Upload folder using huggingface_hub

added_tokens.json

@@ -0,0 +1,4 @@
+{
+  "<|im_end|>": 32000,
+  "<|im_start|>": 32001
+}

config.json

@@ -0,0 +1,28 @@
+{
+  "_name_or_path": "Unbabel/TowerBase-7B-v0.1",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "eos_token_id": 32000,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 4096,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.40.0.dev0",
+  "use_cache": false,
+  "vocab_size": 32002
+}

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "do_sample": true,
+  "eos_token_id": 2,
+  "transformers_version": "4.40.0.dev0"
+}

latest

@@ -0,0 +1 @@
+global_step5624

model-00001-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:419140c230c3066b065684f79909a8973c2f663616dae18206cec205f21e8942
+size 4939001736

model-00002-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aec1f182a8cab4287640bb45035987839589c066c68af85ab091d1b2f12dd254
+size 4947390880

model-00003-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8aba4188a43b9e11083bd4c27091ce9899a46c8b4729ec483913252b6e64e9b3
+size 3590505200

model.safetensors.index.json

@@ -0,0 +1,298 @@
+{
+  "metadata": {
+    "total_size": 13476864000
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00003-of-00003.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.norm.weight": "model-00003-of-00003.safetensors"
+  }
+}

rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f2439da621f14c22b4f733e91bfc9de6b506d28d7b8d6f3eaca2e0b4f24c078
+size 15024

rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9e3fb386557f376b8946af5b8c91f9418f374dddb2ad9da4868b1ef16778c32
+size 15024

rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc7774d06045635bece9e960378fdc6913bf7bbbc903444cc570d1ca6ac25645
+size 15024

rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d98c54a80a914fecf43d06ea81432499f46e70664f1d04651bf339163e30fa9e
+size 15024

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cda74cda78125cfad5a2d4b09c5a8c8327b1a8d68aa04ba8799fe223d26bcee8
+size 1064

special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|im_end|>",
+    "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 +1,58 @@
+{
+  "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
+    },
+    "32000": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "32001": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "bos_token": "<s>",
+  "chat_template": "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "legacy": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

