Upload 13 files

config.json

@@ -0,0 +1,27 @@
+{
+  "architectures": [
+    "OlmoForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "clip_qkv": null,
+  "eos_token_id": 50279,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 8192,
+  "max_position_embeddings": 2048,
+  "model_type": "olmo",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 4,
+  "pad_token_id": 1,
+  "pretraining_tp": 1,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.40.2",
+  "use_cache": true,
+  "vocab_size": 50304
+}

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 50279,
+  "pad_token_id": 1,
+  "transformers_version": "4.40.2"
+}

optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bdcd739af8dfaca51bdd9f9f2d776563a65d4eb5b50488d2b45ee7a03aa5ff2d
+size 16393633958

pytorch_model-00001-of-00002.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ccb7c1855bcc1fae26fe7f0e45668e6c51238151aad5fb267a3bdcd7d9b2453
+size 4967152029

pytorch_model-00002-of-00002.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:48bb8dea8d16c1cdbc7e7a8b1e31491c2f31d4c8643e5fa89b9bdcdb7bf4de5a
+size 3229646818

pytorch_model.bin.index.json

@@ -0,0 +1,233 @@
+{
+  "metadata": {
+    "total_size": 8196718592
+  },
+  "weight_map": {
+    "lm_head.weight": "pytorch_model-00001-of-00002.bin",
+    "model.embed_tokens.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.19.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.20.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.21.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.21.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.22.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.22.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.30.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin"
+  }
+}

rng_state.pth

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

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a06377e686c1b183f5f5c60c3d3432d77774dbfe74b1c84563286245a76e870a
+size 1064

special_tokens_map.json

@@ -0,0 +1,16 @@
+{
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|padding|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer_config.json

@@ -0,0 +1,238 @@
+{
+  "add_bos_token": false,
+  "add_eos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "|||IP_ADDRESS|||",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "1": {
+      "content": "<|padding|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50254": {
+      "content": "                        ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50255": {
+      "content": "                       ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50256": {
+      "content": "                      ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50257": {
+      "content": "                     ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50258": {
+      "content": "                    ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50259": {
+      "content": "                   ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50260": {
+      "content": "                  ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50261": {
+      "content": "                 ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50262": {
+      "content": "                ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50263": {
+      "content": "               ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50264": {
+      "content": "              ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50265": {
+      "content": "             ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50266": {
+      "content": "            ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50267": {
+      "content": "           ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50268": {
+      "content": "          ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50269": {
+      "content": "         ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50270": {
+      "content": "        ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50271": {
+      "content": "       ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50272": {
+      "content": "      ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50273": {
+      "content": "     ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50274": {
+      "content": "    ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50275": {
+      "content": "   ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50276": {
+      "content": "  ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50277": {
+      "content": "|||EMAIL_ADDRESS|||",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50278": {
+      "content": "|||PHONE_NUMBER|||",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50279": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": null,
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|endoftext|>",
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<|padding|>",
+  "tokenizer_class": "GPTNeoXTokenizer",
+  "unk_token": null
+}

