "grad_norm": 2.027357816696167,
+      "learning_rate": 9.467506978574371e-05,
+      "loss": 1.862,
+      "step": 480
+    },
+    {
+      "epoch": 0.38424821002386633,
+      "grad_norm": 1.91681969165802,
+      "learning_rate": 9.45861061870249e-05,
+      "loss": 2.0105,
+      "step": 483
+    },
+    {
+      "epoch": 0.38663484486873506,
+      "grad_norm": 1.8423808813095093,
+      "learning_rate": 9.449644808775902e-05,
+      "loss": 1.9303,
+      "step": 486
+    },
+    {
+      "epoch": 0.38902147971360385,
+      "grad_norm": 1.5539026260375977,
+      "learning_rate": 9.44060968845156e-05,
+      "loss": 1.9071,
+      "step": 489
+    },
+    {
+      "epoch": 0.3914081145584726,
+      "grad_norm": 1.8300362825393677,
+      "learning_rate": 9.431505398466045e-05,
+      "loss": 2.0194,
+      "step": 492
+    },
+    {
+      "epoch": 0.3937947494033413,
+      "grad_norm": 1.6660507917404175,
+      "learning_rate": 9.42233208063336e-05,
+      "loss": 1.8226,
+      "step": 495
+    },
+    {
+      "epoch": 0.39618138424821003,
+      "grad_norm": 1.9865005016326904,
+      "learning_rate": 9.413089877842736e-05,
+      "loss": 2.0689,
+      "step": 498
+    },
+    {
+      "epoch": 0.39856801909307876,
+      "grad_norm": 1.9386086463928223,
+      "learning_rate": 9.403778934056391e-05,
+      "loss": 2.0289,
+      "step": 501
+    },
+    {
+      "epoch": 0.4009546539379475,
+      "grad_norm": 1.9398423433303833,
+      "learning_rate": 9.394399394307303e-05,
+      "loss": 2.2213,
+      "step": 504
+    },
+    {
+      "epoch": 0.4033412887828162,
+      "grad_norm": 1.864970326423645,
+      "learning_rate": 9.384951404696933e-05,
+      "loss": 1.8574,
+      "step": 507
+    },
+    {
+      "epoch": 0.40572792362768495,
+      "grad_norm": 1.9465175867080688,
+      "learning_rate": 9.375435112392969e-05,
+      "loss": 2.0628,
+      "step": 510
+    },
+    {
+      "epoch": 0.4081145584725537,
+      "grad_norm": 1.808294415473938,
+      "learning_rate": 9.365850665627016e-05,
+      "loss": 1.9223,
+      "step": 513
+    },
+    {
+      "epoch": 0.4105011933174224,
+      "grad_norm": 2.4403979778289795,
+      "learning_rate": 9.356198213692297e-05,
+      "loss": 1.8865,
+      "step": 516
+    },
+    {
+      "epoch": 0.4128878281622912,
+      "grad_norm": 1.8101935386657715,
+      "learning_rate": 9.346477906941331e-05,
+      "loss": 1.8335,
+      "step": 519
+    },
+    {
+      "epoch": 0.4152744630071599,
+      "grad_norm": 1.718172311782837,
+      "learning_rate": 9.336689896783573e-05,
+      "loss": 1.8691,
+      "step": 522
+    },
+    {
+      "epoch": 0.41766109785202865,
+      "grad_norm": 1.8754642009735107,
+      "learning_rate": 9.32683433568308e-05,
+      "loss": 1.9212,
+      "step": 525
+    },
+    {
+      "epoch": 0.4200477326968974,
+      "grad_norm": 2.471646785736084,
+      "learning_rate": 9.316911377156117e-05,
+      "loss": 2.0121,
+      "step": 528
+    },
+    {
+      "epoch": 0.4224343675417661,
+      "grad_norm": 2.1250839233398438,
+      "learning_rate": 9.306921175768775e-05,
+      "loss": 1.9309,
+      "step": 531
+    },
+    {
+      "epoch": 0.42482100238663484,
+      "grad_norm": 2.0609402656555176,
+      "learning_rate": 9.29686388713456e-05,
+      "loss": 2.1438,
+      "step": 534
+    },
+    {
+      "epoch": 0.42720763723150357,
+      "grad_norm": 1.7126508951187134,
+      "learning_rate": 9.286739667911972e-05,
+      "loss": 1.9621,
+      "step": 537
+    },
+    {
+      "epoch": 0.4295942720763723,
+      "grad_norm": 1.7493348121643066,
+      "learning_rate": 9.276548675802059e-05,
+      "loss": 2.0063,
+      "step": 540
+    },
+    {
+      "epoch": 0.431980906921241,
+      "grad_norm": 1.8076331615447998,
+      "learning_rate": 9.266291069545972e-05,
+      "loss": 1.9256,
+      "step": 543
+    },
+    {
+      "epoch": 0.4343675417661098,
+      "grad_norm": 1.6762983798980713,
+      "learning_rate": 9.255967008922474e-05,
+      "loss": 2.0052,
+      "step": 546
+    },
+    {
+      "epoch": 0.43675417661097854,
+      "grad_norm": 1.4335881471633911,
+      "learning_rate": 9.245576654745471e-05,
+      "loss": 2.0899,
+      "step": 549
+    },
+    {
+      "epoch": 0.43914081145584727,
+      "grad_norm": 1.6479798555374146,
+      "learning_rate": 9.235120168861496e-05,
+      "loss": 1.7962,
+      "step": 552
+    },
+    {
+      "epoch": 0.441527446300716,
+      "grad_norm": 2.1050121784210205,
+      "learning_rate": 9.224597714147186e-05,
+      "loss": 2.0109,
+      "step": 555
+    },
+    {
+      "epoch": 0.4439140811455847,
+      "grad_norm": 1.6112617254257202,
+      "learning_rate": 9.214009454506753e-05,
+      "loss": 1.8432,
+      "step": 558
+    },
+    {
+      "epoch": 0.44630071599045346,
+      "grad_norm": 1.8361741304397583,
+      "learning_rate": 9.203355554869428e-05,
+      "loss": 1.9433,
+      "step": 561
+    },
+    {
+      "epoch": 0.4486873508353222,
+      "grad_norm": 3.137519121170044,
+      "learning_rate": 9.192636181186888e-05,
+      "loss": 1.7776,
+      "step": 564
+    },
+    {
+      "epoch": 0.4510739856801909,
+      "grad_norm": 1.79214346408844,
+      "learning_rate": 9.181851500430673e-05,
+      "loss": 1.8203,
+      "step": 567
+    },
+    {
+      "epoch": 0.45346062052505964,
+      "grad_norm": 2.010784149169922,
+      "learning_rate": 9.171001680589588e-05,
+      "loss": 1.8505,
+      "step": 570
+    },
+    {
+      "epoch": 0.45584725536992843,
+      "grad_norm": 2.2128775119781494,
+      "learning_rate": 9.160086890667086e-05,
+      "loss": 1.9007,
+      "step": 573
+    },
+    {
+      "epoch": 0.45823389021479716,
+      "grad_norm": 1.6658575534820557,
+      "learning_rate": 9.14910730067863e-05,
+      "loss": 1.9516,
+      "step": 576
+    },
+    {
+      "epoch": 0.4606205250596659,
+      "grad_norm": 1.6554961204528809,
+      "learning_rate": 9.138063081649051e-05,
+      "loss": 2.0685,
+      "step": 579
+    },
+    {
+      "epoch": 0.4630071599045346,
+      "grad_norm": 1.6959861516952515,
+      "learning_rate": 9.126954405609882e-05,
+      "loss": 1.9156,
+      "step": 582
+    },
+    {
+      "epoch": 0.46539379474940334,
+      "grad_norm": 1.795530915260315,
+      "learning_rate": 9.115781445596676e-05,
+      "loss": 1.7886,
+      "step": 585
+    },
+    {
+      "epoch": 0.4677804295942721,
+      "grad_norm": 1.9987218379974365,
+      "learning_rate": 9.104544375646313e-05,
+      "loss": 2.1443,
+      "step": 588
+    },
+    {
+      "epoch": 0.4701670644391408,
+      "grad_norm": 1.8751397132873535,
+      "learning_rate": 9.093243370794291e-05,
