Upload folder using huggingface_hub

checkpoint-4955/config.json

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

checkpoint-4955/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-4955/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f02160cd1501eaed60c5842d50a2c7a252680eaac669941c149a9bc8e15a8733
+size 4950070472

checkpoint-4955/model-00002-of-00006.safetensors

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

checkpoint-4955/model-00003-of-00006.safetensors

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

checkpoint-4955/model-00004-of-00006.safetensors

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

checkpoint-4955/model-00005-of-00006.safetensors

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

checkpoint-4955/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:135a3d1be7237a8a20ea25aa015e9aa4907f58353cf2654d0ec64e93aac7a3b3
+size 4614509080

checkpoint-4955/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-4955/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa9db518900a231510c06e1e724cb3746221040cc7b73f388a258a9958d82fa8
+size 13055369358

checkpoint-4955/rng_state.pth

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

checkpoint-4955/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4484acf9c43ca08110346c1865f9938a6ca484b1abdceedc34a9dea0396043d
+size 1064

checkpoint-4955/trainer_state.json

@@ -0,0 +1,3498 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 4955,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0020181634712411706,
+      "grad_norm": 61.64529800415039,
+      "learning_rate": 2.0161290322580646e-06,
+      "loss": 7.4781,
+      "step": 10
+    },
+    {
+      "epoch": 0.004036326942482341,
+      "grad_norm": 12.924609184265137,
+      "learning_rate": 4.032258064516129e-06,
+      "loss": 5.3628,
+      "step": 20
+    },
+    {
+      "epoch": 0.006054490413723511,
+      "grad_norm": 8.111140251159668,
+      "learning_rate": 6.048387096774194e-06,
+      "loss": 4.5188,
+      "step": 30
+    },
+    {
+      "epoch": 0.008072653884964682,
+      "grad_norm": 8.318406105041504,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 4.2211,
+      "step": 40
+    },
+    {
+      "epoch": 0.010090817356205853,
+      "grad_norm": 4.871032238006592,
+      "learning_rate": 1.0080645161290323e-05,
+      "loss": 3.844,
+      "step": 50
+    },
+    {
+      "epoch": 0.012108980827447022,
+      "grad_norm": 4.057001113891602,
+      "learning_rate": 1.2096774193548388e-05,
+      "loss": 3.5377,
+      "step": 60
+    },
+    {
+      "epoch": 0.014127144298688193,
+      "grad_norm": 3.9526875019073486,
+      "learning_rate": 1.4112903225806454e-05,
+      "loss": 3.203,
+      "step": 70
+    },
+    {
+      "epoch": 0.016145307769929364,
+      "grad_norm": 3.171088695526123,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 3.0832,
+      "step": 80
+    },
+    {
+      "epoch": 0.018163471241170535,
+      "grad_norm": 3.63295841217041,
+      "learning_rate": 1.8145161290322583e-05,
+      "loss": 2.9575,
+      "step": 90
+    },
+    {
+      "epoch": 0.020181634712411706,
+      "grad_norm": 2.8197014331817627,
+      "learning_rate": 2.0161290322580645e-05,
+      "loss": 2.831,
+      "step": 100
+    },
+    {
+      "epoch": 0.022199798183652877,
+      "grad_norm": 3.540992259979248,
+      "learning_rate": 2.217741935483871e-05,
+      "loss": 2.6876,
+      "step": 110
+    },
+    {
+      "epoch": 0.024217961654894045,
+      "grad_norm": 2.642745018005371,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 2.6302,
+      "step": 120
+    },
+    {
+      "epoch": 0.026236125126135216,
+      "grad_norm": 3.0355093479156494,
+      "learning_rate": 2.620967741935484e-05,
+      "loss": 2.5268,
+      "step": 130
+    },
+    {
+      "epoch": 0.028254288597376387,
+      "grad_norm": 2.7031023502349854,
+      "learning_rate": 2.822580645161291e-05,
+      "loss": 2.5202,
+      "step": 140
+    },
+    {
+      "epoch": 0.030272452068617558,
+      "grad_norm": 3.2656476497650146,
+      "learning_rate": 3.024193548387097e-05,
+      "loss": 2.3902,
+      "step": 150
+    },
+    {
+      "epoch": 0.03229061553985873,
+      "grad_norm": 2.836116075515747,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 2.4072,
+      "step": 160
+    },
+    {
+      "epoch": 0.034308779011099896,
+      "grad_norm": 2.732985496520996,
+      "learning_rate": 3.427419354838709e-05,
+      "loss": 2.3159,
+      "step": 170
+    },
+    {
+      "epoch": 0.03632694248234107,
+      "grad_norm": 2.5948660373687744,
+      "learning_rate": 3.6290322580645165e-05,
+      "loss": 2.2847,
+      "step": 180
+    },
+    {
+      "epoch": 0.03834510595358224,
+      "grad_norm": 2.968618869781494,
+      "learning_rate": 3.8306451612903224e-05,
+      "loss": 2.3511,
+      "step": 190
+    },
+    {
+      "epoch": 0.04036326942482341,
+      "grad_norm": 4.173515319824219,
+      "learning_rate": 4.032258064516129e-05,
+      "loss": 2.341,
+      "step": 200
+    },
+    {
+      "epoch": 0.04238143289606458,
+      "grad_norm": 2.294019937515259,
+      "learning_rate": 4.2338709677419356e-05,
+      "loss": 2.2497,
+      "step": 210
+    },
+    {
+      "epoch": 0.044399596367305755,
+      "grad_norm": 2.3338685035705566,
+      "learning_rate": 4.435483870967742e-05,
+      "loss": 2.2591,
+      "step": 220
+    },
+    {
+      "epoch": 0.04641775983854692,
+      "grad_norm": 2.389831066131592,
+      "learning_rate": 4.637096774193548e-05,
+      "loss": 2.2448,
+      "step": 230
+    },
+    {
+      "epoch": 0.04843592330978809,
+      "grad_norm": 2.5201873779296875,
+      "learning_rate": 4.8387096774193554e-05,
+      "loss": 2.272,
+      "step": 240
+    },
+    {
+      "epoch": 0.050454086781029264,
+      "grad_norm": 3.078508138656616,
+      "learning_rate": 5.040322580645161e-05,
+      "loss": 2.1645,
+      "step": 250
+    },
+    {
+      "epoch": 0.05247225025227043,
+      "grad_norm": 2.0750768184661865,
+      "learning_rate": 5.241935483870968e-05,
+      "loss": 2.2286,
+      "step": 260
+    },
+    {
+      "epoch": 0.054490413723511606,
+      "grad_norm": 2.5442333221435547,
+      "learning_rate": 5.443548387096774e-05,
+      "loss": 2.2015,
+      "step": 270
+    },
+    {
+      "epoch": 0.056508577194752774,
+      "grad_norm": 2.4685232639312744,
+      "learning_rate": 5.645161290322582e-05,
+      "loss": 2.1876,
+      "step": 280
+    },
+    {
+      "epoch": 0.05852674066599395,
+      "grad_norm": 2.6516835689544678,
+      "learning_rate": 5.8467741935483876e-05,
+      "loss": 2.0738,
+      "step": 290
+    },
+    {
+      "epoch": 0.060544904137235116,
+      "grad_norm": 2.234955310821533,
+      "learning_rate": 6.048387096774194e-05,
+      "loss": 2.1074,
+      "step": 300
+    },
+    {
+      "epoch": 0.06256306760847628,
+      "grad_norm": 2.285836935043335,
+      "learning_rate": 6.25e-05,
+      "loss": 2.1229,
+      "step": 310
+    },
+    {
+      "epoch": 0.06458123107971746,
+      "grad_norm": 2.150893449783325,
+      "learning_rate": 6.451612903225807e-05,
+      "loss": 2.0795,
+      "step": 320
+    },
+    {
+      "epoch": 0.06659939455095863,
+      "grad_norm": 2.7134313583374023,
+      "learning_rate": 6.653225806451613e-05,
+      "loss": 2.0569,
+      "step": 330
+    },
+    {
+      "epoch": 0.06861755802219979,
+      "grad_norm": 2.2563529014587402,
+      "learning_rate": 6.854838709677419e-05,
+      "loss": 2.054,
+      "step": 340
+    },
+    {
+      "epoch": 0.07063572149344097,
+      "grad_norm": 2.307612419128418,
+      "learning_rate": 7.056451612903226e-05,
+      "loss": 2.2254,
+      "step": 350
+    },
+    {
+      "epoch": 0.07265388496468214,
+      "grad_norm": 2.4022858142852783,
+      "learning_rate": 7.258064516129033e-05,
+      "loss": 2.0501,
+      "step": 360
+    },
+    {
+      "epoch": 0.07467204843592332,
+      "grad_norm": 1.9709093570709229,
+      "learning_rate": 7.45967741935484e-05,
+      "loss": 2.0284,
+      "step": 370
+    },
+    {
+      "epoch": 0.07669021190716448,
+      "grad_norm": 1.8165977001190186,
+      "learning_rate": 7.661290322580645e-05,
+      "loss": 2.0161,
+      "step": 380
+    },
+    {
+      "epoch": 0.07870837537840565,
+      "grad_norm": 2.2313830852508545,
+      "learning_rate": 7.862903225806451e-05,
+      "loss": 2.0064,
+      "step": 390
+    },
+    {
+      "epoch": 0.08072653884964683,
+      "grad_norm": 4.013941764831543,
+      "learning_rate": 8.064516129032258e-05,
+      "loss": 2.0056,
+      "step": 400
+    },
+    {
+      "epoch": 0.08274470232088799,
+      "grad_norm": 2.2855305671691895,
+      "learning_rate": 8.266129032258066e-05,
+      "loss": 2.0044,
+      "step": 410
+    },
+    {
+      "epoch": 0.08476286579212916,
+      "grad_norm": 1.9798551797866821,
+      "learning_rate": 8.467741935483871e-05,
+      "loss": 1.9908,
+      "step": 420
+    },
+    {
+      "epoch": 0.08678102926337034,
+      "grad_norm": 2.0610132217407227,
+      "learning_rate": 8.669354838709678e-05,
+      "loss": 1.9845,
+      "step": 430
+    },
+    {
+      "epoch": 0.08879919273461151,
+      "grad_norm": 2.354846477508545,
+      "learning_rate": 8.870967741935484e-05,
+      "loss": 2.0008,
+      "step": 440
+    },
+    {
+      "epoch": 0.09081735620585267,
+      "grad_norm": 2.430293560028076,
+      "learning_rate": 9.072580645161291e-05,
+      "loss": 1.9855,
+      "step": 450
+    },
+    {
+      "epoch": 0.09283551967709384,
+      "grad_norm": 2.4229743480682373,
+      "learning_rate": 9.274193548387096e-05,
+      "loss": 1.935,
+      "step": 460
+    },
+    {
+      "epoch": 0.09485368314833502,
+      "grad_norm": 1.9692800045013428,
+      "learning_rate": 9.475806451612904e-05,
+      "loss": 2.1015,
+      "step": 470
+    },
+    {
+      "epoch": 0.09687184661957618,
+      "grad_norm": 2.103461503982544,
+      "learning_rate": 9.677419354838711e-05,
+      "loss": 2.0086,
+      "step": 480
+    },
+    {
+      "epoch": 0.09889001009081735,
+      "grad_norm": 1.9599565267562866,
+      "learning_rate": 9.879032258064517e-05,
+      "loss": 2.0889,
+      "step": 490
+    },
+    {
+      "epoch": 0.10090817356205853,
+      "grad_norm": 2.226715087890625,
+      "learning_rate": 9.999995545373623e-05,
+      "loss": 1.9485,
+      "step": 500
+    },
+    {
+      "epoch": 0.1029263370332997,
+      "grad_norm": 2.138767719268799,
+      "learning_rate": 9.999945430918042e-05,
+      "loss": 1.8391,
+      "step": 510
+    },
+    {
+      "epoch": 0.10494450050454086,
+      "grad_norm": 1.9570167064666748,
+      "learning_rate": 9.999839634283869e-05,
+      "loss": 2.004,
+      "step": 520
+    },
+    {
+      "epoch": 0.10696266397578204,
+      "grad_norm": 2.2599446773529053,
+      "learning_rate": 9.999678156649317e-05,
+      "loss": 2.0093,
+      "step": 530
+    },
+    {
+      "epoch": 0.10898082744702321,
+      "grad_norm": 1.9920490980148315,
+      "learning_rate": 9.999460999812691e-05,
+      "loss": 1.8934,
+      "step": 540
+    },
+    {
+      "epoch": 0.11099899091826437,
+      "grad_norm": 1.9292041063308716,
+      "learning_rate": 9.999188166192368e-05,
+      "loss": 1.8937,
+      "step": 550
+    },
+    {
+      "epoch": 0.11301715438950555,
+      "grad_norm": 1.9816484451293945,
+      "learning_rate": 9.998859658826777e-05,
+      "loss": 1.9489,
+      "step": 560
+    },
+    {
+      "epoch": 0.11503531786074672,
+      "grad_norm": 2.503206729888916,
+      "learning_rate": 9.998475481374358e-05,
+      "loss": 1.9871,
+      "step": 570
+    },
+    {
+      "epoch": 0.1170534813319879,
+      "grad_norm": 1.8593742847442627,
+      "learning_rate": 9.998035638113527e-05,
+      "loss": 1.8784,
+      "step": 580
+    },
+    {
+      "epoch": 0.11907164480322906,
+      "grad_norm": 2.2534918785095215,
+      "learning_rate": 9.997540133942624e-05,
+      "loss": 1.7789,
+      "step": 590
+    },
+    {
+      "epoch": 0.12108980827447023,
+      "grad_norm": 1.7839148044586182,
+      "learning_rate": 9.996988974379857e-05,
+      "loss": 1.8833,
+      "step": 600
+    },
+    {
+      "epoch": 0.1231079717457114,
+      "grad_norm": 1.9832431077957153,
+      "learning_rate": 9.996382165563247e-05,
+      "loss": 1.8618,
+      "step": 610
+    },
+    {
+      "epoch": 0.12512613521695257,
+      "grad_norm": 1.8574587106704712,
+      "learning_rate": 9.995719714250556e-05,
+      "loss": 1.9742,
+      "step": 620
+    },
+    {
+      "epoch": 0.12714429868819374,
+      "grad_norm": 2.3903088569641113,
+      "learning_rate": 9.995001627819211e-05,
+      "loss": 1.8395,
+      "step": 630
+    },
+    {
+      "epoch": 0.12916246215943492,
+      "grad_norm": 1.8283036947250366,
+      "learning_rate": 9.99422791426622e-05,
+      "loss": 1.7997,
+      "step": 640
+    },
+    {