trainer_state.json

@@ -0,0 +1,3955 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.9420444444444445,
+  "eval_steps": 500,
+  "global_step": 5624,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.912928587976149,
+      "learning_rate": 1.4e-07,
+      "loss": 1.0098,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.065321089366511,
+      "learning_rate": 2.8e-07,
+      "loss": 1.0032,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.529282680442283,
+      "learning_rate": 4.2e-07,
+      "loss": 0.9767,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.079773534866118,
+      "learning_rate": 5.6e-07,
+      "loss": 0.9341,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.7504240808921168,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 0.8727,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7446189301316752,
+      "learning_rate": 8.4e-07,
+      "loss": 0.7997,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5893489037586176,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.7828,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5798012459841311,
+      "learning_rate": 1.12e-06,
+      "loss": 0.7671,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5143143417454488,
+      "learning_rate": 1.26e-06,
+      "loss": 0.777,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5006881361687121,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 0.7709,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5268772561224019,
+      "learning_rate": 1.54e-06,
+      "loss": 0.7751,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.49059329535011015,
+      "learning_rate": 1.68e-06,
+      "loss": 0.7588,
+      "step": 120
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.548982179156723,
+      "learning_rate": 1.82e-06,
+      "loss": 0.758,
+      "step": 130
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5118740800557817,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 0.7492,
+      "step": 140
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.47988356348194033,
+      "learning_rate": 2.1e-06,
+      "loss": 0.7479,
+      "step": 150
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5324095582498372,
+      "learning_rate": 2.24e-06,
+      "loss": 0.7344,
+      "step": 160
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.49578185528674784,
+      "learning_rate": 2.38e-06,
+      "loss": 0.7379,
+      "step": 170
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4751722809020323,
+      "learning_rate": 2.52e-06,
+      "loss": 0.7515,
+      "step": 180
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4898512842949614,
+      "learning_rate": 2.66e-06,
+      "loss": 0.7428,
+      "step": 190
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4938014103724035,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.7356,
+      "step": 200
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4853179196888149,
+      "learning_rate": 2.94e-06,
+      "loss": 0.7338,
+      "step": 210
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5006261354893382,
+      "learning_rate": 3.08e-06,
+      "loss": 0.7228,
+      "step": 220
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.49494536099466524,
+      "learning_rate": 3.22e-06,
+      "loss": 0.7371,
+      "step": 230
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4745560090617258,
+      "learning_rate": 3.36e-06,
+      "loss": 0.7374,
+      "step": 240
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.458424659300056,
+      "learning_rate": 3.5e-06,
+      "loss": 0.7284,
+      "step": 250
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4918105642778609,
+      "learning_rate": 3.64e-06,
+      "loss": 0.719,
+      "step": 260
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.45994092727545755,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.7328,
+      "step": 270
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4888877840053054,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 0.7257,
+      "step": 280
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4891132357037931,
+      "learning_rate": 4.059999999999999e-06,
+      "loss": 0.7146,
+      "step": 290
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4659780107286472,
+      "learning_rate": 4.2e-06,
+      "loss": 0.7207,
+      "step": 300
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4747662452681582,
+      "learning_rate": 4.34e-06,
+      "loss": 0.7196,
+      "step": 310
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.46183058951309874,
+      "learning_rate": 4.48e-06,
+      "loss": 0.7166,
+      "step": 320
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.47556837186042844,
+      "learning_rate": 4.62e-06,
+      "loss": 0.7138,
+      "step": 330
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4646419935884572,
+      "learning_rate": 4.76e-06,
+      "loss": 0.7166,
+      "step": 340
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.47208612393069765,
+      "learning_rate": 4.9e-06,
+      "loss": 0.7071,
+      "step": 350
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.47395551626034477,
+      "learning_rate": 5.04e-06,
+      "loss": 0.7081,
+      "step": 360
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.46256038389399284,
+      "learning_rate": 5.1799999999999995e-06,
+      "loss": 0.7112,
+      "step": 370
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.44989559880311664,
+      "learning_rate": 5.32e-06,
+      "loss": 0.7157,
+      "step": 380
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4759980664139243,
+      "learning_rate": 5.46e-06,
+      "loss": 0.716,
+      "step": 390
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.47761427911509746,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.6936,
+      "step": 400
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4823631066912239,
+      "learning_rate": 5.739999999999999e-06,
+      "loss": 0.7096,
+      "step": 410
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4692563644972781,
+      "learning_rate": 5.88e-06,
+      "loss": 0.6955,
+      "step": 420
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4758216043542266,
+      "learning_rate": 6.02e-06,
+      "loss": 0.7046,
+      "step": 430
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4607724176991764,
+      "learning_rate": 6.16e-06,
+      "loss": 0.7071,
+      "step": 440
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.47650098464440593,
+      "learning_rate": 6.3e-06,
+      "loss": 0.6948,
+      "step": 450
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4927763843500283,
+      "learning_rate": 6.44e-06,
+      "loss": 0.7138,
+      "step": 460
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.44343044028786904,
+      "learning_rate": 6.58e-06,
+      "loss": 0.7033,
+      "step": 470
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.45708129790603597,
+      "learning_rate": 6.72e-06,
+      "loss": 0.7038,
+      "step": 480
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.47564264251663835,
+      "learning_rate": 6.8599999999999995e-06,
+      "loss": 0.6974,
+      "step": 490
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4561386006973232,
+      "learning_rate": 7e-06,
+      "loss": 0.702,
+      "step": 500
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4318637464381274,
+      "learning_rate": 6.999934216315939e-06,
+      "loss": 0.7054,
+      "step": 510
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.47772094451329594,
+      "learning_rate": 6.999736867736609e-06,
+      "loss": 0.6946,
+      "step": 520
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.45891608711087106,
+      "learning_rate": 6.9994079616804764e-06,
+      "loss": 0.6952,
+      "step": 530
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.46731862765960264,
+      "learning_rate": 6.9989475105113426e-06,
+      "loss": 0.6888,
+      "step": 540
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4667223098464595,
+      "learning_rate": 6.998355531537879e-06,
+      "loss": 0.7017,
+      "step": 550
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.46285196540927176,
+      "learning_rate": 6.997632047012975e-06,
+      "loss": 0.7051,
+      "step": 560
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.48044807815149254,
+      "learning_rate": 6.996777084132904e-06,
+      "loss": 0.701,
+      "step": 570
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.47600970966063727,
+      "learning_rate": 6.995790675036298e-06,
+      "loss": 0.7001,
+      "step": 580
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4494522317826872,
+      "learning_rate": 6.994672856802944e-06,
+      "loss": 0.7042,
+      "step": 590
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4623294450089233,
+      "learning_rate": 6.993423671452386e-06,
+      "loss": 0.69,
+      "step": 600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.43825456028915594,
+      "learning_rate": 6.9920431659423436e-06,
+      "loss": 0.6996,
+      "step": 610
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4568055452742323,
+      "learning_rate": 6.990531392166956e-06,
+      "loss": 0.6939,
+      "step": 620
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4302767633743081,
+      "learning_rate": 6.988888406954821e-06,
+      "loss": 0.6898,
+      "step": 630
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4762852616798798,
+      "learning_rate": 6.9871142720668644e-06,
+      "loss": 0.703,
+      "step": 640
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4572026337069386,
+      "learning_rate": 6.985209054194017e-06,
+      "loss": 0.7004,
+      "step": 650
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.45803902960498666,
+      "learning_rate": 6.983172824954708e-06,
+      "loss": 0.6853,
+      "step": 660
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.44353624606381903,
+      "learning_rate": 6.9810056608921725e-06,
+      "loss": 0.7074,
+      "step": 670
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.44517458769087626,
+      "learning_rate": 6.978707643471573e-06,
+      "loss": 0.6988,