+      "loss": 1.8975,
+      "step": 591
+    },
+    {
+      "epoch": 0.47255369928400953,
+      "grad_norm": 1.921032428741455,
+      "learning_rate": 9.081878607071996e-05,
+      "loss": 1.9969,
+      "step": 594
+    },
+    {
+      "epoch": 0.47494033412887826,
+      "grad_norm": 4.745877742767334,
+      "learning_rate": 9.07045026150396e-05,
+      "loss": 1.9351,
+      "step": 597
+    },
+    {
+      "epoch": 0.477326968973747,
+      "grad_norm": 1.8512818813323975,
+      "learning_rate": 9.058958512105104e-05,
+      "loss": 1.8773,
+      "step": 600
+    },
+    {
+      "epoch": 0.4797136038186158,
+      "grad_norm": 2.7076666355133057,
+      "learning_rate": 9.047403537877971e-05,
+      "loss": 1.8907,
+      "step": 603
+    },
+    {
+      "epoch": 0.4821002386634845,
+      "grad_norm": 1.9602298736572266,
+      "learning_rate": 9.035785518809927e-05,
+      "loss": 1.955,
+      "step": 606
+    },
+    {
+      "epoch": 0.48448687350835323,
+      "grad_norm": 1.8718470335006714,
+      "learning_rate": 9.024104635870368e-05,
+      "loss": 1.9148,
+      "step": 609
+    },
+    {
+      "epoch": 0.48687350835322196,
+      "grad_norm": 1.708706259727478,
+      "learning_rate": 9.012361071007891e-05,
+      "loss": 1.9202,
+      "step": 612
+    },
+    {
+      "epoch": 0.4892601431980907,
+      "grad_norm": 2.0447752475738525,
+      "learning_rate": 9.000555007147469e-05,
+      "loss": 1.9347,
+      "step": 615
+    },
+    {
+      "epoch": 0.4916467780429594,
+      "grad_norm": 2.0071353912353516,
+      "learning_rate": 8.988686628187597e-05,
+      "loss": 1.9392,
+      "step": 618
+    },
+    {
+      "epoch": 0.49403341288782815,
+      "grad_norm": 1.9587384462356567,
+      "learning_rate": 8.976756118997427e-05,
+      "loss": 1.9572,
+      "step": 621
+    },
+    {
+      "epoch": 0.4964200477326969,
+      "grad_norm": 2.829688549041748,
+      "learning_rate": 8.964763665413893e-05,
+      "loss": 1.7861,
+      "step": 624
+    },
+    {
+      "epoch": 0.4988066825775656,
+      "grad_norm": 1.878251075744629,
+      "learning_rate": 8.952709454238808e-05,
+      "loss": 2.0488,
+      "step": 627
+    },
+    {
+      "epoch": 0.5011933174224343,
+      "grad_norm": 1.6959253549575806,
+      "learning_rate": 8.940593673235962e-05,
+      "loss": 2.0164,
+      "step": 630
+    },
+    {
+      "epoch": 0.5035799522673031,
+      "grad_norm": 1.8719433546066284,
+      "learning_rate": 8.928416511128195e-05,
+      "loss": 1.8288,
+      "step": 633
+    },
+    {
+      "epoch": 0.5059665871121718,
+      "grad_norm": 1.9740793704986572,
+      "learning_rate": 8.916178157594453e-05,
+      "loss": 1.8924,
+      "step": 636
+    },
+    {
+      "epoch": 0.5083532219570406,
+      "grad_norm": 2.177084445953369,
+      "learning_rate": 8.903878803266841e-05,
+      "loss": 2.0888,
+      "step": 639
+    },
+    {
+      "epoch": 0.5107398568019093,
+      "grad_norm": 1.6153382062911987,
+      "learning_rate": 8.891518639727649e-05,
+      "loss": 1.7814,
+      "step": 642
+    },
+    {
+      "epoch": 0.513126491646778,
+      "grad_norm": 2.0001626014709473,
+      "learning_rate": 8.879097859506372e-05,
+      "loss": 1.9432,
+      "step": 645
+    },
+    {
+      "epoch": 0.5155131264916468,
+      "grad_norm": 2.0495731830596924,
+      "learning_rate": 8.866616656076696e-05,
+      "loss": 1.8339,
+      "step": 648
+    },
+    {
+      "epoch": 0.5178997613365155,
+      "grad_norm": 1.7960389852523804,
+      "learning_rate": 8.854075223853508e-05,
+      "loss": 1.9401,
+      "step": 651
+    },
+    {
+      "epoch": 0.5202863961813843,
+      "grad_norm": 1.9565211534500122,
+      "learning_rate": 8.841473758189854e-05,
+      "loss": 1.8112,
+      "step": 654
+    },
+    {
+      "epoch": 0.522673031026253,
+      "grad_norm": 2.022819995880127,
+      "learning_rate": 8.828812455373891e-05,
+      "loss": 2.0094,
+      "step": 657
+    },
+    {
+      "epoch": 0.5250596658711217,
+      "grad_norm": 1.8336989879608154,
+      "learning_rate": 8.816091512625843e-05,
+      "loss": 1.9463,
+      "step": 660
+    },
+    {
+      "epoch": 0.5274463007159904,
+      "grad_norm": 2.031402111053467,
+      "learning_rate": 8.803311128094918e-05,
+      "loss": 1.9657,
+      "step": 663
+    },
+    {
+      "epoch": 0.5298329355608592,
+      "grad_norm": 2.0874717235565186,
+      "learning_rate": 8.790471500856228e-05,
+      "loss": 2.0375,
+      "step": 666
+    },
+    {
+      "epoch": 0.5322195704057279,
+      "grad_norm": 1.7618159055709839,
+      "learning_rate": 8.777572830907684e-05,
+      "loss": 2.1104,
+      "step": 669
+    },
+    {
+      "epoch": 0.5346062052505967,
+      "grad_norm": 1.6236835718154907,
+      "learning_rate": 8.764615319166886e-05,
+      "loss": 2.0333,
+      "step": 672
+    },
+    {
+      "epoch": 0.5369928400954654,
+      "grad_norm": 1.613146424293518,
+      "learning_rate": 8.751599167467985e-05,
+      "loss": 1.8055,
+      "step": 675
+    },
+    {
+      "epoch": 0.5393794749403341,
+      "grad_norm": 1.5570297241210938,
+      "learning_rate": 8.738524578558547e-05,
+      "loss": 1.8801,
+      "step": 678
+    },
+    {
+      "epoch": 0.5417661097852029,
+      "grad_norm": 1.7564152479171753,
+      "learning_rate": 8.72539175609639e-05,
+      "loss": 1.8072,
+      "step": 681
+    },
+    {
+      "epoch": 0.5441527446300716,
+      "grad_norm": 1.7274627685546875,
+      "learning_rate": 8.712200904646416e-05,
+      "loss": 1.787,
+      "step": 684
+    },
+    {
+      "epoch": 0.5465393794749404,
+      "grad_norm": 2.1072866916656494,
+      "learning_rate": 8.698952229677422e-05,
+      "loss": 1.8179,
+      "step": 687
+    },
+    {
+      "epoch": 0.548926014319809,
+      "grad_norm": 2.1879334449768066,
+      "learning_rate": 8.685645937558896e-05,
+      "loss": 2.0755,
+      "step": 690
+    },
+    {
+      "epoch": 0.5513126491646778,
+      "grad_norm": 1.7754478454589844,
+      "learning_rate": 8.67228223555781e-05,
+      "loss": 1.8438,
+      "step": 693
+    },
+    {
+      "epoch": 0.5536992840095465,
+      "grad_norm": 2.0312633514404297,
+      "learning_rate": 8.658861331835385e-05,
+      "loss": 1.9058,
+      "step": 696
+    },
+    {
+      "epoch": 0.5560859188544153,
+      "grad_norm": 1.9634501934051514,
+      "learning_rate": 8.645383435443852e-05,
+      "loss": 1.8278,
+      "step": 699
+    },
+    {
+      "epoch": 0.5584725536992841,
+      "grad_norm": 1.915229082107544,
+      "learning_rate": 8.631848756323197e-05,
+      "loss": 1.9127,
+      "step": 702
+    },
+    {