trainer_state.json

@@ -0,0 +1,1708 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.05868090819708105,
+  "eval_steps": 100000000,
+  "global_step": 7200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 8.15012613848348e-06,
+      "grad_norm": 3.4710192680358887,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 11.2605,
+      "step": 1
+    },
+    {
+      "epoch": 0.0002445037841545044,
+      "grad_norm": 5.464046955108643,
+      "learning_rate": 3e-06,
+      "loss": 10.9977,
+      "step": 30
+    },
+    {
+      "epoch": 0.0004890075683090088,
+      "grad_norm": 1.1376756429672241,
+      "learning_rate": 6e-06,
+      "loss": 9.4082,
+      "step": 60
+    },
+    {
+      "epoch": 0.0007335113524635132,
+      "grad_norm": 1.2878344058990479,
+      "learning_rate": 9e-06,
+      "loss": 8.549,
+      "step": 90
+    },
+    {
+      "epoch": 0.0009780151366180175,
+      "grad_norm": 1.2800588607788086,
+      "learning_rate": 1.2e-05,
+      "loss": 8.0361,
+      "step": 120
+    },
+    {
+      "epoch": 0.001222518920772522,
+      "grad_norm": 1.4315314292907715,
+      "learning_rate": 1.5e-05,
+      "loss": 7.653,
+      "step": 150
+    },
+    {
+      "epoch": 0.0014670227049270264,
+      "grad_norm": 1.381317377090454,
+      "learning_rate": 1.8e-05,
+      "loss": 7.4179,
+      "step": 180
+    },
+    {
+      "epoch": 0.0017115264890815308,
+      "grad_norm": 1.7989460229873657,
+      "learning_rate": 2.1e-05,
+      "loss": 7.2012,
+      "step": 210
+    },
+    {
+      "epoch": 0.001956030273236035,
+      "grad_norm": 1.3431414365768433,
+      "learning_rate": 2.4e-05,
+      "loss": 7.0383,
+      "step": 240
+    },
+    {
+      "epoch": 0.0022005340573905395,
+      "grad_norm": 1.028826117515564,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 6.866,
+      "step": 270
+    },
+    {
+      "epoch": 0.002445037841545044,
+      "grad_norm": 1.201025128364563,
+      "learning_rate": 3e-05,
+      "loss": 6.717,
+      "step": 300
+    },
+    {
+      "epoch": 0.0026895416256995483,
+      "grad_norm": 1.1023098230361938,
+      "learning_rate": 3.3e-05,
+      "loss": 6.5535,
+      "step": 330
+    },
+    {
+      "epoch": 0.0029340454098540527,
+      "grad_norm": 1.2839674949645996,
+      "learning_rate": 3.6e-05,
+      "loss": 6.4022,
+      "step": 360
+    },
+    {
+      "epoch": 0.003178549194008557,
+      "grad_norm": 2.267265796661377,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 6.2858,
+      "step": 390
+    },
+    {
+      "epoch": 0.0034230529781630616,
+      "grad_norm": 1.0635628700256348,
+      "learning_rate": 4.2e-05,
+      "loss": 6.1681,
+      "step": 420
+    },
+    {
+      "epoch": 0.0036675567623175656,
+      "grad_norm": 1.263838768005371,
+      "learning_rate": 4.5e-05,
+      "loss": 6.0728,
+      "step": 450
+    },
+    {
+      "epoch": 0.00391206054647207,
+      "grad_norm": 1.4611454010009766,
+      "learning_rate": 4.8e-05,
+      "loss": 5.972,
+      "step": 480
+    },
+    {
+      "epoch": 0.0041565643306265745,
+      "grad_norm": 1.0120561122894287,
+      "learning_rate": 4.999999990869806e-05,
+      "loss": 5.8619,
+      "step": 510
+    },
+    {
+      "epoch": 0.004401068114781079,
+      "grad_norm": 1.1349974870681763,
+      "learning_rate": 4.999999853916893e-05,
+      "loss": 5.785,
+      "step": 540
+    },
+    {
+      "epoch": 0.004645571898935583,
+      "grad_norm": 1.0840613842010498,
+      "learning_rate": 4.9999995526204936e-05,
+      "loss": 5.7071,
+      "step": 570
+    },
+    {
+      "epoch": 0.004890075683090088,
+      "grad_norm": 1.258074402809143,
+      "learning_rate": 4.999999086980628e-05,
+      "loss": 5.6199,
+      "step": 600
+    },
+    {
+      "epoch": 0.005134579467244592,
+      "grad_norm": 1.284726858139038,
+      "learning_rate": 4.999998456997326e-05,
+      "loss": 5.5465,
+      "step": 630
+    },
+    {
+      "epoch": 0.005379083251399097,
+      "grad_norm": 1.2079874277114868,
+      "learning_rate": 4.999997662670628e-05,
+      "loss": 5.4816,
+      "step": 660
+    },
+    {
+      "epoch": 0.005623587035553601,
+      "grad_norm": 1.3364052772521973,
+      "learning_rate": 4.999996704000589e-05,
+      "loss": 5.4079,
+      "step": 690
+    },
+    {
+      "epoch": 0.0058680908197081055,
+      "grad_norm": 0.9860705137252808,
+      "learning_rate": 4.99999558098727e-05,
+      "loss": 5.3598,
+      "step": 720
+    },
+    {
+      "epoch": 0.00611259460386261,
+      "grad_norm": 1.2071930170059204,
+      "learning_rate": 4.9999942936307445e-05,
+      "loss": 5.2884,
+      "step": 750
+    },
+    {
+      "epoch": 0.006357098388017114,
+      "grad_norm": 0.8959563970565796,
+      "learning_rate": 4.9999928419310994e-05,
+      "loss": 5.2391,
+      "step": 780
+    },
+    {
+      "epoch": 0.006601602172171619,
+      "grad_norm": 1.2356096506118774,
+      "learning_rate": 4.999991225888427e-05,
+      "loss": 5.1879,
+      "step": 810
+    },
+    {
+      "epoch": 0.006846105956326123,
+      "grad_norm": 0.9705113172531128,
+      "learning_rate": 4.999989445502837e-05,
+      "loss": 5.1424,
+      "step": 840
+    },
+    {
+      "epoch": 0.007090609740480627,
+      "grad_norm": 0.9504437446594238,
+      "learning_rate": 4.9999875007744436e-05,
+      "loss": 5.0966,
+      "step": 870
+    },
+    {
+      "epoch": 0.007335113524635131,
+      "grad_norm": 0.9488673806190491,
+      "learning_rate": 4.9999853917033756e-05,
+      "loss": 5.0424,
+      "step": 900
+    },
+    {
+      "epoch": 0.007579617308789636,
+      "grad_norm": 0.959373950958252,
+      "learning_rate": 4.999983118289773e-05,
+      "loss": 5.0387,
+      "step": 930
+    },
+    {
+      "epoch": 0.00782412109294414,
+      "grad_norm": 0.8465414643287659,
+      "learning_rate": 4.999980680533782e-05,
+      "loss": 4.9769,
+      "step": 960
+    },
+    {
+      "epoch": 0.008068624877098645,
+      "grad_norm": 0.8328993916511536,
+      "learning_rate": 4.999978078435567e-05,
+      "loss": 4.9335,
+      "step": 990
+    },
+    {
+      "epoch": 0.008313128661253149,
+      "grad_norm": 0.8107655644416809,
+      "learning_rate": 4.999975311995295e-05,
+      "loss": 4.9214,
+      "step": 1020