+      "step": 680
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4616555458392388,
+      "learning_rate": 6.97627885907694e-06,
+      "loss": 0.7034,
+      "step": 690
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4770896081066365,
+      "learning_rate": 6.973719399007923e-06,
+      "loss": 0.6935,
+      "step": 700
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.45665921054521347,
+      "learning_rate": 6.9710293594763545e-06,
+      "loss": 0.6773,
+      "step": 710
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.48834217157342125,
+      "learning_rate": 6.968208841602645e-06,
+      "loss": 0.6974,
+      "step": 720
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4661409470252182,
+      "learning_rate": 6.965257951411967e-06,
+      "loss": 0.6796,
+      "step": 730
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4249423447942054,
+      "learning_rate": 6.962176799830279e-06,
+      "loss": 0.686,
+      "step": 740
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4517631229399239,
+      "learning_rate": 6.958965502680155e-06,
+      "loss": 0.6968,
+      "step": 750
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4334006789419362,
+      "learning_rate": 6.955624180676427e-06,
+      "loss": 0.705,
+      "step": 760
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.44354874837116653,
+      "learning_rate": 6.9521529594216516e-06,
+      "loss": 0.6954,
+      "step": 770
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4606606226964418,
+      "learning_rate": 6.948551969401381e-06,
+      "loss": 0.6965,
+      "step": 780
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.46221163538458165,
+      "learning_rate": 6.94482134597927e-06,
+      "loss": 0.695,
+      "step": 790
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4636824720485381,
+      "learning_rate": 6.940961229391975e-06,
+      "loss": 0.6919,
+      "step": 800
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4450527833539268,
+      "learning_rate": 6.936971764743891e-06,
+      "loss": 0.6977,
+      "step": 810
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4358125416971688,
+      "learning_rate": 6.932853102001694e-06,
+      "loss": 0.6998,
+      "step": 820
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.45623590289661414,
+      "learning_rate": 6.928605395988701e-06,
+      "loss": 0.6954,
+      "step": 830
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4536975058820564,
+      "learning_rate": 6.924228806379058e-06,
+      "loss": 0.6742,
+      "step": 840
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4563719379438227,
+      "learning_rate": 6.919723497691728e-06,
+      "loss": 0.6921,
+      "step": 850
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.45279224746852664,
+      "learning_rate": 6.915089639284313e-06,
+      "loss": 0.6861,
+      "step": 860
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.466062080319079,
+      "learning_rate": 6.910327405346686e-06,
+      "loss": 0.6895,
+      "step": 870
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.443881137156012,
+      "learning_rate": 6.905436974894443e-06,
+      "loss": 0.7008,
+      "step": 880
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.47752762402129206,
+      "learning_rate": 6.900418531762173e-06,
+      "loss": 0.6985,
+      "step": 890
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4542692407893758,
+      "learning_rate": 6.89527226459655e-06,
+      "loss": 0.6822,
+      "step": 900
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4314820719874765,
+      "learning_rate": 6.889998366849237e-06,
+      "loss": 0.691,
+      "step": 910
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4278370127210443,
+      "learning_rate": 6.884597036769621e-06,
+      "loss": 0.689,
+      "step": 920
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.45134601911703476,
+      "learning_rate": 6.879068477397353e-06,
+      "loss": 0.6898,
+      "step": 930
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.45160503192413054,
+      "learning_rate": 6.87341289655472e-06,
+      "loss": 0.6869,
+      "step": 940
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.41025143635863104,
+      "learning_rate": 6.867630506838833e-06,
+      "loss": 0.6984,
+      "step": 950
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.46520301654074564,
+      "learning_rate": 6.861721525613633e-06,
+      "loss": 0.6843,
+      "step": 960
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.451991102882798,
+      "learning_rate": 6.8556861750017235e-06,
+      "loss": 0.6962,
+      "step": 970
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.418111038766468,
+      "learning_rate": 6.849524681876018e-06,
+      "loss": 0.6797,
+      "step": 980
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4403261547939229,
+      "learning_rate": 6.843237277851211e-06,
+      "loss": 0.6965,
+      "step": 990
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.426598785059419,
+      "learning_rate": 6.836824199275074e-06,
+      "loss": 0.6821,
+      "step": 1000
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.42988247771547117,
+      "learning_rate": 6.830285687219569e-06,
+      "loss": 0.6911,
+      "step": 1010
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.452230475071558,
+      "learning_rate": 6.823621987471789e-06,
+      "loss": 0.6851,
+      "step": 1020
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4267205539811686,
+      "learning_rate": 6.816833350524716e-06,
+      "loss": 0.6777,
+      "step": 1030
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.44148424584394874,
+      "learning_rate": 6.809920031567808e-06,
+      "loss": 0.6838,
+      "step": 1040
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.43306877795839893,
+      "learning_rate": 6.802882290477399e-06,
+      "loss": 0.6864,
+      "step": 1050
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4952482617663558,
+      "learning_rate": 6.79572039180694e-06,
+      "loss": 0.6904,
+      "step": 1060
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.45382453893592856,
+      "learning_rate": 6.788434604777048e-06,
+      "loss": 0.6795,
+      "step": 1070
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.452960843334945,
+      "learning_rate": 6.781025203265388e-06,
+      "loss": 0.6891,
+      "step": 1080
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4537364245497661,
+      "learning_rate": 6.773492465796373e-06,
+      "loss": 0.6907,
+      "step": 1090
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.44929090527897886,
+      "learning_rate": 6.765836675530703e-06,
+      "loss": 0.6798,
+      "step": 1100
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.46381413350008455,
+      "learning_rate": 6.758058120254715e-06,
+      "loss": 0.6716,
+      "step": 1110
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4309028536458763,
+      "learning_rate": 6.750157092369563e-06,
+      "loss": 0.6799,
+      "step": 1120
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.43717422966700575,
+      "learning_rate": 6.742133888880233e-06,
+      "loss": 0.6883,
+      "step": 1130
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4459700930425581,
+      "learning_rate": 6.7339888113843696e-06,
+      "loss": 0.6891,
+      "step": 1140
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.44045298948848877,
+      "learning_rate": 6.725722166060951e-06,
+      "loss": 0.6817,
+      "step": 1150
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4485899862146157,
+      "learning_rate": 6.717334263658766e-06,
+      "loss": 0.6897,
+      "step": 1160
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.45682000330961775,
+      "learning_rate": 6.70882541948474e-06,
+      "loss": 0.6776,
+      "step": 1170
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.48037041295136884,
+      "learning_rate": 6.700195953392085e-06,
+      "loss": 0.6872,
+      "step": 1180
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.44334741491819346,
+      "learning_rate": 6.691446189768268e-06,
+      "loss": 0.6798,
+      "step": 1190
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4674740757760583,
+      "learning_rate": 6.682576457522825e-06,
+      "loss": 0.6977,
+      "step": 1200
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4696181980144796,
+      "learning_rate": 6.673587090074993e-06,
+      "loss": 0.6896,
+      "step": 1210
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4593954697303246,
+      "learning_rate": 6.664478425341176e-06,
+      "loss": 0.6749,
+      "step": 1220
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.41647753357217115,
+      "learning_rate": 6.655250805722244e-06,
+      "loss": 0.6894,
+      "step": 1230
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4245409839045758,
+      "learning_rate": 6.645904578090662e-06,
+      "loss": 0.6693,
+      "step": 1240
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.45490183172736,
+      "learning_rate": 6.636440093777451e-06,
+      "loss": 0.6881,
+      "step": 1250
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4633877447287089,
+      "learning_rate": 6.626857708558979e-06,
+      "loss": 0.6953,
+      "step": 1260
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.45069656102358646,
+      "learning_rate": 6.617157782643591e-06,
+      "loss": 0.6787,
+      "step": 1270
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.44438426822862237,
+      "learning_rate": 6.6073406806580646e-06,
+      "loss": 0.6859,
+      "step": 1280
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4335460798475662,
+      "learning_rate": 6.597406771633906e-06,
+      "loss": 0.6829,
+      "step": 1290
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4282672786086354,
+      "learning_rate": 6.587356428993477e-06,
+      "loss": 0.6831,
+      "step": 1300
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.46465171297436636,
+      "learning_rate": 6.577190030535957e-06,
+      "loss": 0.6778,
+      "step": 1310
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4590812961346198,
+      "learning_rate": 6.566907958423142e-06,