+      "epoch": 0.5608591885441527,
+      "grad_norm": 1.7499381303787231,
+      "learning_rate": 8.618257505297886e-05,
+      "loss": 1.9196,
+      "step": 705
+    },
+    {
+      "epoch": 0.5632458233890215,
+      "grad_norm": 2.883967161178589,
+      "learning_rate": 8.604609894073584e-05,
+      "loss": 1.7157,
+      "step": 708
+    },
+    {
+      "epoch": 0.5656324582338902,
+      "grad_norm": 1.8204519748687744,
+      "learning_rate": 8.590906135233854e-05,
+      "loss": 1.845,
+      "step": 711
+    },
+    {
+      "epoch": 0.568019093078759,
+      "grad_norm": 3.1434381008148193,
+      "learning_rate": 8.577146442236857e-05,
+      "loss": 1.9142,
+      "step": 714
+    },
+    {
+      "epoch": 0.5704057279236276,
+      "grad_norm": 1.7107082605361938,
+      "learning_rate": 8.563331029412012e-05,
+      "loss": 1.9358,
+      "step": 717
+    },
+    {
+      "epoch": 0.5727923627684964,
+      "grad_norm": 2.010882616043091,
+      "learning_rate": 8.549460111956664e-05,
+      "loss": 1.8745,
+      "step": 720
+    },
+    {
+      "epoch": 0.5751789976133651,
+      "grad_norm": 2.873108148574829,
+      "learning_rate": 8.535533905932738e-05,
+      "loss": 1.8725,
+      "step": 723
+    },
+    {
+      "epoch": 0.5775656324582339,
+      "grad_norm": 2.0034258365631104,
+      "learning_rate": 8.521552628263362e-05,
+      "loss": 1.8666,
+      "step": 726
+    },
+    {
+      "epoch": 0.5799522673031027,
+      "grad_norm": 1.7691940069198608,
+      "learning_rate": 8.507516496729495e-05,
+      "loss": 1.8096,
+      "step": 729
+    },
+    {
+      "epoch": 0.5823389021479713,
+      "grad_norm": 1.77200186252594,
+      "learning_rate": 8.493425729966534e-05,
+      "loss": 1.9294,
+      "step": 732
+    },
+    {
+      "epoch": 0.5847255369928401,
+      "grad_norm": 1.6291214227676392,
+      "learning_rate": 8.479280547460907e-05,
+      "loss": 1.8242,
+      "step": 735
+    },
+    {
+      "epoch": 0.5871121718377088,
+      "grad_norm": 1.5914812088012695,
+      "learning_rate": 8.465081169546659e-05,
+      "loss": 1.9836,
+      "step": 738
+    },
+    {
+      "epoch": 0.5894988066825776,
+      "grad_norm": 1.6103566884994507,
+      "learning_rate": 8.450827817402011e-05,
+      "loss": 1.9699,
+      "step": 741
+    },
+    {
+      "epoch": 0.5918854415274463,
+      "grad_norm": 2.9773311614990234,
+      "learning_rate": 8.436520713045922e-05,
+      "loss": 1.7708,
+      "step": 744
+    },
+    {
+      "epoch": 0.594272076372315,
+      "grad_norm": 1.5203982591629028,
+      "learning_rate": 8.422160079334628e-05,
+      "loss": 1.8533,
+      "step": 747
+    },
+    {
+      "epoch": 0.5966587112171837,
+      "grad_norm": 3.8291232585906982,
+      "learning_rate": 8.40774613995817e-05,
+      "loss": 1.8964,
+      "step": 750
+    },
+    {
+      "epoch": 0.5990453460620525,
+      "grad_norm": 1.650585412979126,
+      "learning_rate": 8.393279119436912e-05,
+      "loss": 1.9163,
+      "step": 753
+    },
+    {
+      "epoch": 0.6014319809069213,
+      "grad_norm": 1.588335633277893,
+      "learning_rate": 8.378759243118044e-05,
+      "loss": 2.0618,
+      "step": 756
+    },
+    {
+      "epoch": 0.60381861575179,
+      "grad_norm": 1.9786241054534912,
+      "learning_rate": 8.364186737172068e-05,
+      "loss": 1.8235,
+      "step": 759
+    },
+    {
+      "epoch": 0.6062052505966588,
+      "grad_norm": 1.7208276987075806,
+      "learning_rate": 8.349561828589277e-05,
+      "loss": 2.0045,
+      "step": 762
+    },
+    {
+      "epoch": 0.6085918854415274,
+      "grad_norm": 1.713976502418518,
+      "learning_rate": 8.33488474517622e-05,
+      "loss": 1.8602,
+      "step": 765
+    },
+    {
+      "epoch": 0.6109785202863962,
+      "grad_norm": 1.655760407447815,
+      "learning_rate": 8.320155715552155e-05,
+      "loss": 1.8096,
+      "step": 768
+    },
+    {
+      "epoch": 0.6133651551312649,
+      "grad_norm": 1.82340669631958,
+      "learning_rate": 8.305374969145488e-05,
+      "loss": 1.9755,
+      "step": 771
+    },
+    {
+      "epoch": 0.6157517899761337,
+      "grad_norm": 1.6505107879638672,
+      "learning_rate": 8.290542736190188e-05,
+      "loss": 1.7543,
+      "step": 774
+    },
+    {
+      "epoch": 0.6181384248210023,
+      "grad_norm": 1.6107587814331055,
+      "learning_rate": 8.275659247722222e-05,
+      "loss": 1.7788,
+      "step": 777
+    },
+    {
+      "epoch": 0.6205250596658711,
+      "grad_norm": 1.7392557859420776,
+      "learning_rate": 8.260724735575933e-05,
+      "loss": 1.8713,
+      "step": 780
+    },
+    {
+      "epoch": 0.6229116945107399,
+      "grad_norm": 1.8423359394073486,
+      "learning_rate": 8.24573943238045e-05,
+      "loss": 1.9501,
+      "step": 783
+    },
+    {
+      "epoch": 0.6252983293556086,
+      "grad_norm": 1.596801996231079,
+      "learning_rate": 8.230703571556048e-05,
+      "loss": 1.7561,
+      "step": 786
+    },
+    {
+      "epoch": 0.6276849642004774,
+      "grad_norm": 1.6264513731002808,
+      "learning_rate": 8.215617387310524e-05,
+      "loss": 1.812,
+      "step": 789
+    },
+    {
+      "epoch": 0.630071599045346,
+      "grad_norm": 1.8065801858901978,
+      "learning_rate": 8.200481114635536e-05,
+      "loss": 1.8587,
+      "step": 792
+    },
+    {
+      "epoch": 0.6324582338902148,
+      "grad_norm": 1.6027936935424805,
+      "learning_rate": 8.185294989302958e-05,
+      "loss": 1.7951,
+      "step": 795
+    },
+    {
+      "epoch": 0.6348448687350835,
+      "grad_norm": 1.863053560256958,
+      "learning_rate": 8.170059247861194e-05,
+      "loss": 1.791,
+      "step": 798
+    },
+    {
+      "epoch": 0.6372315035799523,
+      "grad_norm": 1.7930762767791748,
+      "learning_rate": 8.154774127631501e-05,
+      "loss": 1.7575,
+      "step": 801
+    },
+    {
+      "epoch": 0.639618138424821,
+      "grad_norm": 2.0538759231567383,
+      "learning_rate": 8.139439866704293e-05,
+      "loss": 1.8417,
+      "step": 804
+    },
+    {
+      "epoch": 0.6420047732696897,
+      "grad_norm": 2.5710806846618652,
+      "learning_rate": 8.124056703935423e-05,
+      "loss": 1.8187,
+      "step": 807
+    },
+    {
+      "epoch": 0.6443914081145584,
+      "grad_norm": 1.6980230808258057,
+      "learning_rate": 8.108624878942477e-05,
+      "loss": 1.8364,
+      "step": 810
+    },
+    {
+      "epoch": 0.6467780429594272,
+      "grad_norm": 1.7313123941421509,
+      "learning_rate": 8.093144632101026e-05,
+      "loss": 1.7538,
+      "step": 813
+    },
+    {
+      "epoch": 0.649164677804296,
+      "grad_norm": 1.6911081075668335,
+      "learning_rate": 8.077616204540897e-05,