+    },
+    {
+      "epoch": 0.008557632445407654,
+      "grad_norm": 0.8149654865264893,
+      "learning_rate": 4.99997238121315e-05,
+      "loss": 4.8651,
+      "step": 1050
+    },
+    {
+      "epoch": 0.008802136229562158,
+      "grad_norm": 0.8837414979934692,
+      "learning_rate": 4.999969286089325e-05,
+      "loss": 4.8327,
+      "step": 1080
+    },
+    {
+      "epoch": 0.009046640013716663,
+      "grad_norm": 1.1360137462615967,
+      "learning_rate": 4.9999660266240235e-05,
+      "loss": 4.7906,
+      "step": 1110
+    },
+    {
+      "epoch": 0.009291143797871167,
+      "grad_norm": 0.7992026209831238,
+      "learning_rate": 4.9999626028174585e-05,
+      "loss": 4.7612,
+      "step": 1140
+    },
+    {
+      "epoch": 0.00953564758202567,
+      "grad_norm": 0.8481825590133667,
+      "learning_rate": 4.999959014669856e-05,
+      "loss": 4.7106,
+      "step": 1170
+    },
+    {
+      "epoch": 0.009780151366180176,
+      "grad_norm": 0.8183879256248474,
+      "learning_rate": 4.9999552621814513e-05,
+      "loss": 4.6993,
+      "step": 1200
+    },
+    {
+      "epoch": 0.010024655150334679,
+      "grad_norm": 0.8460689187049866,
+      "learning_rate": 4.9999513453524917e-05,
+      "loss": 4.6664,
+      "step": 1230
+    },
+    {
+      "epoch": 0.010269158934489184,
+      "grad_norm": 0.8723706007003784,
+      "learning_rate": 4.9999472641832336e-05,
+      "loss": 4.6371,
+      "step": 1260
+    },
+    {
+      "epoch": 0.010513662718643688,
+      "grad_norm": 0.7682787179946899,
+      "learning_rate": 4.999943018673946e-05,
+      "loss": 4.6184,
+      "step": 1290
+    },
+    {
+      "epoch": 0.010758166502798193,
+      "grad_norm": 0.9049955606460571,
+      "learning_rate": 4.999938608824909e-05,
+      "loss": 4.5968,
+      "step": 1320
+    },
+    {
+      "epoch": 0.011002670286952697,
+      "grad_norm": 0.7860899567604065,
+      "learning_rate": 4.999934034636411e-05,
+      "loss": 4.5266,
+      "step": 1350
+    },
+    {
+      "epoch": 0.011247174071107202,
+      "grad_norm": 0.7918768525123596,
+      "learning_rate": 4.999929296108753e-05,
+      "loss": 4.5069,
+      "step": 1380
+    },
+    {
+      "epoch": 0.011491677855261706,
+      "grad_norm": 0.6713089942932129,
+      "learning_rate": 4.9999243932422466e-05,
+      "loss": 4.4663,
+      "step": 1410
+    },
+    {
+      "epoch": 0.011736181639416211,
+      "grad_norm": 0.756101131439209,
+      "learning_rate": 4.999919326037215e-05,
+      "loss": 4.4782,
+      "step": 1440
+    },
+    {
+      "epoch": 0.011980685423570715,
+      "grad_norm": 0.7067996263504028,
+      "learning_rate": 4.99991409449399e-05,
+      "loss": 4.4388,
+      "step": 1470
+    },
+    {
+      "epoch": 0.01222518920772522,
+      "grad_norm": 0.7132194638252258,
+      "learning_rate": 4.999908698612916e-05,
+      "loss": 4.4135,
+      "step": 1500
+    },
+    {
+      "epoch": 0.012469692991879723,
+      "grad_norm": 0.7282501459121704,
+      "learning_rate": 4.9999031383943486e-05,
+      "loss": 4.4057,
+      "step": 1530
+    },
+    {
+      "epoch": 0.012714196776034229,
+      "grad_norm": 0.6851722598075867,
+      "learning_rate": 4.999897413838651e-05,
+      "loss": 4.3612,
+      "step": 1560
+    },
+    {
+      "epoch": 0.012958700560188732,
+      "grad_norm": 0.7953110337257385,
+      "learning_rate": 4.999891524946202e-05,
+      "loss": 4.3268,
+      "step": 1590
+    },
+    {
+      "epoch": 0.013203204344343238,
+      "grad_norm": 0.726719319820404,
+      "learning_rate": 4.999885471717387e-05,
+      "loss": 4.314,
+      "step": 1620
+    },
+    {
+      "epoch": 0.013447708128497741,
+      "grad_norm": 0.7354792356491089,
+      "learning_rate": 4.999879254152605e-05,
+      "loss": 4.2807,
+      "step": 1650
+    },
+    {
+      "epoch": 0.013692211912652246,
+      "grad_norm": 0.7072017788887024,
+      "learning_rate": 4.999872872252265e-05,
+      "loss": 4.282,
+      "step": 1680
+    },
+    {
+      "epoch": 0.01393671569680675,
+      "grad_norm": 0.690112292766571,
+      "learning_rate": 4.999866326016785e-05,
+      "loss": 4.2686,
+      "step": 1710
+    },
+    {
+      "epoch": 0.014181219480961254,
+      "grad_norm": 0.697634756565094,
+      "learning_rate": 4.999859615446596e-05,
+      "loss": 4.2662,
+      "step": 1740
+    },
+    {
+      "epoch": 0.014425723265115759,
+      "grad_norm": 0.6999862790107727,
+      "learning_rate": 4.99985274054214e-05,
+      "loss": 4.2207,
+      "step": 1770
+    },
+    {
+      "epoch": 0.014670227049270262,
+      "grad_norm": 0.6920334100723267,
+      "learning_rate": 4.999845701303868e-05,
+      "loss": 4.2163,
+      "step": 1800
+    },
+    {
+      "epoch": 0.014914730833424768,
+      "grad_norm": 0.6884493827819824,
+      "learning_rate": 4.999838497732243e-05,
+      "loss": 4.209,
+      "step": 1830
+    },
+    {
+      "epoch": 0.015159234617579271,
+      "grad_norm": 0.776447057723999,
+      "learning_rate": 4.999831129827739e-05,
+      "loss": 4.1856,
+      "step": 1860
+    },
+    {
+      "epoch": 0.015403738401733777,
+      "grad_norm": 0.7332949042320251,
+      "learning_rate": 4.9998235975908394e-05,
+      "loss": 4.156,
+      "step": 1890
+    },
+    {
+      "epoch": 0.01564824218588828,
+      "grad_norm": 0.6691558361053467,
+      "learning_rate": 4.99981590102204e-05,
+      "loss": 4.1339,
+      "step": 1920
+    },
+    {
+      "epoch": 0.015892745970042785,
+      "grad_norm": 0.6789201498031616,
+      "learning_rate": 4.9998080401218464e-05,
+      "loss": 4.1344,
+      "step": 1950
+    },
+    {
+      "epoch": 0.01613724975419729,
+      "grad_norm": 0.6654175519943237,
+      "learning_rate": 4.999800014890777e-05,
+      "loss": 4.1295,
+      "step": 1980
+    },
+    {
+      "epoch": 0.016381753538351793,
+      "grad_norm": 0.6632519364356995,
+      "learning_rate": 4.9997918253293555e-05,
+      "loss": 4.1036,
+      "step": 2010
+    },
+    {
+      "epoch": 0.016626257322506298,
+      "grad_norm": 0.6618740558624268,
+      "learning_rate": 4.999783471438124e-05,
+      "loss": 4.0857,
+      "step": 2040
+    },
+    {
+      "epoch": 0.016870761106660803,
+      "grad_norm": 0.6384952068328857,
