Upload folder using huggingface_hub

checkpoint-1990/config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-ace_Latn-coli-align/",
+  "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-1990/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-1990/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dab1ba52a0dcbed9b5315742c600110fdc1df4fe9a936e9d7840174e0740b493
+size 4950070472

checkpoint-1990/model-00002-of-00006.safetensors

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

checkpoint-1990/model-00003-of-00006.safetensors

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

checkpoint-1990/model-00004-of-00006.safetensors

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

checkpoint-1990/model-00005-of-00006.safetensors

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

checkpoint-1990/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95d189204f4a7c654e0b96823d05202ca6230116fe3450954291fbcf93c8b730
+size 4614509080

checkpoint-1990/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-1990/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb6baa4de886068848911efb0617c639ee659fcca4029c3e0bde70dad8a21d7d
+size 13055369358

checkpoint-1990/rng_state.pth

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

checkpoint-1990/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:520a26338c9dc3d5685bd6a4d759b53fe390b7f335dc0e10e816597e74829bca
+size 1064

checkpoint-1990/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a6292fe987426601ff3a282fbc93263d2cc3acea82d69c0e4d76a82ed072116
+size 5432

checkpoint-995/config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-ace_Latn-coli-align/",
+  "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-995/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-995/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2547efa26c3267e2c7c061e57bf03b654c135c03f44c1881c768760a275e03cb
+size 4950070472

checkpoint-995/model-00002-of-00006.safetensors

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

checkpoint-995/model-00003-of-00006.safetensors

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

checkpoint-995/model-00004-of-00006.safetensors

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

checkpoint-995/model-00005-of-00006.safetensors

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

checkpoint-995/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:609696a30d973da38a435585655cb8292d615b2acebcde261f55896357df8607
+size 4614509080

checkpoint-995/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-995/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e36664a22d5c3db6436c4b50939e19a28e352e8d8e720f0536f2391c24d10a9
+size 13055369358

checkpoint-995/rng_state.pth

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

checkpoint-995/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45251de673384cd911a0c6ed2cbc5c9d2588a3d9a33cbd6184e2b9a739e772b4
+size 1064