+      "loss": 1.8258,
+      "step": 816
+    },
+    {
+      "epoch": 0.6515513126491647,
+      "grad_norm": 1.9907560348510742,
+      "learning_rate": 8.062039838142402e-05,
+      "loss": 1.7978,
+      "step": 819
+    },
+    {
+      "epoch": 0.6539379474940334,
+      "grad_norm": 1.9501363039016724,
+      "learning_rate": 8.046415775532585e-05,
+      "loss": 1.8116,
+      "step": 822
+    },
+    {
+      "epoch": 0.6563245823389021,
+      "grad_norm": 1.7937824726104736,
+      "learning_rate": 8.030744260081426e-05,
+      "loss": 1.8347,
+      "step": 825
+    },
+    {
+      "epoch": 0.6587112171837709,
+      "grad_norm": 1.682985782623291,
+      "learning_rate": 8.015025535898073e-05,
+      "loss": 1.8879,
+      "step": 828
+    },
+    {
+      "epoch": 0.6610978520286396,
+      "grad_norm": 1.7274394035339355,
+      "learning_rate": 7.999259847827015e-05,
+      "loss": 1.8931,
+      "step": 831
+    },
+    {
+      "epoch": 0.6634844868735084,
+      "grad_norm": 1.7429416179656982,
+      "learning_rate": 7.983447441444281e-05,
+      "loss": 1.7171,
+      "step": 834
+    },
+    {
+      "epoch": 0.665871121718377,
+      "grad_norm": 1.949879765510559,
+      "learning_rate": 7.967588563053616e-05,
+      "loss": 1.7779,
+      "step": 837
+    },
+    {
+      "epoch": 0.6682577565632458,
+      "grad_norm": 1.5538753271102905,
+      "learning_rate": 7.951683459682641e-05,
+      "loss": 1.8087,
+      "step": 840
+    },
+    {
+      "epoch": 0.6706443914081146,
+      "grad_norm": 1.7967875003814697,
+      "learning_rate": 7.935732379079008e-05,
+      "loss": 2.0304,
+      "step": 843
+    },
+    {
+      "epoch": 0.6730310262529833,
+      "grad_norm": 1.7717353105545044,
+      "learning_rate": 7.919735569706533e-05,
+      "loss": 1.904,
+      "step": 846
+    },
+    {
+      "epoch": 0.6754176610978521,
+      "grad_norm": 1.6083266735076904,
+      "learning_rate": 7.903693280741331e-05,
+      "loss": 1.8501,
+      "step": 849
+    },
+    {
+      "epoch": 0.6778042959427207,
+      "grad_norm": 1.6468119621276855,
+      "learning_rate": 7.887605762067945e-05,
+      "loss": 1.7535,
+      "step": 852
+    },
+    {
+      "epoch": 0.6801909307875895,
+      "grad_norm": 1.6218470335006714,
+      "learning_rate": 7.871473264275429e-05,
+      "loss": 1.7495,
+      "step": 855
+    },
+    {
+      "epoch": 0.6825775656324582,
+      "grad_norm": 1.6359236240386963,
+      "learning_rate": 7.855296038653475e-05,
+      "loss": 2.0507,
+      "step": 858
+    },
+    {
+      "epoch": 0.684964200477327,
+      "grad_norm": 1.4922749996185303,
+      "learning_rate": 7.83907433718847e-05,
+      "loss": 1.7931,
+      "step": 861
+    },
+    {
+      "epoch": 0.6873508353221957,
+      "grad_norm": 1.5041239261627197,
+      "learning_rate": 7.82280841255959e-05,
+      "loss": 1.7704,
+      "step": 864
+    },
+    {
+      "epoch": 0.6897374701670644,
+      "grad_norm": 2.032655954360962,
+      "learning_rate": 7.80649851813486e-05,
+      "loss": 1.9174,
+      "step": 867
+    },
+    {
+      "epoch": 0.6921241050119332,
+      "grad_norm": 1.7632269859313965,
+      "learning_rate": 7.790144907967201e-05,
+      "loss": 1.7885,
+      "step": 870
+    },
+    {
+      "epoch": 0.6945107398568019,
+      "grad_norm": 1.7323729991912842,
+      "learning_rate": 7.773747836790481e-05,
+      "loss": 1.9919,
+      "step": 873
+    },
+    {
+      "epoch": 0.6968973747016707,
+      "grad_norm": 2.3218891620635986,
+      "learning_rate": 7.757307560015538e-05,
+      "loss": 1.7896,
+      "step": 876
+    },
+    {
+      "epoch": 0.6992840095465394,
+      "grad_norm": 1.620492935180664,
+      "learning_rate": 7.740824333726213e-05,
+      "loss": 1.748,
+      "step": 879
+    },
+    {
+      "epoch": 0.7016706443914081,
+      "grad_norm": 1.6885743141174316,
+      "learning_rate": 7.724298414675353e-05,
+      "loss": 1.7732,
+      "step": 882
+    },
+    {
+      "epoch": 0.7040572792362768,
+      "grad_norm": 1.8093699216842651,
+      "learning_rate": 7.707730060280812e-05,
+      "loss": 1.9147,
+      "step": 885
+    },
+    {
+      "epoch": 0.7064439140811456,
+      "grad_norm": 1.7837680578231812,
+      "learning_rate": 7.691119528621444e-05,
+      "loss": 1.8391,
+      "step": 888
+    },
+    {
+      "epoch": 0.7088305489260143,
+      "grad_norm": 1.75551176071167,
+      "learning_rate": 7.674467078433081e-05,
+      "loss": 1.9519,
+      "step": 891
+    },
+    {
+      "epoch": 0.711217183770883,
+      "grad_norm": 1.791812777519226,
+      "learning_rate": 7.657772969104508e-05,
+      "loss": 1.6442,
+      "step": 894
+    },
+    {
+      "epoch": 0.7136038186157518,
+      "grad_norm": 1.6638917922973633,
+      "learning_rate": 7.641037460673412e-05,
+      "loss": 1.6225,
+      "step": 897
+    },
+    {
+      "epoch": 0.7159904534606205,
+      "grad_norm": 1.479506015777588,
+      "learning_rate": 7.624260813822342e-05,
+      "loss": 1.7162,
+      "step": 900
+    },
+    {
+      "epoch": 0.7183770883054893,
+      "grad_norm": 1.9166977405548096,
+      "learning_rate": 7.607443289874642e-05,
+      "loss": 1.8657,
+      "step": 903
+    },
+    {
+      "epoch": 0.720763723150358,
+      "grad_norm": 1.4827370643615723,
+      "learning_rate": 7.590585150790389e-05,
+      "loss": 1.9136,
+      "step": 906
+    },
+    {
+      "epoch": 0.7231503579952268,
+      "grad_norm": 1.5117080211639404,
+      "learning_rate": 7.573686659162293e-05,
+      "loss": 1.8548,
+      "step": 909
+    },
+    {
+      "epoch": 0.7255369928400954,
+      "grad_norm": 1.8482357263565063,
+      "learning_rate": 7.556748078211635e-05,
+      "loss": 1.9555,
+      "step": 912
+    },
+    {
+      "epoch": 0.7279236276849642,
+      "grad_norm": 1.6880775690078735,
+      "learning_rate": 7.53976967178414e-05,
+      "loss": 1.8775,
+      "step": 915
+    },
+    {
+      "epoch": 0.7303102625298329,
+      "grad_norm": 1.5047417879104614,
+      "learning_rate": 7.522751704345887e-05,
+      "loss": 1.8849,
+      "step": 918
+    },
+    {
+      "epoch": 0.7326968973747017,
+      "grad_norm": 1.5307697057724,
+      "learning_rate": 7.505694440979178e-05,
+      "loss": 1.8404,
+      "step": 921
+    },
+    {
+      "epoch": 0.7350835322195705,
+      "grad_norm": 1.508344054222107,
+      "learning_rate": 7.488598147378416e-05,
+      "loss": 1.666,
+      "step": 924
+    },
+    {
+      "epoch": 0.7374701670644391,
+      "grad_norm": 1.7632466554641724,
+      "learning_rate": 7.471463089845956e-05,
+      "loss": 1.7466,
+      "step": 927
+    },
+    {
