Upload folder using huggingface_hub

checkpoint-1042/config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-ary_Arab-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-1042/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-1042/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c7999db0cf31bf98af925cb7b6f4ad8024b9e864a35d596969a73e20f455324
+size 4950070472

checkpoint-1042/model-00002-of-00006.safetensors

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

checkpoint-1042/model-00003-of-00006.safetensors

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

checkpoint-1042/model-00004-of-00006.safetensors

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

checkpoint-1042/model-00005-of-00006.safetensors

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

checkpoint-1042/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:840857203f141ce5771549f01229a9f15fa001fc49a509f633c43a6288bd3878
+size 4614509080

checkpoint-1042/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-1042/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bbde4aca1d2c5ac5400b7b3ec32c1ad7ca39e677dd23deb2f47fd5ce365d7db
+size 13055369358

checkpoint-1042/rng_state.pth

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

checkpoint-1042/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d7958675868298b58d93c029d088813924b8fdcb24ce82dad99471ebd0b58806
+size 1064

checkpoint-1042/trainer_state.json

@@ -0,0 +1,3680 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 1042,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003838771593090211,
+      "grad_norm": 48.289031982421875,
+      "learning_rate": 3.7735849056603773e-06,
+      "loss": 6.4916,
+      "step": 2
+    },
+    {
+      "epoch": 0.007677543186180422,
+      "grad_norm": 29.75079917907715,
+      "learning_rate": 7.547169811320755e-06,
+      "loss": 6.2279,
+      "step": 4
+    },
+    {
+      "epoch": 0.011516314779270634,
+      "grad_norm": 10.823013305664062,
+      "learning_rate": 1.1320754716981132e-05,
+      "loss": 5.5753,
+      "step": 6
+    },
+    {
+      "epoch": 0.015355086372360844,
+      "grad_norm": 7.690505027770996,
+      "learning_rate": 1.509433962264151e-05,
+      "loss": 5.1097,
+      "step": 8
+    },
+    {
+      "epoch": 0.019193857965451054,
+      "grad_norm": 8.298799514770508,
+      "learning_rate": 1.8867924528301888e-05,
+      "loss": 4.9181,
+      "step": 10
+    },
+    {
+      "epoch": 0.023032629558541268,
+      "grad_norm": 5.300838947296143,
+      "learning_rate": 2.2641509433962265e-05,
+      "loss": 4.881,
+      "step": 12
+    },
+    {
+      "epoch": 0.026871401151631478,
+      "grad_norm": 5.326018810272217,
+      "learning_rate": 2.641509433962264e-05,
+      "loss": 4.5184,
+      "step": 14
+    },
+    {
+      "epoch": 0.030710172744721688,
+      "grad_norm": 5.346513271331787,
+      "learning_rate": 3.018867924528302e-05,
+      "loss": 4.5163,
+      "step": 16
+    },
+    {
+      "epoch": 0.0345489443378119,
+      "grad_norm": 3.9342334270477295,
+      "learning_rate": 3.39622641509434e-05,
+      "loss": 4.3626,
+      "step": 18
+    },
+    {
+      "epoch": 0.03838771593090211,
+      "grad_norm": 3.5320026874542236,
+      "learning_rate": 3.7735849056603776e-05,
+      "loss": 4.4012,
+      "step": 20
+    },
+    {
+      "epoch": 0.04222648752399232,
+      "grad_norm": 3.396531581878662,
+      "learning_rate": 4.150943396226415e-05,
+      "loss": 4.2005,
+      "step": 22
+    },
+    {
+      "epoch": 0.046065259117082535,
+      "grad_norm": 2.8504703044891357,
+      "learning_rate": 4.528301886792453e-05,
+      "loss": 4.0846,
+      "step": 24
+    },
+    {
+      "epoch": 0.04990403071017274,
+      "grad_norm": 3.5860769748687744,
+      "learning_rate": 4.9056603773584906e-05,
+      "loss": 4.0599,
+      "step": 26
+    },
+    {
+      "epoch": 0.053742802303262956,
+      "grad_norm": 3.1936397552490234,
+      "learning_rate": 5.283018867924528e-05,
+      "loss": 4.1174,
+      "step": 28
+    },
+    {
+      "epoch": 0.05758157389635317,
+      "grad_norm": 3.2273406982421875,
+      "learning_rate": 5.660377358490566e-05,
+      "loss": 3.7389,
+      "step": 30
+    },
+    {
+      "epoch": 0.061420345489443376,
+      "grad_norm": 2.6602048873901367,
+      "learning_rate": 6.037735849056604e-05,
+      "loss": 4.0485,
+      "step": 32
+    },
+    {
+      "epoch": 0.06525911708253358,
+      "grad_norm": 3.2807395458221436,
+      "learning_rate": 6.415094339622641e-05,
+      "loss": 4.0037,
+      "step": 34
+    },
+    {
+      "epoch": 0.0690978886756238,
+      "grad_norm": 2.3215198516845703,
+      "learning_rate": 6.79245283018868e-05,
+      "loss": 3.9036,
+      "step": 36
+    },
+    {
+      "epoch": 0.07293666026871401,
+      "grad_norm": 2.9224202632904053,
+      "learning_rate": 7.169811320754717e-05,
+      "loss": 3.973,
+      "step": 38
+    },
+    {
+      "epoch": 0.07677543186180422,
+      "grad_norm": 2.602111339569092,
+      "learning_rate": 7.547169811320755e-05,
+      "loss": 3.7376,
+      "step": 40
+    },
+    {
+      "epoch": 0.08061420345489444,
+      "grad_norm": 2.60905122756958,
+      "learning_rate": 7.924528301886794e-05,
+      "loss": 3.8791,
+      "step": 42
+    },
+    {
+      "epoch": 0.08445297504798464,
+      "grad_norm": 3.377249002456665,
+      "learning_rate": 8.30188679245283e-05,
+      "loss": 3.8816,
+      "step": 44
+    },
+    {
+      "epoch": 0.08829174664107485,
+      "grad_norm": 2.953822612762451,
+      "learning_rate": 8.679245283018869e-05,
+      "loss": 3.7942,
+      "step": 46
+    },
+    {
+      "epoch": 0.09213051823416507,
+      "grad_norm": 2.96752667427063,
+      "learning_rate": 9.056603773584906e-05,
+      "loss": 3.5678,
+      "step": 48
+    },
+    {
+      "epoch": 0.09596928982725528,
+      "grad_norm": 3.927953004837036,
+      "learning_rate": 9.433962264150944e-05,
+      "loss": 3.8445,
+      "step": 50
+    },
+    {
+      "epoch": 0.09980806142034548,
+      "grad_norm": 3.235875368118286,
+      "learning_rate": 9.811320754716981e-05,
+      "loss": 3.8813,
+      "step": 52
+    },
+    {
+      "epoch": 0.1036468330134357,
+      "grad_norm": 3.226400375366211,
+      "learning_rate": 9.999974774092106e-05,
+      "loss": 3.5795,
+      "step": 54
+    },
+    {
+      "epoch": 0.10748560460652591,
+      "grad_norm": 3.028534173965454,
+      "learning_rate": 9.999772968356182e-05,
+      "loss": 3.6002,
+      "step": 56
+    },
+    {
+      "epoch": 0.11132437619961612,
+      "grad_norm": 2.740309715270996,
+      "learning_rate": 9.999369365029487e-05,
+      "loss": 3.6699,
+      "step": 58
+    },
+    {
+      "epoch": 0.11516314779270634,
+      "grad_norm": 3.1072158813476562,
+      "learning_rate": 9.998763980401998e-05,
+      "loss": 3.6008,
+      "step": 60
+    },
+    {
+      "epoch": 0.11900191938579655,
+      "grad_norm": 2.9284427165985107,
+      "learning_rate": 9.997956838907852e-05,
+      "loss": 3.6526,
+      "step": 62
+    },
+    {
+      "epoch": 0.12284069097888675,
+      "grad_norm": 3.230877161026001,
+      "learning_rate": 9.996947973124372e-05,
+      "loss": 3.7293,
+      "step": 64
+    },
+    {
+      "epoch": 0.12667946257197696,
+      "grad_norm": 3.2215468883514404,
+      "learning_rate": 9.995737423770745e-05,
+      "loss": 3.4869,
+      "step": 66
+    },
+    {
+      "epoch": 0.13051823416506717,
+      "grad_norm": 2.768615245819092,
+      "learning_rate": 9.994325239706377e-05,
+      "loss": 3.513,
+      "step": 68
+    },
+    {
+      "epoch": 0.1343570057581574,
+      "grad_norm": 2.9997639656066895,
+      "learning_rate": 9.992711477928926e-05,
+      "loss": 3.4237,
+      "step": 70
+    },
+    {
+      "epoch": 0.1381957773512476,
+      "grad_norm": 3.5303409099578857,
+      "learning_rate": 9.990896203571995e-05,
+      "loss": 3.6657,
+      "step": 72
+    },
+    {
+      "epoch": 0.1420345489443378,
+      "grad_norm": 2.925368309020996,
+      "learning_rate": 9.988879489902509e-05,
+      "loss": 3.4966,
+      "step": 74
+    },
+    {
+      "epoch": 0.14587332053742802,
+      "grad_norm": 3.1105096340179443,
+      "learning_rate": 9.986661418317759e-05,
+      "loss": 3.5241,
+      "step": 76
+    },
+    {
+      "epoch": 0.14971209213051823,
+      "grad_norm": 3.1198081970214844,
+      "learning_rate": 9.984242078342107e-05,
+      "loss": 3.6021,
+      "step": 78
+    },
+    {
+      "epoch": 0.15355086372360843,
+      "grad_norm": 2.56496000289917,
+      "learning_rate": 9.981621567623385e-05,
+      "loss": 3.4452,
+      "step": 80
+    },
+    {
+      "epoch": 0.15738963531669867,
+      "grad_norm": 2.585533380508423,
+      "learning_rate": 9.978799991928944e-05,
+      "loss": 3.5144,
+      "step": 82
+    },
+    {
+      "epoch": 0.16122840690978887,
+      "grad_norm": 2.6131768226623535,
+      "learning_rate": 9.97577746514139e-05,
+      "loss": 3.49,
+      "step": 84
+    },
+    {
+      "epoch": 0.16506717850287908,
+      "grad_norm": 3.118385076522827,
+      "learning_rate": 9.972554109253989e-05,
+      "loss": 3.5082,
+      "step": 86
+    },
+    {
+      "epoch": 0.1689059500959693,
+      "grad_norm": 2.592630624771118,
+      "learning_rate": 9.969130054365736e-05,
+      "loss": 3.3943,
+      "step": 88
+    },
+    {
+      "epoch": 0.1727447216890595,
+      "grad_norm": 3.1140449047088623,
+      "learning_rate": 9.965505438676115e-05,
+      "loss": 3.4553,
+      "step": 90
+    },
+    {
+      "epoch": 0.1765834932821497,
+      "grad_norm": 2.3941497802734375,
+      "learning_rate": 9.961680408479507e-05,
+      "loss": 3.2766,
+      "step": 92
+    },
+    {
+      "epoch": 0.18042226487523993,
+      "grad_norm": 2.8205137252807617,
+      "learning_rate": 9.957655118159304e-05,
+      "loss": 3.3713,
+      "step": 94
+    },
+    {
+      "epoch": 0.18426103646833014,
+      "grad_norm": 2.3888957500457764,
+      "learning_rate": 9.953429730181653e-05,
+      "loss": 3.3228,
+      "step": 96
+    },
+    {
+      "epoch": 0.18809980806142035,
+      "grad_norm": 2.363973617553711,
+      "learning_rate": 9.949004415088927e-05,
+      "loss": 3.3273,
+      "step": 98
+    },
+    {
+      "epoch": 0.19193857965451055,
+      "grad_norm": 2.508699655532837,
+      "learning_rate": 9.944379351492818e-05,
+      "loss": 3.2716,
+      "step": 100
+    },
+    {
+      "epoch": 0.19577735124760076,
+      "grad_norm": 2.3598968982696533,
+      "learning_rate": 9.939554726067141e-05,
+      "loss": 3.4032,
+      "step": 102
+    },
+    {
+      "epoch": 0.19961612284069097,
+      "grad_norm": 2.8131492137908936,
+      "learning_rate": 9.934530733540294e-05,
+      "loss": 3.3651,
+      "step": 104
+    },
+    {
+      "epoch": 0.2034548944337812,
+      "grad_norm": 2.7955825328826904,
+      "learning_rate": 9.929307576687403e-05,
+      "loss": 3.3716,
+      "step": 106
+    },
+    {
+      "epoch": 0.2072936660268714,
+      "grad_norm": 2.7270803451538086,
+      "learning_rate": 9.923885466322135e-05,
+      "loss": 3.451,
+      "step": 108
+    },
+    {
+      "epoch": 0.21113243761996162,
+      "grad_norm": 3.13055419921875,
+      "learning_rate": 9.918264621288187e-05,
+      "loss": 3.3958,
+      "step": 110
+    },
+    {
+      "epoch": 0.21497120921305182,
+      "grad_norm": 2.994201898574829,
+      "learning_rate": 9.912445268450458e-05,
+      "loss": 3.3517,
+      "step": 112
+    },
+    {
+      "epoch": 0.21880998080614203,
+      "grad_norm": 2.897017002105713,
+      "learning_rate": 9.906427642685889e-05,
+      "loss": 3.3386,
+      "step": 114
+    },
+    {
+      "epoch": 0.22264875239923224,
+      "grad_norm": 2.668201208114624,
+      "learning_rate": 9.900211986873985e-05,
+      "loss": 3.3498,
+      "step": 116
+    },
+    {
+      "epoch": 0.22648752399232247,
+      "grad_norm": 2.353947877883911,
+      "learning_rate": 9.89379855188701e-05,
+      "loss": 3.2568,
+      "step": 118
+    },
+    {
+      "epoch": 0.23032629558541268,
+      "grad_norm": 2.7698841094970703,
+      "learning_rate": 9.887187596579866e-05,
+      "loss": 3.2583,
+      "step": 120
+    },
+    {
+      "epoch": 0.23416506717850288,
+      "grad_norm": 2.742597818374634,
+      "learning_rate": 9.880379387779637e-05,
+      "loss": 3.1913,
+      "step": 122
+    },
+    {
+      "epoch": 0.2380038387715931,
+      "grad_norm": 2.6558752059936523,
+      "learning_rate": 9.873374200274826e-05,
+      "loss": 3.262,
+      "step": 124
+    },
+    {
+      "epoch": 0.2418426103646833,
+      "grad_norm": 2.544968843460083,
+      "learning_rate": 9.866172316804266e-05,
+      "loss": 3.3072,
+      "step": 126
+    },
+    {
+      "epoch": 0.2456813819577735,
+      "grad_norm": 2.6007907390594482,
+      "learning_rate": 9.8587740280457e-05,
+      "loss": 3.135,
+      "step": 128
+    },
+    {
+      "epoch": 0.2495201535508637,
+      "grad_norm": 2.7021710872650146,
+      "learning_rate": 9.851179632604057e-05,
+      "loss": 3.2714,
+      "step": 130
+    },
+    {
+      "epoch": 0.2533589251439539,
+      "grad_norm": 2.698706865310669,
+      "learning_rate": 9.843389436999396e-05,
+      "loss": 3.2385,
+      "step": 132
+    },
+    {
+      "epoch": 0.2571976967370441,
+      "grad_norm": 2.9197003841400146,
+      "learning_rate": 9.835403755654536e-05,
+      "loss": 3.2911,
+      "step": 134
+    },
+    {
+      "epoch": 0.26103646833013433,
+      "grad_norm": 2.5004656314849854,
+      "learning_rate": 9.827222910882358e-05,
+      "loss": 3.3533,
+      "step": 136
+    },
+    {
+      "epoch": 0.2648752399232246,
+      "grad_norm": 2.7729904651641846,
+      "learning_rate": 9.818847232872815e-05,
+      "loss": 3.2461,
+      "step": 138
+    },
+    {
+      "epoch": 0.2687140115163148,
+      "grad_norm": 2.4783594608306885,
+      "learning_rate": 9.81027705967958e-05,
+      "loss": 3.2066,
+      "step": 140
+    },
+    {
+      "epoch": 0.272552783109405,
+      "grad_norm": 2.5870554447174072,
+      "learning_rate": 9.801512737206422e-05,
+      "loss": 3.3161,
+      "step": 142
+    },
+    {
+      "epoch": 0.2763915547024952,
+      "grad_norm": 2.610823154449463,
+      "learning_rate": 9.792554619193235e-05,
+      "loss": 3.1034,
+      "step": 144
+    },
+    {
+      "epoch": 0.2802303262955854,
+      "grad_norm": 2.5675435066223145,
+      "learning_rate": 9.783403067201762e-05,
+      "loss": 3.0721,
+      "step": 146
+    },
+    {
+      "epoch": 0.2840690978886756,
+      "grad_norm": 2.435715913772583,
+      "learning_rate": 9.774058450601002e-05,
+      "loss": 3.2662,
+      "step": 148
+    },
+    {
+      "epoch": 0.28790786948176583,
+      "grad_norm": 2.705893039703369,
+      "learning_rate": 9.764521146552309e-05,
+      "loss": 3.2292,
+      "step": 150
+    },
+    {
+      "epoch": 0.29174664107485604,
+      "grad_norm": 2.7473392486572266,
+      "learning_rate": 9.754791539994152e-05,
+      "loss": 3.1772,
+      "step": 152
+    },
+    {
+      "epoch": 0.29558541266794625,
+      "grad_norm": 2.6607396602630615,
+      "learning_rate": 9.744870023626599e-05,
+      "loss": 3.1006,
+      "step": 154
+    },
+    {
+      "epoch": 0.29942418426103645,
+      "grad_norm": 2.3230140209198,
+      "learning_rate": 9.734756997895448e-05,
+      "loss": 3.1766,
+      "step": 156
+    },
+    {
+      "epoch": 0.30326295585412666,
+      "grad_norm": 2.8475935459136963,
+      "learning_rate": 9.724452870976085e-05,
+      "loss": 3.2234,
+      "step": 158
+    },
+    {
+      "epoch": 0.30710172744721687,
+      "grad_norm": 2.8263561725616455,
+      "learning_rate": 9.713958058756985e-05,
+      "loss": 2.9765,
+      "step": 160
+    },
+    {
+      "epoch": 0.31094049904030713,
+      "grad_norm": 2.7287673950195312,
+      "learning_rate": 9.703272984822946e-05,
+      "loss": 3.2325,
+      "step": 162
+    },
+    {
+      "epoch": 0.31477927063339733,
+      "grad_norm": 2.7081940174102783,
+      "learning_rate": 9.69239808043799e-05,
+      "loss": 3.1034,
+      "step": 164
+    },
+    {
+      "epoch": 0.31861804222648754,
+      "grad_norm": 2.571908473968506,
+      "learning_rate": 9.681333784527944e-05,
+      "loss": 3.1813,
+      "step": 166
+    },
+    {
+      "epoch": 0.32245681381957775,
+      "grad_norm": 2.3741984367370605,
+      "learning_rate": 9.67008054366274e-05,
+      "loss": 3.1286,
+      "step": 168
+    },
+    {
+      "epoch": 0.32629558541266795,
+      "grad_norm": 2.4630539417266846,
+      "learning_rate": 9.658638812038379e-05,
+      "loss": 3.2526,
+      "step": 170
+    },
+    {
+      "epoch": 0.33013435700575816,
+      "grad_norm": 2.2894673347473145,
+      "learning_rate": 9.647009051458604e-05,
+      "loss": 3.0942,
+      "step": 172
+    },
+    {
+      "epoch": 0.33397312859884837,
+      "grad_norm": 2.4378833770751953,
+      "learning_rate": 9.635191731316262e-05,
+      "loss": 3.0711,
+      "step": 174
+    },
+    {
+      "epoch": 0.3378119001919386,
+      "grad_norm": 2.3150041103363037,
+      "learning_rate": 9.623187328574357e-05,
+      "loss": 3.1246,
+      "step": 176
+    },
+    {
+      "epoch": 0.3416506717850288,
+      "grad_norm": 2.4483959674835205,
+      "learning_rate": 9.6109963277468e-05,
+      "loss": 3.1125,
+      "step": 178
+    },
+    {
+      "epoch": 0.345489443378119,
+      "grad_norm": 2.5114800930023193,
+      "learning_rate": 9.598619220878853e-05,
+      "loss": 3.0962,
+      "step": 180
+    },
+    {
+      "epoch": 0.3493282149712092,
+      "grad_norm": 2.2731025218963623,
+      "learning_rate": 9.586056507527266e-05,
+      "loss": 3.0675,
+      "step": 182
+    },
+    {
+      "epoch": 0.3531669865642994,
+      "grad_norm": 2.7290143966674805,
+      "learning_rate": 9.573308694740121e-05,
+      "loss": 3.1265,
+      "step": 184
+    },
+    {
+      "epoch": 0.3570057581573896,
+      "grad_norm": 2.613831043243408,
+      "learning_rate": 9.560376297036362e-05,
+      "loss": 3.1269,
+      "step": 186
+    },
+    {
+      "epoch": 0.36084452975047987,
+      "grad_norm": 2.2489418983459473,
+      "learning_rate": 9.547259836385031e-05,
+      "loss": 3.0518,
+      "step": 188
+    },
+    {
+      "epoch": 0.3646833013435701,
+      "grad_norm": 2.547145366668701,
+      "learning_rate": 9.533959842184195e-05,
+      "loss": 3.0586,
+      "step": 190
+    },
+    {
+      "epoch": 0.3685220729366603,
+      "grad_norm": 2.385820150375366,
+      "learning_rate": 9.520476851239588e-05,
+      "loss": 3.182,
+      "step": 192
+    },
+    {
+      "epoch": 0.3723608445297505,
+      "grad_norm": 2.3586604595184326,
+      "learning_rate": 9.506811407742937e-05,
+      "loss": 3.1526,
+      "step": 194
+    },
+    {
+      "epoch": 0.3761996161228407,
+      "grad_norm": 2.4252877235412598,
+      "learning_rate": 9.492964063250001e-05,
+      "loss": 3.0067,
