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-mean/",
+  "architectures": [
+    "Olmo2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "eos_token_id": 100257,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 4096,
+  "model_type": "olmo2",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 100277,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 500000,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.48.0.dev0",
+  "use_cache": true,
+  "vocab_size": 100384
+}

checkpoint-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 sha256:7b2f3126ff493c6961622256e4998efa56a3bcf849c0e4a5af61c147a3ab6451
+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:b2a289f3510b997bf969acdbb4da9863d140e27188986d8a60b0adb43de2644e
+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:d33e1c7612d9b01c9ccd954130a4318768a494988644a97e79dce486cca31cce
+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": 84.60143280029297,
+      "learning_rate": 2.0161290322580646e-06,
+      "loss": 7.1336,
+      "step": 10
+    },
+    {
+      "epoch": 0.004036326942482341,
+      "grad_norm": 15.496525764465332,
+      "learning_rate": 4.032258064516129e-06,
+      "loss": 5.515,
+      "step": 20
+    },
+    {
+      "epoch": 0.006054490413723511,
+      "grad_norm": 10.33206844329834,
+      "learning_rate": 6.048387096774194e-06,
+      "loss": 4.8214,
+      "step": 30
+    },
+    {
+      "epoch": 0.008072653884964682,
+      "grad_norm": 19.034317016601562,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 4.4868,
+      "step": 40
+    },
+    {
+      "epoch": 0.010090817356205853,
+      "grad_norm": 4.963449001312256,
+      "learning_rate": 1.0080645161290323e-05,
+      "loss": 4.1247,
+      "step": 50
+    },
+    {
+      "epoch": 0.012108980827447022,
+      "grad_norm": 4.877665042877197,
+      "learning_rate": 1.2096774193548388e-05,
+      "loss": 3.756,
+      "step": 60
+    },
+    {
+      "epoch": 0.014127144298688193,
+      "grad_norm": 4.284060955047607,
+      "learning_rate": 1.4112903225806454e-05,
+      "loss": 3.4368,
+      "step": 70
+    },
+    {
+      "epoch": 0.016145307769929364,
+      "grad_norm": 3.7732603549957275,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 3.2998,
+      "step": 80
+    },
+    {
+      "epoch": 0.018163471241170535,
+      "grad_norm": 3.365959405899048,
+      "learning_rate": 1.8145161290322583e-05,
+      "loss": 3.1228,
+      "step": 90
+    },
+    {
+      "epoch": 0.020181634712411706,
+      "grad_norm": 2.8467206954956055,
+      "learning_rate": 2.0161290322580645e-05,
+      "loss": 2.9786,
+      "step": 100
+    },
+    {
+      "epoch": 0.022199798183652877,
+      "grad_norm": 3.936636209487915,
+      "learning_rate": 2.217741935483871e-05,
+      "loss": 2.815,
+      "step": 110
+    },
+    {
+      "epoch": 0.024217961654894045,
+      "grad_norm": 2.843057155609131,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 2.7479,
+      "step": 120
+    },
+    {
+      "epoch": 0.026236125126135216,
+      "grad_norm": 2.4136409759521484,
+      "learning_rate": 2.620967741935484e-05,
+      "loss": 2.6495,
+      "step": 130
+    },
+    {
+      "epoch": 0.028254288597376387,
+      "grad_norm": 4.50978946685791,
+      "learning_rate": 2.822580645161291e-05,
+      "loss": 2.6704,
+      "step": 140
+    },
+    {
+      "epoch": 0.030272452068617558,
+      "grad_norm": 2.8832032680511475,
+      "learning_rate": 3.024193548387097e-05,
+      "loss": 2.5093,
+      "step": 150
+    },
+    {
+      "epoch": 0.03229061553985873,
+      "grad_norm": 3.834477424621582,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 2.515,
+      "step": 160
+    },
+    {
+      "epoch": 0.034308779011099896,
+      "grad_norm": 2.2256412506103516,
+      "learning_rate": 3.427419354838709e-05,
+      "loss": 2.4173,
+      "step": 170
+    },
+    {
+      "epoch": 0.03632694248234107,
+      "grad_norm": 2.659034013748169,
+      "learning_rate": 3.6290322580645165e-05,
+      "loss": 2.3726,
+      "step": 180
+    },
+    {
+      "epoch": 0.03834510595358224,
+      "grad_norm": 2.66379714012146,
+      "learning_rate": 3.8306451612903224e-05,
+      "loss": 2.4307,
+      "step": 190
+    },
+    {
+      "epoch": 0.04036326942482341,
+      "grad_norm": 2.4209799766540527,
+      "learning_rate": 4.032258064516129e-05,
+      "loss": 2.4125,
+      "step": 200
+    },
+    {
+      "epoch": 0.04238143289606458,
+      "grad_norm": 2.361922025680542,
+      "learning_rate": 4.2338709677419356e-05,
+      "loss": 2.331,
+      "step": 210
+    },
+    {
+      "epoch": 0.044399596367305755,
+      "grad_norm": 2.37805438041687,
+      "learning_rate": 4.435483870967742e-05,
+      "loss": 2.3345,
+      "step": 220
+    },
+    {
+      "epoch": 0.04641775983854692,
+      "grad_norm": 2.6350855827331543,
+      "learning_rate": 4.637096774193548e-05,
+      "loss": 2.3108,
+      "step": 230
+    },
+    {
+      "epoch": 0.04843592330978809,
+      "grad_norm": 2.274717092514038,
+      "learning_rate": 4.8387096774193554e-05,
+      "loss": 2.3286,
+      "step": 240
+    },
+    {
+      "epoch": 0.050454086781029264,
+      "grad_norm": 2.4076735973358154,
+      "learning_rate": 5.040322580645161e-05,
+      "loss": 2.2257,
+      "step": 250
+    },
+    {
+      "epoch": 0.05247225025227043,
+      "grad_norm": 2.0606577396392822,
+      "learning_rate": 5.241935483870968e-05,
+      "loss": 2.2772,
+      "step": 260
+    },
+    {
+      "epoch": 0.054490413723511606,
+      "grad_norm": 2.362431287765503,
+      "learning_rate": 5.443548387096774e-05,
+      "loss": 2.2675,
+      "step": 270
+    },
+    {
+      "epoch": 0.056508577194752774,
+      "grad_norm": 2.4200310707092285,
+      "learning_rate": 5.645161290322582e-05,
+      "loss": 2.236,
+      "step": 280
+    },
+    {
+      "epoch": 0.05852674066599395,
+      "grad_norm": 2.1776976585388184,
+      "learning_rate": 5.8467741935483876e-05,
+      "loss": 2.1248,
+      "step": 290
+    },
+    {
+      "epoch": 0.060544904137235116,
+      "grad_norm": 2.3392269611358643,
+      "learning_rate": 6.048387096774194e-05,
+      "loss": 2.1562,
+      "step": 300
+    },
+    {
+      "epoch": 0.06256306760847628,
+      "grad_norm": 2.3293347358703613,
+      "learning_rate": 6.25e-05,
+      "loss": 2.1715,
+      "step": 310
+    },
+    {
+      "epoch": 0.06458123107971746,
+      "grad_norm": 1.9923762083053589,
+      "learning_rate": 6.451612903225807e-05,
+      "loss": 2.1212,
+      "step": 320
+    },
+    {
+      "epoch": 0.06659939455095863,
+      "grad_norm": 2.042971611022949,
+      "learning_rate": 6.653225806451613e-05,
+      "loss": 2.0921,
+      "step": 330
+    },
+    {
+      "epoch": 0.06861755802219979,
+      "grad_norm": 2.2022523880004883,
+      "learning_rate": 6.854838709677419e-05,
+      "loss": 2.0906,
+      "step": 340
+    },
+    {
+      "epoch": 0.07063572149344097,
+      "grad_norm": 2.102933406829834,
+      "learning_rate": 7.056451612903226e-05,
+      "loss": 2.2652,
+      "step": 350
+    },
+    {
+      "epoch": 0.07265388496468214,
+      "grad_norm": 2.3307838439941406,
+      "learning_rate": 7.258064516129033e-05,
+      "loss": 2.0891,
+      "step": 360
+    },
+    {
+      "epoch": 0.07467204843592332,
+      "grad_norm": 1.9409267902374268,
+      "learning_rate": 7.45967741935484e-05,
+      "loss": 2.0769,
+      "step": 370
+    },
+    {
+      "epoch": 0.07669021190716448,
+      "grad_norm": 1.8024511337280273,
+      "learning_rate": 7.661290322580645e-05,
+      "loss": 2.0587,
+      "step": 380
+    },
+    {
+      "epoch": 0.07870837537840565,
+      "grad_norm": 2.0050477981567383,
+      "learning_rate": 7.862903225806451e-05,
+      "loss": 2.0375,
+      "step": 390
+    },
+    {
+      "epoch": 0.08072653884964683,
+      "grad_norm": 3.634002923965454,
+      "learning_rate": 8.064516129032258e-05,
+      "loss": 2.0089,
+      "step": 400
+    },
+    {
+      "epoch": 0.08274470232088799,
+      "grad_norm": 2.52044415473938,
+      "learning_rate": 8.266129032258066e-05,
+      "loss": 2.0246,
+      "step": 410
+    },
+    {
+      "epoch": 0.08476286579212916,
+      "grad_norm": 2.014629364013672,
+      "learning_rate": 8.467741935483871e-05,
+      "loss": 2.0291,
+      "step": 420
+    },
+    {
+      "epoch": 0.08678102926337034,
+      "grad_norm": 2.003462791442871,
+      "learning_rate": 8.669354838709678e-05,
+      "loss": 2.0223,
+      "step": 430
+    },
+    {
+      "epoch": 0.08879919273461151,
+      "grad_norm": 2.210601329803467,
+      "learning_rate": 8.870967741935484e-05,
+      "loss": 2.0178,
+      "step": 440
+    },
+    {
+      "epoch": 0.09081735620585267,
+      "grad_norm": 2.3255221843719482,
+      "learning_rate": 9.072580645161291e-05,
+      "loss": 2.0062,
+      "step": 450
+    },
+    {
+      "epoch": 0.09283551967709384,
+      "grad_norm": 2.130826950073242,
+      "learning_rate": 9.274193548387096e-05,
+      "loss": 1.9744,
+      "step": 460
+    },
+    {
+      "epoch": 0.09485368314833502,
+      "grad_norm": 2.3463449478149414,
+      "learning_rate": 9.475806451612904e-05,
+      "loss": 2.1566,
+      "step": 470
+    },
+    {
+      "epoch": 0.09687184661957618,
+      "grad_norm": 1.9628673791885376,
+      "learning_rate": 9.677419354838711e-05,
+      "loss": 2.046,
+      "step": 480
+    },
+    {
+      "epoch": 0.09889001009081735,
+      "grad_norm": 1.8357641696929932,
+      "learning_rate": 9.879032258064517e-05,
+      "loss": 2.121,
+      "step": 490
+    },
+    {
+      "epoch": 0.10090817356205853,
+      "grad_norm": 2.052020788192749,
+      "learning_rate": 9.999995545373623e-05,
+      "loss": 1.9792,
+      "step": 500
+    },
+    {
+      "epoch": 0.1029263370332997,
+      "grad_norm": 2.158012866973877,
+      "learning_rate": 9.999945430918042e-05,
+      "loss": 1.8667,
+      "step": 510
+    },
+    {
+      "epoch": 0.10494450050454086,
+      "grad_norm": 1.946934461593628,
+      "learning_rate": 9.999839634283869e-05,
+      "loss": 2.0411,
+      "step": 520
+    },
+    {
+      "epoch": 0.10696266397578204,
+      "grad_norm": 2.1030478477478027,
+      "learning_rate": 9.999678156649317e-05,
+      "loss": 2.0485,
+      "step": 530
+    },
+    {
+      "epoch": 0.10898082744702321,
+      "grad_norm": 1.9499808549880981,
+      "learning_rate": 9.999460999812691e-05,
+      "loss": 1.9301,
+      "step": 540
+    },
+    {
+      "epoch": 0.11099899091826437,
+      "grad_norm": 1.789939045906067,
+      "learning_rate": 9.999188166192368e-05,
+      "loss": 1.924,
+      "step": 550
+    },
+    {
+      "epoch": 0.11301715438950555,
+      "grad_norm": 1.7510501146316528,
+      "learning_rate": 9.998859658826777e-05,
+      "loss": 1.9682,
+      "step": 560
+    },
+    {
+      "epoch": 0.11503531786074672,
+      "grad_norm": 2.5464296340942383,
+      "learning_rate": 9.998475481374358e-05,
+      "loss": 2.0132,
+      "step": 570
+    },
+    {
+      "epoch": 0.1170534813319879,
+      "grad_norm": 1.796861171722412,
+      "learning_rate": 9.998035638113527e-05,
+      "loss": 1.9089,
+      "step": 580
+    },
+    {
+      "epoch": 0.11907164480322906,
+      "grad_norm": 2.1990606784820557,
+      "learning_rate": 9.997540133942624e-05,
+      "loss": 1.8077,
+      "step": 590
+    },