+      "learning_rate": 4.999774953217631e-05,
+      "loss": 4.0755,
+      "step": 2070
+    },
+    {
+      "epoch": 0.01711526489081531,
+      "grad_norm": 0.6808215975761414,
+      "learning_rate": 4.9997662706684345e-05,
+      "loss": 4.0568,
+      "step": 2100
+    },
+    {
+      "epoch": 0.01735976867496981,
+      "grad_norm": 0.6514068841934204,
+      "learning_rate": 4.999757423791107e-05,
+      "loss": 4.0474,
+      "step": 2130
+    },
+    {
+      "epoch": 0.017604272459124316,
+      "grad_norm": 0.6691142320632935,
+      "learning_rate": 4.9997484125862306e-05,
+      "loss": 4.0467,
+      "step": 2160
+    },
+    {
+      "epoch": 0.01784877624327882,
+      "grad_norm": 0.6240984201431274,
+      "learning_rate": 4.999739237054395e-05,
+      "loss": 4.0259,
+      "step": 2190
+    },
+    {
+      "epoch": 0.018093280027433326,
+      "grad_norm": 0.6286123991012573,
+      "learning_rate": 4.9997298971962065e-05,
+      "loss": 4.0098,
+      "step": 2220
+    },
+    {
+      "epoch": 0.018337783811587828,
+      "grad_norm": 0.6232910752296448,
+      "learning_rate": 4.999720393012277e-05,
+      "loss": 4.0203,
+      "step": 2250
+    },
+    {
+      "epoch": 0.018582287595742333,
+      "grad_norm": 0.6792376041412354,
+      "learning_rate": 4.999710724503233e-05,
+      "loss": 3.9909,
+      "step": 2280
+    },
+    {
+      "epoch": 0.01882679137989684,
+      "grad_norm": 0.6824718117713928,
+      "learning_rate": 4.9997008916697075e-05,
+      "loss": 3.9922,
+      "step": 2310
+    },
+    {
+      "epoch": 0.01907129516405134,
+      "grad_norm": 0.6520410776138306,
+      "learning_rate": 4.999690894512349e-05,
+      "loss": 3.9533,
+      "step": 2340
+    },
+    {
+      "epoch": 0.019315798948205846,
+      "grad_norm": 0.6467030048370361,
+      "learning_rate": 4.999680733031814e-05,
+      "loss": 3.9464,
+      "step": 2370
+    },
+    {
+      "epoch": 0.01956030273236035,
+      "grad_norm": 0.6356080770492554,
+      "learning_rate": 4.9996704072287716e-05,
+      "loss": 3.9288,
+      "step": 2400
+    },
+    {
+      "epoch": 0.019804806516514856,
+      "grad_norm": 0.6498362421989441,
+      "learning_rate": 4.9996599171038984e-05,
+      "loss": 3.9345,
+      "step": 2430
+    },
+    {
+      "epoch": 0.020049310300669358,
+      "grad_norm": 0.6107562184333801,
+      "learning_rate": 4.999649262657886e-05,
+      "loss": 3.9368,
+      "step": 2460
+    },
+    {
+      "epoch": 0.020293814084823864,
+      "grad_norm": 0.6478250622749329,
+      "learning_rate": 4.999638443891434e-05,
+      "loss": 3.9273,
+      "step": 2490
+    },
+    {
+      "epoch": 0.02053831786897837,
+      "grad_norm": 0.631263256072998,
+      "learning_rate": 4.999627460805253e-05,
+      "loss": 3.9136,
+      "step": 2520
+    },
+    {
+      "epoch": 0.020782821653132874,
+      "grad_norm": 0.6120920777320862,
+      "learning_rate": 4.999616313400066e-05,
+      "loss": 3.8834,
+      "step": 2550
+    },
+    {
+      "epoch": 0.021027325437287376,
+      "grad_norm": 0.6151197552680969,
+      "learning_rate": 4.999605001676605e-05,
+      "loss": 3.8994,
+      "step": 2580
+    },
+    {
+      "epoch": 0.02127182922144188,
+      "grad_norm": 0.623715877532959,
+      "learning_rate": 4.9995935256356144e-05,
+      "loss": 3.8929,
+      "step": 2610
+    },
+    {
+      "epoch": 0.021516333005596387,
+      "grad_norm": 0.6688769459724426,
+      "learning_rate": 4.9995818852778476e-05,
+      "loss": 3.8499,
+      "step": 2640
+    },
+    {
+      "epoch": 0.021760836789750892,
+      "grad_norm": 0.6272155046463013,
+      "learning_rate": 4.999570080604071e-05,
+      "loss": 3.8861,
+      "step": 2670
+    },
+    {
+      "epoch": 0.022005340573905394,
+      "grad_norm": 0.597653329372406,
+      "learning_rate": 4.99955811161506e-05,
+      "loss": 3.8674,
+      "step": 2700
+    },
+    {
+      "epoch": 0.0222498443580599,
+      "grad_norm": 0.5608483552932739,
+      "learning_rate": 4.9995459783116004e-05,
+      "loss": 3.8493,
+      "step": 2730
+    },
+    {
+      "epoch": 0.022494348142214404,
+      "grad_norm": 0.5991063117980957,
+      "learning_rate": 4.999533680694493e-05,
+      "loss": 3.8454,
+      "step": 2760
+    },
+    {
+      "epoch": 0.02273885192636891,
+      "grad_norm": 0.5738102793693542,
+      "learning_rate": 4.9995212187645416e-05,
+      "loss": 3.8395,
+      "step": 2790
+    },
+    {
+      "epoch": 0.02298335571052341,
+      "grad_norm": 0.6145568490028381,
+      "learning_rate": 4.9995085925225693e-05,
+      "loss": 3.8313,
+      "step": 2820
+    },
+    {
+      "epoch": 0.023227859494677917,
+      "grad_norm": 0.6019515991210938,
+      "learning_rate": 4.999495801969404e-05,
+      "loss": 3.8277,
+      "step": 2850
+    },
+    {
+      "epoch": 0.023472363278832422,
+      "grad_norm": 0.6177758574485779,
+      "learning_rate": 4.9994828471058876e-05,
+      "loss": 3.8083,
+      "step": 2880
+    },
+    {
+      "epoch": 0.023716867062986924,
+      "grad_norm": 0.5616528987884521,
+      "learning_rate": 4.9994697279328714e-05,
+      "loss": 3.8114,
+      "step": 2910
+    },
+    {
+      "epoch": 0.02396137084714143,
+      "grad_norm": 0.6194447875022888,
+      "learning_rate": 4.9994564444512176e-05,
+      "loss": 3.8183,
+      "step": 2940
+    },
+    {
+      "epoch": 0.024205874631295934,
+      "grad_norm": 0.578895092010498,
+      "learning_rate": 4.9994429966618e-05,
+      "loss": 3.7871,
+      "step": 2970
+    },
+    {
+      "epoch": 0.02445037841545044,
+      "grad_norm": 0.6014060378074646,
+      "learning_rate": 4.999429384565502e-05,
+      "loss": 3.7711,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02469488219960494,
+      "grad_norm": 0.5589067339897156,
+      "learning_rate": 4.999415608163217e-05,
+      "loss": 3.7533,
+      "step": 3030
+    },
+    {
+      "epoch": 0.024939385983759447,
+      "grad_norm": 0.5722873210906982,
+      "learning_rate": 4.999401667455854e-05,
+      "loss": 3.7585,
+      "step": 3060
+    },
+    {
+      "epoch": 0.025183889767913952,
+      "grad_norm": 0.5734965801239014,
+      "learning_rate": 4.9993875624443274e-05,
+      "loss": 3.77,
+      "step": 3090
+    },