checkpoint-995/trainer_state.json

@@ -0,0 +1,3512 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 995,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0020100502512562816,
+      "grad_norm": 19.729604721069336,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 6.6152,
+      "step": 2
+    },
+    {
+      "epoch": 0.004020100502512563,
+      "grad_norm": 16.32933807373047,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 6.684,
+      "step": 4
+    },
+    {
+      "epoch": 0.006030150753768844,
+      "grad_norm": 9.964282989501953,
+      "learning_rate": 6e-06,
+      "loss": 6.4803,
+      "step": 6
+    },
+    {
+      "epoch": 0.008040201005025126,
+      "grad_norm": 9.893184661865234,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 5.9697,
+      "step": 8
+    },
+    {
+      "epoch": 0.010050251256281407,
+      "grad_norm": 7.443181037902832,
+      "learning_rate": 1e-05,
+      "loss": 5.7487,
+      "step": 10
+    },
+    {
+      "epoch": 0.012060301507537688,
+      "grad_norm": 5.485438346862793,
+      "learning_rate": 1.2e-05,
+      "loss": 5.6723,
+      "step": 12
+    },
+    {
+      "epoch": 0.01407035175879397,
+      "grad_norm": 4.975802421569824,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 5.4903,
+      "step": 14
+    },
+    {
+      "epoch": 0.016080402010050253,
+      "grad_norm": 5.055726051330566,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 5.2516,
+      "step": 16
+    },
+    {
+      "epoch": 0.018090452261306532,
+      "grad_norm": 4.720687389373779,
+      "learning_rate": 1.8e-05,
+      "loss": 5.0191,
+      "step": 18
+    },
+    {
+      "epoch": 0.020100502512562814,
+      "grad_norm": 3.445399761199951,
+      "learning_rate": 2e-05,
+      "loss": 4.8732,
+      "step": 20
+    },
+    {
+      "epoch": 0.022110552763819097,
+      "grad_norm": 3.533928155899048,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 5.1519,
+      "step": 22
+    },
+    {
+      "epoch": 0.024120603015075376,
+      "grad_norm": 3.4936132431030273,
+      "learning_rate": 2.4e-05,
+      "loss": 4.9552,
+      "step": 24
+    },
+    {
+      "epoch": 0.02613065326633166,
+      "grad_norm": 3.037400722503662,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 4.622,
+      "step": 26
+    },
+    {
+      "epoch": 0.02814070351758794,
+      "grad_norm": 4.790655136108398,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 4.5671,
+      "step": 28
+    },
+    {
+      "epoch": 0.03015075376884422,
+      "grad_norm": 6.313861846923828,
+      "learning_rate": 3e-05,
+      "loss": 4.3285,
+      "step": 30
+    },
+    {
+      "epoch": 0.032160804020100506,
+      "grad_norm": 2.662900447845459,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 4.2636,
+      "step": 32
+    },
+    {
+      "epoch": 0.034170854271356785,
+      "grad_norm": 2.5860118865966797,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 4.3172,
+      "step": 34
+    },
+    {
+      "epoch": 0.036180904522613064,
+      "grad_norm": 2.960880994796753,
+      "learning_rate": 3.6e-05,
+      "loss": 3.9687,
+      "step": 36
+    },
+    {
+      "epoch": 0.03819095477386935,
+      "grad_norm": 3.4453978538513184,
+      "learning_rate": 3.8e-05,
+      "loss": 3.8505,
+      "step": 38
+    },
+    {
+      "epoch": 0.04020100502512563,
+      "grad_norm": 2.6721041202545166,
+      "learning_rate": 4e-05,
+      "loss": 3.971,
+      "step": 40
+    },
+    {
+      "epoch": 0.04221105527638191,
+      "grad_norm": 2.6484851837158203,
+      "learning_rate": 4.2e-05,
+      "loss": 3.8536,
+      "step": 42
+    },
+    {
+      "epoch": 0.044221105527638194,
+      "grad_norm": 6.354692459106445,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 3.8466,
+      "step": 44
+    },
+    {
+      "epoch": 0.04623115577889447,
+      "grad_norm": 2.6469643115997314,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 3.5942,
+      "step": 46
+    },
+    {
+      "epoch": 0.04824120603015075,
+      "grad_norm": 2.7454721927642822,
+      "learning_rate": 4.8e-05,
+      "loss": 3.526,
+      "step": 48
+    },
+    {
+      "epoch": 0.05025125628140704,
+      "grad_norm": 2.445122480392456,
+      "learning_rate": 5e-05,
+      "loss": 3.5655,
+      "step": 50
+    },
+    {
+      "epoch": 0.05226130653266332,
+      "grad_norm": 2.458037853240967,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 3.3741,
+      "step": 52
+    },
+    {
+      "epoch": 0.054271356783919596,
+      "grad_norm": 2.469712972640991,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 3.4174,
+      "step": 54
+    },
+    {
+      "epoch": 0.05628140703517588,
+      "grad_norm": 2.475627899169922,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 3.3984,
+      "step": 56
+    },
+    {
+      "epoch": 0.05829145728643216,
+      "grad_norm": 2.6965248584747314,
+      "learning_rate": 5.8e-05,
+      "loss": 3.5654,
+      "step": 58
+    },
+    {
+      "epoch": 0.06030150753768844,
+      "grad_norm": 2.3575079441070557,
+      "learning_rate": 6e-05,
+      "loss": 3.2454,
+      "step": 60
+    },
+    {
+      "epoch": 0.062311557788944726,
+      "grad_norm": 2.371737480163574,
+      "learning_rate": 6.2e-05,
+      "loss": 3.361,
+      "step": 62
+    },
+    {
+      "epoch": 0.06432160804020101,
+      "grad_norm": 2.144615650177002,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 3.2241,
+      "step": 64
+    },
+    {
+      "epoch": 0.06633165829145729,
+      "grad_norm": 1.9614999294281006,
+      "learning_rate": 6.6e-05,
+      "loss": 3.1206,
+      "step": 66
+    },
+    {
+      "epoch": 0.06834170854271357,
+      "grad_norm": 2.0626800060272217,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 3.3899,
+      "step": 68
+    },
+    {
+      "epoch": 0.07035175879396985,
+      "grad_norm": 2.669525623321533,
+      "learning_rate": 7e-05,
+      "loss": 3.2405,
+      "step": 70
+    },
+    {
+      "epoch": 0.07236180904522613,
+      "grad_norm": 2.620374917984009,
+      "learning_rate": 7.2e-05,
+      "loss": 3.3299,
+      "step": 72
+    },
+    {
+      "epoch": 0.0743718592964824,
+      "grad_norm": 2.2318763732910156,
+      "learning_rate": 7.4e-05,
+      "loss": 2.9798,
+      "step": 74
+    },
+    {
+      "epoch": 0.0763819095477387,
+      "grad_norm": 2.28938889503479,
+      "learning_rate": 7.6e-05,
+      "loss": 3.2799,
+      "step": 76
+    },
+    {
+      "epoch": 0.07839195979899498,
+      "grad_norm": 2.176002264022827,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 3.2466,
+      "step": 78
+    },
+    {
+      "epoch": 0.08040201005025126,
+      "grad_norm": 2.2421767711639404,
+      "learning_rate": 8e-05,
+      "loss": 3.0629,
+      "step": 80
+    },
+    {
+      "epoch": 0.08241206030150754,
+      "grad_norm": 2.7618038654327393,
+      "learning_rate": 8.2e-05,
+      "loss": 3.2719,
+      "step": 82
+    },
+    {
+      "epoch": 0.08442211055276382,
+      "grad_norm": 2.1031386852264404,
+      "learning_rate": 8.4e-05,
+      "loss": 3.0094,
+      "step": 84
+    },
+    {
+      "epoch": 0.0864321608040201,
+      "grad_norm": 2.1169002056121826,
+      "learning_rate": 8.6e-05,
+      "loss": 3.021,
+      "step": 86
+    },
+    {
+      "epoch": 0.08844221105527639,
+      "grad_norm": 2.4154305458068848,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 3.0685,
+      "step": 88
+    },
+    {
+      "epoch": 0.09045226130653267,
+      "grad_norm": 2.4994680881500244,
+      "learning_rate": 9e-05,
+      "loss": 2.8943,
+      "step": 90
+    },
+    {
+      "epoch": 0.09246231155778895,
+      "grad_norm": 3.4030494689941406,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 2.8389,
+      "step": 92
+    },
+    {
+      "epoch": 0.09447236180904522,
+      "grad_norm": 5.54694128036499,
+      "learning_rate": 9.4e-05,
+      "loss": 3.0157,
+      "step": 94
+    },
+    {
+      "epoch": 0.0964824120603015,
+      "grad_norm": 2.363124370574951,
+      "learning_rate": 9.6e-05,
+      "loss": 2.811,
+      "step": 96
+    },
+    {
+      "epoch": 0.09849246231155778,
+      "grad_norm": 2.0617117881774902,
+      "learning_rate": 9.8e-05,
+      "loss": 2.8819,
+      "step": 98
+    },
+    {
+      "epoch": 0.10050251256281408,
+      "grad_norm": 2.354085922241211,
+      "learning_rate": 0.0001,
+      "loss": 3.0962,
+      "step": 100
+    },
+    {
+      "epoch": 0.10251256281407035,
+      "grad_norm": 2.3361897468566895,
+      "learning_rate": 9.999972370327507e-05,
+      "loss": 3.0454,
+      "step": 102
+    },
+    {
+      "epoch": 0.10452261306532663,
+      "grad_norm": 2.2245845794677734,
+      "learning_rate": 9.999889481615387e-05,
+      "loss": 3.1027,
+      "step": 104
+    },
+    {
+      "epoch": 0.10653266331658291,
+      "grad_norm": 2.069746494293213,
+      "learning_rate": 9.999751334779716e-05,
+      "loss": 2.915,
+      "step": 106
+    },
+    {
+      "epoch": 0.10854271356783919,
+      "grad_norm": 2.3541088104248047,
+      "learning_rate": 9.999557931347273e-05,
+      "loss": 2.8328,
+      "step": 108
+    },
+    {
+      "epoch": 0.11055276381909548,
+      "grad_norm": 2.2207984924316406,
+      "learning_rate": 9.999309273455528e-05,
+      "loss": 3.0153,
+      "step": 110
+    },
+    {
+      "epoch": 0.11256281407035176,
+      "grad_norm": 2.1303670406341553,
+      "learning_rate": 9.999005363852618e-05,
+      "loss": 2.8649,
+      "step": 112
+    },
+    {
+      "epoch": 0.11457286432160804,
+      "grad_norm": 2.084247589111328,
+      "learning_rate": 9.998646205897309e-05,
+      "loss": 2.9326,
+      "step": 114
+    },
+    {
+      "epoch": 0.11658291457286432,
+      "grad_norm": 2.121018171310425,
+      "learning_rate": 9.998231803558968e-05,
+      "loss": 2.6179,
+      "step": 116
+    },
+    {
+      "epoch": 0.1185929648241206,
+      "grad_norm": 2.2490546703338623,
+      "learning_rate": 9.997762161417517e-05,
+      "loss": 3.0462,
+      "step": 118
+    },
+    {
+      "epoch": 0.12060301507537688,
+      "grad_norm": 1.9550148248672485,
+      "learning_rate": 9.997237284663379e-05,
+      "loss": 2.8662,
+      "step": 120
+    },
+    {
+      "epoch": 0.12261306532663317,
+      "grad_norm": 2.2603437900543213,
+      "learning_rate": 9.996657179097421e-05,
+      "loss": 2.7448,
+      "step": 122
+    },
+    {
+      "epoch": 0.12462311557788945,
+      "grad_norm": 1.897772192955017,
+      "learning_rate": 9.996021851130897e-05,
+      "loss": 2.8954,
+      "step": 124
+    },
+    {
+      "epoch": 0.12663316582914572,
+      "grad_norm": 2.1712026596069336,
+      "learning_rate": 9.995331307785365e-05,
+      "loss": 2.9001,
+      "step": 126
+    },
+    {
+      "epoch": 0.12864321608040202,
+      "grad_norm": 1.9912605285644531,
+      "learning_rate": 9.994585556692624e-05,
+      "loss": 3.0064,
+      "step": 128
+    },
+    {
+      "epoch": 0.1306532663316583,
+      "grad_norm": 2.094640016555786,
+      "learning_rate": 9.993784606094612e-05,
+      "loss": 2.6635,
+      "step": 130
+    },
+    {
+      "epoch": 0.13266331658291458,
+      "grad_norm": 2.1064651012420654,
+      "learning_rate": 9.992928464843334e-05,
+      "loss": 2.9224,
+      "step": 132
+    },
+    {
+      "epoch": 0.13467336683417086,
+      "grad_norm": 1.9782873392105103,
+      "learning_rate": 9.992017142400751e-05,
+      "loss": 2.8202,
+      "step": 134
+    },
+    {
+      "epoch": 0.13668341708542714,
+      "grad_norm": 2.189976692199707,
+      "learning_rate": 9.991050648838675e-05,
+      "loss": 2.7919,
+      "step": 136
+    },
+    {
+      "epoch": 0.13869346733668342,
+      "grad_norm": 2.0882322788238525,
+      "learning_rate": 9.990028994838673e-05,
+      "loss": 2.868,
+      "step": 138
+    },
+    {
+      "epoch": 0.1407035175879397,
+      "grad_norm": 1.8483355045318604,
+      "learning_rate": 9.988952191691925e-05,
+      "loss": 2.903,
+      "step": 140
+    },
+    {
+      "epoch": 0.14271356783919598,
+      "grad_norm": 2.1383628845214844,
+      "learning_rate": 9.987820251299122e-05,
+      "loss": 2.6627,
+      "step": 142
+    },
+    {
+      "epoch": 0.14472361809045226,
+      "grad_norm": 2.0356757640838623,
+      "learning_rate": 9.986633186170319e-05,
+      "loss": 2.813,
+      "step": 144
+    },
+    {
+      "epoch": 0.14673366834170853,
+      "grad_norm": 2.4072749614715576,
+      "learning_rate": 9.985391009424805e-05,
+      "loss": 2.7325,
+      "step": 146
+    },
+    {
+      "epoch": 0.1487437185929648,
+      "grad_norm": 1.8525023460388184,
+      "learning_rate": 9.984093734790956e-05,
+      "loss": 2.7092,
+      "step": 148