+    {
+      "epoch": 0.12108980827447023,
+      "grad_norm": 1.6746110916137695,
+      "learning_rate": 9.996988974379857e-05,
+      "loss": 1.9023,
+      "step": 600
+    },
+    {
+      "epoch": 0.1231079717457114,
+      "grad_norm": 1.7681794166564941,
+      "learning_rate": 9.996382165563247e-05,
+      "loss": 1.8878,
+      "step": 610
+    },
+    {
+      "epoch": 0.12512613521695257,
+      "grad_norm": 1.6506789922714233,
+      "learning_rate": 9.995719714250556e-05,
+      "loss": 1.993,
+      "step": 620
+    },
+    {
+      "epoch": 0.12714429868819374,
+      "grad_norm": 2.155871868133545,
+      "learning_rate": 9.995001627819211e-05,
+      "loss": 1.847,
+      "step": 630
+    },
+    {
+      "epoch": 0.12916246215943492,
+      "grad_norm": 1.7361620664596558,
+      "learning_rate": 9.99422791426622e-05,
+      "loss": 1.8265,
+      "step": 640
+    },
+    {
+      "epoch": 0.1311806256306761,
+      "grad_norm": 2.3019630908966064,
+      "learning_rate": 9.993398582208093e-05,
+      "loss": 1.8233,
+      "step": 650
+    },
+    {
+      "epoch": 0.13319878910191726,
+      "grad_norm": 1.9044718742370605,
+      "learning_rate": 9.99251364088073e-05,
+      "loss": 1.9208,
+      "step": 660
+    },
+    {
+      "epoch": 0.13521695257315844,
+      "grad_norm": 2.070833683013916,
+      "learning_rate": 9.991573100139334e-05,
+      "loss": 1.8743,
+      "step": 670
+    },
+    {
+      "epoch": 0.13723511604439959,
+      "grad_norm": 1.8902406692504883,
+      "learning_rate": 9.990576970458285e-05,
+      "loss": 1.8131,
+      "step": 680
+    },
+    {
+      "epoch": 0.13925327951564076,
+      "grad_norm": 2.179497718811035,
+      "learning_rate": 9.989525262931045e-05,
+      "loss": 1.816,
+      "step": 690
+    },
+    {
+      "epoch": 0.14127144298688193,
+      "grad_norm": 2.6633455753326416,
+      "learning_rate": 9.988417989270011e-05,
+      "loss": 1.8408,
+      "step": 700
+    },
+    {
+      "epoch": 0.1432896064581231,
+      "grad_norm": 1.8275139331817627,
+      "learning_rate": 9.987255161806402e-05,
+      "loss": 1.7789,
+      "step": 710
+    },
+    {
+      "epoch": 0.14530776992936428,
+      "grad_norm": 1.9536734819412231,
+      "learning_rate": 9.986036793490112e-05,
+      "loss": 1.9016,
+      "step": 720
+    },
+    {
+      "epoch": 0.14732593340060546,
+      "grad_norm": 1.7954963445663452,
+      "learning_rate": 9.984762897889568e-05,
+      "loss": 1.8267,
+      "step": 730
+    },
+    {
+      "epoch": 0.14934409687184663,
+      "grad_norm": 1.7522506713867188,
+      "learning_rate": 9.983433489191581e-05,
+      "loss": 1.8596,
+      "step": 740
+    },
+    {
+      "epoch": 0.15136226034308778,
+      "grad_norm": 1.799124836921692,
+      "learning_rate": 9.98204858220119e-05,
+      "loss": 1.8224,
+      "step": 750
+    },
+    {
+      "epoch": 0.15338042381432895,
+      "grad_norm": 1.8433665037155151,
+      "learning_rate": 9.980608192341488e-05,
+      "loss": 1.8241,
+      "step": 760
+    },
+    {
+      "epoch": 0.15539858728557013,
+      "grad_norm": 1.764257550239563,
+      "learning_rate": 9.979112335653462e-05,
+      "loss": 1.8093,
+      "step": 770
+    },
+    {
+      "epoch": 0.1574167507568113,
+      "grad_norm": 1.8525418043136597,
+      "learning_rate": 9.977561028795803e-05,
+      "loss": 1.7784,
+      "step": 780
+    },
+    {
+      "epoch": 0.15943491422805248,
+      "grad_norm": 2.0689351558685303,
+      "learning_rate": 9.97595428904473e-05,
+      "loss": 1.8073,
+      "step": 790
+    },
+    {
+      "epoch": 0.16145307769929365,
+      "grad_norm": 1.7697125673294067,
+      "learning_rate": 9.974292134293792e-05,
+      "loss": 1.804,
+      "step": 800
+    },
+    {
+      "epoch": 0.16347124117053483,
+      "grad_norm": 1.8966635465621948,
+      "learning_rate": 9.97257458305367e-05,
+      "loss": 1.8198,
+      "step": 810
+    },
+    {
+      "epoch": 0.16548940464177597,
+      "grad_norm": 1.6123894453048706,
+      "learning_rate": 9.970801654451973e-05,
+      "loss": 1.7637,
+      "step": 820
+    },
+    {
+      "epoch": 0.16750756811301715,
+      "grad_norm": 1.6604522466659546,
+      "learning_rate": 9.968973368233022e-05,
+      "loss": 1.7528,
+      "step": 830
+    },
+    {
+      "epoch": 0.16952573158425832,
+      "grad_norm": 1.8791446685791016,
+      "learning_rate": 9.96708974475763e-05,
+      "loss": 1.7687,
+      "step": 840
+    },
+    {
+      "epoch": 0.1715438950554995,
+      "grad_norm": 2.045793056488037,
+      "learning_rate": 9.965150805002878e-05,
+      "loss": 1.8325,
+      "step": 850
+    },
+    {
+      "epoch": 0.17356205852674067,
+      "grad_norm": 1.8702943325042725,
+      "learning_rate": 9.963156570561878e-05,
+      "loss": 1.712,
+      "step": 860
+    },
+    {
+      "epoch": 0.17558022199798184,
+      "grad_norm": 1.9760271310806274,
+      "learning_rate": 9.96110706364354e-05,
+      "loss": 1.8057,
+      "step": 870
+    },
+    {
+      "epoch": 0.17759838546922302,
+      "grad_norm": 1.9553672075271606,
+      "learning_rate": 9.959002307072312e-05,
+      "loss": 1.7453,
+      "step": 880
+    },
+    {
+      "epoch": 0.17961654894046417,
+      "grad_norm": 1.989668607711792,
+      "learning_rate": 9.956842324287936e-05,
+      "loss": 1.9109,
+      "step": 890
+    },
+    {
+      "epoch": 0.18163471241170534,
+      "grad_norm": 1.766709566116333,
+      "learning_rate": 9.954627139345186e-05,
+      "loss": 1.7539,
+      "step": 900
+    },
+    {
+      "epoch": 0.18365287588294651,
+      "grad_norm": 1.933042287826538,
+      "learning_rate": 9.952356776913594e-05,
+      "loss": 1.8372,
+      "step": 910
+    },
+    {
+      "epoch": 0.1856710393541877,
+      "grad_norm": 1.717915415763855,
+      "learning_rate": 9.950031262277183e-05,
+      "loss": 1.8199,
+      "step": 920
+    },
+    {
+      "epoch": 0.18768920282542886,
+      "grad_norm": 9.502272605895996,
+      "learning_rate": 9.947650621334179e-05,
+      "loss": 1.814,
+      "step": 930
+    },
+    {
+      "epoch": 0.18970736629667004,
+      "grad_norm": 2.0936007499694824,
+      "learning_rate": 9.945214880596725e-05,
+      "loss": 1.7535,
+      "step": 940
+    },
+    {
+      "epoch": 0.1917255297679112,
+      "grad_norm": 2.1896564960479736,
+      "learning_rate": 9.94272406719059e-05,
+      "loss": 1.772,
+      "step": 950
+    },
+    {
+      "epoch": 0.19374369323915236,
+      "grad_norm": 1.951393723487854,
+      "learning_rate": 9.940178208854858e-05,
+      "loss": 1.7976,
+      "step": 960
+    },
+    {
+      "epoch": 0.19576185671039353,
+      "grad_norm": 1.8681849241256714,
+      "learning_rate": 9.937577333941626e-05,
+      "loss": 1.662,
+      "step": 970
+    },
+    {
+      "epoch": 0.1977800201816347,
+      "grad_norm": 1.8085284233093262,
+      "learning_rate": 9.934921471415687e-05,
+      "loss": 1.7778,
+      "step": 980
+    },
+    {
+      "epoch": 0.19979818365287588,
+      "grad_norm": 2.0100769996643066,
+      "learning_rate": 9.932210650854205e-05,
+      "loss": 1.7732,
+      "step": 990
+    },
+    {
+      "epoch": 0.20181634712411706,
+      "grad_norm": 1.6803241968154907,
+      "learning_rate": 9.929444902446392e-05,
+      "loss": 1.7685,
+      "step": 1000
+    },
+    {
+      "epoch": 0.20383451059535823,
+      "grad_norm": 1.8448835611343384,
+      "learning_rate": 9.92662425699316e-05,
+      "loss": 1.7719,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2058526740665994,
+      "grad_norm": 1.8499823808670044,
+      "learning_rate": 9.923748745906789e-05,
+      "loss": 1.8417,
+      "step": 1020
+    },
+    {
+      "epoch": 0.20787083753784055,
+      "grad_norm": 1.7802287340164185,
+      "learning_rate": 9.920818401210574e-05,
+      "loss": 1.7677,
+      "step": 1030
+    },
+    {
+      "epoch": 0.20988900100908173,
+      "grad_norm": 2.019920587539673,
+      "learning_rate": 9.917833255538467e-05,
+      "loss": 1.8071,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2119071644803229,
+      "grad_norm": 1.838690161705017,
+      "learning_rate": 9.914793342134711e-05,
+      "loss": 1.7962,
+      "step": 1050
+    },
+    {
+      "epoch": 0.21392532795156408,
+      "grad_norm": 2.0018012523651123,
+      "learning_rate": 9.911698694853477e-05,
+      "loss": 1.7135,
+      "step": 1060
+    },
+    {
+      "epoch": 0.21594349142280525,
+      "grad_norm": 1.905208706855774,
+      "learning_rate": 9.908549348158485e-05,
+      "loss": 1.9039,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21796165489404642,
+      "grad_norm": 2.0252766609191895,
+      "learning_rate": 9.905345337122609e-05,
+      "loss": 1.8479,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2199798183652876,
+      "grad_norm": 1.7580780982971191,
+      "learning_rate": 9.902086697427504e-05,
+      "loss": 1.8081,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22199798183652875,
+      "grad_norm": 1.7557222843170166,
+      "learning_rate": 9.8987734653632e-05,
+      "loss": 1.6225,
+      "step": 1100
+    },
+    {
+      "epoch": 0.22401614530776992,
+      "grad_norm": 1.779552936553955,
+      "learning_rate": 9.895405677827692e-05,
+      "loss": 1.674,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2260343087790111,
+      "grad_norm": 1.6549618244171143,
+      "learning_rate": 9.89198337232654e-05,
+      "loss": 1.6844,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22805247225025227,
+      "grad_norm": 1.816671371459961,
+      "learning_rate": 9.888506586972446e-05,
+      "loss": 1.7142,
+      "step": 1130
+    },
+    {
+      "epoch": 0.23007063572149344,
+      "grad_norm": 1.7347807884216309,
+      "learning_rate": 9.884975360484827e-05,
+      "loss": 1.8952,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23208879919273462,
+      "grad_norm": 1.9305518865585327,
+      "learning_rate": 9.881389732189392e-05,
+      "loss": 1.7851,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2341069626639758,
+      "grad_norm": 1.5616772174835205,
+      "learning_rate": 9.877749742017694e-05,
+      "loss": 1.7088,
+      "step": 1160
+    },
+    {
+      "epoch": 0.23612512613521694,
+      "grad_norm": 1.8221651315689087,
+      "learning_rate": 9.874055430506691e-05,
+      "loss": 1.6248,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23814328960645811,
+      "grad_norm": 1.7022042274475098,
+      "learning_rate": 9.870306838798297e-05,
+      "loss": 1.669,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2401614530776993,
+      "grad_norm": 1.7719357013702393,
+      "learning_rate": 9.866504008638917e-05,
+      "loss": 1.7587,
+      "step": 1190
+    },
+    {
+      "epoch": 0.24217961654894046,
+      "grad_norm": 1.9335434436798096,
+      "learning_rate": 9.862646982378987e-05,
+      "loss": 1.8279,
+      "step": 1200
+    },
+    {
+      "epoch": 0.24419778002018164,
+      "grad_norm": 1.7225123643875122,
+      "learning_rate": 9.8587358029725e-05,
+      "loss": 1.7519,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2462159434914228,
+      "grad_norm": 1.6799899339675903,
+      "learning_rate": 9.854770513976531e-05,
+      "loss": 1.6876,
+      "step": 1220
+    },
+    {
+      "epoch": 0.248234106962664,
+      "grad_norm": 1.7735378742218018,
+      "learning_rate": 9.850751159550746e-05,
