Upload folder using huggingface_hub

config.json

@@ -0,0 +1,29 @@
+{
+  "_name_or_path": "/paratera5-data/public/chai_models/phase2_winner_13b2",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 5120,
+  "initializer_range": 0.02,
+  "intermediate_size": 13824,
+  "max_position_embeddings": 4096,
+  "model_type": "llama",
+  "num_attention_heads": 40,
+  "num_hidden_layers": 40,
+  "num_key_value_heads": 40,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.37.0.dev0",
+  "use_cache": false,
+  "vocab_size": 32000
+}

latest

@@ -0,0 +1 @@
+global_step600

model-00001-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66ffb8d1d341d9c1273fc07b026f703fb4da25f169fa46d309eb8faec194e6c1
+size 4978265800

model-00002-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:310375d29fe16203ab88f4f16927dd83f6cd3f9c00828876459ab84cdb44af09
+size 4970422232

model-00003-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f262efeabbc558c4548f446b8a79d62741f4f40f75d3ea0a7a2c496542a65eb
+size 4970422256

model-00004-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d13d42545ce444841c28869646f39fbe57a7c3ffe669cc333b66e9ba530844b
+size 4933701504

model-00005-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82ad7a6569d0f299d76f1d18c07bdf731bbde971b1fa6cf357f49a8e2a8a741e
+size 4933722216

model-00006-of-00006.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:851ac04d9042b614aafff4ba6a4aadc6b0e845a5b249a8af746addb9fc4fd1f9
+size 1245236920

model.safetensors.index.json

@@ -0,0 +1,370 @@
+{
+  "metadata": {
+    "total_size": 26031728640
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00006-of-00006.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.input_layernorm.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.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_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.input_layernorm.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.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_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-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-00002-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.input_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00006.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.15.input_layernorm.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.self_attn.k_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.post_attention_layernorm.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_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-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.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_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00006.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00006.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00004-of-00006.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-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.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_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00004-of-00006.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00005-of-00006.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00005-of-00006.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00006-of-00006.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00006-of-00006.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-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-00001-of-00006.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00006.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.7.input_layernorm.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.self_attn.k_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00006.safetensors",
+    "model.layers.8.input_layernorm.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.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_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00006.safetensors",
+    "model.layers.9.input_layernorm.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.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_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"
+  }
+}

rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9917f57cf6db9ff973eea259725600ee48691f1a311a86c35d7ddf8e83d23e8f
+size 15984

rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0cc4f70c70839703ce970c1ee45417022e0e83b20bfb2f7333a736bc535e8ea2
+size 15984

rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:538d096ef85541a907692e7ea5870361f3c33d47d2ae36cc8d80fcbee6d6ba67
+size 15984

rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d89a04b316fca29777647ac11482d222694dc8c9b7f28dd628a2a2b636f93048
+size 15984

rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0984343f1736954604090ffc2da611b7513bda5a8c67f3c5f4fef651f2645f48
+size 15984

rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:85c4361c193c2496c66013365189651bdfde23b29e0b3ff425414826ff253d97
+size 15984

rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5dc54ccbf26a6adc177b6b75ed0989fccb1dc837719258579975ddb0a14c1a3e
+size 15984

rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f0cc3f27f578251164d5f20b22210ec59eee8ce74548c0b21bfd06c1cd01224
+size 15984

special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "</s>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723

tokenizer_config.json

@@ -0,0 +1,43 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "split_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