+      "epoch": 0.1311806256306761,
+      "grad_norm": 2.3620362281799316,
+      "learning_rate": 9.993398582208093e-05,
+      "loss": 1.8091,
+      "step": 650
+    },
+    {
+      "epoch": 0.13319878910191726,
+      "grad_norm": 1.9973320960998535,
+      "learning_rate": 9.99251364088073e-05,
+      "loss": 1.8977,
+      "step": 660
+    },
+    {
+      "epoch": 0.13521695257315844,
+      "grad_norm": 2.1321332454681396,
+      "learning_rate": 9.991573100139334e-05,
+      "loss": 1.8635,
+      "step": 670
+    },
+    {
+      "epoch": 0.13723511604439959,
+      "grad_norm": 2.079270362854004,
+      "learning_rate": 9.990576970458285e-05,
+      "loss": 1.8001,
+      "step": 680
+    },
+    {
+      "epoch": 0.13925327951564076,
+      "grad_norm": 2.42185640335083,
+      "learning_rate": 9.989525262931045e-05,
+      "loss": 1.8191,
+      "step": 690
+    },
+    {
+      "epoch": 0.14127144298688193,
+      "grad_norm": 1.972676396369934,
+      "learning_rate": 9.988417989270011e-05,
+      "loss": 1.8399,
+      "step": 700
+    },
+    {
+      "epoch": 0.1432896064581231,
+      "grad_norm": 1.9847174882888794,
+      "learning_rate": 9.987255161806402e-05,
+      "loss": 1.7614,
+      "step": 710
+    },
+    {
+      "epoch": 0.14530776992936428,
+      "grad_norm": 1.9572982788085938,
+      "learning_rate": 9.986036793490112e-05,
+      "loss": 1.8777,
+      "step": 720
+    },
+    {
+      "epoch": 0.14732593340060546,
+      "grad_norm": 1.9790279865264893,
+      "learning_rate": 9.984762897889568e-05,
+      "loss": 1.8118,
+      "step": 730
+    },
+    {
+      "epoch": 0.14934409687184663,
+      "grad_norm": 1.9155077934265137,
+      "learning_rate": 9.983433489191581e-05,
+      "loss": 1.8436,
+      "step": 740
+    },
+    {
+      "epoch": 0.15136226034308778,
+      "grad_norm": 1.9012539386749268,
+      "learning_rate": 9.98204858220119e-05,
+      "loss": 1.8109,
+      "step": 750
+    },
+    {
+      "epoch": 0.15338042381432895,
+      "grad_norm": 2.0491812229156494,
+      "learning_rate": 9.980608192341488e-05,
+      "loss": 1.8072,
+      "step": 760
+    },
+    {
+      "epoch": 0.15539858728557013,
+      "grad_norm": 1.941206932067871,
+      "learning_rate": 9.979112335653462e-05,
+      "loss": 1.7898,
+      "step": 770
+    },
+    {
+      "epoch": 0.1574167507568113,
+      "grad_norm": 2.0589675903320312,
+      "learning_rate": 9.977561028795803e-05,
+      "loss": 1.7524,
+      "step": 780
+    },
+    {
+      "epoch": 0.15943491422805248,
+      "grad_norm": 1.9348995685577393,
+      "learning_rate": 9.97595428904473e-05,
+      "loss": 1.7864,
+      "step": 790
+    },
+    {
+      "epoch": 0.16145307769929365,
+      "grad_norm": 1.8057056665420532,
+      "learning_rate": 9.974292134293792e-05,
+      "loss": 1.7851,
+      "step": 800
+    },
+    {
+      "epoch": 0.16347124117053483,
+      "grad_norm": 1.8641736507415771,
+      "learning_rate": 9.97257458305367e-05,
+      "loss": 1.8094,
+      "step": 810
+    },
+    {
+      "epoch": 0.16548940464177597,
+      "grad_norm": 1.713681936264038,
+      "learning_rate": 9.970801654451973e-05,
+      "loss": 1.7454,
+      "step": 820
+    },
+    {
+      "epoch": 0.16750756811301715,
+      "grad_norm": 1.8647186756134033,
+      "learning_rate": 9.968973368233022e-05,
+      "loss": 1.7415,
+      "step": 830
+    },
+    {
+      "epoch": 0.16952573158425832,
+      "grad_norm": 1.9415329694747925,
+      "learning_rate": 9.96708974475763e-05,
+      "loss": 1.7501,
+      "step": 840
+    },
+    {
+      "epoch": 0.1715438950554995,
+      "grad_norm": 1.8891992568969727,
+      "learning_rate": 9.965150805002878e-05,
+      "loss": 1.8218,
+      "step": 850
+    },
+    {
+      "epoch": 0.17356205852674067,
+      "grad_norm": 1.9533817768096924,
+      "learning_rate": 9.963156570561878e-05,
+      "loss": 1.6947,
+      "step": 860
+    },
+    {
+      "epoch": 0.17558022199798184,
+      "grad_norm": 2.1692941188812256,
+      "learning_rate": 9.96110706364354e-05,
+      "loss": 1.784,
+      "step": 870
+    },
+    {
+      "epoch": 0.17759838546922302,
+      "grad_norm": 2.229025363922119,
+      "learning_rate": 9.959002307072312e-05,
+      "loss": 1.7266,
+      "step": 880
+    },
+    {
+      "epoch": 0.17961654894046417,
+      "grad_norm": 3.380396604537964,
+      "learning_rate": 9.956842324287936e-05,
+      "loss": 1.9071,
+      "step": 890
+    },
+    {
+      "epoch": 0.18163471241170534,
+      "grad_norm": 1.9694714546203613,
+      "learning_rate": 9.954627139345186e-05,
+      "loss": 1.7442,
+      "step": 900
+    },
+    {
+      "epoch": 0.18365287588294651,
+      "grad_norm": 1.9261081218719482,
+      "learning_rate": 9.952356776913594e-05,
+      "loss": 1.8254,
+      "step": 910
+    },
+    {
+      "epoch": 0.1856710393541877,
+      "grad_norm": 1.6406139135360718,
+      "learning_rate": 9.950031262277183e-05,
+      "loss": 1.8027,
+      "step": 920
+    },
+    {
+      "epoch": 0.18768920282542886,
+      "grad_norm": 1.874161720275879,
+      "learning_rate": 9.947650621334179e-05,
+      "loss": 1.8027,
+      "step": 930
+    },
+    {
+      "epoch": 0.18970736629667004,
+      "grad_norm": 2.183828115463257,
+      "learning_rate": 9.945214880596725e-05,
+      "loss": 1.7398,
+      "step": 940
+    },
+    {
+      "epoch": 0.1917255297679112,
+      "grad_norm": 1.954749584197998,
+      "learning_rate": 9.94272406719059e-05,
+      "loss": 1.7675,
+      "step": 950
+    },
+    {
+      "epoch": 0.19374369323915236,
+      "grad_norm": 1.966210126876831,
+      "learning_rate": 9.940178208854858e-05,
+      "loss": 1.7841,
+      "step": 960
+    },
+    {
+      "epoch": 0.19576185671039353,
+      "grad_norm": 2.180774688720703,
+      "learning_rate": 9.937577333941626e-05,
+      "loss": 1.6492,
+      "step": 970
+    },
+    {
+      "epoch": 0.1977800201816347,
+      "grad_norm": 1.921655297279358,
+      "learning_rate": 9.934921471415687e-05,
+      "loss": 1.7661,
+      "step": 980
+    },
+    {
+      "epoch": 0.19979818365287588,
+      "grad_norm": 1.8591227531433105,
+      "learning_rate": 9.932210650854205e-05,
+      "loss": 1.7543,
+      "step": 990
+    },
+    {
+      "epoch": 0.20181634712411706,
+      "grad_norm": 1.6740055084228516,
+      "learning_rate": 9.929444902446392e-05,
+      "loss": 1.7578,
+      "step": 1000
+    },
+    {
+      "epoch": 0.20383451059535823,
+      "grad_norm": 2.937347650527954,
+      "learning_rate": 9.92662425699316e-05,
+      "loss": 1.7566,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2058526740665994,
+      "grad_norm": 1.9954780340194702,
+      "learning_rate": 9.923748745906789e-05,
+      "loss": 1.8376,
+      "step": 1020
+    },
+    {
+      "epoch": 0.20787083753784055,
+      "grad_norm": 2.204568386077881,
+      "learning_rate": 9.920818401210574e-05,
+      "loss": 1.7558,
+      "step": 1030
+    },
+    {
+      "epoch": 0.20988900100908173,
+      "grad_norm": 2.1600749492645264,
+      "learning_rate": 9.917833255538467e-05,
+      "loss": 1.8082,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2119071644803229,
+      "grad_norm": 2.097487449645996,
+      "learning_rate": 9.914793342134711e-05,
+      "loss": 1.7874,
+      "step": 1050
+    },
+    {
+      "epoch": 0.21392532795156408,
+      "grad_norm": 1.796174168586731,
+      "learning_rate": 9.911698694853477e-05,
+      "loss": 1.7015,
+      "step": 1060
+    },
+    {
+      "epoch": 0.21594349142280525,
+      "grad_norm": 1.9651933908462524,
+      "learning_rate": 9.908549348158485e-05,
+      "loss": 1.8868,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21796165489404642,
+      "grad_norm": 2.083768367767334,
+      "learning_rate": 9.905345337122609e-05,
+      "loss": 1.8338,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2199798183652876,
+      "grad_norm": 1.953058123588562,
+      "learning_rate": 9.902086697427504e-05,
+      "loss": 1.792,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22199798183652875,
+      "grad_norm": 1.9650559425354004,
+      "learning_rate": 9.8987734653632e-05,
+      "loss": 1.6179,
+      "step": 1100
+    },
+    {
+      "epoch": 0.22401614530776992,
+      "grad_norm": 1.7365285158157349,
+      "learning_rate": 9.895405677827692e-05,
+      "loss": 1.6512,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2260343087790111,
+      "grad_norm": 1.7468541860580444,
+      "learning_rate": 9.89198337232654e-05,
+      "loss": 1.6747,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22805247225025227,
+      "grad_norm": 1.967139720916748,
+      "learning_rate": 9.888506586972446e-05,
+      "loss": 1.7086,
+      "step": 1130
+    },
+    {
+      "epoch": 0.23007063572149344,
+      "grad_norm": 1.824622631072998,
+      "learning_rate": 9.884975360484827e-05,
+      "loss": 1.887,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23208879919273462,
+      "grad_norm": 2.1145553588867188,
+      "learning_rate": 9.881389732189392e-05,
+      "loss": 1.7733,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2341069626639758,
+      "grad_norm": 1.6385475397109985,
+      "learning_rate": 9.877749742017694e-05,
+      "loss": 1.6961,
+      "step": 1160
+    },
+    {
+      "epoch": 0.23612512613521694,
+      "grad_norm": 1.9020243883132935,
+      "learning_rate": 9.874055430506691e-05,
+      "loss": 1.6124,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23814328960645811,
+      "grad_norm": 1.817199945449829,
+      "learning_rate": 9.870306838798297e-05,
+      "loss": 1.6601,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2401614530776993,
+      "grad_norm": 2.001286268234253,
+      "learning_rate": 9.866504008638917e-05,
+      "loss": 1.7485,
+      "step": 1190
+    },
+    {
+      "epoch": 0.24217961654894046,
+      "grad_norm": 1.9406825304031372,
+      "learning_rate": 9.862646982378987e-05,
+      "loss": 1.8185,
+      "step": 1200
+    },
+    {
+      "epoch": 0.24419778002018164,
+      "grad_norm": 1.7781312465667725,
+      "learning_rate": 9.8587358029725e-05,
+      "loss": 1.751,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2462159434914228,
+      "grad_norm": 1.793482780456543,
+      "learning_rate": 9.854770513976531e-05,
+      "loss": 1.6768,
+      "step": 1220
+    },
+    {
+      "epoch": 0.248234106962664,
+      "grad_norm": 1.9680509567260742,
+      "learning_rate": 9.850751159550746e-05,
+      "loss": 1.6726,
+      "step": 1230
+    },
+    {
+      "epoch": 0.25025227043390513,
+      "grad_norm": 2.0250823497772217,
+      "learning_rate": 9.846677784456918e-05,
+      "loss": 1.7675,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2522704339051463,
+      "grad_norm": 1.7494395971298218,
+      "learning_rate": 9.842550434058421e-05,
+      "loss": 1.6584,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2542885973763875,
+      "grad_norm": 2.1690242290496826,
+      "learning_rate": 9.838369154319728e-05,
+      "loss": 1.802,
+      "step": 1260
+    },
+    {
+      "epoch": 0.25630676084762866,
+      "grad_norm": 1.6883882284164429,
+      "learning_rate": 9.8341339918059e-05,
+      "loss": 1.5858,
+      "step": 1270
+    },
+    {
+      "epoch": 0.25832492431886983,
+      "grad_norm": 1.9592852592468262,
+      "learning_rate": 9.82984499368207e-05,
+      "loss": 1.7107,
+      "step": 1280
+    },
+    {
+      "epoch": 0.260343087790111,
+      "grad_norm": 1.9985136985778809,
+      "learning_rate": 9.825502207712909e-05,
+      "loss": 1.674,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2623612512613522,
+      "grad_norm": 1.6702767610549927,
+      "learning_rate": 9.821105682262099e-05,
+      "loss": 1.6821,
+      "step": 1300
+    },
+    {
+      "epoch": 0.26437941473259335,
+      "grad_norm": 1.7378309965133667,
+      "learning_rate": 9.816655466291803e-05,
+      "loss": 1.7136,
+      "step": 1310
+    },
+    {
+      "epoch": 0.26639757820383453,
+      "grad_norm": 1.7488417625427246,
+      "learning_rate": 9.812151609362102e-05,
+      "loss": 1.5944,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2684157416750757,
+      "grad_norm": 1.7831872701644897,
+      "learning_rate": 9.807594161630458e-05,
+      "loss": 1.575,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2704339051463169,
+      "grad_norm": 2.0005006790161133,
+      "learning_rate": 9.802983173851149e-05,
+      "loss": 1.6744,
+      "step": 1340
+    },
+    {
+      "epoch": 0.272452068617558,
+      "grad_norm": 1.5925371646881104,
+      "learning_rate": 9.798318697374702e-05,
+      "loss": 1.6679,
+      "step": 1350
+    },
+    {
+      "epoch": 0.27447023208879917,
+      "grad_norm": 1.6449369192123413,
+      "learning_rate": 9.79360078414733e-05,
+      "loss": 1.6395,
+      "step": 1360
+    },
+    {
+      "epoch": 0.27648839556004035,