+      "loss": 1.687,
+      "step": 1230
+    },
+    {
+      "epoch": 0.25025227043390513,
+      "grad_norm": 1.9135595560073853,
+      "learning_rate": 9.846677784456918e-05,
+      "loss": 1.785,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2522704339051463,
+      "grad_norm": 1.684078335762024,
+      "learning_rate": 9.842550434058421e-05,
+      "loss": 1.6777,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2542885973763875,
+      "grad_norm": 1.8692346811294556,
+      "learning_rate": 9.838369154319728e-05,
+      "loss": 1.8198,
+      "step": 1260
+    },
+    {
+      "epoch": 0.25630676084762866,
+      "grad_norm": 1.526811957359314,
+      "learning_rate": 9.8341339918059e-05,
+      "loss": 1.6035,
+      "step": 1270
+    },
+    {
+      "epoch": 0.25832492431886983,
+      "grad_norm": 1.8684613704681396,
+      "learning_rate": 9.82984499368207e-05,
+      "loss": 1.7229,
+      "step": 1280
+    },
+    {
+      "epoch": 0.260343087790111,
+      "grad_norm": 1.9074355363845825,
+      "learning_rate": 9.825502207712909e-05,
+      "loss": 1.6801,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2623612512613522,
+      "grad_norm": 1.5898246765136719,
+      "learning_rate": 9.821105682262099e-05,
+      "loss": 1.6945,
+      "step": 1300
+    },
+    {
+      "epoch": 0.26437941473259335,
+      "grad_norm": 1.633530616760254,
+      "learning_rate": 9.816655466291803e-05,
+      "loss": 1.723,
+      "step": 1310
+    },
+    {
+      "epoch": 0.26639757820383453,
+      "grad_norm": 1.637403130531311,
+      "learning_rate": 9.812151609362102e-05,
+      "loss": 1.6019,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2684157416750757,
+      "grad_norm": 1.7958626747131348,
+      "learning_rate": 9.807594161630458e-05,
+      "loss": 1.5707,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2704339051463169,
+      "grad_norm": 1.7324177026748657,
+      "learning_rate": 9.802983173851149e-05,
+      "loss": 1.6786,
+      "step": 1340
+    },
+    {
+      "epoch": 0.272452068617558,
+      "grad_norm": 1.55552339553833,
+      "learning_rate": 9.798318697374702e-05,
+      "loss": 1.6684,
+      "step": 1350
+    },
+    {
+      "epoch": 0.27447023208879917,
+      "grad_norm": 1.7341140508651733,
+      "learning_rate": 9.79360078414733e-05,
+      "loss": 1.6447,
+      "step": 1360
+    },
+    {
+      "epoch": 0.27648839556004035,
+      "grad_norm": 1.8685839176177979,
+      "learning_rate": 9.78882948671034e-05,
+      "loss": 1.6794,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2785065590312815,
+      "grad_norm": 1.783153772354126,
+      "learning_rate": 9.784004858199563e-05,
+      "loss": 1.7196,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2805247225025227,
+      "grad_norm": 1.7783135175704956,
+      "learning_rate": 9.779126952344748e-05,
+      "loss": 1.6273,
+      "step": 1390
+    },
+    {
+      "epoch": 0.28254288597376387,
+      "grad_norm": 1.6643409729003906,
+      "learning_rate": 9.774195823468973e-05,
+      "loss": 1.6116,
+      "step": 1400
+    },
+    {
+      "epoch": 0.28456104944500504,
+      "grad_norm": 1.7578685283660889,
+      "learning_rate": 9.769211526488038e-05,
+      "loss": 1.6021,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2865792129162462,
+      "grad_norm": 1.5535609722137451,
+      "learning_rate": 9.764174116909852e-05,
+      "loss": 1.7308,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2885973763874874,
+      "grad_norm": 1.8649767637252808,
+      "learning_rate": 9.759083650833815e-05,
+      "loss": 1.7038,
+      "step": 1430
+    },
+    {
+      "epoch": 0.29061553985872857,
+      "grad_norm": 1.589314341545105,
+      "learning_rate": 9.753940184950192e-05,
+      "loss": 1.6144,
+      "step": 1440
+    },
+    {
+      "epoch": 0.29263370332996974,
+      "grad_norm": 1.9105316400527954,
+      "learning_rate": 9.748743776539488e-05,
+      "loss": 1.7696,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2946518668012109,
+      "grad_norm": 1.9593696594238281,
+      "learning_rate": 9.743494483471801e-05,
+      "loss": 1.6077,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2966700302724521,
+      "grad_norm": 1.767230749130249,
+      "learning_rate": 9.738192364206185e-05,
+      "loss": 1.6627,
+      "step": 1470
+    },
+    {
+      "epoch": 0.29868819374369326,
+      "grad_norm": 1.7512140274047852,
+      "learning_rate": 9.732837477789993e-05,
+      "loss": 1.6611,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3007063572149344,
+      "grad_norm": 1.9849492311477661,
+      "learning_rate": 9.727429883858227e-05,
+      "loss": 1.6493,
+      "step": 1490
+    },
+    {
+      "epoch": 0.30272452068617556,
+      "grad_norm": 1.7700562477111816,
+      "learning_rate": 9.721969642632865e-05,
+      "loss": 1.5956,
+      "step": 1500
+    },
+    {
+      "epoch": 0.30474268415741673,
+      "grad_norm": 1.5900918245315552,
+      "learning_rate": 9.716456814922196e-05,
+      "loss": 1.5717,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3067608476286579,
+      "grad_norm": 1.5321190357208252,
+      "learning_rate": 9.710891462120141e-05,
+      "loss": 1.6783,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3087790110998991,
+      "grad_norm": 1.8479689359664917,
+      "learning_rate": 9.70527364620557e-05,
+      "loss": 1.7176,
+      "step": 1530
+    },
+    {
+      "epoch": 0.31079717457114026,
+      "grad_norm": 1.5251414775848389,
+      "learning_rate": 9.699603429741615e-05,
+      "loss": 1.6911,
+      "step": 1540
+    },
+    {
+      "epoch": 0.31281533804238143,
+      "grad_norm": 1.658829927444458,
+      "learning_rate": 9.693880875874961e-05,
+      "loss": 1.8104,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3148335015136226,
+      "grad_norm": 1.8885022401809692,
+      "learning_rate": 9.68810604833516e-05,
+      "loss": 1.5729,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3168516649848638,
+      "grad_norm": 1.5589392185211182,
+      "learning_rate": 9.682279011433908e-05,
+      "loss": 1.7939,
+      "step": 1570
+    },
+    {
+      "epoch": 0.31886982845610495,
+      "grad_norm": 1.6108678579330444,
+      "learning_rate": 9.676399830064339e-05,
+      "loss": 1.5793,
+      "step": 1580
+    },
+    {
+      "epoch": 0.32088799192734613,
+      "grad_norm": 1.651929259300232,
+      "learning_rate": 9.670468569700288e-05,
+      "loss": 1.6892,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3229061553985873,
+      "grad_norm": 1.899056077003479,
+      "learning_rate": 9.664485296395578e-05,
+      "loss": 1.5486,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3249243188698285,
+      "grad_norm": 1.6485919952392578,
+      "learning_rate": 9.658450076783274e-05,
+      "loss": 1.6536,
+      "step": 1610
+    },
+    {
+      "epoch": 0.32694248234106965,
+      "grad_norm": 1.6275869607925415,
+      "learning_rate": 9.652362978074947e-05,
+      "loss": 1.6136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.32896064581231077,
+      "grad_norm": 1.6739528179168701,
+      "learning_rate": 9.646224068059917e-05,
+      "loss": 1.7081,
+      "step": 1630
+    },
+    {
+      "epoch": 0.33097880928355194,
+      "grad_norm": 1.6004359722137451,
+      "learning_rate": 9.640033415104508e-05,
+      "loss": 1.6391,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3329969727547931,
+      "grad_norm": 1.5969157218933105,
+      "learning_rate": 9.633791088151283e-05,
+      "loss": 1.5738,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3350151362260343,
+      "grad_norm": 1.7182707786560059,
+      "learning_rate": 9.627497156718271e-05,
+      "loss": 1.7195,
+      "step": 1660
+    },
+    {
+      "epoch": 0.33703329969727547,
+      "grad_norm": 1.680814504623413,
+      "learning_rate": 9.621151690898203e-05,
+      "loss": 1.626,
+      "step": 1670
+    },
+    {
+      "epoch": 0.33905146316851664,
+      "grad_norm": 2.010774612426758,
+      "learning_rate": 9.614754761357718e-05,
+      "loss": 1.8037,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3410696266397578,
+      "grad_norm": 1.6631429195404053,
+      "learning_rate": 9.608306439336592e-05,
+      "loss": 1.7439,
+      "step": 1690
+    },
+    {
+      "epoch": 0.343087790110999,
+      "grad_norm": 1.5902904272079468,
+      "learning_rate": 9.60180679664693e-05,
+      "loss": 1.6691,
+      "step": 1700
+    },
+    {
+      "epoch": 0.34510595358224017,
+      "grad_norm": 1.7272005081176758,
+      "learning_rate": 9.595255905672377e-05,
+      "loss": 1.5521,
+      "step": 1710
+    },
+    {
+      "epoch": 0.34712411705348134,
+      "grad_norm": 1.6618432998657227,
+      "learning_rate": 9.588653839367302e-05,
+      "loss": 1.5553,
+      "step": 1720
+    },
+    {
+      "epoch": 0.3491422805247225,
+      "grad_norm": 1.6251617670059204,
+      "learning_rate": 9.582000671256e-05,
+      "loss": 1.6653,
+      "step": 1730
+    },
+    {
+      "epoch": 0.3511604439959637,
+      "grad_norm": 2.123147487640381,
+      "learning_rate": 9.575296475431855e-05,
+      "loss": 1.6303,
+      "step": 1740
+    },
+    {
+      "epoch": 0.35317860746720486,
+      "grad_norm": 1.7185384035110474,
+      "learning_rate": 9.568541326556527e-05,
+      "loss": 1.5665,
+      "step": 1750
+    },
+    {
+      "epoch": 0.35519677093844604,
+      "grad_norm": 1.6353096961975098,
+      "learning_rate": 9.56173529985912e-05,
+      "loss": 1.6406,
+      "step": 1760
+    },
+    {
+      "epoch": 0.35721493440968716,
+      "grad_norm": 1.5363441705703735,
+      "learning_rate": 9.554878471135339e-05,
+      "loss": 1.6211,
+      "step": 1770
+    },
+    {
+      "epoch": 0.35923309788092833,
+      "grad_norm": 1.7037582397460938,
+      "learning_rate": 9.547970916746649e-05,
+      "loss": 1.617,
+      "step": 1780
+    },
+    {
+      "epoch": 0.3612512613521695,
+      "grad_norm": 1.7400474548339844,
+      "learning_rate": 9.541012713619428e-05,
+      "loss": 1.5177,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3632694248234107,
+      "grad_norm": 1.3855012655258179,
+      "learning_rate": 9.5340039392441e-05,
+      "loss": 1.5964,
+      "step": 1800
+    },
+    {
+      "epoch": 0.36528758829465185,
+      "grad_norm": 1.669772744178772,
+      "learning_rate": 9.526944671674286e-05,
+      "loss": 1.5476,
+      "step": 1810
+    },
+    {
+      "epoch": 0.36730575176589303,
+      "grad_norm": 1.547868251800537,
+      "learning_rate": 9.51983498952592e-05,
+      "loss": 1.7272,
+      "step": 1820
+    },
+    {
+      "epoch": 0.3693239152371342,
+      "grad_norm": 1.9002543687820435,
+      "learning_rate": 9.512674971976385e-05,
+      "loss": 1.4606,
+      "step": 1830
+    },
+    {
+      "epoch": 0.3713420787083754,
+      "grad_norm": 1.597686767578125,
+      "learning_rate": 9.505464698763629e-05,
+      "loss": 1.5597,
+      "step": 1840
+    },
+    {
+      "epoch": 0.37336024217961655,
+      "grad_norm": 1.425994634628296,
+      "learning_rate": 9.49820425018527e-05,
+      "loss": 1.5451,
+      "step": 1850
+    },
+    {
+      "epoch": 0.3753784056508577,
+      "grad_norm": 2.0066096782684326,
+      "learning_rate": 9.49089370709771e-05,
+      "loss": 1.6121,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3773965691220989,