+      "epoch": 0.7398568019093079,
+      "grad_norm": 1.6543248891830444,
+      "learning_rate": 7.454289535287968e-05,
+      "loss": 1.7259,
+      "step": 930
+    },
+    {
+      "epoch": 0.7422434367541766,
+      "grad_norm": 2.329713821411133,
+      "learning_rate": 7.437077751210279e-05,
+      "loss": 1.9443,
+      "step": 933
+    },
+    {
+      "epoch": 0.7446300715990454,
+      "grad_norm": 1.585302472114563,
+      "learning_rate": 7.419828005714194e-05,
+      "loss": 1.8221,
+      "step": 936
+    },
+    {
+      "epoch": 0.747016706443914,
+      "grad_norm": 1.523518681526184,
+      "learning_rate": 7.402540567492337e-05,
+      "loss": 1.7825,
+      "step": 939
+    },
+    {
+      "epoch": 0.7494033412887828,
+      "grad_norm": 1.5896198749542236,
+      "learning_rate": 7.385215705824449e-05,
+      "loss": 1.9895,
+      "step": 942
+    },
+    {
+      "epoch": 0.7517899761336515,
+      "grad_norm": 1.7857471704483032,
+      "learning_rate": 7.367853690573208e-05,
+      "loss": 1.6531,
+      "step": 945
+    },
+    {
+      "epoch": 0.7541766109785203,
+      "grad_norm": 1.8661036491394043,
+      "learning_rate": 7.350454792180016e-05,
+      "loss": 1.7411,
+      "step": 948
+    },
+    {
+      "epoch": 0.7565632458233891,
+      "grad_norm": 1.7312443256378174,
+      "learning_rate": 7.333019281660789e-05,
+      "loss": 1.9503,
+      "step": 951
+    },
+    {
+      "epoch": 0.7589498806682577,
+      "grad_norm": 1.8219163417816162,
+      "learning_rate": 7.31554743060174e-05,
+      "loss": 1.7237,
+      "step": 954
+    },
+    {
+      "epoch": 0.7613365155131265,
+      "grad_norm": 1.7200877666473389,
+      "learning_rate": 7.298039511155138e-05,
+      "loss": 1.8042,
+      "step": 957
+    },
+    {
+      "epoch": 0.7637231503579952,
+      "grad_norm": 1.5407986640930176,
+      "learning_rate": 7.280495796035079e-05,
+      "loss": 1.8225,
+      "step": 960
+    },
+    {
+      "epoch": 0.766109785202864,
+      "grad_norm": 1.5745642185211182,
+      "learning_rate": 7.262916558513237e-05,
+      "loss": 1.6478,
+      "step": 963
+    },
+    {
+      "epoch": 0.7684964200477327,
+      "grad_norm": 1.8857331275939941,
+      "learning_rate": 7.245302072414601e-05,
+      "loss": 1.8026,
+      "step": 966
+    },
+    {
+      "epoch": 0.7708830548926014,
+      "grad_norm": 1.4270589351654053,
+      "learning_rate": 7.227652612113213e-05,
+      "loss": 1.6531,
+      "step": 969
+    },
+    {
+      "epoch": 0.7732696897374701,
+      "grad_norm": 1.530493140220642,
+      "learning_rate": 7.209968452527896e-05,
+      "loss": 1.7553,
+      "step": 972
+    },
+    {
+      "epoch": 0.7756563245823389,
+      "grad_norm": 1.6771613359451294,
+      "learning_rate": 7.192249869117971e-05,
+      "loss": 1.8374,
+      "step": 975
+    },
+    {
+      "epoch": 0.7780429594272077,
+      "grad_norm": 1.7160065174102783,
+      "learning_rate": 7.174497137878966e-05,
+      "loss": 1.7429,
+      "step": 978
+    },
+    {
+      "epoch": 0.7804295942720764,
+      "grad_norm": 1.518904685974121,
+      "learning_rate": 7.156710535338312e-05,
+      "loss": 1.8843,
+      "step": 981
+    },
+    {
+      "epoch": 0.7828162291169452,
+      "grad_norm": 1.7113001346588135,
+      "learning_rate": 7.138890338551048e-05,
+      "loss": 1.8249,
+      "step": 984
+    },
+    {
+      "epoch": 0.7852028639618138,
+      "grad_norm": 1.6391565799713135,
+      "learning_rate": 7.121036825095492e-05,
+      "loss": 1.6807,
+      "step": 987
+    },
+    {
+      "epoch": 0.7875894988066826,
+      "grad_norm": 1.6043463945388794,
+      "learning_rate": 7.103150273068921e-05,
+      "loss": 1.7299,
+      "step": 990
+    },
+    {
+      "epoch": 0.7899761336515513,
+      "grad_norm": 1.5004619359970093,
+      "learning_rate": 7.085230961083249e-05,
+      "loss": 1.8501,
+      "step": 993
+    },
+    {
+      "epoch": 0.7923627684964201,
+      "grad_norm": 1.893096685409546,
+      "learning_rate": 7.067279168260671e-05,
+      "loss": 1.8326,
+      "step": 996
+    },
+    {
+      "epoch": 0.7947494033412887,
+      "grad_norm": 1.5315916538238525,
+      "learning_rate": 7.04929517422933e-05,
+      "loss": 1.7833,
+      "step": 999
+    },
+    {
+      "epoch": 0.7971360381861575,
+      "grad_norm": 1.5632952451705933,
+      "learning_rate": 7.031279259118946e-05,
+      "loss": 1.6346,
+      "step": 1002
+    },
+    {
+      "epoch": 0.7995226730310262,
+      "grad_norm": 1.9762367010116577,
+      "learning_rate": 7.013231703556471e-05,
+      "loss": 1.8849,
+      "step": 1005
+    },
+    {
+      "epoch": 0.801909307875895,
+      "grad_norm": 1.5724525451660156,
+      "learning_rate": 6.995152788661705e-05,
+      "loss": 1.7792,
+      "step": 1008
+    },
+    {
+      "epoch": 0.8042959427207638,
+      "grad_norm": 1.4417201280593872,
+      "learning_rate": 6.977042796042917e-05,
+      "loss": 1.7516,
+      "step": 1011
+    },
+    {
+      "epoch": 0.8066825775656324,
+      "grad_norm": 1.6131057739257812,
+      "learning_rate": 6.958902007792466e-05,
+      "loss": 1.7614,
+      "step": 1014
+    },
+    {
+      "epoch": 0.8090692124105012,
+      "grad_norm": 2.5175702571868896,
+      "learning_rate": 6.940730706482399e-05,
+      "loss": 1.8208,
+      "step": 1017
+    },
+    {
+      "epoch": 0.8114558472553699,
+      "grad_norm": 1.5602608919143677,
+      "learning_rate": 6.922529175160054e-05,
+      "loss": 1.7046,
+      "step": 1020
+    },
+    {
+      "epoch": 0.8138424821002387,
+      "grad_norm": 1.5071675777435303,
+      "learning_rate": 6.904297697343655e-05,
+      "loss": 1.9148,
+      "step": 1023
+    },
+    {
+      "epoch": 0.8162291169451074,
+      "grad_norm": 1.4578834772109985,
+      "learning_rate": 6.886036557017881e-05,
+      "loss": 1.8546,
+      "step": 1026
+    },
+    {
+      "epoch": 0.8186157517899761,
+      "grad_norm": 1.7864019870758057,
+      "learning_rate": 6.867746038629462e-05,
+      "loss": 1.9423,
+      "step": 1029
+    },
+    {
+      "epoch": 0.8210023866348448,
+      "grad_norm": 1.4294066429138184,
+      "learning_rate": 6.849426427082735e-05,
+      "loss": 1.7797,
+      "step": 1032
+    },
+    {
+      "epoch": 0.8233890214797136,
+      "grad_norm": 2.900899648666382,
+      "learning_rate": 6.83107800773521e-05,
+      "loss": 1.8395,
+      "step": 1035
+    },
+    {
+      "epoch": 0.8257756563245824,
+      "grad_norm": 1.5912294387817383,
+      "learning_rate": 6.812701066393124e-05,
+      "loss": 1.7345,
+      "step": 1038
+    },
+    {
+      "epoch": 0.8281622911694511,
+      "grad_norm": 1.6191726922988892,