+    },
+    {
+      "epoch": 0.1507537688442211,
+      "grad_norm": 1.8999531269073486,
+      "learning_rate": 9.982741376606078e-05,
+      "loss": 2.8637,
+      "step": 150
+    },
+    {
+      "epoch": 0.1527638190954774,
+      "grad_norm": 2.2264699935913086,
+      "learning_rate": 9.981333949816259e-05,
+      "loss": 2.8068,
+      "step": 152
+    },
+    {
+      "epoch": 0.15477386934673368,
+      "grad_norm": 2.1256349086761475,
+      "learning_rate": 9.979871469976196e-05,
+      "loss": 2.5258,
+      "step": 154
+    },
+    {
+      "epoch": 0.15678391959798996,
+      "grad_norm": 1.8756039142608643,
+      "learning_rate": 9.978353953249022e-05,
+      "loss": 2.7572,
+      "step": 156
+    },
+    {
+      "epoch": 0.15879396984924624,
+      "grad_norm": 2.4083592891693115,
+      "learning_rate": 9.976781416406136e-05,
+      "loss": 2.564,
+      "step": 158
+    },
+    {
+      "epoch": 0.16080402010050251,
+      "grad_norm": 1.9118754863739014,
+      "learning_rate": 9.975153876827008e-05,
+      "loss": 2.7158,
+      "step": 160
+    },
+    {
+      "epoch": 0.1628140703517588,
+      "grad_norm": 2.419102907180786,
+      "learning_rate": 9.973471352498991e-05,
+      "loss": 2.9075,
+      "step": 162
+    },
+    {
+      "epoch": 0.16482412060301507,
+      "grad_norm": 2.2566428184509277,
+      "learning_rate": 9.971733862017126e-05,
+      "loss": 2.7698,
+      "step": 164
+    },
+    {
+      "epoch": 0.16683417085427135,
+      "grad_norm": 1.980230450630188,
+      "learning_rate": 9.969941424583926e-05,
+      "loss": 2.6454,
+      "step": 166
+    },
+    {
+      "epoch": 0.16884422110552763,
+      "grad_norm": 2.0920395851135254,
+      "learning_rate": 9.96809406000918e-05,
+      "loss": 2.8594,
+      "step": 168
+    },
+    {
+      "epoch": 0.1708542713567839,
+      "grad_norm": 1.8870232105255127,
+      "learning_rate": 9.966191788709716e-05,
+      "loss": 2.5967,
+      "step": 170
+    },
+    {
+      "epoch": 0.1728643216080402,
+      "grad_norm": 2.334998607635498,
+      "learning_rate": 9.964234631709187e-05,
+      "loss": 2.661,
+      "step": 172
+    },
+    {
+      "epoch": 0.1748743718592965,
+      "grad_norm": 1.9071625471115112,
+      "learning_rate": 9.962222610637837e-05,
+      "loss": 2.6161,
+      "step": 174
+    },
+    {
+      "epoch": 0.17688442211055277,
+      "grad_norm": 2.015760898590088,
+      "learning_rate": 9.960155747732259e-05,
+      "loss": 2.5681,
+      "step": 176
+    },
+    {
+      "epoch": 0.17889447236180905,
+      "grad_norm": 1.827253818511963,
+      "learning_rate": 9.958034065835151e-05,
+      "loss": 2.5538,
+      "step": 178
+    },
+    {
+      "epoch": 0.18090452261306533,
+      "grad_norm": 1.9388582706451416,
+      "learning_rate": 9.955857588395065e-05,
+      "loss": 2.7098,
+      "step": 180
+    },
+    {
+      "epoch": 0.1829145728643216,
+      "grad_norm": 2.1864373683929443,
+      "learning_rate": 9.95362633946614e-05,
+      "loss": 2.5615,
+      "step": 182
+    },
+    {
+      "epoch": 0.1849246231155779,
+      "grad_norm": 2.093961477279663,
+      "learning_rate": 9.951340343707852e-05,
+      "loss": 2.7367,
+      "step": 184
+    },
+    {
+      "epoch": 0.18693467336683417,
+      "grad_norm": 1.9437264204025269,
+      "learning_rate": 9.948999626384724e-05,
+      "loss": 2.7414,
+      "step": 186
+    },
+    {
+      "epoch": 0.18894472361809045,
+      "grad_norm": 1.8001607656478882,
+      "learning_rate": 9.946604213366057e-05,
+      "loss": 2.7595,
+      "step": 188
+    },
+    {
+      "epoch": 0.19095477386934673,
+      "grad_norm": 1.8736133575439453,
+      "learning_rate": 9.944154131125642e-05,
+      "loss": 2.2943,
+      "step": 190
+    },
+    {
+      "epoch": 0.192964824120603,
+      "grad_norm": 1.8067212104797363,
+      "learning_rate": 9.941649406741469e-05,
+      "loss": 2.6118,
+      "step": 192
+    },
+    {
+      "epoch": 0.19497487437185929,
+      "grad_norm": 1.9442073106765747,
+      "learning_rate": 9.939090067895422e-05,
+      "loss": 2.5651,
+      "step": 194
+    },
+    {
+      "epoch": 0.19698492462311556,
+      "grad_norm": 1.976509690284729,
+      "learning_rate": 9.936476142872979e-05,
+      "loss": 2.5084,
+      "step": 196
+    },
+    {
+      "epoch": 0.19899497487437187,
+      "grad_norm": 1.9604930877685547,
+      "learning_rate": 9.933807660562898e-05,
+      "loss": 2.3689,
+      "step": 198
+    },
+    {
+      "epoch": 0.20100502512562815,
+      "grad_norm": 2.081547975540161,
+      "learning_rate": 9.931084650456892e-05,
+      "loss": 2.7038,
+      "step": 200
+    },
+    {
+      "epoch": 0.20301507537688443,
+      "grad_norm": 2.1766066551208496,
+      "learning_rate": 9.928307142649316e-05,
+      "loss": 2.8203,
+      "step": 202
+    },
+    {
+      "epoch": 0.2050251256281407,
+      "grad_norm": 2.217263698577881,
+      "learning_rate": 9.925475167836821e-05,
+      "loss": 2.5522,
+      "step": 204
+    },
+    {
+      "epoch": 0.207035175879397,
+      "grad_norm": 1.8877842426300049,
+      "learning_rate": 9.922588757318021e-05,
+      "loss": 2.6209,
+      "step": 206
+    },
+    {
+      "epoch": 0.20904522613065327,
+      "grad_norm": 1.8535610437393188,
+      "learning_rate": 9.919647942993148e-05,
+      "loss": 2.7805,
+      "step": 208
+    },
+    {
+      "epoch": 0.21105527638190955,
+      "grad_norm": 1.9754009246826172,
+      "learning_rate": 9.916652757363698e-05,
+      "loss": 2.7363,
+      "step": 210
+    },
+    {
+      "epoch": 0.21306532663316582,
+      "grad_norm": 2.0575201511383057,
+      "learning_rate": 9.913603233532067e-05,
+      "loss": 2.5019,
+      "step": 212
+    },
+    {
+      "epoch": 0.2150753768844221,
+      "grad_norm": 1.8286898136138916,
+      "learning_rate": 9.910499405201195e-05,
+      "loss": 2.5019,
+      "step": 214
+    },
+    {
+      "epoch": 0.21708542713567838,
+      "grad_norm": 1.954950213432312,
+      "learning_rate": 9.907341306674185e-05,
+      "loss": 2.5339,
+      "step": 216
+    },
+    {
+      "epoch": 0.21909547738693466,
+      "grad_norm": 1.8942043781280518,
+      "learning_rate": 9.90412897285393e-05,
+      "loss": 2.5065,
+      "step": 218
+    },
+    {
+      "epoch": 0.22110552763819097,
+      "grad_norm": 2.165287494659424,
+      "learning_rate": 9.900862439242719e-05,
+      "loss": 2.596,
+      "step": 220
+    },
+    {
+      "epoch": 0.22311557788944725,
+      "grad_norm": 1.8702832460403442,
+      "learning_rate": 9.897541741941858e-05,
+      "loss": 2.6656,
+      "step": 222
+    },
+    {
+      "epoch": 0.22512562814070353,
+      "grad_norm": 1.745975375175476,
+      "learning_rate": 9.894166917651256e-05,
+      "loss": 2.4078,
+      "step": 224
+    },
+    {
+      "epoch": 0.2271356783919598,
+      "grad_norm": 9.129441261291504,
+      "learning_rate": 9.890738003669029e-05,
+      "loss": 2.4112,
+      "step": 226
+    },
+    {
+      "epoch": 0.22914572864321608,
+      "grad_norm": 2.1922168731689453,
+      "learning_rate": 9.887255037891086e-05,
+      "loss": 2.5331,
+      "step": 228
+    },
+    {
+      "epoch": 0.23115577889447236,
+      "grad_norm": 1.9028476476669312,
+      "learning_rate": 9.883718058810707e-05,
+      "loss": 2.4778,
+      "step": 230
+    },
+    {
+      "epoch": 0.23316582914572864,
+      "grad_norm": 1.9489747285842896,
+      "learning_rate": 9.880127105518122e-05,
+      "loss": 2.6777,
+      "step": 232
+    },
+    {
+      "epoch": 0.23517587939698492,
+      "grad_norm": 1.922662377357483,
+      "learning_rate": 9.876482217700078e-05,
+      "loss": 2.5236,
+      "step": 234
+    },
+    {
+      "epoch": 0.2371859296482412,
+      "grad_norm": 1.9473645687103271,
+      "learning_rate": 9.872783435639397e-05,
+      "loss": 2.4375,
+      "step": 236
+    },
+    {
+      "epoch": 0.23919597989949748,
+      "grad_norm": 1.835803747177124,
+      "learning_rate": 9.869030800214532e-05,
+      "loss": 2.6191,
+      "step": 238
+    },
+    {
+      "epoch": 0.24120603015075376,
+      "grad_norm": 2.013805627822876,
+      "learning_rate": 9.865224352899119e-05,
+      "loss": 2.5062,
+      "step": 240
+    },
+    {
+      "epoch": 0.24321608040201004,
+      "grad_norm": 1.8618221282958984,
+      "learning_rate": 9.861364135761517e-05,
+      "loss": 2.3637,
+      "step": 242
+    },
+    {
+      "epoch": 0.24522613065326634,
+      "grad_norm": 2.0696043968200684,
+      "learning_rate": 9.857450191464337e-05,
+      "loss": 2.3941,
+      "step": 244
+    },
+    {
+      "epoch": 0.24723618090452262,
+      "grad_norm": 1.835046410560608,
+      "learning_rate": 9.853482563263981e-05,
+      "loss": 2.9119,
+      "step": 246
+    },
+    {
+      "epoch": 0.2492462311557789,
+      "grad_norm": 1.9712692499160767,
+      "learning_rate": 9.849461295010156e-05,
+      "loss": 2.4119,
+      "step": 248
+    },
+    {
+      "epoch": 0.25125628140703515,
+      "grad_norm": 2.0356907844543457,
+      "learning_rate": 9.84538643114539e-05,
+      "loss": 2.4488,
+      "step": 250
+    },
+    {
+      "epoch": 0.25326633165829143,
+      "grad_norm": 1.983216643333435,
+      "learning_rate": 9.841258016704546e-05,
+      "loss": 2.5618,
+      "step": 252
+    },
+    {
+      "epoch": 0.2552763819095477,
+      "grad_norm": 1.8130682706832886,
+      "learning_rate": 9.837076097314319e-05,
+      "loss": 2.6361,
+      "step": 254
+    },
+    {
+      "epoch": 0.25728643216080405,
+      "grad_norm": 1.7342140674591064,
+      "learning_rate": 9.832840719192736e-05,
+      "loss": 2.3223,
+      "step": 256
+    },
+    {
+      "epoch": 0.2592964824120603,
+      "grad_norm": 1.9762102365493774,
+      "learning_rate": 9.82855192914864e-05,
+      "loss": 2.4629,
+      "step": 258
+    },
+    {
+      "epoch": 0.2613065326633166,
+      "grad_norm": 1.8401470184326172,
+      "learning_rate": 9.824209774581174e-05,
+      "loss": 2.5557,
+      "step": 260
+    },
+    {
+      "epoch": 0.2633165829145729,
+      "grad_norm": 1.8549468517303467,
+      "learning_rate": 9.819814303479267e-05,
+      "loss": 2.6337,
+      "step": 262
+    },
+    {
+      "epoch": 0.26532663316582916,
+      "grad_norm": 2.0843794345855713,
+      "learning_rate": 9.815365564421085e-05,
+      "loss": 2.3528,
+      "step": 264
+    },
+    {
+      "epoch": 0.26733668341708544,
+      "grad_norm": 1.893712043762207,
+      "learning_rate": 9.810863606573513e-05,
+      "loss": 2.5509,
+      "step": 266
+    },
+    {
+      "epoch": 0.2693467336683417,
+      "grad_norm": 2.0922601222991943,
+      "learning_rate": 9.806308479691595e-05,
+      "loss": 2.6126,
+      "step": 268
+    },
+    {
+      "epoch": 0.271356783919598,
+      "grad_norm": 1.8790444135665894,
+      "learning_rate": 9.801700234117999e-05,
+      "loss": 2.5569,
+      "step": 270
+    },
+    {
+      "epoch": 0.2733668341708543,
+      "grad_norm": 1.8532637357711792,
+      "learning_rate": 9.797038920782454e-05,
+      "loss": 2.4948,
+      "step": 272
+    },
+    {
+      "epoch": 0.27537688442211056,
+      "grad_norm": 1.8589162826538086,
+      "learning_rate": 9.792324591201179e-05,
+      "loss": 2.4739,
+      "step": 274
+    },
+    {
+      "epoch": 0.27738693467336684,
+      "grad_norm": 1.8177707195281982,
+      "learning_rate": 9.78755729747633e-05,
+      "loss": 2.4883,
+      "step": 276
+    },
+    {
+      "epoch": 0.2793969849246231,
+      "grad_norm": 1.9046951532363892,
+      "learning_rate": 9.782737092295413e-05,
+      "loss": 2.331,
+      "step": 278
+    },
+    {
+      "epoch": 0.2814070351758794,
+      "grad_norm": 1.8969510793685913,
+      "learning_rate": 9.777864028930705e-05,
+      "loss": 2.4818,
+      "step": 280
+    },
+    {
+      "epoch": 0.2834170854271357,
+      "grad_norm": 2.1642754077911377,
+      "learning_rate": 9.77293816123866e-05,
+      "loss": 2.3212,
+      "step": 282
+    },
+    {
+      "epoch": 0.28542713567839195,
+      "grad_norm": 1.7480413913726807,
+      "learning_rate": 9.767959543659326e-05,
+      "loss": 2.5252,
+      "step": 284
+    },
+    {
+      "epoch": 0.28743718592964823,
+      "grad_norm": 2.0396628379821777,
+      "learning_rate": 9.76292823121573e-05,
+      "loss": 2.6501,
+      "step": 286
+    },
+    {
+      "epoch": 0.2894472361809045,
+      "grad_norm": 1.5687803030014038,
+      "learning_rate": 9.757844279513281e-05,
+      "loss": 2.3029,
+      "step": 288
+    },
+    {
+      "epoch": 0.2914572864321608,
+      "grad_norm": 1.9262704849243164,
+      "learning_rate": 9.752707744739145e-05,
+      "loss": 2.4645,
+      "step": 290