+      "grad_norm": 1.5784003734588623,
+      "learning_rate": 9.483533150915229e-05,
+      "loss": 1.6218,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3794147325933401,
+      "grad_norm": 1.843019723892212,
+      "learning_rate": 9.476122663609086e-05,
+      "loss": 1.6694,
+      "step": 1880
+    },
+    {
+      "epoch": 0.38143289606458125,
+      "grad_norm": 1.7118667364120483,
+      "learning_rate": 9.468662327706594e-05,
+      "loss": 1.5564,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3834510595358224,
+      "grad_norm": 1.6396056413650513,
+      "learning_rate": 9.461152226290212e-05,
+      "loss": 1.6626,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3854692230070636,
+      "grad_norm": 1.4227858781814575,
+      "learning_rate": 9.453592442996614e-05,
+      "loss": 1.5375,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3874873864783047,
+      "grad_norm": 1.6100040674209595,
+      "learning_rate": 9.445983062015761e-05,
+      "loss": 1.4965,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3895055499495459,
+      "grad_norm": 1.6516296863555908,
+      "learning_rate": 9.43832416808996e-05,
+      "loss": 1.5941,
+      "step": 1930
+    },
+    {
+      "epoch": 0.39152371342078707,
+      "grad_norm": 1.6929583549499512,
+      "learning_rate": 9.430615846512923e-05,
+      "loss": 1.6325,
+      "step": 1940
+    },
+    {
+      "epoch": 0.39354187689202824,
+      "grad_norm": 1.5503323078155518,
+      "learning_rate": 9.422858183128808e-05,
+      "loss": 1.6673,
+      "step": 1950
+    },
+    {
+      "epoch": 0.3955600403632694,
+      "grad_norm": 1.7053431272506714,
+      "learning_rate": 9.415051264331285e-05,
+      "loss": 1.4846,
+      "step": 1960
+    },
+    {
+      "epoch": 0.3975782038345106,
+      "grad_norm": 1.7111166715621948,
+      "learning_rate": 9.407195177062549e-05,
+      "loss": 1.598,
+      "step": 1970
+    },
+    {
+      "epoch": 0.39959636730575177,
+      "grad_norm": 1.728541374206543,
+      "learning_rate": 9.399290008812365e-05,
+      "loss": 1.4917,
+      "step": 1980
+    },
+    {
+      "epoch": 0.40161453077699294,
+      "grad_norm": 1.6542996168136597,
+      "learning_rate": 9.391335847617093e-05,
+      "loss": 1.6358,
+      "step": 1990
+    },
+    {
+      "epoch": 0.4036326942482341,
+      "grad_norm": 1.4176194667816162,
+      "learning_rate": 9.383332782058705e-05,
+      "loss": 1.6766,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4056508577194753,
+      "grad_norm": 1.4830904006958008,
+      "learning_rate": 9.375280901263796e-05,
+      "loss": 1.5902,
+      "step": 2010
+    },
+    {
+      "epoch": 0.40766902119071646,
+      "grad_norm": 1.6302123069763184,
+      "learning_rate": 9.367180294902603e-05,
+      "loss": 1.613,
+      "step": 2020
+    },
+    {
+      "epoch": 0.40968718466195764,
+      "grad_norm": 1.4098937511444092,
+      "learning_rate": 9.359031053187988e-05,
+      "loss": 1.5243,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4117053481331988,
+      "grad_norm": 2.8872201442718506,
+      "learning_rate": 9.350833266874451e-05,
+      "loss": 1.5927,
+      "step": 2040
+    },
+    {
+      "epoch": 0.41372351160444,
+      "grad_norm": 2.0234012603759766,
+      "learning_rate": 9.342587027257104e-05,
+      "loss": 1.7196,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4157416750756811,
+      "grad_norm": 1.7716432809829712,
+      "learning_rate": 9.334292426170672e-05,
+      "loss": 1.6426,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4177598385469223,
+      "grad_norm": 1.5563617944717407,
+      "learning_rate": 9.325949555988452e-05,
+      "loss": 1.6418,
+      "step": 2070
+    },
+    {
+      "epoch": 0.41977800201816345,
+      "grad_norm": 1.4168179035186768,
+      "learning_rate": 9.317558509621296e-05,
+      "loss": 1.6293,
+      "step": 2080
+    },
+    {
+      "epoch": 0.42179616548940463,
+      "grad_norm": 1.492793321609497,
+      "learning_rate": 9.309119380516573e-05,
+      "loss": 1.5355,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4238143289606458,
+      "grad_norm": 1.7277498245239258,
+      "learning_rate": 9.300632262657128e-05,
+      "loss": 1.6541,
+      "step": 2100
+    },
+    {
+      "epoch": 0.425832492431887,
+      "grad_norm": 1.7189927101135254,
+      "learning_rate": 9.292097250560232e-05,
+      "loss": 1.7026,
+      "step": 2110
+    },
+    {
+      "epoch": 0.42785065590312815,
+      "grad_norm": 1.4795873165130615,
+      "learning_rate": 9.283514439276539e-05,
+      "loss": 1.5959,
+      "step": 2120
+    },
+    {
+      "epoch": 0.4298688193743693,
+      "grad_norm": 1.5818445682525635,
+      "learning_rate": 9.274883924389018e-05,
+      "loss": 1.6147,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4318869828456105,
+      "grad_norm": 1.4752700328826904,
+      "learning_rate": 9.266205802011892e-05,
+      "loss": 1.6281,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4339051463168517,
+      "grad_norm": 1.4389899969100952,
+      "learning_rate": 9.257480168789565e-05,
+      "loss": 1.5655,
+      "step": 2150
+    },
+    {
+      "epoch": 0.43592330978809285,
+      "grad_norm": 1.4133797883987427,
+      "learning_rate": 9.248707121895555e-05,
+      "loss": 1.6437,
+      "step": 2160
+    },
+    {
+      "epoch": 0.437941473259334,
+      "grad_norm": 1.9490972757339478,
+      "learning_rate": 9.239886759031398e-05,
+      "loss": 1.4673,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4399596367305752,
+      "grad_norm": 1.6025413274765015,
+      "learning_rate": 9.231019178425573e-05,
+      "loss": 1.6627,
+      "step": 2180
+    },
+    {
+      "epoch": 0.4419778002018164,
+      "grad_norm": 1.5877257585525513,
+      "learning_rate": 9.222104478832398e-05,
+      "loss": 1.5995,
+      "step": 2190
+    },
+    {
+      "epoch": 0.4439959636730575,
+      "grad_norm": 1.5336265563964844,
+      "learning_rate": 9.213142759530936e-05,
+      "loss": 1.5419,
+      "step": 2200
+    },
+    {
+      "epoch": 0.44601412714429867,
+      "grad_norm": 1.4506264925003052,
+      "learning_rate": 9.204134120323883e-05,
+      "loss": 1.6579,
+      "step": 2210
+    },
+    {
+      "epoch": 0.44803229061553984,
+      "grad_norm": 1.896984338760376,
+      "learning_rate": 9.195078661536471e-05,
+      "loss": 1.5389,
+      "step": 2220
+    },
+    {
+      "epoch": 0.450050454086781,
+      "grad_norm": 1.4249799251556396,
+      "learning_rate": 9.185976484015333e-05,
+      "loss": 1.5563,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4520686175580222,
+      "grad_norm": 1.5877562761306763,
+      "learning_rate": 9.176827689127389e-05,
+      "loss": 1.6289,
+      "step": 2240
+    },
+    {
+      "epoch": 0.45408678102926336,
+      "grad_norm": 1.5677759647369385,
+      "learning_rate": 9.167632378758719e-05,
+      "loss": 1.4721,
+      "step": 2250
+    },
+    {
+      "epoch": 0.45610494450050454,
+      "grad_norm": 1.6074806451797485,
+      "learning_rate": 9.158390655313422e-05,
+      "loss": 1.6101,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4581231079717457,
+      "grad_norm": 1.6484979391098022,
+      "learning_rate": 9.149102621712482e-05,
+      "loss": 1.51,
+      "step": 2270
+    },
+    {
+      "epoch": 0.4601412714429869,
+      "grad_norm": 1.7982864379882812,
+      "learning_rate": 9.139768381392616e-05,
+      "loss": 1.6383,
+      "step": 2280
+    },
+    {
+      "epoch": 0.46215943491422806,
+      "grad_norm": 2.003589391708374,
+      "learning_rate": 9.130388038305127e-05,
+      "loss": 1.6297,
+      "step": 2290
+    },
+    {
+      "epoch": 0.46417759838546924,
+      "grad_norm": 1.659283995628357,
+      "learning_rate": 9.12096169691474e-05,
+      "loss": 1.6508,
+      "step": 2300
+    },
+    {
+      "epoch": 0.4661957618567104,
+      "grad_norm": 1.4378119707107544,
+      "learning_rate": 9.111489462198448e-05,
+      "loss": 1.5852,
+      "step": 2310
+    },
+    {
+      "epoch": 0.4682139253279516,
+      "grad_norm": 1.7221019268035889,
+      "learning_rate": 9.101971439644335e-05,
+      "loss": 1.5671,
+      "step": 2320
+    },
+    {
+      "epoch": 0.47023208879919276,
+      "grad_norm": 1.6366978883743286,
+      "learning_rate": 9.092407735250404e-05,
+      "loss": 1.6594,
+      "step": 2330
+    },
+    {
+      "epoch": 0.4722502522704339,
+      "grad_norm": 1.7291353940963745,
+      "learning_rate": 9.082798455523396e-05,
+      "loss": 1.5025,
+      "step": 2340
+    },
+    {
+      "epoch": 0.47426841574167505,
+      "grad_norm": 1.6143367290496826,
+      "learning_rate": 9.073143707477607e-05,
+      "loss": 1.6939,
+      "step": 2350
+    },
+    {
+      "epoch": 0.47628657921291623,
+      "grad_norm": 1.5067676305770874,
+      "learning_rate": 9.063443598633688e-05,
+      "loss": 1.5281,
+      "step": 2360
+    },
+    {
+      "epoch": 0.4783047426841574,
+      "grad_norm": 1.5984981060028076,
+      "learning_rate": 9.053698237017459e-05,
+      "loss": 1.5932,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4803229061553986,
+      "grad_norm": 1.7212659120559692,
+      "learning_rate": 9.043907731158699e-05,
+      "loss": 1.5234,
+      "step": 2380
+    },
+    {
+      "epoch": 0.48234106962663975,
+      "grad_norm": 1.4098666906356812,
+      "learning_rate": 9.034072190089932e-05,
+      "loss": 1.5491,
+      "step": 2390
+    },
+    {
+      "epoch": 0.4843592330978809,
+      "grad_norm": 1.656089186668396,
+      "learning_rate": 9.02419172334523e-05,
+      "loss": 1.4821,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4863773965691221,
+      "grad_norm": 1.4160220623016357,
+      "learning_rate": 9.014266440958974e-05,
+      "loss": 1.6342,
+      "step": 2410
+    },
+    {
+      "epoch": 0.4883955600403633,
+      "grad_norm": 1.7586220502853394,
+      "learning_rate": 9.004296453464638e-05,
+      "loss": 1.5503,
+      "step": 2420
+    },
+    {
+      "epoch": 0.49041372351160445,
+      "grad_norm": 1.5569350719451904,
+      "learning_rate": 8.994281871893562e-05,
+      "loss": 1.5558,
+      "step": 2430
+    },
+    {
+      "epoch": 0.4924318869828456,
+      "grad_norm": 1.9452998638153076,
+      "learning_rate": 8.984222807773706e-05,
+      "loss": 1.6341,
+      "step": 2440
+    },
+    {
+      "epoch": 0.4944500504540868,
+      "grad_norm": 1.6458998918533325,
+      "learning_rate": 8.974119373128411e-05,
+      "loss": 1.5798,
+      "step": 2450
+    },
+    {
+      "epoch": 0.496468213925328,
+      "grad_norm": 1.7907352447509766,
+      "learning_rate": 8.963971680475161e-05,
+      "loss": 1.499,
+      "step": 2460
+    },
+    {
+      "epoch": 0.49848637739656915,
+      "grad_norm": 1.6107879877090454,
+      "learning_rate": 8.95377984282431e-05,
+      "loss": 1.5705,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5005045408678103,
+      "grad_norm": 1.5708277225494385,
+      "learning_rate": 8.943543973677846e-05,
+      "loss": 1.6062,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5025227043390514,
+      "grad_norm": 1.5355361700057983,
+      "learning_rate": 8.933264187028109e-05,
+      "loss": 1.6155,
+      "step": 2490
+    },
+    {
+      "epoch": 0.5045408678102926,
+      "grad_norm": 2.4249730110168457,
+      "learning_rate": 8.922940597356532e-05,