+      "learning_rate": 6.79429588930699e-05,
+      "loss": 1.7563,
+      "step": 1041
+    },
+    {
+      "epoch": 0.8305489260143198,
+      "grad_norm": 1.926048755645752,
+      "learning_rate": 6.775862763167142e-05,
+      "loss": 1.7473,
+      "step": 1044
+    },
+    {
+      "epoch": 0.8329355608591885,
+      "grad_norm": 1.5704717636108398,
+      "learning_rate": 6.757401975099262e-05,
+      "loss": 1.6788,
+      "step": 1047
+    },
+    {
+      "epoch": 0.8353221957040573,
+      "grad_norm": 2.986739158630371,
+      "learning_rate": 6.738913812659912e-05,
+      "loss": 1.9091,
+      "step": 1050
+    },
+    {
+      "epoch": 0.837708830548926,
+      "grad_norm": 2.8111684322357178,
+      "learning_rate": 6.720398563832055e-05,
+      "loss": 1.7738,
+      "step": 1053
+    },
+    {
+      "epoch": 0.8400954653937948,
+      "grad_norm": 1.5233489274978638,
+      "learning_rate": 6.701856517020565e-05,
+      "loss": 1.869,
+      "step": 1056
+    },
+    {
+      "epoch": 0.8424821002386634,
+      "grad_norm": 1.679288387298584,
+      "learning_rate": 6.683287961047742e-05,
+      "loss": 1.977,
+      "step": 1059
+    },
+    {
+      "epoch": 0.8448687350835322,
+      "grad_norm": 1.5280253887176514,
+      "learning_rate": 6.664693185148807e-05,
+      "loss": 1.7278,
+      "step": 1062
+    },
+    {
+      "epoch": 0.847255369928401,
+      "grad_norm": 1.5906157493591309,
+      "learning_rate": 6.646072478967397e-05,
+      "loss": 1.8965,
+      "step": 1065
+    },
+    {
+      "epoch": 0.8496420047732697,
+      "grad_norm": 1.6531226634979248,
+      "learning_rate": 6.627426132551058e-05,
+      "loss": 1.77,
+      "step": 1068
+    },
+    {
+      "epoch": 0.8520286396181385,
+      "grad_norm": 1.5413316488265991,
+      "learning_rate": 6.608754436346725e-05,
+      "loss": 1.7051,
+      "step": 1071
+    },
+    {
+      "epoch": 0.8544152744630071,
+      "grad_norm": 1.7978452444076538,
+      "learning_rate": 6.590057681196191e-05,
+      "loss": 1.6797,
+      "step": 1074
+    },
+    {
+      "epoch": 0.8568019093078759,
+      "grad_norm": 1.6512346267700195,
+      "learning_rate": 6.571336158331589e-05,
+      "loss": 1.9775,
+      "step": 1077
+    },
+    {
+      "epoch": 0.8591885441527446,
+      "grad_norm": 1.7628110647201538,
+      "learning_rate": 6.552590159370844e-05,
+      "loss": 1.6468,
+      "step": 1080
+    },
+    {
+      "epoch": 0.8615751789976134,
+      "grad_norm": 1.659655213356018,
+      "learning_rate": 6.53381997631314e-05,
+      "loss": 1.8676,
+      "step": 1083
+    },
+    {
+      "epoch": 0.863961813842482,
+      "grad_norm": 1.5994445085525513,
+      "learning_rate": 6.515025901534364e-05,
+      "loss": 1.7799,
+      "step": 1086
+    },
+    {
+      "epoch": 0.8663484486873508,
+      "grad_norm": 4.107300758361816,
+      "learning_rate": 6.496208227782556e-05,
+      "loss": 1.7622,
+      "step": 1089
+    },
+    {
+      "epoch": 0.8687350835322196,
+      "grad_norm": 1.6633572578430176,
+      "learning_rate": 6.477367248173352e-05,
+      "loss": 1.7943,
+      "step": 1092
+    },
+    {
+      "epoch": 0.8711217183770883,
+      "grad_norm": 1.5715115070343018,
+      "learning_rate": 6.458503256185404e-05,
+      "loss": 1.791,
+      "step": 1095
+    },
+    {
+      "epoch": 0.8735083532219571,
+      "grad_norm": 1.4537943601608276,
+      "learning_rate": 6.439616545655834e-05,
+      "loss": 1.6835,
+      "step": 1098
+    },
+    {
+      "epoch": 0.8758949880668258,
+      "grad_norm": 1.4551641941070557,
+      "learning_rate": 6.420707410775626e-05,
+      "loss": 1.8273,
+      "step": 1101
+    },
+    {
+      "epoch": 0.8782816229116945,
+      "grad_norm": 1.7414036989212036,
+      "learning_rate": 6.401776146085072e-05,
+      "loss": 2.0934,
+      "step": 1104
+    },
+    {
+      "epoch": 0.8806682577565632,
+      "grad_norm": 1.602457046508789,
+      "learning_rate": 6.382823046469167e-05,
+      "loss": 1.8388,
+      "step": 1107
+    },
+    {
+      "epoch": 0.883054892601432,
+      "grad_norm": 1.549912929534912,
+      "learning_rate": 6.363848407153016e-05,
+      "loss": 1.6429,
+      "step": 1110
+    },
+    {
+      "epoch": 0.8854415274463007,
+      "grad_norm": 1.4284569025039673,
+      "learning_rate": 6.344852523697247e-05,
+      "loss": 1.8545,
+      "step": 1113
+    },
+    {
+      "epoch": 0.8878281622911695,
+      "grad_norm": 1.83491849899292,
+      "learning_rate": 6.325835691993394e-05,
+      "loss": 1.574,
+      "step": 1116
+    },
+    {
+      "epoch": 0.8902147971360382,
+      "grad_norm": 1.6155121326446533,
+      "learning_rate": 6.306798208259297e-05,
+      "loss": 1.7415,
+      "step": 1119
+    },
+    {
+      "epoch": 0.8926014319809069,
+      "grad_norm": 1.5498522520065308,
+      "learning_rate": 6.287740369034485e-05,
+      "loss": 1.5622,
+      "step": 1122
+    },
+    {
+      "epoch": 0.8949880668257757,
+      "grad_norm": 1.6388617753982544,
+      "learning_rate": 6.26866247117555e-05,
+      "loss": 1.6144,
+      "step": 1125
+    },
+    {
+      "epoch": 0.8973747016706444,
+      "grad_norm": 1.4823899269104004,
+      "learning_rate": 6.249564811851543e-05,
+      "loss": 1.8225,
+      "step": 1128
+    },
+    {
+      "epoch": 0.8997613365155132,
+      "grad_norm": 1.4698792695999146,
+      "learning_rate": 6.230447688539316e-05,
+      "loss": 1.6339,
+      "step": 1131
+    },
+    {
+      "epoch": 0.9021479713603818,
+      "grad_norm": 1.6909016370773315,
+      "learning_rate": 6.211311399018916e-05,
+      "loss": 1.7918,
+      "step": 1134
+    },
+    {
+      "epoch": 0.9045346062052506,
+      "grad_norm": 1.4534494876861572,
+      "learning_rate": 6.192156241368929e-05,
+      "loss": 1.7715,
+      "step": 1137
+    },
+    {
+      "epoch": 0.9069212410501193,
+      "grad_norm": 1.719106912612915,
+      "learning_rate": 6.172982513961845e-05,
+      "loss": 1.7261,
+      "step": 1140
+    },
+    {
+      "epoch": 0.9093078758949881,
+      "grad_norm": 1.6923831701278687,
+      "learning_rate": 6.153790515459404e-05,
+      "loss": 1.6554,
+      "step": 1143
+    },
+    {
+      "epoch": 0.9116945107398569,
+      "grad_norm": 1.6443957090377808,
+      "learning_rate": 6.13458054480795e-05,
+      "loss": 1.8556,
+      "step": 1146
+    },
+    {
+      "epoch": 0.9140811455847255,
+      "grad_norm": 1.5689623355865479,
+      "learning_rate": 6.115352901233779e-05,
+      "loss": 1.8041,
+      "step": 1149
+    },
+    {
+      "epoch": 0.9164677804295943,
+      "grad_norm": 1.46951425075531,
+      "learning_rate": 6.096107884238458e-05,