trainer_state.json

@@ -0,0 +1,3717 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9127316198445905,
+  "eval_steps": 50,
+  "global_step": 600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-07,
+      "loss": 1.171,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-07,
+      "loss": 1.1933,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5e-07,
+      "loss": 1.1931,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-06,
+      "loss": 1.2064,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2499999999999999e-06,
+      "loss": 1.2553,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5e-06,
+      "loss": 1.1993,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.75e-06,
+      "loss": 1.169,
+      "step": 7
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-06,
+      "loss": 1.136,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.25e-06,
+      "loss": 1.0732,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4999999999999998e-06,
+      "loss": 1.0262,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.75e-06,
+      "loss": 0.9988,
+      "step": 11
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3e-06,
+      "loss": 1.0119,
+      "step": 12
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.25e-06,
+      "loss": 0.9621,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5e-06,
+      "loss": 0.9128,
+      "step": 14
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.7499999999999997e-06,
+      "loss": 0.8979,
+      "step": 15
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4e-06,
+      "loss": 0.8497,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.25e-06,
+      "loss": 0.8423,
+      "step": 17
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.5e-06,
+      "loss": 0.8535,
+      "step": 18
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.749999999999999e-06,
+      "loss": 0.8679,
+      "step": 19
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9999999999999996e-06,
+      "loss": 0.8042,
+      "step": 20
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.25e-06,
+      "loss": 0.8447,
+      "step": 21
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.5e-06,
+      "loss": 0.8084,
+      "step": 22
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.75e-06,
+      "loss": 0.7878,
+      "step": 23
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6e-06,
+      "loss": 0.8129,
+      "step": 24
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.2499999999999995e-06,
+      "loss": 0.7616,
+      "step": 25
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.5e-06,
+      "loss": 0.7954,
+      "step": 26
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.75e-06,
+      "loss": 0.776,
+      "step": 27
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7e-06,
+      "loss": 0.7494,
+      "step": 28
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.25e-06,
+      "loss": 0.7211,
+      "step": 29
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.499999999999999e-06,
+      "loss": 0.7321,
+      "step": 30
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.75e-06,
+      "loss": 0.7115,
+      "step": 31
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8e-06,
+      "loss": 0.7295,
+      "step": 32
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.999944055700695e-06,
+      "loss": 0.7143,
+      "step": 33
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.999776224367659e-06,
+      "loss": 0.7092,
+      "step": 34
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.999496510695501e-06,
+      "loss": 0.7071,
+      "step": 35
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.999104922508408e-06,
+      "loss": 0.7302,
+      "step": 36
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.998601470759944e-06,
+      "loss": 0.7338,
+      "step": 37
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.99798616953274e-06,
+      "loss": 0.7018,
+      "step": 38
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.997259036038092e-06,
+      "loss": 0.7041,
+      "step": 39
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.996420090615486e-06,
+      "loss": 0.7133,
+      "step": 40
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.995469356732032e-06,
+      "loss": 0.7142,
+      "step": 41
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.994406860981797e-06,
+      "loss": 0.7124,
+      "step": 42
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.993232633085074e-06,
+      "loss": 0.6892,
+      "step": 43
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.991946705887537e-06,
+      "loss": 0.6985,
+      "step": 44
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.99054911535934e-06,
+      "loss": 0.6861,
+      "step": 45
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.989039900594089e-06,
+      "loss": 0.7207,
+      "step": 46
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.98741910380777e-06,
+      "loss": 0.698,
+      "step": 47
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.98568677033755e-06,
+      "loss": 0.7224,
+      "step": 48
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.98384294864052e-06,
+      "loss": 0.6799,
+      "step": 49
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.981887690292338e-06,
+      "loss": 0.6708,
+      "step": 50
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 0.6769059300422668,
+      "eval_runtime": 20.2812,
+      "eval_samples_per_second": 40.037,
+      "eval_steps_per_second": 2.515,
+      "step": 50
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.979821049985783e-06,
+      "loss": 0.687,
+      "step": 51
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.977643085529227e-06,
+      "loss": 0.6902,
+      "step": 52
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.975353857845016e-06,
+      "loss": 0.708,
+      "step": 53
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.972953430967771e-06,
+      "loss": 0.6835,
+      "step": 54
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.970441872042592e-06,
+      "loss": 0.6909,
+      "step": 55
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.96781925132318e-06,
+      "loss": 0.6696,
+      "step": 56
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.965085642169875e-06,
+      "loss": 0.681,
+      "step": 57
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.962241121047602e-06,
+      "loss": 0.6629,
+      "step": 58
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.95928576752373e-06,
+      "loss": 0.6938,
+      "step": 59
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.956219664265852e-06,
+      "loss": 0.6173,
+      "step": 60
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.953042897039465e-06,
+      "loss": 0.6545,
+      "step": 61
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.949755554705577e-06,
+      "loss": 0.6837,
+      "step": 62
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.946357729218223e-06,
+      "loss": 0.663,
+      "step": 63
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.942849515621881e-06,
+      "loss": 0.6456,
+      "step": 64
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.939231012048832e-06,
+      "loss": 0.6521,
+      "step": 65
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.935502319716397e-06,
+      "loss": 0.631,
+      "step": 66
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.931663542924115e-06,
+      "loss": 0.6769,
+      "step": 67
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.927714789050827e-06,
+      "loss": 0.6821,
+      "step": 68
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.923656168551665e-06,
+      "loss": 0.6559,
+      "step": 69
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.919487794954972e-06,
+      "loss": 0.669,
+      "step": 70
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.915209784859116e-06,
+      "loss": 0.6716,
+      "step": 71
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.910822257929234e-06,
+      "loss": 0.6727,
+      "step": 72
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.90632533689389e-06,
+      "loss": 0.68,
+      "step": 73
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.901719147541628e-06,
+      "loss": 0.649,
+      "step": 74
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.897003818717468e-06,
+      "loss": 0.6428,
+      "step": 75
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.892179482319294e-06,
+      "loss": 0.6429,
+      "step": 76
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.887246273294166e-06,
+      "loss": 0.6429,
+      "step": 77
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.882204329634543e-06,
+      "loss": 0.6439,
+      "step": 78
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.87705379237443e-06,
+      "loss": 0.6288,
+      "step": 79
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.871794805585425e-06,
+      "loss": 0.6525,
+      "step": 80
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.866427516372693e-06,
+      "loss": 0.6498,
+      "step": 81
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.86095207487085e-06,
+      "loss": 0.6366,
+      "step": 82
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.855368634239768e-06,
+      "loss": 0.6402,
+      "step": 83
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.849677350660282e-06,
+      "loss": 0.6704,
+      "step": 84
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.843878383329828e-06,
+      "loss": 0.6417,
+      "step": 85
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.837971894457989e-06,
+      "loss": 0.6146,
+      "step": 86
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.831958049261955e-06,
+      "loss": 0.6307,
+      "step": 87
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.825837015961904e-06,
+      "loss": 0.6371,
+      "step": 88
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.819608965776295e-06,
+      "loss": 0.6155,
+      "step": 89
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.813274072917081e-06,
+      "loss": 0.6316,
+      "step": 90
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.806832514584832e-06,
+      "loss": 0.627,
+      "step": 91
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.800284470963781e-06,
+      "loss": 0.6345,
+      "step": 92
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.793630125216786e-06,
+      "loss": 0.6448,
+      "step": 93
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.786869663480201e-06,