+      "loss": 1.4927,
+      "step": 2500
+    },
+    {
+      "epoch": 0.5065590312815338,
+      "grad_norm": 1.5792551040649414,
+      "learning_rate": 8.912573319632367e-05,
+      "loss": 1.6917,
+      "step": 2510
+    },
+    {
+      "epoch": 0.508577194752775,
+      "grad_norm": 1.5651421546936035,
+      "learning_rate": 8.90216246931139e-05,
+      "loss": 1.5866,
+      "step": 2520
+    },
+    {
+      "epoch": 0.5105953582240161,
+      "grad_norm": 1.485982060432434,
+      "learning_rate": 8.891708162334635e-05,
+      "loss": 1.5548,
+      "step": 2530
+    },
+    {
+      "epoch": 0.5126135216952573,
+      "grad_norm": 1.4294884204864502,
+      "learning_rate": 8.88121051512709e-05,
+      "loss": 1.5208,
+      "step": 2540
+    },
+    {
+      "epoch": 0.5146316851664985,
+      "grad_norm": 1.7587571144104004,
+      "learning_rate": 8.870669644596402e-05,
+      "loss": 1.5843,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5166498486377397,
+      "grad_norm": 1.257310152053833,
+      "learning_rate": 8.860085668131582e-05,
+      "loss": 1.437,
+      "step": 2560
+    },
+    {
+      "epoch": 0.5186680121089808,
+      "grad_norm": 1.6487629413604736,
+      "learning_rate": 8.84945870360169e-05,
+      "loss": 1.6055,
+      "step": 2570
+    },
+    {
+      "epoch": 0.520686175580222,
+      "grad_norm": 1.415015459060669,
+      "learning_rate": 8.838788869354522e-05,
+      "loss": 1.509,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5227043390514632,
+      "grad_norm": 1.5729633569717407,
+      "learning_rate": 8.828076284215301e-05,
+      "loss": 1.4648,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5247225025227044,
+      "grad_norm": 2.0716793537139893,
+      "learning_rate": 8.817321067485343e-05,
+      "loss": 1.6064,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5267406659939455,
+      "grad_norm": 1.4554495811462402,
+      "learning_rate": 8.806523338940736e-05,
+      "loss": 1.6376,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5287588294651867,
+      "grad_norm": 1.7924822568893433,
+      "learning_rate": 8.795683218831001e-05,
+      "loss": 1.6513,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5307769929364279,
+      "grad_norm": 1.3876484632492065,
+      "learning_rate": 8.78480082787776e-05,
+      "loss": 1.6005,
+      "step": 2630
+    },
+    {
+      "epoch": 0.5327951564076691,
+      "grad_norm": 1.5425324440002441,
+      "learning_rate": 8.773876287273377e-05,
+      "loss": 1.6121,
+      "step": 2640
+    },
+    {
+      "epoch": 0.5348133198789102,
+      "grad_norm": 1.5110645294189453,
+      "learning_rate": 8.762909718679629e-05,
+      "loss": 1.5611,
+      "step": 2650
+    },
+    {
+      "epoch": 0.5368314833501514,
+      "grad_norm": 1.580310583114624,
+      "learning_rate": 8.751901244226332e-05,
+      "loss": 1.636,
+      "step": 2660
+    },
+    {
+      "epoch": 0.5388496468213926,
+      "grad_norm": 1.5676929950714111,
+      "learning_rate": 8.740850986509994e-05,
+      "loss": 1.4185,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5408678102926338,
+      "grad_norm": 1.6521614789962769,
+      "learning_rate": 8.729759068592442e-05,
+      "loss": 1.5195,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5428859737638748,
+      "grad_norm": 1.6536593437194824,
+      "learning_rate": 8.718625613999457e-05,
+      "loss": 1.6065,
+      "step": 2690
+    },
+    {
+      "epoch": 0.544904137235116,
+      "grad_norm": 1.223604679107666,
+      "learning_rate": 8.70745074671939e-05,
+      "loss": 1.5292,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5469223007063572,
+      "grad_norm": 1.5844485759735107,
+      "learning_rate": 8.696234591201793e-05,
+      "loss": 1.5145,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5489404641775983,
+      "grad_norm": 1.4692803621292114,
+      "learning_rate": 8.684977272356024e-05,
+      "loss": 1.5126,
+      "step": 2720
+    },
+    {
+      "epoch": 0.5509586276488395,
+      "grad_norm": 1.509020209312439,
+      "learning_rate": 8.673678915549855e-05,
+      "loss": 1.6746,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5529767911200807,
+      "grad_norm": 1.6188277006149292,
+      "learning_rate": 8.662339646608089e-05,
+      "loss": 1.5323,
+      "step": 2740
+    },
+    {
+      "epoch": 0.5549949545913219,
+      "grad_norm": 1.4918463230133057,
+      "learning_rate": 8.650959591811141e-05,
+      "loss": 1.5413,
+      "step": 2750
+    },
+    {
+      "epoch": 0.557013118062563,
+      "grad_norm": 1.5165650844573975,
+      "learning_rate": 8.639538877893644e-05,
+      "loss": 1.4788,
+      "step": 2760
+    },
+    {
+      "epoch": 0.5590312815338042,
+      "grad_norm": 1.4886066913604736,
+      "learning_rate": 8.628077632043032e-05,
+      "loss": 1.5158,
+      "step": 2770
+    },
+    {
+      "epoch": 0.5610494450050454,
+      "grad_norm": 1.6249706745147705,
+      "learning_rate": 8.616575981898125e-05,
+      "loss": 1.4793,
+      "step": 2780
+    },
+    {
+      "epoch": 0.5630676084762866,
+      "grad_norm": 1.3315315246582031,
+      "learning_rate": 8.605034055547709e-05,
+      "loss": 1.4828,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5650857719475277,
+      "grad_norm": 1.448897123336792,
+      "learning_rate": 8.593451981529108e-05,
+      "loss": 1.5655,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5671039354187689,
+      "grad_norm": 1.4807590246200562,
+      "learning_rate": 8.581829888826754e-05,
+      "loss": 1.5999,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5691220988900101,
+      "grad_norm": 1.7087481021881104,
+      "learning_rate": 8.570167906870745e-05,
+      "loss": 1.591,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5711402623612513,
+      "grad_norm": 1.6135573387145996,
+      "learning_rate": 8.558466165535411e-05,
+      "loss": 1.6063,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5731584258324924,
+      "grad_norm": 1.5271415710449219,
+      "learning_rate": 8.546724795137865e-05,
+      "loss": 1.5007,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5751765893037336,
+      "grad_norm": 1.5794763565063477,
+      "learning_rate": 8.534943926436554e-05,
+      "loss": 1.5415,
+      "step": 2850
+    },
+    {
+      "epoch": 0.5771947527749748,
+      "grad_norm": 1.4942843914031982,
+      "learning_rate": 8.523123690629791e-05,
+      "loss": 1.5891,
+      "step": 2860
+    },
+    {
+      "epoch": 0.579212916246216,
+      "grad_norm": 1.5353363752365112,
+      "learning_rate": 8.511264219354313e-05,
+      "loss": 1.4995,
+      "step": 2870
+    },
+    {
+      "epoch": 0.5812310797174571,
+      "grad_norm": 1.5738115310668945,
+      "learning_rate": 8.4993656446838e-05,
+      "loss": 1.5277,
+      "step": 2880
+    },
+    {
+      "epoch": 0.5832492431886983,
+      "grad_norm": 1.4808515310287476,
+      "learning_rate": 8.48742809912741e-05,
+      "loss": 1.5682,
+      "step": 2890
+    },
+    {
+      "epoch": 0.5852674066599395,
+      "grad_norm": 1.5826464891433716,
+      "learning_rate": 8.475451715628302e-05,
+      "loss": 1.4706,
+      "step": 2900
+    },
+    {
+      "epoch": 0.5872855701311807,
+      "grad_norm": 1.5062007904052734,
+      "learning_rate": 8.463436627562158e-05,
+      "loss": 1.6083,
+      "step": 2910
+    },
+    {
+      "epoch": 0.5893037336024218,
+      "grad_norm": 1.56938898563385,
+      "learning_rate": 8.451382968735693e-05,
+      "loss": 1.4747,
+      "step": 2920
+    },
+    {
+      "epoch": 0.591321897073663,
+      "grad_norm": 1.4118072986602783,
+      "learning_rate": 8.43929087338517e-05,
+      "loss": 1.4768,
+      "step": 2930
+    },
+    {
+      "epoch": 0.5933400605449042,
+      "grad_norm": 1.5036567449569702,
+      "learning_rate": 8.4271604761749e-05,
+      "loss": 1.4414,
+      "step": 2940
+    },
+    {
+      "epoch": 0.5953582240161454,
+      "grad_norm": 1.4589097499847412,
+      "learning_rate": 8.414991912195747e-05,
+      "loss": 1.5648,
+      "step": 2950
+    },
+    {
+      "epoch": 0.5973763874873865,
+      "grad_norm": 1.522581934928894,
+      "learning_rate": 8.402785316963618e-05,
+      "loss": 1.4947,
+      "step": 2960
+    },
+    {
+      "epoch": 0.5993945509586277,
+      "grad_norm": 1.5915330648422241,
+      "learning_rate": 8.390540826417964e-05,
+      "loss": 1.474,
+      "step": 2970
+    },
+    {
+      "epoch": 0.6014127144298688,
+      "grad_norm": 1.5158594846725464,
+      "learning_rate": 8.378258576920253e-05,
+      "loss": 1.5417,
+      "step": 2980
+    },
+    {
+      "epoch": 0.6034308779011099,
+      "grad_norm": 1.5189268589019775,
+      "learning_rate": 8.365938705252459e-05,
+      "loss": 1.4829,
+      "step": 2990
+    },
+    {
+      "epoch": 0.6054490413723511,
+      "grad_norm": 1.3481444120407104,
+      "learning_rate": 8.353581348615538e-05,
+      "loss": 1.5622,
+      "step": 3000
+    },
+    {
+      "epoch": 0.6074672048435923,
+      "grad_norm": 1.482692003250122,
+      "learning_rate": 8.341186644627901e-05,
+      "loss": 1.6286,
+      "step": 3010
+    },
+    {
+      "epoch": 0.6094853683148335,
+      "grad_norm": 1.5743776559829712,
+      "learning_rate": 8.32875473132388e-05,
+      "loss": 1.4813,
+      "step": 3020
+    },
+    {
+      "epoch": 0.6115035317860746,
+      "grad_norm": 1.3886172771453857,
+      "learning_rate": 8.316285747152189e-05,
+      "loss": 1.4297,
+      "step": 3030
+    },
+    {
+      "epoch": 0.6135216952573158,
+      "grad_norm": 1.558706283569336,
+      "learning_rate": 8.30377983097438e-05,
+      "loss": 1.4639,
+      "step": 3040
+    },
+    {
+      "epoch": 0.615539858728557,
+      "grad_norm": 1.537921667098999,
+      "learning_rate": 8.291237122063309e-05,
+      "loss": 1.5532,
+      "step": 3050
+    },
+    {
+      "epoch": 0.6175580221997982,
+      "grad_norm": 1.2276337146759033,
+      "learning_rate": 8.27865776010157e-05,
+      "loss": 1.4639,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6195761856710393,
+      "grad_norm": 1.4864561557769775,
+      "learning_rate": 8.266041885179949e-05,
+      "loss": 1.436,
+      "step": 3070
+    },
+    {
+      "epoch": 0.6215943491422805,
+      "grad_norm": 1.7501429319381714,
+      "learning_rate": 8.253389637795858e-05,
+      "loss": 1.4653,
+      "step": 3080
+    },
+    {
+      "epoch": 0.6236125126135217,
+      "grad_norm": 1.4437967538833618,
+      "learning_rate": 8.240701158851778e-05,
+      "loss": 1.4266,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6256306760847629,
+      "grad_norm": 1.3351330757141113,
+      "learning_rate": 8.227976589653676e-05,
+      "loss": 1.4005,
+      "step": 3100
+    },
+    {
+      "epoch": 0.627648839556004,
+      "grad_norm": 1.4452919960021973,
+      "learning_rate": 8.215216071909448e-05,
+      "loss": 1.5651,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6296670030272452,
+      "grad_norm": 1.4321256875991821,
+      "learning_rate": 8.202419747727333e-05,
+      "loss": 1.4941,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6316851664984864,
+      "grad_norm": 1.3962360620498657,
+      "learning_rate": 8.189587759614325e-05,
+      "loss": 1.4671,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6337033299697276,