+    },
+    {
+      "epoch": 0.29346733668341707,
+      "grad_norm": 1.9408085346221924,
+      "learning_rate": 9.747518683661631e-05,
+      "loss": 2.3497,
+      "step": 292
+    },
+    {
+      "epoch": 0.29547738693467335,
+      "grad_norm": 1.78432297706604,
+      "learning_rate": 9.742277153629564e-05,
+      "loss": 2.2824,
+      "step": 294
+    },
+    {
+      "epoch": 0.2974874371859296,
+      "grad_norm": 2.070885181427002,
+      "learning_rate": 9.736983212571646e-05,
+      "loss": 2.5931,
+      "step": 296
+    },
+    {
+      "epoch": 0.2994974874371859,
+      "grad_norm": 1.657235026359558,
+      "learning_rate": 9.731636918995821e-05,
+      "loss": 2.4344,
+      "step": 298
+    },
+    {
+      "epoch": 0.3015075376884422,
+      "grad_norm": 1.8496333360671997,
+      "learning_rate": 9.726238331988624e-05,
+      "loss": 2.4043,
+      "step": 300
+    },
+    {
+      "epoch": 0.3035175879396985,
+      "grad_norm": 1.8144067525863647,
+      "learning_rate": 9.720787511214533e-05,
+      "loss": 2.5708,
+      "step": 302
+    },
+    {
+      "epoch": 0.3055276381909548,
+      "grad_norm": 1.875706434249878,
+      "learning_rate": 9.715284516915303e-05,
+      "loss": 2.387,
+      "step": 304
+    },
+    {
+      "epoch": 0.3075376884422111,
+      "grad_norm": 1.905147910118103,
+      "learning_rate": 9.709729409909307e-05,
+      "loss": 2.6459,
+      "step": 306
+    },
+    {
+      "epoch": 0.30954773869346736,
+      "grad_norm": 1.919798493385315,
+      "learning_rate": 9.704122251590862e-05,
+      "loss": 2.2947,
+      "step": 308
+    },
+    {
+      "epoch": 0.31155778894472363,
+      "grad_norm": 1.5951811075210571,
+      "learning_rate": 9.698463103929542e-05,
+      "loss": 2.3532,
+      "step": 310
+    },
+    {
+      "epoch": 0.3135678391959799,
+      "grad_norm": 1.6839174032211304,
+      "learning_rate": 9.692752029469512e-05,
+      "loss": 2.2152,
+      "step": 312
+    },
+    {
+      "epoch": 0.3155778894472362,
+      "grad_norm": 2.0012948513031006,
+      "learning_rate": 9.686989091328813e-05,
+      "loss": 2.3561,
+      "step": 314
+    },
+    {
+      "epoch": 0.31758793969849247,
+      "grad_norm": 1.8996232748031616,
+      "learning_rate": 9.681174353198687e-05,
+      "loss": 2.3638,
+      "step": 316
+    },
+    {
+      "epoch": 0.31959798994974875,
+      "grad_norm": 1.8173401355743408,
+      "learning_rate": 9.675307879342854e-05,
+      "loss": 2.2213,
+      "step": 318
+    },
+    {
+      "epoch": 0.32160804020100503,
+      "grad_norm": 1.6612550020217896,
+      "learning_rate": 9.669389734596819e-05,
+      "loss": 2.5352,
+      "step": 320
+    },
+    {
+      "epoch": 0.3236180904522613,
+      "grad_norm": 1.8817943334579468,
+      "learning_rate": 9.663419984367139e-05,
+      "loss": 2.5058,
+      "step": 322
+    },
+    {
+      "epoch": 0.3256281407035176,
+      "grad_norm": 2.1160693168640137,
+      "learning_rate": 9.657398694630712e-05,
+      "loss": 2.2973,
+      "step": 324
+    },
+    {
+      "epoch": 0.32763819095477387,
+      "grad_norm": 1.738549828529358,
+      "learning_rate": 9.651325931934046e-05,
+      "loss": 2.2865,
+      "step": 326
+    },
+    {
+      "epoch": 0.32964824120603015,
+      "grad_norm": 1.8293092250823975,
+      "learning_rate": 9.645201763392513e-05,
+      "loss": 2.1597,
+      "step": 328
+    },
+    {
+      "epoch": 0.3316582914572864,
+      "grad_norm": 1.9268925189971924,
+      "learning_rate": 9.639026256689628e-05,
+      "loss": 2.5048,
+      "step": 330
+    },
+    {
+      "epoch": 0.3336683417085427,
+      "grad_norm": 1.983359932899475,
+      "learning_rate": 9.632799480076278e-05,
+      "loss": 2.3261,
+      "step": 332
+    },
+    {
+      "epoch": 0.335678391959799,
+      "grad_norm": 1.717955470085144,
+      "learning_rate": 9.626521502369984e-05,
+      "loss": 2.6444,
+      "step": 334
+    },
+    {
+      "epoch": 0.33768844221105526,
+      "grad_norm": 1.836305856704712,
+      "learning_rate": 9.620192392954132e-05,
+      "loss": 2.4493,
+      "step": 336
+    },
+    {
+      "epoch": 0.33969849246231154,
+      "grad_norm": 1.8819546699523926,
+      "learning_rate": 9.613812221777212e-05,
+      "loss": 2.6608,
+      "step": 338
+    },
+    {
+      "epoch": 0.3417085427135678,
+      "grad_norm": 1.845680832862854,
+      "learning_rate": 9.607381059352038e-05,
+      "loss": 2.2059,
+      "step": 340
+    },
+    {
+      "epoch": 0.3437185929648241,
+      "grad_norm": 1.878606915473938,
+      "learning_rate": 9.600898976754977e-05,
+      "loss": 2.2733,
+      "step": 342
+    },
+    {
+      "epoch": 0.3457286432160804,
+      "grad_norm": 1.768790364265442,
+      "learning_rate": 9.594366045625154e-05,
+      "loss": 2.3773,
+      "step": 344
+    },
+    {
+      "epoch": 0.34773869346733666,
+      "grad_norm": 1.761987328529358,
+      "learning_rate": 9.587782338163669e-05,
+      "loss": 2.5148,
+      "step": 346
+    },
+    {
+      "epoch": 0.349748743718593,
+      "grad_norm": 1.6543912887573242,
+      "learning_rate": 9.581147927132797e-05,
+      "loss": 2.4609,
+      "step": 348
+    },
+    {
+      "epoch": 0.35175879396984927,
+      "grad_norm": 1.9013664722442627,
+      "learning_rate": 9.574462885855174e-05,
+      "loss": 2.121,
+      "step": 350
+    },
+    {
+      "epoch": 0.35376884422110555,
+      "grad_norm": 3.302755832672119,
+      "learning_rate": 9.567727288213005e-05,
+      "loss": 2.4353,
+      "step": 352
+    },
+    {
+      "epoch": 0.35577889447236183,
+      "grad_norm": 2.773200750350952,
+      "learning_rate": 9.560941208647231e-05,
+      "loss": 2.3111,
+      "step": 354
+    },
+    {
+      "epoch": 0.3577889447236181,
+      "grad_norm": 1.7840498685836792,
+      "learning_rate": 9.554104722156716e-05,
+      "loss": 2.5435,
+      "step": 356
+    },
+    {
+      "epoch": 0.3597989949748744,
+      "grad_norm": 1.8287582397460938,
+      "learning_rate": 9.547217904297411e-05,
+      "loss": 2.3461,
+      "step": 358
+    },
+    {
+      "epoch": 0.36180904522613067,
+      "grad_norm": 1.8139948844909668,
+      "learning_rate": 9.540280831181525e-05,
+      "loss": 2.3423,
+      "step": 360
+    },
+    {
+      "epoch": 0.36381909547738694,
+      "grad_norm": 1.8012893199920654,
+      "learning_rate": 9.533293579476683e-05,
+      "loss": 2.412,
+      "step": 362
+    },
+    {
+      "epoch": 0.3658291457286432,
+      "grad_norm": 1.8645695447921753,
+      "learning_rate": 9.526256226405075e-05,
+      "loss": 2.5261,
+      "step": 364
+    },
+    {
+      "epoch": 0.3678391959798995,
+      "grad_norm": 1.9607142210006714,
+      "learning_rate": 9.519168849742604e-05,
+      "loss": 2.6721,
+      "step": 366
+    },
+    {
+      "epoch": 0.3698492462311558,
+      "grad_norm": 1.772139549255371,
+      "learning_rate": 9.512031527818028e-05,
+      "loss": 2.3517,
+      "step": 368
+    },
+    {
+      "epoch": 0.37185929648241206,
+      "grad_norm": 1.8913545608520508,
+      "learning_rate": 9.504844339512095e-05,
+      "loss": 2.6095,
+      "step": 370
+    },
+    {
+      "epoch": 0.37386934673366834,
+      "grad_norm": 1.8451741933822632,
+      "learning_rate": 9.497607364256672e-05,
+      "loss": 2.5707,
+      "step": 372
+    },
+    {
+      "epoch": 0.3758793969849246,
+      "grad_norm": 1.8256680965423584,
+      "learning_rate": 9.490320682033855e-05,
+      "loss": 2.3579,
+      "step": 374
+    },
+    {
+      "epoch": 0.3778894472361809,
+      "grad_norm": 1.9581091403961182,
+      "learning_rate": 9.482984373375105e-05,
+      "loss": 2.5629,
+      "step": 376
+    },
+    {
+      "epoch": 0.3798994974874372,
+      "grad_norm": 1.998180627822876,
+      "learning_rate": 9.475598519360344e-05,
+      "loss": 2.3817,
+      "step": 378
+    },
+    {
+      "epoch": 0.38190954773869346,
+      "grad_norm": 1.7146025896072388,
+      "learning_rate": 9.468163201617062e-05,
+      "loss": 2.2208,
+      "step": 380
+    },
+    {
+      "epoch": 0.38391959798994973,
+      "grad_norm": 1.8821600675582886,
+      "learning_rate": 9.460678502319418e-05,
+      "loss": 2.4416,
+      "step": 382
+    },
+    {
+      "epoch": 0.385929648241206,
+      "grad_norm": 1.7585586309432983,
+      "learning_rate": 9.453144504187327e-05,
+      "loss": 2.3008,
+      "step": 384
+    },
+    {
+      "epoch": 0.3879396984924623,
+      "grad_norm": 1.6485331058502197,
+      "learning_rate": 9.445561290485549e-05,
+      "loss": 2.383,
+      "step": 386
+    },
+    {
+      "epoch": 0.38994974874371857,
+      "grad_norm": 1.9249005317687988,
+      "learning_rate": 9.437928945022771e-05,
+      "loss": 2.4204,
+      "step": 388
+    },
+    {
+      "epoch": 0.39195979899497485,
+      "grad_norm": 1.8952553272247314,
+      "learning_rate": 9.430247552150673e-05,
+      "loss": 2.2751,
+      "step": 390
+    },
+    {
+      "epoch": 0.39396984924623113,
+      "grad_norm": 2.2544572353363037,
+      "learning_rate": 9.422517196763002e-05,
+      "loss": 2.4784,
+      "step": 392
+    },
+    {
+      "epoch": 0.39597989949748746,
+      "grad_norm": 2.308283567428589,
+      "learning_rate": 9.414737964294636e-05,
+      "loss": 2.2052,
+      "step": 394
+    },
+    {
+      "epoch": 0.39798994974874374,
+      "grad_norm": 1.8085863590240479,
+      "learning_rate": 9.40690994072063e-05,
+      "loss": 2.2588,
+      "step": 396
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9807807207107544,
+      "learning_rate": 9.399033212555275e-05,
+      "loss": 2.2698,
+      "step": 398
+    },
+    {
+      "epoch": 0.4020100502512563,
+      "grad_norm": 1.6799061298370361,
+      "learning_rate": 9.391107866851143e-05,
+      "loss": 2.2507,
+      "step": 400
+    },
+    {
+      "epoch": 0.4040201005025126,
+      "grad_norm": 1.6821558475494385,
+      "learning_rate": 9.383133991198112e-05,
+      "loss": 2.386,
+      "step": 402
+    },
+    {
+      "epoch": 0.40603015075376886,
+      "grad_norm": 1.7488336563110352,
+      "learning_rate": 9.375111673722414e-05,
+      "loss": 2.7501,
+      "step": 404
+    },
+    {
+      "epoch": 0.40804020100502514,
+      "grad_norm": 1.7495909929275513,
+      "learning_rate": 9.367041003085649e-05,
+      "loss": 2.4578,
+      "step": 406
+    },
+    {
+      "epoch": 0.4100502512562814,
+      "grad_norm": 1.773022174835205,
+      "learning_rate": 9.358922068483812e-05,
+      "loss": 2.1896,
+      "step": 408
+    },
+    {
+      "epoch": 0.4120603015075377,
+      "grad_norm": 1.6720446348190308,
+      "learning_rate": 9.350754959646306e-05,
+      "loss": 2.2264,
+      "step": 410
+    },
+    {
+      "epoch": 0.414070351758794,
+      "grad_norm": 1.8488444089889526,
+      "learning_rate": 9.342539766834946e-05,
+      "loss": 2.5527,
+      "step": 412
+    },
+    {
+      "epoch": 0.41608040201005025,
+      "grad_norm": 1.7335339784622192,
+      "learning_rate": 9.334276580842967e-05,
+      "loss": 2.2894,
+      "step": 414
+    },
+    {
+      "epoch": 0.41809045226130653,
+      "grad_norm": 1.670088529586792,
+      "learning_rate": 9.325965492994018e-05,
+      "loss": 2.1557,
+      "step": 416
+    },
+    {
+      "epoch": 0.4201005025125628,
+      "grad_norm": 1.6430253982543945,
+      "learning_rate": 9.317606595141154e-05,
+      "loss": 2.2405,
+      "step": 418
+    },
+    {
+      "epoch": 0.4221105527638191,
+      "grad_norm": 1.7390164136886597,
+      "learning_rate": 9.30919997966582e-05,
+      "loss": 2.3486,
+      "step": 420
+    },
+    {
+      "epoch": 0.42412060301507537,
+      "grad_norm": 1.7880635261535645,
+      "learning_rate": 9.300745739476829e-05,
+      "loss": 2.2626,
+      "step": 422
+    },
+    {
+      "epoch": 0.42613065326633165,
+      "grad_norm": 1.7333531379699707,
+      "learning_rate": 9.292243968009331e-05,
+      "loss": 2.2424,
+      "step": 424
+    },
+    {
+      "epoch": 0.42814070351758793,
+      "grad_norm": 1.6569937467575073,
+      "learning_rate": 9.283694759223796e-05,
+      "loss": 2.1534,
+      "step": 426
+    },
+    {
+      "epoch": 0.4301507537688442,
+      "grad_norm": 1.7097984552383423,
+      "learning_rate": 9.275098207604957e-05,
+      "loss": 2.2569,
+      "step": 428
+    },
+    {
+      "epoch": 0.4321608040201005,
+      "grad_norm": 1.7049428224563599,
+      "learning_rate": 9.266454408160779e-05,
+      "loss": 2.4531,
+      "step": 430
+    },
+    {
+      "epoch": 0.43417085427135677,
+      "grad_norm": 1.5478094816207886,
+      "learning_rate": 9.257763456421398e-05,
+      "loss": 2.2312,
+      "step": 432
+    },
+    {
+      "epoch": 0.43618090452261304,