+      "loss": 0.6185,
+      "step": 94
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.780003274858673e-06,
+      "loss": 0.6294,
+      "step": 95
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.773031151419853e-06,
+      "loss": 0.6362,
+      "step": 96
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.765953488189021e-06,
+      "loss": 0.6586,
+      "step": 97
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.758770483143633e-06,
+      "loss": 0.627,
+      "step": 98
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.75148233720778e-06,
+      "loss": 0.6331,
+      "step": 99
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.744089254246569e-06,
+      "loss": 0.6041,
+      "step": 100
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 0.616474986076355,
+      "eval_runtime": 20.2693,
+      "eval_samples_per_second": 40.061,
+      "eval_steps_per_second": 2.516,
+      "step": 100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.736591441060427e-06,
+      "loss": 0.6055,
+      "step": 101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.728989107379303e-06,
+      "loss": 0.607,
+      "step": 102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.721282465856814e-06,
+      "loss": 0.6246,
+      "step": 103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.71347173206429e-06,
+      "loss": 0.5974,
+      "step": 104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.705557124484746e-06,
+      "loss": 0.5991,
+      "step": 105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.697538864506767e-06,
+      "loss": 0.612,
+      "step": 106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.689417176418326e-06,
+      "loss": 0.6069,
+      "step": 107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681192287400491e-06,
+      "loss": 0.6163,
+      "step": 108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.672864427521097e-06,
+      "loss": 0.588,
+      "step": 109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.664433829728277e-06,
+      "loss": 0.622,
+      "step": 110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.655900729843983e-06,
+      "loss": 0.6169,
+      "step": 111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.647265366557355e-06,
+      "loss": 0.6326,
+      "step": 112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.638527981418074e-06,
+      "loss": 0.6054,
+      "step": 113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.629688818829577e-06,
+      "loss": 0.577,
+      "step": 114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.620748126042249e-06,
+      "loss": 0.624,
+      "step": 115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.611706153146485e-06,
+      "loss": 0.6018,
+      "step": 116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.602563153065702e-06,
+      "loss": 0.5857,
+      "step": 117
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.5933193815492675e-06,
+      "loss": 0.5735,
+      "step": 118
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.583975097165343e-06,
+      "loss": 0.5837,
+      "step": 119
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.574530561293649e-06,
+      "loss": 0.5696,
+      "step": 120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.564986038118155e-06,
+      "loss": 0.5989,
+      "step": 121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.555341794619694e-06,
+      "loss": 0.6055,
+      "step": 122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.545598100568487e-06,
+      "loss": 0.6071,
+      "step": 123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.535755228516601e-06,
+      "loss": 0.5903,
+      "step": 124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.525813453790328e-06,
+      "loss": 0.6042,
+      "step": 125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.515773054482478e-06,
+      "loss": 0.6211,
+      "step": 126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.505634311444602e-06,
+      "loss": 0.5919,
+      "step": 127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.49539750827914e-06,
+      "loss": 0.5752,
+      "step": 128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.48506293133148e-06,
+      "loss": 0.5936,
+      "step": 129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.474630869681954e-06,
+      "loss": 0.6023,
+      "step": 130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.4641016151377545e-06,
+      "loss": 0.5851,
+      "step": 131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.453475462224763e-06,
+      "loss": 0.5902,
+      "step": 132
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.44275270817932e-06,
+      "loss": 0.5633,
+      "step": 133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.431933652939908e-06,
+      "loss": 0.5855,
+      "step": 134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.421018599138756e-06,
+      "loss": 0.5862,
+      "step": 135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.410007852093384e-06,
+      "loss": 0.5414,
+      "step": 136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.398901719798058e-06,
+      "loss": 0.572,
+      "step": 137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.387700512915168e-06,
+      "loss": 0.5721,
+      "step": 138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.376404544766554e-06,
+      "loss": 0.5717,
+      "step": 139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.365014131324725e-06,
+      "loss": 0.5958,
+      "step": 140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.353529591204029e-06,
+      "loss": 0.5773,
+      "step": 141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.3419512456517455e-06,
+      "loss": 0.5705,
+      "step": 142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.330279418539084e-06,
+      "loss": 0.5706,
+      "step": 143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.3185144363521435e-06,
+      "loss": 0.6,
+      "step": 144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.306656628182765e-06,
+      "loss": 0.5863,
+      "step": 145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.294706325719331e-06,
+      "loss": 0.5591,
+      "step": 146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.2826638632374924e-06,
+      "loss": 0.5788,
+      "step": 147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.270529577590812e-06,
+      "loss": 0.6078,
+      "step": 148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.258303808201343e-06,
+      "loss": 0.5722,
+      "step": 149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.245986897050137e-06,
+      "loss": 0.5855,
+      "step": 150
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 0.5726646780967712,
+      "eval_runtime": 20.2436,
+      "eval_samples_per_second": 40.111,
+      "eval_steps_per_second": 2.519,
+      "step": 150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.233579188667675e-06,
+      "loss": 0.5824,
+      "step": 151
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.2210810301242345e-06,
+      "loss": 0.5419,
+      "step": 152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.208492771020175e-06,
+      "loss": 0.5475,
+      "step": 153
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.195814763476164e-06,
+      "loss": 0.5657,
+      "step": 154
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.183047362123328e-06,
+      "loss": 0.5649,
+      "step": 155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.170190924093326e-06,
+      "loss": 0.5452,
+      "step": 156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.157245809008368e-06,
+      "loss": 0.5774,
+      "step": 157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.1442123789711495e-06,
+      "loss": 0.5858,
+      "step": 158
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.131090998554727e-06,
+      "loss": 0.5623,
+      "step": 159
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.117882034792315e-06,
+      "loss": 0.6064,
+      "step": 160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.104585857167027e-06,
+      "loss": 0.5734,
+      "step": 161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.0912028376015315e-06,
+      "loss": 0.5386,
+      "step": 162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.077733350447657e-06,
+      "loss": 0.5603,
+      "step": 163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.064177772475912e-06,
+      "loss": 0.5448,
+      "step": 164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.0505364828649524e-06,
+      "loss": 0.5496,
+      "step": 165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.036809863190972e-06,
+      "loss": 0.5467,
+      "step": 166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.022998297417033e-06,
+      "loss": 0.5632,
+      "step": 167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.0091021718823185e-06,
+      "loss": 0.5699,
+      "step": 168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.99512187529133e-06,
+      "loss": 0.5606,
+      "step": 169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.981057798703019e-06,
+      "loss": 0.5626,
+      "step": 170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.966910335519838e-06,
+      "loss": 0.5388,
+      "step": 171
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.952679881476746e-06,
+      "loss": 0.5598,
+      "step": 172
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.9383668346301325e-06,
+      "loss": 0.5981,
+      "step": 173
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.923971595346686e-06,
+      "loss": 0.5408,
+      "step": 174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.909494566292195e-06,
+      "loss": 0.5157,
+      "step": 175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.89493615242028e-06,
+      "loss": 0.5568,
+      "step": 176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.880296760961076e-06,
+      "loss": 0.5713,
+      "step": 177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.865576801409828e-06,
+      "loss": 0.5853,
+      "step": 178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.8507766855154515e-06,
+      "loss": 0.5758,
+      "step": 179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.8358968272689995e-06,
+      "loss": 0.546,
+      "step": 180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.820937642892096e-06,
+      "loss": 0.5275,
