Upload folder using huggingface_hub

checkpoint-1940/config.json

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

checkpoint-1940/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-1940/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3488443091602791280dc0af33efa0c8084c3dbaa850fd0bfd94d1517cec9f1
+size 4950070472

checkpoint-1940/model-00002-of-00006.safetensors

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

checkpoint-1940/model-00003-of-00006.safetensors

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

checkpoint-1940/model-00004-of-00006.safetensors

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

checkpoint-1940/model-00005-of-00006.safetensors

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

checkpoint-1940/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:673a529d534d8f91fc38f1075bb12afff441edb45a25ffe1644356a5ce6e1d6f
+size 4614509080

checkpoint-1940/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-1940/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b095be0fef3c421409c0f8429b10d53d7a92c0b9d373504de938c33a4be36ad
+size 13055369358

checkpoint-1940/rng_state.pth

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

checkpoint-1940/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8fea9108509b080df955250c61ab0463fa30b7ddd90ea66dd24ea5707f81069
+size 1064

checkpoint-1940/trainer_state.json

@@ -0,0 +1,3428 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 1940,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002061855670103093,
+      "grad_norm": 4.401242256164551,
+      "learning_rate": 2.061855670103093e-06,
+      "loss": 5.7831,
+      "step": 4
+    },
+    {
+      "epoch": 0.004123711340206186,
+      "grad_norm": 4.165046691894531,
+      "learning_rate": 4.123711340206186e-06,
+      "loss": 5.6527,
+      "step": 8
+    },
+    {
+      "epoch": 0.006185567010309278,
+      "grad_norm": 3.4843931198120117,
+      "learning_rate": 6.185567010309279e-06,
+      "loss": 5.5896,
+      "step": 12
+    },
+    {
+      "epoch": 0.008247422680412371,
+      "grad_norm": 3.29345965385437,
+      "learning_rate": 8.247422680412371e-06,
+      "loss": 5.3939,
+      "step": 16
+    },
+    {
+      "epoch": 0.010309278350515464,
+      "grad_norm": 3.1640310287475586,
+      "learning_rate": 1.0309278350515464e-05,
+      "loss": 5.0285,
+      "step": 20
+    },
+    {
+      "epoch": 0.012371134020618556,
+      "grad_norm": 3.238795757293701,
+      "learning_rate": 1.2371134020618558e-05,
+      "loss": 5.3506,
+      "step": 24
+    },
+    {
+      "epoch": 0.01443298969072165,
+      "grad_norm": 2.786606550216675,
+      "learning_rate": 1.4432989690721649e-05,
+      "loss": 4.8622,
+      "step": 28
+    },
+    {
+      "epoch": 0.016494845360824743,
+      "grad_norm": 2.689506769180298,
+      "learning_rate": 1.6494845360824743e-05,
+      "loss": 4.7998,
+      "step": 32
+    },
+    {
+      "epoch": 0.018556701030927835,
+      "grad_norm": 3.6669228076934814,
+      "learning_rate": 1.8556701030927837e-05,
+      "loss": 4.8852,
+      "step": 36
+    },
+    {
+      "epoch": 0.020618556701030927,
+      "grad_norm": 2.614325761795044,
+      "learning_rate": 2.0618556701030927e-05,
+      "loss": 4.5481,
+      "step": 40
+    },
+    {
+      "epoch": 0.02268041237113402,
+      "grad_norm": 2.6969540119171143,
+      "learning_rate": 2.268041237113402e-05,
+      "loss": 4.4287,
+      "step": 44
+    },
+    {
+      "epoch": 0.024742268041237112,
+      "grad_norm": 2.5721373558044434,
+      "learning_rate": 2.4742268041237116e-05,
+      "loss": 4.2744,
+      "step": 48
+    },
+    {
+      "epoch": 0.026804123711340205,
+      "grad_norm": 2.7030205726623535,
+      "learning_rate": 2.6804123711340206e-05,
+      "loss": 4.2424,
+      "step": 52
+    },
+    {
+      "epoch": 0.0288659793814433,
+      "grad_norm": 2.5551116466522217,
+      "learning_rate": 2.8865979381443297e-05,
+      "loss": 4.1868,
+      "step": 56
+    },
+    {
+      "epoch": 0.030927835051546393,
+      "grad_norm": 2.6516504287719727,
+      "learning_rate": 3.0927835051546395e-05,
+      "loss": 3.9569,
+      "step": 60
+    },
+    {
+      "epoch": 0.032989690721649485,
+      "grad_norm": 2.516420602798462,
+      "learning_rate": 3.2989690721649485e-05,
+      "loss": 4.0263,
+      "step": 64
+    },
+    {
+      "epoch": 0.03505154639175258,
+      "grad_norm": 2.4037609100341797,
+      "learning_rate": 3.5051546391752576e-05,
+      "loss": 4.001,
+      "step": 68
+    },
+    {
+      "epoch": 0.03711340206185567,
+      "grad_norm": 2.3336949348449707,
+      "learning_rate": 3.7113402061855674e-05,
+      "loss": 3.7654,
+      "step": 72
+    },
+    {
+      "epoch": 0.03917525773195876,
+      "grad_norm": 2.5022168159484863,
+      "learning_rate": 3.9175257731958764e-05,
+      "loss": 3.7868,
+      "step": 76
+    },
+    {
+      "epoch": 0.041237113402061855,
+      "grad_norm": 2.496216297149658,
+      "learning_rate": 4.1237113402061855e-05,
+      "loss": 3.7257,
+      "step": 80
+    },
+    {
+      "epoch": 0.04329896907216495,
+      "grad_norm": 2.3523142337799072,
+      "learning_rate": 4.329896907216495e-05,
+      "loss": 3.6547,
+      "step": 84
+    },
+    {
+      "epoch": 0.04536082474226804,
+      "grad_norm": 2.4950451850891113,
+      "learning_rate": 4.536082474226804e-05,
+      "loss": 3.5265,
+      "step": 88
+    },
+    {
+      "epoch": 0.04742268041237113,
+      "grad_norm": 2.5210957527160645,
+      "learning_rate": 4.7422680412371134e-05,
+      "loss": 3.5772,
+      "step": 92
+    },
+    {
+      "epoch": 0.049484536082474224,
+      "grad_norm": 2.3704540729522705,
+      "learning_rate": 4.948453608247423e-05,
+      "loss": 3.6708,
+      "step": 96
+    },
+    {
+      "epoch": 0.05154639175257732,
+      "grad_norm": 2.361452102661133,
+      "learning_rate": 5.1546391752577315e-05,
+      "loss": 3.4935,
+      "step": 100
+    },
+    {
+      "epoch": 0.05360824742268041,
+      "grad_norm": 2.4770519733428955,
+      "learning_rate": 5.360824742268041e-05,
+      "loss": 3.4953,
+      "step": 104
+    },
+    {
+      "epoch": 0.05567010309278351,
+      "grad_norm": 2.6673684120178223,
+      "learning_rate": 5.567010309278351e-05,
+      "loss": 3.4362,
+      "step": 108
+    },
+    {
+      "epoch": 0.0577319587628866,
+      "grad_norm": 2.3836777210235596,
+      "learning_rate": 5.7731958762886594e-05,
+      "loss": 3.4341,
+      "step": 112
+    },
+    {
+      "epoch": 0.05979381443298969,
+      "grad_norm": 2.4742355346679688,
+      "learning_rate": 5.979381443298969e-05,
+      "loss": 3.4443,
+      "step": 116
+    },
+    {
+      "epoch": 0.061855670103092786,
+      "grad_norm": 2.513885974884033,
+      "learning_rate": 6.185567010309279e-05,
+      "loss": 3.3867,
+      "step": 120
+    },
+    {
+      "epoch": 0.06391752577319587,
+      "grad_norm": 2.633009433746338,
+      "learning_rate": 6.391752577319587e-05,
+      "loss": 3.3804,
+      "step": 124
+    },
+    {
+      "epoch": 0.06597938144329897,
+      "grad_norm": 2.7852959632873535,
+      "learning_rate": 6.597938144329897e-05,
+      "loss": 3.4522,
+      "step": 128
+    },
+    {
+      "epoch": 0.06804123711340206,
+      "grad_norm": 2.9940905570983887,
+      "learning_rate": 6.804123711340207e-05,
+      "loss": 3.3606,
+      "step": 132
+    },
+    {
+      "epoch": 0.07010309278350516,
+      "grad_norm": 2.4730873107910156,
+      "learning_rate": 7.010309278350515e-05,
+      "loss": 3.2392,
+      "step": 136
+    },
+    {
+      "epoch": 0.07216494845360824,
+      "grad_norm": 2.1951076984405518,
+      "learning_rate": 7.216494845360825e-05,
+      "loss": 3.4019,
+      "step": 140
+    },
+    {
+      "epoch": 0.07422680412371134,
+      "grad_norm": 1.9874905347824097,
+      "learning_rate": 7.422680412371135e-05,
+      "loss": 3.1746,
+      "step": 144
+    },
+    {
+      "epoch": 0.07628865979381444,
+      "grad_norm": 2.2700319290161133,
+      "learning_rate": 7.628865979381443e-05,
+      "loss": 3.1944,
+      "step": 148
+    },
+    {
+      "epoch": 0.07835051546391752,
+      "grad_norm": 2.2276201248168945,
+      "learning_rate": 7.835051546391753e-05,
+      "loss": 3.2753,
+      "step": 152
+    },
+    {
+      "epoch": 0.08041237113402062,
+      "grad_norm": 2.150624990463257,
+      "learning_rate": 8.041237113402063e-05,
+      "loss": 3.2761,
+      "step": 156
+    },
+    {
+      "epoch": 0.08247422680412371,
+      "grad_norm": 2.2154808044433594,
+      "learning_rate": 8.247422680412371e-05,
+      "loss": 3.1575,
+      "step": 160
+    },
+    {
+      "epoch": 0.08453608247422681,
+      "grad_norm": 2.6334950923919678,
+      "learning_rate": 8.453608247422681e-05,
+      "loss": 3.185,
+      "step": 164
+    },
+    {
+      "epoch": 0.0865979381443299,
+      "grad_norm": 2.70011305809021,
+      "learning_rate": 8.65979381443299e-05,
+      "loss": 3.1316,
+      "step": 168
+    },
+    {
+      "epoch": 0.088659793814433,
+      "grad_norm": 2.533686637878418,
+      "learning_rate": 8.865979381443299e-05,
+      "loss": 3.1387,
+      "step": 172
+    },
+    {
+      "epoch": 0.09072164948453608,
+      "grad_norm": 2.154860019683838,
+      "learning_rate": 9.072164948453609e-05,
+      "loss": 3.1592,
+      "step": 176
+    },
+    {
+      "epoch": 0.09278350515463918,
+      "grad_norm": 2.1940603256225586,
+      "learning_rate": 9.278350515463918e-05,
+      "loss": 3.1773,
+      "step": 180
+    },
+    {
+      "epoch": 0.09484536082474226,
+      "grad_norm": 2.257709264755249,
+      "learning_rate": 9.484536082474227e-05,
+      "loss": 3.1052,
+      "step": 184
+    },
+    {
+      "epoch": 0.09690721649484536,
+      "grad_norm": 2.08589768409729,
+      "learning_rate": 9.690721649484537e-05,
+      "loss": 3.0373,
+      "step": 188
+    },
+    {
+      "epoch": 0.09896907216494845,
+      "grad_norm": 2.060812473297119,
+      "learning_rate": 9.896907216494846e-05,
+      "loss": 3.112,
+      "step": 192
+    },
+    {
+      "epoch": 0.10103092783505155,
+      "grad_norm": 2.3649418354034424,
+      "learning_rate": 9.999992735780168e-05,
+      "loss": 3.198,
+      "step": 196
+    },
+    {
+      "epoch": 0.10309278350515463,
+      "grad_norm": 2.339231252670288,
+      "learning_rate": 9.999934622148157e-05,
+      "loss": 3.1736,
+      "step": 200
+    },
+    {
+      "epoch": 0.10515463917525773,
+      "grad_norm": 2.037454128265381,
+      "learning_rate": 9.999818395559577e-05,
+      "loss": 3.1006,
+      "step": 204
+    },
+    {
+      "epoch": 0.10721649484536082,
+      "grad_norm": 2.343181848526001,
+      "learning_rate": 9.999644057365295e-05,
+      "loss": 3.1055,
+      "step": 208
+    },
+    {
+      "epoch": 0.10927835051546392,
+      "grad_norm": 2.3469326496124268,
+      "learning_rate": 9.999411609591603e-05,
+      "loss": 3.0211,
+      "step": 212
+    },
+    {
+      "epoch": 0.11134020618556702,
+      "grad_norm": 2.2599899768829346,
+      "learning_rate": 9.999121054940182e-05,
+      "loss": 3.0236,
+      "step": 216
+    },
+    {
+      "epoch": 0.1134020618556701,
+      "grad_norm": 2.114220142364502,
+      "learning_rate": 9.998772396788072e-05,
+      "loss": 3.0895,
+      "step": 220
+    },
+    {
+      "epoch": 0.1154639175257732,
+      "grad_norm": 2.0050344467163086,
+      "learning_rate": 9.998365639187638e-05,
+      "loss": 3.0542,
+      "step": 224
+    },
+    {
+      "epoch": 0.11752577319587629,
+      "grad_norm": 2.1245343685150146,
+      "learning_rate": 9.997900786866519e-05,
+      "loss": 3.0631,
+      "step": 228
+    },
+    {
+      "epoch": 0.11958762886597939,
+      "grad_norm": 2.0271108150482178,
+      "learning_rate": 9.997377845227576e-05,
+      "loss": 3.1621,
+      "step": 232
+    },
+    {
+      "epoch": 0.12164948453608247,
+      "grad_norm": 2.075441837310791,
+      "learning_rate": 9.996796820348822e-05,
+      "loss": 2.9906,
+      "step": 236
+    },
+    {
+      "epoch": 0.12371134020618557,
+      "grad_norm": 2.118668556213379,
+      "learning_rate": 9.996157718983362e-05,
+      "loss": 3.0718,
+      "step": 240
+    },
+    {
+      "epoch": 0.12577319587628866,
+      "grad_norm": 2.0540637969970703,
+      "learning_rate": 9.995460548559307e-05,
+      "loss": 3.077,
+      "step": 244
+    },
+    {
+      "epoch": 0.12783505154639174,
+      "grad_norm": 2.2028098106384277,
+      "learning_rate": 9.99470531717969e-05,
+      "loss": 3.1064,
+      "step": 248
+    },
+    {
+      "epoch": 0.12989690721649486,
+      "grad_norm": 2.0607352256774902,