+      "grad_norm": 1.706602692604065,
+      "learning_rate": 9.249025448438076e-05,
+      "loss": 2.4432,
+      "step": 434
+    },
+    {
+      "epoch": 0.4381909547738693,
+      "grad_norm": 1.701011300086975,
+      "learning_rate": 9.24024048078213e-05,
+      "loss": 2.1473,
+      "step": 436
+    },
+    {
+      "epoch": 0.4402010050251256,
+      "grad_norm": 1.7655497789382935,
+      "learning_rate": 9.231408650543874e-05,
+      "loss": 2.433,
+      "step": 438
+    },
+    {
+      "epoch": 0.44221105527638194,
+      "grad_norm": 1.853971004486084,
+      "learning_rate": 9.22253005533154e-05,
+      "loss": 2.3029,
+      "step": 440
+    },
+    {
+      "epoch": 0.4442211055276382,
+      "grad_norm": 1.8369559049606323,
+      "learning_rate": 9.213604793270196e-05,
+      "loss": 2.1837,
+      "step": 442
+    },
+    {
+      "epoch": 0.4462311557788945,
+      "grad_norm": 1.6373860836029053,
+      "learning_rate": 9.204632963000671e-05,
+      "loss": 2.4237,
+      "step": 444
+    },
+    {
+      "epoch": 0.4482412060301508,
+      "grad_norm": 1.7693603038787842,
+      "learning_rate": 9.195614663678458e-05,
+      "loss": 2.4229,
+      "step": 446
+    },
+    {
+      "epoch": 0.45025125628140705,
+      "grad_norm": 1.6722673177719116,
+      "learning_rate": 9.186549994972618e-05,
+      "loss": 2.3788,
+      "step": 448
+    },
+    {
+      "epoch": 0.45226130653266333,
+      "grad_norm": 1.6546918153762817,
+      "learning_rate": 9.177439057064683e-05,
+      "loss": 2.2079,
+      "step": 450
+    },
+    {
+      "epoch": 0.4542713567839196,
+      "grad_norm": 1.8333224058151245,
+      "learning_rate": 9.168281950647545e-05,
+      "loss": 2.5018,
+      "step": 452
+    },
+    {
+      "epoch": 0.4562814070351759,
+      "grad_norm": 1.7532380819320679,
+      "learning_rate": 9.159078776924346e-05,
+      "loss": 2.0344,
+      "step": 454
+    },
+    {
+      "epoch": 0.45829145728643217,
+      "grad_norm": 1.647209644317627,
+      "learning_rate": 9.149829637607353e-05,
+      "loss": 2.2324,
+      "step": 456
+    },
+    {
+      "epoch": 0.46030150753768845,
+      "grad_norm": 1.6968568563461304,
+      "learning_rate": 9.140534634916846e-05,
+      "loss": 2.1635,
+      "step": 458
+    },
+    {
+      "epoch": 0.4623115577889447,
+      "grad_norm": 1.6668813228607178,
+      "learning_rate": 9.131193871579975e-05,
+      "loss": 2.334,
+      "step": 460
+    },
+    {
+      "epoch": 0.464321608040201,
+      "grad_norm": 1.6174511909484863,
+      "learning_rate": 9.121807450829632e-05,
+      "loss": 2.1475,
+      "step": 462
+    },
+    {
+      "epoch": 0.4663316582914573,
+      "grad_norm": 1.8468537330627441,
+      "learning_rate": 9.112375476403312e-05,
+      "loss": 2.2984,
+      "step": 464
+    },
+    {
+      "epoch": 0.46834170854271356,
+      "grad_norm": 1.7187474966049194,
+      "learning_rate": 9.102898052541958e-05,
+      "loss": 2.4139,
+      "step": 466
+    },
+    {
+      "epoch": 0.47035175879396984,
+      "grad_norm": 1.7762575149536133,
+      "learning_rate": 9.093375283988819e-05,
+      "loss": 2.1496,
+      "step": 468
+    },
+    {
+      "epoch": 0.4723618090452261,
+      "grad_norm": 1.9211578369140625,
+      "learning_rate": 9.083807275988284e-05,
+      "loss": 2.1667,
+      "step": 470
+    },
+    {
+      "epoch": 0.4743718592964824,
+      "grad_norm": 3.2748539447784424,
+      "learning_rate": 9.074194134284726e-05,
+      "loss": 2.2991,
+      "step": 472
+    },
+    {
+      "epoch": 0.4763819095477387,
+      "grad_norm": 1.751917839050293,
+      "learning_rate": 9.064535965121324e-05,
+      "loss": 2.2903,
+      "step": 474
+    },
+    {
+      "epoch": 0.47839195979899496,
+      "grad_norm": 1.7564281225204468,
+      "learning_rate": 9.054832875238903e-05,
+      "loss": 2.5954,
+      "step": 476
+    },
+    {
+      "epoch": 0.48040201005025124,
+      "grad_norm": 1.6731312274932861,
+      "learning_rate": 9.045084971874738e-05,
+      "loss": 2.3209,
+      "step": 478
+    },
+    {
+      "epoch": 0.4824120603015075,
+      "grad_norm": 2.067354679107666,
+      "learning_rate": 9.035292362761381e-05,
+      "loss": 2.1253,
+      "step": 480
+    },
+    {
+      "epoch": 0.4844221105527638,
+      "grad_norm": 1.6500976085662842,
+      "learning_rate": 9.025455156125466e-05,
+      "loss": 2.0942,
+      "step": 482
+    },
+    {
+      "epoch": 0.4864321608040201,
+      "grad_norm": 1.7590264081954956,
+      "learning_rate": 9.015573460686509e-05,
+      "loss": 2.3827,
+      "step": 484
+    },
+    {
+      "epoch": 0.4884422110552764,
+      "grad_norm": 1.8388062715530396,
+      "learning_rate": 9.005647385655718e-05,
+      "loss": 2.2311,
+      "step": 486
+    },
+    {
+      "epoch": 0.4904522613065327,
+      "grad_norm": 1.7108007669448853,
+      "learning_rate": 8.995677040734769e-05,
+      "loss": 2.1734,
+      "step": 488
+    },
+    {
+      "epoch": 0.49246231155778897,
+      "grad_norm": 2.4704267978668213,
+      "learning_rate": 8.985662536114613e-05,
+      "loss": 2.2604,
+      "step": 490
+    },
+    {
+      "epoch": 0.49447236180904525,
+      "grad_norm": 1.7523339986801147,
+      "learning_rate": 8.97560398247424e-05,
+      "loss": 2.3831,
+      "step": 492
+    },
+    {
+      "epoch": 0.4964824120603015,
+      "grad_norm": 1.5778342485427856,
+      "learning_rate": 8.965501490979467e-05,
+      "loss": 2.118,
+      "step": 494
+    },
+    {
+      "epoch": 0.4984924623115578,
+      "grad_norm": 1.5618497133255005,
+      "learning_rate": 8.955355173281708e-05,
+      "loss": 2.3317,
+      "step": 496
+    },
+    {
+      "epoch": 0.5005025125628141,
+      "grad_norm": 2.049302339553833,
+      "learning_rate": 8.945165141516734e-05,
+      "loss": 2.2127,
+      "step": 498
+    },
+    {
+      "epoch": 0.5025125628140703,
+      "grad_norm": 2.8442394733428955,
+      "learning_rate": 8.934931508303445e-05,
+      "loss": 2.1562,
+      "step": 500
+    },
+    {
+      "epoch": 0.5045226130653266,
+      "grad_norm": 1.7337098121643066,
+      "learning_rate": 8.924654386742613e-05,
+      "loss": 2.2541,
+      "step": 502
+    },
+    {
+      "epoch": 0.5065326633165829,
+      "grad_norm": 1.6087921857833862,
+      "learning_rate": 8.914333890415639e-05,
+      "loss": 2.3858,
+      "step": 504
+    },
+    {
+      "epoch": 0.5085427135678392,
+      "grad_norm": 1.7101848125457764,
+      "learning_rate": 8.903970133383297e-05,
+      "loss": 2.0366,
+      "step": 506
+    },
+    {
+      "epoch": 0.5105527638190954,
+      "grad_norm": 1.640261173248291,
+      "learning_rate": 8.89356323018447e-05,
+      "loss": 2.3109,
+      "step": 508
+    },
+    {
+      "epoch": 0.5125628140703518,
+      "grad_norm": 1.7887450456619263,
+      "learning_rate": 8.883113295834892e-05,
+      "loss": 2.23,
+      "step": 510
+    },
+    {
+      "epoch": 0.5145728643216081,
+      "grad_norm": 1.772544503211975,
+      "learning_rate": 8.872620445825868e-05,
+      "loss": 2.3887,
+      "step": 512
+    },
+    {
+      "epoch": 0.5165829145728643,
+      "grad_norm": 1.6293214559555054,
+      "learning_rate": 8.862084796122998e-05,
+      "loss": 2.2669,
+      "step": 514
+    },
+    {
+      "epoch": 0.5185929648241207,
+      "grad_norm": 1.9295259714126587,
+      "learning_rate": 8.851506463164907e-05,
+      "loss": 2.4984,
+      "step": 516
+    },
+    {
+      "epoch": 0.5206030150753769,
+      "grad_norm": 1.5557610988616943,
+      "learning_rate": 8.84088556386194e-05,
+      "loss": 2.2621,
+      "step": 518
+    },
+    {
+      "epoch": 0.5226130653266332,
+      "grad_norm": 1.9299404621124268,
+      "learning_rate": 8.83022221559489e-05,
+      "loss": 2.5963,
+      "step": 520
+    },
+    {
+      "epoch": 0.5246231155778894,
+      "grad_norm": 1.7403899431228638,
+      "learning_rate": 8.819516536213683e-05,
+      "loss": 2.261,
+      "step": 522
+    },
+    {
+      "epoch": 0.5266331658291458,
+      "grad_norm": 1.6017705202102661,
+      "learning_rate": 8.808768644036085e-05,
+      "loss": 2.1345,
+      "step": 524
+    },
+    {
+      "epoch": 0.528643216080402,
+      "grad_norm": 1.560952067375183,
+      "learning_rate": 8.797978657846391e-05,
+      "loss": 2.1304,
+      "step": 526
+    },
+    {
+      "epoch": 0.5306532663316583,
+      "grad_norm": 1.63511061668396,
+      "learning_rate": 8.787146696894118e-05,
+      "loss": 2.2639,
+      "step": 528
+    },
+    {
+      "epoch": 0.5326633165829145,
+      "grad_norm": 1.6902211904525757,
+      "learning_rate": 8.776272880892675e-05,
+      "loss": 2.4526,
+      "step": 530
+    },
+    {
+      "epoch": 0.5346733668341709,
+      "grad_norm": 2.2787492275238037,
+      "learning_rate": 8.765357330018056e-05,
+      "loss": 2.1836,
+      "step": 532
+    },
+    {
+      "epoch": 0.5366834170854271,
+      "grad_norm": 1.6377997398376465,
+      "learning_rate": 8.754400164907497e-05,
+      "loss": 2.1207,
+      "step": 534
+    },
+    {
+      "epoch": 0.5386934673366834,
+      "grad_norm": 1.829947829246521,
+      "learning_rate": 8.74340150665815e-05,
+      "loss": 2.4133,
+      "step": 536
+    },
+    {
+      "epoch": 0.5407035175879397,
+      "grad_norm": 1.8088699579238892,
+      "learning_rate": 8.732361476825752e-05,
+      "loss": 2.4477,
+      "step": 538
+    },
+    {
+      "epoch": 0.542713567839196,
+      "grad_norm": 1.8463410139083862,
+      "learning_rate": 8.721280197423258e-05,
+      "loss": 2.5268,
+      "step": 540
+    },
+    {
+      "epoch": 0.5447236180904522,
+      "grad_norm": 1.7526369094848633,
+      "learning_rate": 8.710157790919522e-05,
+      "loss": 2.2297,
+      "step": 542
+    },
+    {
+      "epoch": 0.5467336683417086,
+      "grad_norm": 1.548508882522583,
+      "learning_rate": 8.69899438023792e-05,
+      "loss": 2.3638,
+      "step": 544
+    },
+    {
+      "epoch": 0.5487437185929648,
+      "grad_norm": 1.6327900886535645,
+      "learning_rate": 8.687790088755008e-05,
+      "loss": 2.5126,
+      "step": 546
+    },
+    {
+      "epoch": 0.5507537688442211,
+      "grad_norm": 4.006368160247803,
+      "learning_rate": 8.676545040299145e-05,
+      "loss": 2.4828,
+      "step": 548
+    },
+    {
+      "epoch": 0.5527638190954773,
+      "grad_norm": 1.6274868249893188,
+      "learning_rate": 8.665259359149132e-05,
+      "loss": 2.2662,
+      "step": 550
+    },
+    {
+      "epoch": 0.5547738693467337,
+      "grad_norm": 1.686354398727417,
+      "learning_rate": 8.653933170032842e-05,
+      "loss": 2.4134,
+      "step": 552
+    },
+    {
+      "epoch": 0.5567839195979899,
+      "grad_norm": 1.62911856174469,
+      "learning_rate": 8.642566598125831e-05,
+      "loss": 2.275,
+      "step": 554
+    },
+    {
+      "epoch": 0.5587939698492462,
+      "grad_norm": 1.634544849395752,
+      "learning_rate": 8.631159769049965e-05,
+      "loss": 2.2874,
+      "step": 556
+    },
+    {
+      "epoch": 0.5608040201005026,
+      "grad_norm": 1.5775830745697021,
+      "learning_rate": 8.619712808872024e-05,
+      "loss": 2.2704,
+      "step": 558
+    },
+    {
+      "epoch": 0.5628140703517588,
+      "grad_norm": 1.618634581565857,
+      "learning_rate": 8.60822584410231e-05,
+      "loss": 2.545,
+      "step": 560
+    },
+    {
+      "epoch": 0.5648241206030151,
+      "grad_norm": 1.692204475402832,
+      "learning_rate": 8.596699001693255e-05,
+      "loss": 2.2853,
+      "step": 562
+    },
+    {
+      "epoch": 0.5668341708542713,
+      "grad_norm": 1.717124342918396,
+      "learning_rate": 8.585132409038013e-05,
+      "loss": 2.1552,
+      "step": 564
+    },
+    {
+      "epoch": 0.5688442211055277,
+      "grad_norm": 1.6397979259490967,
+      "learning_rate": 8.573526193969046e-05,
+      "loss": 2.3056,
+      "step": 566
+    },
+    {
+      "epoch": 0.5708542713567839,
+      "grad_norm": 1.6189571619033813,
+      "learning_rate": 8.561880484756725e-05,
+      "loss": 2.1406,
+      "step": 568
+    },
+    {
+      "epoch": 0.5728643216080402,
+      "grad_norm": 1.9604424238204956,
+      "learning_rate": 8.550195410107902e-05,
+      "loss": 2.4183,
+      "step": 570
+    },
+    {
+      "epoch": 0.5748743718592965,
+      "grad_norm": 1.6223558187484741,
+      "learning_rate": 8.538471099164493e-05,
+      "loss": 2.1032,
+      "step": 572
+    },
+    {
+      "epoch": 0.5768844221105528,
+      "grad_norm": 1.8475818634033203,
+      "learning_rate": 8.526707681502044e-05,
+      "loss": 2.1361,
+      "step": 574
+    },
+    {
+      "epoch": 0.578894472361809,
+      "grad_norm": 1.6998199224472046,