+      "step": 181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.805899550825285e-06,
+      "loss": 0.5285,
+      "step": 182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.790782971716327e-06,
+      "loss": 0.5554,
+      "step": 183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.775588328408435e-06,
+      "loss": 0.534,
+      "step": 184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.760316045928448e-06,
+      "loss": 0.5305,
+      "step": 185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.744966551474935e-06,
+      "loss": 0.5357,
+      "step": 186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.7295402744062504e-06,
+      "loss": 0.5448,
+      "step": 187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.714037646228529e-06,
+      "loss": 0.5503,
+      "step": 188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.698459100583603e-06,
+      "loss": 0.5637,
+      "step": 189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.682805073236883e-06,
+      "loss": 0.5868,
+      "step": 190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6670760020651665e-06,
+      "loss": 0.5531,
+      "step": 191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.651272327044385e-06,
+      "loss": 0.5439,
+      "step": 192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.635394490237299e-06,
+      "loss": 0.5799,
+      "step": 193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.619442935781141e-06,
+      "loss": 0.5535,
+      "step": 194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603418109875174e-06,
+      "loss": 0.5494,
+      "step": 195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.58732046076823e-06,
+      "loss": 0.5546,
+      "step": 196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.571150438746157e-06,
+      "loss": 0.5761,
+      "step": 197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.554908496119232e-06,
+      "loss": 0.5544,
+      "step": 198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.538595087209504e-06,
+      "loss": 0.5644,
+      "step": 199
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.52221066833809e-06,
+      "loss": 0.5519,
+      "step": 200
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 0.5564766526222229,
+      "eval_runtime": 20.2313,
+      "eval_samples_per_second": 40.136,
+      "eval_steps_per_second": 2.521,
+      "step": 200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.505755697812405e-06,
+      "loss": 0.5609,
+      "step": 201
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.489230635913346e-06,
+      "loss": 0.5404,
+      "step": 202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.472635944882421e-06,
+      "loss": 0.5063,
+      "step": 203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.455972088908807e-06,
+      "loss": 0.5667,
+      "step": 204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.43923953411638e-06,
+      "loss": 0.5303,
+      "step": 205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.422438748550666e-06,
+      "loss": 0.5529,
+      "step": 206
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.405570202165754e-06,
+      "loss": 0.5481,
+      "step": 207
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.388634366811145e-06,
+      "loss": 0.5639,
+      "step": 208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.371631716218562e-06,
+      "loss": 0.5285,
+      "step": 209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.354562725988691e-06,
+      "loss": 0.5846,
+      "step": 210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.3374278735778834e-06,
+      "loss": 0.5814,
+      "step": 211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.3202276382847925e-06,
+      "loss": 0.525,
+      "step": 212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.302962501236975e-06,
+      "loss": 0.5504,
+      "step": 213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.285632945377429e-06,
+      "loss": 0.5549,
+      "step": 214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.268239455451082e-06,
+      "loss": 0.5411,
+      "step": 215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.250782517991241e-06,
+      "loss": 0.5251,
+      "step": 216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.2332626213059686e-06,
+      "loss": 0.5376,
+      "step": 217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.215680255464441e-06,
+      "loss": 0.5573,
+      "step": 218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.198035912283224e-06,
+      "loss": 0.5427,
+      "step": 219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.180330085312526e-06,
+      "loss": 0.561,
+      "step": 220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.162563269822391e-06,
+      "loss": 0.5625,
+      "step": 221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.144735962788837e-06,
+      "loss": 0.5555,
+      "step": 222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.126848662879967e-06,
+      "loss": 0.5658,
+      "step": 223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.108901870442009e-06,
+      "loss": 0.5642,
+      "step": 224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.09089608748533e-06,
+      "loss": 0.5413,
+      "step": 225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.072831817670382e-06,
+      "loss": 0.5394,
+      "step": 226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.054709566293626e-06,
+      "loss": 0.5397,
+      "step": 227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.036529840273388e-06,
+      "loss": 0.5701,
+      "step": 228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.018293148135684e-06,
+      "loss": 0.5632,
+      "step": 229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6e-06,
+      "loss": 0.5492,
+      "step": 230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.98165090756501e-06,
+      "loss": 0.5409,
+      "step": 231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.963246384094273e-06,
+      "loss": 0.5207,
+      "step": 232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.944786944401874e-06,
+      "loss": 0.5153,
+      "step": 233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.926273104838025e-06,
+      "loss": 0.5628,
+      "step": 234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.907705383274616e-06,
+      "loss": 0.5387,
+      "step": 235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.889084299090731e-06,
+      "loss": 0.552,
+      "step": 236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.870410373158125e-06,
+      "loss": 0.5473,
+      "step": 237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.85168412782665e-06,
+      "loss": 0.5455,
+      "step": 238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.832906086909641e-06,
+      "loss": 0.541,
+      "step": 239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.8140767756692705e-06,
+      "loss": 0.5405,
+      "step": 240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.795196720801849e-06,
+      "loss": 0.5421,
+      "step": 241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.7762664504230965e-06,
+      "loss": 0.5345,
+      "step": 242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.75728649405337e-06,
+      "loss": 0.5287,
+      "step": 243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.73825738260285e-06,
+      "loss": 0.5414,
+      "step": 244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7191796483566865e-06,
+      "loss": 0.5427,
+      "step": 245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.70005382496012e-06,
+      "loss": 0.5582,
+      "step": 246
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.680880447403544e-06,
+      "loss": 0.5183,
+      "step": 247
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.661660052007546e-06,
+      "loss": 0.5319,
+      "step": 248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.6423931764079e-06,
+      "loss": 0.53,
+      "step": 249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.623080359540536e-06,
+      "loss": 0.5378,
+      "step": 250
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.544645369052887,
+      "eval_runtime": 20.274,
+      "eval_samples_per_second": 40.051,
+      "eval_steps_per_second": 2.516,
+      "step": 250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.6037221416264545e-06,
+      "loss": 0.5536,
+      "step": 251
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.584319064156627e-06,
+      "loss": 0.5427,
+      "step": 252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.5648716698768395e-06,
+      "loss": 0.5224,
+      "step": 253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.545380502772514e-06,
+      "loss": 0.5579,
+      "step": 254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.525846108053495e-06,
+      "loss": 0.5457,
+      "step": 255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.506269032138794e-06,
+      "loss": 0.5763,
+      "step": 256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.48664982264131e-06,
+      "loss": 0.565,
+      "step": 257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4669890283525045e-06,
+      "loss": 0.5151,
+      "step": 258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.44728719922706e-06,
+      "loss": 0.5771,
+      "step": 259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.427544886367487e-06,
+      "loss": 0.5131,
+      "step": 260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.407762642008718e-06,
+      "loss": 0.5431,
+      "step": 261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.387941019502649e-06,
+      "loss": 0.5393,
+      "step": 262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.368080573302675e-06,
+      "loss": 0.5516,
+      "step": 263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.348181858948167e-06,
+      "loss": 0.5478,
+      "step": 264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.328245433048942e-06,
+      "loss": 0.5263,
+      "step": 265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.308271853269687e-06,
+      "loss": 0.5432,
+      "step": 266
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.288261678314367e-06,
+      "loss": 0.5477,
+      "step": 267
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.268215467910587e-06,
+      "loss": 0.5605,
+      "step": 268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.248133782793947e-06,
+      "loss": 0.536,