+      "step": 196
+    },
+    {
+      "epoch": 0.3800383877159309,
+      "grad_norm": 2.4425666332244873,
+      "learning_rate": 9.478935376658307e-05,
+      "loss": 3.0842,
+      "step": 198
+    },
+    {
+      "epoch": 0.3838771593090211,
+      "grad_norm": 2.5915610790252686,
+      "learning_rate": 9.4647259141846e-05,
+      "loss": 3.0993,
+      "step": 200
+    },
+    {
+      "epoch": 0.3877159309021113,
+      "grad_norm": 2.3467631340026855,
+      "learning_rate": 9.450336249341976e-05,
+      "loss": 3.1839,
+      "step": 202
+    },
+    {
+      "epoch": 0.3915547024952015,
+      "grad_norm": 2.370368719100952,
+      "learning_rate": 9.435766962916748e-05,
+      "loss": 3.1194,
+      "step": 204
+    },
+    {
+      "epoch": 0.39539347408829173,
+      "grad_norm": 2.5787832736968994,
+      "learning_rate": 9.421018642944997e-05,
+      "loss": 3.1111,
+      "step": 206
+    },
+    {
+      "epoch": 0.39923224568138194,
+      "grad_norm": 2.549276351928711,
+      "learning_rate": 9.406091884688836e-05,
+      "loss": 3.0373,
+      "step": 208
+    },
+    {
+      "epoch": 0.40307101727447214,
+      "grad_norm": 2.4557924270629883,
+      "learning_rate": 9.390987290612395e-05,
+      "loss": 3.0727,
+      "step": 210
+    },
+    {
+      "epoch": 0.4069097888675624,
+      "grad_norm": 2.4945881366729736,
+      "learning_rate": 9.375705470357493e-05,
+      "loss": 3.0119,
+      "step": 212
+    },
+    {
+      "epoch": 0.4107485604606526,
+      "grad_norm": 2.5190112590789795,
+      "learning_rate": 9.360247040719039e-05,
+      "loss": 3.0296,
+      "step": 214
+    },
+    {
+      "epoch": 0.4145873320537428,
+      "grad_norm": 2.3181045055389404,
+      "learning_rate": 9.344612625620132e-05,
+      "loss": 3.0463,
+      "step": 216
+    },
+    {
+      "epoch": 0.418426103646833,
+      "grad_norm": 2.344257116317749,
+      "learning_rate": 9.32880285608689e-05,
+      "loss": 3.0091,
+      "step": 218
+    },
+    {
+      "epoch": 0.42226487523992323,
+      "grad_norm": 2.0085527896881104,
+      "learning_rate": 9.312818370222962e-05,
+      "loss": 2.9649,
+      "step": 220
+    },
+    {
+      "epoch": 0.42610364683301344,
+      "grad_norm": 2.92118763923645,
+      "learning_rate": 9.296659813183794e-05,
+      "loss": 3.0637,
+      "step": 222
+    },
+    {
+      "epoch": 0.42994241842610365,
+      "grad_norm": 2.6241536140441895,
+      "learning_rate": 9.280327837150571e-05,
+      "loss": 3.0458,
+      "step": 224
+    },
+    {
+      "epoch": 0.43378119001919385,
+      "grad_norm": 2.025529146194458,
+      "learning_rate": 9.26382310130391e-05,
+      "loss": 3.0004,
+      "step": 226
+    },
+    {
+      "epoch": 0.43761996161228406,
+      "grad_norm": 2.682751178741455,
+      "learning_rate": 9.247146271797245e-05,
+      "loss": 3.0602,
+      "step": 228
+    },
+    {
+      "epoch": 0.44145873320537427,
+      "grad_norm": 2.2250850200653076,
+      "learning_rate": 9.23029802172994e-05,
+      "loss": 2.9436,
+      "step": 230
+    },
+    {
+      "epoch": 0.44529750479846447,
+      "grad_norm": 2.490293264389038,
+      "learning_rate": 9.213279031120129e-05,
+      "loss": 2.7951,
+      "step": 232
+    },
+    {
+      "epoch": 0.4491362763915547,
+      "grad_norm": 2.5657706260681152,
+      "learning_rate": 9.19608998687726e-05,
+      "loss": 3.1099,
+      "step": 234
+    },
+    {
+      "epoch": 0.45297504798464494,
+      "grad_norm": 2.199953079223633,
+      "learning_rate": 9.178731582774381e-05,
+      "loss": 2.9906,
+      "step": 236
+    },
+    {
+      "epoch": 0.45681381957773515,
+      "grad_norm": 2.581700086593628,
+      "learning_rate": 9.161204519420126e-05,
+      "loss": 2.9742,
+      "step": 238
+    },
+    {
+      "epoch": 0.46065259117082535,
+      "grad_norm": 2.086547374725342,
+      "learning_rate": 9.143509504230449e-05,
+      "loss": 3.0655,
+      "step": 240
+    },
+    {
+      "epoch": 0.46449136276391556,
+      "grad_norm": 2.3075525760650635,
+      "learning_rate": 9.125647251400067e-05,
+      "loss": 2.9537,
+      "step": 242
+    },
+    {
+      "epoch": 0.46833013435700577,
+      "grad_norm": 2.2352142333984375,
+      "learning_rate": 9.107618481873632e-05,
+      "loss": 2.9135,
+      "step": 244
+    },
+    {
+      "epoch": 0.472168905950096,
+      "grad_norm": 2.1639482975006104,
+      "learning_rate": 9.089423923316636e-05,
+      "loss": 3.0237,
+      "step": 246
+    },
+    {
+      "epoch": 0.4760076775431862,
+      "grad_norm": 2.307377576828003,
+      "learning_rate": 9.07106431008604e-05,
+      "loss": 2.9091,
+      "step": 248
+    },
+    {
+      "epoch": 0.4798464491362764,
+      "grad_norm": 2.010719060897827,
+      "learning_rate": 9.052540383200633e-05,
+      "loss": 2.8782,
+      "step": 250
+    },
+    {
+      "epoch": 0.4836852207293666,
+      "grad_norm": 2.6130549907684326,
+      "learning_rate": 9.033852890311128e-05,
+      "loss": 3.0429,
+      "step": 252
+    },
+    {
+      "epoch": 0.4875239923224568,
+      "grad_norm": 2.3005809783935547,
+      "learning_rate": 9.01500258566998e-05,
+      "loss": 2.9648,
+      "step": 254
+    },
+    {
+      "epoch": 0.491362763915547,
+      "grad_norm": 2.362955331802368,
+      "learning_rate": 8.995990230100949e-05,
+      "loss": 2.908,
+      "step": 256
+    },
+    {
+      "epoch": 0.4952015355086372,
+      "grad_norm": 2.1333205699920654,
+      "learning_rate": 8.976816590968387e-05,
+      "loss": 3.0275,
+      "step": 258
+    },
+    {
+      "epoch": 0.4990403071017274,
+      "grad_norm": 2.328183650970459,
+      "learning_rate": 8.957482442146272e-05,
+      "loss": 2.9645,
+      "step": 260
+    },
+    {
+      "epoch": 0.5028790786948176,
+      "grad_norm": 2.229396343231201,
+      "learning_rate": 8.937988563986963e-05,
+      "loss": 2.8474,
+      "step": 262
+    },
+    {
+      "epoch": 0.5067178502879078,
+      "grad_norm": 2.283524513244629,
+      "learning_rate": 8.918335743289718e-05,
+      "loss": 3.0276,
+      "step": 264
+    },
+    {
+      "epoch": 0.510556621880998,
+      "grad_norm": 2.0872859954833984,
+      "learning_rate": 8.898524773268926e-05,
+      "loss": 3.0058,
+      "step": 266
+    },
+    {
+      "epoch": 0.5143953934740882,
+      "grad_norm": 2.427462100982666,
+      "learning_rate": 8.8785564535221e-05,
+      "loss": 2.971,
+      "step": 268
+    },
+    {
+      "epoch": 0.5182341650671785,
+      "grad_norm": 2.280747413635254,
+      "learning_rate": 8.858431589997597e-05,
+      "loss": 2.9367,
+      "step": 270
+    },
+    {
+      "epoch": 0.5220729366602687,
+      "grad_norm": 2.2372734546661377,
+      "learning_rate": 8.838150994962093e-05,
+      "loss": 3.0049,
+      "step": 272
+    },
+    {
+      "epoch": 0.525911708253359,
+      "grad_norm": 2.379145622253418,
+      "learning_rate": 8.817715486967803e-05,
+      "loss": 3.0165,
+      "step": 274
+    },
+    {
+      "epoch": 0.5297504798464492,
+      "grad_norm": 2.5526132583618164,
+      "learning_rate": 8.797125890819428e-05,
+      "loss": 2.9192,
+      "step": 276
+    },
+    {
+      "epoch": 0.5335892514395394,
+      "grad_norm": 1.9904067516326904,
+      "learning_rate": 8.776383037540888e-05,
+      "loss": 2.9803,
+      "step": 278
+    },
+    {
+      "epoch": 0.5374280230326296,
+      "grad_norm": 1.8862968683242798,
+      "learning_rate": 8.755487764341757e-05,
+      "loss": 2.8934,
+      "step": 280
+    },
+    {
+      "epoch": 0.5412667946257198,
+      "grad_norm": 2.738217830657959,
+      "learning_rate": 8.734440914583486e-05,
+      "loss": 2.9304,
+      "step": 282
+    },
+    {
+      "epoch": 0.54510556621881,
+      "grad_norm": 2.059157609939575,
+      "learning_rate": 8.713243337745365e-05,
+      "loss": 2.9255,
+      "step": 284
+    },
+    {
+      "epoch": 0.5489443378119002,
+      "grad_norm": 2.4487457275390625,
+      "learning_rate": 8.691895889390227e-05,
+      "loss": 2.906,
+      "step": 286
+    },
+    {
+      "epoch": 0.5527831094049904,
+      "grad_norm": 1.9257361888885498,
+      "learning_rate": 8.670399431129926e-05,
+      "loss": 2.7553,
+      "step": 288
+    },
+    {
+      "epoch": 0.5566218809980806,
+      "grad_norm": 2.365708589553833,
+      "learning_rate": 8.648754830590552e-05,
+      "loss": 3.0451,
+      "step": 290
+    },
+    {
+      "epoch": 0.5604606525911708,
+      "grad_norm": 2.44295072555542,
+      "learning_rate": 8.626962961377423e-05,
+      "loss": 2.9814,
+      "step": 292
+    },
+    {
+      "epoch": 0.564299424184261,
+      "grad_norm": 2.14540433883667,
+      "learning_rate": 8.605024703039816e-05,
+      "loss": 2.9126,
+      "step": 294
+    },
+    {
+      "epoch": 0.5681381957773513,
+      "grad_norm": 2.1933960914611816,
+      "learning_rate": 8.582940941035476e-05,
+      "loss": 2.9439,
+      "step": 296
+    },
+    {
+      "epoch": 0.5719769673704415,
+      "grad_norm": 2.231813907623291,
+      "learning_rate": 8.560712566694862e-05,
+      "loss": 3.0361,
+      "step": 298
+    },
+    {
+      "epoch": 0.5758157389635317,
+      "grad_norm": 2.183107852935791,
+      "learning_rate": 8.538340477185192e-05,
+      "loss": 2.9277,
+      "step": 300
+    },
+    {
+      "epoch": 0.5796545105566219,
+      "grad_norm": 2.7278597354888916,
+      "learning_rate": 8.515825575474221e-05,
+      "loss": 2.9476,
+      "step": 302
+    },
+    {
+      "epoch": 0.5834932821497121,
+      "grad_norm": 2.6327316761016846,
+      "learning_rate": 8.493168770293792e-05,
+      "loss": 2.961,
+      "step": 304
+    },
+    {
+      "epoch": 0.5873320537428023,
+      "grad_norm": 2.3474512100219727,
+      "learning_rate": 8.47037097610317e-05,
+      "loss": 2.9233,
+      "step": 306
+    },
+    {
+      "epoch": 0.5911708253358925,
+      "grad_norm": 2.31101131439209,
+      "learning_rate": 8.447433113052123e-05,
+      "loss": 2.9289,
+      "step": 308
+    },
+    {
+      "epoch": 0.5950095969289827,
+      "grad_norm": 2.4171314239501953,
+      "learning_rate": 8.42435610694379e-05,
+      "loss": 2.9037,
+      "step": 310
+    },
+    {
+      "epoch": 0.5988483685220729,
+      "grad_norm": 2.2114620208740234,
+      "learning_rate": 8.401140889197305e-05,
+      "loss": 2.8163,
+      "step": 312
+    },
+    {
+      "epoch": 0.6026871401151631,
+      "grad_norm": 2.3434977531433105,
+      "learning_rate": 8.377788396810223e-05,
+      "loss": 2.8856,
+      "step": 314
+    },
+    {
+      "epoch": 0.6065259117082533,
+      "grad_norm": 2.4382877349853516,
+      "learning_rate": 8.354299572320679e-05,
+      "loss": 2.8265,
+      "step": 316
+    },
+    {
+      "epoch": 0.6103646833013435,
+      "grad_norm": 1.9080297946929932,
+      "learning_rate": 8.330675363769356e-05,
+      "loss": 2.9292,
+      "step": 318
+    },
+    {
+      "epoch": 0.6142034548944337,
+      "grad_norm": 2.439094066619873,
+      "learning_rate": 8.306916724661224e-05,
+      "loss": 2.8653,
+      "step": 320
+    },
+    {
+      "epoch": 0.6180422264875239,
+      "grad_norm": 2.5351178646087646,
+      "learning_rate": 8.283024613927054e-05,
+      "loss": 2.8837,
+      "step": 322
+    },
+    {
+      "epoch": 0.6218809980806143,
+      "grad_norm": 2.589860439300537,
+      "learning_rate": 8.258999995884705e-05,
+      "loss": 2.8083,
+      "step": 324
+    },
+    {
+      "epoch": 0.6257197696737045,
+      "grad_norm": 2.439771890640259,
+      "learning_rate": 8.234843840200217e-05,
+      "loss": 2.8731,
+      "step": 326
+    },
+    {
+      "epoch": 0.6295585412667947,
+      "grad_norm": 2.1145169734954834,
+      "learning_rate": 8.210557121848663e-05,
+      "loss": 2.8095,
+      "step": 328
+    },
+    {
+      "epoch": 0.6333973128598849,
+      "grad_norm": 2.5235095024108887,
+      "learning_rate": 8.186140821074801e-05,
+      "loss": 3.0236,
+      "step": 330
+    },
+    {
+      "epoch": 0.6372360844529751,
+      "grad_norm": 2.062345027923584,
+      "learning_rate": 8.161595923353516e-05,
+      "loss": 2.8697,
+      "step": 332
+    },
+    {
+      "epoch": 0.6410748560460653,
+      "grad_norm": 2.3722496032714844,
+      "learning_rate": 8.136923419350031e-05,
+      "loss": 2.8812,
+      "step": 334
+    },
+    {
+      "epoch": 0.6449136276391555,
+      "grad_norm": 2.054664134979248,
+      "learning_rate": 8.112124304879937e-05,
+      "loss": 2.8569,
+      "step": 336
+    },
+    {
+      "epoch": 0.6487523992322457,
+      "grad_norm": 2.16813325881958,
+      "learning_rate": 8.087199580868996e-05,
+      "loss": 2.874,
+      "step": 338
+    },
+    {
+      "epoch": 0.6525911708253359,
+      "grad_norm": 2.463815689086914,
+      "learning_rate": 8.062150253312735e-05,
+      "loss": 2.9572,
+      "step": 340
+    },
+    {
+      "epoch": 0.6564299424184261,
+      "grad_norm": 2.1177213191986084,
+      "learning_rate": 8.03697733323585e-05,
+      "loss": 2.9028,
+      "step": 342
+    },
+    {
+      "epoch": 0.6602687140115163,
+      "grad_norm": 2.3246190547943115,
+      "learning_rate": 8.011681836651402e-05,
+      "loss": 2.8694,
+      "step": 344
+    },
+    {
+      "epoch": 0.6641074856046065,
+      "grad_norm": 2.157296895980835,
+      "learning_rate": 7.986264784519801e-05,
+      "loss": 2.8129,
+      "step": 346
+    },
+    {
+      "epoch": 0.6679462571976967,
+      "grad_norm": 2.1130011081695557,
+      "learning_rate": 7.960727202707605e-05,
+      "loss": 2.8877,
+      "step": 348
+    },
+    {
+      "epoch": 0.6717850287907869,
+      "grad_norm": 2.256319046020508,
+      "learning_rate": 7.935070121946115e-05,
+      "loss": 2.8947,
+      "step": 350
+    },
+    {
+      "epoch": 0.6756238003838771,
+      "grad_norm": 2.183379650115967,
+      "learning_rate": 7.909294577789766e-05,
+      "loss": 2.8678,
+      "step": 352
+    },
+    {
+      "epoch": 0.6794625719769674,
+      "grad_norm": 2.4424221515655518,
+      "learning_rate": 7.883401610574336e-05,
+      "loss": 2.9269,
+      "step": 354
+    },
+    {
+      "epoch": 0.6833013435700576,
+      "grad_norm": 1.933434247970581,
+      "learning_rate": 7.857392265374963e-05,
+      "loss": 2.8906,
+      "step": 356
+    },
+    {
+      "epoch": 0.6871401151631478,
+      "grad_norm": 2.322988271713257,
+      "learning_rate": 7.83126759196395e-05,
+      "loss": 2.8756,
+      "step": 358
+    },
+    {
+      "epoch": 0.690978886756238,
+      "grad_norm": 2.2925362586975098,
+      "learning_rate": 7.805028644768407e-05,
+      "loss": 2.8704,
+      "step": 360
+    },
+    {
+      "epoch": 0.6948176583493282,
+      "grad_norm": 2.209590196609497,
+      "learning_rate": 7.778676482827686e-05,
+      "loss": 2.8794,
+      "step": 362
+    },
+    {
+      "epoch": 0.6986564299424184,
+      "grad_norm": 1.9942725896835327,
+      "learning_rate": 7.752212169750641e-05,
+      "loss": 2.8943,
+      "step": 364
+    },
+    {
+      "epoch": 0.7024952015355086,
+      "grad_norm": 2.1021668910980225,
+      "learning_rate": 7.725636773672693e-05,
+      "loss": 2.8838,
+      "step": 366
+    },
+    {
+      "epoch": 0.7063339731285988,
+      "grad_norm": 2.3661866188049316,
+      "learning_rate": 7.69895136721273e-05,
+      "loss": 2.7449,
+      "step": 368
+    },
+    {
+      "epoch": 0.710172744721689,
+      "grad_norm": 2.2460644245147705,
+      "learning_rate": 7.672157027429803e-05,
+      "loss": 2.8648,
+      "step": 370
+    },
+    {
+      "epoch": 0.7140115163147792,
+      "grad_norm": 2.0501885414123535,
+      "learning_rate": 7.645254835779657e-05,
+      "loss": 2.9297,
+      "step": 372
+    },
+    {
+      "epoch": 0.7178502879078695,
+      "grad_norm": 2.259514570236206,
+      "learning_rate": 7.618245878071091e-05,
+      "loss": 2.9449,
+      "step": 374
+    },
+    {
+      "epoch": 0.7216890595009597,
+      "grad_norm": 2.193086862564087,
+      "learning_rate": 7.591131244422118e-05,
+      "loss": 2.8749,
+      "step": 376
+    },
+    {
+      "epoch": 0.72552783109405,
+      "grad_norm": 2.0527806282043457,
+      "learning_rate": 7.563912029215983e-05,
+      "loss": 2.7844,
+      "step": 378
+    },
+    {
+      "epoch": 0.7293666026871402,
+      "grad_norm": 1.9866085052490234,
+      "learning_rate": 7.536589331056976e-05,
+      "loss": 2.9124,
+      "step": 380
+    },
+    {
+      "epoch": 0.7332053742802304,
+      "grad_norm": 2.272838830947876,
+      "learning_rate": 7.509164252726107e-05,
+      "loss": 2.9241,
+      "step": 382
+    },
+    {
+      "epoch": 0.7370441458733206,
+      "grad_norm": 2.093259334564209,
+      "learning_rate": 7.481637901136577e-05,
+      "loss": 2.8545,
+      "step": 384
+    },
+    {
+      "epoch": 0.7408829174664108,
+      "grad_norm": 1.9920848608016968,
+      "learning_rate": 7.454011387289127e-05,
+      "loss": 2.8148,
+      "step": 386
+    },
+    {
+      "epoch": 0.744721689059501,
+      "grad_norm": 7.795645236968994,
+      "learning_rate": 7.42628582622717e-05,
+      "loss": 2.7107,
+      "step": 388
+    },
+    {
+      "epoch": 0.7485604606525912,
+      "grad_norm": 2.090559482574463,
+      "learning_rate": 7.398462336991801e-05,
+      "loss": 2.7163,
+      "step": 390
+    },
+    {
+      "epoch": 0.7523992322456814,
+      "grad_norm": 2.06302809715271,
+      "learning_rate": 7.370542042576634e-05,
+      "loss": 2.7883,
+      "step": 392
+    },
+    {
+      "epoch": 0.7562380038387716,
+      "grad_norm": 2.325312376022339,
+      "learning_rate": 7.342526069882465e-05,
+      "loss": 2.9318,
+      "step": 394
+    },
+    {
+      "epoch": 0.7600767754318618,
+      "grad_norm": 2.163017988204956,
+      "learning_rate": 7.314415549671795e-05,
+      "loss": 2.877,
+      "step": 396
+    },
+    {
+      "epoch": 0.763915547024952,
+      "grad_norm": 2.363990545272827,
+      "learning_rate": 7.286211616523193e-05,
+      "loss": 2.9007,
+      "step": 398
+    },
+    {
+      "epoch": 0.7677543186180422,
+      "grad_norm": 2.189789295196533,
+      "learning_rate": 7.257915408785498e-05,
+      "loss": 2.9451,
+      "step": 400
+    },
+    {
+      "epoch": 0.7715930902111324,
+      "grad_norm": 2.284350872039795,
+      "learning_rate": 7.229528068531882e-05,
+      "loss": 2.8626,
+      "step": 402
+    },
+    {
+      "epoch": 0.7754318618042226,
+      "grad_norm": 2.0571820735931396,
+      "learning_rate": 7.201050741513736e-05,
+      "loss": 2.6964,
+      "step": 404
+    },
+    {
+      "epoch": 0.7792706333973128,
+      "grad_norm": 1.95001220703125,
+      "learning_rate": 7.17248457711445e-05,
+      "loss": 2.8581,
+      "step": 406
+    },
+    {
+      "epoch": 0.783109404990403,
+      "grad_norm": 1.9939192533493042,
+      "learning_rate": 7.14383072830301e-05,
+      "loss": 2.8433,
+      "step": 408
+    },
+    {
+      "epoch": 0.7869481765834933,