+      "loss": 0.6701,
+      "step": 1320
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4180631333820519,
+      "learning_rate": 6.5565105991650815e-06,
+      "loss": 0.6825,
+      "step": 1330
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.42684427340923925,
+      "learning_rate": 6.545998343605544e-06,
+      "loss": 0.6823,
+      "step": 1340
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6515643833482546,
+      "learning_rate": 6.5353715869073275e-06,
+      "loss": 0.6748,
+      "step": 1350
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.42995190312179654,
+      "learning_rate": 6.524630728537408e-06,
+      "loss": 0.6896,
+      "step": 1360
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4307066820527156,
+      "learning_rate": 6.513776172251919e-06,
+      "loss": 0.6821,
+      "step": 1370
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4401373902110004,
+      "learning_rate": 6.5028083260809735e-06,
+      "loss": 0.6729,
+      "step": 1380
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.420372235119902,
+      "learning_rate": 6.491727602313334e-06,
+      "loss": 0.6812,
+      "step": 1390
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.44387468527179835,
+      "learning_rate": 6.4805344174808986e-06,
+      "loss": 0.6713,
+      "step": 1400
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4224291568526637,
+      "learning_rate": 6.4692291923430634e-06,
+      "loss": 0.6928,
+      "step": 1410
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.42342827072921446,
+      "learning_rate": 6.457812351870889e-06,
+      "loss": 0.6925,
+      "step": 1420
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.4614687139520872,
+      "learning_rate": 6.446284325231132e-06,
+      "loss": 0.6804,
+      "step": 1430
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.4513094113300999,
+      "learning_rate": 6.434645545770116e-06,
+      "loss": 0.649,
+      "step": 1440
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.46129242006354043,
+      "learning_rate": 6.422896450997434e-06,
+      "loss": 0.6244,
+      "step": 1450
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.44352477273420793,
+      "learning_rate": 6.411037482569509e-06,
+      "loss": 0.6231,
+      "step": 1460
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.43347730975194065,
+      "learning_rate": 6.399069086272988e-06,
+      "loss": 0.6163,
+      "step": 1470
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5042235757137699,
+      "learning_rate": 6.386991712007985e-06,
+      "loss": 0.6295,
+      "step": 1480
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.4635765704926019,
+      "learning_rate": 6.374805813771171e-06,
+      "loss": 0.6145,
+      "step": 1490
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.4672283056367441,
+      "learning_rate": 6.362511849638706e-06,
+      "loss": 0.6248,
+      "step": 1500
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.44386378239345664,
+      "learning_rate": 6.3501102817490184e-06,
+      "loss": 0.6208,
+      "step": 1510
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.45014512458671113,
+      "learning_rate": 6.337601576285438e-06,
+      "loss": 0.6241,
+      "step": 1520
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.47077991205008496,
+      "learning_rate": 6.324986203458665e-06,
+      "loss": 0.637,
+      "step": 1530
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.43971957336428713,
+      "learning_rate": 6.3122646374891014e-06,
+      "loss": 0.6274,
+      "step": 1540
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.45398595356146343,
+      "learning_rate": 6.299437356589018e-06,
+      "loss": 0.6172,
+      "step": 1550
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.4638039927896387,
+      "learning_rate": 6.2865048429445835e-06,
+      "loss": 0.6162,
+      "step": 1560
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.456884430778857,
+      "learning_rate": 6.273467582697736e-06,
+      "loss": 0.6358,
+      "step": 1570
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.4513273711536076,
+      "learning_rate": 6.260326065927908e-06,
+      "loss": 0.6256,
+      "step": 1580
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.4585546365167011,
+      "learning_rate": 6.247080786633608e-06,
+      "loss": 0.6343,
+      "step": 1590
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.4837809920582229,
+      "learning_rate": 6.233732242713847e-06,
+      "loss": 0.6205,
+      "step": 1600
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.45062031874118463,
+      "learning_rate": 6.220280935949423e-06,
+      "loss": 0.6181,
+      "step": 1610
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.4934582241182996,
+      "learning_rate": 6.206727371984055e-06,
+      "loss": 0.6101,
+      "step": 1620
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.45848465100131724,
+      "learning_rate": 6.193072060305386e-06,
+      "loss": 0.6274,
+      "step": 1630
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.49225379713590917,
+      "learning_rate": 6.17931551422582e-06,
+      "loss": 0.6287,
+      "step": 1640
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.43783738072351636,
+      "learning_rate": 6.165458250863233e-06,
+      "loss": 0.6322,
+      "step": 1650
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.45111919610212603,
+      "learning_rate": 6.15150079112153e-06,
+      "loss": 0.6343,
+      "step": 1660
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.7283719867926337,
+      "learning_rate": 6.137443659671066e-06,
+      "loss": 0.6245,
+      "step": 1670
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.4317614230374671,
+      "learning_rate": 6.123287384928924e-06,
+      "loss": 0.6252,
+      "step": 1680
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.43630742763076885,
+      "learning_rate": 6.1090324990390505e-06,
+      "loss": 0.6281,
+      "step": 1690
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.49179102646470696,
+      "learning_rate": 6.09467953785225e-06,
+      "loss": 0.6304,
+      "step": 1700
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.4269421327683836,
+      "learning_rate": 6.080229040906045e-06,
+      "loss": 0.6205,
+      "step": 1710
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.44873848635658836,
+      "learning_rate": 6.065681551404392e-06,
+      "loss": 0.6203,
+      "step": 1720
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.43522811508044484,
+      "learning_rate": 6.051037616197267e-06,
+      "loss": 0.6233,
+      "step": 1730
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.43363424076560303,
+      "learning_rate": 6.036297785760099e-06,
+      "loss": 0.6274,
+      "step": 1740
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.4420787259752861,
+      "learning_rate": 6.0214626141730895e-06,
+      "loss": 0.6388,
+      "step": 1750
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.445119846862499,
+      "learning_rate": 6.006532659100377e-06,
+      "loss": 0.6107,
+      "step": 1760
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.4380767674114949,
+      "learning_rate": 5.991508481769071e-06,
+      "loss": 0.6341,
+      "step": 1770
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.44003117819419657,
+      "learning_rate": 5.976390646948166e-06,
+      "loss": 0.6344,
+      "step": 1780
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.45806509086322245,
+      "learning_rate": 5.961179722927302e-06,
+      "loss": 0.6283,
+      "step": 1790
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.4545928600817147,
+      "learning_rate": 5.9458762814954016e-06,
+      "loss": 0.6254,
+      "step": 1800
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.4438181707408447,
+      "learning_rate": 5.930480897919185e-06,
+      "loss": 0.631,
+      "step": 1810
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.44695115171581695,
+      "learning_rate": 5.9149941509215366e-06,
+      "loss": 0.6338,
+      "step": 1820
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.4280430227739119,
+      "learning_rate": 5.899416622659754e-06,
+      "loss": 0.6182,
+      "step": 1830
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.458726186518369,
+      "learning_rate": 5.883748898703666e-06,
+      "loss": 0.6162,
+      "step": 1840
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.43445566304338457,
+      "learning_rate": 5.8679915680136155e-06,
+      "loss": 0.6228,
+      "step": 1850
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.44895947980462597,
+      "learning_rate": 5.852145222918326e-06,
+      "loss": 0.6373,
+      "step": 1860
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.43403817083393664,
+      "learning_rate": 5.83621045909263e-06,
+      "loss": 0.6376,
+      "step": 1870
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.4673939224968789,
+      "learning_rate": 5.820187875535083e-06,
+      "loss": 0.6215,
+      "step": 1880
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.46323588428022766,
+      "learning_rate": 5.804078074545439e-06,
+      "loss": 0.6187,
+      "step": 1890
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.4530033509696719,
+      "learning_rate": 5.7878816617020204e-06,
+      "loss": 0.6239,
+      "step": 1900
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.4317929663828983,
+      "learning_rate": 5.771599245838943e-06,
+      "loss": 0.6168,
+      "step": 1910
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.436592310414347,
+      "learning_rate": 5.7552314390232364e-06,
+      "loss": 0.6179,
+      "step": 1920
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.4702835623046126,
+      "learning_rate": 5.738778856531832e-06,
+      "loss": 0.6272,
+      "step": 1930
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.4619318889613922,
+      "learning_rate": 5.72224211682844e-06,
+      "loss": 0.6256,
+      "step": 1940
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.49429029776316813,
+      "learning_rate": 5.705621841540292e-06,