+    {
+      "epoch": 0.025428393552068457,
+      "grad_norm": 0.5524207353591919,
+      "learning_rate": 4.9993732931295646e-05,
+      "loss": 3.718,
+      "step": 3120
+    },
+    {
+      "epoch": 0.02567289733622296,
+      "grad_norm": 0.5776082277297974,
+      "learning_rate": 4.999358859512503e-05,
+      "loss": 3.7573,
+      "step": 3150
+    },
+    {
+      "epoch": 0.025917401120377465,
+      "grad_norm": 0.5799595713615417,
+      "learning_rate": 4.9993442615940936e-05,
+      "loss": 3.7552,
+      "step": 3180
+    },
+    {
+      "epoch": 0.02616190490453197,
+      "grad_norm": 0.5820346474647522,
+      "learning_rate": 4.999329499375292e-05,
+      "loss": 3.7394,
+      "step": 3210
+    },
+    {
+      "epoch": 0.026406408688686475,
+      "grad_norm": 0.5450282096862793,
+      "learning_rate": 4.999314572857074e-05,
+      "loss": 3.7393,
+      "step": 3240
+    },
+    {
+      "epoch": 0.026650912472840977,
+      "grad_norm": 0.6213249564170837,
+      "learning_rate": 4.9992994820404174e-05,
+      "loss": 3.7191,
+      "step": 3270
+    },
+    {
+      "epoch": 0.026895416256995482,
+      "grad_norm": 0.5940688848495483,
+      "learning_rate": 4.999284226926314e-05,
+      "loss": 3.719,
+      "step": 3300
+    },
+    {
+      "epoch": 0.027139920041149988,
+      "grad_norm": 0.5779993534088135,
+      "learning_rate": 4.999268807515768e-05,
+      "loss": 3.7092,
+      "step": 3330
+    },
+    {
+      "epoch": 0.027384423825304493,
+      "grad_norm": 0.5899255871772766,
+      "learning_rate": 4.999253223809792e-05,
+      "loss": 3.6939,
+      "step": 3360
+    },
+    {
+      "epoch": 0.027628927609458995,
+      "grad_norm": 0.6330375671386719,
+      "learning_rate": 4.999237475809411e-05,
+      "loss": 3.7102,
+      "step": 3390
+    },
+    {
+      "epoch": 0.0278734313936135,
+      "grad_norm": 0.5771914720535278,
+      "learning_rate": 4.99922156351566e-05,
+      "loss": 3.7196,
+      "step": 3420
+    },
+    {
+      "epoch": 0.028117935177768005,
+      "grad_norm": 0.5769143104553223,
+      "learning_rate": 4.999205486929586e-05,
+      "loss": 3.7127,
+      "step": 3450
+    },
+    {
+      "epoch": 0.028362438961922507,
+      "grad_norm": 0.5581954121589661,
+      "learning_rate": 4.999189246052245e-05,
+      "loss": 3.689,
+      "step": 3480
+    },
+    {
+      "epoch": 0.028606942746077012,
+      "grad_norm": 0.6041043400764465,
+      "learning_rate": 4.999172840884704e-05,
+      "loss": 3.6831,
+      "step": 3510
+    },
+    {
+      "epoch": 0.028851446530231518,
+      "grad_norm": 0.5458335280418396,
+      "learning_rate": 4.999156271428043e-05,
+      "loss": 3.6694,
+      "step": 3540
+    },
+    {
+      "epoch": 0.029095950314386023,
+      "grad_norm": 0.5470607280731201,
+      "learning_rate": 4.9991395376833496e-05,
+      "loss": 3.6702,
+      "step": 3570
+    },
+    {
+      "epoch": 0.029340454098540525,
+      "grad_norm": 0.5878787040710449,
+      "learning_rate": 4.999122639651725e-05,
+      "loss": 3.6492,
+      "step": 3600
+    },
+    {
+      "epoch": 0.02958495788269503,
+      "grad_norm": 0.5691691637039185,
+      "learning_rate": 4.9991055773342795e-05,
+      "loss": 3.6812,
+      "step": 3630
+    },
+    {
+      "epoch": 0.029829461666849535,
+      "grad_norm": 0.5548356771469116,
+      "learning_rate": 4.9990883507321354e-05,
+      "loss": 3.645,
+      "step": 3660
+    },
+    {
+      "epoch": 0.03007396545100404,
+      "grad_norm": 0.5686156153678894,
+      "learning_rate": 4.999070959846424e-05,
+      "loss": 3.6505,
+      "step": 3690
+    },
+    {
+      "epoch": 0.030318469235158543,
+      "grad_norm": 0.5596534013748169,
+      "learning_rate": 4.999053404678289e-05,
+      "loss": 3.6532,
+      "step": 3720
+    },
+    {
+      "epoch": 0.030562973019313048,
+      "grad_norm": 0.54421067237854,
+      "learning_rate": 4.999035685228884e-05,
+      "loss": 3.6238,
+      "step": 3750
+    },
+    {
+      "epoch": 0.030807476803467553,
+      "grad_norm": 0.56732177734375,
+      "learning_rate": 4.999017801499375e-05,
+      "loss": 3.6449,
+      "step": 3780
+    },
+    {
+      "epoch": 0.03105198058762206,
+      "grad_norm": 0.5608410239219666,
+      "learning_rate": 4.998999753490937e-05,
+      "loss": 3.6344,
+      "step": 3810
+    },
+    {
+      "epoch": 0.03129648437177656,
+      "grad_norm": 0.5919491648674011,
+      "learning_rate": 4.998981541204757e-05,
+      "loss": 3.6213,
+      "step": 3840
+    },
+    {
+      "epoch": 0.031540988155931066,
+      "grad_norm": 0.5795233249664307,
+      "learning_rate": 4.998963164642031e-05,
+      "loss": 3.6239,
+      "step": 3870
+    },
+    {
+      "epoch": 0.03178549194008557,
+      "grad_norm": 0.5669205784797668,
+      "learning_rate": 4.9989446238039676e-05,
+      "loss": 3.5831,
+      "step": 3900
+    },
+    {
+      "epoch": 0.032029995724240076,
+      "grad_norm": 0.5817368626594543,
+      "learning_rate": 4.998925918691786e-05,
+      "loss": 3.6074,
+      "step": 3930
+    },
+    {
+      "epoch": 0.03227449950839458,
+      "grad_norm": 0.5662333369255066,
+      "learning_rate": 4.998907049306715e-05,
+      "loss": 3.5874,
+      "step": 3960
+    },
+    {
+      "epoch": 0.03251900329254909,
+      "grad_norm": 0.5641735792160034,
+      "learning_rate": 4.998888015649996e-05,
+      "loss": 3.6056,
+      "step": 3990
+    },
+    {
+      "epoch": 0.032763507076703585,
+      "grad_norm": 0.524918794631958,
+      "learning_rate": 4.99886881772288e-05,
+      "loss": 3.5963,
+      "step": 4020
+    },
+    {
+      "epoch": 0.03300801086085809,
+      "grad_norm": 0.5625722408294678,
+      "learning_rate": 4.998849455526628e-05,
+      "loss": 3.5917,
+      "step": 4050
+    },
+    {
+      "epoch": 0.033252514645012596,
+      "grad_norm": 0.5612478852272034,
+      "learning_rate": 4.998829929062515e-05,
+      "loss": 3.5792,
+      "step": 4080
+    },
+    {
+      "epoch": 0.0334970184291671,
+      "grad_norm": 0.5586293935775757,
+      "learning_rate": 4.998810238331822e-05,
+      "loss": 3.5708,
+      "step": 4110
+    },
+    {
+      "epoch": 0.033741522213321606,
+      "grad_norm": 0.53324955701828,
+      "learning_rate": 4.998790383335845e-05,