+      "learning_rate": 8.51490528712831e-05,
+      "loss": 2.0701,
+      "step": 576
+    },
+    {
+      "epoch": 0.5809045226130654,
+      "grad_norm": 1.5956889390945435,
+      "learning_rate": 8.503064046481803e-05,
+      "loss": 2.3293,
+      "step": 578
+    },
+    {
+      "epoch": 0.5829145728643216,
+      "grad_norm": 1.544945478439331,
+      "learning_rate": 8.491184090430364e-05,
+      "loss": 2.0934,
+      "step": 580
+    },
+    {
+      "epoch": 0.5849246231155779,
+      "grad_norm": 1.5254535675048828,
+      "learning_rate": 8.479265550269714e-05,
+      "loss": 1.9323,
+      "step": 582
+    },
+    {
+      "epoch": 0.5869346733668341,
+      "grad_norm": 1.6815849542617798,
+      "learning_rate": 8.467308557721996e-05,
+      "loss": 2.2834,
+      "step": 584
+    },
+    {
+      "epoch": 0.5889447236180905,
+      "grad_norm": 1.7338416576385498,
+      "learning_rate": 8.455313244934324e-05,
+      "loss": 2.2117,
+      "step": 586
+    },
+    {
+      "epoch": 0.5909547738693467,
+      "grad_norm": 1.6180495023727417,
+      "learning_rate": 8.443279744477324e-05,
+      "loss": 2.1526,
+      "step": 588
+    },
+    {
+      "epoch": 0.592964824120603,
+      "grad_norm": 1.6239418983459473,
+      "learning_rate": 8.43120818934367e-05,
+      "loss": 2.1047,
+      "step": 590
+    },
+    {
+      "epoch": 0.5949748743718593,
+      "grad_norm": 1.6448568105697632,
+      "learning_rate": 8.419098712946601e-05,
+      "loss": 2.0582,
+      "step": 592
+    },
+    {
+      "epoch": 0.5969849246231156,
+      "grad_norm": 1.722649097442627,
+      "learning_rate": 8.406951449118469e-05,
+      "loss": 1.9733,
+      "step": 594
+    },
+    {
+      "epoch": 0.5989949748743718,
+      "grad_norm": 1.7211109399795532,
+      "learning_rate": 8.394766532109242e-05,
+      "loss": 2.2033,
+      "step": 596
+    },
+    {
+      "epoch": 0.6010050251256281,
+      "grad_norm": 1.55821692943573,
+      "learning_rate": 8.382544096585027e-05,
+      "loss": 1.8462,
+      "step": 598
+    },
+    {
+      "epoch": 0.6030150753768844,
+      "grad_norm": 1.5059350728988647,
+      "learning_rate": 8.370284277626577e-05,
+      "loss": 2.0388,
+      "step": 600
+    },
+    {
+      "epoch": 0.6050251256281407,
+      "grad_norm": 1.7469459772109985,
+      "learning_rate": 8.357987210727808e-05,
+      "loss": 2.367,
+      "step": 602
+    },
+    {
+      "epoch": 0.607035175879397,
+      "grad_norm": 1.6946009397506714,
+      "learning_rate": 8.345653031794292e-05,
+      "loss": 2.3654,
+      "step": 604
+    },
+    {
+      "epoch": 0.6090452261306533,
+      "grad_norm": 2.0396924018859863,
+      "learning_rate": 8.333281877141758e-05,
+      "loss": 2.2726,
+      "step": 606
+    },
+    {
+      "epoch": 0.6110552763819096,
+      "grad_norm": 1.6228737831115723,
+      "learning_rate": 8.320873883494585e-05,
+      "loss": 2.2077,
+      "step": 608
+    },
+    {
+      "epoch": 0.6130653266331658,
+      "grad_norm": 1.6517311334609985,
+      "learning_rate": 8.308429187984297e-05,
+      "loss": 2.2134,
+      "step": 610
+    },
+    {
+      "epoch": 0.6150753768844222,
+      "grad_norm": 1.6150524616241455,
+      "learning_rate": 8.295947928148036e-05,
+      "loss": 2.1454,
+      "step": 612
+    },
+    {
+      "epoch": 0.6170854271356784,
+      "grad_norm": 1.5641722679138184,
+      "learning_rate": 8.283430241927052e-05,
+      "loss": 1.9578,
+      "step": 614
+    },
+    {
+      "epoch": 0.6190954773869347,
+      "grad_norm": 1.668264389038086,
+      "learning_rate": 8.270876267665173e-05,
+      "loss": 2.1463,
+      "step": 616
+    },
+    {
+      "epoch": 0.6211055276381909,
+      "grad_norm": 1.6941081285476685,
+      "learning_rate": 8.258286144107276e-05,
+      "loss": 2.2608,
+      "step": 618
+    },
+    {
+      "epoch": 0.6231155778894473,
+      "grad_norm": 1.9184767007827759,
+      "learning_rate": 8.24566001039776e-05,
+      "loss": 2.2404,
+      "step": 620
+    },
+    {
+      "epoch": 0.6251256281407035,
+      "grad_norm": 1.5730127096176147,
+      "learning_rate": 8.232998006078997e-05,
+      "loss": 2.1935,
+      "step": 622
+    },
+    {
+      "epoch": 0.6271356783919598,
+      "grad_norm": 1.4828630685806274,
+      "learning_rate": 8.220300271089807e-05,
+      "loss": 2.2226,
+      "step": 624
+    },
+    {
+      "epoch": 0.629145728643216,
+      "grad_norm": 1.5179678201675415,
+      "learning_rate": 8.207566945763885e-05,
+      "loss": 2.0261,
+      "step": 626
+    },
+    {
+      "epoch": 0.6311557788944724,
+      "grad_norm": 1.4846012592315674,
+      "learning_rate": 8.19479817082828e-05,
+      "loss": 2.1515,
+      "step": 628
+    },
+    {
+      "epoch": 0.6331658291457286,
+      "grad_norm": 1.6346242427825928,
+      "learning_rate": 8.181994087401819e-05,
+      "loss": 2.0147,
+      "step": 630
+    },
+    {
+      "epoch": 0.6351758793969849,
+      "grad_norm": 1.5853772163391113,
+      "learning_rate": 8.169154836993551e-05,
+      "loss": 2.1816,
+      "step": 632
+    },
+    {
+      "epoch": 0.6371859296482412,
+      "grad_norm": 1.573282241821289,
+      "learning_rate": 8.156280561501195e-05,
+      "loss": 2.0802,
+      "step": 634
+    },
+    {
+      "epoch": 0.6391959798994975,
+      "grad_norm": 1.5579333305358887,
+      "learning_rate": 8.143371403209554e-05,
+      "loss": 2.1474,
+      "step": 636
+    },
+    {
+      "epoch": 0.6412060301507537,
+      "grad_norm": 1.6341716051101685,
+      "learning_rate": 8.130427504788955e-05,
+      "loss": 2.293,
+      "step": 638
+    },
+    {
+      "epoch": 0.6432160804020101,
+      "grad_norm": 1.6196293830871582,
+      "learning_rate": 8.117449009293668e-05,
+      "loss": 2.0744,
+      "step": 640
+    },
+    {
+      "epoch": 0.6452261306532663,
+      "grad_norm": 1.702324628829956,
+      "learning_rate": 8.104436060160324e-05,
+      "loss": 2.158,
+      "step": 642
+    },
+    {
+      "epoch": 0.6472361809045226,
+      "grad_norm": 1.734678030014038,
+      "learning_rate": 8.091388801206333e-05,
+      "loss": 2.2079,
+      "step": 644
+    },
+    {
+      "epoch": 0.6492462311557788,
+      "grad_norm": 1.667091727256775,
+      "learning_rate": 8.07830737662829e-05,
+      "loss": 2.1447,
+      "step": 646
+    },
+    {
+      "epoch": 0.6512562814070352,
+      "grad_norm": 1.7557045221328735,
+      "learning_rate": 8.06519193100039e-05,
+      "loss": 2.2181,
+      "step": 648
+    },
+    {
+      "epoch": 0.6532663316582915,
+      "grad_norm": 1.6288976669311523,
+      "learning_rate": 8.052042609272817e-05,
+      "loss": 2.0902,
+      "step": 650
+    },
+    {
+      "epoch": 0.6552763819095477,
+      "grad_norm": 1.6819061040878296,
+      "learning_rate": 8.038859556770151e-05,
+      "loss": 2.3705,
+      "step": 652
+    },
+    {
+      "epoch": 0.6572864321608041,
+      "grad_norm": 1.7110100984573364,
+      "learning_rate": 8.025642919189762e-05,
+      "loss": 2.1783,
+      "step": 654
+    },
+    {
+      "epoch": 0.6592964824120603,
+      "grad_norm": 1.7823935747146606,
+      "learning_rate": 8.012392842600198e-05,
+      "loss": 2.3344,
+      "step": 656
+    },
+    {
+      "epoch": 0.6613065326633166,
+      "grad_norm": 1.661436915397644,
+      "learning_rate": 7.99910947343957e-05,
+      "loss": 2.2976,
+      "step": 658
+    },
+    {
+      "epoch": 0.6633165829145728,
+      "grad_norm": 1.6022391319274902,
+      "learning_rate": 7.985792958513931e-05,
+      "loss": 1.9313,
+      "step": 660
+    },
+    {
+      "epoch": 0.6653266331658292,
+      "grad_norm": 1.5373127460479736,
+      "learning_rate": 7.972443444995663e-05,
+      "loss": 2.182,
+      "step": 662
+    },
+    {
+      "epoch": 0.6673366834170854,
+      "grad_norm": 1.713374376296997,
+      "learning_rate": 7.959061080421839e-05,
+      "loss": 2.3038,
+      "step": 664
+    },
+    {
+      "epoch": 0.6693467336683417,
+      "grad_norm": 1.645225167274475,
+      "learning_rate": 7.9456460126926e-05,
+      "loss": 2.2538,
+      "step": 666
+    },
+    {
+      "epoch": 0.671356783919598,
+      "grad_norm": 1.480999231338501,
+      "learning_rate": 7.932198390069515e-05,
+      "loss": 1.9686,
+      "step": 668
+    },
+    {
+      "epoch": 0.6733668341708543,
+      "grad_norm": 3.5348572731018066,
+      "learning_rate": 7.91871836117395e-05,
+      "loss": 2.3065,
+      "step": 670
+    },
+    {
+      "epoch": 0.6753768844221105,
+      "grad_norm": 1.6705471277236938,
+      "learning_rate": 7.905206074985416e-05,
+      "loss": 2.1366,
+      "step": 672
+    },
+    {
+      "epoch": 0.6773869346733669,
+      "grad_norm": 1.6539703607559204,
+      "learning_rate": 7.891661680839932e-05,
+      "loss": 2.111,
+      "step": 674
+    },
+    {
+      "epoch": 0.6793969849246231,
+      "grad_norm": 1.5140559673309326,
+      "learning_rate": 7.878085328428369e-05,
+      "loss": 2.1999,
+      "step": 676
+    },
+    {
+      "epoch": 0.6814070351758794,
+      "grad_norm": 1.6765365600585938,
+      "learning_rate": 7.86447716779479e-05,
+      "loss": 1.9244,
+      "step": 678
+    },
+    {
+      "epoch": 0.6834170854271356,
+      "grad_norm": 1.6503502130508423,
+      "learning_rate": 7.85083734933481e-05,
+      "loss": 2.0321,
+      "step": 680
+    },
+    {
+      "epoch": 0.685427135678392,
+      "grad_norm": 1.5550287961959839,
+      "learning_rate": 7.83716602379391e-05,
+      "loss": 1.9953,
+      "step": 682
+    },
+    {
+      "epoch": 0.6874371859296482,
+      "grad_norm": 1.5494416952133179,
+      "learning_rate": 7.823463342265792e-05,
+      "loss": 2.1868,
+      "step": 684
+    },
+    {
+      "epoch": 0.6894472361809045,
+      "grad_norm": 1.5480455160140991,
+      "learning_rate": 7.809729456190698e-05,
+      "loss": 2.2091,
+      "step": 686
+    },
+    {
+      "epoch": 0.6914572864321608,
+      "grad_norm": 1.5905412435531616,
+      "learning_rate": 7.795964517353735e-05,
+      "loss": 2.0997,
+      "step": 688
+    },
+    {
+      "epoch": 0.6934673366834171,
+      "grad_norm": 1.6843619346618652,
+      "learning_rate": 7.782168677883206e-05,
+      "loss": 2.2762,
+      "step": 690
+    },
+    {
+      "epoch": 0.6954773869346733,
+      "grad_norm": 1.9944095611572266,
+      "learning_rate": 7.76834209024892e-05,
+      "loss": 1.8889,
+      "step": 692
+    },
+    {
+      "epoch": 0.6974874371859296,
+      "grad_norm": 1.6074081659317017,
+      "learning_rate": 7.754484907260513e-05,
+      "loss": 2.2474,
+      "step": 694
+    },
+    {
+      "epoch": 0.699497487437186,
+      "grad_norm": 1.506535530090332,
+      "learning_rate": 7.740597282065756e-05,
+      "loss": 2.1043,
+      "step": 696
+    },
+    {
+      "epoch": 0.7015075376884422,
+      "grad_norm": 1.5604645013809204,
+      "learning_rate": 7.726679368148864e-05,
+      "loss": 2.0762,
+      "step": 698
+    },
+    {
+      "epoch": 0.7035175879396985,
+      "grad_norm": 1.6519221067428589,
+      "learning_rate": 7.712731319328798e-05,
+      "loss": 2.123,
+      "step": 700
+    },
+    {
+      "epoch": 0.7055276381909548,
+      "grad_norm": 1.775421142578125,
+      "learning_rate": 7.698753289757565e-05,
+      "loss": 2.125,
+      "step": 702
+    },
+    {
+      "epoch": 0.7075376884422111,
+      "grad_norm": 1.552018404006958,
+      "learning_rate": 7.684745433918518e-05,
+      "loss": 1.952,
+      "step": 704
+    },
+    {
+      "epoch": 0.7095477386934673,
+      "grad_norm": 1.8088091611862183,
+      "learning_rate": 7.670707906624644e-05,
+      "loss": 2.2094,
+      "step": 706
+    },
+    {
+      "epoch": 0.7115577889447237,
+      "grad_norm": 1.5663000345230103,
+      "learning_rate": 7.656640863016857e-05,
+      "loss": 2.2196,
+      "step": 708
+    },
+    {
+      "epoch": 0.7135678391959799,
+      "grad_norm": 1.6653417348861694,
+      "learning_rate": 7.642544458562278e-05,
+      "loss": 2.1193,
+      "step": 710
+    },
+    {
+      "epoch": 0.7155778894472362,
+      "grad_norm": 1.5582131147384644,
+      "learning_rate": 7.628418849052523e-05,
+      "loss": 2.0609,
+      "step": 712
+    },
+    {
+      "epoch": 0.7175879396984924,
+      "grad_norm": 1.6001904010772705,
+      "learning_rate": 7.614264190601981e-05,
+      "loss": 1.9337,
+      "step": 714
+    },
+    {
+      "epoch": 0.7195979899497488,
+      "grad_norm": 1.5969533920288086,
+      "learning_rate": 7.600080639646077e-05,
+      "loss": 1.9728,
+      "step": 716
+    },
+    {
+      "epoch": 0.721608040201005,
+      "grad_norm": 1.5485557317733765,
+      "learning_rate": 7.585868352939563e-05,
+      "loss": 2.0791,