+      "loss": 0.6283,
+      "step": 1950
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.47054367378052575,
+      "learning_rate": 5.688918655434783e-06,
+      "loss": 0.6156,
+      "step": 1960
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.45638233691668284,
+      "learning_rate": 5.67213318639598e-06,
+      "loss": 0.6257,
+      "step": 1970
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.43819489071261747,
+      "learning_rate": 5.655266065401021e-06,
+      "loss": 0.6255,
+      "step": 1980
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.45603698357049277,
+      "learning_rate": 5.638317926496398e-06,
+      "loss": 0.6267,
+      "step": 1990
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.45518318702227223,
+      "learning_rate": 5.6212894067741176e-06,
+      "loss": 0.6357,
+      "step": 2000
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.4402683023420712,
+      "learning_rate": 5.604181146347758e-06,
+      "loss": 0.6311,
+      "step": 2010
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.4498808898227514,
+      "learning_rate": 5.5869937883284065e-06,
+      "loss": 0.6213,
+      "step": 2020
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.46040698115780887,
+      "learning_rate": 5.569727978800478e-06,
+      "loss": 0.6223,
+      "step": 2030
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.44168864627397236,
+      "learning_rate": 5.552384366797435e-06,
+      "loss": 0.6268,
+      "step": 2040
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.45494321235524204,
+      "learning_rate": 5.534963604277388e-06,
+      "loss": 0.6193,
+      "step": 2050
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.44543538788588954,
+      "learning_rate": 5.517466346098587e-06,
+      "loss": 0.6311,
+      "step": 2060
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.45370006917207745,
+      "learning_rate": 5.4998932499948055e-06,
+      "loss": 0.6263,
+      "step": 2070
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.4457705866746906,
+      "learning_rate": 5.482244976550616e-06,
+      "loss": 0.6267,
+      "step": 2080
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.44178347775287935,
+      "learning_rate": 5.464522189176559e-06,
+      "loss": 0.6168,
+      "step": 2090
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.4510685099498634,
+      "learning_rate": 5.446725554084202e-06,
+      "loss": 0.6071,
+      "step": 2100
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.4463056440103558,
+      "learning_rate": 5.4288557402611e-06,
+      "loss": 0.6193,
+      "step": 2110
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.4450825773000299,
+      "learning_rate": 5.410913419445647e-06,
+      "loss": 0.6114,
+      "step": 2120
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.4609214677792106,
+      "learning_rate": 5.3928992661018194e-06,
+      "loss": 0.6255,
+      "step": 2130
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.48687583594807843,
+      "learning_rate": 5.374813957393832e-06,
+      "loss": 0.6286,
+      "step": 2140
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.47549284042607015,
+      "learning_rate": 5.356658173160674e-06,
+      "loss": 0.6143,
+      "step": 2150
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.49532165280916113,
+      "learning_rate": 5.338432595890562e-06,
+      "loss": 0.6249,
+      "step": 2160
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.45253915740067313,
+      "learning_rate": 5.320137910695275e-06,
+      "loss": 0.6257,
+      "step": 2170
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.43721435814923637,
+      "learning_rate": 5.301774805284408e-06,
+      "loss": 0.6178,
+      "step": 2180
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.4683301857922748,
+      "learning_rate": 5.2833439699395175e-06,
+      "loss": 0.6173,
+      "step": 2190
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.43871464981194036,
+      "learning_rate": 5.264846097488175e-06,
+      "loss": 0.6214,
+      "step": 2200
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.4524085111628937,
+      "learning_rate": 5.246281883277922e-06,
+      "loss": 0.6346,
+      "step": 2210
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.4468406698869542,
+      "learning_rate": 5.227652025150132e-06,
+      "loss": 0.614,
+      "step": 2220
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.468252187542662,
+      "learning_rate": 5.208957223413776e-06,
+      "loss": 0.6057,
+      "step": 2230
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.46458186348478814,
+      "learning_rate": 5.1901981808191e-06,
+      "loss": 0.6192,
+      "step": 2240
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.4589397282179608,
+      "learning_rate": 5.1713756025312095e-06,
+      "loss": 0.6197,
+      "step": 2250
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4733441471283767,
+      "learning_rate": 5.1524901961035555e-06,
+      "loss": 0.6146,
+      "step": 2260
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.49573981085967583,
+      "learning_rate": 5.1335426714513436e-06,
+      "loss": 0.6205,
+      "step": 2270
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.45753588591278177,
+      "learning_rate": 5.114533740824848e-06,
+      "loss": 0.6194,
+      "step": 2280
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.44981584915327405,
+      "learning_rate": 5.095464118782631e-06,
+      "loss": 0.6285,
+      "step": 2290
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.44941448245640475,
+      "learning_rate": 5.076334522164687e-06,
+      "loss": 0.6183,
+      "step": 2300
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.46348841235648264,
+      "learning_rate": 5.057145670065498e-06,
+      "loss": 0.6178,
+      "step": 2310
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.4819885899523623,
+      "learning_rate": 5.037898283806995e-06,
+      "loss": 0.6209,
+      "step": 2320
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.45974762343297226,
+      "learning_rate": 5.018593086911453e-06,
+      "loss": 0.6144,
+      "step": 2330
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.4832719455105882,
+      "learning_rate": 4.999230805074284e-06,
+      "loss": 0.6255,
+      "step": 2340
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.4580501245903807,
+      "learning_rate": 4.979812166136764e-06,
+      "loss": 0.622,
+      "step": 2350
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.4869292416366864,
+      "learning_rate": 4.960337900058668e-06,
+      "loss": 0.6295,
+      "step": 2360
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.44734991176527494,
+      "learning_rate": 4.940808738890834e-06,
+      "loss": 0.61,
+      "step": 2370
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.4836741219786191,
+      "learning_rate": 4.921225416747647e-06,
+      "loss": 0.6131,
+      "step": 2380
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.43868937063180397,
+      "learning_rate": 4.901588669779433e-06,
+      "loss": 0.6261,
+      "step": 2390
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.4549440779907735,
+      "learning_rate": 4.881899236144797e-06,
+      "loss": 0.6216,
+      "step": 2400
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.4561309327019534,
+      "learning_rate": 4.862157855982875e-06,
+      "loss": 0.6262,
+      "step": 2410
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.4521274007767562,
+      "learning_rate": 4.8423652713855e-06,
+      "loss": 0.6214,
+      "step": 2420
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.4876373591113174,
+      "learning_rate": 4.822522226369323e-06,
+      "loss": 0.6303,
+      "step": 2430
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.4403247558369275,
+      "learning_rate": 4.802629466847827e-06,
+      "loss": 0.6236,
+      "step": 2440
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.4392883872725244,
+      "learning_rate": 4.782687740603308e-06,
+      "loss": 0.6125,
+      "step": 2450
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.44359149108855517,
+      "learning_rate": 4.762697797258742e-06,
+      "loss": 0.6208,
+      "step": 2460
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.45892783125410747,
+      "learning_rate": 4.742660388249629e-06,
+      "loss": 0.6146,
+      "step": 2470
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.46353318895549067,
+      "learning_rate": 4.722576266795729e-06,
+      "loss": 0.6199,
+      "step": 2480
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.4642990741363008,
+      "learning_rate": 4.702446187872758e-06,
+      "loss": 0.6182,
+      "step": 2490
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.44827792507065956,
+      "learning_rate": 4.682270908184003e-06,
+      "loss": 0.6246,
+      "step": 2500
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.45544933714150454,
+      "learning_rate": 4.662051186131876e-06,
+      "loss": 0.6256,
+      "step": 2510
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.4485500362120205,
+      "learning_rate": 4.641787781789412e-06,
+      "loss": 0.6181,
+      "step": 2520
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.42631048877270405,
+      "learning_rate": 4.6214814568716894e-06,
+      "loss": 0.6331,
+      "step": 2530
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.4714279586473698,
+      "learning_rate": 4.601132974707202e-06,
+      "loss": 0.628,
+      "step": 2540
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.4228608375782349,
+      "learning_rate": 4.5807431002091605e-06,
+      "loss": 0.6054,
+      "step": 2550
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.46872660848782277,
+      "learning_rate": 4.560312599846746e-06,
+      "loss": 0.6102,
+      "step": 2560
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.4379038714391558,
+      "learning_rate": 4.539842241616287e-06,
+      "loss": 0.6143,
+      "step": 2570
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.4719919574560488,
+      "learning_rate": 4.519332795012404e-06,
+      "loss": 0.6197,
+      "step": 2580