+      "grad_norm": 1.870029091835022,
+      "learning_rate": 9.78882948671034e-05,
+      "loss": 1.6746,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2785065590312815,
+      "grad_norm": 1.8943411111831665,
+      "learning_rate": 9.784004858199563e-05,
+      "loss": 1.7118,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2805247225025227,
+      "grad_norm": 1.9735866785049438,
+      "learning_rate": 9.779126952344748e-05,
+      "loss": 1.6215,
+      "step": 1390
+    },
+    {
+      "epoch": 0.28254288597376387,
+      "grad_norm": 1.719292402267456,
+      "learning_rate": 9.774195823468973e-05,
+      "loss": 1.6028,
+      "step": 1400
+    },
+    {
+      "epoch": 0.28456104944500504,
+      "grad_norm": 1.7490273714065552,
+      "learning_rate": 9.769211526488038e-05,
+      "loss": 1.5878,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2865792129162462,
+      "grad_norm": 1.621436357498169,
+      "learning_rate": 9.764174116909852e-05,
+      "loss": 1.7188,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2885973763874874,
+      "grad_norm": 1.7977508306503296,
+      "learning_rate": 9.759083650833815e-05,
+      "loss": 1.6966,
+      "step": 1430
+    },
+    {
+      "epoch": 0.29061553985872857,
+      "grad_norm": 1.6182868480682373,
+      "learning_rate": 9.753940184950192e-05,
+      "loss": 1.6072,
+      "step": 1440
+    },
+    {
+      "epoch": 0.29263370332996974,
+      "grad_norm": 2.094125747680664,
+      "learning_rate": 9.748743776539488e-05,
+      "loss": 1.7654,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2946518668012109,
+      "grad_norm": 2.0340657234191895,
+      "learning_rate": 9.743494483471801e-05,
+      "loss": 1.5987,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2966700302724521,
+      "grad_norm": 1.8374106884002686,
+      "learning_rate": 9.738192364206185e-05,
+      "loss": 1.6468,
+      "step": 1470
+    },
+    {
+      "epoch": 0.29868819374369326,
+      "grad_norm": 1.8509396314620972,
+      "learning_rate": 9.732837477789993e-05,
+      "loss": 1.6514,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3007063572149344,
+      "grad_norm": 2.0533978939056396,
+      "learning_rate": 9.727429883858227e-05,
+      "loss": 1.6375,
+      "step": 1490
+    },
+    {
+      "epoch": 0.30272452068617556,
+      "grad_norm": 1.7004354000091553,
+      "learning_rate": 9.721969642632865e-05,
+      "loss": 1.5852,
+      "step": 1500
+    },
+    {
+      "epoch": 0.30474268415741673,
+      "grad_norm": 1.8703371286392212,
+      "learning_rate": 9.716456814922196e-05,
+      "loss": 1.5644,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3067608476286579,
+      "grad_norm": 1.6673510074615479,
+      "learning_rate": 9.710891462120141e-05,
+      "loss": 1.6704,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3087790110998991,
+      "grad_norm": 1.8909664154052734,
+      "learning_rate": 9.70527364620557e-05,
+      "loss": 1.7009,
+      "step": 1530
+    },
+    {
+      "epoch": 0.31079717457114026,
+      "grad_norm": 1.6285436153411865,
+      "learning_rate": 9.699603429741615e-05,
+      "loss": 1.6874,
+      "step": 1540
+    },
+    {
+      "epoch": 0.31281533804238143,
+      "grad_norm": 1.7128878831863403,
+      "learning_rate": 9.693880875874961e-05,
+      "loss": 1.8054,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3148335015136226,
+      "grad_norm": 2.0219228267669678,
+      "learning_rate": 9.68810604833516e-05,
+      "loss": 1.5689,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3168516649848638,
+      "grad_norm": 1.5529290437698364,
+      "learning_rate": 9.682279011433908e-05,
+      "loss": 1.7903,
+      "step": 1570
+    },
+    {
+      "epoch": 0.31886982845610495,
+      "grad_norm": 1.7940829992294312,
+      "learning_rate": 9.676399830064339e-05,
+      "loss": 1.5716,
+      "step": 1580
+    },
+    {
+      "epoch": 0.32088799192734613,
+      "grad_norm": 1.6057904958724976,
+      "learning_rate": 9.670468569700288e-05,
+      "loss": 1.6821,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3229061553985873,
+      "grad_norm": 1.8908942937850952,
+      "learning_rate": 9.664485296395578e-05,
+      "loss": 1.541,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3249243188698285,
+      "grad_norm": 1.8195271492004395,
+      "learning_rate": 9.658450076783274e-05,
+      "loss": 1.65,
+      "step": 1610
+    },
+    {
+      "epoch": 0.32694248234106965,
+      "grad_norm": 6.518181800842285,
+      "learning_rate": 9.652362978074947e-05,
+      "loss": 1.6047,
+      "step": 1620
+    },
+    {
+      "epoch": 0.32896064581231077,
+      "grad_norm": 1.5660922527313232,
+      "learning_rate": 9.646224068059917e-05,
+      "loss": 1.7102,
+      "step": 1630
+    },
+    {
+      "epoch": 0.33097880928355194,
+      "grad_norm": 1.6389318704605103,
+      "learning_rate": 9.640033415104508e-05,
+      "loss": 1.6255,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3329969727547931,
+      "grad_norm": 1.6905416250228882,
+      "learning_rate": 9.633791088151283e-05,
+      "loss": 1.5718,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3350151362260343,
+      "grad_norm": 1.7866181135177612,
+      "learning_rate": 9.627497156718271e-05,
+      "loss": 1.7042,
+      "step": 1660
+    },
+    {
+      "epoch": 0.33703329969727547,
+      "grad_norm": 1.7515119314193726,
+      "learning_rate": 9.621151690898203e-05,
+      "loss": 1.6239,
+      "step": 1670
+    },
+    {
+      "epoch": 0.33905146316851664,
+      "grad_norm": 1.9881833791732788,
+      "learning_rate": 9.614754761357718e-05,
+      "loss": 1.7982,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3410696266397578,
+      "grad_norm": 1.8311055898666382,
+      "learning_rate": 9.608306439336592e-05,
+      "loss": 1.7399,
+      "step": 1690
+    },
+    {
+      "epoch": 0.343087790110999,
+      "grad_norm": 1.688659906387329,
+      "learning_rate": 9.60180679664693e-05,
+      "loss": 1.6666,
+      "step": 1700
+    },
+    {
+      "epoch": 0.34510595358224017,
+      "grad_norm": 1.8084384202957153,
+      "learning_rate": 9.595255905672377e-05,
+      "loss": 1.5487,
+      "step": 1710
+    },
+    {
+      "epoch": 0.34712411705348134,
+      "grad_norm": 1.7272626161575317,
+      "learning_rate": 9.588653839367302e-05,
+      "loss": 1.551,
+      "step": 1720
+    },
+    {
+      "epoch": 0.3491422805247225,
+      "grad_norm": 1.7780117988586426,
+      "learning_rate": 9.582000671256e-05,
+      "loss": 1.6598,
+      "step": 1730
+    },
+    {
+      "epoch": 0.3511604439959637,
+      "grad_norm": 2.405978202819824,
+      "learning_rate": 9.575296475431855e-05,
+      "loss": 1.6297,
+      "step": 1740
+    },
+    {
+      "epoch": 0.35317860746720486,
+      "grad_norm": 1.6499069929122925,
+      "learning_rate": 9.568541326556527e-05,
+      "loss": 1.5609,
+      "step": 1750
+    },
+    {
+      "epoch": 0.35519677093844604,
+      "grad_norm": 1.7063804864883423,
+      "learning_rate": 9.56173529985912e-05,
+      "loss": 1.6281,
+      "step": 1760
+    },
+    {
+      "epoch": 0.35721493440968716,
+      "grad_norm": 2.427915573120117,
+      "learning_rate": 9.554878471135339e-05,
+      "loss": 1.6166,
+      "step": 1770
+    },
+    {
+      "epoch": 0.35923309788092833,
+      "grad_norm": 1.7606953382492065,
+      "learning_rate": 9.547970916746649e-05,
+      "loss": 1.6152,
+      "step": 1780
+    },
+    {
+      "epoch": 0.3612512613521695,
+      "grad_norm": 1.7059513330459595,
+      "learning_rate": 9.541012713619428e-05,
+      "loss": 1.5078,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3632694248234107,
+      "grad_norm": 1.4295872449874878,
+      "learning_rate": 9.5340039392441e-05,
+      "loss": 1.5864,
+      "step": 1800
+    },
+    {
+      "epoch": 0.36528758829465185,
+      "grad_norm": 1.8844209909439087,
+      "learning_rate": 9.526944671674286e-05,
+      "loss": 1.5401,
+      "step": 1810
+    },
+    {
+      "epoch": 0.36730575176589303,
+      "grad_norm": 1.6048489809036255,
+      "learning_rate": 9.51983498952592e-05,
+      "loss": 1.7246,
+      "step": 1820
+    },
+    {
+      "epoch": 0.3693239152371342,
+      "grad_norm": 2.0559234619140625,
+      "learning_rate": 9.512674971976385e-05,
+      "loss": 1.4579,
+      "step": 1830
+    },
+    {
+      "epoch": 0.3713420787083754,
+      "grad_norm": 1.7094260454177856,
+      "learning_rate": 9.505464698763629e-05,
+      "loss": 1.5574,
+      "step": 1840
+    },
+    {
+      "epoch": 0.37336024217961655,
+      "grad_norm": 1.5393606424331665,
+      "learning_rate": 9.49820425018527e-05,
+      "loss": 1.5389,
+      "step": 1850
+    },
+    {
+      "epoch": 0.3753784056508577,
+      "grad_norm": 2.1417794227600098,
+      "learning_rate": 9.49089370709771e-05,
+      "loss": 1.6108,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3773965691220989,
+      "grad_norm": 1.6064629554748535,
+      "learning_rate": 9.483533150915229e-05,
+      "loss": 1.6211,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3794147325933401,
+      "grad_norm": 2.044330596923828,
+      "learning_rate": 9.476122663609086e-05,
+      "loss": 1.6693,
+      "step": 1880
+    },
+    {
+      "epoch": 0.38143289606458125,
+      "grad_norm": 1.7574785947799683,
+      "learning_rate": 9.468662327706594e-05,
+      "loss": 1.5502,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3834510595358224,
+      "grad_norm": 1.8084129095077515,
+      "learning_rate": 9.461152226290212e-05,
+      "loss": 1.6606,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3854692230070636,
+      "grad_norm": 1.4581211805343628,
+      "learning_rate": 9.453592442996614e-05,
+      "loss": 1.5317,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3874873864783047,
+      "grad_norm": 1.8625178337097168,
+      "learning_rate": 9.445983062015761e-05,
+      "loss": 1.4944,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3895055499495459,
+      "grad_norm": 1.7608696222305298,
+      "learning_rate": 9.43832416808996e-05,
+      "loss": 1.5913,
+      "step": 1930
+    },
+    {
+      "epoch": 0.39152371342078707,
+      "grad_norm": 1.9058195352554321,
+      "learning_rate": 9.430615846512923e-05,
+      "loss": 1.6255,
+      "step": 1940
+    },
+    {
+      "epoch": 0.39354187689202824,
+      "grad_norm": 1.6042306423187256,
+      "learning_rate": 9.422858183128808e-05,
+      "loss": 1.6525,
+      "step": 1950
+    },
+    {
+      "epoch": 0.3955600403632694,
+      "grad_norm": 1.6958731412887573,
+      "learning_rate": 9.415051264331285e-05,
+      "loss": 1.4745,
+      "step": 1960
+    },
+    {
+      "epoch": 0.3975782038345106,
+      "grad_norm": 1.7777879238128662,
+      "learning_rate": 9.407195177062549e-05,
+      "loss": 1.5836,
+      "step": 1970
+    },
+    {
+      "epoch": 0.39959636730575177,
+      "grad_norm": 1.7249339818954468,
+      "learning_rate": 9.399290008812365e-05,
+      "loss": 1.4844,
+      "step": 1980
+    },
+    {
+      "epoch": 0.40161453077699294,
+      "grad_norm": 1.854981541633606,
+      "learning_rate": 9.391335847617093e-05,
+      "loss": 1.6211,
+      "step": 1990
+    },
+    {
+      "epoch": 0.4036326942482341,
+      "grad_norm": 1.4041742086410522,
+      "learning_rate": 9.383332782058705e-05,
+      "loss": 1.6664,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4056508577194753,
+      "grad_norm": 1.605724573135376,
+      "learning_rate": 9.375280901263796e-05,
+      "loss": 1.5706,
+      "step": 2010
+    },
+    {
+      "epoch": 0.40766902119071646,
+      "grad_norm": 1.7073047161102295,
+      "learning_rate": 9.367180294902603e-05,
+      "loss": 1.6047,
+      "step": 2020
+    },
+    {
+      "epoch": 0.40968718466195764,
+      "grad_norm": 1.5172196626663208,
+      "learning_rate": 9.359031053187988e-05,
+      "loss": 1.5207,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4117053481331988,
+      "grad_norm": 1.9370439052581787,
+      "learning_rate": 9.350833266874451e-05,
+      "loss": 1.5746,
+      "step": 2040
+    },
+    {
+      "epoch": 0.41372351160444,
+      "grad_norm": 2.1937825679779053,
+      "learning_rate": 9.342587027257104e-05,
+      "loss": 1.7112,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4157416750756811,
+      "grad_norm": 1.7659974098205566,
+      "learning_rate": 9.334292426170672e-05,
+      "loss": 1.6329,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4177598385469223,
+      "grad_norm": 1.5360724925994873,
+      "learning_rate": 9.325949555988452e-05,
+      "loss": 1.6289,
+      "step": 2070
+    },
+    {
+      "epoch": 0.41977800201816345,
+      "grad_norm": 1.4664136171340942,
+      "learning_rate": 9.317558509621296e-05,
+      "loss": 1.6237,
+      "step": 2080
+    },
+    {
+      "epoch": 0.42179616548940463,