+      "learning_rate": 9.993892033622374e-05,
+      "loss": 3.0582,
+      "step": 252
+    },
+    {
+      "epoch": 0.13195876288659794,
+      "grad_norm": 2.054838180541992,
+      "learning_rate": 9.993020707339939e-05,
+      "loss": 2.991,
+      "step": 256
+    },
+    {
+      "epoch": 0.13402061855670103,
+      "grad_norm": 1.9367263317108154,
+      "learning_rate": 9.992091348459591e-05,
+      "loss": 2.9583,
+      "step": 260
+    },
+    {
+      "epoch": 0.1360824742268041,
+      "grad_norm": 1.9889065027236938,
+      "learning_rate": 9.99110396778303e-05,
+      "loss": 2.9845,
+      "step": 264
+    },
+    {
+      "epoch": 0.13814432989690723,
+      "grad_norm": 2.249467134475708,
+      "learning_rate": 9.990058576786325e-05,
+      "loss": 3.0511,
+      "step": 268
+    },
+    {
+      "epoch": 0.1402061855670103,
+      "grad_norm": 2.011195659637451,
+      "learning_rate": 9.98895518761979e-05,
+      "loss": 2.9925,
+      "step": 272
+    },
+    {
+      "epoch": 0.1422680412371134,
+      "grad_norm": 2.0568485260009766,
+      "learning_rate": 9.987793813107833e-05,
+      "loss": 2.9491,
+      "step": 276
+    },
+    {
+      "epoch": 0.14432989690721648,
+      "grad_norm": 1.9748948812484741,
+      "learning_rate": 9.986574466748812e-05,
+      "loss": 3.0667,
+      "step": 280
+    },
+    {
+      "epoch": 0.1463917525773196,
+      "grad_norm": 2.234513282775879,
+      "learning_rate": 9.985297162714877e-05,
+      "loss": 2.8514,
+      "step": 284
+    },
+    {
+      "epoch": 0.14845360824742268,
+      "grad_norm": 1.9163928031921387,
+      "learning_rate": 9.983961915851804e-05,
+      "loss": 3.1231,
+      "step": 288
+    },
+    {
+      "epoch": 0.15051546391752577,
+      "grad_norm": 1.9212491512298584,
+      "learning_rate": 9.982568741678823e-05,
+      "loss": 2.9029,
+      "step": 292
+    },
+    {
+      "epoch": 0.15257731958762888,
+      "grad_norm": 2.065171241760254,
+      "learning_rate": 9.981117656388445e-05,
+      "loss": 2.9171,
+      "step": 296
+    },
+    {
+      "epoch": 0.15463917525773196,
+      "grad_norm": 1.979385495185852,
+      "learning_rate": 9.979608676846258e-05,
+      "loss": 3.0767,
+      "step": 300
+    },
+    {
+      "epoch": 0.15670103092783505,
+      "grad_norm": 1.9221059083938599,
+      "learning_rate": 9.978041820590743e-05,
+      "loss": 3.0627,
+      "step": 304
+    },
+    {
+      "epoch": 0.15876288659793814,
+      "grad_norm": 2.0437042713165283,
+      "learning_rate": 9.97641710583307e-05,
+      "loss": 2.938,
+      "step": 308
+    },
+    {
+      "epoch": 0.16082474226804125,
+      "grad_norm": 1.9110627174377441,
+      "learning_rate": 9.974734551456881e-05,
+      "loss": 2.9462,
+      "step": 312
+    },
+    {
+      "epoch": 0.16288659793814433,
+      "grad_norm": 1.8492796421051025,
+      "learning_rate": 9.972994177018074e-05,
+      "loss": 3.0006,
+      "step": 316
+    },
+    {
+      "epoch": 0.16494845360824742,
+      "grad_norm": 1.9782150983810425,
+      "learning_rate": 9.971196002744575e-05,
+      "loss": 2.9449,
+      "step": 320
+    },
+    {
+      "epoch": 0.1670103092783505,
+      "grad_norm": 1.8963398933410645,
+      "learning_rate": 9.969340049536099e-05,
+      "loss": 2.9932,
+      "step": 324
+    },
+    {
+      "epoch": 0.16907216494845362,
+      "grad_norm": 2.0604023933410645,
+      "learning_rate": 9.967426338963917e-05,
+      "loss": 2.8904,
+      "step": 328
+    },
+    {
+      "epoch": 0.1711340206185567,
+      "grad_norm": 2.0546135902404785,
+      "learning_rate": 9.965454893270592e-05,
+      "loss": 3.0603,
+      "step": 332
+    },
+    {
+      "epoch": 0.1731958762886598,
+      "grad_norm": 1.9946964979171753,
+      "learning_rate": 9.963425735369736e-05,
+      "loss": 2.917,
+      "step": 336
+    },
+    {
+      "epoch": 0.17525773195876287,
+      "grad_norm": 1.8940315246582031,
+      "learning_rate": 9.961338888845725e-05,
+      "loss": 2.9194,
+      "step": 340
+    },
+    {
+      "epoch": 0.177319587628866,
+      "grad_norm": 1.9344205856323242,
+      "learning_rate": 9.959194377953447e-05,
+      "loss": 2.9884,
+      "step": 344
+    },
+    {
+      "epoch": 0.17938144329896907,
+      "grad_norm": 1.9542304277420044,
+      "learning_rate": 9.956992227617995e-05,
+      "loss": 2.9051,
+      "step": 348
+    },
+    {
+      "epoch": 0.18144329896907216,
+      "grad_norm": 1.9588041305541992,
+      "learning_rate": 9.954732463434402e-05,
+      "loss": 2.7897,
+      "step": 352
+    },
+    {
+      "epoch": 0.18350515463917524,
+      "grad_norm": 1.7251778841018677,
+      "learning_rate": 9.952415111667324e-05,
+      "loss": 2.7511,
+      "step": 356
+    },
+    {
+      "epoch": 0.18556701030927836,
+      "grad_norm": 1.930525302886963,
+      "learning_rate": 9.950040199250746e-05,
+      "loss": 2.8544,
+      "step": 360
+    },
+    {
+      "epoch": 0.18762886597938144,
+      "grad_norm": 2.0370941162109375,
+      "learning_rate": 9.947607753787667e-05,
+      "loss": 3.0097,
+      "step": 364
+    },
+    {
+      "epoch": 0.18969072164948453,
+      "grad_norm": 2.0568158626556396,
+      "learning_rate": 9.945117803549774e-05,
+      "loss": 2.8412,
+      "step": 368
+    },
+    {
+      "epoch": 0.19175257731958764,
+      "grad_norm": 1.9520659446716309,
+      "learning_rate": 9.942570377477121e-05,
+      "loss": 2.8904,
+      "step": 372
+    },
+    {
+      "epoch": 0.19381443298969073,
+      "grad_norm": 1.9777973890304565,
+      "learning_rate": 9.939965505177786e-05,
+      "loss": 2.9093,
+      "step": 376
+    },
+    {
+      "epoch": 0.1958762886597938,
+      "grad_norm": 2.0709102153778076,
+      "learning_rate": 9.937303216927534e-05,
+      "loss": 2.7659,
+      "step": 380
+    },
+    {
+      "epoch": 0.1979381443298969,
+      "grad_norm": 1.919057846069336,
+      "learning_rate": 9.934583543669453e-05,
+      "loss": 2.8857,
+      "step": 384
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9447135925292969,
+      "learning_rate": 9.931806517013612e-05,
+      "loss": 2.7969,
+      "step": 388
+    },
+    {
+      "epoch": 0.2020618556701031,
+      "grad_norm": 1.9707754850387573,
+      "learning_rate": 9.928972169236676e-05,
+      "loss": 2.9706,
+      "step": 392
+    },
+    {
+      "epoch": 0.20412371134020618,
+      "grad_norm": 2.1095566749572754,
+      "learning_rate": 9.926080533281543e-05,
+      "loss": 2.8079,
+      "step": 396
+    },
+    {
+      "epoch": 0.20618556701030927,
+      "grad_norm": 1.9369230270385742,
+      "learning_rate": 9.923131642756954e-05,
+      "loss": 2.759,
+      "step": 400
+    },
+    {
+      "epoch": 0.20824742268041238,
+      "grad_norm": 1.8451541662216187,
+      "learning_rate": 9.920125531937107e-05,
+      "loss": 2.8428,
+      "step": 404
+    },
+    {
+      "epoch": 0.21030927835051547,
+      "grad_norm": 1.8733794689178467,
+      "learning_rate": 9.917062235761259e-05,
+      "loss": 2.7717,
+      "step": 408
+    },
+    {
+      "epoch": 0.21237113402061855,
+      "grad_norm": 1.8459867238998413,
+      "learning_rate": 9.913941789833311e-05,
+      "loss": 2.8008,
+      "step": 412
+    },
+    {
+      "epoch": 0.21443298969072164,
+      "grad_norm": 1.8867549896240234,
+      "learning_rate": 9.910764230421406e-05,
+      "loss": 2.7071,
+      "step": 416
+    },
+    {
+      "epoch": 0.21649484536082475,
+      "grad_norm": 1.8722797632217407,
+      "learning_rate": 9.907529594457504e-05,
+      "loss": 2.8076,
+      "step": 420
+    },
+    {
+      "epoch": 0.21855670103092784,
+      "grad_norm": 1.9473061561584473,
+      "learning_rate": 9.904237919536945e-05,
+      "loss": 2.8613,
+      "step": 424
+    },
+    {
+      "epoch": 0.22061855670103092,
+      "grad_norm": 1.8926331996917725,
+      "learning_rate": 9.900889243918024e-05,
+      "loss": 2.7264,
+      "step": 428
+    },
+    {
+      "epoch": 0.22268041237113403,
+      "grad_norm": 2.013500928878784,
+      "learning_rate": 9.897483606521536e-05,
+      "loss": 2.7307,
+      "step": 432
+    },
+    {
+      "epoch": 0.22474226804123712,
+      "grad_norm": 1.7798408269882202,
+      "learning_rate": 9.894021046930333e-05,
+      "loss": 2.7089,
+      "step": 436
+    },
+    {
+      "epoch": 0.2268041237113402,
+      "grad_norm": 1.8650320768356323,
+      "learning_rate": 9.890501605388853e-05,
+      "loss": 2.812,
+      "step": 440
+    },
+    {
+      "epoch": 0.2288659793814433,
+      "grad_norm": 1.8471648693084717,
+      "learning_rate": 9.886925322802663e-05,
+      "loss": 2.8819,
+      "step": 444
+    },
+    {
+      "epoch": 0.2309278350515464,
+      "grad_norm": 1.7936201095581055,
+      "learning_rate": 9.883292240737978e-05,
+      "loss": 2.8319,
+      "step": 448
+    },
+    {
+      "epoch": 0.2329896907216495,
+      "grad_norm": 1.89383065700531,
+      "learning_rate": 9.87960240142118e-05,
+      "loss": 2.7643,
+      "step": 452
+    },
+    {
+      "epoch": 0.23505154639175257,
+      "grad_norm": 1.7932506799697876,
+      "learning_rate": 9.875855847738319e-05,
+      "loss": 2.7604,
+      "step": 456
+    },
+    {
+      "epoch": 0.23711340206185566,
+      "grad_norm": 1.8939586877822876,
+      "learning_rate": 9.872052623234632e-05,
+      "loss": 2.6743,
+      "step": 460
+    },
+    {
+      "epoch": 0.23917525773195877,
+      "grad_norm": 1.8440996408462524,
+      "learning_rate": 9.868192772114016e-05,
+      "loss": 2.6986,
+      "step": 464
+    },
+    {
+      "epoch": 0.24123711340206186,
+      "grad_norm": 2.047183036804199,
+      "learning_rate": 9.864276339238534e-05,
+      "loss": 2.7008,
+      "step": 468
+    },
+    {
+      "epoch": 0.24329896907216494,
+      "grad_norm": 1.7253282070159912,
+      "learning_rate": 9.860303370127876e-05,
+      "loss": 2.7944,
+      "step": 472
+    },
+    {
+      "epoch": 0.24536082474226803,
+      "grad_norm": 1.8298146724700928,
+      "learning_rate": 9.856273910958847e-05,
+      "loss": 2.637,
+      "step": 476
+    },
+    {
+      "epoch": 0.24742268041237114,
+      "grad_norm": 1.7853765487670898,
+      "learning_rate": 9.852188008564813e-05,
+      "loss": 2.7162,
+      "step": 480
+    },
+    {
+      "epoch": 0.24948453608247423,
+      "grad_norm": 1.8834904432296753,
+      "learning_rate": 9.84804571043517e-05,
+      "loss": 2.8762,
+      "step": 484
+    },
+    {
+      "epoch": 0.2515463917525773,
+      "grad_norm": 1.9652595520019531,
+      "learning_rate": 9.843847064714785e-05,
+      "loss": 2.7794,
+      "step": 488
+    },
+    {
+      "epoch": 0.2536082474226804,
+      "grad_norm": 1.983886957168579,
+      "learning_rate": 9.839592120203441e-05,
+      "loss": 2.8331,
+      "step": 492
+    },
+    {
+      "epoch": 0.2556701030927835,
+      "grad_norm": 1.9896190166473389,
+      "learning_rate": 9.835280926355261e-05,
+      "loss": 2.8015,
+      "step": 496
+    },
+    {
+      "epoch": 0.25773195876288657,
+      "grad_norm": 1.70455002784729,
+      "learning_rate": 9.83091353327815e-05,
+      "loss": 2.7135,
+      "step": 500
+    },
+    {
+      "epoch": 0.2597938144329897,
+      "grad_norm": 2.0131478309631348,
+      "learning_rate": 9.826489991733194e-05,
+      "loss": 2.7678,
+      "step": 504
+    },
+    {
+      "epoch": 0.2618556701030928,
+      "grad_norm": 2.189298152923584,
+      "learning_rate": 9.822010353134081e-05,
+      "loss": 2.8352,
+      "step": 508
+    },
+    {
+      "epoch": 0.2639175257731959,
+      "grad_norm": 2.022023916244507,
+      "learning_rate": 9.817474669546501e-05,
+      "loss": 2.7118,
+      "step": 512
+    },
+    {
+      "epoch": 0.26597938144329897,
+      "grad_norm": 2.1792008876800537,
+      "learning_rate": 9.812882993687539e-05,
+      "loss": 2.8367,
+      "step": 516
+    },
+    {
+      "epoch": 0.26804123711340205,
+      "grad_norm": 1.8317099809646606,
+      "learning_rate": 9.808235378925066e-05,
+      "loss": 2.7326,
+      "step": 520
+    },
+    {
+      "epoch": 0.27010309278350514,
+      "grad_norm": 1.7824715375900269,
+      "learning_rate": 9.803531879277113e-05,
+      "loss": 2.7147,
+      "step": 524
+    },
+    {
+      "epoch": 0.2721649484536082,
+      "grad_norm": 1.7565789222717285,
+      "learning_rate": 9.798772549411252e-05,
+      "loss": 2.8029,
+      "step": 528
+    },
+    {
+      "epoch": 0.27422680412371137,
+      "grad_norm": 1.805213212966919,
+      "learning_rate": 9.793957444643951e-05,
+      "loss": 2.6505,
+      "step": 532
+    },
+    {
+      "epoch": 0.27628865979381445,