+      "grad_norm": 2.1089251041412354,
+      "learning_rate": 7.115090351587454e-05,
+      "loss": 2.8486,
+      "step": 410
+    },
+    {
+      "epoch": 0.7907869481765835,
+      "grad_norm": 2.033663749694824,
+      "learning_rate": 7.086264606968214e-05,
+      "loss": 2.828,
+      "step": 412
+    },
+    {
+      "epoch": 0.7946257197696737,
+      "grad_norm": 2.1510426998138428,
+      "learning_rate": 7.05735465789128e-05,
+      "loss": 2.9666,
+      "step": 414
+    },
+    {
+      "epoch": 0.7984644913627639,
+      "grad_norm": 2.126054048538208,
+      "learning_rate": 7.028361671201245e-05,
+      "loss": 2.8362,
+      "step": 416
+    },
+    {
+      "epoch": 0.8023032629558541,
+      "grad_norm": 2.3474514484405518,
+      "learning_rate": 6.99928681709422e-05,
+      "loss": 2.7777,
+      "step": 418
+    },
+    {
+      "epoch": 0.8061420345489443,
+      "grad_norm": 2.0623159408569336,
+      "learning_rate": 6.970131269070592e-05,
+      "loss": 2.9225,
+      "step": 420
+    },
+    {
+      "epoch": 0.8099808061420346,
+      "grad_norm": 2.257369041442871,
+      "learning_rate": 6.940896203887659e-05,
+      "loss": 2.801,
+      "step": 422
+    },
+    {
+      "epoch": 0.8138195777351248,
+      "grad_norm": 2.3267722129821777,
+      "learning_rate": 6.911582801512147e-05,
+      "loss": 2.8882,
+      "step": 424
+    },
+    {
+      "epoch": 0.817658349328215,
+      "grad_norm": 2.18564772605896,
+      "learning_rate": 6.882192245072569e-05,
+      "loss": 2.8164,
+      "step": 426
+    },
+    {
+      "epoch": 0.8214971209213052,
+      "grad_norm": 3.0314860343933105,
+      "learning_rate": 6.852725720811488e-05,
+      "loss": 2.7877,
+      "step": 428
+    },
+    {
+      "epoch": 0.8253358925143954,
+      "grad_norm": 2.14628005027771,
+      "learning_rate": 6.823184418037624e-05,
+      "loss": 2.812,
+      "step": 430
+    },
+    {
+      "epoch": 0.8291746641074856,
+      "grad_norm": 2.2532942295074463,
+      "learning_rate": 6.793569529077864e-05,
+      "loss": 2.7552,
+      "step": 432
+    },
+    {
+      "epoch": 0.8330134357005758,
+      "grad_norm": 2.1687734127044678,
+      "learning_rate": 6.76388224922913e-05,
+      "loss": 2.7738,
+      "step": 434
+    },
+    {
+      "epoch": 0.836852207293666,
+      "grad_norm": 2.3331027030944824,
+      "learning_rate": 6.734123776710137e-05,
+      "loss": 2.8104,
+      "step": 436
+    },
+    {
+      "epoch": 0.8406909788867563,
+      "grad_norm": 2.2633862495422363,
+      "learning_rate": 6.704295312613037e-05,
+      "loss": 2.7378,
+      "step": 438
+    },
+    {
+      "epoch": 0.8445297504798465,
+      "grad_norm": 2.0122358798980713,
+      "learning_rate": 6.674398060854931e-05,
+      "loss": 2.7827,
+      "step": 440
+    },
+    {
+      "epoch": 0.8483685220729367,
+      "grad_norm": 1.988330364227295,
+      "learning_rate": 6.644433228129287e-05,
+      "loss": 2.827,
+      "step": 442
+    },
+    {
+      "epoch": 0.8522072936660269,
+      "grad_norm": 2.3183484077453613,
+      "learning_rate": 6.614402023857232e-05,
+      "loss": 2.8699,
+      "step": 444
+    },
+    {
+      "epoch": 0.8560460652591171,
+      "grad_norm": 1.986506462097168,
+      "learning_rate": 6.584305660138734e-05,
+      "loss": 2.8937,
+      "step": 446
+    },
+    {
+      "epoch": 0.8598848368522073,
+      "grad_norm": 2.1625592708587646,
+      "learning_rate": 6.554145351703689e-05,
+      "loss": 2.8388,
+      "step": 448
+    },
+    {
+      "epoch": 0.8637236084452975,
+      "grad_norm": 2.391918897628784,
+      "learning_rate": 6.523922315862887e-05,
+      "loss": 2.7143,
+      "step": 450
+    },
+    {
+      "epoch": 0.8675623800383877,
+      "grad_norm": 2.2226004600524902,
+      "learning_rate": 6.493637772458879e-05,
+      "loss": 2.8072,
+      "step": 452
+    },
+    {
+      "epoch": 0.8714011516314779,
+      "grad_norm": 2.3496720790863037,
+      "learning_rate": 6.463292943816746e-05,
+      "loss": 2.9179,
+      "step": 454
+    },
+    {
+      "epoch": 0.8752399232245681,
+      "grad_norm": 2.0147578716278076,
+      "learning_rate": 6.432889054694764e-05,
+      "loss": 2.7736,
+      "step": 456
+    },
+    {
+      "epoch": 0.8790786948176583,
+      "grad_norm": 2.150322198867798,
+      "learning_rate": 6.402427332234965e-05,
+      "loss": 2.71,
+      "step": 458
+    },
+    {
+      "epoch": 0.8829174664107485,
+      "grad_norm": 1.9796924591064453,
+      "learning_rate": 6.371909005913618e-05,
+      "loss": 2.846,
+      "step": 460
+    },
+    {
+      "epoch": 0.8867562380038387,
+      "grad_norm": 1.9230434894561768,
+      "learning_rate": 6.341335307491595e-05,
+      "loss": 2.7577,
+      "step": 462
+    },
+    {
+      "epoch": 0.8905950095969289,
+      "grad_norm": 2.1865625381469727,
+      "learning_rate": 6.310707470964668e-05,
+      "loss": 2.7676,
+      "step": 464
+    },
+    {
+      "epoch": 0.8944337811900192,
+      "grad_norm": 2.251591682434082,
+      "learning_rate": 6.280026732513689e-05,
+      "loss": 2.7558,
+      "step": 466
+    },
+    {
+      "epoch": 0.8982725527831094,
+      "grad_norm": 2.048290252685547,
+      "learning_rate": 6.249294330454705e-05,
+      "loss": 2.806,
+      "step": 468
+    },
+    {
+      "epoch": 0.9021113243761996,
+      "grad_norm": 2.071037769317627,
+      "learning_rate": 6.21851150518898e-05,
+      "loss": 2.7993,
+      "step": 470
+    },
+    {
+      "epoch": 0.9059500959692899,
+      "grad_norm": 2.1514012813568115,
+      "learning_rate": 6.187679499152919e-05,
+      "loss": 2.8195,
+      "step": 472
+    },
+    {
+      "epoch": 0.9097888675623801,
+      "grad_norm": 2.073908567428589,
+      "learning_rate": 6.15679955676794e-05,
+      "loss": 2.7419,
+      "step": 474
+    },
+    {
+      "epoch": 0.9136276391554703,
+      "grad_norm": 2.136406660079956,
+      "learning_rate": 6.125872924390226e-05,
+      "loss": 2.6854,
+      "step": 476
+    },
+    {
+      "epoch": 0.9174664107485605,
+      "grad_norm": 1.851083755493164,
+      "learning_rate": 6.094900850260439e-05,
+      "loss": 2.8086,
+      "step": 478
+    },
+    {
+      "epoch": 0.9213051823416507,
+      "grad_norm": 2.3391993045806885,
+      "learning_rate": 6.0638845844533254e-05,
+      "loss": 2.8492,
+      "step": 480
+    },
+    {
+      "epoch": 0.9251439539347409,
+      "grad_norm": 2.0181124210357666,
+      "learning_rate": 6.032825378827273e-05,
+      "loss": 2.7973,
+      "step": 482
+    },
+    {
+      "epoch": 0.9289827255278311,
+      "grad_norm": 1.977287769317627,
+      "learning_rate": 6.0017244869737746e-05,
+      "loss": 2.7837,
+      "step": 484
+    },
+    {
+      "epoch": 0.9328214971209213,
+      "grad_norm": 2.0987181663513184,
+      "learning_rate": 5.970583164166838e-05,
+      "loss": 2.7967,
+      "step": 486
+    },
+    {
+      "epoch": 0.9366602687140115,
+      "grad_norm": 2.025846242904663,
+      "learning_rate": 5.9394026673123157e-05,
+      "loss": 2.8064,
+      "step": 488
+    },
+    {
+      "epoch": 0.9404990403071017,
+      "grad_norm": 2.065599203109741,
+      "learning_rate": 5.908184254897182e-05,
+      "loss": 2.7712,
+      "step": 490
+    },
+    {
+      "epoch": 0.944337811900192,
+      "grad_norm": 2.3943252563476562,
+      "learning_rate": 5.8769291869387335e-05,
+      "loss": 2.8494,
+      "step": 492
+    },
+    {
+      "epoch": 0.9481765834932822,
+      "grad_norm": 2.1195807456970215,
+      "learning_rate": 5.845638724933728e-05,
+      "loss": 2.7855,
+      "step": 494
+    },
+    {
+      "epoch": 0.9520153550863724,
+      "grad_norm": 1.8809967041015625,
+      "learning_rate": 5.814314131807486e-05,
+      "loss": 2.79,
+      "step": 496
+    },
+    {
+      "epoch": 0.9558541266794626,
+      "grad_norm": 1.9404385089874268,
+      "learning_rate": 5.7829566718628946e-05,
+      "loss": 2.7431,
+      "step": 498
+    },
+    {
+      "epoch": 0.9596928982725528,
+      "grad_norm": 1.913078784942627,
+      "learning_rate": 5.751567610729398e-05,
+      "loss": 2.7221,
+      "step": 500
+    },
+    {
+      "epoch": 0.963531669865643,
+      "grad_norm": 1.974898338317871,
+      "learning_rate": 5.7201482153119015e-05,
+      "loss": 2.7087,
+      "step": 502
+    },
+    {
+      "epoch": 0.9673704414587332,
+      "grad_norm": 2.0164756774902344,
+      "learning_rate": 5.6886997537396494e-05,
+      "loss": 2.6904,
+      "step": 504
+    },
+    {
+      "epoch": 0.9712092130518234,
+      "grad_norm": 2.159647226333618,
+      "learning_rate": 5.6572234953150305e-05,
+      "loss": 2.8019,
+      "step": 506
+    },
+    {
+      "epoch": 0.9750479846449136,
+      "grad_norm": 1.9416415691375732,
+      "learning_rate": 5.6257207104623524e-05,
+      "loss": 2.7502,
+      "step": 508
+    },
+    {
+      "epoch": 0.9788867562380038,
+      "grad_norm": 2.1035892963409424,
+      "learning_rate": 5.594192670676568e-05,
+      "loss": 2.6954,
+      "step": 510
+    },
+    {
+      "epoch": 0.982725527831094,
+      "grad_norm": 1.8888871669769287,
+      "learning_rate": 5.5626406484719506e-05,
+      "loss": 2.751,
+      "step": 512
+    },
+    {
+      "epoch": 0.9865642994241842,
+      "grad_norm": 1.9274441003799438,
+      "learning_rate": 5.5310659173307366e-05,
+      "loss": 2.7768,
+      "step": 514
+    },
+    {
+      "epoch": 0.9904030710172744,
+      "grad_norm": 1.9080277681350708,
+      "learning_rate": 5.499469751651728e-05,
+      "loss": 2.8421,
+      "step": 516
+    },
+    {
+      "epoch": 0.9942418426103646,
+      "grad_norm": 1.9564564228057861,
+      "learning_rate": 5.467853426698853e-05,
+      "loss": 2.8099,
+      "step": 518
+    },
+    {
+      "epoch": 0.9980806142034548,
+      "grad_norm": 1.8488651514053345,
+      "learning_rate": 5.43621821854969e-05,
+      "loss": 2.733,
+      "step": 520
+    },
+    {
+      "epoch": 1.0019193857965452,
+      "grad_norm": 1.8804161548614502,
+      "learning_rate": 5.4045654040439776e-05,
+      "loss": 2.7516,
+      "step": 522
+    },
+    {
+      "epoch": 1.0057581573896353,
+      "grad_norm": 1.8232853412628174,
+      "learning_rate": 5.3728962607320654e-05,
+      "loss": 2.544,
+      "step": 524
+    },
+    {
+      "epoch": 1.0095969289827256,
+      "grad_norm": 1.684367299079895,
+      "learning_rate": 5.341212066823355e-05,
+      "loss": 2.548,
+      "step": 526
+    },
+    {
+      "epoch": 1.0134357005758157,
+      "grad_norm": 1.7816741466522217,
+      "learning_rate": 5.309514101134715e-05,
+      "loss": 2.5254,
+      "step": 528
+    },
+    {
+      "epoch": 1.017274472168906,
+      "grad_norm": 2.167954683303833,
+      "learning_rate": 5.2778036430388554e-05,
+      "loss": 2.3372,
+      "step": 530
+    },
+    {
+      "epoch": 1.021113243761996,
+      "grad_norm": 2.0625479221343994,
+      "learning_rate": 5.246081972412702e-05,
+      "loss": 2.6017,
+      "step": 532
+    },
+    {
+      "epoch": 1.0249520153550864,
+      "grad_norm": 1.869403600692749,
+      "learning_rate": 5.214350369585731e-05,
+      "loss": 2.4747,
+      "step": 534
+    },
+    {
+      "epoch": 1.0287907869481765,
+      "grad_norm": 1.9606835842132568,
+      "learning_rate": 5.182610115288295e-05,
+      "loss": 2.5672,
+      "step": 536
+    },
+    {
+      "epoch": 1.0326295585412668,
+      "grad_norm": 1.9363110065460205,
+      "learning_rate": 5.150862490599934e-05,
+      "loss": 2.5577,
+      "step": 538
+    },
+    {
+      "epoch": 1.036468330134357,
+      "grad_norm": 1.8159326314926147,
+      "learning_rate": 5.119108776897664e-05,
+      "loss": 2.5811,
+      "step": 540
+    },
+    {
+      "epoch": 1.0403071017274472,
+      "grad_norm": 1.9988152980804443,
+      "learning_rate": 5.0873502558042665e-05,
+      "loss": 2.4306,
+      "step": 542
+    },
+    {
+      "epoch": 1.0441458733205373,
+      "grad_norm": 1.9837391376495361,
+      "learning_rate": 5.055588209136548e-05,
+      "loss": 2.6292,
+      "step": 544
+    },
+    {
+      "epoch": 1.0479846449136276,
+      "grad_norm": 1.7319928407669067,
+      "learning_rate": 5.0238239188536215e-05,
+      "loss": 2.4065,
+      "step": 546
+    },
+    {
+      "epoch": 1.051823416506718,
+      "grad_norm": 1.8957133293151855,
+      "learning_rate": 4.992058667005149e-05,
+      "loss": 2.6305,
+      "step": 548
+    },
+    {
+      "epoch": 1.055662188099808,
+      "grad_norm": 1.9159317016601562,
+      "learning_rate": 4.9602937356796014e-05,
+      "loss": 2.4567,
+      "step": 550
+    },
+    {
+      "epoch": 1.0595009596928984,
+      "grad_norm": 1.9933502674102783,
+      "learning_rate": 4.92853040695252e-05,
+      "loss": 2.5151,
+      "step": 552
+    },
+    {
+      "epoch": 1.0633397312859885,
+      "grad_norm": 2.0190207958221436,
+      "learning_rate": 4.896769962834754e-05,
+      "loss": 2.4625,
+      "step": 554
+    },
+    {
+      "epoch": 1.0671785028790788,
+      "grad_norm": 2.0344066619873047,
+      "learning_rate": 4.865013685220733e-05,
+      "loss": 2.4921,
+      "step": 556
+    },
+    {
+      "epoch": 1.0710172744721689,
+      "grad_norm": 2.000459909439087,
+      "learning_rate": 4.83326285583672e-05,
+      "loss": 2.6122,
+      "step": 558
+    },
+    {
+      "epoch": 1.0748560460652592,
+      "grad_norm": 2.0316357612609863,
+      "learning_rate": 4.801518756189074e-05,
+      "loss": 2.4042,
+      "step": 560
+    },
+    {
+      "epoch": 1.0786948176583493,
+      "grad_norm": 1.9071227312088013,
+      "learning_rate": 4.76978266751254e-05,
+      "loss": 2.5865,
+      "step": 562
+    },
+    {
+      "epoch": 1.0825335892514396,
+      "grad_norm": 1.8110324144363403,
+      "learning_rate": 4.7380558707185284e-05,
+      "loss": 2.2694,
+      "step": 564
+    },
+    {
+      "epoch": 1.0863723608445297,
+      "grad_norm": 1.8338054418563843,
+      "learning_rate": 4.7063396463434104e-05,
+      "loss": 2.5412,
+      "step": 566
+    },
+    {
+      "epoch": 1.09021113243762,
+      "grad_norm": 1.961002230644226,
+      "learning_rate": 4.674635274496847e-05,
+      "loss": 2.6088,
+      "step": 568
+    },
+    {
+      "epoch": 1.0940499040307101,
+      "grad_norm": 1.7149364948272705,
+      "learning_rate": 4.642944034810113e-05,
+      "loss": 2.521,
+      "step": 570
+    },
+    {
+      "epoch": 1.0978886756238004,
+      "grad_norm": 1.9060455560684204,
+      "learning_rate": 4.61126720638445e-05,
+      "loss": 2.2973,
+      "step": 572
+    },
+    {
+      "epoch": 1.1017274472168905,
+      "grad_norm": 1.8988847732543945,
+      "learning_rate": 4.5796060677394444e-05,
+      "loss": 2.4553,
+      "step": 574
+    },
+    {
+      "epoch": 1.1055662188099808,
+      "grad_norm": 1.9789772033691406,
+      "learning_rate": 4.547961896761419e-05,
+      "loss": 2.4774,
+      "step": 576
+    },
+    {
+      "epoch": 1.109404990403071,
+      "grad_norm": 1.8079990148544312,
+      "learning_rate": 4.51633597065186e-05,
+      "loss": 2.4428,
+      "step": 578
+    },
+    {
+      "epoch": 1.1132437619961613,
+      "grad_norm": 1.9464575052261353,
+      "learning_rate": 4.484729565875865e-05,
+      "loss": 2.4523,
+      "step": 580
+    },
+    {
+      "epoch": 1.1170825335892514,
+      "grad_norm": 1.917179822921753,
+      "learning_rate": 4.4531439581106295e-05,
+      "loss": 2.4285,
+      "step": 582
+    },
+    {
+      "epoch": 1.1209213051823417,
+      "grad_norm": 1.8777834177017212,
+      "learning_rate": 4.421580422193943e-05,
+      "loss": 2.4706,
+      "step": 584
+    },
+    {
+      "epoch": 1.1247600767754318,
+      "grad_norm": 1.7496826648712158,
+      "learning_rate": 4.390040232072755e-05,
+      "loss": 2.4908,
+      "step": 586
+    },
+    {
+      "epoch": 1.128598848368522,
+      "grad_norm": 1.7248530387878418,
+      "learning_rate": 4.358524660751746e-05,
+      "loss": 2.3502,
+      "step": 588
+    },
+    {
+      "epoch": 1.1324376199616122,
+      "grad_norm": 1.9547333717346191,
+      "learning_rate": 4.327034980241941e-05,
+      "loss": 2.5444,
+      "step": 590
+    },
+    {
+      "epoch": 1.1362763915547025,
+      "grad_norm": 2.112178325653076,
+      "learning_rate": 4.2955724615093844e-05,
+      "loss": 2.5955,
+      "step": 592
+    },
+    {
+      "epoch": 1.1401151631477928,
+      "grad_norm": 1.8485419750213623,
+      "learning_rate": 4.264138374423835e-05,
+      "loss": 2.2975,
+      "step": 594
+    },
+    {
+      "epoch": 1.143953934740883,
+      "grad_norm": 1.9420745372772217,
+      "learning_rate": 4.232733987707505e-05,
+      "loss": 2.5706,
+      "step": 596
+    },
+    {
+      "epoch": 1.147792706333973,
+      "grad_norm": 1.8905168771743774,
+      "learning_rate": 4.2013605688838656e-05,
+      "loss": 2.3751,
+      "step": 598
+    },
+    {
+      "epoch": 1.1516314779270633,
+      "grad_norm": 1.7361751794815063,
+      "learning_rate": 4.1700193842264814e-05,
+      "loss": 2.4757,
+      "step": 600
+    },
+    {
+      "epoch": 1.1554702495201536,
+      "grad_norm": 1.778198003768921,
+      "learning_rate": 4.138711698707899e-05,
+      "loss": 2.4159,
+      "step": 602
+    },
+    {
+      "epoch": 1.1593090211132437,
+      "grad_norm": 1.834826111793518,
+      "learning_rate": 4.107438775948598e-05,
+      "loss": 2.4044,
+      "step": 604
+    },
+    {
+      "epoch": 1.163147792706334,
+      "grad_norm": 1.765609622001648,
+      "learning_rate": 4.076201878165985e-05,
+      "loss": 2.3987,
+      "step": 606
+    },
+    {
+      "epoch": 1.1669865642994242,
+      "grad_norm": 2.0101146697998047,
+      "learning_rate": 4.0450022661234496e-05,
+      "loss": 2.5001,
+      "step": 608
+    },
+    {
+      "epoch": 1.1708253358925145,
+      "grad_norm": 2.0628116130828857,
+      "learning_rate": 4.0138411990794786e-05,
+      "loss": 2.5032,
+      "step": 610
+    },
+    {
+      "epoch": 1.1746641074856046,
+      "grad_norm": 1.866971731185913,
+      "learning_rate": 3.982719934736832e-05,
+      "loss": 2.3883,
+      "step": 612
+    },
+    {
+      "epoch": 1.1785028790786949,
+      "grad_norm": 2.0846517086029053,
+      "learning_rate": 3.9516397291917754e-05,
+      "loss": 2.4801,
+      "step": 614
+    },
+    {
+      "epoch": 1.182341650671785,
+      "grad_norm": 1.861687421798706,
+      "learning_rate": 3.920601836883389e-05,
+      "loss": 2.2688,
+      "step": 616
+    },
+    {
+      "epoch": 1.1861804222648753,
+      "grad_norm": 1.7908990383148193,
+      "learning_rate": 3.889607510542936e-05,
+      "loss": 2.3655,
+      "step": 618
+    },
+    {
+      "epoch": 1.1900191938579654,
+      "grad_norm": 2.128852605819702,
+      "learning_rate": 3.858658001143289e-05,
+      "loss": 2.5408,
+      "step": 620
+    },
+    {
+      "epoch": 1.1938579654510557,
+      "grad_norm": 1.8755757808685303,
+      "learning_rate": 3.82775455784846e-05,