+      "grad_norm": 1.5282961130142212,
+      "learning_rate": 9.309119380516573e-05,
+      "loss": 1.5247,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4238143289606458,
+      "grad_norm": 1.7818903923034668,
+      "learning_rate": 9.300632262657128e-05,
+      "loss": 1.6479,
+      "step": 2100
+    },
+    {
+      "epoch": 0.425832492431887,
+      "grad_norm": 1.8099353313446045,
+      "learning_rate": 9.292097250560232e-05,
+      "loss": 1.692,
+      "step": 2110
+    },
+    {
+      "epoch": 0.42785065590312815,
+      "grad_norm": 1.6518584489822388,
+      "learning_rate": 9.283514439276539e-05,
+      "loss": 1.5806,
+      "step": 2120
+    },
+    {
+      "epoch": 0.4298688193743693,
+      "grad_norm": 1.6262339353561401,
+      "learning_rate": 9.274883924389018e-05,
+      "loss": 1.6018,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4318869828456105,
+      "grad_norm": 1.858011245727539,
+      "learning_rate": 9.266205802011892e-05,
+      "loss": 1.6162,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4339051463168517,
+      "grad_norm": 1.5540958642959595,
+      "learning_rate": 9.257480168789565e-05,
+      "loss": 1.5558,
+      "step": 2150
+    },
+    {
+      "epoch": 0.43592330978809285,
+      "grad_norm": 1.5216200351715088,
+      "learning_rate": 9.248707121895555e-05,
+      "loss": 1.6317,
+      "step": 2160
+    },
+    {
+      "epoch": 0.437941473259334,
+      "grad_norm": 2.030132532119751,
+      "learning_rate": 9.239886759031398e-05,
+      "loss": 1.457,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4399596367305752,
+      "grad_norm": 1.6100101470947266,
+      "learning_rate": 9.231019178425573e-05,
+      "loss": 1.6372,
+      "step": 2180
+    },
+    {
+      "epoch": 0.4419778002018164,
+      "grad_norm": 1.6417200565338135,
+      "learning_rate": 9.222104478832398e-05,
+      "loss": 1.5867,
+      "step": 2190
+    },
+    {
+      "epoch": 0.4439959636730575,
+      "grad_norm": 1.5133955478668213,
+      "learning_rate": 9.213142759530936e-05,
+      "loss": 1.5338,
+      "step": 2200
+    },
+    {
+      "epoch": 0.44601412714429867,
+      "grad_norm": 1.543249249458313,
+      "learning_rate": 9.204134120323883e-05,
+      "loss": 1.6463,
+      "step": 2210
+    },
+    {
+      "epoch": 0.44803229061553984,
+      "grad_norm": 2.0193662643432617,
+      "learning_rate": 9.195078661536471e-05,
+      "loss": 1.5299,
+      "step": 2220
+    },
+    {
+      "epoch": 0.450050454086781,
+      "grad_norm": 1.4934704303741455,
+      "learning_rate": 9.185976484015333e-05,
+      "loss": 1.5422,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4520686175580222,
+      "grad_norm": 1.6684024333953857,
+      "learning_rate": 9.176827689127389e-05,
+      "loss": 1.62,
+      "step": 2240
+    },
+    {
+      "epoch": 0.45408678102926336,
+      "grad_norm": 1.7710318565368652,
+      "learning_rate": 9.167632378758719e-05,
+      "loss": 1.4557,
+      "step": 2250
+    },
+    {
+      "epoch": 0.45610494450050454,
+      "grad_norm": 1.717943549156189,
+      "learning_rate": 9.158390655313422e-05,
+      "loss": 1.601,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4581231079717457,
+      "grad_norm": 1.751428246498108,
+      "learning_rate": 9.149102621712482e-05,
+      "loss": 1.5032,
+      "step": 2270
+    },
+    {
+      "epoch": 0.4601412714429869,
+      "grad_norm": 1.8777353763580322,
+      "learning_rate": 9.139768381392616e-05,
+      "loss": 1.6255,
+      "step": 2280
+    },
+    {
+      "epoch": 0.46215943491422806,
+      "grad_norm": 1.7452517747879028,
+      "learning_rate": 9.130388038305127e-05,
+      "loss": 1.6209,
+      "step": 2290
+    },
+    {
+      "epoch": 0.46417759838546924,
+      "grad_norm": 1.7831089496612549,
+      "learning_rate": 9.12096169691474e-05,
+      "loss": 1.6401,
+      "step": 2300
+    },
+    {
+      "epoch": 0.4661957618567104,
+      "grad_norm": 1.591808795928955,
+      "learning_rate": 9.111489462198448e-05,
+      "loss": 1.5767,
+      "step": 2310
+    },
+    {
+      "epoch": 0.4682139253279516,
+      "grad_norm": 1.6915345191955566,
+      "learning_rate": 9.101971439644335e-05,
+      "loss": 1.556,
+      "step": 2320
+    },
+    {
+      "epoch": 0.47023208879919276,
+      "grad_norm": 1.8185473680496216,
+      "learning_rate": 9.092407735250404e-05,
+      "loss": 1.6477,
+      "step": 2330
+    },
+    {
+      "epoch": 0.4722502522704339,
+      "grad_norm": 1.7397855520248413,
+      "learning_rate": 9.082798455523396e-05,
+      "loss": 1.4885,
+      "step": 2340
+    },
+    {
+      "epoch": 0.47426841574167505,
+      "grad_norm": 1.7972854375839233,
+      "learning_rate": 9.073143707477607e-05,
+      "loss": 1.6802,
+      "step": 2350
+    },
+    {
+      "epoch": 0.47628657921291623,
+      "grad_norm": 1.7079814672470093,
+      "learning_rate": 9.063443598633688e-05,
+      "loss": 1.5201,
+      "step": 2360
+    },
+    {
+      "epoch": 0.4783047426841574,
+      "grad_norm": 1.7126438617706299,
+      "learning_rate": 9.053698237017459e-05,
+      "loss": 1.5861,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4803229061553986,
+      "grad_norm": 1.7915846109390259,
+      "learning_rate": 9.043907731158699e-05,
+      "loss": 1.5139,
+      "step": 2380
+    },
+    {
+      "epoch": 0.48234106962663975,
+      "grad_norm": 1.5365029573440552,
+      "learning_rate": 9.034072190089932e-05,
+      "loss": 1.5428,
+      "step": 2390
+    },
+    {
+      "epoch": 0.4843592330978809,
+      "grad_norm": 1.720226526260376,
+      "learning_rate": 9.02419172334523e-05,
+      "loss": 1.4767,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4863773965691221,
+      "grad_norm": 1.9600703716278076,
+      "learning_rate": 9.014266440958974e-05,
+      "loss": 1.6188,
+      "step": 2410
+    },
+    {
+      "epoch": 0.4883955600403633,
+      "grad_norm": 1.8720204830169678,
+      "learning_rate": 9.004296453464638e-05,
+      "loss": 1.5432,
+      "step": 2420
+    },
+    {
+      "epoch": 0.49041372351160445,
+      "grad_norm": 1.6241135597229004,
+      "learning_rate": 8.994281871893562e-05,
+      "loss": 1.5496,
+      "step": 2430
+    },
+    {
+      "epoch": 0.4924318869828456,
+      "grad_norm": 1.8902565240859985,
+      "learning_rate": 8.984222807773706e-05,
+      "loss": 1.6235,
+      "step": 2440
+    },
+    {
+      "epoch": 0.4944500504540868,
+      "grad_norm": 1.925911784172058,
+      "learning_rate": 8.974119373128411e-05,
+      "loss": 1.5734,
+      "step": 2450
+    },
+    {
+      "epoch": 0.496468213925328,
+      "grad_norm": 2.0046586990356445,
+      "learning_rate": 8.963971680475161e-05,
+      "loss": 1.5009,
+      "step": 2460
+    },
+    {
+      "epoch": 0.49848637739656915,
+      "grad_norm": 1.6803854703903198,
+      "learning_rate": 8.95377984282431e-05,
+      "loss": 1.5605,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5005045408678103,
+      "grad_norm": 1.5540610551834106,
+      "learning_rate": 8.943543973677846e-05,
+      "loss": 1.5961,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5025227043390514,
+      "grad_norm": 1.625119686126709,
+      "learning_rate": 8.933264187028109e-05,
+      "loss": 1.604,
+      "step": 2490
+    },
+    {
+      "epoch": 0.5045408678102926,
+      "grad_norm": 1.7552522420883179,
+      "learning_rate": 8.922940597356532e-05,
+      "loss": 1.4812,
+      "step": 2500
+    },
+    {
+      "epoch": 0.5065590312815338,
+      "grad_norm": 1.7819663286209106,
+      "learning_rate": 8.912573319632367e-05,
+      "loss": 1.6794,
+      "step": 2510
+    },
+    {
+      "epoch": 0.508577194752775,
+      "grad_norm": 1.715986728668213,
+      "learning_rate": 8.90216246931139e-05,
+      "loss": 1.5846,
+      "step": 2520
+    },
+    {
+      "epoch": 0.5105953582240161,
+      "grad_norm": 1.6441289186477661,
+      "learning_rate": 8.891708162334635e-05,
+      "loss": 1.5525,
+      "step": 2530
+    },
+    {
+      "epoch": 0.5126135216952573,
+      "grad_norm": 1.7777501344680786,
+      "learning_rate": 8.88121051512709e-05,
+      "loss": 1.5136,
+      "step": 2540
+    },
+    {
+      "epoch": 0.5146316851664985,
+      "grad_norm": 1.7654690742492676,
+      "learning_rate": 8.870669644596402e-05,
+      "loss": 1.574,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5166498486377397,
+      "grad_norm": 1.329132080078125,
+      "learning_rate": 8.860085668131582e-05,
+      "loss": 1.4299,
+      "step": 2560
+    },
+    {
+      "epoch": 0.5186680121089808,
+      "grad_norm": 1.7770514488220215,
+      "learning_rate": 8.84945870360169e-05,
+      "loss": 1.6012,
+      "step": 2570
+    },
+    {
+      "epoch": 0.520686175580222,
+      "grad_norm": 1.421560287475586,
+      "learning_rate": 8.838788869354522e-05,
+      "loss": 1.5075,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5227043390514632,
+      "grad_norm": 2.0325329303741455,
+      "learning_rate": 8.828076284215301e-05,
+      "loss": 1.4582,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5247225025227044,
+      "grad_norm": 2.303136110305786,
+      "learning_rate": 8.817321067485343e-05,
+      "loss": 1.6037,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5267406659939455,
+      "grad_norm": 1.5532690286636353,
+      "learning_rate": 8.806523338940736e-05,
+      "loss": 1.6264,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5287588294651867,
+      "grad_norm": 1.9267081022262573,
+      "learning_rate": 8.795683218831001e-05,
+      "loss": 1.6513,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5307769929364279,
+      "grad_norm": 1.4832985401153564,
+      "learning_rate": 8.78480082787776e-05,
+      "loss": 1.5968,
+      "step": 2630
+    },
+    {
+      "epoch": 0.5327951564076691,
+      "grad_norm": 1.6407015323638916,
+      "learning_rate": 8.773876287273377e-05,
+      "loss": 1.6084,
+      "step": 2640
+    },
+    {
+      "epoch": 0.5348133198789102,
+      "grad_norm": 1.6609562635421753,
+      "learning_rate": 8.762909718679629e-05,
+      "loss": 1.5557,
+      "step": 2650
+    },
+    {
+      "epoch": 0.5368314833501514,
+      "grad_norm": 1.6559169292449951,
+      "learning_rate": 8.751901244226332e-05,
+      "loss": 1.623,
+      "step": 2660
+    },
+    {
+      "epoch": 0.5388496468213926,
+      "grad_norm": 1.5652551651000977,
+      "learning_rate": 8.740850986509994e-05,
+      "loss": 1.4157,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5408678102926338,
+      "grad_norm": 1.9073171615600586,
+      "learning_rate": 8.729759068592442e-05,
+      "loss": 1.5152,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5428859737638748,
+      "grad_norm": 1.72683846950531,
+      "learning_rate": 8.718625613999457e-05,
+      "loss": 1.6011,
+      "step": 2690
+    },
+    {
+      "epoch": 0.544904137235116,
+      "grad_norm": 1.3139292001724243,
+      "learning_rate": 8.70745074671939e-05,
+      "loss": 1.5194,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5469223007063572,
+      "grad_norm": 1.6831001043319702,
+      "learning_rate": 8.696234591201793e-05,
+      "loss": 1.5078,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5489404641775983,
+      "grad_norm": 1.5346101522445679,
+      "learning_rate": 8.684977272356024e-05,
+      "loss": 1.4988,
+      "step": 2720
+    },
+    {
+      "epoch": 0.5509586276488395,
+      "grad_norm": 1.6062910556793213,
+      "learning_rate": 8.673678915549855e-05,
+      "loss": 1.6626,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5529767911200807,
+      "grad_norm": 1.7970744371414185,
+      "learning_rate": 8.662339646608089e-05,
+      "loss": 1.5251,
+      "step": 2740
+    },
+    {
+      "epoch": 0.5549949545913219,
+      "grad_norm": 1.687828540802002,
+      "learning_rate": 8.650959591811141e-05,
+      "loss": 1.5361,
+      "step": 2750
+    },
+    {
+      "epoch": 0.557013118062563,
+      "grad_norm": 1.8278907537460327,
+      "learning_rate": 8.639538877893644e-05,
+      "loss": 1.4754,
+      "step": 2760
+    },
+    {
+      "epoch": 0.5590312815338042,
+      "grad_norm": 1.5675766468048096,
+      "learning_rate": 8.628077632043032e-05,
+      "loss": 1.5059,
+      "step": 2770
+    },
+    {
+      "epoch": 0.5610494450050454,
+      "grad_norm": 1.5290266275405884,
+      "learning_rate": 8.616575981898125e-05,
+      "loss": 1.4684,
+      "step": 2780
+    },
+    {
+      "epoch": 0.5630676084762866,
+      "grad_norm": 1.3652130365371704,
+      "learning_rate": 8.605034055547709e-05,
+      "loss": 1.4736,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5650857719475277,
+      "grad_norm": 1.4992094039916992,
+      "learning_rate": 8.593451981529108e-05,
+      "loss": 1.5559,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5671039354187689,
+      "grad_norm": 1.5727661848068237,