+      "grad_norm": 1.8401093482971191,
+      "learning_rate": 9.789086620939936e-05,
+      "loss": 2.7262,
+      "step": 536
+    },
+    {
+      "epoch": 0.27835051546391754,
+      "grad_norm": 1.7650929689407349,
+      "learning_rate": 9.784160134911541e-05,
+      "loss": 2.6988,
+      "step": 540
+    },
+    {
+      "epoch": 0.2804123711340206,
+      "grad_norm": 1.902664065361023,
+      "learning_rate": 9.77917804381805e-05,
+      "loss": 2.7028,
+      "step": 544
+    },
+    {
+      "epoch": 0.2824742268041237,
+      "grad_norm": 1.6840468645095825,
+      "learning_rate": 9.774140405565024e-05,
+      "loss": 2.7141,
+      "step": 548
+    },
+    {
+      "epoch": 0.2845360824742268,
+      "grad_norm": 1.8542104959487915,
+      "learning_rate": 9.769047278703644e-05,
+      "loss": 2.7512,
+      "step": 552
+    },
+    {
+      "epoch": 0.2865979381443299,
+      "grad_norm": 1.9419035911560059,
+      "learning_rate": 9.763898722430015e-05,
+      "loss": 2.7085,
+      "step": 556
+    },
+    {
+      "epoch": 0.28865979381443296,
+      "grad_norm": 1.8914318084716797,
+      "learning_rate": 9.758694796584483e-05,
+      "loss": 2.7119,
+      "step": 560
+    },
+    {
+      "epoch": 0.2907216494845361,
+      "grad_norm": 2.028513193130493,
+      "learning_rate": 9.753435561650946e-05,
+      "loss": 2.7293,
+      "step": 564
+    },
+    {
+      "epoch": 0.2927835051546392,
+      "grad_norm": 1.7596354484558105,
+      "learning_rate": 9.748121078756137e-05,
+      "loss": 2.6624,
+      "step": 568
+    },
+    {
+      "epoch": 0.2948453608247423,
+      "grad_norm": 1.955361247062683,
+      "learning_rate": 9.742751409668929e-05,
+      "loss": 2.6958,
+      "step": 572
+    },
+    {
+      "epoch": 0.29690721649484536,
+      "grad_norm": 1.980281949043274,
+      "learning_rate": 9.737326616799605e-05,
+      "loss": 2.6921,
+      "step": 576
+    },
+    {
+      "epoch": 0.29896907216494845,
+      "grad_norm": 2.0537736415863037,
+      "learning_rate": 9.731846763199144e-05,
+      "loss": 2.7734,
+      "step": 580
+    },
+    {
+      "epoch": 0.30103092783505153,
+      "grad_norm": 2.0544698238372803,
+      "learning_rate": 9.726311912558474e-05,
+      "loss": 2.622,
+      "step": 584
+    },
+    {
+      "epoch": 0.3030927835051546,
+      "grad_norm": 1.8842084407806396,
+      "learning_rate": 9.720722129207746e-05,
+      "loss": 2.6809,
+      "step": 588
+    },
+    {
+      "epoch": 0.30515463917525776,
+      "grad_norm": 1.9250359535217285,
+      "learning_rate": 9.715077478115574e-05,
+      "loss": 2.7359,
+      "step": 592
+    },
+    {
+      "epoch": 0.30721649484536084,
+      "grad_norm": 1.7907671928405762,
+      "learning_rate": 9.709378024888292e-05,
+      "loss": 2.7575,
+      "step": 596
+    },
+    {
+      "epoch": 0.30927835051546393,
+      "grad_norm": 1.8203668594360352,
+      "learning_rate": 9.703623835769178e-05,
+      "loss": 2.7181,
+      "step": 600
+    },
+    {
+      "epoch": 0.311340206185567,
+      "grad_norm": 1.837925672531128,
+      "learning_rate": 9.697814977637696e-05,
+      "loss": 2.743,
+      "step": 604
+    },
+    {
+      "epoch": 0.3134020618556701,
+      "grad_norm": 1.7200803756713867,
+      "learning_rate": 9.691951518008715e-05,
+      "loss": 2.6844,
+      "step": 608
+    },
+    {
+      "epoch": 0.3154639175257732,
+      "grad_norm": 1.9397820234298706,
+      "learning_rate": 9.686033525031719e-05,
+      "loss": 2.5767,
+      "step": 612
+    },
+    {
+      "epoch": 0.31752577319587627,
+      "grad_norm": 1.9038435220718384,
+      "learning_rate": 9.680061067490021e-05,
+      "loss": 2.6503,
+      "step": 616
+    },
+    {
+      "epoch": 0.31958762886597936,
+      "grad_norm": 1.9289252758026123,
+      "learning_rate": 9.674034214799964e-05,
+      "loss": 2.6878,
+      "step": 620
+    },
+    {
+      "epoch": 0.3216494845360825,
+      "grad_norm": 1.7512463331222534,
+      "learning_rate": 9.667953037010108e-05,
+      "loss": 2.7551,
+      "step": 624
+    },
+    {
+      "epoch": 0.3237113402061856,
+      "grad_norm": 1.9182549715042114,
+      "learning_rate": 9.661817604800421e-05,
+      "loss": 2.7986,
+      "step": 628
+    },
+    {
+      "epoch": 0.32577319587628867,
+      "grad_norm": 1.8200740814208984,
+      "learning_rate": 9.655627989481458e-05,
+      "loss": 2.6923,
+      "step": 632
+    },
+    {
+      "epoch": 0.32783505154639175,
+      "grad_norm": 1.8738125562667847,
+      "learning_rate": 9.649384262993525e-05,
+      "loss": 2.7518,
+      "step": 636
+    },
+    {
+      "epoch": 0.32989690721649484,
+      "grad_norm": 1.9261345863342285,
+      "learning_rate": 9.64308649790586e-05,
+      "loss": 2.6037,
+      "step": 640
+    },
+    {
+      "epoch": 0.3319587628865979,
+      "grad_norm": 1.7359293699264526,
+      "learning_rate": 9.636734767415763e-05,
+      "loss": 2.7102,
+      "step": 644
+    },
+    {
+      "epoch": 0.334020618556701,
+      "grad_norm": 2.001077175140381,
+      "learning_rate": 9.630329145347767e-05,
+      "loss": 2.7068,
+      "step": 648
+    },
+    {
+      "epoch": 0.33608247422680415,
+      "grad_norm": 1.7505935430526733,
+      "learning_rate": 9.623869706152777e-05,
+      "loss": 2.6348,
+      "step": 652
+    },
+    {
+      "epoch": 0.33814432989690724,
+      "grad_norm": 1.9153965711593628,
+      "learning_rate": 9.617356524907193e-05,
+      "loss": 2.6874,
+      "step": 656
+    },
+    {
+      "epoch": 0.3402061855670103,
+      "grad_norm": 1.805038332939148,
+      "learning_rate": 9.61078967731205e-05,
+      "loss": 2.5985,
+      "step": 660
+    },
+    {
+      "epoch": 0.3422680412371134,
+      "grad_norm": 1.7926819324493408,
+      "learning_rate": 9.604169239692133e-05,
+      "loss": 2.7203,
+      "step": 664
+    },
+    {
+      "epoch": 0.3443298969072165,
+      "grad_norm": 1.8701540231704712,
+      "learning_rate": 9.597495288995089e-05,
+      "loss": 2.6819,
+      "step": 668
+    },
+    {
+      "epoch": 0.3463917525773196,
+      "grad_norm": 2.0685527324676514,
+      "learning_rate": 9.590767902790529e-05,
+      "loss": 2.7401,
+      "step": 672
+    },
+    {
+      "epoch": 0.34845360824742266,
+      "grad_norm": 1.7549872398376465,
+      "learning_rate": 9.583987159269143e-05,
+      "loss": 2.5833,
+      "step": 676
+    },
+    {
+      "epoch": 0.35051546391752575,
+      "grad_norm": 1.8161414861679077,
+      "learning_rate": 9.577153137241765e-05,
+      "loss": 2.7274,
+      "step": 680
+    },
+    {
+      "epoch": 0.3525773195876289,
+      "grad_norm": 1.7566990852355957,
+      "learning_rate": 9.570265916138484e-05,
+      "loss": 2.7536,
+      "step": 684
+    },
+    {
+      "epoch": 0.354639175257732,
+      "grad_norm": 1.8565185070037842,
+      "learning_rate": 9.563325576007701e-05,
+      "loss": 2.6785,
+      "step": 688
+    },
+    {
+      "epoch": 0.35670103092783506,
+      "grad_norm": 1.7326369285583496,
+      "learning_rate": 9.556332197515207e-05,
+      "loss": 2.7096,
+      "step": 692
+    },
+    {
+      "epoch": 0.35876288659793815,
+      "grad_norm": 1.705234408378601,
+      "learning_rate": 9.549285861943247e-05,
+      "loss": 2.7314,
+      "step": 696
+    },
+    {
+      "epoch": 0.36082474226804123,
+      "grad_norm": 1.9541505575180054,
+      "learning_rate": 9.542186651189569e-05,
+      "loss": 2.6204,
+      "step": 700
+    },
+    {
+      "epoch": 0.3628865979381443,
+      "grad_norm": 1.8346929550170898,
+      "learning_rate": 9.535034647766476e-05,
+      "loss": 2.6913,
+      "step": 704
+    },
+    {
+      "epoch": 0.3649484536082474,
+      "grad_norm": 1.9132497310638428,
+      "learning_rate": 9.527829934799869e-05,
+      "loss": 2.7041,
+      "step": 708
+    },
+    {
+      "epoch": 0.3670103092783505,
+      "grad_norm": 1.763213038444519,
+      "learning_rate": 9.520572596028278e-05,
+      "loss": 2.6216,
+      "step": 712
+    },
+    {
+      "epoch": 0.36907216494845363,
+      "grad_norm": 1.9224257469177246,
+      "learning_rate": 9.513262715801887e-05,
+      "loss": 2.6632,
+      "step": 716
+    },
+    {
+      "epoch": 0.3711340206185567,
+      "grad_norm": 1.8817148208618164,
+      "learning_rate": 9.505900379081559e-05,
+      "loss": 2.6868,
+      "step": 720
+    },
+    {
+      "epoch": 0.3731958762886598,
+      "grad_norm": 1.7395341396331787,
+      "learning_rate": 9.498485671437842e-05,
+      "loss": 2.5171,
+      "step": 724
+    },
+    {
+      "epoch": 0.3752577319587629,
+      "grad_norm": 1.7571380138397217,
+      "learning_rate": 9.491018679049981e-05,
+      "loss": 2.6194,
+      "step": 728
+    },
+    {
+      "epoch": 0.37731958762886597,
+      "grad_norm": 1.8208237886428833,
+      "learning_rate": 9.48349948870491e-05,
+      "loss": 2.6101,
+      "step": 732
+    },
+    {
+      "epoch": 0.37938144329896906,
+      "grad_norm": 1.888909935951233,
+      "learning_rate": 9.47592818779625e-05,
+      "loss": 2.6569,
+      "step": 736
+    },
+    {
+      "epoch": 0.38144329896907214,
+      "grad_norm": 1.7806968688964844,
+      "learning_rate": 9.468304864323288e-05,
+      "loss": 2.5329,
+      "step": 740
+    },
+    {
+      "epoch": 0.3835051546391753,
+      "grad_norm": 1.8010936975479126,
+      "learning_rate": 9.460629606889952e-05,
+      "loss": 2.6971,
+      "step": 744
+    },
+    {
+      "epoch": 0.38556701030927837,
+      "grad_norm": 1.7562525272369385,
+      "learning_rate": 9.452902504703793e-05,
+      "loss": 2.676,
+      "step": 748
+    },
+    {
+      "epoch": 0.38762886597938145,
+      "grad_norm": 1.828466534614563,
+      "learning_rate": 9.445123647574936e-05,
+      "loss": 2.5952,
+      "step": 752
+    },
+    {
+      "epoch": 0.38969072164948454,
+      "grad_norm": 1.7644858360290527,
+      "learning_rate": 9.437293125915037e-05,
+      "loss": 2.6475,
+      "step": 756
+    },
+    {
+      "epoch": 0.3917525773195876,
+      "grad_norm": 1.6723216772079468,
+      "learning_rate": 9.429411030736242e-05,
+      "loss": 2.6823,
+      "step": 760
+    },
+    {
+      "epoch": 0.3938144329896907,
+      "grad_norm": 1.7757784128189087,
+      "learning_rate": 9.421477453650118e-05,
+      "loss": 2.5711,
+      "step": 764
+    },
+    {
+      "epoch": 0.3958762886597938,
+      "grad_norm": 1.647575855255127,
+      "learning_rate": 9.413492486866598e-05,
+      "loss": 2.6141,
+      "step": 768
+    },
+    {
+      "epoch": 0.3979381443298969,
+      "grad_norm": 1.9347645044326782,
+      "learning_rate": 9.405456223192897e-05,
+      "loss": 2.6708,
+      "step": 772
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.838582158088684,
+      "learning_rate": 9.397368756032445e-05,
+      "loss": 2.7496,
+      "step": 776
+    },
+    {
+      "epoch": 0.4020618556701031,
+      "grad_norm": 1.8382269144058228,
+      "learning_rate": 9.389230179383801e-05,
+      "loss": 2.5654,
+      "step": 780
+    },
+    {
+      "epoch": 0.4041237113402062,
+      "grad_norm": 1.8177680969238281,
+      "learning_rate": 9.381040587839548e-05,
+      "loss": 2.6553,
+      "step": 784
+    },
+    {
+      "epoch": 0.4061855670103093,
+      "grad_norm": 1.7643057107925415,
+      "learning_rate": 9.372800076585207e-05,
+      "loss": 2.5951,
+      "step": 788
+    },
+    {
+      "epoch": 0.40824742268041236,
+      "grad_norm": 1.8181430101394653,
+      "learning_rate": 9.364508741398127e-05,
+      "loss": 2.6452,
+      "step": 792
+    },
+    {
+      "epoch": 0.41030927835051545,
+      "grad_norm": 1.8445159196853638,
+      "learning_rate": 9.356166678646366e-05,
+      "loss": 2.6172,
+      "step": 796
+    },
+    {
+      "epoch": 0.41237113402061853,
+      "grad_norm": 1.6617608070373535,
+      "learning_rate": 9.347773985287578e-05,
+      "loss": 2.5781,
+      "step": 800
+    },
+    {
+      "epoch": 0.4144329896907217,
+      "grad_norm": 1.9522396326065063,
+      "learning_rate": 9.339330758867883e-05,
+      "loss": 2.6309,
+      "step": 804
+    },
+    {
+      "epoch": 0.41649484536082476,
+      "grad_norm": 2.5656485557556152,
+      "learning_rate": 9.330837097520738e-05,
+      "loss": 2.6967,
+      "step": 808
+    },
+    {
+      "epoch": 0.41855670103092785,
+      "grad_norm": 1.6885048151016235,
+      "learning_rate": 9.322293099965784e-05,
+      "loss": 2.617,
+      "step": 812
+    },
+    {
+      "epoch": 0.42061855670103093,
+      "grad_norm": 1.7637752294540405,
+      "learning_rate": 9.313698865507713e-05,
+      "loss": 2.6538,
+      "step": 816
+    },