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.4560470541146194,
+      "learning_rate": 4.498785030999068e-06,
+      "loss": 0.6132,
+      "step": 2590
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.48502107778992737,
+      "learning_rate": 4.478199721980633e-06,
+      "loss": 0.631,
+      "step": 2600
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.45288928959662245,
+      "learning_rate": 4.457577641772792e-06,
+      "loss": 0.6148,
+      "step": 2610
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.45740004712492455,
+      "learning_rate": 4.436919565573495e-06,
+      "loss": 0.613,
+      "step": 2620
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.4680089016865197,
+      "learning_rate": 4.416226269933802e-06,
+      "loss": 0.6109,
+      "step": 2630
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.4498754217059588,
+      "learning_rate": 4.395498532728697e-06,
+      "loss": 0.627,
+      "step": 2640
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.490510820257092,
+      "learning_rate": 4.374737133127847e-06,
+      "loss": 0.6287,
+      "step": 2650
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.4384793154811805,
+      "learning_rate": 4.35394285156631e-06,
+      "loss": 0.6265,
+      "step": 2660
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.42053564372682345,
+      "learning_rate": 4.3331164697151995e-06,
+      "loss": 0.6123,
+      "step": 2670
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.44499220286710817,
+      "learning_rate": 4.3122587704523015e-06,
+      "loss": 0.6196,
+      "step": 2680
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.4681953108721627,
+      "learning_rate": 4.291370537832641e-06,
+      "loss": 0.6301,
+      "step": 2690
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.4245150987038812,
+      "learning_rate": 4.2704525570590185e-06,
+      "loss": 0.6203,
+      "step": 2700
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.4738423212960381,
+      "learning_rate": 4.2495056144524824e-06,
+      "loss": 0.6159,
+      "step": 2710
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.49926406862961464,
+      "learning_rate": 4.228530497422779e-06,
+      "loss": 0.6193,
+      "step": 2720
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.4423739374256911,
+      "learning_rate": 4.207527994438748e-06,
+      "loss": 0.617,
+      "step": 2730
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.44692873617751755,
+      "learning_rate": 4.186498894998689e-06,
+      "loss": 0.6135,
+      "step": 2740
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.4358994979972626,
+      "learning_rate": 4.165443989600678e-06,
+      "loss": 0.6121,
+      "step": 2750
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.46452930431844286,
+      "learning_rate": 4.144364069712854e-06,
+      "loss": 0.6167,
+      "step": 2760
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.4816111574015236,
+      "learning_rate": 4.123259927743669e-06,
+      "loss": 0.6203,
+      "step": 2770
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.45232518080467465,
+      "learning_rate": 4.102132357012098e-06,
+      "loss": 0.6199,
+      "step": 2780
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.45515782747165817,
+      "learning_rate": 4.08098215171782e-06,
+      "loss": 0.6174,
+      "step": 2790
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.44933646029392305,
+      "learning_rate": 4.059810106911363e-06,
+      "loss": 0.6188,
+      "step": 2800
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.45633219759975596,
+      "learning_rate": 4.038617018464217e-06,
+      "loss": 0.6168,
+      "step": 2810
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.4663774750339656,
+      "learning_rate": 4.017403683038914e-06,
+      "loss": 0.6199,
+      "step": 2820
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.4565589400061048,
+      "learning_rate": 3.996170898059087e-06,
+      "loss": 0.6187,
+      "step": 2830
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.45638098232431645,
+      "learning_rate": 3.97491946167949e-06,
+      "loss": 0.6133,
+      "step": 2840
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.4330737687010161,
+      "learning_rate": 3.9536501727559956e-06,
+      "loss": 0.6179,
+      "step": 2850
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.44620897297773393,
+      "learning_rate": 3.932363830815563e-06,
+      "loss": 0.606,
+      "step": 2860
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.4727298461430969,
+      "learning_rate": 3.911061236026192e-06,
+      "loss": 0.5804,
+      "step": 2870
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.5332182751767908,
+      "learning_rate": 3.889743189166831e-06,
+      "loss": 0.5552,
+      "step": 2880
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.471875496548638,
+      "learning_rate": 3.868410491597286e-06,
+      "loss": 0.5467,
+      "step": 2890
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.4869637805163024,
+      "learning_rate": 3.847063945228094e-06,
+      "loss": 0.5691,
+      "step": 2900
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.4714418364302173,
+      "learning_rate": 3.825704352490375e-06,
+      "loss": 0.5788,
+      "step": 2910
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.49636094733662106,
+      "learning_rate": 3.804332516305672e-06,
+      "loss": 0.5583,
+      "step": 2920
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.48087980189754664,
+      "learning_rate": 3.782949240055768e-06,
+      "loss": 0.5632,
+      "step": 2930
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.4873147689537464,
+      "learning_rate": 3.7615553275524852e-06,
+      "loss": 0.5602,
+      "step": 2940
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.4603275098510104,
+      "learning_rate": 3.74015158300747e-06,
+      "loss": 0.5641,
+      "step": 2950
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.5162191764305892,
+      "learning_rate": 3.7187388110019604e-06,
+      "loss": 0.5628,
+      "step": 2960
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.49005627074608765,
+      "learning_rate": 3.697317816456546e-06,
+      "loss": 0.559,
+      "step": 2970
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.4585568665283943,
+      "learning_rate": 3.6758894046009037e-06,
+      "loss": 0.547,
+      "step": 2980
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.4506260874603515,
+      "learning_rate": 3.6544543809435346e-06,
+      "loss": 0.5433,
+      "step": 2990
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.46595533436834136,
+      "learning_rate": 3.6330135512414822e-06,
+      "loss": 0.5666,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.4690150184503375,
+      "learning_rate": 3.6115677214700397e-06,
+      "loss": 0.5596,
+      "step": 3010
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.4683369095498927,
+      "learning_rate": 3.5901176977924606e-06,
+      "loss": 0.5458,
+      "step": 3020
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4710288608351933,
+      "learning_rate": 3.568664286529646e-06,
+      "loss": 0.5507,
+      "step": 3030
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4928542807361932,
+      "learning_rate": 3.5472082941298433e-06,
+      "loss": 0.5665,
+      "step": 3040
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.4972921543225756,
+      "learning_rate": 3.5257505271383217e-06,
+      "loss": 0.5586,
+      "step": 3050
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.4855107426051562,
+      "learning_rate": 3.504291792167063e-06,
+      "loss": 0.5615,
+      "step": 3060
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.4623236179613674,
+      "learning_rate": 3.4828328958644326e-06,
+      "loss": 0.5638,
+      "step": 3070
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.46028741167647896,
+      "learning_rate": 3.4613746448848622e-06,
+      "loss": 0.5464,
+      "step": 3080
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.46156508300115645,
+      "learning_rate": 3.439917845858524e-06,
+      "loss": 0.567,
+      "step": 3090
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.5669489602625127,
+      "learning_rate": 3.418463305361013e-06,
+      "loss": 0.5524,
+      "step": 3100
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.49099941076825016,
+      "learning_rate": 3.3970118298830207e-06,
+      "loss": 0.5591,
+      "step": 3110
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.5207064606888653,
+      "learning_rate": 3.3755642258000265e-06,
+      "loss": 0.5538,
+      "step": 3120
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.4830219809120518,
+      "learning_rate": 3.3541212993419773e-06,
+      "loss": 0.5475,
+      "step": 3130
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.4801836621711601,
+      "learning_rate": 3.3326838565629895e-06,
+      "loss": 0.5413,
+      "step": 3140
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.47387958333244534,
+      "learning_rate": 3.31125270331104e-06,
+      "loss": 0.5537,
+      "step": 3150
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.5090490511350312,
+      "learning_rate": 3.289828645197681e-06,
+      "loss": 0.5567,
+      "step": 3160
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.5286353188714713,
+      "learning_rate": 3.2684124875677518e-06,
+      "loss": 0.5589,
+      "step": 3170
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.4927074981163475,
+      "learning_rate": 3.247005035469109e-06,
+      "loss": 0.5697,
+      "step": 3180
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.47340856305327644,
+      "learning_rate": 3.2256070936223603e-06,
+      "loss": 0.5687,
+      "step": 3190
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.5115028667136483,
+      "learning_rate": 3.2042194663906193e-06,
+      "loss": 0.5625,
+      "step": 3200
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.4723602653535651,
+      "learning_rate": 3.182842957749263e-06,
+      "loss": 0.5633,
+      "step": 3210
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.4679538952450783,