+      "loss": 2.3866,
+      "step": 622
+    },
+    {
+      "epoch": 1.1976967370441458,
+      "grad_norm": 1.8754103183746338,
+      "learning_rate": 3.796898427963163e-05,
+      "loss": 2.4923,
+      "step": 624
+    },
+    {
+      "epoch": 1.2015355086372361,
+      "grad_norm": 1.8576866388320923,
+      "learning_rate": 3.7660908568824805e-05,
+      "loss": 2.6046,
+      "step": 626
+    },
+    {
+      "epoch": 1.2053742802303262,
+      "grad_norm": 1.7710564136505127,
+      "learning_rate": 3.735333088041596e-05,
+      "loss": 2.4535,
+      "step": 628
+    },
+    {
+      "epoch": 1.2092130518234165,
+      "grad_norm": 1.7709726095199585,
+      "learning_rate": 3.704626362865612e-05,
+      "loss": 2.5116,
+      "step": 630
+    },
+    {
+      "epoch": 1.2130518234165066,
+      "grad_norm": 1.9364629983901978,
+      "learning_rate": 3.673971920719431e-05,
+      "loss": 2.4581,
+      "step": 632
+    },
+    {
+      "epoch": 1.216890595009597,
+      "grad_norm": 1.938812255859375,
+      "learning_rate": 3.643370998857748e-05,
+      "loss": 2.4839,
+      "step": 634
+    },
+    {
+      "epoch": 1.220729366602687,
+      "grad_norm": 1.914156198501587,
+      "learning_rate": 3.612824832375109e-05,
+      "loss": 2.2989,
+      "step": 636
+    },
+    {
+      "epoch": 1.2245681381957774,
+      "grad_norm": 1.5851590633392334,
+      "learning_rate": 3.5823346541560495e-05,
+      "loss": 2.2631,
+      "step": 638
+    },
+    {
+      "epoch": 1.2284069097888675,
+      "grad_norm": 2.01133131980896,
+      "learning_rate": 3.551901694825352e-05,
+      "loss": 2.3921,
+      "step": 640
+    },
+    {
+      "epoch": 1.2322456813819578,
+      "grad_norm": 1.9045090675354004,
+      "learning_rate": 3.521527182698365e-05,
+      "loss": 2.3766,
+      "step": 642
+    },
+    {
+      "epoch": 1.236084452975048,
+      "grad_norm": 1.786575436592102,
+      "learning_rate": 3.491212343731428e-05,
+      "loss": 2.5844,
+      "step": 644
+    },
+    {
+      "epoch": 1.2399232245681382,
+      "grad_norm": 2.0299885272979736,
+      "learning_rate": 3.4609584014723904e-05,
+      "loss": 2.4664,
+      "step": 646
+    },
+    {
+      "epoch": 1.2437619961612283,
+      "grad_norm": 2.086113691329956,
+      "learning_rate": 3.430766577011231e-05,
+      "loss": 2.4601,
+      "step": 648
+    },
+    {
+      "epoch": 1.2476007677543186,
+      "grad_norm": 1.7769064903259277,
+      "learning_rate": 3.400638088930766e-05,
+      "loss": 2.4849,
+      "step": 650
+    },
+    {
+      "epoch": 1.251439539347409,
+      "grad_norm": 2.0460405349731445,
+      "learning_rate": 3.370574153257474e-05,
+      "loss": 2.6407,
+      "step": 652
+    },
+    {
+      "epoch": 1.255278310940499,
+      "grad_norm": 1.994273066520691,
+      "learning_rate": 3.340575983412412e-05,
+      "loss": 2.3818,
+      "step": 654
+    },
+    {
+      "epoch": 1.2591170825335891,
+      "grad_norm": 1.7842358350753784,
+      "learning_rate": 3.3106447901622324e-05,
+      "loss": 2.494,
+      "step": 656
+    },
+    {
+      "epoch": 1.2629558541266794,
+      "grad_norm": 1.8710514307022095,
+      "learning_rate": 3.280781781570328e-05,
+      "loss": 2.3716,
+      "step": 658
+    },
+    {
+      "epoch": 1.2667946257197698,
+      "grad_norm": 1.8952568769454956,
+      "learning_rate": 3.250988162948067e-05,
+      "loss": 2.5571,
+      "step": 660
+    },
+    {
+      "epoch": 1.2706333973128598,
+      "grad_norm": 1.9719605445861816,
+      "learning_rate": 3.221265136806139e-05,
+      "loss": 2.4698,
+      "step": 662
+    },
+    {
+      "epoch": 1.2744721689059502,
+      "grad_norm": 1.9866466522216797,
+      "learning_rate": 3.1916139028060315e-05,
+      "loss": 2.4504,
+      "step": 664
+    },
+    {
+      "epoch": 1.2783109404990403,
+      "grad_norm": 2.0619564056396484,
+      "learning_rate": 3.162035657711604e-05,
+      "loss": 2.5778,
+      "step": 666
+    },
+    {
+      "epoch": 1.2821497120921306,
+      "grad_norm": 1.7766873836517334,
+      "learning_rate": 3.132531595340779e-05,
+      "loss": 2.5757,
+      "step": 668
+    },
+    {
+      "epoch": 1.2859884836852207,
+      "grad_norm": 1.8886102437973022,
+      "learning_rate": 3.1031029065173704e-05,
+      "loss": 2.5952,
+      "step": 670
+    },
+    {
+      "epoch": 1.289827255278311,
+      "grad_norm": 1.8903406858444214,
+      "learning_rate": 3.073750779023014e-05,
+      "loss": 2.5322,
+      "step": 672
+    },
+    {
+      "epoch": 1.293666026871401,
+      "grad_norm": 1.7394644021987915,
+      "learning_rate": 3.0444763975492208e-05,
+      "loss": 2.593,
+      "step": 674
+    },
+    {
+      "epoch": 1.2975047984644914,
+      "grad_norm": 1.9238591194152832,
+      "learning_rate": 3.0152809436495734e-05,
+      "loss": 2.2963,
+      "step": 676
+    },
+    {
+      "epoch": 1.3013435700575815,
+      "grad_norm": 1.8425519466400146,
+      "learning_rate": 2.9861655956920286e-05,
+      "loss": 2.3245,
+      "step": 678
+    },
+    {
+      "epoch": 1.3051823416506718,
+      "grad_norm": 1.892540693283081,
+      "learning_rate": 2.95713152881136e-05,
+      "loss": 2.4921,
+      "step": 680
+    },
+    {
+      "epoch": 1.309021113243762,
+      "grad_norm": 1.8045918941497803,
+      "learning_rate": 2.9281799148617262e-05,
+      "loss": 2.4807,
+      "step": 682
+    },
+    {
+      "epoch": 1.3128598848368522,
+      "grad_norm": 1.925676703453064,
+      "learning_rate": 2.8993119223693755e-05,
+      "loss": 2.5433,
+      "step": 684
+    },
+    {
+      "epoch": 1.3166986564299425,
+      "grad_norm": 2.1154892444610596,
+      "learning_rate": 2.8705287164854755e-05,
+      "loss": 2.5387,
+      "step": 686
+    },
+    {
+      "epoch": 1.3205374280230326,
+      "grad_norm": 1.7754255533218384,
+      "learning_rate": 2.8418314589391027e-05,
+      "loss": 2.4555,
+      "step": 688
+    },
+    {
+      "epoch": 1.3243761996161227,
+      "grad_norm": 1.999812364578247,
+      "learning_rate": 2.8132213079903336e-05,
+      "loss": 2.5424,
+      "step": 690
+    },
+    {
+      "epoch": 1.328214971209213,
+      "grad_norm": 1.8174470663070679,
+      "learning_rate": 2.7846994183835074e-05,
+      "loss": 2.5658,
+      "step": 692
+    },
+    {
+      "epoch": 1.3320537428023034,
+      "grad_norm": 1.8758563995361328,
+      "learning_rate": 2.7562669413006148e-05,
+      "loss": 2.3513,
+      "step": 694
+    },
+    {
+      "epoch": 1.3358925143953935,
+      "grad_norm": 1.809468150138855,
+      "learning_rate": 2.7279250243148412e-05,
+      "loss": 2.359,
+      "step": 696
+    },
+    {
+      "epoch": 1.3397312859884836,
+      "grad_norm": 1.7983132600784302,
+      "learning_rate": 2.6996748113442394e-05,
+      "loss": 2.3086,
+      "step": 698
+    },
+    {
+      "epoch": 1.3435700575815739,
+      "grad_norm": 1.9790072441101074,
+      "learning_rate": 2.6715174426055662e-05,
+      "loss": 2.5407,
+      "step": 700
+    },
+    {
+      "epoch": 1.3474088291746642,
+      "grad_norm": 1.73479163646698,
+      "learning_rate": 2.6434540545682584e-05,
+      "loss": 2.3909,
+      "step": 702
+    },
+    {
+      "epoch": 1.3512476007677543,
+      "grad_norm": 1.7946439981460571,
+      "learning_rate": 2.6154857799085643e-05,
+      "loss": 2.3164,
+      "step": 704
+    },
+    {
+      "epoch": 1.3550863723608444,
+      "grad_norm": 1.8529855012893677,
+      "learning_rate": 2.5876137474638323e-05,
+      "loss": 2.5285,
+      "step": 706
+    },
+    {
+      "epoch": 1.3589251439539347,
+      "grad_norm": 1.9864492416381836,
+      "learning_rate": 2.55983908218694e-05,
+      "loss": 2.5069,
+      "step": 708
+    },
+    {
+      "epoch": 1.362763915547025,
+      "grad_norm": 1.7641496658325195,
+      "learning_rate": 2.5321629051008976e-05,
+      "loss": 2.3275,
+      "step": 710
+    },
+    {
+      "epoch": 1.3666026871401151,
+      "grad_norm": 1.7185853719711304,
+      "learning_rate": 2.5045863332535945e-05,
+      "loss": 2.2071,
+      "step": 712
+    },
+    {
+      "epoch": 1.3704414587332054,
+      "grad_norm": 1.7760812044143677,
+      "learning_rate": 2.4771104796727273e-05,
+      "loss": 2.4337,
+      "step": 714
+    },
+    {
+      "epoch": 1.3742802303262955,
+      "grad_norm": 1.7367132902145386,
+      "learning_rate": 2.449736453320854e-05,
+      "loss": 2.4869,
+      "step": 716
+    },
+    {
+      "epoch": 1.3781190019193859,
+      "grad_norm": 1.9105916023254395,
+      "learning_rate": 2.4224653590506607e-05,
+      "loss": 2.4345,
+      "step": 718
+    },
+    {
+      "epoch": 1.381957773512476,
+      "grad_norm": 1.7931406497955322,
+      "learning_rate": 2.3952982975603496e-05,
+      "loss": 2.3898,
+      "step": 720
+    },
+    {
+      "epoch": 1.3857965451055663,
+      "grad_norm": 1.8123571872711182,
+      "learning_rate": 2.3682363653492178e-05,
+      "loss": 2.4744,
+      "step": 722
+    },
+    {
+      "epoch": 1.3896353166986564,
+      "grad_norm": 1.8735435009002686,
+      "learning_rate": 2.3412806546734055e-05,
+      "loss": 2.4267,
+      "step": 724
+    },
+    {
+      "epoch": 1.3934740882917467,
+      "grad_norm": 1.9823077917099,
+      "learning_rate": 2.3144322535018127e-05,
+      "loss": 2.4519,
+      "step": 726
+    },
+    {
+      "epoch": 1.3973128598848368,
+      "grad_norm": 1.8242884874343872,
+      "learning_rate": 2.2876922454721694e-05,
+      "loss": 2.3486,
+      "step": 728
+    },
+    {
+      "epoch": 1.401151631477927,
+      "grad_norm": 1.7422012090682983,
+      "learning_rate": 2.261061709847327e-05,
+      "loss": 2.471,
+      "step": 730
+    },
+    {
+      "epoch": 1.4049904030710172,
+      "grad_norm": 1.8321129083633423,
+      "learning_rate": 2.2345417214716748e-05,
+      "loss": 2.4468,
+      "step": 732
+    },
+    {
+      "epoch": 1.4088291746641075,
+      "grad_norm": 1.9305740594863892,
+      "learning_rate": 2.208133350727764e-05,
+      "loss": 2.4354,
+      "step": 734
+    },
+    {
+      "epoch": 1.4126679462571978,
+      "grad_norm": 1.728279948234558,
+      "learning_rate": 2.1818376634931153e-05,
+      "loss": 2.5314,
+      "step": 736
+    },
+    {
+      "epoch": 1.416506717850288,
+      "grad_norm": 1.9136515855789185,
+      "learning_rate": 2.1556557210971847e-05,
+      "loss": 2.4653,
+      "step": 738
+    },
+    {
+      "epoch": 1.420345489443378,
+      "grad_norm": 1.7468599081039429,
+      "learning_rate": 2.129588580278533e-05,
+      "loss": 2.4217,
+      "step": 740
+    },
+    {
+      "epoch": 1.4241842610364683,
+      "grad_norm": 1.827812910079956,
+      "learning_rate": 2.103637293142175e-05,
+      "loss": 2.3471,
+      "step": 742
+    },
+    {
+      "epoch": 1.4280230326295587,
+      "grad_norm": 1.8303539752960205,
+      "learning_rate": 2.077802907117119e-05,
+      "loss": 2.4945,
+      "step": 744
+    },
+    {
+      "epoch": 1.4318618042226487,
+      "grad_norm": 1.9014883041381836,
+      "learning_rate": 2.0520864649140765e-05,
+      "loss": 2.4279,
+      "step": 746
+    },
+    {
+      "epoch": 1.4357005758157388,
+      "grad_norm": 1.861970067024231,
+      "learning_rate": 2.0264890044833995e-05,
+      "loss": 2.5185,
+      "step": 748
+    },
+    {
+      "epoch": 1.4395393474088292,
+      "grad_norm": 1.7734556198120117,
+      "learning_rate": 2.0010115589731613e-05,
+      "loss": 2.3451,
+      "step": 750
+    },
+    {
+      "epoch": 1.4433781190019195,
+      "grad_norm": 1.8593039512634277,
+      "learning_rate": 1.9756551566874837e-05,
+      "loss": 2.4111,
+      "step": 752
+    },
+    {
+      "epoch": 1.4472168905950096,
+      "grad_norm": 2.0306501388549805,
+      "learning_rate": 1.9504208210450127e-05,
+      "loss": 2.1885,
+      "step": 754
+    },
+    {
+      "epoch": 1.4510556621880997,
+      "grad_norm": 1.8446749448776245,
+      "learning_rate": 1.9253095705376218e-05,
+      "loss": 2.5532,
+      "step": 756
+    },
+    {
+      "epoch": 1.45489443378119,
+      "grad_norm": 1.7542308568954468,
+      "learning_rate": 1.9003224186893e-05,
+      "loss": 2.4208,
+      "step": 758
+    },
+    {
+      "epoch": 1.4587332053742803,
+      "grad_norm": 1.7439130544662476,
+      "learning_rate": 1.875460374015253e-05,
+      "loss": 2.3911,
+      "step": 760
+    },
+    {
+      "epoch": 1.4625719769673704,
+      "grad_norm": 1.713499903678894,
+      "learning_rate": 1.850724439981186e-05,
+      "loss": 2.3793,
+      "step": 762
+    },
+    {
+      "epoch": 1.4664107485604607,
+      "grad_norm": 1.8578466176986694,
+      "learning_rate": 1.82611561496281e-05,
+      "loss": 2.529,
+      "step": 764
+    },
+    {
+      "epoch": 1.4702495201535508,
+      "grad_norm": 1.881801724433899,
+      "learning_rate": 1.801634892205545e-05,
+      "loss": 2.3416,
+      "step": 766
+    },
+    {
+      "epoch": 1.4740882917466411,
+      "grad_norm": 1.628251314163208,
+      "learning_rate": 1.7772832597844285e-05,
+      "loss": 2.3498,
+      "step": 768
+    },
+    {
+      "epoch": 1.4779270633397312,
+      "grad_norm": 1.7867623567581177,
+      "learning_rate": 1.753061700564243e-05,
+      "loss": 2.5108,
+      "step": 770
+    },
+    {
+      "epoch": 1.4817658349328215,
+      "grad_norm": 1.7731565237045288,
+      "learning_rate": 1.728971192159836e-05,
+      "loss": 2.4097,
+      "step": 772
+    },
+    {
+      "epoch": 1.4856046065259116,
+      "grad_norm": 1.7546172142028809,
+      "learning_rate": 1.705012706896668e-05,
+      "loss": 2.3188,
+      "step": 774
+    },
+    {
+      "epoch": 1.489443378119002,
+      "grad_norm": 1.7338075637817383,
+      "learning_rate": 1.6811872117715673e-05,
+      "loss": 2.4417,
+      "step": 776
+    },
+    {
+      "epoch": 1.493282149712092,
+      "grad_norm": 1.7118301391601562,
+      "learning_rate": 1.6574956684137045e-05,
+      "loss": 2.4761,
+      "step": 778
+    },
+    {
+      "epoch": 1.4971209213051824,
+      "grad_norm": 1.8096641302108765,
+      "learning_rate": 1.633939033045766e-05,
+      "loss": 2.4536,
+      "step": 780
+    },
+    {
+      "epoch": 1.5009596928982725,
+      "grad_norm": 1.7673901319503784,
+      "learning_rate": 1.610518256445382e-05,
+      "loss": 2.4854,
+      "step": 782
+    },
+    {
+      "epoch": 1.5047984644913628,
+      "grad_norm": 1.8908624649047852,
+      "learning_rate": 1.5872342839067306e-05,
+      "loss": 2.3752,
+      "step": 784
+    },
+    {
+      "epoch": 1.508637236084453,
+      "grad_norm": 1.7519292831420898,
+      "learning_rate": 1.5640880552023955e-05,
+      "loss": 2.5641,
+      "step": 786
+    },
+    {
+      "epoch": 1.5124760076775432,
+      "grad_norm": 1.697787880897522,
+      "learning_rate": 1.541080504545433e-05,
+      "loss": 2.3805,
+      "step": 788
+    },
+    {
+      "epoch": 1.5163147792706333,
+      "grad_norm": 1.7646615505218506,
+      "learning_rate": 1.5182125605516707e-05,
+      "loss": 2.5407,
+      "step": 790
+    },
+    {
+      "epoch": 1.5201535508637236,
+      "grad_norm": 1.865901231765747,
+      "learning_rate": 1.4954851462022118e-05,
+      "loss": 2.4108,
+      "step": 792
+    },
+    {
+      "epoch": 1.523992322456814,
+      "grad_norm": 1.6581934690475464,
+      "learning_rate": 1.4728991788062052e-05,
+      "loss": 2.2233,
+      "step": 794
+    },
+    {
+      "epoch": 1.527831094049904,
+      "grad_norm": 1.936775803565979,
+      "learning_rate": 1.4504555699638034e-05,
+      "loss": 2.2673,
+      "step": 796
+    },
+    {
+      "epoch": 1.5316698656429941,
+      "grad_norm": 1.8059828281402588,
+      "learning_rate": 1.4281552255293739e-05,
+      "loss": 2.4363,
+      "step": 798
+    },
+    {
+      "epoch": 1.5355086372360844,
+      "grad_norm": 1.6729570627212524,
+      "learning_rate": 1.405999045574945e-05,
+      "loss": 2.3638,
+      "step": 800
+    },
+    {
+      "epoch": 1.5393474088291748,
+      "grad_norm": 1.5310014486312866,
+      "learning_rate": 1.3839879243538678e-05,
+      "loss": 2.1237,
+      "step": 802
+    },
+    {
+      "epoch": 1.5431861804222649,
+      "grad_norm": 1.6975595951080322,
+      "learning_rate": 1.362122750264727e-05,
+      "loss": 2.434,
+      "step": 804
+    },
+    {
+      "epoch": 1.547024952015355,
+      "grad_norm": 1.9747432470321655,
+      "learning_rate": 1.3404044058154835e-05,
+      "loss": 2.4547,
+      "step": 806
+    },
+    {
+      "epoch": 1.5508637236084453,
+      "grad_norm": 1.931599736213684,
+      "learning_rate": 1.3188337675878609e-05,
+      "loss": 2.2581,
+      "step": 808
+    },
+    {
+      "epoch": 1.5547024952015356,
+      "grad_norm": 1.690551519393921,
+      "learning_rate": 1.2974117062019503e-05,
+      "loss": 2.4887,
+      "step": 810
+    },
+    {
+      "epoch": 1.5585412667946257,
+      "grad_norm": 2.5367448329925537,
+      "learning_rate": 1.2761390862810907e-05,
+      "loss": 2.3701,
+      "step": 812
+    },
+    {
+      "epoch": 1.5623800383877158,
+      "grad_norm": 1.7968242168426514,
+      "learning_rate": 1.2550167664169566e-05,
+      "loss": 2.4853,
+      "step": 814
+    },
+    {
+      "epoch": 1.566218809980806,
+      "grad_norm": 1.73814058303833,
+      "learning_rate": 1.2340455991349093e-05,
+      "loss": 2.4246,
+      "step": 816
+    },
+    {
+      "epoch": 1.5700575815738964,
+      "grad_norm": 1.79641592502594,
+      "learning_rate": 1.2132264308595875e-05,
+      "loss": 2.4943,
+      "step": 818
+    },
+    {
+      "epoch": 1.5738963531669867,
+      "grad_norm": 1.7317153215408325,
+      "learning_rate": 1.1925601018807497e-05,
+      "loss": 2.5241,
+      "step": 820
+    },
+    {
+      "epoch": 1.5777351247600768,
+      "grad_norm": 1.8299314975738525,
+      "learning_rate": 1.1720474463193443e-05,
+      "loss": 2.4614,
+      "step": 822
+    },
+    {
+      "epoch": 1.581573896353167,
+      "grad_norm": 1.6449967622756958,
+      "learning_rate": 1.1516892920938626e-05,
+      "loss": 2.4743,
+      "step": 824
+    },
+    {
+      "epoch": 1.5854126679462572,
+      "grad_norm": 1.9421472549438477,
+      "learning_rate": 1.131486460886908e-05,
+      "loss": 2.4542,
+      "step": 826
+    },
+    {
+      "epoch": 1.5892514395393476,
+      "grad_norm": 1.8020849227905273,
+      "learning_rate": 1.1114397681120387e-05,
+      "loss": 2.4273,
+      "step": 828
+    },
+    {
+      "epoch": 1.5930902111324377,
+      "grad_norm": 1.8233286142349243,
+      "learning_rate": 1.0915500228808523e-05,
+      "loss": 2.4576,
+      "step": 830
+    },
+    {
+      "epoch": 1.5969289827255277,
+      "grad_norm": 1.8505898714065552,
+      "learning_rate": 1.0718180279703371e-05,
+      "loss": 2.3293,
+      "step": 832
+    },
+    {
+      "epoch": 1.600767754318618,
+      "grad_norm": 1.885059118270874,
+      "learning_rate": 1.0522445797904607e-05,