+      "loss": 1.6515,
+      "step": 1152
+    },
+    {
+      "epoch": 0.918854415274463,
+      "grad_norm": 1.5866256952285767,
+      "learning_rate": 6.0768457935941817e-05,
+      "loss": 1.8007,
+      "step": 1155
+    },
+    {
+      "epoch": 0.9212410501193318,
+      "grad_norm": 1.9523183107376099,
+      "learning_rate": 6.0575669293390954e-05,
+      "loss": 1.8313,
+      "step": 1158
+    },
+    {
+      "epoch": 0.9236276849642004,
+      "grad_norm": 1.6216835975646973,
+      "learning_rate": 6.038271591772615e-05,
+      "loss": 1.8399,
+      "step": 1161
+    },
+    {
+      "epoch": 0.9260143198090692,
+      "grad_norm": 1.8278216123580933,
+      "learning_rate": 6.0189600814507604e-05,
+      "loss": 1.9067,
+      "step": 1164
+    },
+    {
+      "epoch": 0.9284009546539379,
+      "grad_norm": 1.2732576131820679,
+      "learning_rate": 5.9996326991814654e-05,
+      "loss": 1.6042,
+      "step": 1167
+    },
+    {
+      "epoch": 0.9307875894988067,
+      "grad_norm": 1.75897216796875,
+      "learning_rate": 5.980289746019892e-05,
+      "loss": 1.8796,
+      "step": 1170
+    },
+    {
+      "epoch": 0.9331742243436754,
+      "grad_norm": 1.5754450559616089,
+      "learning_rate": 5.9609315232637483e-05,
+      "loss": 1.5981,
+      "step": 1173
+    },
+    {
+      "epoch": 0.9355608591885441,
+      "grad_norm": 1.7975653409957886,
+      "learning_rate": 5.941558332448589e-05,
+      "loss": 1.638,
+      "step": 1176
+    },
+    {
+      "epoch": 0.9379474940334129,
+      "grad_norm": 1.6017708778381348,
+      "learning_rate": 5.922170475343125e-05,
+      "loss": 1.6904,
+      "step": 1179
+    },
+    {
+      "epoch": 0.9403341288782816,
+      "grad_norm": 1.6121450662612915,
+      "learning_rate": 5.9027682539445104e-05,
+      "loss": 1.69,
+      "step": 1182
+    },
+    {
+      "epoch": 0.9427207637231504,
+      "grad_norm": 1.5319181680679321,
+      "learning_rate": 5.883351970473654e-05,
+      "loss": 1.8468,
+      "step": 1185
+    },
+    {
+      "epoch": 0.9451073985680191,
+      "grad_norm": 1.7490177154541016,
+      "learning_rate": 5.863921927370498e-05,
+      "loss": 1.7477,
+      "step": 1188
+    },
+    {
+      "epoch": 0.9474940334128878,
+      "grad_norm": 1.546705722808838,
+      "learning_rate": 5.8444784272893175e-05,
+      "loss": 1.6837,
+      "step": 1191
+    },
+    {
+      "epoch": 0.9498806682577565,
+      "grad_norm": 1.576828956604004,
+      "learning_rate": 5.8250217730939973e-05,
+      "loss": 1.6976,
+      "step": 1194
+    },
+    {
+      "epoch": 0.9522673031026253,
+      "grad_norm": 1.443121075630188,
+      "learning_rate": 5.8055522678533225e-05,
+      "loss": 1.6661,
+      "step": 1197
+    },
+    {
+      "epoch": 0.954653937947494,
+      "grad_norm": 1.619332194328308,
+      "learning_rate": 5.786070214836254e-05,
+      "loss": 1.6282,
+      "step": 1200
+    },
+    {
+      "epoch": 0.9570405727923628,
+      "grad_norm": 1.516404628753662,
+      "learning_rate": 5.7665759175072034e-05,
+      "loss": 1.8743,
+      "step": 1203
+    },
+    {
+      "epoch": 0.9594272076372315,
+      "grad_norm": 1.842561960220337,
+      "learning_rate": 5.747069679521305e-05,
+      "loss": 1.7742,
+      "step": 1206
+    },
+    {
+      "epoch": 0.9618138424821002,
+      "grad_norm": 1.5098801851272583,
+      "learning_rate": 5.727551804719693e-05,
+      "loss": 1.7087,
+      "step": 1209
+    },
+    {
+      "epoch": 0.964200477326969,
+      "grad_norm": 1.611047387123108,
+      "learning_rate": 5.708022597124758e-05,
+      "loss": 1.6934,
+      "step": 1212
+    },
+    {
+      "epoch": 0.9665871121718377,
+      "grad_norm": 1.5745174884796143,
+      "learning_rate": 5.688482360935423e-05,
+      "loss": 1.8729,
+      "step": 1215
+    },
+    {
+      "epoch": 0.9689737470167065,
+      "grad_norm": 1.4478429555892944,
+      "learning_rate": 5.668931400522396e-05,
+      "loss": 1.801,
+      "step": 1218
+    },
+    {
+      "epoch": 0.9713603818615751,
+      "grad_norm": 2.9802427291870117,
+      "learning_rate": 5.649370020423431e-05,
+      "loss": 1.6937,
+      "step": 1221
+    },
+    {
+      "epoch": 0.9737470167064439,
+      "grad_norm": 1.6159980297088623,
+      "learning_rate": 5.629798525338589e-05,
+      "loss": 1.712,
+      "step": 1224
+    },
+    {
+      "epoch": 0.9761336515513126,
+      "grad_norm": 1.6687465906143188,
+      "learning_rate": 5.6102172201254835e-05,
+      "loss": 1.7582,
+      "step": 1227
+    },
+    {
+      "epoch": 0.9785202863961814,
+      "grad_norm": 1.318992018699646,
+      "learning_rate": 5.5906264097945407e-05,
+      "loss": 1.7913,
+      "step": 1230
+    },
+    {
+      "epoch": 0.9809069212410502,
+      "grad_norm": 1.6671638488769531,
+      "learning_rate": 5.5710263995042434e-05,
+      "loss": 1.8547,
+      "step": 1233
+    },
+    {
+      "epoch": 0.9832935560859188,
+      "grad_norm": 1.460523247718811,
+      "learning_rate": 5.551417494556376e-05,
+      "loss": 1.7699,
+      "step": 1236
+    },
+    {
+      "epoch": 0.9856801909307876,
+      "grad_norm": 1.4564532041549683,
+      "learning_rate": 5.531800000391275e-05,
+      "loss": 1.7636,
+      "step": 1239
+    },
+    {
+      "epoch": 0.9880668257756563,
+      "grad_norm": 1.7387609481811523,
+      "learning_rate": 5.5121742225830665e-05,
+      "loss": 1.8602,
+      "step": 1242
+    },
+    {
+      "epoch": 0.9904534606205251,
+      "grad_norm": 1.8153102397918701,
+      "learning_rate": 5.4925404668349076e-05,
+      "loss": 1.797,
+      "step": 1245
+    },
+    {
+      "epoch": 0.9928400954653938,
+      "grad_norm": 1.4723916053771973,
+      "learning_rate": 5.472899038974225e-05,
+      "loss": 1.7051,
+      "step": 1248
+    },
+    {
+      "epoch": 0.9952267303102625,
+      "grad_norm": 2.961282968521118,
+      "learning_rate": 5.45325024494795e-05,
+      "loss": 1.7661,
+      "step": 1251
+    },
+    {
+      "epoch": 0.9976133651551312,
+      "grad_norm": 1.6182713508605957,
+      "learning_rate": 5.433594390817756e-05,
+      "loss": 1.875,
+      "step": 1254
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.34993839263916,
+      "learning_rate": 5.413931782755283e-05,
+      "loss": 1.8514,
+      "step": 1257
+    }
+  ],
+  "logging_steps": 3,
+  "max_steps": 2514,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 1257,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.127117763309732e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-1257/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba2b4600d8e12f46b8c1ffc58c8d4c06c4c5cf6613ee34768ecf329421dbd41c
+size 5432