+      "learning_rate": 3.1614783712557156e-06,
+      "loss": 0.5572,
+      "step": 3220
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.48051919166640805,
+      "learning_rate": 3.1401265100192383e-06,
+      "loss": 0.5648,
+      "step": 3230
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.4594423765819446,
+      "learning_rate": 3.1187881766707425e-06,
+      "loss": 0.5595,
+      "step": 3240
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.49220125939314296,
+      "learning_rate": 3.0974641733326154e-06,
+      "loss": 0.5479,
+      "step": 3250
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4944265110257382,
+      "learning_rate": 3.0761553015885717e-06,
+      "loss": 0.5502,
+      "step": 3260
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.495744161270211,
+      "learning_rate": 3.0548623624535165e-06,
+      "loss": 0.5629,
+      "step": 3270
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.478561888744776,
+      "learning_rate": 3.0335861563434403e-06,
+      "loss": 0.5597,
+      "step": 3280
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4946624980435279,
+      "learning_rate": 3.012327483045325e-06,
+      "loss": 0.556,
+      "step": 3290
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.4913013156645444,
+      "learning_rate": 2.9910871416870855e-06,
+      "loss": 0.5638,
+      "step": 3300
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.46629667333688474,
+      "learning_rate": 2.9698659307075224e-06,
+      "loss": 0.5508,
+      "step": 3310
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.47577400823898375,
+      "learning_rate": 2.948664647826318e-06,
+      "loss": 0.5518,
+      "step": 3320
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.48528006049817207,
+      "learning_rate": 2.9274840900140375e-06,
+      "loss": 0.5582,
+      "step": 3330
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.5499143301618472,
+      "learning_rate": 2.906325053462181e-06,
+      "loss": 0.548,
+      "step": 3340
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.4772816560553211,
+      "learning_rate": 2.8851883335532496e-06,
+      "loss": 0.5523,
+      "step": 3350
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.49887071761697505,
+      "learning_rate": 2.8640747248308445e-06,
+      "loss": 0.5544,
+      "step": 3360
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.4853842631362592,
+      "learning_rate": 2.8429850209698053e-06,
+      "loss": 0.5558,
+      "step": 3370
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.45895465861964546,
+      "learning_rate": 2.8219200147463677e-06,
+      "loss": 0.5598,
+      "step": 3380
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.4662802877247775,
+      "learning_rate": 2.8008804980083695e-06,
+      "loss": 0.5551,
+      "step": 3390
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.4881083174435456,
+      "learning_rate": 2.7798672616454785e-06,
+      "loss": 0.5511,
+      "step": 3400
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.5016617932642891,
+      "learning_rate": 2.75888109555947e-06,
+      "loss": 0.5438,
+      "step": 3410
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.4831166076149674,
+      "learning_rate": 2.7379227886345244e-06,
+      "loss": 0.5598,
+      "step": 3420
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.4953933886035155,
+      "learning_rate": 2.716993128707581e-06,
+      "loss": 0.5609,
+      "step": 3430
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.503170266490847,
+      "learning_rate": 2.696092902538716e-06,
+      "loss": 0.5488,
+      "step": 3440
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.5098380667106547,
+      "learning_rate": 2.675222895781574e-06,
+      "loss": 0.5539,
+      "step": 3450
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.49948084086860606,
+      "learning_rate": 2.6543838929538285e-06,
+      "loss": 0.5581,
+      "step": 3460
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.4872613273522286,
+      "learning_rate": 2.6335766774076965e-06,
+      "loss": 0.5562,
+      "step": 3470
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.47926716145131487,
+      "learning_rate": 2.6128020313004875e-06,
+      "loss": 0.5561,
+      "step": 3480
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.49339314189894584,
+      "learning_rate": 2.592060735565206e-06,
+      "loss": 0.5633,
+      "step": 3490
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.4888816777932096,
+      "learning_rate": 2.5713535698811926e-06,
+      "loss": 0.5623,
+      "step": 3500
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.47873225411797143,
+      "learning_rate": 2.550681312644815e-06,
+      "loss": 0.5629,
+      "step": 3510
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.4985498589688127,
+      "learning_rate": 2.5300447409402104e-06,
+      "loss": 0.5517,
+      "step": 3520
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.4699404709889953,
+      "learning_rate": 2.509444630510071e-06,
+      "loss": 0.5542,
+      "step": 3530
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.5471742855253533,
+      "learning_rate": 2.4888817557264883e-06,
+      "loss": 0.5573,
+      "step": 3540
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.4890601716460387,
+      "learning_rate": 2.468356889561835e-06,
+      "loss": 0.5496,
+      "step": 3550
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.4884550896007432,
+      "learning_rate": 2.4478708035597206e-06,
+      "loss": 0.5517,
+      "step": 3560
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.53082092791935,
+      "learning_rate": 2.427424267805977e-06,
+      "loss": 0.5643,
+      "step": 3570
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4588900957688972,
+      "learning_rate": 2.407018050899719e-06,
+      "loss": 0.5588,
+      "step": 3580
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4930240761419014,
+      "learning_rate": 2.3866529199244454e-06,
+      "loss": 0.5534,
+      "step": 3590
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.4995410840918172,
+      "learning_rate": 2.36632964041921e-06,
+      "loss": 0.5526,
+      "step": 3600
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.4889682103736911,
+      "learning_rate": 2.3460489763498393e-06,
+      "loss": 0.5575,
+      "step": 3610
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.47254332660748083,
+      "learning_rate": 2.3258116900802188e-06,
+      "loss": 0.5641,
+      "step": 3620
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.5271806756431864,
+      "learning_rate": 2.3056185423436304e-06,
+      "loss": 0.5515,
+      "step": 3630
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.5014716634327129,
+      "learning_rate": 2.2854702922141627e-06,
+      "loss": 0.5578,
+      "step": 3640
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.48930981901485066,
+      "learning_rate": 2.265367697078168e-06,
+      "loss": 0.5648,
+      "step": 3650
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.4822043988267899,
+      "learning_rate": 2.245311512605801e-06,
+      "loss": 0.5554,
+      "step": 3660
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.4978119631671631,
+      "learning_rate": 2.2253024927226053e-06,
+      "loss": 0.5586,
+      "step": 3670
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.49756480432664524,
+      "learning_rate": 2.2053413895811764e-06,
+      "loss": 0.5578,
+      "step": 3680
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.4671920108876918,
+      "learning_rate": 2.1854289535328864e-06,
+      "loss": 0.5557,
+      "step": 3690
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.513655855548841,
+      "learning_rate": 2.165565933099682e-06,
+      "loss": 0.5589,
+      "step": 3700
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.46274876339767745,
+      "learning_rate": 2.1457530749459373e-06,
+      "loss": 0.5588,
+      "step": 3710
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.48340392958868733,
+      "learning_rate": 2.1259911238503988e-06,
+      "loss": 0.5481,
+      "step": 3720
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.5024001177410511,
+      "learning_rate": 2.1062808226781767e-06,
+      "loss": 0.5604,
+      "step": 3730
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.4794062865649958,
+      "learning_rate": 2.0866229123528305e-06,
+      "loss": 0.552,
+      "step": 3740
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.49502474291815657,
+      "learning_rate": 2.0670181318285076e-06,
+      "loss": 0.5526,
+      "step": 3750
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.4912138589836612,
+      "learning_rate": 2.0474672180621754e-06,
+      "loss": 0.5433,
+      "step": 3760
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.46287983551015915,
+      "learning_rate": 2.027970905985908e-06,
+      "loss": 0.5607,
+      "step": 3770
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.4818908530273005,
+      "learning_rate": 2.008529928479269e-06,
+      "loss": 0.5552,
+      "step": 3780
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.49475825963312386,
+      "learning_rate": 1.9891450163417574e-06,
+      "loss": 0.5473,
+      "step": 3790
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.5090335613659759,
+      "learning_rate": 1.9698168982653334e-06,
+      "loss": 0.5469,
+      "step": 3800
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.48712846229296525,
+      "learning_rate": 1.950546300807037e-06,
+      "loss": 0.5526,
+      "step": 3810
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.5087151308068611,
+      "learning_rate": 1.931333948361664e-06,
+      "loss": 0.563,
+      "step": 3820
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.4770122954574883,
+      "learning_rate": 1.9121805631345406e-06,
+      "loss": 0.5588,
+      "step": 3830
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.49875337542296333,
+      "learning_rate": 1.8930868651143776e-06,
+      "loss": 0.5556,
+      "step": 3840
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.46661280379905284,