+      "loss": 2.4214,
+      "step": 834
+    },
+    {
+      "epoch": 1.6046065259117084,
+      "grad_norm": 1.697663426399231,
+      "learning_rate": 1.0328304683520306e-05,
+      "loss": 2.449,
+      "step": 836
+    },
+    {
+      "epoch": 1.6084452975047985,
+      "grad_norm": 1.6720285415649414,
+      "learning_rate": 1.0135764772348106e-05,
+      "loss": 2.4358,
+      "step": 838
+    },
+    {
+      "epoch": 1.6122840690978886,
+      "grad_norm": 1.7693917751312256,
+      "learning_rate": 9.944833835558887e-06,
+      "loss": 2.5039,
+      "step": 840
+    },
+    {
+      "epoch": 1.616122840690979,
+      "grad_norm": 1.7624363899230957,
+      "learning_rate": 9.755519579383204e-06,
+      "loss": 2.4609,
+      "step": 842
+    },
+    {
+      "epoch": 1.6199616122840692,
+      "grad_norm": 1.6836014986038208,
+      "learning_rate": 9.567829644800141e-06,
+      "loss": 2.4885,
+      "step": 844
+    },
+    {
+      "epoch": 1.6238003838771593,
+      "grad_norm": 1.7030502557754517,
+      "learning_rate": 9.381771607229e-06,
+      "loss": 2.4847,
+      "step": 846
+    },
+    {
+      "epoch": 1.6276391554702494,
+      "grad_norm": 1.9006402492523193,
+      "learning_rate": 9.197352976223495e-06,
+      "loss": 2.4292,
+      "step": 848
+    },
+    {
+      "epoch": 1.6314779270633397,
+      "grad_norm": 1.6497300863265991,
+      "learning_rate": 9.014581195168725e-06,
+      "loss": 2.4066,
+      "step": 850
+    },
+    {
+      "epoch": 1.63531669865643,
+      "grad_norm": 1.8096089363098145,
+      "learning_rate": 8.83346364098061e-06,
+      "loss": 2.3377,
+      "step": 852
+    },
+    {
+      "epoch": 1.6391554702495201,
+      "grad_norm": 1.703515648841858,
+      "learning_rate": 8.654007623808335e-06,
+      "loss": 2.5303,
+      "step": 854
+    },
+    {
+      "epoch": 1.6429942418426102,
+      "grad_norm": 1.6932040452957153,
+      "learning_rate": 8.476220386739153e-06,
+      "loss": 2.2796,
+      "step": 856
+    },
+    {
+      "epoch": 1.6468330134357005,
+      "grad_norm": 1.7195641994476318,
+      "learning_rate": 8.30010910550611e-06,
+      "loss": 2.1625,
+      "step": 858
+    },
+    {
+      "epoch": 1.6506717850287909,
+      "grad_norm": 1.6625498533248901,
+      "learning_rate": 8.125680888198396e-06,
+      "loss": 2.5335,
+      "step": 860
+    },
+    {
+      "epoch": 1.654510556621881,
+      "grad_norm": 1.7404083013534546,
+      "learning_rate": 7.95294277497452e-06,
+      "loss": 2.5084,
+      "step": 862
+    },
+    {
+      "epoch": 1.658349328214971,
+      "grad_norm": 1.6975153684616089,
+      "learning_rate": 7.781901737778013e-06,
+      "loss": 2.3985,
+      "step": 864
+    },
+    {
+      "epoch": 1.6621880998080614,
+      "grad_norm": 1.8806735277175903,
+      "learning_rate": 7.612564680056233e-06,
+      "loss": 2.3835,
+      "step": 866
+    },
+    {
+      "epoch": 1.6660268714011517,
+      "grad_norm": 1.7741533517837524,
+      "learning_rate": 7.4449384364815464e-06,
+      "loss": 2.3336,
+      "step": 868
+    },
+    {
+      "epoch": 1.669865642994242,
+      "grad_norm": 1.7553293704986572,
+      "learning_rate": 7.2790297726755716e-06,
+      "loss": 2.303,
+      "step": 870
+    },
+    {
+      "epoch": 1.673704414587332,
+      "grad_norm": 1.678241491317749,
+      "learning_rate": 7.114845384936108e-06,
+      "loss": 2.3424,
+      "step": 872
+    },
+    {
+      "epoch": 1.6775431861804222,
+      "grad_norm": 1.7955679893493652,
+      "learning_rate": 6.9523918999668255e-06,
+      "loss": 2.2851,
+      "step": 874
+    },
+    {
+      "epoch": 1.6813819577735125,
+      "grad_norm": 1.733140468597412,
+      "learning_rate": 6.7916758746098155e-06,
+      "loss": 2.3875,
+      "step": 876
+    },
+    {
+      "epoch": 1.6852207293666028,
+      "grad_norm": 1.778879165649414,
+      "learning_rate": 6.632703795580946e-06,
+      "loss": 2.4897,
+      "step": 878
+    },
+    {
+      "epoch": 1.689059500959693,
+      "grad_norm": 1.8510472774505615,
+      "learning_rate": 6.475482079208112e-06,
+      "loss": 2.3219,
+      "step": 880
+    },
+    {
+      "epoch": 1.692898272552783,
+      "grad_norm": 1.5964735746383667,
+      "learning_rate": 6.320017071172113e-06,
+      "loss": 2.2506,
+      "step": 882
+    },
+    {
+      "epoch": 1.6967370441458733,
+      "grad_norm": 1.7399059534072876,
+      "learning_rate": 6.1663150462507034e-06,
+      "loss": 2.3867,
+      "step": 884
+    },
+    {
+      "epoch": 1.7005758157389637,
+      "grad_norm": 1.704087495803833,
+      "learning_rate": 6.014382208065234e-06,
+      "loss": 2.4008,
+      "step": 886
+    },
+    {
+      "epoch": 1.7044145873320538,
+      "grad_norm": 1.610163927078247,
+      "learning_rate": 5.864224688830283e-06,
+      "loss": 2.234,
+      "step": 888
+    },
+    {
+      "epoch": 1.7082533589251438,
+      "grad_norm": 1.5860165357589722,
+      "learning_rate": 5.715848549106145e-06,
+      "loss": 2.2767,
+      "step": 890
+    },
+    {
+      "epoch": 1.7120921305182342,
+      "grad_norm": 1.7125369310379028,
+      "learning_rate": 5.569259777554287e-06,
+      "loss": 2.4252,
+      "step": 892
+    },
+    {
+      "epoch": 1.7159309021113245,
+      "grad_norm": 1.8192423582077026,
+      "learning_rate": 5.424464290695496e-06,
+      "loss": 2.419,
+      "step": 894
+    },
+    {
+      "epoch": 1.7197696737044146,
+      "grad_norm": 1.7669687271118164,
+      "learning_rate": 5.281467932671253e-06,
+      "loss": 2.386,
+      "step": 896
+    },
+    {
+      "epoch": 1.7236084452975047,
+      "grad_norm": 1.677895426750183,
+      "learning_rate": 5.140276475007711e-06,
+      "loss": 2.5122,
+      "step": 898
+    },
+    {
+      "epoch": 1.727447216890595,
+      "grad_norm": 1.746677279472351,
+      "learning_rate": 5.000895616382828e-06,
+      "loss": 2.3692,
+      "step": 900
+    },
+    {
+      "epoch": 1.7312859884836853,
+      "grad_norm": 1.6640464067459106,
+      "learning_rate": 4.863330982396319e-06,
+      "loss": 2.317,
+      "step": 902
+    },
+    {
+      "epoch": 1.7351247600767754,
+      "grad_norm": 1.6886448860168457,
+      "learning_rate": 4.727588125342669e-06,
+      "loss": 2.2828,
+      "step": 904
+    },
+    {
+      "epoch": 1.7389635316698655,
+      "grad_norm": 1.8009768724441528,
+      "learning_rate": 4.593672523986936e-06,
+      "loss": 2.3412,
+      "step": 906
+    },
+    {
+      "epoch": 1.7428023032629558,
+      "grad_norm": 1.8097089529037476,
+      "learning_rate": 4.461589583343678e-06,
+      "loss": 2.3527,
+      "step": 908
+    },
+    {
+      "epoch": 1.7466410748560461,
+      "grad_norm": 1.6583417654037476,
+      "learning_rate": 4.331344634458812e-06,
+      "loss": 2.3027,
+      "step": 910
+    },
+    {
+      "epoch": 1.7504798464491362,
+      "grad_norm": 1.6135237216949463,
+      "learning_rate": 4.2029429341943984e-06,
+      "loss": 2.3895,
+      "step": 912
+    },
+    {
+      "epoch": 1.7543186180422263,
+      "grad_norm": 1.6572721004486084,
+      "learning_rate": 4.076389665016522e-06,
+      "loss": 2.4186,
+      "step": 914
+    },
+    {
+      "epoch": 1.7581573896353166,
+      "grad_norm": 1.652147650718689,
+      "learning_rate": 3.951689934786068e-06,
+      "loss": 2.3758,
+      "step": 916
+    },
+    {
+      "epoch": 1.761996161228407,
+      "grad_norm": 1.785905122756958,
+      "learning_rate": 3.828848776552596e-06,
+      "loss": 2.3179,
+      "step": 918
+    },
+    {
+      "epoch": 1.7658349328214973,
+      "grad_norm": 1.8040014505386353,
+      "learning_rate": 3.707871148351183e-06,
+      "loss": 2.4267,
+      "step": 920
+    },
+    {
+      "epoch": 1.7696737044145874,
+      "grad_norm": 1.7442965507507324,
+      "learning_rate": 3.5887619330023435e-06,
+      "loss": 2.4342,
+      "step": 922
+    },
+    {
+      "epoch": 1.7735124760076775,
+      "grad_norm": 1.6539864540100098,
+      "learning_rate": 3.4715259379148656e-06,
+      "loss": 2.3907,
+      "step": 924
+    },
+    {
+      "epoch": 1.7773512476007678,
+      "grad_norm": 1.7002536058425903,
+      "learning_rate": 3.3561678948919095e-06,
+      "loss": 2.2729,
+      "step": 926
+    },
+    {
+      "epoch": 1.781190019193858,
+      "grad_norm": 1.7664355039596558,
+      "learning_rate": 3.2426924599399056e-06,
+      "loss": 2.244,
+      "step": 928
+    },
+    {
+      "epoch": 1.7850287907869482,
+      "grad_norm": 1.5973105430603027,
+      "learning_rate": 3.131104213080688e-06,
+      "loss": 2.4728,
+      "step": 930
+    },
+    {
+      "epoch": 1.7888675623800383,
+      "grad_norm": 1.7511796951293945,
+      "learning_rate": 3.021407658166636e-06,
+      "loss": 2.3096,
+      "step": 932
+    },
+    {
+      "epoch": 1.7927063339731286,
+      "grad_norm": 1.6772406101226807,
+      "learning_rate": 2.913607222698905e-06,
+      "loss": 2.3705,
+      "step": 934
+    },
+    {
+      "epoch": 1.796545105566219,
+      "grad_norm": 1.704423427581787,
+      "learning_rate": 2.807707257648662e-06,
+      "loss": 2.3784,
+      "step": 936
+    },
+    {
+      "epoch": 1.800383877159309,
+      "grad_norm": 1.6434552669525146,
+      "learning_rate": 2.703712037281564e-06,
+      "loss": 2.4451,
+      "step": 938
+    },
+    {
+      "epoch": 1.8042226487523991,
+      "grad_norm": 1.5974692106246948,
+      "learning_rate": 2.601625758985182e-06,
+      "loss": 2.4799,
+      "step": 940
+    },
+    {
+      "epoch": 1.8080614203454894,
+      "grad_norm": 1.7015239000320435,
+      "learning_rate": 2.5014525430995916e-06,
+      "loss": 2.2772,
+      "step": 942
+    },
+    {
+      "epoch": 1.8119001919385798,
+      "grad_norm": 1.8971093893051147,
+      "learning_rate": 2.4031964327511313e-06,
+      "loss": 2.4349,
+      "step": 944
+    },
+    {
+      "epoch": 1.8157389635316699,
+      "grad_norm": 1.7988215684890747,
+      "learning_rate": 2.3068613936891137e-06,
+      "loss": 2.4552,
+      "step": 946
+    },
+    {
+      "epoch": 1.81957773512476,
+      "grad_norm": 1.6101404428482056,
+      "learning_rate": 2.2124513141258574e-06,
+      "loss": 2.408,
+      "step": 948
+    },
+    {
+      "epoch": 1.8234165067178503,
+      "grad_norm": 1.686816930770874,
+      "learning_rate": 2.1199700045797077e-06,
+      "loss": 2.4018,
+      "step": 950
+    },
+    {
+      "epoch": 1.8272552783109406,
+      "grad_norm": 1.6202586889266968,
+      "learning_rate": 2.0294211977212317e-06,
+      "loss": 2.4629,
+      "step": 952
+    },
+    {
+      "epoch": 1.8310940499040307,
+      "grad_norm": 1.7056370973587036,
+      "learning_rate": 1.9408085482225945e-06,
+      "loss": 2.512,
+      "step": 954
+    },
+    {
+      "epoch": 1.8349328214971208,
+      "grad_norm": 1.7047405242919922,
+      "learning_rate": 1.8541356326100433e-06,
+      "loss": 2.3859,
+      "step": 956
+    },
+    {
+      "epoch": 1.838771593090211,
+      "grad_norm": 1.5985552072525024,
+      "learning_rate": 1.7694059491195013e-06,
+      "loss": 2.4665,
+      "step": 958
+    },
+    {
+      "epoch": 1.8426103646833014,
+      "grad_norm": 1.6778897047042847,
+      "learning_rate": 1.6866229175554748e-06,
+      "loss": 2.3102,
+      "step": 960
+    },
+    {
+      "epoch": 1.8464491362763915,
+      "grad_norm": 1.6465007066726685,
+      "learning_rate": 1.6057898791529303e-06,
+      "loss": 2.2886,
+      "step": 962
+    },
+    {
+      "epoch": 1.8502879078694816,
+      "grad_norm": 1.5736628770828247,
+      "learning_rate": 1.526910096442491e-06,
+      "loss": 2.2019,
+      "step": 964
+    },
+    {
+      "epoch": 1.854126679462572,
+      "grad_norm": 1.5600590705871582,
+      "learning_rate": 1.4499867531187373e-06,
+      "loss": 2.2312,
+      "step": 966
+    },
+    {
+      "epoch": 1.8579654510556622,
+      "grad_norm": 1.828391432762146,
+      "learning_rate": 1.3750229539117144e-06,
+      "loss": 2.3583,
+      "step": 968
+    },
+    {
+      "epoch": 1.8618042226487526,
+      "grad_norm": 1.5677580833435059,
+      "learning_rate": 1.3020217244616273e-06,
+      "loss": 2.3495,
+      "step": 970
+    },
+    {
+      "epoch": 1.8656429942418427,
+      "grad_norm": 1.7349492311477661,
+      "learning_rate": 1.2309860111967053e-06,
+      "loss": 2.3854,
+      "step": 972
+    },
+    {
+      "epoch": 1.8694817658349328,
+      "grad_norm": 1.5582528114318848,
+      "learning_rate": 1.1619186812142857e-06,
+      "loss": 2.4444,
+      "step": 974
+    },
+    {
+      "epoch": 1.873320537428023,
+      "grad_norm": 1.6760716438293457,
+      "learning_rate": 1.094822522165101e-06,
+      "loss": 2.5214,
+      "step": 976
+    },
+    {
+      "epoch": 1.8771593090211134,
+      "grad_norm": 1.7502938508987427,
+      "learning_rate": 1.0297002421407797e-06,
+      "loss": 2.4415,
+      "step": 978
+    },
+    {
+      "epoch": 1.8809980806142035,
+      "grad_norm": 1.6573673486709595,
+      "learning_rate": 9.665544695645012e-07,
+      "loss": 2.3996,
+      "step": 980
+    },
+    {
+      "epoch": 1.8848368522072936,
+      "grad_norm": 1.6866267919540405,
+      "learning_rate": 9.053877530849464e-07,
+      "loss": 2.3104,
+      "step": 982
+    },
+    {
+      "epoch": 1.888675623800384,
+      "grad_norm": 1.7685641050338745,
+      "learning_rate": 8.462025614734192e-07,
+      "loss": 2.4496,
+      "step": 984
+    },
+    {
+      "epoch": 1.8925143953934742,
+      "grad_norm": 1.6532257795333862,
+      "learning_rate": 7.890012835242044e-07,
+      "loss": 2.4791,
+      "step": 986
+    },
+    {
+      "epoch": 1.8963531669865643,
+      "grad_norm": 1.67217218875885,
+      "learning_rate": 7.337862279581332e-07,
+      "loss": 2.4586,
+      "step": 988
+    },
+    {
+      "epoch": 1.9001919385796544,
+      "grad_norm": 1.7168689966201782,
+      "learning_rate": 6.805596233294576e-07,
+      "loss": 2.3798,
+      "step": 990
+    },
+    {
+      "epoch": 1.9040307101727447,
+      "grad_norm": 1.7827550172805786,
+      "learning_rate": 6.293236179358175e-07,
+      "loss": 2.485,
+      "step": 992
+    },
+    {
+      "epoch": 1.907869481765835,
+      "grad_norm": 1.5891352891921997,
+      "learning_rate": 5.800802797316152e-07,
+      "loss": 2.1959,
+      "step": 994
+    },
+    {
+      "epoch": 1.9117082533589251,
+      "grad_norm": 1.6550140380859375,
+      "learning_rate": 5.328315962444874e-07,
+      "loss": 2.5058,
+      "step": 996
+    },
+    {
+      "epoch": 1.9155470249520152,
+      "grad_norm": 1.6463745832443237,
+      "learning_rate": 4.875794744951423e-07,
+      "loss": 2.1304,
+      "step": 998
+    },
+    {
+      "epoch": 1.9193857965451055,
+      "grad_norm": 1.6227257251739502,
+      "learning_rate": 4.4432574092032053e-07,
+      "loss": 2.2414,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9232245681381959,
+      "grad_norm": 1.7231581211090088,
+      "learning_rate": 4.0307214129914896e-07,
+      "loss": 2.4267,
+      "step": 1002
+    },
+    {
+      "epoch": 1.927063339731286,
+      "grad_norm": 1.6770871877670288,
+      "learning_rate": 3.6382034068263015e-07,
+      "loss": 2.3622,
+      "step": 1004
+    },
+    {
+      "epoch": 1.930902111324376,
+      "grad_norm": 1.7246891260147095,
+      "learning_rate": 3.265719233264575e-07,
+      "loss": 2.4724,
+      "step": 1006
+    },
+    {
+      "epoch": 1.9347408829174664,
+      "grad_norm": 1.694502830505371,
+      "learning_rate": 2.913283926270771e-07,
+      "loss": 2.4177,
+      "step": 1008
+    },
+    {
+      "epoch": 1.9385796545105567,
+      "grad_norm": 1.7070053815841675,
+      "learning_rate": 2.580911710609923e-07,
+      "loss": 2.5533,
+      "step": 1010
+    },
+    {
+      "epoch": 1.9424184261036468,
+      "grad_norm": 1.6291368007659912,
+      "learning_rate": 2.2686160012735936e-07,
+      "loss": 2.267,
+      "step": 1012
+    },
+    {
+      "epoch": 1.9462571976967369,
+      "grad_norm": 1.5740125179290771,
+      "learning_rate": 1.9764094029385284e-07,
+      "loss": 2.3001,
+      "step": 1014
+    },
+    {
+      "epoch": 1.9500959692898272,
+      "grad_norm": 1.6735721826553345,
+      "learning_rate": 1.704303709457733e-07,
+      "loss": 2.2846,
+      "step": 1016
+    },
+    {
+      "epoch": 1.9539347408829175,
+      "grad_norm": 1.7732954025268555,
+      "learning_rate": 1.452309903384519e-07,
+      "loss": 2.308,
+      "step": 1018
+    },
+    {
+      "epoch": 1.9577735124760078,
+      "grad_norm": 1.748245358467102,
+      "learning_rate": 1.220438155529302e-07,
+      "loss": 2.4971,
+      "step": 1020
+    },
+    {
+      "epoch": 1.961612284069098,
+      "grad_norm": 1.6858606338500977,
+      "learning_rate": 1.0086978245490985e-07,
+      "loss": 2.4105,
+      "step": 1022
+    },
+    {
+      "epoch": 1.965451055662188,
+      "grad_norm": 1.763435959815979,
+      "learning_rate": 8.170974565696599e-08,
+      "loss": 2.4337,
+      "step": 1024
+    },
+    {
+      "epoch": 1.9692898272552783,
+      "grad_norm": 1.6636366844177246,
+      "learning_rate": 6.456447848406932e-08,
+      "loss": 2.4137,
+      "step": 1026
+    },
+    {
+      "epoch": 1.9731285988483687,
+      "grad_norm": 1.641494631767273,
+      "learning_rate": 4.94346729423556e-08,
+      "loss": 2.4322,
+      "step": 1028
+    },
+    {
+      "epoch": 1.9769673704414588,
+      "grad_norm": 1.7012929916381836,
+      "learning_rate": 3.632093969121453e-08,
+      "loss": 2.4342,
+      "step": 1030
+    },
+    {
+      "epoch": 1.9808061420345489,
+      "grad_norm": 1.6317704916000366,
+      "learning_rate": 2.522380801863733e-08,
+      "loss": 2.3635,
+      "step": 1032
+    },
+    {
+      "epoch": 1.9846449136276392,
+      "grad_norm": 1.5718603134155273,
+      "learning_rate": 1.6143725819850463e-08,
+      "loss": 2.1745,
+      "step": 1034
+    },
+    {
+      "epoch": 1.9884836852207295,
+      "grad_norm": 1.7624621391296387,
+      "learning_rate": 9.081059579235662e-09,
+      "loss": 2.3599,
+      "step": 1036
+    },
+    {
+      "epoch": 1.9923224568138196,
+      "grad_norm": 1.759146809577942,
+      "learning_rate": 4.036094355541753e-09,
+      "loss": 2.4143,
+      "step": 1038
+    },
+    {
+      "epoch": 1.9961612284069097,
+      "grad_norm": 1.8646202087402344,
+      "learning_rate": 1.0090337703771991e-09,
+      "loss": 2.4176,
+      "step": 1040
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 2.1013474464416504,
+      "learning_rate": 0.0,
+      "loss": 2.1832,
+      "step": 1042
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 1042,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 521,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.762124413890724e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-1042/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6937747f26bef28460fcdec34b04d43de882bbe87ba71ec6a7684200eddd5c51
+size 5432