+      "loss": 3.5686,
+      "step": 4140
+    },
+    {
+      "epoch": 0.03398602599747611,
+      "grad_norm": 0.5210742950439453,
+      "learning_rate": 4.9987703640758894e-05,
+      "loss": 3.575,
+      "step": 4170
+    },
+    {
+      "epoch": 0.03423052978163062,
+      "grad_norm": 0.5591189861297607,
+      "learning_rate": 4.99875018055327e-05,
+      "loss": 3.5717,
+      "step": 4200
+    },
+    {
+      "epoch": 0.034475033565785115,
+      "grad_norm": 0.5435970425605774,
+      "learning_rate": 4.998729832769315e-05,
+      "loss": 3.5638,
+      "step": 4230
+    },
+    {
+      "epoch": 0.03471953734993962,
+      "grad_norm": 0.5489551424980164,
+      "learning_rate": 4.998709320725361e-05,
+      "loss": 3.574,
+      "step": 4260
+    },
+    {
+      "epoch": 0.034964041134094126,
+      "grad_norm": 0.5275290012359619,
+      "learning_rate": 4.998688644422756e-05,
+      "loss": 3.5695,
+      "step": 4290
+    },
+    {
+      "epoch": 0.03520854491824863,
+      "grad_norm": 0.583881139755249,
+      "learning_rate": 4.998667803862861e-05,
+      "loss": 3.5703,
+      "step": 4320
+    },
+    {
+      "epoch": 0.03545304870240314,
+      "grad_norm": 0.5317121744155884,
+      "learning_rate": 4.9986467990470445e-05,
+      "loss": 3.5668,
+      "step": 4350
+    },
+    {
+      "epoch": 0.03569755248655764,
+      "grad_norm": 0.5981696248054504,
+      "learning_rate": 4.998625629976688e-05,
+      "loss": 3.5411,
+      "step": 4380
+    },
+    {
+      "epoch": 0.03594205627071215,
+      "grad_norm": 0.55133455991745,
+      "learning_rate": 4.998604296653182e-05,
+      "loss": 3.521,
+      "step": 4410
+    },
+    {
+      "epoch": 0.03618656005486665,
+      "grad_norm": 0.5481094717979431,
+      "learning_rate": 4.99858279907793e-05,
+      "loss": 3.5421,
+      "step": 4440
+    },
+    {
+      "epoch": 0.03643106383902115,
+      "grad_norm": 0.5218031406402588,
+      "learning_rate": 4.998561137252346e-05,
+      "loss": 3.5305,
+      "step": 4470
+    },
+    {
+      "epoch": 0.036675567623175656,
+      "grad_norm": 0.5458360910415649,
+      "learning_rate": 4.9985393111778525e-05,
+      "loss": 3.5332,
+      "step": 4500
+    },
+    {
+      "epoch": 0.03692007140733016,
+      "grad_norm": 0.5501233339309692,
+      "learning_rate": 4.998517320855884e-05,
+      "loss": 3.5485,
+      "step": 4530
+    },
+    {
+      "epoch": 0.03716457519148467,
+      "grad_norm": 0.5359978675842285,
+      "learning_rate": 4.998495166287887e-05,
+      "loss": 3.534,
+      "step": 4560
+    },
+    {
+      "epoch": 0.03740907897563917,
+      "grad_norm": 0.5447133183479309,
+      "learning_rate": 4.998472847475318e-05,
+      "loss": 3.5176,
+      "step": 4590
+    },
+    {
+      "epoch": 0.03765358275979368,
+      "grad_norm": 0.5201069712638855,
+      "learning_rate": 4.998450364419643e-05,
+      "loss": 3.518,
+      "step": 4620
+    },
+    {
+      "epoch": 0.03789808654394818,
+      "grad_norm": 0.5251840353012085,
+      "learning_rate": 4.998427717122342e-05,
+      "loss": 3.5021,
+      "step": 4650
+    },
+    {
+      "epoch": 0.03814259032810268,
+      "grad_norm": 0.5204648375511169,
+      "learning_rate": 4.9984049055849024e-05,
+      "loss": 3.505,
+      "step": 4680
+    },
+    {
+      "epoch": 0.038387094112257186,
+      "grad_norm": 0.5137141942977905,
+      "learning_rate": 4.9983819298088234e-05,
+      "loss": 3.4997,
+      "step": 4710
+    },
+    {
+      "epoch": 0.03863159789641169,
+      "grad_norm": 0.5232805609703064,
+      "learning_rate": 4.9983587897956166e-05,
+      "loss": 3.5049,
+      "step": 4740
+    },
+    {
+      "epoch": 0.0388761016805662,
+      "grad_norm": 0.5415229201316833,
+      "learning_rate": 4.998335485546802e-05,
+      "loss": 3.5123,
+      "step": 4770
+    },
+    {
+      "epoch": 0.0391206054647207,
+      "grad_norm": 0.5097187161445618,
+      "learning_rate": 4.998312017063912e-05,
+      "loss": 3.4839,
+      "step": 4800
+    },
+    {
+      "epoch": 0.03936510924887521,
+      "grad_norm": 0.5637670755386353,
+      "learning_rate": 4.9982883843484895e-05,
+      "loss": 3.5084,
+      "step": 4830
+    },
+    {
+      "epoch": 0.03960961303302971,
+      "grad_norm": 0.5436129570007324,
+      "learning_rate": 4.998264587402088e-05,
+      "loss": 3.5184,
+      "step": 4860
+    },
+    {
+      "epoch": 0.03985411681718422,
+      "grad_norm": 0.5133230090141296,
+      "learning_rate": 4.9982406262262726e-05,
+      "loss": 3.48,
+      "step": 4890
+    },
+    {
+      "epoch": 0.040098620601338716,
+      "grad_norm": 0.5678510069847107,
+      "learning_rate": 4.9982165008226175e-05,
+      "loss": 3.4899,
+      "step": 4920
+    },
+    {
+      "epoch": 0.04034312438549322,
+      "grad_norm": 0.5121597647666931,
+      "learning_rate": 4.998192211192708e-05,
+      "loss": 3.4942,
+      "step": 4950
+    },
+    {
+      "epoch": 0.04058762816964773,
+      "grad_norm": 0.537929117679596,
+      "learning_rate": 4.9981677573381423e-05,
+      "loss": 3.4721,
+      "step": 4980
+    },
+    {
+      "epoch": 0.04083213195380223,
+      "grad_norm": 0.537289023399353,
+      "learning_rate": 4.9981431392605274e-05,
+      "loss": 3.4919,
+      "step": 5010
+    },
+    {
+      "epoch": 0.04107663573795674,
+      "grad_norm": 0.5594364404678345,
+      "learning_rate": 4.998118356961481e-05,
+      "loss": 3.4665,
+      "step": 5040
+    },
+    {
+      "epoch": 0.04132113952211124,
+      "grad_norm": 0.5009840726852417,
+      "learning_rate": 4.998093410442632e-05,
+      "loss": 3.4524,
+      "step": 5070
+    },
+    {
+      "epoch": 0.04156564330626575,
+      "grad_norm": 0.53801429271698,
+      "learning_rate": 4.998068299705623e-05,
+      "loss": 3.4629,
+      "step": 5100
+    },
+    {
+      "epoch": 0.041810147090420254,
+      "grad_norm": 0.5134426355361938,
+      "learning_rate": 4.9980430247521014e-05,
+      "loss": 3.4504,
+      "step": 5130
+    },
+    {
+      "epoch": 0.04205465087457475,
+      "grad_norm": 0.5095092058181763,
+      "learning_rate": 4.998017585583731e-05,
+      "loss": 3.4684,
+      "step": 5160
+    },
+    {
+      "epoch": 0.04229915465872926,