+      "learning_rate": 8.581829888826754e-05,
+      "loss": 1.5884,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5691220988900101,
+      "grad_norm": 1.8026494979858398,
+      "learning_rate": 8.570167906870745e-05,
+      "loss": 1.5782,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5711402623612513,
+      "grad_norm": 1.701768159866333,
+      "learning_rate": 8.558466165535411e-05,
+      "loss": 1.598,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5731584258324924,
+      "grad_norm": 1.6463897228240967,
+      "learning_rate": 8.546724795137865e-05,
+      "loss": 1.4889,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5751765893037336,
+      "grad_norm": 1.5795738697052002,
+      "learning_rate": 8.534943926436554e-05,
+      "loss": 1.5306,
+      "step": 2850
+    },
+    {
+      "epoch": 0.5771947527749748,
+      "grad_norm": 1.5358326435089111,
+      "learning_rate": 8.523123690629791e-05,
+      "loss": 1.5801,
+      "step": 2860
+    },
+    {
+      "epoch": 0.579212916246216,
+      "grad_norm": 1.566599726676941,
+      "learning_rate": 8.511264219354313e-05,
+      "loss": 1.4904,
+      "step": 2870
+    },
+    {
+      "epoch": 0.5812310797174571,
+      "grad_norm": 1.3994381427764893,
+      "learning_rate": 8.4993656446838e-05,
+      "loss": 1.5097,
+      "step": 2880
+    },
+    {
+      "epoch": 0.5832492431886983,
+      "grad_norm": 1.5844892263412476,
+      "learning_rate": 8.48742809912741e-05,
+      "loss": 1.561,
+      "step": 2890
+    },
+    {
+      "epoch": 0.5852674066599395,
+      "grad_norm": 1.592240333557129,
+      "learning_rate": 8.475451715628302e-05,
+      "loss": 1.4621,
+      "step": 2900
+    },
+    {
+      "epoch": 0.5872855701311807,
+      "grad_norm": 1.835154414176941,
+      "learning_rate": 8.463436627562158e-05,
+      "loss": 1.5978,
+      "step": 2910
+    },
+    {
+      "epoch": 0.5893037336024218,
+      "grad_norm": 1.5988234281539917,
+      "learning_rate": 8.451382968735693e-05,
+      "loss": 1.4611,
+      "step": 2920
+    },
+    {
+      "epoch": 0.591321897073663,
+      "grad_norm": 1.551164984703064,
+      "learning_rate": 8.43929087338517e-05,
+      "loss": 1.4682,
+      "step": 2930
+    },
+    {
+      "epoch": 0.5933400605449042,
+      "grad_norm": 1.537986159324646,
+      "learning_rate": 8.4271604761749e-05,
+      "loss": 1.437,
+      "step": 2940
+    },
+    {
+      "epoch": 0.5953582240161454,
+      "grad_norm": 1.5694961547851562,
+      "learning_rate": 8.414991912195747e-05,
+      "loss": 1.5569,
+      "step": 2950
+    },
+    {
+      "epoch": 0.5973763874873865,
+      "grad_norm": 1.6652250289916992,
+      "learning_rate": 8.402785316963618e-05,
+      "loss": 1.4936,
+      "step": 2960
+    },
+    {
+      "epoch": 0.5993945509586277,
+      "grad_norm": 1.690019965171814,
+      "learning_rate": 8.390540826417964e-05,
+      "loss": 1.4642,
+      "step": 2970
+    },
+    {
+      "epoch": 0.6014127144298688,
+      "grad_norm": 1.532211422920227,
+      "learning_rate": 8.378258576920253e-05,
+      "loss": 1.5379,
+      "step": 2980
+    },
+    {
+      "epoch": 0.6034308779011099,
+      "grad_norm": 1.6568650007247925,
+      "learning_rate": 8.365938705252459e-05,
+      "loss": 1.4731,
+      "step": 2990
+    },
+    {
+      "epoch": 0.6054490413723511,
+      "grad_norm": 1.3985533714294434,
+      "learning_rate": 8.353581348615538e-05,
+      "loss": 1.5571,
+      "step": 3000
+    },
+    {
+      "epoch": 0.6074672048435923,
+      "grad_norm": 1.7050758600234985,
+      "learning_rate": 8.341186644627901e-05,
+      "loss": 1.6194,
+      "step": 3010
+    },
+    {
+      "epoch": 0.6094853683148335,
+      "grad_norm": 1.656028389930725,
+      "learning_rate": 8.32875473132388e-05,
+      "loss": 1.4723,
+      "step": 3020
+    },
+    {
+      "epoch": 0.6115035317860746,
+      "grad_norm": 1.4241358041763306,
+      "learning_rate": 8.316285747152189e-05,
+      "loss": 1.4154,
+      "step": 3030
+    },
+    {
+      "epoch": 0.6135216952573158,
+      "grad_norm": 1.6242072582244873,
+      "learning_rate": 8.30377983097438e-05,
+      "loss": 1.4525,
+      "step": 3040
+    },
+    {
+      "epoch": 0.615539858728557,
+      "grad_norm": 1.6200363636016846,
+      "learning_rate": 8.291237122063309e-05,
+      "loss": 1.5451,
+      "step": 3050
+    },
+    {
+      "epoch": 0.6175580221997982,
+      "grad_norm": 1.2775624990463257,
+      "learning_rate": 8.27865776010157e-05,
+      "loss": 1.454,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6195761856710393,
+      "grad_norm": 1.5615248680114746,
+      "learning_rate": 8.266041885179949e-05,
+      "loss": 1.4296,
+      "step": 3070
+    },
+    {
+      "epoch": 0.6215943491422805,
+      "grad_norm": 1.7214854955673218,
+      "learning_rate": 8.253389637795858e-05,
+      "loss": 1.4538,
+      "step": 3080
+    },
+    {
+      "epoch": 0.6236125126135217,
+      "grad_norm": 1.4547313451766968,
+      "learning_rate": 8.240701158851778e-05,
+      "loss": 1.4107,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6256306760847629,
+      "grad_norm": 1.4714813232421875,
+      "learning_rate": 8.227976589653676e-05,
+      "loss": 1.3942,
+      "step": 3100
+    },
+    {
+      "epoch": 0.627648839556004,
+      "grad_norm": 1.4874024391174316,
+      "learning_rate": 8.215216071909448e-05,
+      "loss": 1.5679,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6296670030272452,
+      "grad_norm": 1.5745984315872192,
+      "learning_rate": 8.202419747727333e-05,
+      "loss": 1.4826,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6316851664984864,
+      "grad_norm": 1.5688875913619995,
+      "learning_rate": 8.189587759614325e-05,
+      "loss": 1.4611,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6337033299697276,
+      "grad_norm": 1.5299983024597168,
+      "learning_rate": 8.176720250474594e-05,
+      "loss": 1.4565,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6357214934409687,
+      "grad_norm": 1.5063962936401367,
+      "learning_rate": 8.163817363607894e-05,
+      "loss": 1.5174,
+      "step": 3150
+    },
+    {
+      "epoch": 0.6377396569122099,
+      "grad_norm": 1.7101162672042847,
+      "learning_rate": 8.150879242707962e-05,
+      "loss": 1.4651,
+      "step": 3160
+    },
+    {
+      "epoch": 0.6397578203834511,
+      "grad_norm": 1.6039749383926392,
+      "learning_rate": 8.137906031860925e-05,
+      "loss": 1.5918,
+      "step": 3170
+    },
+    {
+      "epoch": 0.6417759838546923,
+      "grad_norm": 1.54131019115448,
+      "learning_rate": 8.124897875543684e-05,
+      "loss": 1.4304,
+      "step": 3180
+    },
+    {
+      "epoch": 0.6437941473259334,
+      "grad_norm": 1.422153353691101,
+      "learning_rate": 8.111854918622321e-05,
+      "loss": 1.6108,
+      "step": 3190
+    },
+    {
+      "epoch": 0.6458123107971746,
+      "grad_norm": 1.6443060636520386,
+      "learning_rate": 8.098777306350469e-05,
+      "loss": 1.4497,
+      "step": 3200
+    },
+    {
+      "epoch": 0.6478304742684158,
+      "grad_norm": 1.5704275369644165,
+      "learning_rate": 8.08566518436771e-05,
+      "loss": 1.5172,
+      "step": 3210
+    },
+    {
+      "epoch": 0.649848637739657,
+      "grad_norm": 1.7546700239181519,
+      "learning_rate": 8.072518698697938e-05,
+      "loss": 1.4498,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6518668012108981,
+      "grad_norm": 1.650565266609192,
+      "learning_rate": 8.059337995747743e-05,
+      "loss": 1.4536,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6538849646821393,
+      "grad_norm": 1.516605019569397,
+      "learning_rate": 8.046123222304781e-05,
+      "loss": 1.5499,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6559031281533805,
+      "grad_norm": 1.4544646739959717,
+      "learning_rate": 8.032874525536131e-05,
+      "loss": 1.4791,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6579212916246215,
+      "grad_norm": 1.590570092201233,
+      "learning_rate": 8.019592052986665e-05,
+      "loss": 1.3705,
+      "step": 3260
+    },
+    {
+      "epoch": 0.6599394550958627,
+      "grad_norm": 1.3931959867477417,
+      "learning_rate": 8.006275952577397e-05,
+      "loss": 1.5409,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6619576185671039,
+      "grad_norm": 1.676604986190796,
+      "learning_rate": 7.992926372603842e-05,
+      "loss": 1.4835,
+      "step": 3280
+    },
+    {
+      "epoch": 0.6639757820383451,
+      "grad_norm": 1.832785725593567,
+      "learning_rate": 7.979543461734362e-05,
+      "loss": 1.4715,
+      "step": 3290
+    },
+    {
+      "epoch": 0.6659939455095862,
+      "grad_norm": 1.3548346757888794,
+      "learning_rate": 7.966127369008512e-05,
+      "loss": 1.4553,
+      "step": 3300
+    },
+    {
+      "epoch": 0.6680121089808274,
+      "grad_norm": 1.397743821144104,
+      "learning_rate": 7.952678243835376e-05,
+      "loss": 1.4793,
+      "step": 3310
+    },
+    {
+      "epoch": 0.6700302724520686,
+      "grad_norm": 1.5748425722122192,
+      "learning_rate": 7.939196235991904e-05,
+      "loss": 1.4791,
+      "step": 3320
+    },
+    {
+      "epoch": 0.6720484359233098,
+      "grad_norm": 1.672951340675354,
+      "learning_rate": 7.925681495621253e-05,
+      "loss": 1.5467,
+      "step": 3330
+    },
+    {
+      "epoch": 0.6740665993945509,
+      "grad_norm": 1.3464049100875854,
+      "learning_rate": 7.912134173231098e-05,
+      "loss": 1.4887,
+      "step": 3340
+    },
+    {
+      "epoch": 0.6760847628657921,
+      "grad_norm": 1.761104941368103,
+      "learning_rate": 7.898554419691974e-05,
+      "loss": 1.4937,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6781029263370333,
+      "grad_norm": 1.5034103393554688,
+      "learning_rate": 7.884942386235582e-05,
+      "loss": 1.3636,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6801210898082745,
+      "grad_norm": 1.463498592376709,
+      "learning_rate": 7.871298224453113e-05,
+      "loss": 1.3987,
+      "step": 3370
+    },
+    {
+      "epoch": 0.6821392532795156,
+      "grad_norm": 1.4649510383605957,
+      "learning_rate": 7.857622086293557e-05,
+      "loss": 1.5976,
+      "step": 3380
+    },
+    {
+      "epoch": 0.6841574167507568,
+      "grad_norm": 1.6214101314544678,
+      "learning_rate": 7.843914124062006e-05,
+      "loss": 1.4602,
+      "step": 3390
+    },
+    {
+      "epoch": 0.686175580221998,
+      "grad_norm": 1.752223253250122,
+      "learning_rate": 7.830174490417972e-05,
+      "loss": 1.4979,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6881937436932392,
+      "grad_norm": 1.4902223348617554,
+      "learning_rate": 7.816403338373666e-05,
+      "loss": 1.4157,
+      "step": 3410
+    },
+    {
+      "epoch": 0.6902119071644803,
+      "grad_norm": 1.4046521186828613,
+      "learning_rate": 7.802600821292314e-05,
+      "loss": 1.3817,
+      "step": 3420
+    },
+    {
+      "epoch": 0.6922300706357215,
+      "grad_norm": 1.5104862451553345,
+      "learning_rate": 7.78876709288644e-05,
+      "loss": 1.4425,
+      "step": 3430
+    },
+    {
+      "epoch": 0.6942482341069627,
+      "grad_norm": 1.6517716646194458,
+      "learning_rate": 7.774902307216148e-05,
+      "loss": 1.5526,
+      "step": 3440
+    },
+    {
+      "epoch": 0.6962663975782039,
+      "grad_norm": 1.43543541431427,
+      "learning_rate": 7.76100661868742e-05,
+      "loss": 1.5054,
+      "step": 3450
+    },
+    {
+      "epoch": 0.698284561049445,
+      "grad_norm": 1.665265679359436,
+      "learning_rate": 7.747080182050388e-05,
+      "loss": 1.4123,
+      "step": 3460
+    },
+    {
+      "epoch": 0.7003027245206862,
+      "grad_norm": 1.5892359018325806,
+      "learning_rate": 7.733123152397609e-05,
+      "loss": 1.4904,
+      "step": 3470
+    },
+    {
+      "epoch": 0.7023208879919274,
+      "grad_norm": 1.5370794534683228,
+      "learning_rate": 7.719135685162342e-05,
+      "loss": 1.3999,
+      "step": 3480
+    },
+    {
+      "epoch": 0.7043390514631686,
+      "grad_norm": 1.509031891822815,
+      "learning_rate": 7.705117936116822e-05,
+      "loss": 1.5462,
+      "step": 3490
+    },
+    {
+      "epoch": 0.7063572149344097,
+      "grad_norm": 1.753881812095642,
+      "learning_rate": 7.691070061370507e-05,
+      "loss": 1.5522,
+      "step": 3500
+    },
+    {
+      "epoch": 0.7083753784056509,
+      "grad_norm": 1.5018798112869263,
+      "learning_rate": 7.676992217368364e-05,
+      "loss": 1.4837,
+      "step": 3510
+    },
+    {
+      "epoch": 0.7103935418768921,
+      "grad_norm": 1.5720365047454834,
+      "learning_rate": 7.662884560889105e-05,
+      "loss": 1.3675,
+      "step": 3520
+    },
+    {
+      "epoch": 0.7124117053481333,
+      "grad_norm": 1.3165663480758667,
+      "learning_rate": 7.648747249043457e-05,