+      "grad_norm": 1.4179112911224365,
+      "learning_rate": 8.176720250474594e-05,
+      "loss": 1.4636,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6357214934409687,
+      "grad_norm": 1.4552099704742432,
+      "learning_rate": 8.163817363607894e-05,
+      "loss": 1.5253,
+      "step": 3150
+    },
+    {
+      "epoch": 0.6377396569122099,
+      "grad_norm": 1.6266930103302002,
+      "learning_rate": 8.150879242707962e-05,
+      "loss": 1.4704,
+      "step": 3160
+    },
+    {
+      "epoch": 0.6397578203834511,
+      "grad_norm": 1.480033278465271,
+      "learning_rate": 8.137906031860925e-05,
+      "loss": 1.5921,
+      "step": 3170
+    },
+    {
+      "epoch": 0.6417759838546923,
+      "grad_norm": 1.4520894289016724,
+      "learning_rate": 8.124897875543684e-05,
+      "loss": 1.4444,
+      "step": 3180
+    },
+    {
+      "epoch": 0.6437941473259334,
+      "grad_norm": 1.3743687868118286,
+      "learning_rate": 8.111854918622321e-05,
+      "loss": 1.6175,
+      "step": 3190
+    },
+    {
+      "epoch": 0.6458123107971746,
+      "grad_norm": 1.5462607145309448,
+      "learning_rate": 8.098777306350469e-05,
+      "loss": 1.4526,
+      "step": 3200
+    },
+    {
+      "epoch": 0.6478304742684158,
+      "grad_norm": 1.4813790321350098,
+      "learning_rate": 8.08566518436771e-05,
+      "loss": 1.5253,
+      "step": 3210
+    },
+    {
+      "epoch": 0.649848637739657,
+      "grad_norm": 1.5957119464874268,
+      "learning_rate": 8.072518698697938e-05,
+      "loss": 1.4505,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6518668012108981,
+      "grad_norm": 1.5894341468811035,
+      "learning_rate": 8.059337995747743e-05,
+      "loss": 1.4643,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6538849646821393,
+      "grad_norm": 1.4520728588104248,
+      "learning_rate": 8.046123222304781e-05,
+      "loss": 1.5529,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6559031281533805,
+      "grad_norm": 1.379225730895996,
+      "learning_rate": 8.032874525536131e-05,
+      "loss": 1.4944,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6579212916246215,
+      "grad_norm": 1.4799381494522095,
+      "learning_rate": 8.019592052986665e-05,
+      "loss": 1.3809,
+      "step": 3260
+    },
+    {
+      "epoch": 0.6599394550958627,
+      "grad_norm": 1.3304007053375244,
+      "learning_rate": 8.006275952577397e-05,
+      "loss": 1.5455,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6619576185671039,
+      "grad_norm": 1.4968523979187012,
+      "learning_rate": 7.992926372603842e-05,
+      "loss": 1.4879,
+      "step": 3280
+    },
+    {
+      "epoch": 0.6639757820383451,
+      "grad_norm": 1.582527756690979,
+      "learning_rate": 7.979543461734362e-05,
+      "loss": 1.4796,
+      "step": 3290
+    },
+    {
+      "epoch": 0.6659939455095862,
+      "grad_norm": 1.3123412132263184,
+      "learning_rate": 7.966127369008512e-05,
+      "loss": 1.4623,
+      "step": 3300
+    },
+    {
+      "epoch": 0.6680121089808274,
+      "grad_norm": 1.3255366086959839,
+      "learning_rate": 7.952678243835376e-05,
+      "loss": 1.4871,
+      "step": 3310
+    },
+    {
+      "epoch": 0.6700302724520686,
+      "grad_norm": 1.375701904296875,
+      "learning_rate": 7.939196235991904e-05,
+      "loss": 1.49,
+      "step": 3320
+    },
+    {
+      "epoch": 0.6720484359233098,
+      "grad_norm": 1.5660688877105713,
+      "learning_rate": 7.925681495621253e-05,
+      "loss": 1.5556,
+      "step": 3330
+    },
+    {
+      "epoch": 0.6740665993945509,
+      "grad_norm": 1.2856935262680054,
+      "learning_rate": 7.912134173231098e-05,
+      "loss": 1.4971,
+      "step": 3340
+    },
+    {
+      "epoch": 0.6760847628657921,
+      "grad_norm": 1.7174125909805298,
+      "learning_rate": 7.898554419691974e-05,
+      "loss": 1.506,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6781029263370333,
+      "grad_norm": 1.4166866540908813,
+      "learning_rate": 7.884942386235582e-05,
+      "loss": 1.367,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6801210898082745,
+      "grad_norm": 1.3765437602996826,
+      "learning_rate": 7.871298224453113e-05,
+      "loss": 1.4017,
+      "step": 3370
+    },
+    {
+      "epoch": 0.6821392532795156,
+      "grad_norm": 1.4745761156082153,
+      "learning_rate": 7.857622086293557e-05,
+      "loss": 1.6014,
+      "step": 3380
+    },
+    {
+      "epoch": 0.6841574167507568,
+      "grad_norm": 1.5563029050827026,
+      "learning_rate": 7.843914124062006e-05,
+      "loss": 1.4713,
+      "step": 3390
+    },
+    {
+      "epoch": 0.686175580221998,
+      "grad_norm": 1.6175247430801392,
+      "learning_rate": 7.830174490417972e-05,
+      "loss": 1.5117,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6881937436932392,
+      "grad_norm": 1.4464702606201172,
+      "learning_rate": 7.816403338373666e-05,
+      "loss": 1.4251,
+      "step": 3410
+    },
+    {
+      "epoch": 0.6902119071644803,
+      "grad_norm": 1.3877936601638794,
+      "learning_rate": 7.802600821292314e-05,
+      "loss": 1.3907,
+      "step": 3420
+    },
+    {
+      "epoch": 0.6922300706357215,
+      "grad_norm": 1.4214582443237305,
+      "learning_rate": 7.78876709288644e-05,
+      "loss": 1.4608,
+      "step": 3430
+    },
+    {
+      "epoch": 0.6942482341069627,
+      "grad_norm": 1.3867719173431396,
+      "learning_rate": 7.774902307216148e-05,
+      "loss": 1.5583,
+      "step": 3440
+    },
+    {
+      "epoch": 0.6962663975782039,
+      "grad_norm": 1.345284104347229,
+      "learning_rate": 7.76100661868742e-05,
+      "loss": 1.5107,
+      "step": 3450
+    },
+    {
+      "epoch": 0.698284561049445,
+      "grad_norm": 1.4984257221221924,
+      "learning_rate": 7.747080182050388e-05,
+      "loss": 1.4186,
+      "step": 3460
+    },
+    {
+      "epoch": 0.7003027245206862,
+      "grad_norm": 1.4104434251785278,
+      "learning_rate": 7.733123152397609e-05,
+      "loss": 1.4989,
+      "step": 3470
+    },
+    {
+      "epoch": 0.7023208879919274,
+      "grad_norm": 1.3834434747695923,
+      "learning_rate": 7.719135685162342e-05,
+      "loss": 1.4089,
+      "step": 3480
+    },
+    {
+      "epoch": 0.7043390514631686,
+      "grad_norm": 1.4428340196609497,
+      "learning_rate": 7.705117936116822e-05,
+      "loss": 1.5516,
+      "step": 3490
+    },
+    {
+      "epoch": 0.7063572149344097,
+      "grad_norm": 1.5729140043258667,
+      "learning_rate": 7.691070061370507e-05,
+      "loss": 1.5622,
+      "step": 3500
+    },
+    {
+      "epoch": 0.7083753784056509,
+      "grad_norm": 1.3539918661117554,
+      "learning_rate": 7.676992217368364e-05,
+      "loss": 1.4938,
+      "step": 3510
+    },
+    {
+      "epoch": 0.7103935418768921,
+      "grad_norm": 1.540166974067688,
+      "learning_rate": 7.662884560889105e-05,
+      "loss": 1.3785,
+      "step": 3520
+    },
+    {
+      "epoch": 0.7124117053481333,
+      "grad_norm": 1.3016571998596191,
+      "learning_rate": 7.648747249043457e-05,
+      "loss": 1.5543,
+      "step": 3530
+    },
+    {
+      "epoch": 0.7144298688193743,
+      "grad_norm": 1.4095804691314697,
+      "learning_rate": 7.634580439272401e-05,
+      "loss": 1.5495,
+      "step": 3540
+    },
+    {
+      "epoch": 0.7164480322906155,
+      "grad_norm": 1.280922770500183,
+      "learning_rate": 7.620384289345425e-05,
+      "loss": 1.5104,
+      "step": 3550
+    },
+    {
+      "epoch": 0.7184661957618567,
+      "grad_norm": 1.3793776035308838,
+      "learning_rate": 7.606158957358769e-05,
+      "loss": 1.5002,
+      "step": 3560
+    },
+    {
+      "epoch": 0.7204843592330978,
+      "grad_norm": 1.5305780172348022,
+      "learning_rate": 7.591904601733655e-05,
+      "loss": 1.5148,
+      "step": 3570
+    },
+    {
+      "epoch": 0.722502522704339,
+      "grad_norm": 1.3183702230453491,
+      "learning_rate": 7.577621381214529e-05,
+      "loss": 1.4927,
+      "step": 3580
+    },
+    {
+      "epoch": 0.7245206861755802,
+      "grad_norm": 1.2844158411026,
+      "learning_rate": 7.563309454867295e-05,
+      "loss": 1.4843,
+      "step": 3590
+    },
+    {
+      "epoch": 0.7265388496468214,
+      "grad_norm": 1.252433180809021,
+      "learning_rate": 7.548968982077542e-05,
+      "loss": 1.4047,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7285570131180625,
+      "grad_norm": 1.1570249795913696,
+      "learning_rate": 7.534600122548765e-05,
+      "loss": 1.4062,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7305751765893037,
+      "grad_norm": 1.3853447437286377,
+      "learning_rate": 7.520203036300588e-05,
+      "loss": 1.4593,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7325933400605449,
+      "grad_norm": 1.6483482122421265,
+      "learning_rate": 7.505777883666993e-05,
+      "loss": 1.6098,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7346115035317861,
+      "grad_norm": 1.3295645713806152,
+      "learning_rate": 7.491324825294514e-05,
+      "loss": 1.4467,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7366296670030272,
+      "grad_norm": 1.336206078529358,
+      "learning_rate": 7.476844022140464e-05,
+      "loss": 1.5218,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7386478304742684,
+      "grad_norm": 1.2594281435012817,
+      "learning_rate": 7.462335635471136e-05,
+      "loss": 1.4216,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7406659939455096,
+      "grad_norm": 2.950639009475708,
+      "learning_rate": 7.44779982686001e-05,
+      "loss": 1.4549,
+      "step": 3670
+    },
+    {
+      "epoch": 0.7426841574167508,
+      "grad_norm": 1.4663448333740234,
+      "learning_rate": 7.43323675818595e-05,
+      "loss": 1.3631,
+      "step": 3680
+    },
+    {
+      "epoch": 0.7447023208879919,
+      "grad_norm": 1.5686269998550415,
+      "learning_rate": 7.418646591631404e-05,
+      "loss": 1.3991,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7467204843592331,
+      "grad_norm": 1.5072070360183716,
+      "learning_rate": 7.404029489680598e-05,
+      "loss": 1.4257,
+      "step": 3700
+    },
+    {
+      "epoch": 0.7487386478304743,
+      "grad_norm": 1.391025185585022,
+      "learning_rate": 7.389385615117723e-05,
+      "loss": 1.4345,
+      "step": 3710
+    },
+    {
+      "epoch": 0.7507568113017155,
+      "grad_norm": 1.382051944732666,
+      "learning_rate": 7.37471513102513e-05,
+      "loss": 1.4038,
+      "step": 3720
+    },
+    {
+      "epoch": 0.7527749747729566,
+      "grad_norm": 1.3509782552719116,
+      "learning_rate": 7.360018200781502e-05,
+      "loss": 1.4457,
+      "step": 3730
+    },
+    {
+      "epoch": 0.7547931382441978,
+      "grad_norm": 1.6141653060913086,
+      "learning_rate": 7.345294988060046e-05,
+      "loss": 1.5944,
+      "step": 3740
+    },
+    {
+      "epoch": 0.756811301715439,
+      "grad_norm": 1.4731391668319702,
+      "learning_rate": 7.330545656826662e-05,
+      "loss": 1.4548,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7588294651866802,
+      "grad_norm": 1.3180512189865112,
+      "learning_rate": 7.315770371338126e-05,
+      "loss": 1.3972,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7608476286579213,
+      "grad_norm": 1.3752028942108154,
+      "learning_rate": 7.300969296140244e-05,
+      "loss": 1.5337,
+      "step": 3770