+      "grad_norm": 0.5233584642410278,
+      "learning_rate": 4.9979919822021824e-05,
+      "loss": 3.4726,
+      "step": 5190
+    },
+    {
+      "epoch": 0.04254365844288376,
+      "grad_norm": 0.5375773906707764,
+      "learning_rate": 4.99796621460914e-05,
+      "loss": 3.4502,
+      "step": 5220
+    },
+    {
+      "epoch": 0.04278816222703827,
+      "grad_norm": 0.5292348265647888,
+      "learning_rate": 4.9979402828062963e-05,
+      "loss": 3.4494,
+      "step": 5250
+    },
+    {
+      "epoch": 0.04303266601119277,
+      "grad_norm": 0.5170340538024902,
+      "learning_rate": 4.997914186795358e-05,
+      "loss": 3.4671,
+      "step": 5280
+    },
+    {
+      "epoch": 0.04327716979534728,
+      "grad_norm": 0.5027357339859009,
+      "learning_rate": 4.9978879265780385e-05,
+      "loss": 3.4424,
+      "step": 5310
+    },
+    {
+      "epoch": 0.043521673579501784,
+      "grad_norm": 0.4898991584777832,
+      "learning_rate": 4.997861502156066e-05,
+      "loss": 3.4515,
+      "step": 5340
+    },
+    {
+      "epoch": 0.04376617736365628,
+      "grad_norm": 0.5362656712532043,
+      "learning_rate": 4.997834913531176e-05,
+      "loss": 3.4209,
+      "step": 5370
+    },
+    {
+      "epoch": 0.04401068114781079,
+      "grad_norm": 0.5150277614593506,
+      "learning_rate": 4.9978081607051176e-05,
+      "loss": 3.4575,
+      "step": 5400
+    },
+    {
+      "epoch": 0.04425518493196529,
+      "grad_norm": 0.5086013674736023,
+      "learning_rate": 4.997781243679648e-05,
+      "loss": 3.4377,
+      "step": 5430
+    },
+    {
+      "epoch": 0.0444996887161198,
+      "grad_norm": 0.538194477558136,
+      "learning_rate": 4.9977541624565374e-05,
+      "loss": 3.4282,
+      "step": 5460
+    },
+    {
+      "epoch": 0.0447441925002743,
+      "grad_norm": 0.5045614838600159,
+      "learning_rate": 4.9977269170375665e-05,
+      "loss": 3.4109,
+      "step": 5490
+    },
+    {
+      "epoch": 0.04498869628442881,
+      "grad_norm": 0.5368480086326599,
+      "learning_rate": 4.997699507424526e-05,
+      "loss": 3.4341,
+      "step": 5520
+    },
+    {
+      "epoch": 0.045233200068583314,
+      "grad_norm": 0.6185830235481262,
+      "learning_rate": 4.997671933619218e-05,
+      "loss": 3.4203,
+      "step": 5550
+    },
+    {
+      "epoch": 0.04547770385273782,
+      "grad_norm": 0.4984918534755707,
+      "learning_rate": 4.9976441956234546e-05,
+      "loss": 3.4309,
+      "step": 5580
+    },
+    {
+      "epoch": 0.04572220763689232,
+      "grad_norm": 0.5066754221916199,
+      "learning_rate": 4.99761629343906e-05,
+      "loss": 3.3986,
+      "step": 5610
+    },
+    {
+      "epoch": 0.04596671142104682,
+      "grad_norm": 0.5132448673248291,
+      "learning_rate": 4.9975882270678676e-05,
+      "loss": 3.4126,
+      "step": 5640
+    },
+    {
+      "epoch": 0.04621121520520133,
+      "grad_norm": 0.5501627922058105,
+      "learning_rate": 4.997559996511723e-05,
+      "loss": 3.4057,
+      "step": 5670
+    },
+    {
+      "epoch": 0.04645571898935583,
+      "grad_norm": 0.4679185152053833,
+      "learning_rate": 4.997531601772481e-05,
+      "loss": 3.428,
+      "step": 5700
+    },
+    {
+      "epoch": 0.04670022277351034,
+      "grad_norm": 0.5020308494567871,
+      "learning_rate": 4.9975030428520084e-05,
+      "loss": 3.3977,
+      "step": 5730
+    },
+    {
+      "epoch": 0.046944726557664844,
+      "grad_norm": 0.5141638517379761,
+      "learning_rate": 4.997474319752184e-05,
+      "loss": 3.4044,
+      "step": 5760
+    },
+    {
+      "epoch": 0.04718923034181935,
+      "grad_norm": 0.5127893090248108,
+      "learning_rate": 4.997445432474895e-05,
+      "loss": 3.4071,
+      "step": 5790
+    },
+    {
+      "epoch": 0.04743373412597385,
+      "grad_norm": 0.5528402924537659,
+      "learning_rate": 4.9974163810220406e-05,
+      "loss": 3.3891,
+      "step": 5820
+    },
+    {
+      "epoch": 0.04767823791012835,
+      "grad_norm": 0.5092859864234924,
+      "learning_rate": 4.99738716539553e-05,
+      "loss": 3.4118,
+      "step": 5850
+    },
+    {
+      "epoch": 0.04792274169428286,
+      "grad_norm": 0.49626457691192627,
+      "learning_rate": 4.997357785597284e-05,
+      "loss": 3.3834,
+      "step": 5880
+    },
+    {
+      "epoch": 0.048167245478437364,
+      "grad_norm": 0.5132192969322205,
+      "learning_rate": 4.997328241629234e-05,
+      "loss": 3.3944,
+      "step": 5910
+    },
+    {
+      "epoch": 0.04841174926259187,
+      "grad_norm": 0.5104256868362427,
+      "learning_rate": 4.997298533493323e-05,
+      "loss": 3.4001,
+      "step": 5940
+    },
+    {
+      "epoch": 0.048656253046746374,
+      "grad_norm": 0.507205069065094,
+      "learning_rate": 4.997268661191503e-05,
+      "loss": 3.3867,
+      "step": 5970
+    },
+    {
+      "epoch": 0.04890075683090088,
+      "grad_norm": 0.6091153025627136,
+      "learning_rate": 4.9972386247257385e-05,
+      "loss": 3.3984,
+      "step": 6000
+    },
+    {
+      "epoch": 0.049145260615055385,
+      "grad_norm": 0.4803309142589569,
+      "learning_rate": 4.9972084240980025e-05,
+      "loss": 3.3703,
+      "step": 6030
+    },
+    {
+      "epoch": 0.04938976439920988,
+      "grad_norm": 0.515164852142334,
+      "learning_rate": 4.997178059310281e-05,
+      "loss": 3.3733,
+      "step": 6060
+    },
+    {
+      "epoch": 0.04963426818336439,
+      "grad_norm": 0.515275776386261,
+      "learning_rate": 4.997147530364571e-05,
+      "loss": 3.3802,
+      "step": 6090
+    },
+    {
+      "epoch": 0.049878771967518894,
+      "grad_norm": 0.5258405208587646,
+      "learning_rate": 4.9971168372628793e-05,
+      "loss": 3.3705,
+      "step": 6120
+    },
+    {
+      "epoch": 0.0501232757516734,
+      "grad_norm": 0.49037066102027893,
+      "learning_rate": 4.997085980007222e-05,
+      "loss": 3.37,
+      "step": 6150
+    },
+    {
+      "epoch": 0.050367779535827904,
+      "grad_norm": 0.47182220220565796,
+      "learning_rate": 4.99705495859963e-05,
+      "loss": 3.3705,
+      "step": 6180
+    },
+    {
+      "epoch": 0.05061228331998241,
+      "grad_norm": 0.4967211186885834,
+      "learning_rate": 4.99702377304214e-05,
+      "loss": 3.3743,
+      "step": 6210