+      "step": 718
+    },
+    {
+      "epoch": 0.7236180904522613,
+      "grad_norm": 1.6326680183410645,
+      "learning_rate": 7.571627487554769e-05,
+      "loss": 2.1784,
+      "step": 720
+    },
+    {
+      "epoch": 0.7256281407035176,
+      "grad_norm": 1.7753201723098755,
+      "learning_rate": 7.55735820087987e-05,
+      "loss": 2.1085,
+      "step": 722
+    },
+    {
+      "epoch": 0.7276381909547739,
+      "grad_norm": 1.6765743494033813,
+      "learning_rate": 7.543060650617158e-05,
+      "loss": 2.2751,
+      "step": 724
+    },
+    {
+      "epoch": 0.7296482412060301,
+      "grad_norm": 1.7378075122833252,
+      "learning_rate": 7.528734994781283e-05,
+      "loss": 2.5291,
+      "step": 726
+    },
+    {
+      "epoch": 0.7316582914572864,
+      "grad_norm": 1.5884289741516113,
+      "learning_rate": 7.514381391697517e-05,
+      "loss": 2.1576,
+      "step": 728
+    },
+    {
+      "epoch": 0.7336683417085427,
+      "grad_norm": 1.4826475381851196,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 2.0913,
+      "step": 730
+    },
+    {
+      "epoch": 0.735678391959799,
+      "grad_norm": 1.5756173133850098,
+      "learning_rate": 7.48559097862999e-05,
+      "loss": 1.977,
+      "step": 732
+    },
+    {
+      "epoch": 0.7376884422110552,
+      "grad_norm": 1.5923422574996948,
+      "learning_rate": 7.471154486834105e-05,
+      "loss": 2.1248,
+      "step": 734
+    },
+    {
+      "epoch": 0.7396984924623116,
+      "grad_norm": 1.8260492086410522,
+      "learning_rate": 7.456690684162557e-05,
+      "loss": 2.1871,
+      "step": 736
+    },
+    {
+      "epoch": 0.7417085427135678,
+      "grad_norm": 1.6289384365081787,
+      "learning_rate": 7.442199730467402e-05,
+      "loss": 1.9642,
+      "step": 738
+    },
+    {
+      "epoch": 0.7437185929648241,
+      "grad_norm": 1.423568606376648,
+      "learning_rate": 7.427681785900761e-05,
+      "loss": 1.7796,
+      "step": 740
+    },
+    {
+      "epoch": 0.7457286432160805,
+      "grad_norm": 1.7022663354873657,
+      "learning_rate": 7.413137010913054e-05,
+      "loss": 2.2234,
+      "step": 742
+    },
+    {
+      "epoch": 0.7477386934673367,
+      "grad_norm": 1.6905540227890015,
+      "learning_rate": 7.398565566251232e-05,
+      "loss": 2.1867,
+      "step": 744
+    },
+    {
+      "epoch": 0.749748743718593,
+      "grad_norm": 1.545426607131958,
+      "learning_rate": 7.383967612956988e-05,
+      "loss": 2.0914,
+      "step": 746
+    },
+    {
+      "epoch": 0.7517587939698492,
+      "grad_norm": 1.6682387590408325,
+      "learning_rate": 7.369343312364993e-05,
+      "loss": 2.2405,
+      "step": 748
+    },
+    {
+      "epoch": 0.7537688442211056,
+      "grad_norm": 1.634537935256958,
+      "learning_rate": 7.354692826101102e-05,
+      "loss": 1.9596,
+      "step": 750
+    },
+    {
+      "epoch": 0.7557788944723618,
+      "grad_norm": 1.6119855642318726,
+      "learning_rate": 7.340016316080565e-05,
+      "loss": 2.1663,
+      "step": 752
+    },
+    {
+      "epoch": 0.7577889447236181,
+      "grad_norm": 1.621778130531311,
+      "learning_rate": 7.325313944506254e-05,
+      "loss": 1.9821,
+      "step": 754
+    },
+    {
+      "epoch": 0.7597989949748744,
+      "grad_norm": 2.1226937770843506,
+      "learning_rate": 7.310585873866848e-05,
+      "loss": 2.2974,
+      "step": 756
+    },
+    {
+      "epoch": 0.7618090452261307,
+      "grad_norm": 1.5477526187896729,
+      "learning_rate": 7.295832266935059e-05,
+      "loss": 1.8883,
+      "step": 758
+    },
+    {
+      "epoch": 0.7638190954773869,
+      "grad_norm": 1.5489342212677002,
+      "learning_rate": 7.281053286765815e-05,
+      "loss": 2.0525,
+      "step": 760
+    },
+    {
+      "epoch": 0.7658291457286432,
+      "grad_norm": 2.4198858737945557,
+      "learning_rate": 7.26624909669447e-05,
+      "loss": 2.0754,
+      "step": 762
+    },
+    {
+      "epoch": 0.7678391959798995,
+      "grad_norm": 1.5412720441818237,
+      "learning_rate": 7.251419860334994e-05,
+      "loss": 1.8809,
+      "step": 764
+    },
+    {
+      "epoch": 0.7698492462311558,
+      "grad_norm": 1.656874179840088,
+      "learning_rate": 7.236565741578163e-05,
+      "loss": 2.0115,
+      "step": 766
+    },
+    {
+      "epoch": 0.771859296482412,
+      "grad_norm": 1.5036085844039917,
+      "learning_rate": 7.221686904589754e-05,
+      "loss": 2.2604,
+      "step": 768
+    },
+    {
+      "epoch": 0.7738693467336684,
+      "grad_norm": 1.4638508558273315,
+      "learning_rate": 7.20678351380872e-05,
+      "loss": 1.9812,
+      "step": 770
+    },
+    {
+      "epoch": 0.7758793969849246,
+      "grad_norm": 1.5553574562072754,
+      "learning_rate": 7.191855733945387e-05,
+      "loss": 2.0646,
+      "step": 772
+    },
+    {
+      "epoch": 0.7778894472361809,
+      "grad_norm": 1.5192738771438599,
+      "learning_rate": 7.176903729979621e-05,
+      "loss": 2.3253,
+      "step": 774
+    },
+    {
+      "epoch": 0.7798994974874371,
+      "grad_norm": 1.702390432357788,
+      "learning_rate": 7.161927667159013e-05,
+      "loss": 2.0733,
+      "step": 776
+    },
+    {
+      "epoch": 0.7819095477386935,
+      "grad_norm": 1.551355004310608,
+      "learning_rate": 7.146927710997047e-05,
+      "loss": 2.0301,
+      "step": 778
+    },
+    {
+      "epoch": 0.7839195979899497,
+      "grad_norm": 1.6954432725906372,
+      "learning_rate": 7.13190402727127e-05,
+      "loss": 1.9302,
+      "step": 780
+    },
+    {
+      "epoch": 0.785929648241206,
+      "grad_norm": 1.5084172487258911,
+      "learning_rate": 7.116856782021468e-05,
+      "loss": 2.4387,
+      "step": 782
+    },
+    {
+      "epoch": 0.7879396984924623,
+      "grad_norm": 1.5950528383255005,
+      "learning_rate": 7.101786141547828e-05,
+      "loss": 2.1753,
+      "step": 784
+    },
+    {
+      "epoch": 0.7899497487437186,
+      "grad_norm": 1.674777865409851,
+      "learning_rate": 7.08669227240909e-05,
+      "loss": 2.0804,
+      "step": 786
+    },
+    {
+      "epoch": 0.7919597989949749,
+      "grad_norm": 1.5074516534805298,
+      "learning_rate": 7.071575341420719e-05,
+      "loss": 2.2549,
+      "step": 788
+    },
+    {
+      "epoch": 0.7939698492462312,
+      "grad_norm": 1.6578811407089233,
+      "learning_rate": 7.056435515653059e-05,
+      "loss": 2.1851,
+      "step": 790
+    },
+    {
+      "epoch": 0.7959798994974875,
+      "grad_norm": 1.6693928241729736,
+      "learning_rate": 7.041272962429477e-05,
+      "loss": 2.2608,
+      "step": 792
+    },
+    {
+      "epoch": 0.7979899497487437,
+      "grad_norm": 1.6492758989334106,
+      "learning_rate": 7.026087849324527e-05,
+      "loss": 1.9814,
+      "step": 794
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.59951651096344,
+      "learning_rate": 7.010880344162088e-05,
+      "loss": 2.0959,
+      "step": 796
+    },
+    {
+      "epoch": 0.8020100502512563,
+      "grad_norm": 1.6797847747802734,
+      "learning_rate": 6.995650615013516e-05,
+      "loss": 2.242,
+      "step": 798
+    },
+    {
+      "epoch": 0.8040201005025126,
+      "grad_norm": 1.566300868988037,
+      "learning_rate": 6.980398830195785e-05,
+      "loss": 2.2452,
+      "step": 800
+    },
+    {
+      "epoch": 0.8060301507537688,
+      "grad_norm": 1.471531629562378,
+      "learning_rate": 6.965125158269619e-05,
+      "loss": 1.7636,
+      "step": 802
+    },
+    {
+      "epoch": 0.8080402010050252,
+      "grad_norm": 1.5839449167251587,
+      "learning_rate": 6.94982976803764e-05,
+      "loss": 2.299,
+      "step": 804
+    },
+    {
+      "epoch": 0.8100502512562814,
+      "grad_norm": 1.5502268075942993,
+      "learning_rate": 6.934512828542497e-05,
+      "loss": 1.8683,
+      "step": 806
+    },
+    {
+      "epoch": 0.8120603015075377,
+      "grad_norm": 1.532381296157837,
+      "learning_rate": 6.919174509065004e-05,
+      "loss": 1.8511,
+      "step": 808
+    },
+    {
+      "epoch": 0.8140703517587939,
+      "grad_norm": 1.4679884910583496,
+      "learning_rate": 6.903814979122249e-05,
+      "loss": 2.1163,
+      "step": 810
+    },
+    {
+      "epoch": 0.8160804020100503,
+      "grad_norm": 1.5164811611175537,
+      "learning_rate": 6.888434408465751e-05,
+      "loss": 2.1253,
+      "step": 812
+    },
+    {
+      "epoch": 0.8180904522613065,
+      "grad_norm": 1.6110063791275024,
+      "learning_rate": 6.873032967079561e-05,
+      "loss": 2.2175,
+      "step": 814
+    },
+    {
+      "epoch": 0.8201005025125628,
+      "grad_norm": 1.4853484630584717,
+      "learning_rate": 6.85761082517839e-05,
+      "loss": 2.1247,
+      "step": 816
+    },
+    {
+      "epoch": 0.8221105527638191,
+      "grad_norm": 1.7195366621017456,
+      "learning_rate": 6.842168153205734e-05,
+      "loss": 2.2131,
+      "step": 818
+    },
+    {
+      "epoch": 0.8241206030150754,
+      "grad_norm": 1.622910976409912,
+      "learning_rate": 6.826705121831976e-05,
+      "loss": 2.0826,
+      "step": 820
+    },
+    {
+      "epoch": 0.8261306532663316,
+      "grad_norm": 1.5393376350402832,
+      "learning_rate": 6.811221901952513e-05,
+      "loss": 1.9764,
+      "step": 822
+    },
+    {
+      "epoch": 0.828140703517588,
+      "grad_norm": 1.475002408027649,
+      "learning_rate": 6.795718664685868e-05,
+      "loss": 1.9606,
+      "step": 824
+    },
+    {
+      "epoch": 0.8301507537688442,
+      "grad_norm": 1.607079267501831,
+      "learning_rate": 6.780195581371784e-05,
+      "loss": 2.3341,
+      "step": 826
+    },
+    {
+      "epoch": 0.8321608040201005,
+      "grad_norm": 1.5544415712356567,
+      "learning_rate": 6.764652823569344e-05,
+      "loss": 2.3198,
+      "step": 828
+    },
+    {
+      "epoch": 0.8341708542713567,
+      "grad_norm": 1.5592552423477173,
+      "learning_rate": 6.749090563055076e-05,
+      "loss": 1.8339,
+      "step": 830
+    },
+    {
+      "epoch": 0.8361809045226131,
+      "grad_norm": 1.6058601140975952,
+      "learning_rate": 6.733508971821036e-05,
+      "loss": 2.0026,
+      "step": 832
+    },
+    {
+      "epoch": 0.8381909547738694,
+      "grad_norm": 1.5740671157836914,
+      "learning_rate": 6.717908222072935e-05,
+      "loss": 2.1512,
+      "step": 834
+    },
+    {
+      "epoch": 0.8402010050251256,
+      "grad_norm": 1.4987316131591797,
+      "learning_rate": 6.702288486228216e-05,
+      "loss": 2.1372,
+      "step": 836
+    },
+    {
+      "epoch": 0.842211055276382,
+      "grad_norm": 1.501441478729248,
+      "learning_rate": 6.686649936914152e-05,
+      "loss": 2.1216,
+      "step": 838
+    },
+    {
+      "epoch": 0.8442211055276382,
+      "grad_norm": 1.522459626197815,
+      "learning_rate": 6.670992746965938e-05,
+      "loss": 2.2921,
+      "step": 840
+    },
+    {
+      "epoch": 0.8462311557788945,
+      "grad_norm": 1.5145554542541504,
+      "learning_rate": 6.65531708942479e-05,
+      "loss": 2.0768,
+      "step": 842
+    },
+    {
+      "epoch": 0.8482412060301507,
+      "grad_norm": 1.7844620943069458,
+      "learning_rate": 6.639623137536023e-05,
+      "loss": 2.0404,
+      "step": 844
+    },
+    {
+      "epoch": 0.8502512562814071,
+      "grad_norm": 1.5324727296829224,
+      "learning_rate": 6.623911064747133e-05,
+      "loss": 1.9641,
+      "step": 846
+    },
+    {
+      "epoch": 0.8522613065326633,
+      "grad_norm": 1.5896046161651611,
+      "learning_rate": 6.608181044705892e-05,
+      "loss": 1.922,
+      "step": 848
+    },
+    {
+      "epoch": 0.8542713567839196,
+      "grad_norm": 1.531399130821228,
+      "learning_rate": 6.592433251258423e-05,
+      "loss": 1.9917,
+      "step": 850
+    },
+    {
+      "epoch": 0.8562814070351759,
+      "grad_norm": 1.3786967992782593,
+      "learning_rate": 6.576667858447272e-05,
+      "loss": 2.0952,
+      "step": 852
+    },
+    {
+      "epoch": 0.8582914572864322,
+      "grad_norm": 1.5708434581756592,
+      "learning_rate": 6.560885040509499e-05,
+      "loss": 2.0783,
+      "step": 854
+    },
+    {
+      "epoch": 0.8603015075376884,
+      "grad_norm": 1.5222376585006714,
+      "learning_rate": 6.545084971874738e-05,
+      "loss": 1.9323,
+      "step": 856
+    },
+    {
+      "epoch": 0.8623115577889447,
+      "grad_norm": 1.591243028640747,
+      "learning_rate": 6.529267827163277e-05,
+      "loss": 2.0764,
+      "step": 858
+    },
+    {
+      "epoch": 0.864321608040201,
+      "grad_norm": 1.5842283964157104,
+      "learning_rate": 6.51343378118413e-05,
+      "loss": 2.183,
+      "step": 860
+    },
+    {
+      "epoch": 0.8663316582914573,
+      "grad_norm": 1.5385973453521729,