checkpoint-521/config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-ary_Arab-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-521/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-521/model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a11665acf61223de9237080f288f7532fadb98f7f4ebce49602c7e456a663226
+size 4950070472

checkpoint-521/model-00002-of-00006.safetensors

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

checkpoint-521/model-00003-of-00006.safetensors

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

checkpoint-521/model-00004-of-00006.safetensors

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

checkpoint-521/model-00005-of-00006.safetensors

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

checkpoint-521/model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc8342358cd1ca8e192fd7121d52184f5636ebf5c02e029df74a6be21861ad65
+size 4614509080

checkpoint-521/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-521/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:32a0172ac69f2ad932e2333a4e72e44287bbdc0a6715a506a89071184435e4e8
+size 13055369358

checkpoint-521/rng_state.pth

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

checkpoint-521/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0783d630f8ce8aef70d72f47e1602b69a36f3f4929db1a03884530798b9b38d7
+size 1064

checkpoint-521/trainer_state.json

@@ -0,0 +1,1853 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 521,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003838771593090211,
+      "grad_norm": 48.289031982421875,
+      "learning_rate": 3.7735849056603773e-06,
+      "loss": 6.4916,
+      "step": 2
+    },
+    {
+      "epoch": 0.007677543186180422,
+      "grad_norm": 29.75079917907715,
+      "learning_rate": 7.547169811320755e-06,
+      "loss": 6.2279,
+      "step": 4
+    },
+    {
+      "epoch": 0.011516314779270634,
+      "grad_norm": 10.823013305664062,
+      "learning_rate": 1.1320754716981132e-05,
+      "loss": 5.5753,
+      "step": 6
+    },
+    {
+      "epoch": 0.015355086372360844,
+      "grad_norm": 7.690505027770996,
+      "learning_rate": 1.509433962264151e-05,
+      "loss": 5.1097,
+      "step": 8
+    },
+    {
+      "epoch": 0.019193857965451054,
+      "grad_norm": 8.298799514770508,
+      "learning_rate": 1.8867924528301888e-05,
+      "loss": 4.9181,
+      "step": 10
+    },
+    {
+      "epoch": 0.023032629558541268,
+      "grad_norm": 5.300838947296143,
+      "learning_rate": 2.2641509433962265e-05,
+      "loss": 4.881,
+      "step": 12
+    },
+    {
+      "epoch": 0.026871401151631478,
+      "grad_norm": 5.326018810272217,
+      "learning_rate": 2.641509433962264e-05,
+      "loss": 4.5184,
+      "step": 14
+    },
+    {
+      "epoch": 0.030710172744721688,
+      "grad_norm": 5.346513271331787,
+      "learning_rate": 3.018867924528302e-05,
+      "loss": 4.5163,
+      "step": 16
+    },
+    {
+      "epoch": 0.0345489443378119,
+      "grad_norm": 3.9342334270477295,
+      "learning_rate": 3.39622641509434e-05,
+      "loss": 4.3626,
+      "step": 18
+    },
+    {
+      "epoch": 0.03838771593090211,
+      "grad_norm": 3.5320026874542236,
+      "learning_rate": 3.7735849056603776e-05,
+      "loss": 4.4012,
+      "step": 20
+    },
+    {
+      "epoch": 0.04222648752399232,
+      "grad_norm": 3.396531581878662,
+      "learning_rate": 4.150943396226415e-05,
+      "loss": 4.2005,
+      "step": 22
+    },
+    {
+      "epoch": 0.046065259117082535,
+      "grad_norm": 2.8504703044891357,
+      "learning_rate": 4.528301886792453e-05,
+      "loss": 4.0846,
+      "step": 24
+    },
+    {
+      "epoch": 0.04990403071017274,
+      "grad_norm": 3.5860769748687744,
+      "learning_rate": 4.9056603773584906e-05,
+      "loss": 4.0599,
+      "step": 26
+    },
+    {
+      "epoch": 0.053742802303262956,
+      "grad_norm": 3.1936397552490234,
+      "learning_rate": 5.283018867924528e-05,
+      "loss": 4.1174,
+      "step": 28
+    },
+    {
+      "epoch": 0.05758157389635317,
+      "grad_norm": 3.2273406982421875,
+      "learning_rate": 5.660377358490566e-05,
+      "loss": 3.7389,
+      "step": 30
+    },
+    {
+      "epoch": 0.061420345489443376,
+      "grad_norm": 2.6602048873901367,
+      "learning_rate": 6.037735849056604e-05,
+      "loss": 4.0485,
+      "step": 32
+    },
+    {
+      "epoch": 0.06525911708253358,
+      "grad_norm": 3.2807395458221436,
+      "learning_rate": 6.415094339622641e-05,
+      "loss": 4.0037,
+      "step": 34
+    },
+    {
+      "epoch": 0.0690978886756238,
+      "grad_norm": 2.3215198516845703,
+      "learning_rate": 6.79245283018868e-05,
+      "loss": 3.9036,
+      "step": 36
+    },
+    {
+      "epoch": 0.07293666026871401,
+      "grad_norm": 2.9224202632904053,
+      "learning_rate": 7.169811320754717e-05,
+      "loss": 3.973,
+      "step": 38
+    },
+    {
+      "epoch": 0.07677543186180422,
+      "grad_norm": 2.602111339569092,
+      "learning_rate": 7.547169811320755e-05,
+      "loss": 3.7376,
+      "step": 40
+    },
+    {
+      "epoch": 0.08061420345489444,
+      "grad_norm": 2.60905122756958,
+      "learning_rate": 7.924528301886794e-05,
+      "loss": 3.8791,
+      "step": 42
+    },
+    {
+      "epoch": 0.08445297504798464,
+      "grad_norm": 3.377249002456665,
+      "learning_rate": 8.30188679245283e-05,
+      "loss": 3.8816,
+      "step": 44
+    },
+    {
+      "epoch": 0.08829174664107485,
+      "grad_norm": 2.953822612762451,
+      "learning_rate": 8.679245283018869e-05,
+      "loss": 3.7942,
+      "step": 46
+    },
+    {
+      "epoch": 0.09213051823416507,
+      "grad_norm": 2.96752667427063,
+      "learning_rate": 9.056603773584906e-05,
+      "loss": 3.5678,
+      "step": 48
+    },
+    {
+      "epoch": 0.09596928982725528,
+      "grad_norm": 3.927953004837036,
+      "learning_rate": 9.433962264150944e-05,
+      "loss": 3.8445,
+      "step": 50
+    },
+    {
+      "epoch": 0.09980806142034548,
+      "grad_norm": 3.235875368118286,
+      "learning_rate": 9.811320754716981e-05,
+      "loss": 3.8813,
+      "step": 52
+    },
+    {
+      "epoch": 0.1036468330134357,
+      "grad_norm": 3.226400375366211,
+      "learning_rate": 9.999974774092106e-05,
+      "loss": 3.5795,
+      "step": 54
+    },
+    {
+      "epoch": 0.10748560460652591,
+      "grad_norm": 3.028534173965454,
+      "learning_rate": 9.999772968356182e-05,
+      "loss": 3.6002,
+      "step": 56
+    },
+    {
+      "epoch": 0.11132437619961612,
+      "grad_norm": 2.740309715270996,
+      "learning_rate": 9.999369365029487e-05,
+      "loss": 3.6699,
+      "step": 58
+    },
+    {
+      "epoch": 0.11516314779270634,
+      "grad_norm": 3.1072158813476562,
+      "learning_rate": 9.998763980401998e-05,
+      "loss": 3.6008,
+      "step": 60
+    },
+    {
+      "epoch": 0.11900191938579655,
+      "grad_norm": 2.9284427165985107,
+      "learning_rate": 9.997956838907852e-05,
+      "loss": 3.6526,
+      "step": 62
+    },
+    {
+      "epoch": 0.12284069097888675,
+      "grad_norm": 3.230877161026001,
+      "learning_rate": 9.996947973124372e-05,
+      "loss": 3.7293,
+      "step": 64
+    },
+    {
+      "epoch": 0.12667946257197696,
+      "grad_norm": 3.2215468883514404,
+      "learning_rate": 9.995737423770745e-05,
+      "loss": 3.4869,
+      "step": 66
+    },
+    {
+      "epoch": 0.13051823416506717,
+      "grad_norm": 2.768615245819092,
+      "learning_rate": 9.994325239706377e-05,
+      "loss": 3.513,
+      "step": 68
+    },
+    {
+      "epoch": 0.1343570057581574,
+      "grad_norm": 2.9997639656066895,
+      "learning_rate": 9.992711477928926e-05,
+      "loss": 3.4237,
+      "step": 70
+    },
+    {
+      "epoch": 0.1381957773512476,
+      "grad_norm": 3.5303409099578857,
+      "learning_rate": 9.990896203571995e-05,
+      "loss": 3.6657,
+      "step": 72
+    },
+    {
+      "epoch": 0.1420345489443378,
+      "grad_norm": 2.925368309020996,
+      "learning_rate": 9.988879489902509e-05,
+      "loss": 3.4966,
+      "step": 74
+    },
+    {
+      "epoch": 0.14587332053742802,
+      "grad_norm": 3.1105096340179443,
+      "learning_rate": 9.986661418317759e-05,
+      "loss": 3.5241,
+      "step": 76
+    },
+    {
+      "epoch": 0.14971209213051823,
+      "grad_norm": 3.1198081970214844,
+      "learning_rate": 9.984242078342107e-05,
+      "loss": 3.6021,
+      "step": 78
+    },
+    {
+      "epoch": 0.15355086372360843,
+      "grad_norm": 2.56496000289917,
+      "learning_rate": 9.981621567623385e-05,
+      "loss": 3.4452,
+      "step": 80
+    },
+    {
+      "epoch": 0.15738963531669867,
+      "grad_norm": 2.585533380508423,
+      "learning_rate": 9.978799991928944e-05,
+      "loss": 3.5144,
+      "step": 82
+    },
+    {
+      "epoch": 0.16122840690978887,
+      "grad_norm": 2.6131768226623535,
+      "learning_rate": 9.97577746514139e-05,
+      "loss": 3.49,
+      "step": 84
+    },
+    {
+      "epoch": 0.16506717850287908,
+      "grad_norm": 3.118385076522827,
+      "learning_rate": 9.972554109253989e-05,
+      "loss": 3.5082,
+      "step": 86
+    },
+    {
+      "epoch": 0.1689059500959693,
+      "grad_norm": 2.592630624771118,
+      "learning_rate": 9.969130054365736e-05,
+      "loss": 3.3943,
+      "step": 88
+    },
+    {
+      "epoch": 0.1727447216890595,
+      "grad_norm": 3.1140449047088623,
+      "learning_rate": 9.965505438676115e-05,
+      "loss": 3.4553,
+      "step": 90
+    },
+    {
+      "epoch": 0.1765834932821497,
+      "grad_norm": 2.3941497802734375,
+      "learning_rate": 9.961680408479507e-05,
+      "loss": 3.2766,
+      "step": 92
+    },
+    {
+      "epoch": 0.18042226487523993,
+      "grad_norm": 2.8205137252807617,
+      "learning_rate": 9.957655118159304e-05,
+      "loss": 3.3713,
+      "step": 94
+    },
+    {
+      "epoch": 0.18426103646833014,
+      "grad_norm": 2.3888957500457764,
+      "learning_rate": 9.953429730181653e-05,
+      "loss": 3.3228,
+      "step": 96
+    },
+    {
+      "epoch": 0.18809980806142035,
+      "grad_norm": 2.363973617553711,
+      "learning_rate": 9.949004415088927e-05,
+      "loss": 3.3273,
+      "step": 98
+    },
+    {
+      "epoch": 0.19193857965451055,
+      "grad_norm": 2.508699655532837,
+      "learning_rate": 9.944379351492818e-05,
+      "loss": 3.2716,
+      "step": 100
+    },
+    {
+      "epoch": 0.19577735124760076,
+      "grad_norm": 2.3598968982696533,
+      "learning_rate": 9.939554726067141e-05,
+      "loss": 3.4032,
+      "step": 102
+    },
+    {
+      "epoch": 0.19961612284069097,
+      "grad_norm": 2.8131492137908936,
+      "learning_rate": 9.934530733540294e-05,
+      "loss": 3.3651,
+      "step": 104
+    },
+    {
+      "epoch": 0.2034548944337812,
+      "grad_norm": 2.7955825328826904,
+      "learning_rate": 9.929307576687403e-05,
+      "loss": 3.3716,
+      "step": 106
+    },
+    {
+      "epoch": 0.2072936660268714,
+      "grad_norm": 2.7270803451538086,
+      "learning_rate": 9.923885466322135e-05,
+      "loss": 3.451,
+      "step": 108
+    },
+    {
+      "epoch": 0.21113243761996162,
+      "grad_norm": 3.13055419921875,
+      "learning_rate": 9.918264621288187e-05,
+      "loss": 3.3958,
+      "step": 110
+    },
+    {
+      "epoch": 0.21497120921305182,
+      "grad_norm": 2.994201898574829,
+      "learning_rate": 9.912445268450458e-05,
+      "loss": 3.3517,
+      "step": 112
+    },
+    {
+      "epoch": 0.21880998080614203,
+      "grad_norm": 2.897017002105713,
+      "learning_rate": 9.906427642685889e-05,
+      "loss": 3.3386,
+      "step": 114
+    },
+    {
+      "epoch": 0.22264875239923224,
+      "grad_norm": 2.668201208114624,
+      "learning_rate": 9.900211986873985e-05,
+      "loss": 3.3498,
+      "step": 116
+    },
+    {
+      "epoch": 0.22648752399232247,
+      "grad_norm": 2.353947877883911,
+      "learning_rate": 9.89379855188701e-05,
+      "loss": 3.2568,
+      "step": 118
+    },
+    {
+      "epoch": 0.23032629558541268,
+      "grad_norm": 2.7698841094970703,
+      "learning_rate": 9.887187596579866e-05,
+      "loss": 3.2583,
+      "step": 120
+    },
+    {
+      "epoch": 0.23416506717850288,
+      "grad_norm": 2.742597818374634,
+      "learning_rate": 9.880379387779637e-05,
+      "loss": 3.1913,
+      "step": 122
+    },
+    {
+      "epoch": 0.2380038387715931,
+      "grad_norm": 2.6558752059936523,
+      "learning_rate": 9.873374200274826e-05,
+      "loss": 3.262,
+      "step": 124
+    },
+    {
+      "epoch": 0.2418426103646833,
+      "grad_norm": 2.544968843460083,
+      "learning_rate": 9.866172316804266e-05,
+      "loss": 3.3072,
+      "step": 126
+    },
+    {
+      "epoch": 0.2456813819577735,
+      "grad_norm": 2.6007907390594482,
+      "learning_rate": 9.8587740280457e-05,
+      "loss": 3.135,
+      "step": 128
+    },
+    {
+      "epoch": 0.2495201535508637,
+      "grad_norm": 2.7021710872650146,
+      "learning_rate": 9.851179632604057e-05,
+      "loss": 3.2714,
+      "step": 130
+    },
+    {
+      "epoch": 0.2533589251439539,
+      "grad_norm": 2.698706865310669,
+      "learning_rate": 9.843389436999396e-05,
+      "loss": 3.2385,
+      "step": 132
+    },
+    {
+      "epoch": 0.2571976967370441,
+      "grad_norm": 2.9197003841400146,
+      "learning_rate": 9.835403755654536e-05,
+      "loss": 3.2911,
+      "step": 134
+    },
+    {
+      "epoch": 0.26103646833013433,
+      "grad_norm": 2.5004656314849854,
+      "learning_rate": 9.827222910882358e-05,
+      "loss": 3.3533,
+      "step": 136
+    },
+    {
+      "epoch": 0.2648752399232246,
+      "grad_norm": 2.7729904651641846,
+      "learning_rate": 9.818847232872815e-05,
+      "loss": 3.2461,
+      "step": 138
+    },
+    {
+      "epoch": 0.2687140115163148,
+      "grad_norm": 2.4783594608306885,
+      "learning_rate": 9.81027705967958e-05,
+      "loss": 3.2066,
+      "step": 140
+    },
+    {
+      "epoch": 0.272552783109405,
+      "grad_norm": 2.5870554447174072,
+      "learning_rate": 9.801512737206422e-05,
+      "loss": 3.3161,
+      "step": 142
+    },
+    {
+      "epoch": 0.2763915547024952,
+      "grad_norm": 2.610823154449463,
+      "learning_rate": 9.792554619193235e-05,
+      "loss": 3.1034,
+      "step": 144
+    },
+    {
+      "epoch": 0.2802303262955854,
+      "grad_norm": 2.5675435066223145,
+      "learning_rate": 9.783403067201762e-05,
+      "loss": 3.0721,
+      "step": 146
+    },
+    {
+      "epoch": 0.2840690978886756,
+      "grad_norm": 2.435715913772583,
+      "learning_rate": 9.774058450601002e-05,
+      "loss": 3.2662,
+      "step": 148
+    },
+    {
+      "epoch": 0.28790786948176583,
+      "grad_norm": 2.705893039703369,
+      "learning_rate": 9.764521146552309e-05,
+      "loss": 3.2292,
+      "step": 150
+    },
+    {
+      "epoch": 0.29174664107485604,
+      "grad_norm": 2.7473392486572266,
+      "learning_rate": 9.754791539994152e-05,
+      "loss": 3.1772,
+      "step": 152
+    },
+    {
+      "epoch": 0.29558541266794625,
+      "grad_norm": 2.6607396602630615,
+      "learning_rate": 9.744870023626599e-05,
+      "loss": 3.1006,
+      "step": 154
+    },
+    {
+      "epoch": 0.29942418426103645,
+      "grad_norm": 2.3230140209198,
+      "learning_rate": 9.734756997895448e-05,
+      "loss": 3.1766,
+      "step": 156
+    },
+    {
+      "epoch": 0.30326295585412666,
+      "grad_norm": 2.8475935459136963,
+      "learning_rate": 9.724452870976085e-05,
+      "loss": 3.2234,
+      "step": 158
+    },
+    {
+      "epoch": 0.30710172744721687,
+      "grad_norm": 2.8263561725616455,
+      "learning_rate": 9.713958058756985e-05,
+      "loss": 2.9765,
+      "step": 160
+    },
+    {
+      "epoch": 0.31094049904030713,
+      "grad_norm": 2.7287673950195312,
+      "learning_rate": 9.703272984822946e-05,
+      "loss": 3.2325,
+      "step": 162
+    },
+    {
+      "epoch": 0.31477927063339733,
+      "grad_norm": 2.7081940174102783,
+      "learning_rate": 9.69239808043799e-05,
+      "loss": 3.1034,
+      "step": 164
+    },
+    {
+      "epoch": 0.31861804222648754,
+      "grad_norm": 2.571908473968506,
+      "learning_rate": 9.681333784527944e-05,