+    },
+    {
+      "epoch": 0.7628657921291625,
+      "grad_norm": 1.5503935813903809,
+      "learning_rate": 7.286142596066044e-05,
+      "loss": 1.4658,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7648839556004037,
+      "grad_norm": 1.3432625532150269,
+      "learning_rate": 7.271290436233916e-05,
+      "loss": 1.5027,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7669021190716448,
+      "grad_norm": 1.6639512777328491,
+      "learning_rate": 7.25641298204579e-05,
+      "loss": 1.4577,
+      "step": 3800
+    },
+    {
+      "epoch": 0.768920282542886,
+      "grad_norm": 1.5050352811813354,
+      "learning_rate": 7.241510399185287e-05,
+      "loss": 1.4345,
+      "step": 3810
+    },
+    {
+      "epoch": 0.7709384460141272,
+      "grad_norm": 1.6508047580718994,
+      "learning_rate": 7.226582853615874e-05,
+      "loss": 1.359,
+      "step": 3820
+    },
+    {
+      "epoch": 0.7729566094853683,
+      "grad_norm": 1.489529013633728,
+      "learning_rate": 7.211630511579015e-05,
+      "loss": 1.4296,
+      "step": 3830
+    },
+    {
+      "epoch": 0.7749747729566094,
+      "grad_norm": 1.270341396331787,
+      "learning_rate": 7.196653539592326e-05,
+      "loss": 1.4177,
+      "step": 3840
+    },
+    {
+      "epoch": 0.7769929364278506,
+      "grad_norm": 1.2509267330169678,
+      "learning_rate": 7.181652104447711e-05,
+      "loss": 1.4425,
+      "step": 3850
+    },
+    {
+      "epoch": 0.7790110998990918,
+      "grad_norm": 1.4303765296936035,
+      "learning_rate": 7.166626373209514e-05,
+      "loss": 1.4735,
+      "step": 3860
+    },
+    {
+      "epoch": 0.781029263370333,
+      "grad_norm": 1.3523200750350952,
+      "learning_rate": 7.15157651321265e-05,
+      "loss": 1.5438,
+      "step": 3870
+    },
+    {
+      "epoch": 0.7830474268415741,
+      "grad_norm": 1.3462188243865967,
+      "learning_rate": 7.136502692060746e-05,
+      "loss": 1.5316,
+      "step": 3880
+    },
+    {
+      "epoch": 0.7850655903128153,
+      "grad_norm": 1.3250367641448975,
+      "learning_rate": 7.121405077624276e-05,
+      "loss": 1.4727,
+      "step": 3890
+    },
+    {
+      "epoch": 0.7870837537840565,
+      "grad_norm": 1.267572045326233,
+      "learning_rate": 7.106283838038685e-05,
+      "loss": 1.4804,
+      "step": 3900
+    },
+    {
+      "epoch": 0.7891019172552977,
+      "grad_norm": 1.2590205669403076,
+      "learning_rate": 7.091139141702527e-05,
+      "loss": 1.3984,
+      "step": 3910
+    },
+    {
+      "epoch": 0.7911200807265388,
+      "grad_norm": 1.4484533071517944,
+      "learning_rate": 7.075971157275575e-05,
+      "loss": 1.4728,
+      "step": 3920
+    },
+    {
+      "epoch": 0.79313824419778,
+      "grad_norm": 1.4239338636398315,
+      "learning_rate": 7.06078005367696e-05,
+      "loss": 1.4305,
+      "step": 3930
+    },
+    {
+      "epoch": 0.7951564076690212,
+      "grad_norm": 1.4230101108551025,
+      "learning_rate": 7.045566000083278e-05,
+      "loss": 1.5003,
+      "step": 3940
+    },
+    {
+      "epoch": 0.7971745711402624,
+      "grad_norm": 1.2686034440994263,
+      "learning_rate": 7.030329165926706e-05,
+      "loss": 1.4169,
+      "step": 3950
+    },
+    {
+      "epoch": 0.7991927346115035,
+      "grad_norm": 1.5182725191116333,
+      "learning_rate": 7.01506972089312e-05,
+      "loss": 1.5385,
+      "step": 3960
+    },
+    {
+      "epoch": 0.8012108980827447,
+      "grad_norm": 1.4067586660385132,
+      "learning_rate": 6.999787834920202e-05,
+      "loss": 1.3908,
+      "step": 3970
+    },
+    {
+      "epoch": 0.8032290615539859,
+      "grad_norm": 1.2837873697280884,
+      "learning_rate": 6.984483678195553e-05,
+      "loss": 1.4505,
+      "step": 3980
+    },
+    {
+      "epoch": 0.805247225025227,
+      "grad_norm": 1.4042167663574219,
+      "learning_rate": 6.969157421154789e-05,
+      "loss": 1.4897,
+      "step": 3990
+    },
+    {
+      "epoch": 0.8072653884964682,
+      "grad_norm": 1.6277389526367188,
+      "learning_rate": 6.95380923447965e-05,
+      "loss": 1.4117,
+      "step": 4000
+    },
+    {
+      "epoch": 0.8092835519677094,
+      "grad_norm": 1.572359323501587,
+      "learning_rate": 6.938439289096095e-05,
+      "loss": 1.4158,
+      "step": 4010
+    },
+    {
+      "epoch": 0.8113017154389506,
+      "grad_norm": 1.5279749631881714,
+      "learning_rate": 6.923047756172401e-05,
+      "loss": 1.455,
+      "step": 4020
+    },
+    {
+      "epoch": 0.8133198789101918,
+      "grad_norm": 1.298711895942688,
+      "learning_rate": 6.907634807117257e-05,
+      "loss": 1.3315,
+      "step": 4030
+    },
+    {
+      "epoch": 0.8153380423814329,
+      "grad_norm": 1.2476575374603271,
+      "learning_rate": 6.892200613577852e-05,
+      "loss": 1.514,
+      "step": 4040
+    },
+    {
+      "epoch": 0.8173562058526741,
+      "grad_norm": 1.289162278175354,
+      "learning_rate": 6.876745347437964e-05,
+      "loss": 1.4861,
+      "step": 4050
+    },
+    {
+      "epoch": 0.8193743693239153,
+      "grad_norm": 1.2558218240737915,
+      "learning_rate": 6.861269180816052e-05,
+      "loss": 1.3809,
+      "step": 4060
+    },
+    {
+      "epoch": 0.8213925327951564,
+      "grad_norm": 1.2769505977630615,
+      "learning_rate": 6.845772286063332e-05,
+      "loss": 1.4678,
+      "step": 4070
+    },
+    {
+      "epoch": 0.8234106962663976,
+      "grad_norm": 1.4346575736999512,
+      "learning_rate": 6.830254835761856e-05,
+      "loss": 1.4251,
+      "step": 4080
+    },
+    {
+      "epoch": 0.8254288597376388,
+      "grad_norm": 1.4318442344665527,
+      "learning_rate": 6.814717002722602e-05,
+      "loss": 1.5475,
+      "step": 4090
+    },
+    {
+      "epoch": 0.82744702320888,
+      "grad_norm": 1.2699110507965088,
+      "learning_rate": 6.799158959983536e-05,
+      "loss": 1.4722,
+      "step": 4100
+    },
+    {
+      "epoch": 0.829465186680121,
+      "grad_norm": 1.5809247493743896,
+      "learning_rate": 6.78358088080769e-05,
+      "loss": 1.5189,
+      "step": 4110
+    },
+    {
+      "epoch": 0.8314833501513622,
+      "grad_norm": 1.4829879999160767,
+      "learning_rate": 6.767982938681239e-05,
+      "loss": 1.5611,
+      "step": 4120
+    },
+    {
+      "epoch": 0.8335015136226034,
+      "grad_norm": 1.251018762588501,
+      "learning_rate": 6.752365307311556e-05,
+      "loss": 1.4698,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8355196770938446,
+      "grad_norm": 1.1748842000961304,
+      "learning_rate": 6.736728160625284e-05,
+      "loss": 1.5476,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8375378405650857,
+      "grad_norm": 1.4876999855041504,
+      "learning_rate": 6.721071672766406e-05,
+      "loss": 1.4378,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8395560040363269,
+      "grad_norm": 1.4530222415924072,
+      "learning_rate": 6.705396018094297e-05,
+      "loss": 1.4794,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8415741675075681,
+      "grad_norm": 1.2909533977508545,
+      "learning_rate": 6.689701371181781e-05,
+      "loss": 1.3877,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8435923309788093,
+      "grad_norm": 1.4188405275344849,
+      "learning_rate": 6.673987906813191e-05,
+      "loss": 1.3629,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8456104944500504,
+      "grad_norm": 1.498369812965393,
+      "learning_rate": 6.658255799982424e-05,
+      "loss": 1.4719,
+      "step": 4190
+    },
+    {
+      "epoch": 0.8476286579212916,
+      "grad_norm": 1.3397397994995117,
+      "learning_rate": 6.642505225890987e-05,
+      "loss": 1.3999,
+      "step": 4200
+    },
+    {
+      "epoch": 0.8496468213925328,
+      "grad_norm": 1.3505257368087769,
+      "learning_rate": 6.626736359946052e-05,
+      "loss": 1.4824,
+      "step": 4210
+    },
+    {
+      "epoch": 0.851664984863774,
+      "grad_norm": 1.312658667564392,
+      "learning_rate": 6.610949377758497e-05,
+      "loss": 1.4902,
+      "step": 4220
+    },
+    {
+      "epoch": 0.8536831483350151,
+      "grad_norm": 1.4204723834991455,
+      "learning_rate": 6.595144455140952e-05,
+      "loss": 1.4635,
+      "step": 4230
+    },
+    {
+      "epoch": 0.8557013118062563,
+      "grad_norm": 1.2958680391311646,
+      "learning_rate": 6.579321768105845e-05,
+      "loss": 1.4672,
+      "step": 4240
+    },
+    {
+      "epoch": 0.8577194752774975,
+      "grad_norm": 1.4016082286834717,
+      "learning_rate": 6.563481492863436e-05,
+      "loss": 1.4476,
+      "step": 4250
+    },
+    {
+      "epoch": 0.8597376387487387,
+      "grad_norm": 1.3869260549545288,
+      "learning_rate": 6.547623805819854e-05,
+      "loss": 1.4194,
+      "step": 4260
+    },
+    {
+      "epoch": 0.8617558022199798,
+      "grad_norm": 1.2927767038345337,
+      "learning_rate": 6.531748883575143e-05,
+      "loss": 1.4523,
+      "step": 4270
+    },
+    {
+      "epoch": 0.863773965691221,
+      "grad_norm": 2.418339729309082,
+      "learning_rate": 6.51585690292128e-05,
+      "loss": 1.3799,
+      "step": 4280
+    },
+    {
+      "epoch": 0.8657921291624622,
+      "grad_norm": 1.6442086696624756,
+      "learning_rate": 6.499948040840219e-05,
+      "loss": 1.4596,
+      "step": 4290
+    },
+    {
+      "epoch": 0.8678102926337034,
+      "grad_norm": 1.377209186553955,
+      "learning_rate": 6.484022474501914e-05,
+      "loss": 1.4226,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8698284561049445,
+      "grad_norm": 1.3288756608963013,
+      "learning_rate": 6.468080381262347e-05,
+      "loss": 1.4244,
+      "step": 4310
+    },
+    {
+      "epoch": 0.8718466195761857,
+      "grad_norm": 1.6183438301086426,
+      "learning_rate": 6.45212193866155e-05,
+      "loss": 1.4628,
+      "step": 4320
+    },
+    {
+      "epoch": 0.8738647830474269,
+      "grad_norm": 1.3636276721954346,
+      "learning_rate": 6.436147324421635e-05,
+      "loss": 1.4082,
+      "step": 4330
+    },
+    {
+      "epoch": 0.875882946518668,
+      "grad_norm": 1.5240586996078491,
+      "learning_rate": 6.420156716444805e-05,
+      "loss": 1.4043,
+      "step": 4340
+    },
+    {
+      "epoch": 0.8779011099899092,
+      "grad_norm": 1.2216293811798096,
+      "learning_rate": 6.404150292811386e-05,
+      "loss": 1.3972,
+      "step": 4350
+    },
+    {
+      "epoch": 0.8799192734611504,
+      "grad_norm": 1.447521448135376,
+      "learning_rate": 6.388128231777828e-05,
+      "loss": 1.5133,
+      "step": 4360
+    },
+    {
+      "epoch": 0.8819374369323916,
+      "grad_norm": 1.3184555768966675,
+      "learning_rate": 6.372090711774732e-05,
+      "loss": 1.3777,
+      "step": 4370
+    },
+    {
+      "epoch": 0.8839556004036327,
+      "grad_norm": 1.3064029216766357,
+      "learning_rate": 6.356037911404858e-05,
+      "loss": 1.4235,
+      "step": 4380
+    },
+    {
+      "epoch": 0.8859737638748738,
+      "grad_norm": 1.252462387084961,
+      "learning_rate": 6.339970009441137e-05,
+      "loss": 1.4071,
+      "step": 4390
+    },
+    {
+      "epoch": 0.887991927346115,
+      "grad_norm": 1.3973218202590942,
+      "learning_rate": 6.323887184824678e-05,
+      "loss": 1.397,
+      "step": 4400
+    },
+    {
+      "epoch": 0.8900100908173562,
+      "grad_norm": 1.3893166780471802,