+      "learning_rate": 6.497583008933097e-05,
+      "loss": 2.2194,
+      "step": 862
+    },
+    {
+      "epoch": 0.8683417085427135,
+      "grad_norm": 1.5147958993911743,
+      "learning_rate": 6.481715685590836e-05,
+      "loss": 1.9953,
+      "step": 864
+    },
+    {
+      "epoch": 0.8703517587939699,
+      "grad_norm": 1.5883772373199463,
+      "learning_rate": 6.465831986520927e-05,
+      "loss": 2.0174,
+      "step": 866
+    },
+    {
+      "epoch": 0.8723618090452261,
+      "grad_norm": 1.5510574579238892,
+      "learning_rate": 6.449932087267932e-05,
+      "loss": 1.9516,
+      "step": 868
+    },
+    {
+      "epoch": 0.8743718592964824,
+      "grad_norm": 1.4354772567749023,
+      "learning_rate": 6.434016163555452e-05,
+      "loss": 1.9101,
+      "step": 870
+    },
+    {
+      "epoch": 0.8763819095477386,
+      "grad_norm": 1.5250461101531982,
+      "learning_rate": 6.418084391284192e-05,
+      "loss": 2.1036,
+      "step": 872
+    },
+    {
+      "epoch": 0.878391959798995,
+      "grad_norm": 1.6079366207122803,
+      "learning_rate": 6.402136946530014e-05,
+      "loss": 2.1401,
+      "step": 874
+    },
+    {
+      "epoch": 0.8804020100502512,
+      "grad_norm": 1.5236170291900635,
+      "learning_rate": 6.386174005541986e-05,
+      "loss": 1.9832,
+      "step": 876
+    },
+    {
+      "epoch": 0.8824120603015075,
+      "grad_norm": 1.5333629846572876,
+      "learning_rate": 6.370195744740442e-05,
+      "loss": 1.8619,
+      "step": 878
+    },
+    {
+      "epoch": 0.8844221105527639,
+      "grad_norm": 1.5753098726272583,
+      "learning_rate": 6.354202340715026e-05,
+      "loss": 1.8309,
+      "step": 880
+    },
+    {
+      "epoch": 0.8864321608040201,
+      "grad_norm": 1.5884308815002441,
+      "learning_rate": 6.338193970222744e-05,
+      "loss": 2.1437,
+      "step": 882
+    },
+    {
+      "epoch": 0.8884422110552764,
+      "grad_norm": 1.5323386192321777,
+      "learning_rate": 6.322170810186012e-05,
+      "loss": 2.1074,
+      "step": 884
+    },
+    {
+      "epoch": 0.8904522613065327,
+      "grad_norm": 1.4975218772888184,
+      "learning_rate": 6.306133037690693e-05,
+      "loss": 2.0906,
+      "step": 886
+    },
+    {
+      "epoch": 0.892462311557789,
+      "grad_norm": 1.5518718957901,
+      "learning_rate": 6.290080829984147e-05,
+      "loss": 1.9955,
+      "step": 888
+    },
+    {
+      "epoch": 0.8944723618090452,
+      "grad_norm": 1.5773143768310547,
+      "learning_rate": 6.274014364473274e-05,
+      "loss": 2.2483,
+      "step": 890
+    },
+    {
+      "epoch": 0.8964824120603015,
+      "grad_norm": 1.533818006515503,
+      "learning_rate": 6.257933818722543e-05,
+      "loss": 2.093,
+      "step": 892
+    },
+    {
+      "epoch": 0.8984924623115578,
+      "grad_norm": 1.442911148071289,
+      "learning_rate": 6.241839370452041e-05,
+      "loss": 2.1291,
+      "step": 894
+    },
+    {
+      "epoch": 0.9005025125628141,
+      "grad_norm": 1.498679518699646,
+      "learning_rate": 6.2257311975355e-05,
+      "loss": 1.9726,
+      "step": 896
+    },
+    {
+      "epoch": 0.9025125628140703,
+      "grad_norm": 1.647017002105713,
+      "learning_rate": 6.209609477998338e-05,
+      "loss": 1.9182,
+      "step": 898
+    },
+    {
+      "epoch": 0.9045226130653267,
+      "grad_norm": 1.546736240386963,
+      "learning_rate": 6.19347439001569e-05,
+      "loss": 2.0869,
+      "step": 900
+    },
+    {
+      "epoch": 0.9065326633165829,
+      "grad_norm": 1.4182671308517456,
+      "learning_rate": 6.177326111910429e-05,
+      "loss": 2.1178,
+      "step": 902
+    },
+    {
+      "epoch": 0.9085427135678392,
+      "grad_norm": 1.685409426689148,
+      "learning_rate": 6.161164822151213e-05,
+      "loss": 2.1423,
+      "step": 904
+    },
+    {
+      "epoch": 0.9105527638190954,
+      "grad_norm": 1.488713026046753,
+      "learning_rate": 6.144990699350497e-05,
+      "loss": 2.2579,
+      "step": 906
+    },
+    {
+      "epoch": 0.9125628140703518,
+      "grad_norm": 1.6088262796401978,
+      "learning_rate": 6.128803922262573e-05,
+      "loss": 2.1192,
+      "step": 908
+    },
+    {
+      "epoch": 0.914572864321608,
+      "grad_norm": 1.6858761310577393,
+      "learning_rate": 6.112604669781572e-05,
+      "loss": 1.9913,
+      "step": 910
+    },
+    {
+      "epoch": 0.9165829145728643,
+      "grad_norm": 1.4918874502182007,
+      "learning_rate": 6.096393120939516e-05,
+      "loss": 2.0041,
+      "step": 912
+    },
+    {
+      "epoch": 0.9185929648241206,
+      "grad_norm": 1.501802921295166,
+      "learning_rate": 6.08016945490432e-05,
+      "loss": 2.1627,
+      "step": 914
+    },
+    {
+      "epoch": 0.9206030150753769,
+      "grad_norm": 1.6295005083084106,
+      "learning_rate": 6.063933850977811e-05,
+      "loss": 2.281,
+      "step": 916
+    },
+    {
+      "epoch": 0.9226130653266331,
+      "grad_norm": 1.4057841300964355,
+      "learning_rate": 6.04768648859376e-05,
+      "loss": 2.0616,
+      "step": 918
+    },
+    {
+      "epoch": 0.9246231155778895,
+      "grad_norm": 1.386192798614502,
+      "learning_rate": 6.031427547315889e-05,
+      "loss": 1.9191,
+      "step": 920
+    },
+    {
+      "epoch": 0.9266331658291457,
+      "grad_norm": 1.4363282918930054,
+      "learning_rate": 6.015157206835881e-05,
+      "loss": 1.94,
+      "step": 922
+    },
+    {
+      "epoch": 0.928643216080402,
+      "grad_norm": 1.6516571044921875,
+      "learning_rate": 5.9988756469714135e-05,
+      "loss": 2.3657,
+      "step": 924
+    },
+    {
+      "epoch": 0.9306532663316583,
+      "grad_norm": 1.465562105178833,
+      "learning_rate": 5.982583047664151e-05,
+      "loss": 2.1271,
+      "step": 926
+    },
+    {
+      "epoch": 0.9326633165829146,
+      "grad_norm": 1.5715078115463257,
+      "learning_rate": 5.9662795889777666e-05,
+      "loss": 2.1258,
+      "step": 928
+    },
+    {
+      "epoch": 0.9346733668341709,
+      "grad_norm": 3.0183351039886475,
+      "learning_rate": 5.949965451095951e-05,
+      "loss": 2.2448,
+      "step": 930
+    },
+    {
+      "epoch": 0.9366834170854271,
+      "grad_norm": 1.5689525604248047,
+      "learning_rate": 5.933640814320417e-05,
+      "loss": 2.161,
+      "step": 932
+    },
+    {
+      "epoch": 0.9386934673366835,
+      "grad_norm": 1.5857359170913696,
+      "learning_rate": 5.917305859068912e-05,
+      "loss": 2.2529,
+      "step": 934
+    },
+    {
+      "epoch": 0.9407035175879397,
+      "grad_norm": 1.4421300888061523,
+      "learning_rate": 5.900960765873222e-05,
+      "loss": 2.1033,
+      "step": 936
+    },
+    {
+      "epoch": 0.942713567839196,
+      "grad_norm": 1.5819547176361084,
+      "learning_rate": 5.8846057153771786e-05,
+      "loss": 2.2916,
+      "step": 938
+    },
+    {
+      "epoch": 0.9447236180904522,
+      "grad_norm": 1.4737604856491089,
+      "learning_rate": 5.868240888334653e-05,
+      "loss": 2.005,
+      "step": 940
+    },
+    {
+      "epoch": 0.9467336683417086,
+      "grad_norm": 1.4487162828445435,
+      "learning_rate": 5.8518664656075706e-05,
+      "loss": 1.9979,
+      "step": 942
+    },
+    {
+      "epoch": 0.9487437185929648,
+      "grad_norm": 1.5933785438537598,
+      "learning_rate": 5.835482628163909e-05,
+      "loss": 1.7228,
+      "step": 944
+    },
+    {
+      "epoch": 0.9507537688442211,
+      "grad_norm": 1.6304993629455566,
+      "learning_rate": 5.819089557075689e-05,
+      "loss": 2.109,
+      "step": 946
+    },
+    {
+      "epoch": 0.9527638190954774,
+      "grad_norm": 1.5096608400344849,
+      "learning_rate": 5.802687433516989e-05,
+      "loss": 1.9898,
+      "step": 948
+    },
+    {
+      "epoch": 0.9547738693467337,
+      "grad_norm": 1.611997127532959,
+      "learning_rate": 5.786276438761927e-05,
+      "loss": 1.9248,
+      "step": 950
+    },
+    {
+      "epoch": 0.9567839195979899,
+      "grad_norm": 1.6030073165893555,
+      "learning_rate": 5.7698567541826675e-05,
+      "loss": 2.0681,
+      "step": 952
+    },
+    {
+      "epoch": 0.9587939698492463,
+      "grad_norm": 1.569161057472229,
+      "learning_rate": 5.753428561247416e-05,
+      "loss": 1.919,
+      "step": 954
+    },
+    {
+      "epoch": 0.9608040201005025,
+      "grad_norm": 1.5151729583740234,
+      "learning_rate": 5.7369920415184064e-05,
+      "loss": 2.1896,
+      "step": 956
+    },
+    {
+      "epoch": 0.9628140703517588,
+      "grad_norm": 1.6666862964630127,
+      "learning_rate": 5.7205473766499005e-05,
+      "loss": 2.0416,
+      "step": 958
+    },
+    {
+      "epoch": 0.964824120603015,
+      "grad_norm": 1.5797557830810547,
+      "learning_rate": 5.704094748386184e-05,
+      "loss": 1.9272,
+      "step": 960
+    },
+    {
+      "epoch": 0.9668341708542714,
+      "grad_norm": 1.508834719657898,
+      "learning_rate": 5.6876343385595446e-05,
+      "loss": 1.7583,
+      "step": 962
+    },
+    {
+      "epoch": 0.9688442211055276,
+      "grad_norm": 1.4971197843551636,
+      "learning_rate": 5.6711663290882776e-05,
+      "loss": 1.9919,
+      "step": 964
+    },
+    {
+      "epoch": 0.9708542713567839,
+      "grad_norm": 1.6074587106704712,
+      "learning_rate": 5.6546909019746666e-05,
+      "loss": 1.982,
+      "step": 966
+    },
+    {
+      "epoch": 0.9728643216080402,
+      "grad_norm": 3.3276638984680176,
+      "learning_rate": 5.6382082393029746e-05,
+      "loss": 2.0718,
+      "step": 968
+    },
+    {
+      "epoch": 0.9748743718592965,
+      "grad_norm": 1.4836559295654297,
+      "learning_rate": 5.621718523237427e-05,
+      "loss": 2.196,
+      "step": 970
+    },
+    {
+      "epoch": 0.9768844221105528,
+      "grad_norm": 1.502540946006775,
+      "learning_rate": 5.605221936020207e-05,
+      "loss": 2.1674,
+      "step": 972
+    },
+    {
+      "epoch": 0.978894472361809,
+      "grad_norm": 1.4632574319839478,
+      "learning_rate": 5.588718659969438e-05,
+      "loss": 2.2086,
+      "step": 974
+    },
+    {
+      "epoch": 0.9809045226130654,
+      "grad_norm": 1.5661799907684326,
+      "learning_rate": 5.57220887747716e-05,
+      "loss": 2.0454,
+      "step": 976
+    },
+    {
+      "epoch": 0.9829145728643216,
+      "grad_norm": 1.5829559564590454,
+      "learning_rate": 5.5556927710073314e-05,
+      "loss": 1.9525,
+      "step": 978
+    },
+    {
+      "epoch": 0.9849246231155779,
+      "grad_norm": 1.6690107583999634,
+      "learning_rate": 5.539170523093794e-05,
+      "loss": 2.0355,
+      "step": 980
+    },
+    {
+      "epoch": 0.9869346733668342,
+      "grad_norm": 1.524163842201233,
+      "learning_rate": 5.522642316338268e-05,
+      "loss": 2.0668,
+      "step": 982
+    },
+    {
+      "epoch": 0.9889447236180905,
+      "grad_norm": 1.5150009393692017,
+      "learning_rate": 5.506108333408329e-05,
+      "loss": 2.1167,
+      "step": 984
+    },
+    {
+      "epoch": 0.9909547738693467,
+      "grad_norm": 1.5068144798278809,
+      "learning_rate": 5.489568757035391e-05,
+      "loss": 2.0714,
+      "step": 986
+    },
+    {
+      "epoch": 0.992964824120603,
+      "grad_norm": 1.5889198780059814,
+      "learning_rate": 5.473023770012686e-05,
+      "loss": 2.0012,
+      "step": 988
+    },
+    {
+      "epoch": 0.9949748743718593,
+      "grad_norm": 1.4362760782241821,
+      "learning_rate": 5.456473555193242e-05,
+      "loss": 1.9559,
+      "step": 990
+    },
+    {
+      "epoch": 0.9969849246231156,
+      "grad_norm": 1.4204431772232056,
+      "learning_rate": 5.4399182954878656e-05,
+      "loss": 1.8115,
+      "step": 992
+    },
+    {
+      "epoch": 0.9989949748743718,
+      "grad_norm": 1.500132441520691,
+      "learning_rate": 5.4233581738631165e-05,
+      "loss": 2.2726,
+      "step": 994
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 1990,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 995,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.6840475756671795e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-995/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a6292fe987426601ff3a282fbc93263d2cc3acea82d69c0e4d76a82ed072116
+size 5432

config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-ace_Latn-coli-align/",
+  "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:dab1ba52a0dcbed9b5315742c600110fdc1df4fe9a936e9d7840174e0740b493
+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 sha256:95d189204f4a7c654e0b96823d05202ca6230116fe3450954291fbcf93c8b730
+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 sha256:4a6292fe987426601ff3a282fbc93263d2cc3acea82d69c0e4d76a82ed072116
+size 5432