+      "loss": 3.1813,
+      "step": 166
+    },
+    {
+      "epoch": 0.32245681381957775,
+      "grad_norm": 2.3741984367370605,
+      "learning_rate": 9.67008054366274e-05,
+      "loss": 3.1286,
+      "step": 168
+    },
+    {
+      "epoch": 0.32629558541266795,
+      "grad_norm": 2.4630539417266846,
+      "learning_rate": 9.658638812038379e-05,
+      "loss": 3.2526,
+      "step": 170
+    },
+    {
+      "epoch": 0.33013435700575816,
+      "grad_norm": 2.2894673347473145,
+      "learning_rate": 9.647009051458604e-05,
+      "loss": 3.0942,
+      "step": 172
+    },
+    {
+      "epoch": 0.33397312859884837,
+      "grad_norm": 2.4378833770751953,
+      "learning_rate": 9.635191731316262e-05,
+      "loss": 3.0711,
+      "step": 174
+    },
+    {
+      "epoch": 0.3378119001919386,
+      "grad_norm": 2.3150041103363037,
+      "learning_rate": 9.623187328574357e-05,
+      "loss": 3.1246,
+      "step": 176
+    },
+    {
+      "epoch": 0.3416506717850288,
+      "grad_norm": 2.4483959674835205,
+      "learning_rate": 9.6109963277468e-05,
+      "loss": 3.1125,
+      "step": 178
+    },
+    {
+      "epoch": 0.345489443378119,
+      "grad_norm": 2.5114800930023193,
+      "learning_rate": 9.598619220878853e-05,
+      "loss": 3.0962,
+      "step": 180
+    },
+    {
+      "epoch": 0.3493282149712092,
+      "grad_norm": 2.2731025218963623,
+      "learning_rate": 9.586056507527266e-05,
+      "loss": 3.0675,
+      "step": 182
+    },
+    {
+      "epoch": 0.3531669865642994,
+      "grad_norm": 2.7290143966674805,
+      "learning_rate": 9.573308694740121e-05,
+      "loss": 3.1265,
+      "step": 184
+    },
+    {
+      "epoch": 0.3570057581573896,
+      "grad_norm": 2.613831043243408,
+      "learning_rate": 9.560376297036362e-05,
+      "loss": 3.1269,
+      "step": 186
+    },
+    {
+      "epoch": 0.36084452975047987,
+      "grad_norm": 2.2489418983459473,
+      "learning_rate": 9.547259836385031e-05,
+      "loss": 3.0518,
+      "step": 188
+    },
+    {
+      "epoch": 0.3646833013435701,
+      "grad_norm": 2.547145366668701,
+      "learning_rate": 9.533959842184195e-05,
+      "loss": 3.0586,
+      "step": 190
+    },
+    {
+      "epoch": 0.3685220729366603,
+      "grad_norm": 2.385820150375366,
+      "learning_rate": 9.520476851239588e-05,
+      "loss": 3.182,
+      "step": 192
+    },
+    {
+      "epoch": 0.3723608445297505,
+      "grad_norm": 2.3586604595184326,
+      "learning_rate": 9.506811407742937e-05,
+      "loss": 3.1526,
+      "step": 194
+    },
+    {
+      "epoch": 0.3761996161228407,
+      "grad_norm": 2.4252877235412598,
+      "learning_rate": 9.492964063250001e-05,
+      "loss": 3.0067,
+      "step": 196
+    },
+    {
+      "epoch": 0.3800383877159309,
+      "grad_norm": 2.4425666332244873,
+      "learning_rate": 9.478935376658307e-05,
+      "loss": 3.0842,
+      "step": 198
+    },
+    {
+      "epoch": 0.3838771593090211,
+      "grad_norm": 2.5915610790252686,
+      "learning_rate": 9.4647259141846e-05,
+      "loss": 3.0993,
+      "step": 200
+    },
+    {
+      "epoch": 0.3877159309021113,
+      "grad_norm": 2.3467631340026855,
+      "learning_rate": 9.450336249341976e-05,
+      "loss": 3.1839,
+      "step": 202
+    },
+    {
+      "epoch": 0.3915547024952015,
+      "grad_norm": 2.370368719100952,
+      "learning_rate": 9.435766962916748e-05,
+      "loss": 3.1194,
+      "step": 204
+    },
+    {
+      "epoch": 0.39539347408829173,
+      "grad_norm": 2.5787832736968994,
+      "learning_rate": 9.421018642944997e-05,
+      "loss": 3.1111,
+      "step": 206
+    },
+    {
+      "epoch": 0.39923224568138194,
+      "grad_norm": 2.549276351928711,
+      "learning_rate": 9.406091884688836e-05,
+      "loss": 3.0373,
+      "step": 208
+    },
+    {
+      "epoch": 0.40307101727447214,
+      "grad_norm": 2.4557924270629883,
+      "learning_rate": 9.390987290612395e-05,
+      "loss": 3.0727,
+      "step": 210
+    },
+    {
+      "epoch": 0.4069097888675624,
+      "grad_norm": 2.4945881366729736,
+      "learning_rate": 9.375705470357493e-05,
+      "loss": 3.0119,
+      "step": 212
+    },
+    {
+      "epoch": 0.4107485604606526,
+      "grad_norm": 2.5190112590789795,
+      "learning_rate": 9.360247040719039e-05,
+      "loss": 3.0296,
+      "step": 214
+    },
+    {
+      "epoch": 0.4145873320537428,
+      "grad_norm": 2.3181045055389404,
+      "learning_rate": 9.344612625620132e-05,
+      "loss": 3.0463,
+      "step": 216
+    },
+    {
+      "epoch": 0.418426103646833,
+      "grad_norm": 2.344257116317749,
+      "learning_rate": 9.32880285608689e-05,
+      "loss": 3.0091,
+      "step": 218
+    },
+    {
+      "epoch": 0.42226487523992323,
+      "grad_norm": 2.0085527896881104,
+      "learning_rate": 9.312818370222962e-05,
+      "loss": 2.9649,
+      "step": 220
+    },
+    {
+      "epoch": 0.42610364683301344,
+      "grad_norm": 2.92118763923645,
+      "learning_rate": 9.296659813183794e-05,
+      "loss": 3.0637,
+      "step": 222
+    },
+    {
+      "epoch": 0.42994241842610365,
+      "grad_norm": 2.6241536140441895,
+      "learning_rate": 9.280327837150571e-05,
+      "loss": 3.0458,
+      "step": 224
+    },
+    {
+      "epoch": 0.43378119001919385,
+      "grad_norm": 2.025529146194458,
+      "learning_rate": 9.26382310130391e-05,
+      "loss": 3.0004,
+      "step": 226
+    },
+    {
+      "epoch": 0.43761996161228406,
+      "grad_norm": 2.682751178741455,
+      "learning_rate": 9.247146271797245e-05,
+      "loss": 3.0602,
+      "step": 228
+    },
+    {
+      "epoch": 0.44145873320537427,
+      "grad_norm": 2.2250850200653076,
+      "learning_rate": 9.23029802172994e-05,
+      "loss": 2.9436,
+      "step": 230
+    },
+    {
+      "epoch": 0.44529750479846447,
+      "grad_norm": 2.490293264389038,
+      "learning_rate": 9.213279031120129e-05,
+      "loss": 2.7951,
+      "step": 232
+    },
+    {
+      "epoch": 0.4491362763915547,
+      "grad_norm": 2.5657706260681152,
+      "learning_rate": 9.19608998687726e-05,
+      "loss": 3.1099,
+      "step": 234
+    },
+    {
+      "epoch": 0.45297504798464494,
+      "grad_norm": 2.199953079223633,
+      "learning_rate": 9.178731582774381e-05,
+      "loss": 2.9906,
+      "step": 236
+    },
+    {
+      "epoch": 0.45681381957773515,
+      "grad_norm": 2.581700086593628,
+      "learning_rate": 9.161204519420126e-05,
+      "loss": 2.9742,
+      "step": 238
+    },
+    {
+      "epoch": 0.46065259117082535,
+      "grad_norm": 2.086547374725342,
+      "learning_rate": 9.143509504230449e-05,
+      "loss": 3.0655,
+      "step": 240
+    },
+    {
+      "epoch": 0.46449136276391556,
+      "grad_norm": 2.3075525760650635,
+      "learning_rate": 9.125647251400067e-05,
+      "loss": 2.9537,
+      "step": 242
+    },
+    {
+      "epoch": 0.46833013435700577,
+      "grad_norm": 2.2352142333984375,
+      "learning_rate": 9.107618481873632e-05,
+      "loss": 2.9135,
+      "step": 244
+    },
+    {
+      "epoch": 0.472168905950096,
+      "grad_norm": 2.1639482975006104,
+      "learning_rate": 9.089423923316636e-05,
+      "loss": 3.0237,
+      "step": 246
+    },
+    {
+      "epoch": 0.4760076775431862,
+      "grad_norm": 2.307377576828003,
+      "learning_rate": 9.07106431008604e-05,
+      "loss": 2.9091,
+      "step": 248
+    },
+    {
+      "epoch": 0.4798464491362764,
+      "grad_norm": 2.010719060897827,
+      "learning_rate": 9.052540383200633e-05,
+      "loss": 2.8782,
+      "step": 250
+    },
+    {
+      "epoch": 0.4836852207293666,
+      "grad_norm": 2.6130549907684326,
+      "learning_rate": 9.033852890311128e-05,
+      "loss": 3.0429,
+      "step": 252
+    },
+    {
+      "epoch": 0.4875239923224568,
+      "grad_norm": 2.3005809783935547,
+      "learning_rate": 9.01500258566998e-05,
+      "loss": 2.9648,
+      "step": 254
+    },
+    {
+      "epoch": 0.491362763915547,
+      "grad_norm": 2.362955331802368,
+      "learning_rate": 8.995990230100949e-05,
+      "loss": 2.908,
+      "step": 256
+    },
+    {
+      "epoch": 0.4952015355086372,
+      "grad_norm": 2.1333205699920654,
+      "learning_rate": 8.976816590968387e-05,
+      "loss": 3.0275,
+      "step": 258
+    },
+    {
+      "epoch": 0.4990403071017274,
+      "grad_norm": 2.328183650970459,
+      "learning_rate": 8.957482442146272e-05,
+      "loss": 2.9645,
+      "step": 260
+    },
+    {
+      "epoch": 0.5028790786948176,
+      "grad_norm": 2.229396343231201,
+      "learning_rate": 8.937988563986963e-05,
+      "loss": 2.8474,
+      "step": 262
+    },
+    {
+      "epoch": 0.5067178502879078,
+      "grad_norm": 2.283524513244629,
+      "learning_rate": 8.918335743289718e-05,
+      "loss": 3.0276,
+      "step": 264
+    },
+    {
+      "epoch": 0.510556621880998,
+      "grad_norm": 2.0872859954833984,
+      "learning_rate": 8.898524773268926e-05,
+      "loss": 3.0058,
+      "step": 266
+    },
+    {
+      "epoch": 0.5143953934740882,
+      "grad_norm": 2.427462100982666,
+      "learning_rate": 8.8785564535221e-05,
+      "loss": 2.971,
+      "step": 268
+    },
+    {
+      "epoch": 0.5182341650671785,
+      "grad_norm": 2.280747413635254,
+      "learning_rate": 8.858431589997597e-05,
+      "loss": 2.9367,
+      "step": 270
+    },
+    {
+      "epoch": 0.5220729366602687,
+      "grad_norm": 2.2372734546661377,
+      "learning_rate": 8.838150994962093e-05,
+      "loss": 3.0049,
+      "step": 272
+    },
+    {
+      "epoch": 0.525911708253359,
+      "grad_norm": 2.379145622253418,
+      "learning_rate": 8.817715486967803e-05,
+      "loss": 3.0165,
+      "step": 274
+    },
+    {
+      "epoch": 0.5297504798464492,
+      "grad_norm": 2.5526132583618164,
+      "learning_rate": 8.797125890819428e-05,
+      "loss": 2.9192,
+      "step": 276
+    },
+    {
+      "epoch": 0.5335892514395394,
+      "grad_norm": 1.9904067516326904,
+      "learning_rate": 8.776383037540888e-05,
+      "loss": 2.9803,
+      "step": 278
+    },
+    {
+      "epoch": 0.5374280230326296,
+      "grad_norm": 1.8862968683242798,
+      "learning_rate": 8.755487764341757e-05,
+      "loss": 2.8934,
+      "step": 280
+    },
+    {
+      "epoch": 0.5412667946257198,
+      "grad_norm": 2.738217830657959,
+      "learning_rate": 8.734440914583486e-05,
+      "loss": 2.9304,
+      "step": 282
+    },
+    {
+      "epoch": 0.54510556621881,
+      "grad_norm": 2.059157609939575,
+      "learning_rate": 8.713243337745365e-05,
+      "loss": 2.9255,
+      "step": 284
+    },
+    {
+      "epoch": 0.5489443378119002,
+      "grad_norm": 2.4487457275390625,
+      "learning_rate": 8.691895889390227e-05,
+      "loss": 2.906,
+      "step": 286
+    },
+    {
+      "epoch": 0.5527831094049904,
+      "grad_norm": 1.9257361888885498,
+      "learning_rate": 8.670399431129926e-05,
+      "loss": 2.7553,
+      "step": 288
+    },
+    {
+      "epoch": 0.5566218809980806,
+      "grad_norm": 2.365708589553833,
+      "learning_rate": 8.648754830590552e-05,
+      "loss": 3.0451,
+      "step": 290
+    },
+    {
+      "epoch": 0.5604606525911708,
+      "grad_norm": 2.44295072555542,
+      "learning_rate": 8.626962961377423e-05,
+      "loss": 2.9814,
+      "step": 292
+    },
+    {
+      "epoch": 0.564299424184261,
+      "grad_norm": 2.14540433883667,
+      "learning_rate": 8.605024703039816e-05,
+      "loss": 2.9126,
+      "step": 294
+    },
+    {
+      "epoch": 0.5681381957773513,
+      "grad_norm": 2.1933960914611816,
+      "learning_rate": 8.582940941035476e-05,
+      "loss": 2.9439,
+      "step": 296
+    },
+    {
+      "epoch": 0.5719769673704415,
+      "grad_norm": 2.231813907623291,
+      "learning_rate": 8.560712566694862e-05,
+      "loss": 3.0361,
+      "step": 298
+    },
+    {
+      "epoch": 0.5758157389635317,
+      "grad_norm": 2.183107852935791,
+      "learning_rate": 8.538340477185192e-05,
+      "loss": 2.9277,
+      "step": 300
+    },
+    {
+      "epoch": 0.5796545105566219,
+      "grad_norm": 2.7278597354888916,
+      "learning_rate": 8.515825575474221e-05,
+      "loss": 2.9476,
+      "step": 302
+    },
+    {
+      "epoch": 0.5834932821497121,
+      "grad_norm": 2.6327316761016846,
+      "learning_rate": 8.493168770293792e-05,
+      "loss": 2.961,
+      "step": 304
+    },
+    {
+      "epoch": 0.5873320537428023,
+      "grad_norm": 2.3474512100219727,
+      "learning_rate": 8.47037097610317e-05,
+      "loss": 2.9233,
+      "step": 306
+    },
+    {
+      "epoch": 0.5911708253358925,
+      "grad_norm": 2.31101131439209,
+      "learning_rate": 8.447433113052123e-05,
+      "loss": 2.9289,
+      "step": 308
+    },
+    {
+      "epoch": 0.5950095969289827,
+      "grad_norm": 2.4171314239501953,
+      "learning_rate": 8.42435610694379e-05,
+      "loss": 2.9037,
+      "step": 310
+    },
+    {
+      "epoch": 0.5988483685220729,
+      "grad_norm": 2.2114620208740234,
+      "learning_rate": 8.401140889197305e-05,
+      "loss": 2.8163,
+      "step": 312
+    },
+    {
+      "epoch": 0.6026871401151631,
+      "grad_norm": 2.3434977531433105,
+      "learning_rate": 8.377788396810223e-05,
+      "loss": 2.8856,
+      "step": 314
+    },
+    {
+      "epoch": 0.6065259117082533,
+      "grad_norm": 2.4382877349853516,
+      "learning_rate": 8.354299572320679e-05,
+      "loss": 2.8265,
+      "step": 316
+    },
+    {
+      "epoch": 0.6103646833013435,
+      "grad_norm": 1.9080297946929932,
+      "learning_rate": 8.330675363769356e-05,
+      "loss": 2.9292,
+      "step": 318
+    },
+    {
+      "epoch": 0.6142034548944337,
+      "grad_norm": 2.439094066619873,
+      "learning_rate": 8.306916724661224e-05,
+      "loss": 2.8653,
+      "step": 320
+    },
+    {
+      "epoch": 0.6180422264875239,
+      "grad_norm": 2.5351178646087646,
+      "learning_rate": 8.283024613927054e-05,
+      "loss": 2.8837,
+      "step": 322
+    },
+    {
+      "epoch": 0.6218809980806143,
+      "grad_norm": 2.589860439300537,
+      "learning_rate": 8.258999995884705e-05,
+      "loss": 2.8083,
+      "step": 324
+    },
+    {
+      "epoch": 0.6257197696737045,
+      "grad_norm": 2.439771890640259,
+      "learning_rate": 8.234843840200217e-05,
+      "loss": 2.8731,
+      "step": 326
+    },
+    {
+      "epoch": 0.6295585412667947,
+      "grad_norm": 2.1145169734954834,
+      "learning_rate": 8.210557121848663e-05,
+      "loss": 2.8095,
+      "step": 328
+    },
+    {
+      "epoch": 0.6333973128598849,
+      "grad_norm": 2.5235095024108887,
+      "learning_rate": 8.186140821074801e-05,
+      "loss": 3.0236,
+      "step": 330
+    },
+    {
+      "epoch": 0.6372360844529751,
+      "grad_norm": 2.062345027923584,
+      "learning_rate": 8.161595923353516e-05,
+      "loss": 2.8697,
+      "step": 332
+    },
+    {
+      "epoch": 0.6410748560460653,
+      "grad_norm": 2.3722496032714844,
+      "learning_rate": 8.136923419350031e-05,
+      "loss": 2.8812,
+      "step": 334
+    },
+    {
+      "epoch": 0.6449136276391555,
+      "grad_norm": 2.054664134979248,
+      "learning_rate": 8.112124304879937e-05,
+      "loss": 2.8569,
+      "step": 336
+    },
+    {
+      "epoch": 0.6487523992322457,
+      "grad_norm": 2.16813325881958,
+      "learning_rate": 8.087199580868996e-05,
+      "loss": 2.874,
+      "step": 338
+    },
+    {
+      "epoch": 0.6525911708253359,
+      "grad_norm": 2.463815689086914,
+      "learning_rate": 8.062150253312735e-05,
+      "loss": 2.9572,
+      "step": 340
+    },
+    {
+      "epoch": 0.6564299424184261,
+      "grad_norm": 2.1177213191986084,
+      "learning_rate": 8.03697733323585e-05,
+      "loss": 2.9028,
+      "step": 342
+    },
+    {
+      "epoch": 0.6602687140115163,
+      "grad_norm": 2.3246190547943115,
+      "learning_rate": 8.011681836651402e-05,
+      "loss": 2.8694,
+      "step": 344
+    },
+    {
+      "epoch": 0.6641074856046065,
+      "grad_norm": 2.157296895980835,
+      "learning_rate": 7.986264784519801e-05,
+      "loss": 2.8129,
+      "step": 346
+    },
+    {
+      "epoch": 0.6679462571976967,
+      "grad_norm": 2.1130011081695557,
+      "learning_rate": 7.960727202707605e-05,