+    {
+      "epoch": 0.422680412371134,
+      "grad_norm": 1.7737189531326294,
+      "learning_rate": 9.305054494035106e-05,
+      "loss": 2.6113,
+      "step": 820
+    },
+    {
+      "epoch": 0.4247422680412371,
+      "grad_norm": 1.8070309162139893,
+      "learning_rate": 9.296360086019272e-05,
+      "loss": 2.5118,
+      "step": 824
+    },
+    {
+      "epoch": 0.4268041237113402,
+      "grad_norm": 1.73598313331604,
+      "learning_rate": 9.287615742513086e-05,
+      "loss": 2.5826,
+      "step": 828
+    },
+    {
+      "epoch": 0.4288659793814433,
+      "grad_norm": 1.688819169998169,
+      "learning_rate": 9.278821565149806e-05,
+      "loss": 2.5375,
+      "step": 832
+    },
+    {
+      "epoch": 0.4309278350515464,
+      "grad_norm": 1.7314485311508179,
+      "learning_rate": 9.269977656141898e-05,
+      "loss": 2.5819,
+      "step": 836
+    },
+    {
+      "epoch": 0.4329896907216495,
+      "grad_norm": 1.657240390777588,
+      "learning_rate": 9.261084118279847e-05,
+      "loss": 2.6643,
+      "step": 840
+    },
+    {
+      "epoch": 0.4350515463917526,
+      "grad_norm": 1.7966091632843018,
+      "learning_rate": 9.25214105493096e-05,
+      "loss": 2.7022,
+      "step": 844
+    },
+    {
+      "epoch": 0.43711340206185567,
+      "grad_norm": 1.732354760169983,
+      "learning_rate": 9.243148570038164e-05,
+      "loss": 2.6098,
+      "step": 848
+    },
+    {
+      "epoch": 0.43917525773195876,
+      "grad_norm": 1.737281322479248,
+      "learning_rate": 9.234106768118809e-05,
+      "loss": 2.5879,
+      "step": 852
+    },
+    {
+      "epoch": 0.44123711340206184,
+      "grad_norm": 1.7843495607376099,
+      "learning_rate": 9.225015754263431e-05,
+      "loss": 2.5793,
+      "step": 856
+    },
+    {
+      "epoch": 0.44329896907216493,
+      "grad_norm": 1.7821077108383179,
+      "learning_rate": 9.215875634134552e-05,
+      "loss": 2.6578,
+      "step": 860
+    },
+    {
+      "epoch": 0.44536082474226807,
+      "grad_norm": 1.8063935041427612,
+      "learning_rate": 9.206686513965445e-05,
+      "loss": 2.7075,
+      "step": 864
+    },
+    {
+      "epoch": 0.44742268041237115,
+      "grad_norm": 1.5605168342590332,
+      "learning_rate": 9.19744850055889e-05,
+      "loss": 2.5988,
+      "step": 868
+    },
+    {
+      "epoch": 0.44948453608247424,
+      "grad_norm": 2.078798294067383,
+      "learning_rate": 9.188161701285949e-05,
+      "loss": 2.6126,
+      "step": 872
+    },
+    {
+      "epoch": 0.4515463917525773,
+      "grad_norm": 1.7260762453079224,
+      "learning_rate": 9.178826224084705e-05,
+      "loss": 2.6352,
+      "step": 876
+    },
+    {
+      "epoch": 0.4536082474226804,
+      "grad_norm": 1.7088130712509155,
+      "learning_rate": 9.169442177459011e-05,
+      "loss": 2.6324,
+      "step": 880
+    },
+    {
+      "epoch": 0.4556701030927835,
+      "grad_norm": 1.8762470483779907,
+      "learning_rate": 9.160009670477234e-05,
+      "loss": 2.517,
+      "step": 884
+    },
+    {
+      "epoch": 0.4577319587628866,
+      "grad_norm": 1.9028972387313843,
+      "learning_rate": 9.150528812770981e-05,
+      "loss": 2.5386,
+      "step": 888
+    },
+    {
+      "epoch": 0.45979381443298967,
+      "grad_norm": 1.5713077783584595,
+      "learning_rate": 9.140999714533827e-05,
+      "loss": 2.6347,
+      "step": 892
+    },
+    {
+      "epoch": 0.4618556701030928,
+      "grad_norm": 1.603164792060852,
+      "learning_rate": 9.131422486520034e-05,
+      "loss": 2.5138,
+      "step": 896
+    },
+    {
+      "epoch": 0.4639175257731959,
+      "grad_norm": 1.7028288841247559,
+      "learning_rate": 9.121797240043267e-05,
+      "loss": 2.5366,
+      "step": 900
+    },
+    {
+      "epoch": 0.465979381443299,
+      "grad_norm": 1.7099529504776,
+      "learning_rate": 9.11212408697529e-05,
+      "loss": 2.6465,
+      "step": 904
+    },
+    {
+      "epoch": 0.46804123711340206,
+      "grad_norm": 1.7134875059127808,
+      "learning_rate": 9.102403139744683e-05,
+      "loss": 2.6331,
+      "step": 908
+    },
+    {
+      "epoch": 0.47010309278350515,
+      "grad_norm": 1.6669530868530273,
+      "learning_rate": 9.092634511335519e-05,
+      "loss": 2.5381,
+      "step": 912
+    },
+    {
+      "epoch": 0.47216494845360824,
+      "grad_norm": 1.7473549842834473,
+      "learning_rate": 9.082818315286055e-05,
+      "loss": 2.584,
+      "step": 916
+    },
+    {
+      "epoch": 0.4742268041237113,
+      "grad_norm": 1.673309087753296,
+      "learning_rate": 9.07295466568742e-05,
+      "loss": 2.4649,
+      "step": 920
+    },
+    {
+      "epoch": 0.4762886597938144,
+      "grad_norm": 1.8044261932373047,
+      "learning_rate": 9.063043677182283e-05,
+      "loss": 2.6512,
+      "step": 924
+    },
+    {
+      "epoch": 0.47835051546391755,
+      "grad_norm": 1.7258899211883545,
+      "learning_rate": 9.053085464963518e-05,
+      "loss": 2.5977,
+      "step": 928
+    },
+    {
+      "epoch": 0.48041237113402063,
+      "grad_norm": 1.7187895774841309,
+      "learning_rate": 9.043080144772868e-05,
+      "loss": 2.5948,
+      "step": 932
+    },
+    {
+      "epoch": 0.4824742268041237,
+      "grad_norm": 1.797176718711853,
+      "learning_rate": 9.033027832899601e-05,
+      "loss": 2.659,
+      "step": 936
+    },
+    {
+      "epoch": 0.4845360824742268,
+      "grad_norm": 1.6816799640655518,
+      "learning_rate": 9.022928646179159e-05,
+      "loss": 2.6123,
+      "step": 940
+    },
+    {
+      "epoch": 0.4865979381443299,
+      "grad_norm": 1.7294647693634033,
+      "learning_rate": 9.012782701991795e-05,
+      "loss": 2.5824,
+      "step": 944
+    },
+    {
+      "epoch": 0.488659793814433,
+      "grad_norm": 1.6819491386413574,
+      "learning_rate": 9.002590118261216e-05,
+      "loss": 2.5717,
+      "step": 948
+    },
+    {
+      "epoch": 0.49072164948453606,
+      "grad_norm": 1.667133092880249,
+      "learning_rate": 8.992351013453204e-05,
+      "loss": 2.6282,
+      "step": 952
+    },
+    {
+      "epoch": 0.4927835051546392,
+      "grad_norm": 1.6454589366912842,
+      "learning_rate": 8.982065506574247e-05,
+      "loss": 2.5939,
+      "step": 956
+    },
+    {
+      "epoch": 0.4948453608247423,
+      "grad_norm": 1.8332362174987793,
+      "learning_rate": 8.971733717170148e-05,
+      "loss": 2.4949,
+      "step": 960
+    },
+    {
+      "epoch": 0.49690721649484537,
+      "grad_norm": 6.3988037109375,
+      "learning_rate": 8.961355765324648e-05,
+      "loss": 2.6143,
+      "step": 964
+    },
+    {
+      "epoch": 0.49896907216494846,
+      "grad_norm": 1.6663529872894287,
+      "learning_rate": 8.950931771658014e-05,
+      "loss": 2.5308,
+      "step": 968
+    },
+    {
+      "epoch": 0.5010309278350515,
+      "grad_norm": 1.6272313594818115,
+      "learning_rate": 8.940461857325647e-05,
+      "loss": 2.4458,
+      "step": 972
+    },
+    {
+      "epoch": 0.5030927835051546,
+      "grad_norm": 1.758596658706665,
+      "learning_rate": 8.929946144016677e-05,
+      "loss": 2.517,
+      "step": 976
+    },
+    {
+      "epoch": 0.5051546391752577,
+      "grad_norm": 1.6491873264312744,
+      "learning_rate": 8.919384753952538e-05,
+      "loss": 2.5408,
+      "step": 980
+    },
+    {
+      "epoch": 0.5072164948453608,
+      "grad_norm": 1.5984055995941162,
+      "learning_rate": 8.908777809885557e-05,
+      "loss": 2.5359,
+      "step": 984
+    },
+    {
+      "epoch": 0.5092783505154639,
+      "grad_norm": 1.651795744895935,
+      "learning_rate": 8.898125435097521e-05,
+      "loss": 2.5151,
+      "step": 988
+    },
+    {
+      "epoch": 0.511340206185567,
+      "grad_norm": 1.6190180778503418,
+      "learning_rate": 8.887427753398248e-05,
+      "loss": 2.6733,
+      "step": 992
+    },
+    {
+      "epoch": 0.51340206185567,
+      "grad_norm": 1.6881654262542725,
+      "learning_rate": 8.876684889124145e-05,
+      "loss": 2.5783,
+      "step": 996
+    },
+    {
+      "epoch": 0.5154639175257731,
+      "grad_norm": 1.786693811416626,
+      "learning_rate": 8.865896967136766e-05,
+      "loss": 2.6061,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5175257731958763,
+      "grad_norm": 1.841579556465149,
+      "learning_rate": 8.855064112821361e-05,
+      "loss": 2.6615,
+      "step": 1004
+    },
+    {
+      "epoch": 0.5195876288659794,
+      "grad_norm": 1.5715938806533813,
+      "learning_rate": 8.844186452085412e-05,
+      "loss": 2.6044,
+      "step": 1008
+    },
+    {
+      "epoch": 0.5216494845360825,
+      "grad_norm": 1.616853952407837,
+      "learning_rate": 8.83326411135718e-05,
+      "loss": 2.538,
+      "step": 1012
+    },
+    {
+      "epoch": 0.5237113402061856,
+      "grad_norm": 1.7380198240280151,
+      "learning_rate": 8.822297217584225e-05,
+      "loss": 2.5838,
+      "step": 1016
+    },
+    {
+      "epoch": 0.5257731958762887,
+      "grad_norm": 1.703635334968567,
+      "learning_rate": 8.81128589823194e-05,
+      "loss": 2.5035,
+      "step": 1020
+    },
+    {
+      "epoch": 0.5278350515463918,
+      "grad_norm": 1.9964070320129395,
+      "learning_rate": 8.80023028128206e-05,
+      "loss": 2.544,
+      "step": 1024
+    },
+    {
+      "epoch": 0.5298969072164949,
+      "grad_norm": 1.5779422521591187,
+      "learning_rate": 8.789130495231186e-05,
+      "loss": 2.6569,
+      "step": 1028
+    },
+    {
+      "epoch": 0.5319587628865979,
+      "grad_norm": 1.771633505821228,
+      "learning_rate": 8.77798666908928e-05,
+      "loss": 2.5315,
+      "step": 1032
+    },
+    {
+      "epoch": 0.534020618556701,
+      "grad_norm": 1.7763731479644775,
+      "learning_rate": 8.766798932378172e-05,
+      "loss": 2.5138,
+      "step": 1036
+    },
+    {
+      "epoch": 0.5360824742268041,
+      "grad_norm": 1.699804425239563,
+      "learning_rate": 8.755567415130058e-05,
+      "loss": 2.5168,
+      "step": 1040
+    },
+    {
+      "epoch": 0.5381443298969072,
+      "grad_norm": 1.8998210430145264,
+      "learning_rate": 8.744292247885975e-05,
+      "loss": 2.4795,
+      "step": 1044
+    },
+    {
+      "epoch": 0.5402061855670103,
+      "grad_norm": 1.7309194803237915,
+      "learning_rate": 8.732973561694297e-05,
+      "loss": 2.5866,
+      "step": 1048
+    },
+    {
+      "epoch": 0.5422680412371134,
+      "grad_norm": 1.667391061782837,
+      "learning_rate": 8.721611488109212e-05,
+      "loss": 2.5016,
+      "step": 1052
+    },
+    {
+      "epoch": 0.5443298969072164,
+      "grad_norm": 1.7690322399139404,
+      "learning_rate": 8.710206159189182e-05,
+      "loss": 2.6189,
+      "step": 1056
+    },
+    {
+      "epoch": 0.5463917525773195,
+      "grad_norm": 1.6563546657562256,
+      "learning_rate": 8.698757707495417e-05,
+      "loss": 2.4559,
+      "step": 1060
+    },
+    {
+      "epoch": 0.5484536082474227,
+      "grad_norm": 1.8325282335281372,
+      "learning_rate": 8.687266266090333e-05,
+      "loss": 2.546,
+      "step": 1064
+    },
+    {
+      "epoch": 0.5505154639175258,
+      "grad_norm": 1.6227245330810547,
+      "learning_rate": 8.675731968536002e-05,
+      "loss": 2.5709,
+      "step": 1068
+    },
+    {
+      "epoch": 0.5525773195876289,
+      "grad_norm": 1.6123030185699463,
+      "learning_rate": 8.664154948892607e-05,
+      "loss": 2.5047,
+      "step": 1072
+    },
+    {
+      "epoch": 0.554639175257732,
+      "grad_norm": 1.631011724472046,
+      "learning_rate": 8.65253534171687e-05,
+      "loss": 2.4568,
+      "step": 1076
+    },
+    {
+      "epoch": 0.5567010309278351,
+      "grad_norm": 1.746801495552063,
+      "learning_rate": 8.640873282060506e-05,
+      "loss": 2.5825,
+      "step": 1080
+    },
+    {
+      "epoch": 0.5587628865979382,
+      "grad_norm": 1.727461338043213,
+      "learning_rate": 8.629168905468641e-05,
+      "loss": 2.5724,
+      "step": 1084
+    },
+    {
+      "epoch": 0.5608247422680412,
+      "grad_norm": 1.7059168815612793,
+      "learning_rate": 8.617422347978239e-05,
+      "loss": 2.5469,
+      "step": 1088
+    },
+    {
+      "epoch": 0.5628865979381443,
+      "grad_norm": 1.6282740831375122,
+      "learning_rate": 8.605633746116519e-05,
+      "loss": 2.5503,
+      "step": 1092
+    },
+    {
+      "epoch": 0.5649484536082474,
+      "grad_norm": 1.5870437622070312,
+      "learning_rate": 8.593803236899379e-05,
+      "loss": 2.5664,
+      "step": 1096
+    },
+    {
+      "epoch": 0.5670103092783505,
+      "grad_norm": 1.7030541896820068,
+      "learning_rate": 8.581930957829786e-05,
+      "loss": 2.4574,