+      "loss": 1.5472,
+      "step": 3530
+    },
+    {
+      "epoch": 0.7144298688193743,
+      "grad_norm": 1.523558497428894,
+      "learning_rate": 7.634580439272401e-05,
+      "loss": 1.5398,
+      "step": 3540
+    },
+    {
+      "epoch": 0.7164480322906155,
+      "grad_norm": 1.283341884613037,
+      "learning_rate": 7.620384289345425e-05,
+      "loss": 1.5009,
+      "step": 3550
+    },
+    {
+      "epoch": 0.7184661957618567,
+      "grad_norm": 1.6135417222976685,
+      "learning_rate": 7.606158957358769e-05,
+      "loss": 1.4926,
+      "step": 3560
+    },
+    {
+      "epoch": 0.7204843592330978,
+      "grad_norm": 1.5714365243911743,
+      "learning_rate": 7.591904601733655e-05,
+      "loss": 1.5098,
+      "step": 3570
+    },
+    {
+      "epoch": 0.722502522704339,
+      "grad_norm": 1.4282561540603638,
+      "learning_rate": 7.577621381214529e-05,
+      "loss": 1.4814,
+      "step": 3580
+    },
+    {
+      "epoch": 0.7245206861755802,
+      "grad_norm": 1.384590744972229,
+      "learning_rate": 7.563309454867295e-05,
+      "loss": 1.4716,
+      "step": 3590
+    },
+    {
+      "epoch": 0.7265388496468214,
+      "grad_norm": 1.3100378513336182,
+      "learning_rate": 7.548968982077542e-05,
+      "loss": 1.3972,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7285570131180625,
+      "grad_norm": 1.2014504671096802,
+      "learning_rate": 7.534600122548765e-05,
+      "loss": 1.392,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7305751765893037,
+      "grad_norm": 1.5689890384674072,
+      "learning_rate": 7.520203036300588e-05,
+      "loss": 1.4531,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7325933400605449,
+      "grad_norm": 1.7837287187576294,
+      "learning_rate": 7.505777883666993e-05,
+      "loss": 1.6061,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7346115035317861,
+      "grad_norm": 1.3635213375091553,
+      "learning_rate": 7.491324825294514e-05,
+      "loss": 1.4351,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7366296670030272,
+      "grad_norm": 1.3929572105407715,
+      "learning_rate": 7.476844022140464e-05,
+      "loss": 1.4991,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7386478304742684,
+      "grad_norm": 1.3147705793380737,
+      "learning_rate": 7.462335635471136e-05,
+      "loss": 1.4049,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7406659939455096,
+      "grad_norm": 1.5074125528335571,
+      "learning_rate": 7.44779982686001e-05,
+      "loss": 1.4351,
+      "step": 3670
+    },
+    {
+      "epoch": 0.7426841574167508,
+      "grad_norm": 1.4625370502471924,
+      "learning_rate": 7.43323675818595e-05,
+      "loss": 1.3535,
+      "step": 3680
+    },
+    {
+      "epoch": 0.7447023208879919,
+      "grad_norm": 1.6273128986358643,
+      "learning_rate": 7.418646591631404e-05,
+      "loss": 1.3886,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7467204843592331,
+      "grad_norm": 1.5632721185684204,
+      "learning_rate": 7.404029489680598e-05,
+      "loss": 1.4134,
+      "step": 3700
+    },
+    {
+      "epoch": 0.7487386478304743,
+      "grad_norm": 1.491721510887146,
+      "learning_rate": 7.389385615117723e-05,
+      "loss": 1.4279,
+      "step": 3710
+    },
+    {
+      "epoch": 0.7507568113017155,
+      "grad_norm": 1.5549407005310059,
+      "learning_rate": 7.37471513102513e-05,
+      "loss": 1.3888,
+      "step": 3720
+    },
+    {
+      "epoch": 0.7527749747729566,
+      "grad_norm": 1.4470081329345703,
+      "learning_rate": 7.360018200781502e-05,
+      "loss": 1.4272,
+      "step": 3730
+    },
+    {
+      "epoch": 0.7547931382441978,
+      "grad_norm": 1.686854600906372,
+      "learning_rate": 7.345294988060046e-05,
+      "loss": 1.5853,
+      "step": 3740
+    },
+    {
+      "epoch": 0.756811301715439,
+      "grad_norm": 1.5734236240386963,
+      "learning_rate": 7.330545656826662e-05,
+      "loss": 1.44,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7588294651866802,
+      "grad_norm": 1.3938251733779907,
+      "learning_rate": 7.315770371338126e-05,
+      "loss": 1.3882,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7608476286579213,
+      "grad_norm": 1.5192281007766724,
+      "learning_rate": 7.300969296140244e-05,
+      "loss": 1.5221,
+      "step": 3770
+    },
+    {
+      "epoch": 0.7628657921291625,
+      "grad_norm": 1.7236813306808472,
+      "learning_rate": 7.286142596066044e-05,
+      "loss": 1.4553,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7648839556004037,
+      "grad_norm": 1.402596116065979,
+      "learning_rate": 7.271290436233916e-05,
+      "loss": 1.4925,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7669021190716448,
+      "grad_norm": 1.5744677782058716,
+      "learning_rate": 7.25641298204579e-05,
+      "loss": 1.4484,
+      "step": 3800
+    },
+    {
+      "epoch": 0.768920282542886,
+      "grad_norm": 1.640576720237732,
+      "learning_rate": 7.241510399185287e-05,
+      "loss": 1.4277,
+      "step": 3810
+    },
+    {
+      "epoch": 0.7709384460141272,
+      "grad_norm": 1.6124114990234375,
+      "learning_rate": 7.226582853615874e-05,
+      "loss": 1.3545,
+      "step": 3820
+    },
+    {
+      "epoch": 0.7729566094853683,
+      "grad_norm": 1.7142858505249023,
+      "learning_rate": 7.211630511579015e-05,
+      "loss": 1.4184,
+      "step": 3830
+    },
+    {
+      "epoch": 0.7749747729566094,
+      "grad_norm": 1.3614004850387573,
+      "learning_rate": 7.196653539592326e-05,
+      "loss": 1.4101,
+      "step": 3840
+    },
+    {
+      "epoch": 0.7769929364278506,
+      "grad_norm": 1.3301514387130737,
+      "learning_rate": 7.181652104447711e-05,
+      "loss": 1.4297,
+      "step": 3850
+    },
+    {
+      "epoch": 0.7790110998990918,
+      "grad_norm": 1.441169023513794,
+      "learning_rate": 7.166626373209514e-05,
+      "loss": 1.4615,
+      "step": 3860
+    },
+    {
+      "epoch": 0.781029263370333,
+      "grad_norm": 1.4002233743667603,
+      "learning_rate": 7.15157651321265e-05,
+      "loss": 1.5348,
+      "step": 3870
+    },
+    {
+      "epoch": 0.7830474268415741,
+      "grad_norm": 1.4642239809036255,
+      "learning_rate": 7.136502692060746e-05,
+      "loss": 1.5266,
+      "step": 3880
+    },
+    {
+      "epoch": 0.7850655903128153,
+      "grad_norm": 1.4477249383926392,
+      "learning_rate": 7.121405077624276e-05,
+      "loss": 1.468,
+      "step": 3890
+    },
+    {
+      "epoch": 0.7870837537840565,
+      "grad_norm": 1.3571120500564575,
+      "learning_rate": 7.106283838038685e-05,
+      "loss": 1.472,
+      "step": 3900
+    },
+    {
+      "epoch": 0.7891019172552977,
+      "grad_norm": 1.3332570791244507,
+      "learning_rate": 7.091139141702527e-05,
+      "loss": 1.3955,
+      "step": 3910
+    },
+    {
+      "epoch": 0.7911200807265388,
+      "grad_norm": 1.5350067615509033,
+      "learning_rate": 7.075971157275575e-05,
+      "loss": 1.4683,
+      "step": 3920
+    },
+    {
+      "epoch": 0.79313824419778,
+      "grad_norm": 1.4552041292190552,
+      "learning_rate": 7.06078005367696e-05,
+      "loss": 1.4189,
+      "step": 3930
+    },
+    {
+      "epoch": 0.7951564076690212,
+      "grad_norm": 1.4249471426010132,
+      "learning_rate": 7.045566000083278e-05,
+      "loss": 1.4861,
+      "step": 3940
+    },
+    {
+      "epoch": 0.7971745711402624,
+      "grad_norm": 1.3417631387710571,
+      "learning_rate": 7.030329165926706e-05,
+      "loss": 1.4106,
+      "step": 3950
+    },
+    {
+      "epoch": 0.7991927346115035,
+      "grad_norm": 1.616866111755371,
+      "learning_rate": 7.01506972089312e-05,
+      "loss": 1.5251,
+      "step": 3960
+    },
+    {
+      "epoch": 0.8012108980827447,
+      "grad_norm": 1.5517117977142334,
+      "learning_rate": 6.999787834920202e-05,
+      "loss": 1.388,
+      "step": 3970
+    },
+    {
+      "epoch": 0.8032290615539859,
+      "grad_norm": 1.3729262351989746,
+      "learning_rate": 6.984483678195553e-05,
+      "loss": 1.4466,
+      "step": 3980
+    },
+    {
+      "epoch": 0.805247225025227,
+      "grad_norm": 1.5104920864105225,
+      "learning_rate": 6.969157421154789e-05,
+      "loss": 1.488,
+      "step": 3990
+    },
+    {
+      "epoch": 0.8072653884964682,
+      "grad_norm": 1.644098162651062,
+      "learning_rate": 6.95380923447965e-05,
+      "loss": 1.3932,
+      "step": 4000
+    },
+    {
+      "epoch": 0.8092835519677094,
+      "grad_norm": 1.4966418743133545,
+      "learning_rate": 6.938439289096095e-05,
+      "loss": 1.4094,
+      "step": 4010
+    },
+    {
+      "epoch": 0.8113017154389506,
+      "grad_norm": 1.6125532388687134,
+      "learning_rate": 6.923047756172401e-05,
+      "loss": 1.4399,
+      "step": 4020
+    },
+    {
+      "epoch": 0.8133198789101918,
+      "grad_norm": 1.3540160655975342,
+      "learning_rate": 6.907634807117257e-05,
+      "loss": 1.3183,
+      "step": 4030
+    },
+    {
+      "epoch": 0.8153380423814329,
+      "grad_norm": 1.390666127204895,
+      "learning_rate": 6.892200613577852e-05,
+      "loss": 1.5065,
+      "step": 4040
+    },
+    {
+      "epoch": 0.8173562058526741,
+      "grad_norm": 1.4623992443084717,
+      "learning_rate": 6.876745347437964e-05,
+      "loss": 1.4661,
+      "step": 4050
+    },
+    {
+      "epoch": 0.8193743693239153,
+      "grad_norm": 1.251976728439331,
+      "learning_rate": 6.861269180816052e-05,
+      "loss": 1.3724,
+      "step": 4060
+    },
+    {
+      "epoch": 0.8213925327951564,
+      "grad_norm": 1.4126683473587036,
+      "learning_rate": 6.845772286063332e-05,
+      "loss": 1.4495,
+      "step": 4070
+    },
+    {
+      "epoch": 0.8234106962663976,
+      "grad_norm": 1.3776429891586304,
+      "learning_rate": 6.830254835761856e-05,
+      "loss": 1.4128,
+      "step": 4080
+    },
+    {
+      "epoch": 0.8254288597376388,
+      "grad_norm": 1.5709989070892334,
+      "learning_rate": 6.814717002722602e-05,
+      "loss": 1.5365,
+      "step": 4090
+    },
+    {
+      "epoch": 0.82744702320888,
+      "grad_norm": 1.326343059539795,
+      "learning_rate": 6.799158959983536e-05,
+      "loss": 1.4633,
+      "step": 4100
+    },
+    {
+      "epoch": 0.829465186680121,
+      "grad_norm": 1.4777864217758179,
+      "learning_rate": 6.78358088080769e-05,
+      "loss": 1.5106,
+      "step": 4110
+    },
+    {
+      "epoch": 0.8314833501513622,
+      "grad_norm": 1.595955729484558,
+      "learning_rate": 6.767982938681239e-05,
+      "loss": 1.5467,
+      "step": 4120
+    },
+    {
+      "epoch": 0.8335015136226034,
+      "grad_norm": 1.611190915107727,
+      "learning_rate": 6.752365307311556e-05,
+      "loss": 1.4587,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8355196770938446,
+      "grad_norm": 1.2045243978500366,
+      "learning_rate": 6.736728160625284e-05,
+      "loss": 1.5403,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8375378405650857,
+      "grad_norm": 1.5161255598068237,
+      "learning_rate": 6.721071672766406e-05,
+      "loss": 1.4287,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8395560040363269,
+      "grad_norm": 1.5501888990402222,
+      "learning_rate": 6.705396018094297e-05,
+      "loss": 1.4729,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8415741675075681,
+      "grad_norm": 1.350813388824463,
+      "learning_rate": 6.689701371181781e-05,
+      "loss": 1.3742,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8435923309788093,
+      "grad_norm": 1.5314749479293823,
+      "learning_rate": 6.673987906813191e-05,
+      "loss": 1.355,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8456104944500504,
+      "grad_norm": 1.5933685302734375,
+      "learning_rate": 6.658255799982424e-05,
+      "loss": 1.4609,
+      "step": 4190
+    },
+    {
+      "epoch": 0.8476286579212916,
+      "grad_norm": 1.429506778717041,
+      "learning_rate": 6.642505225890987e-05,
+      "loss": 1.3823,
+      "step": 4200
+    },
+    {
+      "epoch": 0.8496468213925328,
+      "grad_norm": 1.5170999765396118,
+      "learning_rate": 6.626736359946052e-05,
+      "loss": 1.4732,
+      "step": 4210
+    },
+    {
+      "epoch": 0.851664984863774,
+      "grad_norm": 1.3810912370681763,
+      "learning_rate": 6.610949377758497e-05,
+      "loss": 1.4792,
+      "step": 4220
+    },
+    {
+      "epoch": 0.8536831483350151,
+      "grad_norm": 1.5815091133117676,
+      "learning_rate": 6.595144455140952e-05,
+      "loss": 1.4539,
+      "step": 4230
+    },
+    {
+      "epoch": 0.8557013118062563,
+      "grad_norm": 1.4099366664886475,
+      "learning_rate": 6.579321768105845e-05,
+      "loss": 1.459,
+      "step": 4240
+    },
+    {
+      "epoch": 0.8577194752774975,
+      "grad_norm": 1.529607892036438,
+      "learning_rate": 6.563481492863436e-05,
+      "loss": 1.434,
+      "step": 4250