+      "loss": 2.8877,
+      "step": 348
+    },
+    {
+      "epoch": 0.6717850287907869,
+      "grad_norm": 2.256319046020508,
+      "learning_rate": 7.935070121946115e-05,
+      "loss": 2.8947,
+      "step": 350
+    },
+    {
+      "epoch": 0.6756238003838771,
+      "grad_norm": 2.183379650115967,
+      "learning_rate": 7.909294577789766e-05,
+      "loss": 2.8678,
+      "step": 352
+    },
+    {
+      "epoch": 0.6794625719769674,
+      "grad_norm": 2.4424221515655518,
+      "learning_rate": 7.883401610574336e-05,
+      "loss": 2.9269,
+      "step": 354
+    },
+    {
+      "epoch": 0.6833013435700576,
+      "grad_norm": 1.933434247970581,
+      "learning_rate": 7.857392265374963e-05,
+      "loss": 2.8906,
+      "step": 356
+    },
+    {
+      "epoch": 0.6871401151631478,
+      "grad_norm": 2.322988271713257,
+      "learning_rate": 7.83126759196395e-05,
+      "loss": 2.8756,
+      "step": 358
+    },
+    {
+      "epoch": 0.690978886756238,
+      "grad_norm": 2.2925362586975098,
+      "learning_rate": 7.805028644768407e-05,
+      "loss": 2.8704,
+      "step": 360
+    },
+    {
+      "epoch": 0.6948176583493282,
+      "grad_norm": 2.209590196609497,
+      "learning_rate": 7.778676482827686e-05,
+      "loss": 2.8794,
+      "step": 362
+    },
+    {
+      "epoch": 0.6986564299424184,
+      "grad_norm": 1.9942725896835327,
+      "learning_rate": 7.752212169750641e-05,
+      "loss": 2.8943,
+      "step": 364
+    },
+    {
+      "epoch": 0.7024952015355086,
+      "grad_norm": 2.1021668910980225,
+      "learning_rate": 7.725636773672693e-05,
+      "loss": 2.8838,
+      "step": 366
+    },
+    {
+      "epoch": 0.7063339731285988,
+      "grad_norm": 2.3661866188049316,
+      "learning_rate": 7.69895136721273e-05,
+      "loss": 2.7449,
+      "step": 368
+    },
+    {
+      "epoch": 0.710172744721689,
+      "grad_norm": 2.2460644245147705,
+      "learning_rate": 7.672157027429803e-05,
+      "loss": 2.8648,
+      "step": 370
+    },
+    {
+      "epoch": 0.7140115163147792,
+      "grad_norm": 2.0501885414123535,
+      "learning_rate": 7.645254835779657e-05,
+      "loss": 2.9297,
+      "step": 372
+    },
+    {
+      "epoch": 0.7178502879078695,
+      "grad_norm": 2.259514570236206,
+      "learning_rate": 7.618245878071091e-05,
+      "loss": 2.9449,
+      "step": 374
+    },
+    {
+      "epoch": 0.7216890595009597,
+      "grad_norm": 2.193086862564087,
+      "learning_rate": 7.591131244422118e-05,
+      "loss": 2.8749,
+      "step": 376
+    },
+    {
+      "epoch": 0.72552783109405,
+      "grad_norm": 2.0527806282043457,
+      "learning_rate": 7.563912029215983e-05,
+      "loss": 2.7844,
+      "step": 378
+    },
+    {
+      "epoch": 0.7293666026871402,
+      "grad_norm": 1.9866085052490234,
+      "learning_rate": 7.536589331056976e-05,
+      "loss": 2.9124,
+      "step": 380
+    },
+    {
+      "epoch": 0.7332053742802304,
+      "grad_norm": 2.272838830947876,
+      "learning_rate": 7.509164252726107e-05,
+      "loss": 2.9241,
+      "step": 382
+    },
+    {
+      "epoch": 0.7370441458733206,
+      "grad_norm": 2.093259334564209,
+      "learning_rate": 7.481637901136577e-05,
+      "loss": 2.8545,
+      "step": 384
+    },
+    {
+      "epoch": 0.7408829174664108,
+      "grad_norm": 1.9920848608016968,
+      "learning_rate": 7.454011387289127e-05,
+      "loss": 2.8148,
+      "step": 386
+    },
+    {
+      "epoch": 0.744721689059501,
+      "grad_norm": 7.795645236968994,
+      "learning_rate": 7.42628582622717e-05,
+      "loss": 2.7107,
+      "step": 388
+    },
+    {
+      "epoch": 0.7485604606525912,
+      "grad_norm": 2.090559482574463,
+      "learning_rate": 7.398462336991801e-05,
+      "loss": 2.7163,
+      "step": 390
+    },
+    {
+      "epoch": 0.7523992322456814,
+      "grad_norm": 2.06302809715271,
+      "learning_rate": 7.370542042576634e-05,
+      "loss": 2.7883,
+      "step": 392
+    },
+    {
+      "epoch": 0.7562380038387716,
+      "grad_norm": 2.325312376022339,
+      "learning_rate": 7.342526069882465e-05,
+      "loss": 2.9318,
+      "step": 394
+    },
+    {
+      "epoch": 0.7600767754318618,
+      "grad_norm": 2.163017988204956,
+      "learning_rate": 7.314415549671795e-05,
+      "loss": 2.877,
+      "step": 396
+    },
+    {
+      "epoch": 0.763915547024952,
+      "grad_norm": 2.363990545272827,
+      "learning_rate": 7.286211616523193e-05,
+      "loss": 2.9007,
+      "step": 398
+    },
+    {
+      "epoch": 0.7677543186180422,
+      "grad_norm": 2.189789295196533,
+      "learning_rate": 7.257915408785498e-05,
+      "loss": 2.9451,
+      "step": 400
+    },
+    {
+      "epoch": 0.7715930902111324,
+      "grad_norm": 2.284350872039795,
+      "learning_rate": 7.229528068531882e-05,
+      "loss": 2.8626,
+      "step": 402
+    },
+    {
+      "epoch": 0.7754318618042226,
+      "grad_norm": 2.0571820735931396,
+      "learning_rate": 7.201050741513736e-05,
+      "loss": 2.6964,
+      "step": 404
+    },
+    {
+      "epoch": 0.7792706333973128,
+      "grad_norm": 1.95001220703125,
+      "learning_rate": 7.17248457711445e-05,
+      "loss": 2.8581,
+      "step": 406
+    },
+    {
+      "epoch": 0.783109404990403,
+      "grad_norm": 1.9939192533493042,
+      "learning_rate": 7.14383072830301e-05,
+      "loss": 2.8433,
+      "step": 408
+    },
+    {
+      "epoch": 0.7869481765834933,
+      "grad_norm": 2.1089251041412354,
+      "learning_rate": 7.115090351587454e-05,
+      "loss": 2.8486,
+      "step": 410
+    },
+    {
+      "epoch": 0.7907869481765835,
+      "grad_norm": 2.033663749694824,
+      "learning_rate": 7.086264606968214e-05,
+      "loss": 2.828,
+      "step": 412
+    },
+    {
+      "epoch": 0.7946257197696737,
+      "grad_norm": 2.1510426998138428,
+      "learning_rate": 7.05735465789128e-05,
+      "loss": 2.9666,
+      "step": 414
+    },
+    {
+      "epoch": 0.7984644913627639,
+      "grad_norm": 2.126054048538208,
+      "learning_rate": 7.028361671201245e-05,
+      "loss": 2.8362,
+      "step": 416
+    },
+    {
+      "epoch": 0.8023032629558541,
+      "grad_norm": 2.3474514484405518,
+      "learning_rate": 6.99928681709422e-05,
+      "loss": 2.7777,
+      "step": 418
+    },
+    {
+      "epoch": 0.8061420345489443,
+      "grad_norm": 2.0623159408569336,
+      "learning_rate": 6.970131269070592e-05,
+      "loss": 2.9225,
+      "step": 420
+    },
+    {
+      "epoch": 0.8099808061420346,
+      "grad_norm": 2.257369041442871,
+      "learning_rate": 6.940896203887659e-05,
+      "loss": 2.801,
+      "step": 422
+    },
+    {
+      "epoch": 0.8138195777351248,
+      "grad_norm": 2.3267722129821777,
+      "learning_rate": 6.911582801512147e-05,
+      "loss": 2.8882,
+      "step": 424
+    },
+    {
+      "epoch": 0.817658349328215,
+      "grad_norm": 2.18564772605896,
+      "learning_rate": 6.882192245072569e-05,
+      "loss": 2.8164,
+      "step": 426
+    },
+    {
+      "epoch": 0.8214971209213052,
+      "grad_norm": 3.0314860343933105,
+      "learning_rate": 6.852725720811488e-05,
+      "loss": 2.7877,
+      "step": 428
+    },
+    {
+      "epoch": 0.8253358925143954,
+      "grad_norm": 2.14628005027771,
+      "learning_rate": 6.823184418037624e-05,
+      "loss": 2.812,
+      "step": 430
+    },
+    {
+      "epoch": 0.8291746641074856,
+      "grad_norm": 2.2532942295074463,
+      "learning_rate": 6.793569529077864e-05,
+      "loss": 2.7552,
+      "step": 432
+    },
+    {
+      "epoch": 0.8330134357005758,
+      "grad_norm": 2.1687734127044678,
+      "learning_rate": 6.76388224922913e-05,
+      "loss": 2.7738,
+      "step": 434
+    },
+    {
+      "epoch": 0.836852207293666,
+      "grad_norm": 2.3331027030944824,
+      "learning_rate": 6.734123776710137e-05,
+      "loss": 2.8104,
+      "step": 436
+    },
+    {
+      "epoch": 0.8406909788867563,
+      "grad_norm": 2.2633862495422363,
+      "learning_rate": 6.704295312613037e-05,
+      "loss": 2.7378,
+      "step": 438
+    },
+    {
+      "epoch": 0.8445297504798465,
+      "grad_norm": 2.0122358798980713,
+      "learning_rate": 6.674398060854931e-05,
+      "loss": 2.7827,
+      "step": 440
+    },
+    {
+      "epoch": 0.8483685220729367,
+      "grad_norm": 1.988330364227295,
+      "learning_rate": 6.644433228129287e-05,
+      "loss": 2.827,
+      "step": 442
+    },
+    {
+      "epoch": 0.8522072936660269,
+      "grad_norm": 2.3183484077453613,
+      "learning_rate": 6.614402023857232e-05,
+      "loss": 2.8699,
+      "step": 444
+    },
+    {
+      "epoch": 0.8560460652591171,
+      "grad_norm": 1.986506462097168,
+      "learning_rate": 6.584305660138734e-05,
+      "loss": 2.8937,
+      "step": 446
+    },
+    {
+      "epoch": 0.8598848368522073,
+      "grad_norm": 2.1625592708587646,
+      "learning_rate": 6.554145351703689e-05,
+      "loss": 2.8388,
+      "step": 448
+    },
+    {
+      "epoch": 0.8637236084452975,
+      "grad_norm": 2.391918897628784,
+      "learning_rate": 6.523922315862887e-05,
+      "loss": 2.7143,
+      "step": 450
+    },
+    {
+      "epoch": 0.8675623800383877,
+      "grad_norm": 2.2226004600524902,
+      "learning_rate": 6.493637772458879e-05,
+      "loss": 2.8072,
+      "step": 452
+    },
+    {
+      "epoch": 0.8714011516314779,
+      "grad_norm": 2.3496720790863037,
+      "learning_rate": 6.463292943816746e-05,
+      "loss": 2.9179,
+      "step": 454
+    },
+    {
+      "epoch": 0.8752399232245681,
+      "grad_norm": 2.0147578716278076,
+      "learning_rate": 6.432889054694764e-05,
+      "loss": 2.7736,
+      "step": 456
+    },
+    {
+      "epoch": 0.8790786948176583,
+      "grad_norm": 2.150322198867798,
+      "learning_rate": 6.402427332234965e-05,
+      "loss": 2.71,
+      "step": 458
+    },
+    {
+      "epoch": 0.8829174664107485,
+      "grad_norm": 1.9796924591064453,
+      "learning_rate": 6.371909005913618e-05,
+      "loss": 2.846,
+      "step": 460
+    },
+    {
+      "epoch": 0.8867562380038387,
+      "grad_norm": 1.9230434894561768,
+      "learning_rate": 6.341335307491595e-05,
+      "loss": 2.7577,
+      "step": 462
+    },
+    {
+      "epoch": 0.8905950095969289,
+      "grad_norm": 2.1865625381469727,
+      "learning_rate": 6.310707470964668e-05,
+      "loss": 2.7676,
+      "step": 464
+    },
+    {
+      "epoch": 0.8944337811900192,
+      "grad_norm": 2.251591682434082,
+      "learning_rate": 6.280026732513689e-05,
+      "loss": 2.7558,
+      "step": 466
+    },
+    {
+      "epoch": 0.8982725527831094,
+      "grad_norm": 2.048290252685547,
+      "learning_rate": 6.249294330454705e-05,
+      "loss": 2.806,
+      "step": 468
+    },
+    {
+      "epoch": 0.9021113243761996,
+      "grad_norm": 2.071037769317627,
+      "learning_rate": 6.21851150518898e-05,
+      "loss": 2.7993,
+      "step": 470
+    },
+    {
+      "epoch": 0.9059500959692899,
+      "grad_norm": 2.1514012813568115,
+      "learning_rate": 6.187679499152919e-05,
+      "loss": 2.8195,
+      "step": 472
+    },
+    {
+      "epoch": 0.9097888675623801,
+      "grad_norm": 2.073908567428589,
+      "learning_rate": 6.15679955676794e-05,
+      "loss": 2.7419,
+      "step": 474
+    },
+    {
+      "epoch": 0.9136276391554703,
+      "grad_norm": 2.136406660079956,
+      "learning_rate": 6.125872924390226e-05,
+      "loss": 2.6854,
+      "step": 476
+    },
+    {
+      "epoch": 0.9174664107485605,
+      "grad_norm": 1.851083755493164,
+      "learning_rate": 6.094900850260439e-05,
+      "loss": 2.8086,
+      "step": 478
+    },
+    {
+      "epoch": 0.9213051823416507,
+      "grad_norm": 2.3391993045806885,
+      "learning_rate": 6.0638845844533254e-05,
+      "loss": 2.8492,
+      "step": 480
+    },
+    {
+      "epoch": 0.9251439539347409,
+      "grad_norm": 2.0181124210357666,
+      "learning_rate": 6.032825378827273e-05,
+      "loss": 2.7973,
+      "step": 482
+    },
+    {
+      "epoch": 0.9289827255278311,
+      "grad_norm": 1.977287769317627,
+      "learning_rate": 6.0017244869737746e-05,
+      "loss": 2.7837,
+      "step": 484
+    },
+    {
+      "epoch": 0.9328214971209213,
+      "grad_norm": 2.0987181663513184,
+      "learning_rate": 5.970583164166838e-05,
+      "loss": 2.7967,
+      "step": 486
+    },
+    {
+      "epoch": 0.9366602687140115,
+      "grad_norm": 2.025846242904663,
+      "learning_rate": 5.9394026673123157e-05,
+      "loss": 2.8064,
+      "step": 488
+    },
+    {
+      "epoch": 0.9404990403071017,
+      "grad_norm": 2.065599203109741,
+      "learning_rate": 5.908184254897182e-05,
+      "loss": 2.7712,
+      "step": 490
+    },
+    {
+      "epoch": 0.944337811900192,
+      "grad_norm": 2.3943252563476562,
+      "learning_rate": 5.8769291869387335e-05,
+      "loss": 2.8494,
+      "step": 492
+    },
+    {
+      "epoch": 0.9481765834932822,
+      "grad_norm": 2.1195807456970215,
+      "learning_rate": 5.845638724933728e-05,
+      "loss": 2.7855,
+      "step": 494
+    },
+    {
+      "epoch": 0.9520153550863724,
+      "grad_norm": 1.8809967041015625,
+      "learning_rate": 5.814314131807486e-05,
+      "loss": 2.79,
+      "step": 496
+    },
+    {
+      "epoch": 0.9558541266794626,
+      "grad_norm": 1.9404385089874268,
+      "learning_rate": 5.7829566718628946e-05,
+      "loss": 2.7431,
+      "step": 498
+    },
+    {
+      "epoch": 0.9596928982725528,
+      "grad_norm": 1.913078784942627,
+      "learning_rate": 5.751567610729398e-05,
+      "loss": 2.7221,
+      "step": 500
+    },
+    {
+      "epoch": 0.963531669865643,
+      "grad_norm": 1.974898338317871,
+      "learning_rate": 5.7201482153119015e-05,
+      "loss": 2.7087,
+      "step": 502
+    },
+    {
+      "epoch": 0.9673704414587332,
+      "grad_norm": 2.0164756774902344,
+      "learning_rate": 5.6886997537396494e-05,
+      "loss": 2.6904,
+      "step": 504
+    },
+    {
+      "epoch": 0.9712092130518234,
+      "grad_norm": 2.159647226333618,
+      "learning_rate": 5.6572234953150305e-05,
+      "loss": 2.8019,
+      "step": 506
+    },
+    {
+      "epoch": 0.9750479846449136,
+      "grad_norm": 1.9416415691375732,
+      "learning_rate": 5.6257207104623524e-05,
+      "loss": 2.7502,
+      "step": 508
+    },
+    {
+      "epoch": 0.9788867562380038,
+      "grad_norm": 2.1035892963409424,
+      "learning_rate": 5.594192670676568e-05,
+      "loss": 2.6954,
+      "step": 510
+    },
+    {
+      "epoch": 0.982725527831094,
+      "grad_norm": 1.8888871669769287,
+      "learning_rate": 5.5626406484719506e-05,
+      "loss": 2.751,
+      "step": 512
+    },
+    {
+      "epoch": 0.9865642994241842,
+      "grad_norm": 1.9274441003799438,
+      "learning_rate": 5.5310659173307366e-05,
+      "loss": 2.7768,
+      "step": 514
+    },
+    {
+      "epoch": 0.9904030710172744,
+      "grad_norm": 1.9080277681350708,
+      "learning_rate": 5.499469751651728e-05,
+      "loss": 2.8421,
+      "step": 516
+    },
+    {
+      "epoch": 0.9942418426103646,
+      "grad_norm": 1.9564564228057861,
+      "learning_rate": 5.467853426698853e-05,
+      "loss": 2.8099,
+      "step": 518
+    },
+    {
+      "epoch": 0.9980806142034548,
+      "grad_norm": 1.8488651514053345,
+      "learning_rate": 5.43621821854969e-05,
+      "loss": 2.733,
+      "step": 520
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 1042,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 521,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.81062206945362e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-521/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6937747f26bef28460fcdec34b04d43de882bbe87ba71ec6a7684200eddd5c51
+size 5432

config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "/mnt/parscratch/users/acp23ay/private/models/OLMo-2-1124-7B-ary_Arab-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 sha256:3c7999db0cf31bf98af925cb7b6f4ad8024b9e864a35d596969a73e20f455324
+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 sha256:840857203f141ce5771549f01229a9f15fa001fc49a509f633c43a6288bd3878
+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 sha256:6937747f26bef28460fcdec34b04d43de882bbe87ba71ec6a7684200eddd5c51
+size 5432