+      "step": 269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.228017184692351e-06,
+      "loss": 0.5339,
+      "step": 270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.207866236310286e-06,
+      "loss": 0.5691,
+      "step": 271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.1876815013131e-06,
+      "loss": 0.5627,
+      "step": 272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.167463544311218e-06,
+      "loss": 0.5429,
+      "step": 273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.147212930844361e-06,
+      "loss": 0.5142,
+      "step": 274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.126930227365719e-06,
+      "loss": 0.5284,
+      "step": 275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.106616001226108e-06,
+      "loss": 0.5293,
+      "step": 276
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.086270820658103e-06,
+      "loss": 0.5616,
+      "step": 277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.065895254760139e-06,
+      "loss": 0.5402,
+      "step": 278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.045489873480596e-06,
+      "loss": 0.5738,
+      "step": 279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.02505524760185e-06,
+      "loss": 0.5377,
+      "step": 280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.004591948724316e-06,
+      "loss": 0.543,
+      "step": 281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.984100549250454e-06,
+      "loss": 0.5374,
+      "step": 282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.963581622368753e-06,
+      "loss": 0.5351,
+      "step": 283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.943035742037709e-06,
+      "loss": 0.5339,
+      "step": 284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.922463482969761e-06,
+      "loss": 0.5258,
+      "step": 285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.901865420615218e-06,
+      "loss": 0.5288,
+      "step": 286
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.881242131146163e-06,
+      "loss": 0.5445,
+      "step": 287
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8605941914403355e-06,
+      "loss": 0.5147,
+      "step": 288
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.839922179064993e-06,
+      "loss": 0.5395,
+      "step": 289
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.819226672260763e-06,
+      "loss": 0.5912,
+      "step": 290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.798508249925456e-06,
+      "loss": 0.5475,
+      "step": 291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.777767491597883e-06,
+      "loss": 0.5369,
+      "step": 292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7570049774416405e-06,
+      "loss": 0.5359,
+      "step": 293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.736221288228881e-06,
+      "loss": 0.558,
+      "step": 294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.71541700532407e-06,
+      "loss": 0.5399,
+      "step": 295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.694592710667722e-06,
+      "loss": 0.5255,
+      "step": 296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.673748986760122e-06,
+      "loss": 0.5621,
+      "step": 297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.652886416645037e-06,
+      "loss": 0.5378,
+      "step": 298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6320055838934e-06,
+      "loss": 0.5279,
+      "step": 299
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.611107072586988e-06,
+      "loss": 0.5354,
+      "step": 300
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 0.5399758815765381,
+      "eval_runtime": 20.2602,
+      "eval_samples_per_second": 40.079,
+      "eval_steps_per_second": 2.517,
+      "step": 300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5901914673020886e-06,
+      "loss": 0.5329,
+      "step": 301
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.569259353093141e-06,
+      "loss": 0.5258,
+      "step": 302
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.548311315476378e-06,
+      "loss": 0.5317,
+      "step": 303
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.527347940413441e-06,
+      "loss": 0.5301,
+      "step": 304
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.506369814294997e-06,
+      "loss": 0.539,
+      "step": 305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.485377523924327e-06,
+      "loss": 0.5841,
+      "step": 306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.4643716565009205e-06,
+      "loss": 0.552,
+      "step": 307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.443352799604044e-06,
+      "loss": 0.5149,
+      "step": 308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.422321541176308e-06,
+      "loss": 0.5542,
+      "step": 309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.40127846950722e-06,
+      "loss": 0.5427,
+      "step": 310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.38022417321673e-06,
+      "loss": 0.5485,
+      "step": 311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.359159241238766e-06,
+      "loss": 0.5387,
+      "step": 312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.338084262804757e-06,
+      "loss": 0.5586,
+      "step": 313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3169998274271535e-06,
+      "loss": 0.4948,
+      "step": 314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.295906524882937e-06,
+      "loss": 0.4486,
+      "step": 315
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.274804945197124e-06,
+      "loss": 0.4642,
+      "step": 316
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2536956786262574e-06,
+      "loss": 0.4591,
+      "step": 317
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.232579315641903e-06,
+      "loss": 0.4295,
+      "step": 318
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.211456446914125e-06,
+      "loss": 0.4422,
+      "step": 319
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1903276632949695e-06,
+      "loss": 0.464,
+      "step": 320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.169193555801932e-06,
+      "loss": 0.4293,
+      "step": 321
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.148054715601433e-06,
+      "loss": 0.4368,
+      "step": 322
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.126911733992271e-06,
+      "loss": 0.4495,
+      "step": 323
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.105765202389093e-06,
+      "loss": 0.4396,
+      "step": 324
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.084615712305847e-06,
+      "loss": 0.4307,
+      "step": 325
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.063463855339232e-06,
+      "loss": 0.432,
+      "step": 326
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.042310223152158e-06,
+      "loss": 0.4475,
+      "step": 327
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0211554074571905e-06,
+      "loss": 0.4292,
+      "step": 328
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4e-06,
+      "loss": 0.4457,
+      "step": 329
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.978844592542808e-06,
+      "loss": 0.4325,
+      "step": 330
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.957689776847842e-06,
+      "loss": 0.4415,
+      "step": 331
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.936536144660768e-06,
+      "loss": 0.4464,
+      "step": 332
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.915384287694154e-06,
+      "loss": 0.4397,
+      "step": 333
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.894234797610906e-06,
+      "loss": 0.4458,
+      "step": 334
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.87308826600773e-06,
+      "loss": 0.4415,
+      "step": 335
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.851945284398567e-06,
+      "loss": 0.4255,
+      "step": 336
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.8308064441980675e-06,
+      "loss": 0.4191,
+      "step": 337
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.8096723367050306e-06,
+      "loss": 0.423,
+      "step": 338
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.7885435530858745e-06,
+      "loss": 0.4148,
+      "step": 339
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.7674206843580965e-06,
+      "loss": 0.4113,
+      "step": 340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.7463043213737418e-06,
+      "loss": 0.4287,
+      "step": 341
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.7251950548028763e-06,
+      "loss": 0.3999,
+      "step": 342
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.704093475117063e-06,
+      "loss": 0.4107,
+      "step": 343
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.6830001725728453e-06,
+      "loss": 0.4478,
+      "step": 344
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.661915737195243e-06,
+      "loss": 0.4519,
+      "step": 345
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.640840758761233e-06,
+      "loss": 0.4304,
+      "step": 346
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.61977582678327e-06,
+      "loss": 0.434,
+      "step": 347
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.59872153049278e-06,
+      "loss": 0.4349,
+      "step": 348
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.5776784588236934e-06,
+      "loss": 0.4222,
+      "step": 349
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.556647200395956e-06,
+      "loss": 0.4345,
+      "step": 350
+    },
+    {
+      "epoch": 1.12,
+      "eval_loss": 0.5446686744689941,
+      "eval_runtime": 20.2909,
+      "eval_samples_per_second": 40.018,
+      "eval_steps_per_second": 2.513,
+      "step": 350
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.5356283434990783e-06,
+      "loss": 0.4362,
+      "step": 351
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.5146224760756726e-06,
+      "loss": 0.4428,
+      "step": 352
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.4936301857050025e-06,
+      "loss": 0.436,
+      "step": 353
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.4726520595865585e-06,
+      "loss": 0.4456,
+      "step": 354
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.4516886845236217e-06,
+      "loss": 0.4599,
+      "step": 355
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.4307406469068596e-06,
+      "loss": 0.4351,
+      "step": 356
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.409808532697912e-06,