+    },
+    {
+      "epoch": 0.8597376387487387,
+      "grad_norm": 1.4503751993179321,
+      "learning_rate": 6.547623805819854e-05,
+      "loss": 1.3988,
+      "step": 4260
+    },
+    {
+      "epoch": 0.8617558022199798,
+      "grad_norm": 1.3069409132003784,
+      "learning_rate": 6.531748883575143e-05,
+      "loss": 1.4464,
+      "step": 4270
+    },
+    {
+      "epoch": 0.863773965691221,
+      "grad_norm": 1.5749943256378174,
+      "learning_rate": 6.51585690292128e-05,
+      "loss": 1.3765,
+      "step": 4280
+    },
+    {
+      "epoch": 0.8657921291624622,
+      "grad_norm": 1.7546286582946777,
+      "learning_rate": 6.499948040840219e-05,
+      "loss": 1.4531,
+      "step": 4290
+    },
+    {
+      "epoch": 0.8678102926337034,
+      "grad_norm": 1.4653708934783936,
+      "learning_rate": 6.484022474501914e-05,
+      "loss": 1.421,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8698284561049445,
+      "grad_norm": 1.4347306489944458,
+      "learning_rate": 6.468080381262347e-05,
+      "loss": 1.4107,
+      "step": 4310
+    },
+    {
+      "epoch": 0.8718466195761857,
+      "grad_norm": 1.6876280307769775,
+      "learning_rate": 6.45212193866155e-05,
+      "loss": 1.4524,
+      "step": 4320
+    },
+    {
+      "epoch": 0.8738647830474269,
+      "grad_norm": 1.4129562377929688,
+      "learning_rate": 6.436147324421635e-05,
+      "loss": 1.3881,
+      "step": 4330
+    },
+    {
+      "epoch": 0.875882946518668,
+      "grad_norm": 1.5428842306137085,
+      "learning_rate": 6.420156716444805e-05,
+      "loss": 1.3926,
+      "step": 4340
+    },
+    {
+      "epoch": 0.8779011099899092,
+      "grad_norm": 1.4280976057052612,
+      "learning_rate": 6.404150292811386e-05,
+      "loss": 1.3836,
+      "step": 4350
+    },
+    {
+      "epoch": 0.8799192734611504,
+      "grad_norm": 1.5553336143493652,
+      "learning_rate": 6.388128231777828e-05,
+      "loss": 1.5083,
+      "step": 4360
+    },
+    {
+      "epoch": 0.8819374369323916,
+      "grad_norm": 1.3947172164916992,
+      "learning_rate": 6.372090711774732e-05,
+      "loss": 1.3669,
+      "step": 4370
+    },
+    {
+      "epoch": 0.8839556004036327,
+      "grad_norm": 1.413853645324707,
+      "learning_rate": 6.356037911404858e-05,
+      "loss": 1.4119,
+      "step": 4380
+    },
+    {
+      "epoch": 0.8859737638748738,
+      "grad_norm": 1.3096851110458374,
+      "learning_rate": 6.339970009441137e-05,
+      "loss": 1.391,
+      "step": 4390
+    },
+    {
+      "epoch": 0.887991927346115,
+      "grad_norm": 1.2649918794631958,
+      "learning_rate": 6.323887184824678e-05,
+      "loss": 1.3865,
+      "step": 4400
+    },
+    {
+      "epoch": 0.8900100908173562,
+      "grad_norm": 1.4351869821548462,
+      "learning_rate": 6.307789616662778e-05,
+      "loss": 1.3554,
+      "step": 4410
+    },
+    {
+      "epoch": 0.8920282542885973,
+      "grad_norm": 1.4802148342132568,
+      "learning_rate": 6.291677484226929e-05,
+      "loss": 1.5499,
+      "step": 4420
+    },
+    {
+      "epoch": 0.8940464177598385,
+      "grad_norm": 1.6230851411819458,
+      "learning_rate": 6.275550966950814e-05,
+      "loss": 1.4822,
+      "step": 4430
+    },
+    {
+      "epoch": 0.8960645812310797,
+      "grad_norm": 1.442018985748291,
+      "learning_rate": 6.259410244428318e-05,
+      "loss": 1.383,
+      "step": 4440
+    },
+    {
+      "epoch": 0.8980827447023209,
+      "grad_norm": 1.493252158164978,
+      "learning_rate": 6.243255496411519e-05,
+      "loss": 1.4048,
+      "step": 4450
+    },
+    {
+      "epoch": 0.900100908173562,
+      "grad_norm": 1.5293428897857666,
+      "learning_rate": 6.227086902808697e-05,
+      "loss": 1.5437,
+      "step": 4460
+    },
+    {
+      "epoch": 0.9021190716448032,
+      "grad_norm": 1.3244189023971558,
+      "learning_rate": 6.210904643682318e-05,
+      "loss": 1.4652,
+      "step": 4470
+    },
+    {
+      "epoch": 0.9041372351160444,
+      "grad_norm": 1.4349359273910522,
+      "learning_rate": 6.194708899247037e-05,
+      "loss": 1.4024,
+      "step": 4480
+    },
+    {
+      "epoch": 0.9061553985872856,
+      "grad_norm": 1.5854321718215942,
+      "learning_rate": 6.178499849867689e-05,
+      "loss": 1.4496,
+      "step": 4490
+    },
+    {
+      "epoch": 0.9081735620585267,
+      "grad_norm": 1.4568113088607788,
+      "learning_rate": 6.162277676057284e-05,
+      "loss": 1.4378,
+      "step": 4500
+    },
+    {
+      "epoch": 0.9101917255297679,
+      "grad_norm": 1.3261117935180664,
+      "learning_rate": 6.146042558474987e-05,
+      "loss": 1.4257,
+      "step": 4510
+    },
+    {
+      "epoch": 0.9122098890010091,
+      "grad_norm": 1.4419591426849365,
+      "learning_rate": 6.129794677924113e-05,
+      "loss": 1.4045,
+      "step": 4520
+    },
+    {
+      "epoch": 0.9142280524722503,
+      "grad_norm": 1.3974449634552002,
+      "learning_rate": 6.113534215350116e-05,
+      "loss": 1.4281,
+      "step": 4530
+    },
+    {
+      "epoch": 0.9162462159434914,
+      "grad_norm": 1.3392093181610107,
+      "learning_rate": 6.097261351838569e-05,
+      "loss": 1.5913,
+      "step": 4540
+    },
+    {
+      "epoch": 0.9182643794147326,
+      "grad_norm": 1.3982704877853394,
+      "learning_rate": 6.0809762686131474e-05,
+      "loss": 1.391,
+      "step": 4550
+    },
+    {
+      "epoch": 0.9202825428859738,
+      "grad_norm": 1.2605746984481812,
+      "learning_rate": 6.064679147033614e-05,
+      "loss": 1.4911,
+      "step": 4560
+    },
+    {
+      "epoch": 0.922300706357215,
+      "grad_norm": 1.5178440809249878,
+      "learning_rate": 6.0483701685937954e-05,
+      "loss": 1.4428,
+      "step": 4570
+    },
+    {
+      "epoch": 0.9243188698284561,
+      "grad_norm": 1.522623062133789,
+      "learning_rate": 6.0320495149195644e-05,
+      "loss": 1.395,
+      "step": 4580
+    },
+    {
+      "epoch": 0.9263370332996973,
+      "grad_norm": 1.7341537475585938,
+      "learning_rate": 6.015717367766815e-05,
+      "loss": 1.4924,
+      "step": 4590
+    },
+    {
+      "epoch": 0.9283551967709385,
+      "grad_norm": 1.4069305658340454,
+      "learning_rate": 5.999373909019437e-05,
+      "loss": 1.4476,
+      "step": 4600
+    },
+    {
+      "epoch": 0.9303733602421796,
+      "grad_norm": 1.2393527030944824,
+      "learning_rate": 5.9830193206872974e-05,
+      "loss": 1.4227,
+      "step": 4610
+    },
+    {
+      "epoch": 0.9323915237134208,
+      "grad_norm": 1.2959623336791992,
+      "learning_rate": 5.966653784904207e-05,
+      "loss": 1.4123,
+      "step": 4620
+    },
+    {
+      "epoch": 0.934409687184662,
+      "grad_norm": 1.6350857019424438,
+      "learning_rate": 5.950277483925889e-05,
+      "loss": 1.4116,
+      "step": 4630
+    },
+    {
+      "epoch": 0.9364278506559032,
+      "grad_norm": 1.1435925960540771,
+      "learning_rate": 5.933890600127958e-05,
+      "loss": 1.4417,
+      "step": 4640
+    },
+    {
+      "epoch": 0.9384460141271443,
+      "grad_norm": 1.4868639707565308,
+      "learning_rate": 5.917493316003884e-05,
+      "loss": 1.4769,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9404641775983855,
+      "grad_norm": 1.4463169574737549,
+      "learning_rate": 5.90108581416296e-05,
+      "loss": 1.4507,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9424823410696267,
+      "grad_norm": 1.3405494689941406,
+      "learning_rate": 5.8846682773282694e-05,
+      "loss": 1.4446,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9445005045408678,
+      "grad_norm": 1.4846607446670532,
+      "learning_rate": 5.868240888334653e-05,
+      "loss": 1.439,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9465186680121089,
+      "grad_norm": 1.2763575315475464,
+      "learning_rate": 5.851803830126666e-05,
+      "loss": 1.5239,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9485368314833501,
+      "grad_norm": 1.274949550628662,
+      "learning_rate": 5.835357285756552e-05,
+      "loss": 1.374,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9505549949545913,
+      "grad_norm": 1.277209758758545,
+      "learning_rate": 5.8189014383821914e-05,
+      "loss": 1.3033,
+      "step": 4710
+    },
+    {
+      "epoch": 0.9525731584258325,
+      "grad_norm": 1.395630955696106,
+      "learning_rate": 5.8024364712650724e-05,
+      "loss": 1.3979,
+      "step": 4720
+    },
+    {
+      "epoch": 0.9545913218970736,
+      "grad_norm": 1.2895914316177368,
+      "learning_rate": 5.785962567768243e-05,
+      "loss": 1.376,
+      "step": 4730
+    },
+    {
+      "epoch": 0.9566094853683148,
+      "grad_norm": 1.2383575439453125,
+      "learning_rate": 5.769479911354273e-05,
+      "loss": 1.4396,
+      "step": 4740
+    },
+    {
+      "epoch": 0.958627648839556,
+      "grad_norm": 1.3582751750946045,
+      "learning_rate": 5.7529886855832096e-05,
+      "loss": 1.497,
+      "step": 4750
+    },
+    {
+      "epoch": 0.9606458123107972,
+      "grad_norm": 1.4777917861938477,
+      "learning_rate": 5.736489074110533e-05,
+      "loss": 1.2924,
+      "step": 4760
+    },
+    {
+      "epoch": 0.9626639757820383,
+      "grad_norm": 1.448093056678772,
+      "learning_rate": 5.71998126068511e-05,
+      "loss": 1.3533,
+      "step": 4770
+    },
+    {
+      "epoch": 0.9646821392532795,
+      "grad_norm": 1.336315393447876,
+      "learning_rate": 5.7034654291471524e-05,
+      "loss": 1.4275,
+      "step": 4780
+    },
+    {
+      "epoch": 0.9667003027245207,
+      "grad_norm": 1.2323698997497559,
+      "learning_rate": 5.686941763426161e-05,
+      "loss": 1.4587,
+      "step": 4790
+    },
+    {
+      "epoch": 0.9687184661957619,
+      "grad_norm": 1.4090189933776855,
+      "learning_rate": 5.670410447538889e-05,
+      "loss": 1.3976,
+      "step": 4800
+    },
+    {
+      "epoch": 0.970736629667003,
+      "grad_norm": 1.471298336982727,
+      "learning_rate": 5.653871665587278e-05,
+      "loss": 1.4002,
+      "step": 4810
+    },
+    {
+      "epoch": 0.9727547931382442,
+      "grad_norm": 1.2523037195205688,
+      "learning_rate": 5.6373256017564215e-05,
+      "loss": 1.3906,
+      "step": 4820
+    },
+    {
+      "epoch": 0.9747729566094854,
+      "grad_norm": 1.441237211227417,
+      "learning_rate": 5.620772440312508e-05,
+      "loss": 1.3976,
+      "step": 4830
+    },
+    {
+      "epoch": 0.9767911200807265,
+      "grad_norm": 1.7172768115997314,
+      "learning_rate": 5.6042123656007685e-05,
+      "loss": 1.4364,
+      "step": 4840
+    },
+    {
+      "epoch": 0.9788092835519677,
+      "grad_norm": 1.4941967725753784,
+      "learning_rate": 5.587645562043422e-05,
+      "loss": 1.4107,
+      "step": 4850
+    },
+    {
+      "epoch": 0.9808274470232089,
+      "grad_norm": 1.4575670957565308,
+      "learning_rate": 5.5710722141376245e-05,
+      "loss": 1.342,
+      "step": 4860
+    },
+    {
+      "epoch": 0.9828456104944501,
+      "grad_norm": 1.235521674156189,
+      "learning_rate": 5.5544925064534145e-05,
+      "loss": 1.2921,
+      "step": 4870
+    },
+    {
+      "epoch": 0.9848637739656912,
+      "grad_norm": 1.5723048448562622,
+      "learning_rate": 5.537906623631657e-05,
+      "loss": 1.5273,
+      "step": 4880
+    },
+    {
+      "epoch": 0.9868819374369324,
+      "grad_norm": 1.4097591638565063,
+      "learning_rate": 5.521314750381983e-05,
+      "loss": 1.3714,
+      "step": 4890
+    },
+    {
+      "epoch": 0.9889001009081736,
+      "grad_norm": 1.4014616012573242,
+      "learning_rate": 5.5047170714807406e-05,
+      "loss": 1.3598,
+      "step": 4900
+    },
+    {
+      "epoch": 0.9909182643794148,
+      "grad_norm": 1.2477079629898071,
+      "learning_rate": 5.4881137717689315e-05,
+      "loss": 1.3501,
+      "step": 4910
+    },
+    {
+      "epoch": 0.992936427850656,
+      "grad_norm": 1.3461527824401855,
+      "learning_rate": 5.471505036150154e-05,
+      "loss": 1.3813,
+      "step": 4920
+    },
+    {
+      "epoch": 0.9949545913218971,
+      "grad_norm": 1.6173443794250488,
+      "learning_rate": 5.454891049588544e-05,
+      "loss": 1.5266,
+      "step": 4930
+    },
+    {
+      "epoch": 0.9969727547931383,
+      "grad_norm": 1.351027011871338,
+      "learning_rate": 5.438271997106712e-05,
+      "loss": 1.395,
+      "step": 4940
+    },
+    {
+      "epoch": 0.9989909182643795,
+      "grad_norm": 1.210114598274231,
+      "learning_rate": 5.421648063783689e-05,
+      "loss": 1.284,
+      "step": 4950
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9910,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 4955,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.387441374475059e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-4955/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6912a52347623c3b2ccb76cdcbcb6d6cb5bb72087ff419ea2fcbc76b59b74f2
+size 5432