+      "step": 1100
+    },
+    {
+      "epoch": 0.5690721649484536,
+      "grad_norm": 1.6299339532852173,
+      "learning_rate": 8.570017046896197e-05,
+      "loss": 2.5025,
+      "step": 1104
+    },
+    {
+      "epoch": 0.5711340206185567,
+      "grad_norm": 1.648901104927063,
+      "learning_rate": 8.558061642570936e-05,
+      "loss": 2.5028,
+      "step": 1108
+    },
+    {
+      "epoch": 0.5731958762886598,
+      "grad_norm": 1.5893465280532837,
+      "learning_rate": 8.5460648838086e-05,
+      "loss": 2.4432,
+      "step": 1112
+    },
+    {
+      "epoch": 0.5752577319587628,
+      "grad_norm": 1.686439871788025,
+      "learning_rate": 8.534026910044435e-05,
+      "loss": 2.4683,
+      "step": 1116
+    },
+    {
+      "epoch": 0.5773195876288659,
+      "grad_norm": 1.687406301498413,
+      "learning_rate": 8.521947861192723e-05,
+      "loss": 2.4717,
+      "step": 1120
+    },
+    {
+      "epoch": 0.5793814432989691,
+      "grad_norm": 1.7713639736175537,
+      "learning_rate": 8.509827877645144e-05,
+      "loss": 2.6082,
+      "step": 1124
+    },
+    {
+      "epoch": 0.5814432989690722,
+      "grad_norm": 1.7374292612075806,
+      "learning_rate": 8.49766710026916e-05,
+      "loss": 2.4267,
+      "step": 1128
+    },
+    {
+      "epoch": 0.5835051546391753,
+      "grad_norm": 1.5629856586456299,
+      "learning_rate": 8.48546567040636e-05,
+      "loss": 2.4698,
+      "step": 1132
+    },
+    {
+      "epoch": 0.5855670103092784,
+      "grad_norm": 1.6608953475952148,
+      "learning_rate": 8.473223729870834e-05,
+      "loss": 2.4221,
+      "step": 1136
+    },
+    {
+      "epoch": 0.5876288659793815,
+      "grad_norm": 1.6824475526809692,
+      "learning_rate": 8.460941420947514e-05,
+      "loss": 2.5411,
+      "step": 1140
+    },
+    {
+      "epoch": 0.5896907216494846,
+      "grad_norm": 1.590063214302063,
+      "learning_rate": 8.448618886390522e-05,
+      "loss": 2.4957,
+      "step": 1144
+    },
+    {
+      "epoch": 0.5917525773195876,
+      "grad_norm": 1.6329950094223022,
+      "learning_rate": 8.436256269421515e-05,
+      "loss": 2.489,
+      "step": 1148
+    },
+    {
+      "epoch": 0.5938144329896907,
+      "grad_norm": 1.7138408422470093,
+      "learning_rate": 8.423853713728016e-05,
+      "loss": 2.4355,
+      "step": 1152
+    },
+    {
+      "epoch": 0.5958762886597938,
+      "grad_norm": 1.707245111465454,
+      "learning_rate": 8.411411363461745e-05,
+      "loss": 2.4351,
+      "step": 1156
+    },
+    {
+      "epoch": 0.5979381443298969,
+      "grad_norm": 1.6063274145126343,
+      "learning_rate": 8.398929363236948e-05,
+      "loss": 2.4771,
+      "step": 1160
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.62711763381958,
+      "learning_rate": 8.386407858128706e-05,
+      "loss": 2.4136,
+      "step": 1164
+    },
+    {
+      "epoch": 0.6020618556701031,
+      "grad_norm": 1.7263373136520386,
+      "learning_rate": 8.373846993671261e-05,
+      "loss": 2.6003,
+      "step": 1168
+    },
+    {
+      "epoch": 0.6041237113402061,
+      "grad_norm": 1.6953507661819458,
+      "learning_rate": 8.361246915856314e-05,
+      "loss": 2.5567,
+      "step": 1172
+    },
+    {
+      "epoch": 0.6061855670103092,
+      "grad_norm": 1.6107468605041504,
+      "learning_rate": 8.348607771131336e-05,
+      "loss": 2.3889,
+      "step": 1176
+    },
+    {
+      "epoch": 0.6082474226804123,
+      "grad_norm": 1.5828678607940674,
+      "learning_rate": 8.335929706397863e-05,
+      "loss": 2.4063,
+      "step": 1180
+    },
+    {
+      "epoch": 0.6103092783505155,
+      "grad_norm": 1.6693129539489746,
+      "learning_rate": 8.323212869009782e-05,
+      "loss": 2.4474,
+      "step": 1184
+    },
+    {
+      "epoch": 0.6123711340206186,
+      "grad_norm": 1.575134515762329,
+      "learning_rate": 8.310457406771635e-05,
+      "loss": 2.5156,
+      "step": 1188
+    },
+    {
+      "epoch": 0.6144329896907217,
+      "grad_norm": 1.600196123123169,
+      "learning_rate": 8.297663467936882e-05,
+      "loss": 2.4151,
+      "step": 1192
+    },
+    {
+      "epoch": 0.6164948453608248,
+      "grad_norm": 1.6348146200180054,
+      "learning_rate": 8.28483120120619e-05,
+      "loss": 2.4336,
+      "step": 1196
+    },
+    {
+      "epoch": 0.6185567010309279,
+      "grad_norm": 1.6424046754837036,
+      "learning_rate": 8.271960755725702e-05,
+      "loss": 2.45,
+      "step": 1200
+    },
+    {
+      "epoch": 0.6206185567010309,
+      "grad_norm": 1.6451953649520874,
+      "learning_rate": 8.2590522810853e-05,
+      "loss": 2.4414,
+      "step": 1204
+    },
+    {
+      "epoch": 0.622680412371134,
+      "grad_norm": 1.6768877506256104,
+      "learning_rate": 8.246105927316874e-05,
+      "loss": 2.459,
+      "step": 1208
+    },
+    {
+      "epoch": 0.6247422680412371,
+      "grad_norm": 1.7277884483337402,
+      "learning_rate": 8.233121844892568e-05,
+      "loss": 2.4398,
+      "step": 1212
+    },
+    {
+      "epoch": 0.6268041237113402,
+      "grad_norm": 1.6326000690460205,
+      "learning_rate": 8.220100184723038e-05,
+      "loss": 2.5278,
+      "step": 1216
+    },
+    {
+      "epoch": 0.6288659793814433,
+      "grad_norm": 1.6363636255264282,
+      "learning_rate": 8.2070410981557e-05,
+      "loss": 2.466,
+      "step": 1220
+    },
+    {
+      "epoch": 0.6309278350515464,
+      "grad_norm": 1.7234989404678345,
+      "learning_rate": 8.193944736972963e-05,
+      "loss": 2.4129,
+      "step": 1224
+    },
+    {
+      "epoch": 0.6329896907216495,
+      "grad_norm": 1.6150864362716675,
+      "learning_rate": 8.180811253390472e-05,
+      "loss": 2.4872,
+      "step": 1228
+    },
+    {
+      "epoch": 0.6350515463917525,
+      "grad_norm": 1.7305302619934082,
+      "learning_rate": 8.167640800055335e-05,
+      "loss": 2.5246,
+      "step": 1232
+    },
+    {
+      "epoch": 0.6371134020618556,
+      "grad_norm": 1.7068026065826416,
+      "learning_rate": 8.15443353004435e-05,
+      "loss": 2.4428,
+      "step": 1236
+    },
+    {
+      "epoch": 0.6391752577319587,
+      "grad_norm": 1.7080657482147217,
+      "learning_rate": 8.141189596862225e-05,
+      "loss": 2.3974,
+      "step": 1240
+    },
+    {
+      "epoch": 0.6412371134020619,
+      "grad_norm": 1.690262794494629,
+      "learning_rate": 8.127909154439796e-05,
+      "loss": 2.5146,
+      "step": 1244
+    },
+    {
+      "epoch": 0.643298969072165,
+      "grad_norm": 1.7586865425109863,
+      "learning_rate": 8.114592357132236e-05,
+      "loss": 2.459,
+      "step": 1248
+    },
+    {
+      "epoch": 0.6453608247422681,
+      "grad_norm": 1.6623564958572388,
+      "learning_rate": 8.101239359717263e-05,
+      "loss": 2.4779,
+      "step": 1252
+    },
+    {
+      "epoch": 0.6474226804123712,
+      "grad_norm": 1.694476842880249,
+      "learning_rate": 8.087850317393335e-05,
+      "loss": 2.4863,
+      "step": 1256
+    },
+    {
+      "epoch": 0.6494845360824743,
+      "grad_norm": 1.7273415327072144,
+      "learning_rate": 8.074425385777857e-05,
+      "loss": 2.5528,
+      "step": 1260
+    },
+    {
+      "epoch": 0.6515463917525773,
+      "grad_norm": 1.61915922164917,
+      "learning_rate": 8.06096472090536e-05,
+      "loss": 2.5224,
+      "step": 1264
+    },
+    {
+      "epoch": 0.6536082474226804,
+      "grad_norm": 1.64993417263031,
+      "learning_rate": 8.047468479225699e-05,
+      "loss": 2.475,
+      "step": 1268
+    },
+    {
+      "epoch": 0.6556701030927835,
+      "grad_norm": 1.6789292097091675,
+      "learning_rate": 8.033936817602225e-05,
+      "loss": 2.5414,
+      "step": 1272
+    },
+    {
+      "epoch": 0.6577319587628866,
+      "grad_norm": 1.534625768661499,
+      "learning_rate": 8.020369893309969e-05,
+      "loss": 2.5186,
+      "step": 1276
+    },
+    {
+      "epoch": 0.6597938144329897,
+      "grad_norm": 1.6226372718811035,
+      "learning_rate": 8.006767864033805e-05,
+      "loss": 2.5079,
+      "step": 1280
+    },
+    {
+      "epoch": 0.6618556701030928,
+      "grad_norm": 1.6352249383926392,
+      "learning_rate": 7.993130887866631e-05,
+      "loss": 2.4257,
+      "step": 1284
+    },
+    {
+      "epoch": 0.6639175257731958,
+      "grad_norm": 1.6394164562225342,
+      "learning_rate": 7.97945912330752e-05,
+      "loss": 2.4692,
+      "step": 1288
+    },
+    {
+      "epoch": 0.6659793814432989,
+      "grad_norm": 1.7224717140197754,
+      "learning_rate": 7.965752729259881e-05,
+      "loss": 2.5426,
+      "step": 1292
+    },
+    {
+      "epoch": 0.668041237113402,
+      "grad_norm": 1.662453532218933,
+      "learning_rate": 7.952011865029614e-05,
+      "loss": 2.5418,
+      "step": 1296
+    },
+    {
+      "epoch": 0.6701030927835051,
+      "grad_norm": 1.5507217645645142,
+      "learning_rate": 7.938236690323255e-05,
+      "loss": 2.3688,
+      "step": 1300
+    },
+    {
+      "epoch": 0.6721649484536083,
+      "grad_norm": 1.6199603080749512,
+      "learning_rate": 7.924427365246125e-05,
+      "loss": 2.4501,
+      "step": 1304
+    },
+    {
+      "epoch": 0.6742268041237114,
+      "grad_norm": 1.5297261476516724,
+      "learning_rate": 7.910584050300465e-05,
+      "loss": 2.4709,
+      "step": 1308
+    },
+    {
+      "epoch": 0.6762886597938145,
+      "grad_norm": 1.5516424179077148,
+      "learning_rate": 7.896706906383568e-05,
+      "loss": 2.4583,
+      "step": 1312
+    },
+    {
+      "epoch": 0.6783505154639176,
+      "grad_norm": 1.5144871473312378,
+      "learning_rate": 7.882796094785918e-05,
+      "loss": 2.4077,
+      "step": 1316
+    },
+    {
+      "epoch": 0.6804123711340206,
+      "grad_norm": 1.6823511123657227,
+      "learning_rate": 7.868851777189306e-05,
+      "loss": 2.4394,
+      "step": 1320
+    },
+    {
+      "epoch": 0.6824742268041237,
+      "grad_norm": 1.5894042253494263,
+      "learning_rate": 7.854874115664957e-05,
+      "loss": 2.4256,
+      "step": 1324
+    },
+    {
+      "epoch": 0.6845360824742268,
+      "grad_norm": 1.682361125946045,
+      "learning_rate": 7.84086327267164e-05,
+      "loss": 2.4082,
+      "step": 1328
+    },
+    {
+      "epoch": 0.6865979381443299,
+      "grad_norm": 1.6730499267578125,
+      "learning_rate": 7.826819411053787e-05,
+      "loss": 2.43,
+      "step": 1332
+    },
+    {
+      "epoch": 0.688659793814433,
+      "grad_norm": 1.5576122999191284,
+      "learning_rate": 7.812742694039599e-05,
+      "loss": 2.4109,
+      "step": 1336
+    },
+    {
+      "epoch": 0.6907216494845361,
+      "grad_norm": 1.735579013824463,
+      "learning_rate": 7.798633285239141e-05,
+      "loss": 2.5218,
+      "step": 1340
+    },
+    {
+      "epoch": 0.6927835051546392,
+      "grad_norm": 1.5843349695205688,
+      "learning_rate": 7.784491348642452e-05,
+      "loss": 2.4067,
+      "step": 1344
+    },
+    {
+      "epoch": 0.6948453608247422,
+      "grad_norm": 1.5844374895095825,
+      "learning_rate": 7.770317048617631e-05,
+      "loss": 2.4879,
+      "step": 1348
+    },
+    {
+      "epoch": 0.6969072164948453,
+      "grad_norm": 1.6641557216644287,
+      "learning_rate": 7.756110549908924e-05,
+      "loss": 2.4427,
+      "step": 1352
+    },
+    {
+      "epoch": 0.6989690721649484,
+      "grad_norm": 1.567859172821045,
+      "learning_rate": 7.741872017634824e-05,
+      "loss": 2.458,
+      "step": 1356
+    },
+    {
+      "epoch": 0.7010309278350515,
+      "grad_norm": 1.649179220199585,
+      "learning_rate": 7.727601617286132e-05,
+      "loss": 2.4483,
+      "step": 1360
+    },
+    {
+      "epoch": 0.7030927835051546,
+      "grad_norm": 1.6399873495101929,
+      "learning_rate": 7.713299514724051e-05,
+      "loss": 2.5019,
+      "step": 1364
+    },
+    {
+      "epoch": 0.7051546391752578,
+      "grad_norm": 1.724725365638733,
+      "learning_rate": 7.698965876178246e-05,
+      "loss": 2.4014,
+      "step": 1368
+    },
+    {
+      "epoch": 0.7072164948453609,
+      "grad_norm": 1.734067440032959,
+      "learning_rate": 7.68460086824492e-05,
+      "loss": 2.3332,
+      "step": 1372
+    },
+    {
+      "epoch": 0.709278350515464,
+      "grad_norm": 1.5960577726364136,
+      "learning_rate": 7.67020465788487e-05,
+      "loss": 2.4735,
+      "step": 1376
+    },
+    {
+      "epoch": 0.711340206185567,
+      "grad_norm": 1.6822943687438965,
+      "learning_rate": 7.655777412421554e-05,
+      "loss": 2.515,
+      "step": 1380
+    },
+    {
+      "epoch": 0.7134020618556701,
+      "grad_norm": 1.7459498643875122,
+      "learning_rate": 7.641319299539145e-05,