+      "loss": 0.4355,
+      "step": 357
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.3888929274130117e-06,
+      "loss": 0.4175,
+      "step": 358
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.3679944161066e-06,
+      "loss": 0.4451,
+      "step": 359
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.3471135833549626e-06,
+      "loss": 0.4228,
+      "step": 360
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.3262510132398783e-06,
+      "loss": 0.4121,
+      "step": 361
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.3054072893322785e-06,
+      "loss": 0.4258,
+      "step": 362
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.2845829946759312e-06,
+      "loss": 0.428,
+      "step": 363
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.263778711771119e-06,
+      "loss": 0.4413,
+      "step": 364
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.2429950225583604e-06,
+      "loss": 0.4413,
+      "step": 365
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.2222325084021166e-06,
+      "loss": 0.4311,
+      "step": 366
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.2014917500745437e-06,
+      "loss": 0.4339,
+      "step": 367
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.180773327739238e-06,
+      "loss": 0.4252,
+      "step": 368
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.1600778209350063e-06,
+      "loss": 0.4225,
+      "step": 369
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.1394058085596663e-06,
+      "loss": 0.4153,
+      "step": 370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.1187578688538373e-06,
+      "loss": 0.4357,
+      "step": 371
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.098134579384783e-06,
+      "loss": 0.4141,
+      "step": 372
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.077536517030239e-06,
+      "loss": 0.4339,
+      "step": 373
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0569642579622904e-06,
+      "loss": 0.43,
+      "step": 374
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0364183776312474e-06,
+      "loss": 0.433,
+      "step": 375
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.015899450749546e-06,
+      "loss": 0.4263,
+      "step": 376
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.995408051275683e-06,
+      "loss": 0.4269,
+      "step": 377
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9749447523981498e-06,
+      "loss": 0.4203,
+      "step": 378
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9545101265194048e-06,
+      "loss": 0.4418,
+      "step": 379
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9341047452398604e-06,
+      "loss": 0.4454,
+      "step": 380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9137291793418958e-06,
+      "loss": 0.4377,
+      "step": 381
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.893383998773892e-06,
+      "loss": 0.4703,
+      "step": 382
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.873069772634281e-06,
+      "loss": 0.4278,
+      "step": 383
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.85278706915564e-06,
+      "loss": 0.4473,
+      "step": 384
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.832536455688782e-06,
+      "loss": 0.4169,
+      "step": 385
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.812318498686902e-06,
+      "loss": 0.4152,
+      "step": 386
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7921337636897136e-06,
+      "loss": 0.4001,
+      "step": 387
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7719828153076486e-06,
+      "loss": 0.4363,
+      "step": 388
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.751866217206052e-06,
+      "loss": 0.4035,
+      "step": 389
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7317845320894124e-06,
+      "loss": 0.4358,
+      "step": 390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.711738321685635e-06,
+      "loss": 0.4162,
+      "step": 391
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.6917281467303133e-06,
+      "loss": 0.4593,
+      "step": 392
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.6717545669510593e-06,
+      "loss": 0.4203,
+      "step": 393
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.6518181410518327e-06,
+      "loss": 0.451,
+      "step": 394
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.631919426697325e-06,
+      "loss": 0.4532,
+      "step": 395
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.6120589804973504e-06,
+      "loss": 0.46,
+      "step": 396
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.5922373579912825e-06,
+      "loss": 0.4266,
+      "step": 397
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.5724551136325126e-06,
+      "loss": 0.4085,
+      "step": 398
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.5527128007729404e-06,
+      "loss": 0.4435,
+      "step": 399
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.5330109716474955e-06,
+      "loss": 0.4204,
+      "step": 400
+    },
+    {
+      "epoch": 1.28,
+      "eval_loss": 0.5436715483665466,
+      "eval_runtime": 20.2552,
+      "eval_samples_per_second": 40.088,
+      "eval_steps_per_second": 2.518,
+      "step": 400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.5133501773586904e-06,
+      "loss": 0.4335,
+      "step": 401
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.493730967861205e-06,
+      "loss": 0.3994,
+      "step": 402
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.474153891946505e-06,
+      "loss": 0.4503,
+      "step": 403
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.4546194972274847e-06,
+      "loss": 0.4138,
+      "step": 404
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.4351283301231614e-06,
+      "loss": 0.4225,
+      "step": 405
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.4156809358433725e-06,
+      "loss": 0.434,
+      "step": 406
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.396277858373546e-06,
+      "loss": 0.4191,
+      "step": 407
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3769196404594647e-06,
+      "loss": 0.4516,
+      "step": 408
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3576068235920993e-06,
+      "loss": 0.4376,
+      "step": 409
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.3383399479924544e-06,
+      "loss": 0.4235,
+      "step": 410
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.319119552596455e-06,
+      "loss": 0.4409,
+      "step": 411
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.299946175039881e-06,
+      "loss": 0.4088,
+      "step": 412
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.280820351643313e-06,
+      "loss": 0.4093,
+      "step": 413
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2617426173971528e-06,
+      "loss": 0.4343,
+      "step": 414
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.24271350594663e-06,
+      "loss": 0.452,
+      "step": 415
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.223733549576903e-06,
+      "loss": 0.4149,
+      "step": 416
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2048032791981513e-06,
+      "loss": 0.4081,
+      "step": 417
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.1859232243307296e-06,
+      "loss": 0.4345,
+      "step": 418
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.1670939130903583e-06,
+      "loss": 0.3928,
+      "step": 419
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.14831587217335e-06,
+      "loss": 0.4428,
+      "step": 420
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.1295896268418747e-06,
+      "loss": 0.4255,
+      "step": 421
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.1109157009092697e-06,
+      "loss": 0.4159,
+      "step": 422
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.0922946167253853e-06,
+      "loss": 0.4591,
+      "step": 423
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.073726895161975e-06,
+      "loss": 0.4041,
+      "step": 424
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.055213055598126e-06,
+      "loss": 0.4201,
+      "step": 425
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.036753615905728e-06,
+      "loss": 0.4174,
+      "step": 426
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.0183490924349912e-06,
+      "loss": 0.4379,
+      "step": 427
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.0000000000000008e-06,
+      "loss": 0.3877,
+      "step": 428
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9817068518643154e-06,
+      "loss": 0.4416,
+      "step": 429
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9634701597266134e-06,
+      "loss": 0.4631,
+      "step": 430
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9452904337063757e-06,
+      "loss": 0.4077,
+      "step": 431
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9271681823296177e-06,
+      "loss": 0.4365,
+      "step": 432
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9091039125146715e-06,
+      "loss": 0.4345,
+      "step": 433
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.89109812955799e-06,
+      "loss": 0.3845,
+      "step": 434
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8731513371200344e-06,
+      "loss": 0.4188,
+      "step": 435
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.855264037211163e-06,
+      "loss": 0.4209,
+      "step": 436
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.837436730177611e-06,
+      "loss": 0.4211,
+      "step": 437
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8196699146874734e-06,
+      "loss": 0.4061,
+      "step": 438
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.801964087716776e-06,
+      "loss": 0.4164,
+      "step": 439
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7843197445355589e-06,
+      "loss": 0.4148,
+      "step": 440
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7667373786940307e-06,
+      "loss": 0.4212,
+      "step": 441
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7492174820087594e-06,
+      "loss": 0.4445,
+      "step": 442
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7317605445489175e-06,
+      "loss": 0.4249,
+      "step": 443
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7143670546225715e-06,
+      "loss": 0.4041,
+      "step": 444
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.697037498763025e-06,
+      "loss": 0.4122,
+      "step": 445