checkpoint-9910/config.json

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

checkpoint-9910/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-9910/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4357ddac2b9e221ae783e6f333d464cb82ceda695fc0576ab745f89afdc0fb3a
+size 4950070472

checkpoint-9910/model-00002-of-00006.safetensors

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

checkpoint-9910/model-00003-of-00006.safetensors

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

checkpoint-9910/model-00004-of-00006.safetensors

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

checkpoint-9910/model-00005-of-00006.safetensors

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

checkpoint-9910/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42dd47e6d6ade5c0772b4ded16e59b54e63573e93ab1d5557d5123e60cebbc2f
+size 4614509080

checkpoint-9910/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-9910/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27152cad40c80f3e4e4d89b8a0b5c4a0e520c687e3154588e91adb15a0cc2d07
+size 13055369358

checkpoint-9910/rng_state.pth

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

checkpoint-9910/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62175ba90aed8860e6e358d54eacbce4d29da45c99caa1fd61b0dfcff9e66a73
+size 1064

checkpoint-9910/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6912a52347623c3b2ccb76cdcbcb6d6cb5bb72087ff419ea2fcbc76b59b74f2
+size 5432

config.json

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

generation_config.json

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

model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4357ddac2b9e221ae783e6f333d464cb82ceda695fc0576ab745f89afdc0fb3a
+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 sha256:42dd47e6d6ade5c0772b4ded16e59b54e63573e93ab1d5557d5123e60cebbc2f
+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:f6912a52347623c3b2ccb76cdcbcb6d6cb5bb72087ff419ea2fcbc76b59b74f2
+size 5432