+      "loss": 2.4009,
+      "step": 1384
+    },
+    {
+      "epoch": 0.7154639175257732,
+      "grad_norm": 1.5728472471237183,
+      "learning_rate": 7.626830487280573e-05,
+      "loss": 2.3888,
+      "step": 1388
+    },
+    {
+      "epoch": 0.7175257731958763,
+      "grad_norm": 1.621801495552063,
+      "learning_rate": 7.612311144045588e-05,
+      "loss": 2.4036,
+      "step": 1392
+    },
+    {
+      "epoch": 0.7195876288659794,
+      "grad_norm": 1.6353496313095093,
+      "learning_rate": 7.597761438588784e-05,
+      "loss": 2.4504,
+      "step": 1396
+    },
+    {
+      "epoch": 0.7216494845360825,
+      "grad_norm": 1.5395163297653198,
+      "learning_rate": 7.583181540017656e-05,
+      "loss": 2.3526,
+      "step": 1400
+    },
+    {
+      "epoch": 0.7237113402061855,
+      "grad_norm": 1.6341197490692139,
+      "learning_rate": 7.568571617790624e-05,
+      "loss": 2.4689,
+      "step": 1404
+    },
+    {
+      "epoch": 0.7257731958762886,
+      "grad_norm": 1.5934208631515503,
+      "learning_rate": 7.553931841715061e-05,
+      "loss": 2.5371,
+      "step": 1408
+    },
+    {
+      "epoch": 0.7278350515463917,
+      "grad_norm": 1.6311752796173096,
+      "learning_rate": 7.539262381945323e-05,
+      "loss": 2.4693,
+      "step": 1412
+    },
+    {
+      "epoch": 0.7298969072164948,
+      "grad_norm": 1.6778818368911743,
+      "learning_rate": 7.52456340898078e-05,
+      "loss": 2.4041,
+      "step": 1416
+    },
+    {
+      "epoch": 0.7319587628865979,
+      "grad_norm": 1.5758112668991089,
+      "learning_rate": 7.509835093663821e-05,
+      "loss": 2.4328,
+      "step": 1420
+    },
+    {
+      "epoch": 0.734020618556701,
+      "grad_norm": 1.6166560649871826,
+      "learning_rate": 7.495077607177872e-05,
+      "loss": 2.386,
+      "step": 1424
+    },
+    {
+      "epoch": 0.7360824742268042,
+      "grad_norm": 1.4999052286148071,
+      "learning_rate": 7.48029112104541e-05,
+      "loss": 2.3724,
+      "step": 1428
+    },
+    {
+      "epoch": 0.7381443298969073,
+      "grad_norm": 1.6186021566390991,
+      "learning_rate": 7.465475807125968e-05,
+      "loss": 2.3181,
+      "step": 1432
+    },
+    {
+      "epoch": 0.7402061855670103,
+      "grad_norm": 1.7516733407974243,
+      "learning_rate": 7.450631837614138e-05,
+      "loss": 2.4565,
+      "step": 1436
+    },
+    {
+      "epoch": 0.7422680412371134,
+      "grad_norm": 1.647589087486267,
+      "learning_rate": 7.435759385037565e-05,
+      "loss": 2.4047,
+      "step": 1440
+    },
+    {
+      "epoch": 0.7443298969072165,
+      "grad_norm": 1.5819942951202393,
+      "learning_rate": 7.420858622254946e-05,
+      "loss": 2.455,
+      "step": 1444
+    },
+    {
+      "epoch": 0.7463917525773196,
+      "grad_norm": 1.5764845609664917,
+      "learning_rate": 7.405929722454026e-05,
+      "loss": 2.3826,
+      "step": 1448
+    },
+    {
+      "epoch": 0.7484536082474227,
+      "grad_norm": 1.6311144828796387,
+      "learning_rate": 7.39097285914957e-05,
+      "loss": 2.3991,
+      "step": 1452
+    },
+    {
+      "epoch": 0.7505154639175258,
+      "grad_norm": 1.5981923341751099,
+      "learning_rate": 7.375988206181365e-05,
+      "loss": 2.4517,
+      "step": 1456
+    },
+    {
+      "epoch": 0.7525773195876289,
+      "grad_norm": 1.5369350910186768,
+      "learning_rate": 7.360975937712185e-05,
+      "loss": 2.4181,
+      "step": 1460
+    },
+    {
+      "epoch": 0.7546391752577319,
+      "grad_norm": 1.6119554042816162,
+      "learning_rate": 7.345936228225769e-05,
+      "loss": 2.4652,
+      "step": 1464
+    },
+    {
+      "epoch": 0.756701030927835,
+      "grad_norm": 1.7482960224151611,
+      "learning_rate": 7.330869252524804e-05,
+      "loss": 2.4213,
+      "step": 1468
+    },
+    {
+      "epoch": 0.7587628865979381,
+      "grad_norm": 1.676172137260437,
+      "learning_rate": 7.315775185728877e-05,
+      "loss": 2.3816,
+      "step": 1472
+    },
+    {
+      "epoch": 0.7608247422680412,
+      "grad_norm": 4.144953727722168,
+      "learning_rate": 7.300654203272454e-05,
+      "loss": 2.412,
+      "step": 1476
+    },
+    {
+      "epoch": 0.7628865979381443,
+      "grad_norm": 1.5259236097335815,
+      "learning_rate": 7.285506480902831e-05,
+      "loss": 2.4274,
+      "step": 1480
+    },
+    {
+      "epoch": 0.7649484536082474,
+      "grad_norm": 1.5517241954803467,
+      "learning_rate": 7.270332194678097e-05,
+      "loss": 2.3633,
+      "step": 1484
+    },
+    {
+      "epoch": 0.7670103092783506,
+      "grad_norm": 1.583932638168335,
+      "learning_rate": 7.255131520965087e-05,
+      "loss": 2.3964,
+      "step": 1488
+    },
+    {
+      "epoch": 0.7690721649484537,
+      "grad_norm": 1.5848530530929565,
+      "learning_rate": 7.239904636437325e-05,
+      "loss": 2.42,
+      "step": 1492
+    },
+    {
+      "epoch": 0.7711340206185567,
+      "grad_norm": 1.5512683391571045,
+      "learning_rate": 7.224651718072984e-05,
+      "loss": 2.4654,
+      "step": 1496
+    },
+    {
+      "epoch": 0.7731958762886598,
+      "grad_norm": 1.5761094093322754,
+      "learning_rate": 7.20937294315282e-05,
+      "loss": 2.4011,
+      "step": 1500
+    },
+    {
+      "epoch": 0.7752577319587629,
+      "grad_norm": 1.519158124923706,
+      "learning_rate": 7.194068489258109e-05,
+      "loss": 2.4315,
+      "step": 1504
+    },
+    {
+      "epoch": 0.777319587628866,
+      "grad_norm": 1.5774180889129639,
+      "learning_rate": 7.178738534268591e-05,
+      "loss": 2.4667,
+      "step": 1508
+    },
+    {
+      "epoch": 0.7793814432989691,
+      "grad_norm": 1.6394808292388916,
+      "learning_rate": 7.163383256360398e-05,
+      "loss": 2.4861,
+      "step": 1512
+    },
+    {
+      "epoch": 0.7814432989690722,
+      "grad_norm": 1.549655795097351,
+      "learning_rate": 7.14800283400398e-05,
+      "loss": 2.3323,
+      "step": 1516
+    },
+    {
+      "epoch": 0.7835051546391752,
+      "grad_norm": 1.6154853105545044,
+      "learning_rate": 7.132597445962042e-05,
+      "loss": 2.5378,
+      "step": 1520
+    },
+    {
+      "epoch": 0.7855670103092783,
+      "grad_norm": 1.6021648645401,
+      "learning_rate": 7.117167271287453e-05,
+      "loss": 2.4249,
+      "step": 1524
+    },
+    {
+      "epoch": 0.7876288659793814,
+      "grad_norm": 1.665974736213684,
+      "learning_rate": 7.101712489321169e-05,
+      "loss": 2.4566,
+      "step": 1528
+    },
+    {
+      "epoch": 0.7896907216494845,
+      "grad_norm": 1.6290616989135742,
+      "learning_rate": 7.086233279690158e-05,
+      "loss": 2.3565,
+      "step": 1532
+    },
+    {
+      "epoch": 0.7917525773195876,
+      "grad_norm": 1.5475256443023682,
+      "learning_rate": 7.070729822305298e-05,
+      "loss": 2.4542,
+      "step": 1536
+    },
+    {
+      "epoch": 0.7938144329896907,
+      "grad_norm": 1.6090917587280273,
+      "learning_rate": 7.055202297359293e-05,
+      "loss": 2.4498,
+      "step": 1540
+    },
+    {
+      "epoch": 0.7958762886597938,
+      "grad_norm": 1.5616388320922852,
+      "learning_rate": 7.039650885324582e-05,
+      "loss": 2.3708,
+      "step": 1544
+    },
+    {
+      "epoch": 0.797938144329897,
+      "grad_norm": 1.711113452911377,
+      "learning_rate": 7.024075766951233e-05,
+      "loss": 2.5234,
+      "step": 1548
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.6016921997070312,
+      "learning_rate": 7.008477123264848e-05,
+      "loss": 2.4755,
+      "step": 1552
+    },
+    {
+      "epoch": 0.8020618556701031,
+      "grad_norm": 1.5505585670471191,
+      "learning_rate": 6.99285513556446e-05,
+      "loss": 2.3435,
+      "step": 1556
+    },
+    {
+      "epoch": 0.8041237113402062,
+      "grad_norm": 1.532699704170227,
+      "learning_rate": 6.977209985420419e-05,
+      "loss": 2.4878,
+      "step": 1560
+    },
+    {
+      "epoch": 0.8061855670103093,
+      "grad_norm": 2.197500705718994,
+      "learning_rate": 6.961541854672293e-05,
+      "loss": 2.3707,
+      "step": 1564
+    },
+    {
+      "epoch": 0.8082474226804124,
+      "grad_norm": 1.6112059354782104,
+      "learning_rate": 6.945850925426742e-05,
+      "loss": 2.4803,
+      "step": 1568
+    },
+    {
+      "epoch": 0.8103092783505155,
+      "grad_norm": 1.6318986415863037,
+      "learning_rate": 6.930137380055403e-05,
+      "loss": 2.315,
+      "step": 1572
+    },
+    {
+      "epoch": 0.8123711340206186,
+      "grad_norm": 1.607735514640808,
+      "learning_rate": 6.914401401192789e-05,
+      "loss": 2.3804,
+      "step": 1576
+    },
+    {
+      "epoch": 0.8144329896907216,
+      "grad_norm": 1.474547266960144,
+      "learning_rate": 6.898643171734137e-05,
+      "loss": 2.3736,
+      "step": 1580
+    },
+    {
+      "epoch": 0.8164948453608247,
+      "grad_norm": 1.9736120700836182,
+      "learning_rate": 6.882862874833305e-05,
+      "loss": 2.3804,
+      "step": 1584
+    },
+    {
+      "epoch": 0.8185567010309278,
+      "grad_norm": 1.5962013006210327,
+      "learning_rate": 6.867060693900631e-05,
+      "loss": 2.4503,
+      "step": 1588
+    },
+    {
+      "epoch": 0.8206185567010309,
+      "grad_norm": 1.565443754196167,
+      "learning_rate": 6.851236812600808e-05,
+      "loss": 2.4805,
+      "step": 1592
+    },
+    {
+      "epoch": 0.822680412371134,
+      "grad_norm": 1.6651315689086914,
+      "learning_rate": 6.835391414850748e-05,
+      "loss": 2.4496,
+      "step": 1596
+    },
+    {
+      "epoch": 0.8247422680412371,
+      "grad_norm": 1.5796141624450684,
+      "learning_rate": 6.819524684817438e-05,
+      "loss": 2.4143,
+      "step": 1600
+    },
+    {
+      "epoch": 0.8268041237113402,
+      "grad_norm": 1.513759732246399,
+      "learning_rate": 6.803636806915812e-05,
+      "loss": 2.4383,
+      "step": 1604
+    },
+    {
+      "epoch": 0.8288659793814434,
+      "grad_norm": 1.556221842765808,
+      "learning_rate": 6.787727965806591e-05,
+      "loss": 2.4026,
+      "step": 1608
+    },
+    {
+      "epoch": 0.8309278350515464,
+      "grad_norm": 1.5572640895843506,
+      "learning_rate": 6.771798346394157e-05,
+      "loss": 2.4022,
+      "step": 1612
+    },
+    {
+      "epoch": 0.8329896907216495,
+      "grad_norm": 1.602418065071106,
+      "learning_rate": 6.755848133824383e-05,
+      "loss": 2.4555,
+      "step": 1616
+    },
+    {
+      "epoch": 0.8350515463917526,
+      "grad_norm": 1.676102638244629,
+      "learning_rate": 6.739877513482497e-05,
+      "loss": 2.3695,
+      "step": 1620
+    },
+    {
+      "epoch": 0.8371134020618557,
+      "grad_norm": 1.748834252357483,
+      "learning_rate": 6.72388667099092e-05,
+      "loss": 2.2801,
+      "step": 1624
+    },
+    {
+      "epoch": 0.8391752577319588,
+      "grad_norm": 1.5271378755569458,
+      "learning_rate": 6.707875792207108e-05,
+      "loss": 2.2845,
+      "step": 1628
+    },
+    {
+      "epoch": 0.8412371134020619,
+      "grad_norm": 1.7049288749694824,
+      "learning_rate": 6.691845063221393e-05,
+      "loss": 2.4126,
+      "step": 1632
+    },
+    {
+      "epoch": 0.843298969072165,
+      "grad_norm": 1.590855360031128,
+      "learning_rate": 6.675794670354826e-05,
+      "loss": 2.3571,
+      "step": 1636
+    },
+    {
+      "epoch": 0.845360824742268,
+      "grad_norm": 1.7120413780212402,
+      "learning_rate": 6.659724800157002e-05,
+      "loss": 2.4111,
+      "step": 1640
+    },
+    {
+      "epoch": 0.8474226804123711,
+      "grad_norm": 1.5834614038467407,
+      "learning_rate": 6.643635639403897e-05,
+      "loss": 2.4045,
+      "step": 1644
+    },
+    {
+      "epoch": 0.8494845360824742,
+      "grad_norm": 1.4899585247039795,
+      "learning_rate": 6.627527375095697e-05,
+      "loss": 2.2914,
+      "step": 1648
+    },
+    {
+      "epoch": 0.8515463917525773,
+      "grad_norm": 1.573357105255127,
+      "learning_rate": 6.611400194454619e-05,
+      "loss": 2.3746,
+      "step": 1652
+    },
+    {
+      "epoch": 0.8536082474226804,
+      "grad_norm": 1.528746247291565,
+      "learning_rate": 6.595254284922748e-05,
+      "loss": 2.3568,
+      "step": 1656
+    },
+    {
+      "epoch": 0.8556701030927835,
+      "grad_norm": 1.6635682582855225,
+      "learning_rate": 6.579089834159844e-05,
+      "loss": 2.3868,
+      "step": 1660
+    },
+    {
+      "epoch": 0.8577319587628865,
+      "grad_norm": 1.5613417625427246,
+      "learning_rate": 6.562907030041168e-05,
+      "loss": 2.377,
+      "step": 1664
+    },
+    {
+      "epoch": 0.8597938144329897,
+      "grad_norm": 1.6135239601135254,