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6797723617152077e-06,
+      "loss": 0.4299,
+      "step": 446
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.662572126422117e-06,
+      "loss": 0.4175,
+      "step": 447
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.645437274011309e-06,
+      "loss": 0.425,
+      "step": 448
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6283682837814387e-06,
+      "loss": 0.4413,
+      "step": 449
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.611365633188856e-06,
+      "loss": 0.417,
+      "step": 450
+    },
+    {
+      "epoch": 1.43,
+      "eval_loss": 0.5418304204940796,
+      "eval_runtime": 20.2567,
+      "eval_samples_per_second": 40.085,
+      "eval_steps_per_second": 2.518,
+      "step": 450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.5944297978342473e-06,
+      "loss": 0.4245,
+      "step": 451
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.577561251449334e-06,
+      "loss": 0.433,
+      "step": 452
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.5607604658836184e-06,
+      "loss": 0.4076,
+      "step": 453
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.544027911091193e-06,
+      "loss": 0.4075,
+      "step": 454
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.527364055117579e-06,
+      "loss": 0.4168,
+      "step": 455
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5107693640866543e-06,
+      "loss": 0.4428,
+      "step": 456
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.494244302187595e-06,
+      "loss": 0.4229,
+      "step": 457
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4777893316619112e-06,
+      "loss": 0.4156,
+      "step": 458
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4614049127904951e-06,
+      "loss": 0.4174,
+      "step": 459
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4450915038807674e-06,
+      "loss": 0.4194,
+      "step": 460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4288495612538425e-06,
+      "loss": 0.4077,
+      "step": 461
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.41267953923177e-06,
+      "loss": 0.4208,
+      "step": 462
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.396581890124826e-06,
+      "loss": 0.4359,
+      "step": 463
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.38055706421886e-06,
+      "loss": 0.4225,
+      "step": 464
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3646055097627004e-06,
+      "loss": 0.4369,
+      "step": 465
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.348727672955616e-06,
+      "loss": 0.4316,
+      "step": 466
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.332923997934834e-06,
+      "loss": 0.4077,
+      "step": 467
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3171949267631166e-06,
+      "loss": 0.4084,
+      "step": 468
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3015408994163973e-06,
+      "loss": 0.4184,
+      "step": 469
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2859623537714716e-06,
+      "loss": 0.4181,
+      "step": 470
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2704597255937492e-06,
+      "loss": 0.4561,
+      "step": 471
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.255033448525066e-06,
+      "loss": 0.4527,
+      "step": 472
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2396839540715527e-06,
+      "loss": 0.4339,
+      "step": 473
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2244116715915635e-06,
+      "loss": 0.4224,
+      "step": 474
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2092170282836738e-06,
+      "loss": 0.4184,
+      "step": 475
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1941004491747145e-06,
+      "loss": 0.4335,
+      "step": 476
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1790623571079042e-06,
+      "loss": 0.4284,
+      "step": 477
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1641031727309995e-06,
+      "loss": 0.4312,
+      "step": 478
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1492233144845497e-06,
+      "loss": 0.4201,
+      "step": 479
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1344231985901708e-06,
+      "loss": 0.4405,
+      "step": 480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1197032390389254e-06,
+      "loss": 0.412,
+      "step": 481
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.1050638475797191e-06,
+      "loss": 0.4424,
+      "step": 482
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.090505433707805e-06,
+      "loss": 0.4234,
+      "step": 483
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0760284046533134e-06,
+      "loss": 0.433,
+      "step": 484
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0616331653698672e-06,
+      "loss": 0.4113,
+      "step": 485
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0473201185232538e-06,
+      "loss": 0.4328,
+      "step": 486
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.033089664480162e-06,
+      "loss": 0.4482,
+      "step": 487
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0189422012969814e-06,
+      "loss": 0.4159,
+      "step": 488
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0048781247086696e-06,
+      "loss": 0.4146,
+      "step": 489
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.908978281176824e-07,
+      "loss": 0.4519,
+      "step": 490
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.770017025829674e-07,
+      "loss": 0.4253,
+      "step": 491
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.631901368090276e-07,
+      "loss": 0.4544,
+      "step": 492
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.494635171350478e-07,
+      "loss": 0.4072,
+      "step": 493
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.358222275240884e-07,
+      "loss": 0.4067,
+      "step": 494
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.22266649552343e-07,
+      "loss": 0.4515,
+      "step": 495
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.087971623984679e-07,
+      "loss": 0.4241,
+      "step": 496
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.954141428329722e-07,
+      "loss": 0.4325,
+      "step": 497
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.821179652076853e-07,
+      "loss": 0.4338,
+      "step": 498
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.689090014452731e-07,
+      "loss": 0.4252,
+      "step": 499
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.557876210288508e-07,
+      "loss": 0.4161,
+      "step": 500
+    },
+    {
+      "epoch": 1.59,
+      "eval_loss": 0.5409616827964783,
+      "eval_runtime": 20.2688,
+      "eval_samples_per_second": 40.062,
+      "eval_steps_per_second": 2.516,
+      "step": 500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.427541909916312e-07,
+      "loss": 0.4114,
+      "step": 501
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.298090759066743e-07,
+      "loss": 0.4518,
+      "step": 502
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.169526378766712e-07,
+      "loss": 0.4471,
+      "step": 503
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.041852365238346e-07,
+      "loss": 0.4274,
+      "step": 504
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.915072289798246e-07,
+      "loss": 0.4132,
+      "step": 505
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.789189698757654e-07,
+      "loss": 0.4155,
+      "step": 506
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.664208113323245e-07,
+      "loss": 0.42,
+      "step": 507
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.540131029498633e-07,
+      "loss": 0.4125,
+      "step": 508
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.416961917986571e-07,
+      "loss": 0.4476,
+      "step": 509
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.294704224091885e-07,
+      "loss": 0.412,
+      "step": 510
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.173361367625075e-07,
+      "loss": 0.4278,
+      "step": 511
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.052936742806692e-07,
+      "loss": 0.418,
+      "step": 512
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.933433718172357e-07,
+      "loss": 0.459,
+      "step": 513
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.814855636478558e-07,
+      "loss": 0.3998,
+      "step": 514
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.697205814609148e-07,
+      "loss": 0.424,
+      "step": 515
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.580487543482549e-07,
+      "loss": 0.4605,
+      "step": 516
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.464704087959703e-07,
+      "loss": 0.4591,
+      "step": 517
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.349858686752747e-07,
+      "loss": 0.4165,
+      "step": 518
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.235954552334464e-07,
+      "loss": 0.4577,
+      "step": 519
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.122994870848308e-07,
+      "loss": 0.4225,
+      "step": 520
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.010982802019429e-07,
+      "loss": 0.4233,
+      "step": 521
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.899921479066146e-07,
+      "loss": 0.4266,
+      "step": 522
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.789814008612439e-07,
+      "loss": 0.4291,
+      "step": 523
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.680663470600917e-07,
+      "loss": 0.4292,
+      "step": 524
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.572472918206785e-07,
+      "loss": 0.4277,
+      "step": 525
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.465245377752362e-07,
+      "loss": 0.4184,
+      "step": 526
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.358983848622451e-07,
+      "loss": 0.4441,
+      "step": 527
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.253691303180456e-07,
+      "loss": 0.4269,
+      "step": 528
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.149370686685204e-07,
+      "loss": 0.407,
+      "step": 529
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.046024917208602e-07,
+      "loss": 0.4238,
+      "step": 530
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.943656885553973e-07,
+      "loss": 0.4299,
+      "step": 531
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.842269455175221e-07,
+      "loss": 0.44,
+      "step": 532