+    },
+    {
+      "epoch": 0.050856787104136915,
+      "grad_norm": 0.7469291090965271,
+      "learning_rate": 4.9969924233368036e-05,
+      "loss": 3.3732,
+      "step": 6240
+    },
+    {
+      "epoch": 0.05110129088829142,
+      "grad_norm": 0.5237293839454651,
+      "learning_rate": 4.996960909485681e-05,
+      "loss": 3.3646,
+      "step": 6270
+    },
+    {
+      "epoch": 0.05134579467244592,
+      "grad_norm": 0.48839274048805237,
+      "learning_rate": 4.9969292314908446e-05,
+      "loss": 3.3665,
+      "step": 6300
+    },
+    {
+      "epoch": 0.051590298456600424,
+      "grad_norm": 0.48733407258987427,
+      "learning_rate": 4.996897389354376e-05,
+      "loss": 3.3646,
+      "step": 6330
+    },
+    {
+      "epoch": 0.05183480224075493,
+      "grad_norm": 0.4938340187072754,
+      "learning_rate": 4.9968653830783684e-05,
+      "loss": 3.3542,
+      "step": 6360
+    },
+    {
+      "epoch": 0.052079306024909434,
+      "grad_norm": 0.5001193284988403,
+      "learning_rate": 4.996833212664927e-05,
+      "loss": 3.3583,
+      "step": 6390
+    },
+    {
+      "epoch": 0.05232380980906394,
+      "grad_norm": 0.4772029221057892,
+      "learning_rate": 4.996800878116166e-05,
+      "loss": 3.336,
+      "step": 6420
+    },
+    {
+      "epoch": 0.052568313593218445,
+      "grad_norm": 0.49077659845352173,
+      "learning_rate": 4.99676837943421e-05,
+      "loss": 3.3681,
+      "step": 6450
+    },
+    {
+      "epoch": 0.05281281737737295,
+      "grad_norm": 0.5102148056030273,
+      "learning_rate": 4.996735716621196e-05,
+      "loss": 3.3437,
+      "step": 6480
+    },
+    {
+      "epoch": 0.05305732116152745,
+      "grad_norm": 0.5012289881706238,
+      "learning_rate": 4.996702889679272e-05,
+      "loss": 3.3536,
+      "step": 6510
+    },
+    {
+      "epoch": 0.053301824945681954,
+      "grad_norm": 0.4870162308216095,
+      "learning_rate": 4.996669898610595e-05,
+      "loss": 3.3513,
+      "step": 6540
+    },
+    {
+      "epoch": 0.05354632872983646,
+      "grad_norm": 0.48452699184417725,
+      "learning_rate": 4.996636743417334e-05,
+      "loss": 3.3415,
+      "step": 6570
+    },
+    {
+      "epoch": 0.053790832513990965,
+      "grad_norm": 0.4973999559879303,
+      "learning_rate": 4.996603424101669e-05,
+      "loss": 3.3304,
+      "step": 6600
+    },
+    {
+      "epoch": 0.05403533629814547,
+      "grad_norm": 0.4833717942237854,
+      "learning_rate": 4.996569940665789e-05,
+      "loss": 3.3424,
+      "step": 6630
+    },
+    {
+      "epoch": 0.054279840082299975,
+      "grad_norm": 0.5074206590652466,
+      "learning_rate": 4.996536293111896e-05,
+      "loss": 3.3302,
+      "step": 6660
+    },
+    {
+      "epoch": 0.05452434386645448,
+      "grad_norm": 0.4932290315628052,
+      "learning_rate": 4.996502481442202e-05,
+      "loss": 3.3388,
+      "step": 6690
+    },
+    {
+      "epoch": 0.054768847650608986,
+      "grad_norm": 0.48740679025650024,
+      "learning_rate": 4.9964685056589314e-05,
+      "loss": 3.3182,
+      "step": 6720
+    },
+    {
+      "epoch": 0.055013351434763484,
+      "grad_norm": 0.48752760887145996,
+      "learning_rate": 4.996434365764314e-05,
+      "loss": 3.3065,
+      "step": 6750
+    },
+    {
+      "epoch": 0.05525785521891799,
+      "grad_norm": 0.50692218542099,
+      "learning_rate": 4.996400061760597e-05,
+      "loss": 3.3379,
+      "step": 6780
+    },
+    {
+      "epoch": 0.055502359003072495,
+      "grad_norm": 0.479159414768219,
+      "learning_rate": 4.996365593650033e-05,
+      "loss": 3.3317,
+      "step": 6810
+    },
+    {
+      "epoch": 0.055746862787227,
+      "grad_norm": 0.498662531375885,
+      "learning_rate": 4.99633096143489e-05,
+      "loss": 3.3306,
+      "step": 6840
+    },
+    {
+      "epoch": 0.055991366571381505,
+      "grad_norm": 1.4371449947357178,
+      "learning_rate": 4.9962961651174436e-05,
+      "loss": 3.3334,
+      "step": 6870
+    },
+    {
+      "epoch": 0.05623587035553601,
+      "grad_norm": 0.49862873554229736,
+      "learning_rate": 4.9962612046999827e-05,
+      "loss": 3.3142,
+      "step": 6900
+    },
+    {
+      "epoch": 0.056480374139690516,
+      "grad_norm": 0.4759610593318939,
+      "learning_rate": 4.996226080184803e-05,
+      "loss": 3.3238,
+      "step": 6930
+    },
+    {
+      "epoch": 0.056724877923845014,
+      "grad_norm": 0.4844242334365845,
+      "learning_rate": 4.996190791574215e-05,
+      "loss": 3.3197,
+      "step": 6960
+    },
+    {
+      "epoch": 0.05696938170799952,
+      "grad_norm": 0.46844130754470825,
+      "learning_rate": 4.996155338870538e-05,
+      "loss": 3.2949,
+      "step": 6990
+    },
+    {
+      "epoch": 0.057213885492154025,
+      "grad_norm": 0.4850478768348694,
+      "learning_rate": 4.9961197220761035e-05,
+      "loss": 3.3143,
+      "step": 7020
+    },
+    {
+      "epoch": 0.05745838927630853,
+      "grad_norm": 0.4838846027851105,
+      "learning_rate": 4.996083941193252e-05,
+      "loss": 3.3015,
+      "step": 7050
+    },
+    {
+      "epoch": 0.057702893060463036,
+      "grad_norm": 0.49992483854293823,
+      "learning_rate": 4.9960479962243367e-05,
+      "loss": 3.3099,
+      "step": 7080
+    },
+    {
+      "epoch": 0.05794739684461754,
+      "grad_norm": 0.49964553117752075,
+      "learning_rate": 4.996011887171719e-05,
+      "loss": 3.3046,
+      "step": 7110
+    },
+    {
+      "epoch": 0.058191900628772046,
+      "grad_norm": 0.4723115563392639,
+      "learning_rate": 4.995975614037773e-05,
+      "loss": 3.3009,
+      "step": 7140
+    },
+    {
+      "epoch": 0.05843640441292655,
+      "grad_norm": 0.48575958609580994,
+      "learning_rate": 4.995939176824883e-05,
+      "loss": 3.3018,
+      "step": 7170
+    },
+    {
+      "epoch": 0.05868090819708105,
+      "grad_norm": 0.5264491438865662,
+      "learning_rate": 4.995902575535446e-05,
+      "loss": 3.2877,
+      "step": 7200
+    }
+  ],
+  "logging_steps": 30,
+  "max_steps": 368091,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "total_flos": 6.887340836388864e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f909d8205eae6d012dc965e337249e1d014ef52c5b7115301085202b50cb6566
+size 4984