+      "learning_rate": 6.546706060655298e-05,
+      "loss": 2.4542,
+      "step": 1668
+    },
+    {
+      "epoch": 0.8618556701030928,
+      "grad_norm": 1.5151220560073853,
+      "learning_rate": 6.530487114301944e-05,
+      "loss": 2.2057,
+      "step": 1672
+    },
+    {
+      "epoch": 0.8639175257731959,
+      "grad_norm": 1.6280299425125122,
+      "learning_rate": 6.514250379489753e-05,
+      "loss": 2.3594,
+      "step": 1676
+    },
+    {
+      "epoch": 0.865979381443299,
+      "grad_norm": 1.5538045167922974,
+      "learning_rate": 6.49799604493413e-05,
+      "loss": 2.4155,
+      "step": 1680
+    },
+    {
+      "epoch": 0.8680412371134021,
+      "grad_norm": 1.5412131547927856,
+      "learning_rate": 6.481724299555029e-05,
+      "loss": 2.3285,
+      "step": 1684
+    },
+    {
+      "epoch": 0.8701030927835052,
+      "grad_norm": 1.7468520402908325,
+      "learning_rate": 6.465435332474768e-05,
+      "loss": 2.2616,
+      "step": 1688
+    },
+    {
+      "epoch": 0.8721649484536083,
+      "grad_norm": 1.4656267166137695,
+      "learning_rate": 6.449129333015834e-05,
+      "loss": 2.2932,
+      "step": 1692
+    },
+    {
+      "epoch": 0.8742268041237113,
+      "grad_norm": 1.554957628250122,
+      "learning_rate": 6.432806490698671e-05,
+      "loss": 2.4188,
+      "step": 1696
+    },
+    {
+      "epoch": 0.8762886597938144,
+      "grad_norm": 1.6073874235153198,
+      "learning_rate": 6.416466995239485e-05,
+      "loss": 2.2739,
+      "step": 1700
+    },
+    {
+      "epoch": 0.8783505154639175,
+      "grad_norm": 1.6184226274490356,
+      "learning_rate": 6.400111036548037e-05,
+      "loss": 2.3879,
+      "step": 1704
+    },
+    {
+      "epoch": 0.8804123711340206,
+      "grad_norm": 1.5735539197921753,
+      "learning_rate": 6.383738804725438e-05,
+      "loss": 2.3191,
+      "step": 1708
+    },
+    {
+      "epoch": 0.8824742268041237,
+      "grad_norm": 1.6026270389556885,
+      "learning_rate": 6.36735049006193e-05,
+      "loss": 2.5497,
+      "step": 1712
+    },
+    {
+      "epoch": 0.8845360824742268,
+      "grad_norm": 1.6282131671905518,
+      "learning_rate": 6.350946283034696e-05,
+      "loss": 2.4152,
+      "step": 1716
+    },
+    {
+      "epoch": 0.8865979381443299,
+      "grad_norm": 1.553600549697876,
+      "learning_rate": 6.334526374305615e-05,
+      "loss": 2.3417,
+      "step": 1720
+    },
+    {
+      "epoch": 0.8886597938144329,
+      "grad_norm": 1.5668585300445557,
+      "learning_rate": 6.318090954719074e-05,
+      "loss": 2.3811,
+      "step": 1724
+    },
+    {
+      "epoch": 0.8907216494845361,
+      "grad_norm": 1.7151823043823242,
+      "learning_rate": 6.301640215299735e-05,
+      "loss": 2.2925,
+      "step": 1728
+    },
+    {
+      "epoch": 0.8927835051546392,
+      "grad_norm": 1.5615097284317017,
+      "learning_rate": 6.285174347250322e-05,
+      "loss": 2.3591,
+      "step": 1732
+    },
+    {
+      "epoch": 0.8948453608247423,
+      "grad_norm": 1.6054131984710693,
+      "learning_rate": 6.26869354194939e-05,
+      "loss": 2.3084,
+      "step": 1736
+    },
+    {
+      "epoch": 0.8969072164948454,
+      "grad_norm": 1.606719732284546,
+      "learning_rate": 6.252197990949108e-05,
+      "loss": 2.4081,
+      "step": 1740
+    },
+    {
+      "epoch": 0.8989690721649485,
+      "grad_norm": 1.6772600412368774,
+      "learning_rate": 6.235687885973032e-05,
+      "loss": 2.3733,
+      "step": 1744
+    },
+    {
+      "epoch": 0.9010309278350516,
+      "grad_norm": 1.508259654045105,
+      "learning_rate": 6.219163418913872e-05,
+      "loss": 2.3318,
+      "step": 1748
+    },
+    {
+      "epoch": 0.9030927835051547,
+      "grad_norm": 1.5906274318695068,
+      "learning_rate": 6.202624781831268e-05,
+      "loss": 2.297,
+      "step": 1752
+    },
+    {
+      "epoch": 0.9051546391752577,
+      "grad_norm": 1.5747168064117432,
+      "learning_rate": 6.186072166949552e-05,
+      "loss": 2.4498,
+      "step": 1756
+    },
+    {
+      "epoch": 0.9072164948453608,
+      "grad_norm": 1.5119340419769287,
+      "learning_rate": 6.16950576665552e-05,
+      "loss": 2.3045,
+      "step": 1760
+    },
+    {
+      "epoch": 0.9092783505154639,
+      "grad_norm": 1.5936625003814697,
+      "learning_rate": 6.152925773496189e-05,
+      "loss": 2.3989,
+      "step": 1764
+    },
+    {
+      "epoch": 0.911340206185567,
+      "grad_norm": 1.6385735273361206,
+      "learning_rate": 6.13633238017656e-05,
+      "loss": 2.3415,
+      "step": 1768
+    },
+    {
+      "epoch": 0.9134020618556701,
+      "grad_norm": 1.6806796789169312,
+      "learning_rate": 6.119725779557386e-05,
+      "loss": 2.37,
+      "step": 1772
+    },
+    {
+      "epoch": 0.9154639175257732,
+      "grad_norm": 1.5404062271118164,
+      "learning_rate": 6.103106164652924e-05,
+      "loss": 2.476,
+      "step": 1776
+    },
+    {
+      "epoch": 0.9175257731958762,
+      "grad_norm": 1.5629997253417969,
+      "learning_rate": 6.086473728628691e-05,
+      "loss": 2.3843,
+      "step": 1780
+    },
+    {
+      "epoch": 0.9195876288659793,
+      "grad_norm": 1.5114765167236328,
+      "learning_rate": 6.069828664799221e-05,
+      "loss": 2.3972,
+      "step": 1784
+    },
+    {
+      "epoch": 0.9216494845360824,
+      "grad_norm": 1.5878251791000366,
+      "learning_rate": 6.053171166625817e-05,
+      "loss": 2.3686,
+      "step": 1788
+    },
+    {
+      "epoch": 0.9237113402061856,
+      "grad_norm": 1.6494956016540527,
+      "learning_rate": 6.036501427714304e-05,
+      "loss": 2.4627,
+      "step": 1792
+    },
+    {
+      "epoch": 0.9257731958762887,
+      "grad_norm": 1.5250719785690308,
+      "learning_rate": 6.0198196418127804e-05,
+      "loss": 2.332,
+      "step": 1796
+    },
+    {
+      "epoch": 0.9278350515463918,
+      "grad_norm": 1.5805870294570923,
+      "learning_rate": 6.0031260028093596e-05,
+      "loss": 2.2881,
+      "step": 1800
+    },
+    {
+      "epoch": 0.9298969072164949,
+      "grad_norm": 1.6690597534179688,
+      "learning_rate": 5.98642070472992e-05,
+      "loss": 2.2946,
+      "step": 1804
+    },
+    {
+      "epoch": 0.931958762886598,
+      "grad_norm": 1.5214745998382568,
+      "learning_rate": 5.969703941735858e-05,
+      "loss": 2.3984,
+      "step": 1808
+    },
+    {
+      "epoch": 0.934020618556701,
+      "grad_norm": 1.6358650922775269,
+      "learning_rate": 5.952975908121815e-05,
+      "loss": 2.3758,
+      "step": 1812
+    },
+    {
+      "epoch": 0.9360824742268041,
+      "grad_norm": 1.9057590961456299,
+      "learning_rate": 5.936236798313431e-05,
+      "loss": 2.3347,
+      "step": 1816
+    },
+    {
+      "epoch": 0.9381443298969072,
+      "grad_norm": 1.6489713191986084,
+      "learning_rate": 5.9194868068650845e-05,
+      "loss": 2.2552,
+      "step": 1820
+    },
+    {
+      "epoch": 0.9402061855670103,
+      "grad_norm": 1.5380582809448242,
+      "learning_rate": 5.902726128457625e-05,
+      "loss": 2.2898,
+      "step": 1824
+    },
+    {
+      "epoch": 0.9422680412371134,
+      "grad_norm": 1.5303738117218018,
+      "learning_rate": 5.885954957896115e-05,
+      "loss": 2.2896,
+      "step": 1828
+    },
+    {
+      "epoch": 0.9443298969072165,
+      "grad_norm": 1.577243685722351,
+      "learning_rate": 5.8691734901075634e-05,
+      "loss": 2.3281,
+      "step": 1832
+    },
+    {
+      "epoch": 0.9463917525773196,
+      "grad_norm": 1.545732021331787,
+      "learning_rate": 5.852381920138663e-05,
+      "loss": 2.2197,
+      "step": 1836
+    },
+    {
+      "epoch": 0.9484536082474226,
+      "grad_norm": 1.5541272163391113,
+      "learning_rate": 5.835580443153522e-05,
+      "loss": 2.2776,
+      "step": 1840
+    },
+    {
+      "epoch": 0.9505154639175257,
+      "grad_norm": 1.5543440580368042,
+      "learning_rate": 5.818769254431395e-05,
+      "loss": 2.3597,
+      "step": 1844
+    },
+    {
+      "epoch": 0.9525773195876288,
+      "grad_norm": 1.5079783201217651,
+      "learning_rate": 5.80194854936441e-05,
+      "loss": 2.3476,
+      "step": 1848
+    },
+    {
+      "epoch": 0.954639175257732,
+      "grad_norm": 1.503811240196228,
+      "learning_rate": 5.7851185234553064e-05,
+      "loss": 2.3318,
+      "step": 1852
+    },
+    {
+      "epoch": 0.9567010309278351,
+      "grad_norm": 1.532036542892456,
+      "learning_rate": 5.768279372315153e-05,
+      "loss": 2.2278,
+      "step": 1856
+    },
+    {
+      "epoch": 0.9587628865979382,
+      "grad_norm": 1.5316864252090454,
+      "learning_rate": 5.7514312916610814e-05,
+      "loss": 2.3835,
+      "step": 1860
+    },
+    {
+      "epoch": 0.9608247422680413,
+      "grad_norm": 1.5309964418411255,
+      "learning_rate": 5.7345744773140086e-05,
+      "loss": 2.3779,
+      "step": 1864
+    },
+    {
+      "epoch": 0.9628865979381444,
+      "grad_norm": 1.577561855316162,
+      "learning_rate": 5.71770912519636e-05,
+      "loss": 2.3334,
+      "step": 1868
+    },
+    {
+      "epoch": 0.9649484536082474,
+      "grad_norm": 1.5817878246307373,
+      "learning_rate": 5.7008354313297926e-05,
+      "loss": 2.28,
+      "step": 1872
+    },
+    {
+      "epoch": 0.9670103092783505,
+      "grad_norm": 1.5670493841171265,
+      "learning_rate": 5.683953591832922e-05,
+      "loss": 2.304,
+      "step": 1876
+    },
+    {
+      "epoch": 0.9690721649484536,
+      "grad_norm": 1.5366458892822266,
+      "learning_rate": 5.667063802919032e-05,
+      "loss": 2.4577,
+      "step": 1880
+    },
+    {
+      "epoch": 0.9711340206185567,
+      "grad_norm": 1.5330809354782104,
+      "learning_rate": 5.6501662608938014e-05,
+      "loss": 2.4178,
+      "step": 1884
+    },
+    {
+      "epoch": 0.9731958762886598,
+      "grad_norm": 1.4714388847351074,
+      "learning_rate": 5.633261162153027e-05,
+      "loss": 2.3252,
+      "step": 1888
+    },
+    {
+      "epoch": 0.9752577319587629,
+      "grad_norm": 1.5558996200561523,
+      "learning_rate": 5.6163487031803305e-05,
+      "loss": 2.3597,
+      "step": 1892
+    },
+    {
+      "epoch": 0.977319587628866,
+      "grad_norm": 1.5362873077392578,
+      "learning_rate": 5.5994290805448826e-05,
+      "loss": 2.2979,
+      "step": 1896
+    },
+    {
+      "epoch": 0.979381443298969,
+      "grad_norm": 1.5119233131408691,
+      "learning_rate": 5.582502490899111e-05,
+      "loss": 2.3582,
+      "step": 1900
+    },
+    {
+      "epoch": 0.9814432989690721,
+      "grad_norm": 1.6467958688735962,
+      "learning_rate": 5.565569130976422e-05,
+      "loss": 2.3366,
+      "step": 1904
+    },
+    {
+      "epoch": 0.9835051546391752,
+      "grad_norm": 1.6580411195755005,
+      "learning_rate": 5.548629197588913e-05,
+      "loss": 2.2901,
+      "step": 1908
+    },
+    {
+      "epoch": 0.9855670103092784,
+      "grad_norm": 1.4866169691085815,
+      "learning_rate": 5.5316828876250795e-05,
+      "loss": 2.3313,
+      "step": 1912
+    },
+    {
+      "epoch": 0.9876288659793815,
+      "grad_norm": 1.6460485458374023,
+      "learning_rate": 5.514730398047533e-05,
+      "loss": 2.2917,
+      "step": 1916
+    },
+    {
+      "epoch": 0.9896907216494846,
+      "grad_norm": 1.4895657300949097,
+      "learning_rate": 5.497771925890706e-05,
+      "loss": 2.3709,
+      "step": 1920
+    },
+    {
+      "epoch": 0.9917525773195877,
+      "grad_norm": 1.4158916473388672,
+      "learning_rate": 5.48080766825857e-05,
+      "loss": 2.1582,
+      "step": 1924
+    },
+    {
+      "epoch": 0.9938144329896907,
+      "grad_norm": 1.611509919166565,
+      "learning_rate": 5.463837822322333e-05,
+      "loss": 2.5048,
+      "step": 1928
+    },
+    {
+      "epoch": 0.9958762886597938,
+      "grad_norm": 1.6175845861434937,
+      "learning_rate": 5.446862585318161e-05,
+      "loss": 2.3836,
+      "step": 1932
+    },
+    {
+      "epoch": 0.9979381443298969,
+      "grad_norm": 1.6315233707427979,
+      "learning_rate": 5.429882154544875e-05,
+      "loss": 2.2926,
+      "step": 1936
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.427371025085449,
+      "learning_rate": 5.4128967273616625e-05,
+      "loss": 2.3215,
+      "step": 1940
+    }
+  ],
+  "logging_steps": 4,
+  "max_steps": 3880,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 1940,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.2824010606012006e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-1940/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd29156f56d8d4d72e8a360e74ba51b8d8b0f0946df0ed655e1d99ccd3aa14d1
+size 5432