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7418654620967216e-07,
+      "loss": 0.4051,
+      "step": 533
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.6424477148339924e-07,
+      "loss": 0.4094,
+      "step": 534
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5440189943151395e-07,
+      "loss": 0.448,
+      "step": 535
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4465820538030653e-07,
+      "loss": 0.4432,
+      "step": 536
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.3501396188184446e-07,
+      "loss": 0.4343,
+      "step": 537
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.2546943870635135e-07,
+      "loss": 0.4085,
+      "step": 538
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.16024902834657e-07,
+      "loss": 0.4303,
+      "step": 539
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.0668061845073207e-07,
+      "loss": 0.4229,
+      "step": 540
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.974368469342977e-07,
+      "loss": 0.3915,
+      "step": 541
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8829384685351573e-07,
+      "loss": 0.448,
+      "step": 542
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.792518739577497e-07,
+      "loss": 0.4459,
+      "step": 543
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7031118117042227e-07,
+      "loss": 0.418,
+      "step": 544
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.6147201858192623e-07,
+      "loss": 0.4231,
+      "step": 545
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.52734633442644e-07,
+      "loss": 0.4253,
+      "step": 546
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.4409927015601703e-07,
+      "loss": 0.4241,
+      "step": 547
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.355661702717216e-07,
+      "loss": 0.4268,
+      "step": 548
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2713557247890447e-07,
+      "loss": 0.4503,
+      "step": 549
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.18807712599507e-07,
+      "loss": 0.4219,
+      "step": 550
+    },
+    {
+      "epoch": 1.75,
+      "eval_loss": 0.5402326583862305,
+      "eval_runtime": 20.2294,
+      "eval_samples_per_second": 40.14,
+      "eval_steps_per_second": 2.521,
+      "step": 550
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.10582823581675e-07,
+      "loss": 0.4165,
+      "step": 551
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.0246113549323185e-07,
+      "loss": 0.4205,
+      "step": 552
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.9444287551525505e-07,
+      "loss": 0.4266,
+      "step": 553
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.865282679357097e-07,
+      "loss": 0.4536,
+      "step": 554
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7871753414318557e-07,
+      "loss": 0.4358,
+      "step": 555
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.710108926206969e-07,
+      "loss": 0.4157,
+      "step": 556
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.6340855893957337e-07,
+      "loss": 0.4285,
+      "step": 557
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.5591074575343016e-07,
+      "loss": 0.4378,
+      "step": 558
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4851766279222033e-07,
+      "loss": 0.41,
+      "step": 559
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.412295168563667e-07,
+      "loss": 0.4278,
+      "step": 560
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.3404651181097778e-07,
+      "loss": 0.4299,
+      "step": 561
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2696884858014552e-07,
+      "loss": 0.4284,
+      "step": 562
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1999672514132617e-07,
+      "loss": 0.426,
+      "step": 563
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.131303365197983e-07,
+      "loss": 0.4299,
+      "step": 564
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.063698747832139e-07,
+      "loss": 0.4232,
+      "step": 565
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9971552903621868e-07,
+      "loss": 0.4224,
+      "step": 566
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9316748541516882e-07,
+      "loss": 0.3978,
+      "step": 567
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8672592708291934e-07,
+      "loss": 0.4318,
+      "step": 568
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.803910342237045e-07,
+      "loss": 0.4226,
+      "step": 569
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7416298403809625e-07,
+      "loss": 0.4183,
+      "step": 570
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.680419507380444e-07,
+      "loss": 0.4297,
+      "step": 571
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6202810554201097e-07,
+      "loss": 0.4372,
+      "step": 572
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.561216166701711e-07,
+      "loss": 0.4015,
+      "step": 573
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5032264933971806e-07,
+      "loss": 0.4547,
+      "step": 574
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4463136576023184e-07,
+      "loss": 0.4241,
+      "step": 575
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.3904792512914942e-07,
+      "loss": 0.4298,
+      "step": 576
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.3357248362730755e-07,
+      "loss": 0.4399,
+      "step": 577
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.28205194414575e-07,
+      "loss": 0.4303,
+      "step": 578
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.229462076255694e-07,
+      "loss": 0.4212,
+      "step": 579
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1779567036545569e-07,
+      "loss": 0.4225,
+      "step": 580
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1275372670583338e-07,
+      "loss": 0.4194,
+      "step": 581
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0782051768070477e-07,
+      "loss": 0.4573,
+      "step": 582
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0299618128253129e-07,
+      "loss": 0.4204,
+      "step": 583
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.828085245837181e-08,
+      "loss": 0.4088,
+      "step": 584
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.367466310611095e-08,
+      "loss": 0.4537,
+      "step": 585
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.9177742070766e-08,
+      "loss": 0.4299,
+      "step": 586
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.479021514088547e-08,
+      "loss": 0.4376,
+      "step": 587
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.051220504502864e-08,
+      "loss": 0.4256,
+      "step": 588
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.634383144833468e-08,
+      "loss": 0.4429,
+      "step": 589
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.228521094917317e-08,
+      "loss": 0.423,
+      "step": 590
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.833645707588509e-08,
+      "loss": 0.3991,
+      "step": 591
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.449768028360303e-08,
+      "loss": 0.4319,
+      "step": 592
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.076898795116792e-08,
+      "loss": 0.4169,
+      "step": 593
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.715048437811809e-08,
+      "loss": 0.4372,
+      "step": 594
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.36422707817783e-08,
+      "loss": 0.4195,
+      "step": 595
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.0244445294422845e-08,
+      "loss": 0.4105,
+      "step": 596
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.695710296053601e-08,
+      "loss": 0.4279,
+      "step": 597
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.378033573414841e-08,
+      "loss": 0.4392,
+      "step": 598
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.071423247626926e-08,
+      "loss": 0.4049,
+      "step": 599
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.775887895239726e-08,
+      "loss": 0.4121,
+      "step": 600
+    },
+    {
+      "epoch": 1.91,
+      "eval_loss": 0.5397063493728638,
+      "eval_runtime": 20.2502,
+      "eval_samples_per_second": 40.098,
+      "eval_steps_per_second": 2.518,
+      "step": 600
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 626,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 50,
+  "total_flos": 251453096755200.0,
+  "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:dd04f2bb8d9a21e80a2b7bf8a5337a5386bcbd70a570e6daad2ea5277b705ce7
+size 6776

zero_to_fp32.py

@@ -0,0 +1,587 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example: python zero_to_fp32.py . pytorch_model.bin
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+
+    total_files = len(files)
+    state_dicts = []
+    for f in files:
+        state_dict = torch.load(f, map_location=device)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    if zero_stage <= 2:
+        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    elif zero_stage == 3:
+        # if there is more than one param group, there will be multiple flattened tensors - one
+        # flattened tensor per group - for simplicity merge them into a single tensor
+        #
+        # XXX: could make the script more memory efficient for when there are multiple groups - it
+        # will require matching the sub-lists of param_shapes for each param group flattened tensor
+
+        fp32_flat_groups = [
+            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
+        ]
+
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel()
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = fp32_flat_groups[0].numel() * world_size
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    for name, shape in param_shapes.items():
+
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # XXX: memory usage doubles here
+        state_dict[name] = torch.cat(
+            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
+            0).narrow(0, 0, unpartitioned_numel).view(shape)
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
+    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+    """
+
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+    print(f"Saving fp32 state dict to {output_file}")
+    torch.save(state_dict, output_file)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument(
+        "output_file",
+        type=str,
+        help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)