+      "learning_rate": 6.307789616662778e-05,
+      "loss": 1.3642,
+      "step": 4410
+    },
+    {
+      "epoch": 0.8920282542885973,
+      "grad_norm": 1.329697847366333,
+      "learning_rate": 6.291677484226929e-05,
+      "loss": 1.5611,
+      "step": 4420
+    },
+    {
+      "epoch": 0.8940464177598385,
+      "grad_norm": 1.4498480558395386,
+      "learning_rate": 6.275550966950814e-05,
+      "loss": 1.49,
+      "step": 4430
+    },
+    {
+      "epoch": 0.8960645812310797,
+      "grad_norm": 1.3522253036499023,
+      "learning_rate": 6.259410244428318e-05,
+      "loss": 1.3945,
+      "step": 4440
+    },
+    {
+      "epoch": 0.8980827447023209,
+      "grad_norm": 1.3933300971984863,
+      "learning_rate": 6.243255496411519e-05,
+      "loss": 1.4169,
+      "step": 4450
+    },
+    {
+      "epoch": 0.900100908173562,
+      "grad_norm": 1.4387352466583252,
+      "learning_rate": 6.227086902808697e-05,
+      "loss": 1.5595,
+      "step": 4460
+    },
+    {
+      "epoch": 0.9021190716448032,
+      "grad_norm": 1.339574933052063,
+      "learning_rate": 6.210904643682318e-05,
+      "loss": 1.4787,
+      "step": 4470
+    },
+    {
+      "epoch": 0.9041372351160444,
+      "grad_norm": 1.3527144193649292,
+      "learning_rate": 6.194708899247037e-05,
+      "loss": 1.4132,
+      "step": 4480
+    },
+    {
+      "epoch": 0.9061553985872856,
+      "grad_norm": 1.471655249595642,
+      "learning_rate": 6.178499849867689e-05,
+      "loss": 1.4548,
+      "step": 4490
+    },
+    {
+      "epoch": 0.9081735620585267,
+      "grad_norm": 1.5146980285644531,
+      "learning_rate": 6.162277676057284e-05,
+      "loss": 1.4628,
+      "step": 4500
+    },
+    {
+      "epoch": 0.9101917255297679,
+      "grad_norm": 1.2837083339691162,
+      "learning_rate": 6.146042558474987e-05,
+      "loss": 1.4305,
+      "step": 4510
+    },
+    {
+      "epoch": 0.9122098890010091,
+      "grad_norm": 1.2286232709884644,
+      "learning_rate": 6.129794677924113e-05,
+      "loss": 1.4211,
+      "step": 4520
+    },
+    {
+      "epoch": 0.9142280524722503,
+      "grad_norm": 1.3042244911193848,
+      "learning_rate": 6.113534215350116e-05,
+      "loss": 1.4328,
+      "step": 4530
+    },
+    {
+      "epoch": 0.9162462159434914,
+      "grad_norm": 1.2900363206863403,
+      "learning_rate": 6.097261351838569e-05,
+      "loss": 1.591,
+      "step": 4540
+    },
+    {
+      "epoch": 0.9182643794147326,
+      "grad_norm": 1.315226674079895,
+      "learning_rate": 6.0809762686131474e-05,
+      "loss": 1.3962,
+      "step": 4550
+    },
+    {
+      "epoch": 0.9202825428859738,
+      "grad_norm": 1.2552099227905273,
+      "learning_rate": 6.064679147033614e-05,
+      "loss": 1.5005,
+      "step": 4560
+    },
+    {
+      "epoch": 0.922300706357215,
+      "grad_norm": 1.4059544801712036,
+      "learning_rate": 6.0483701685937954e-05,
+      "loss": 1.4515,
+      "step": 4570
+    },
+    {
+      "epoch": 0.9243188698284561,
+      "grad_norm": 1.4410680532455444,
+      "learning_rate": 6.0320495149195644e-05,
+      "loss": 1.4045,
+      "step": 4580
+    },
+    {
+      "epoch": 0.9263370332996973,
+      "grad_norm": 1.1760427951812744,
+      "learning_rate": 6.015717367766815e-05,
+      "loss": 1.5034,
+      "step": 4590
+    },
+    {
+      "epoch": 0.9283551967709385,
+      "grad_norm": 1.3556947708129883,
+      "learning_rate": 5.999373909019437e-05,
+      "loss": 1.4571,
+      "step": 4600
+    },
+    {
+      "epoch": 0.9303733602421796,
+      "grad_norm": 1.2202001810073853,
+      "learning_rate": 5.9830193206872974e-05,
+      "loss": 1.4304,
+      "step": 4610
+    },
+    {
+      "epoch": 0.9323915237134208,
+      "grad_norm": 1.2118003368377686,
+      "learning_rate": 5.966653784904207e-05,
+      "loss": 1.4254,
+      "step": 4620
+    },
+    {
+      "epoch": 0.934409687184662,
+      "grad_norm": 1.5171030759811401,
+      "learning_rate": 5.950277483925889e-05,
+      "loss": 1.4243,
+      "step": 4630
+    },
+    {
+      "epoch": 0.9364278506559032,
+      "grad_norm": 1.0997145175933838,
+      "learning_rate": 5.933890600127958e-05,
+      "loss": 1.4485,
+      "step": 4640
+    },
+    {
+      "epoch": 0.9384460141271443,
+      "grad_norm": 1.3518075942993164,
+      "learning_rate": 5.917493316003884e-05,
+      "loss": 1.4907,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9404641775983855,
+      "grad_norm": 1.3248131275177002,
+      "learning_rate": 5.90108581416296e-05,
+      "loss": 1.4629,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9424823410696267,
+      "grad_norm": 1.2627878189086914,
+      "learning_rate": 5.8846682773282694e-05,
+      "loss": 1.451,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9445005045408678,
+      "grad_norm": 1.4110251665115356,
+      "learning_rate": 5.868240888334653e-05,
+      "loss": 1.4425,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9465186680121089,
+      "grad_norm": 1.2221981287002563,
+      "learning_rate": 5.851803830126666e-05,
+      "loss": 1.5313,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9485368314833501,
+      "grad_norm": 1.2160431146621704,
+      "learning_rate": 5.835357285756552e-05,
+      "loss": 1.3897,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9505549949545913,
+      "grad_norm": 1.2338696718215942,
+      "learning_rate": 5.8189014383821914e-05,
+      "loss": 1.311,
+      "step": 4710
+    },
+    {
+      "epoch": 0.9525731584258325,
+      "grad_norm": 1.3585981130599976,
+      "learning_rate": 5.8024364712650724e-05,
+      "loss": 1.4082,
+      "step": 4720
+    },
+    {
+      "epoch": 0.9545913218970736,
+      "grad_norm": 1.922635555267334,
+      "learning_rate": 5.785962567768243e-05,
+      "loss": 1.3854,
+      "step": 4730
+    },
+    {
+      "epoch": 0.9566094853683148,
+      "grad_norm": 1.230394959449768,
+      "learning_rate": 5.769479911354273e-05,
+      "loss": 1.4562,
+      "step": 4740
+    },
+    {
+      "epoch": 0.958627648839556,
+      "grad_norm": 1.2346444129943848,
+      "learning_rate": 5.7529886855832096e-05,
+      "loss": 1.5012,
+      "step": 4750
+    },
+    {
+      "epoch": 0.9606458123107972,
+      "grad_norm": 1.4491231441497803,
+      "learning_rate": 5.736489074110533e-05,
+      "loss": 1.2942,
+      "step": 4760
+    },
+    {
+      "epoch": 0.9626639757820383,
+      "grad_norm": 1.3211175203323364,
+      "learning_rate": 5.71998126068511e-05,
+      "loss": 1.3634,
+      "step": 4770
+    },
+    {
+      "epoch": 0.9646821392532795,
+      "grad_norm": 1.215053915977478,
+      "learning_rate": 5.7034654291471524e-05,
+      "loss": 1.434,
+      "step": 4780
+    },
+    {
+      "epoch": 0.9667003027245207,
+      "grad_norm": 1.1765618324279785,
+      "learning_rate": 5.686941763426161e-05,
+      "loss": 1.4677,
+      "step": 4790
+    },
+    {
+      "epoch": 0.9687184661957619,
+      "grad_norm": 1.3212573528289795,
+      "learning_rate": 5.670410447538889e-05,
+      "loss": 1.4113,
+      "step": 4800
+    },
+    {
+      "epoch": 0.970736629667003,
+      "grad_norm": 1.455395221710205,
+      "learning_rate": 5.653871665587278e-05,
+      "loss": 1.4146,
+      "step": 4810
+    },
+    {
+      "epoch": 0.9727547931382442,
+      "grad_norm": 1.19629967212677,
+      "learning_rate": 5.6373256017564215e-05,
+      "loss": 1.3996,
+      "step": 4820
+    },
+    {
+      "epoch": 0.9747729566094854,
+      "grad_norm": 1.6926542520523071,
+      "learning_rate": 5.620772440312508e-05,
+      "loss": 1.4043,
+      "step": 4830
+    },
+    {
+      "epoch": 0.9767911200807265,
+      "grad_norm": 1.3891587257385254,
+      "learning_rate": 5.6042123656007685e-05,
+      "loss": 1.4503,
+      "step": 4840
+    },
+    {
+      "epoch": 0.9788092835519677,
+      "grad_norm": 1.368962287902832,
+      "learning_rate": 5.587645562043422e-05,
+      "loss": 1.427,
+      "step": 4850
+    },
+    {
+      "epoch": 0.9808274470232089,
+      "grad_norm": 1.3921308517456055,
+      "learning_rate": 5.5710722141376245e-05,
+      "loss": 1.3451,
+      "step": 4860
+    },
+    {
+      "epoch": 0.9828456104944501,
+      "grad_norm": 1.1947131156921387,
+      "learning_rate": 5.5544925064534145e-05,
+      "loss": 1.3041,
+      "step": 4870
+    },
+    {
+      "epoch": 0.9848637739656912,
+      "grad_norm": 1.3686097860336304,
+      "learning_rate": 5.537906623631657e-05,
+      "loss": 1.5366,
+      "step": 4880
+    },
+    {
+      "epoch": 0.9868819374369324,
+      "grad_norm": 1.3779717683792114,
+      "learning_rate": 5.521314750381983e-05,
+      "loss": 1.3769,
+      "step": 4890
+    },
+    {
+      "epoch": 0.9889001009081736,
+      "grad_norm": 1.2698273658752441,
+      "learning_rate": 5.5047170714807406e-05,
+      "loss": 1.3719,
+      "step": 4900
+    },
+    {
+      "epoch": 0.9909182643794148,
+      "grad_norm": 1.1771740913391113,
+      "learning_rate": 5.4881137717689315e-05,
+      "loss": 1.3579,
+      "step": 4910
+    },
+    {
+      "epoch": 0.992936427850656,
+      "grad_norm": 1.2798762321472168,
+      "learning_rate": 5.471505036150154e-05,
+      "loss": 1.3889,
+      "step": 4920
+    },
+    {
+      "epoch": 0.9949545913218971,
+      "grad_norm": 1.7384541034698486,
+      "learning_rate": 5.454891049588544e-05,
+      "loss": 1.531,
+      "step": 4930
+    },
+    {
+      "epoch": 0.9969727547931383,
+      "grad_norm": 1.2217923402786255,
+      "learning_rate": 5.438271997106712e-05,
+      "loss": 1.406,
+      "step": 4940
+    },
+    {
+      "epoch": 0.9989909182643795,
+      "grad_norm": 1.1362420320510864,
+      "learning_rate": 5.421648063783689e-05,
+      "loss": 1.2914,
+      "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:f5e972b7d316d195c84f74bbc2d36e72d11d4d74af3f7e1b7c30401c068aa6a6
+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-mean/",
+  "architectures": [
+    "Olmo2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "eos_token_id": 100257,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 4096,
+  "model_type": "olmo2",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 100277,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 500000,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.48.0.dev0",
+  "use_cache": true,
+  "vocab_size": 100384
+}

checkpoint-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 sha256:54f832dd260a16e59fbe9b37038eefe30e61dd8b9e4bc3ac50a799e8a3a5a3d1
+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:d9def91ea383873d3fa8086e772cd8351b7aa2c5ca36d095f2c3a6f9a4e98652
+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 sha256:1f43d7ccef4745c85fb4856dce1956dc442d1fadce11297ba2f3958a010551b2
+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:f5e972b7d316d195c84f74bbc2d36e72d11d4d74af3f7e1b7c30401c068aa6a6
+size 5432

config.json

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

generation_config.json

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

model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54f832dd260a16e59fbe9b37038eefe30e61dd8b9e4bc3ac50a799e8a3a5a3d1
+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:d9def91ea383873d3fa8086e772cd8351b7aa2c5ca36d095f2c3a6f9a4e98652
+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:f5e972b7d316d195c84f74bbc2d36e72d11d4d74af3f7e1b7c30401c068aa6a6
+size 5432