checkpoint-2514/config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-shn_Mymr-coli-mean/",
+  "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-2514/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-2514/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e68269fa15f4e9d67cdc2ea13db402f6bd861ef208d22940103ae16dcc7fed7
+size 4950070472

checkpoint-2514/model-00002-of-00006.safetensors

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

checkpoint-2514/model-00003-of-00006.safetensors

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

checkpoint-2514/model-00004-of-00006.safetensors

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

checkpoint-2514/model-00005-of-00006.safetensors

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

checkpoint-2514/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:04abe5dec827857e66a73157a668898195bb05f90d50a2cc691ae1b2aa075825
+size 4614509080

checkpoint-2514/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-2514/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e82a64bc3a353d15aad94e1cfb61e98dc7373afd2a53fb07cd56d87e935c888d
+size 13055369358

checkpoint-2514/rng_state.pth

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

checkpoint-2514/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:495d531179727c69e207f95ad286abd825844ea668463a8ef0048eebda73f74f
+size 1064

checkpoint-2514/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba2b4600d8e12f46b8c1ffc58c8d4c06c4c5cf6613ee34768ecf329421dbd41c
+size 5432

config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-shn_Mymr-coli-mean/",
+  "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 sha256:4e68269fa15f4e9d67cdc2ea13db402f6bd861ef208d22940103ae16dcc7fed7
+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 sha256:04abe5dec827857e66a73157a668898195bb05f90d50a2cc691ae1b2aa075825
+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:ba2b4600d8e12f46b8c1ffc58c8d4c06c4c5cf6613ee34768ecf329421dbd41c
+size 5432