checkpoint-3880/config.json

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

checkpoint-3880/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-3880/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:21b176d8ba5bc10d2304c139ea49b2e1d2844858082a3440d9a971463319527a
+size 4950070472

checkpoint-3880/model-00002-of-00006.safetensors

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

checkpoint-3880/model-00003-of-00006.safetensors

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

checkpoint-3880/model-00004-of-00006.safetensors

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

checkpoint-3880/model-00005-of-00006.safetensors

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

checkpoint-3880/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0dcd7e221fbdc7391161ba71c2d6d809f373d8bc53be5a7c84124eca7ed1688
+size 4614509080

checkpoint-3880/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-3880/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2355f51bbffb34e29f130bc864d31ea9b93db9d57ce4c30a53e2e6cd511fbaad
+size 13055369358

checkpoint-3880/rng_state.pth

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

checkpoint-3880/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9c26e6c575c0e4d1ecece4f16d34ff84e080b56c126d61c216d9060adb9f5f4
+size 1064

checkpoint-3880/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd29156f56d8d4d72e8a360e74ba51b8d8b0f0946df0ed655e1d99ccd3aa14d1
+size 5432

config.json

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

generation_config.json

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

model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:21b176d8ba5bc10d2304c139ea49b2e1d2844858082a3440d9a971463319527a
+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:f0dcd7e221fbdc7391161ba71c2d6d809f373d8bc53be5a7c84124eca7ed1688
+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:dd29156f56d8d4d72e8a360e74ba51b8d8b0f0946df0ed655e1d99ccd3aa14d1
+size 5432