+      "learning_rate": 1.8740535720462034e-06,
+      "loss": 0.5518,
+      "step": 3850
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.49444595207088565,
+      "learning_rate": 1.8550813994043814e-06,
+      "loss": 0.5679,
+      "step": 3860
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.48381227476419236,
+      "learning_rate": 1.8361710603657162e-06,
+      "loss": 0.5572,
+      "step": 3870
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.5055312948711096,
+      "learning_rate": 1.8173232657826508e-06,
+      "loss": 0.5538,
+      "step": 3880
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.4686625212413926,
+      "learning_rate": 1.7985387241565343e-06,
+      "loss": 0.559,
+      "step": 3890
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.4804255341689684,
+      "learning_rate": 1.7798181416109966e-06,
+      "loss": 0.544,
+      "step": 3900
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.5090131219052505,
+      "learning_rate": 1.7611622218654e-06,
+      "loss": 0.5565,
+      "step": 3910
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.4823380469403731,
+      "learning_rate": 1.7425716662083936e-06,
+      "loss": 0.5586,
+      "step": 3920
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.5039478306212927,
+      "learning_rate": 1.7240471734715416e-06,
+      "loss": 0.5582,
+      "step": 3930
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.48106143586192984,
+      "learning_rate": 1.7055894400030597e-06,
+      "loss": 0.5527,
+      "step": 3940
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.4948095621947108,
+      "learning_rate": 1.6871991596416367e-06,
+      "loss": 0.5534,
+      "step": 3950
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.47985601211032985,
+      "learning_rate": 1.668877023690356e-06,
+      "loss": 0.5514,
+      "step": 3960
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.5044751224020304,
+      "learning_rate": 1.6506237208907045e-06,
+      "loss": 0.5541,
+      "step": 3970
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.5080452899508979,
+      "learning_rate": 1.6324399373966833e-06,
+      "loss": 0.5506,
+      "step": 3980
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4931986436565961,
+      "learning_rate": 1.6143263567490192e-06,
+      "loss": 0.5736,
+      "step": 3990
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.4684816221900875,
+      "learning_rate": 1.596283659849464e-06,
+      "loss": 0.556,
+      "step": 4000
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.4785014812413059,
+      "learning_rate": 1.5783125249352016e-06,
+      "loss": 0.5579,
+      "step": 4010
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.5116019647376474,
+      "learning_rate": 1.5604136275533513e-06,
+      "loss": 0.5552,
+      "step": 4020
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.5395436792240803,
+      "learning_rate": 1.5425876405355793e-06,
+      "loss": 0.5384,
+      "step": 4030
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4900436595350879,
+      "learning_rate": 1.5248352339727968e-06,
+      "loss": 0.5622,
+      "step": 4040
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.47513280378884526,
+      "learning_rate": 1.5071570751899785e-06,
+      "loss": 0.5636,
+      "step": 4050
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.4839906292088417,
+      "learning_rate": 1.4895538287210727e-06,
+      "loss": 0.5527,
+      "step": 4060
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.5376958097507211,
+      "learning_rate": 1.4720261562840272e-06,
+      "loss": 0.5635,
+      "step": 4070
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.48771290149288943,
+      "learning_rate": 1.4545747167559066e-06,
+      "loss": 0.564,
+      "step": 4080
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4854524808894032,
+      "learning_rate": 1.4372001661481314e-06,
+      "loss": 0.5598,
+      "step": 4090
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.4700143505212195,
+      "learning_rate": 1.4199031575818126e-06,
+      "loss": 0.5375,
+      "step": 4100
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.4915439052479703,
+      "learning_rate": 1.4026843412632083e-06,
+      "loss": 0.5548,
+      "step": 4110
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.4869720592283153,
+      "learning_rate": 1.385544364459273e-06,
+      "loss": 0.5571,
+      "step": 4120
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.4716126280570366,
+      "learning_rate": 1.3684838714733317e-06,
+      "loss": 0.5516,
+      "step": 4130
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.4965381533290548,
+      "learning_rate": 1.3515035036208578e-06,
+      "loss": 0.5578,
+      "step": 4140
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.49674828915458996,
+      "learning_rate": 1.3346038992053705e-06,
+      "loss": 0.5498,
+      "step": 4150
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.47680857026122736,
+      "learning_rate": 1.3177856934944328e-06,
+      "loss": 0.5531,
+      "step": 4160
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.4870948629881832,
+      "learning_rate": 1.3010495186957768e-06,
+      "loss": 0.552,
+      "step": 4170
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.483089196852953,
+      "learning_rate": 1.2843960039335355e-06,
+      "loss": 0.5564,
+      "step": 4180
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.5140997811965615,
+      "learning_rate": 1.2678257752245992e-06,
+      "loss": 0.5504,
+      "step": 4190
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.4779902409617231,
+      "learning_rate": 1.2513394554550753e-06,
+      "loss": 0.5478,
+      "step": 4200
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.47680861915825756,
+      "learning_rate": 1.2349376643568792e-06,
+      "loss": 0.5555,
+      "step": 4210
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.47618772244534097,
+      "learning_rate": 1.218621018484434e-06,
+      "loss": 0.5509,
+      "step": 4220
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.46991117646305874,
+      "learning_rate": 1.202390131191501e-06,
+      "loss": 0.5572,
+      "step": 4230
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.48145576248836425,
+      "learning_rate": 1.1862456126081136e-06,
+      "loss": 0.562,
+      "step": 4240
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.49862994419451123,
+      "learning_rate": 1.170188069617649e-06,
+      "loss": 0.5574,
+      "step": 4250
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.5025682535998525,
+      "learning_rate": 1.1542181058340122e-06,
+      "loss": 0.5569,
+      "step": 4260
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.47850092658350835,
+      "learning_rate": 1.1383363215789488e-06,
+      "loss": 0.5543,
+      "step": 4270
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.5044422425999335,
+      "learning_rate": 1.1225433138594741e-06,
+      "loss": 0.5599,
+      "step": 4280
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.47419325850109234,
+      "learning_rate": 1.1068396763454339e-06,
+      "loss": 0.5586,
+      "step": 4290
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.560597143802205,
+      "learning_rate": 1.0912259993471857e-06,
+      "loss": 0.5524,
+      "step": 4300
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.5148468793364267,
+      "learning_rate": 1.0757028697934152e-06,
+      "loss": 0.5084,
+      "step": 4310
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.5017714203601242,
+      "learning_rate": 1.060270871209064e-06,
+      "loss": 0.5156,
+      "step": 4320
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.49357251631602217,
+      "learning_rate": 1.0449305836934003e-06,
+      "loss": 0.5109,
+      "step": 4330
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.4936913138076729,
+      "learning_rate": 1.02968258389821e-06,
+      "loss": 0.5158,
+      "step": 4340
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.5049259973539401,
+      "learning_rate": 1.0145274450061254e-06,
+      "loss": 0.5217,
+      "step": 4350
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.517079836314341,
+      "learning_rate": 9.994657367090686e-07,
+      "loss": 0.5136,
+      "step": 4360
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.4837364294449262,
+      "learning_rate": 9.844980251868449e-07,
+      "loss": 0.518,
+      "step": 4370
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.4869343961795407,
+      "learning_rate": 9.696248730858605e-07,
+      "loss": 0.5132,
+      "step": 4380
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.5085658265111329,
+      "learning_rate": 9.54846839497964e-07,
+      "loss": 0.5165,
+      "step": 4390
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.47424129042024027,
+      "learning_rate": 9.401644799394382e-07,
+      "loss": 0.5215,
+      "step": 4400
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.4991885159298539,
+      "learning_rate": 9.255783463301111e-07,
+      "loss": 0.5092,
+      "step": 4410
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.47972707851164975,
+      "learning_rate": 9.110889869726167e-07,
+      "loss": 0.5289,
+      "step": 4420
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.48477312158187885,
+      "learning_rate": 8.966969465317753e-07,
+      "loss": 0.5373,
+      "step": 4430
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5150113149802942,
+      "learning_rate": 8.824027660141253e-07,
+      "loss": 0.5144,
+      "step": 4440
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.5012820847152873,
+      "learning_rate": 8.682069827475828e-07,
+      "loss": 0.5232,
+      "step": 4450
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.536197598669663,
+      "learning_rate": 8.541101303612473e-07,
+      "loss": 0.5312,
+      "step": 4460
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.47456874746453287,
+      "learning_rate": 8.401127387653379e-07,
+      "loss": 0.5021,
+      "step": 4470
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.5022494921077733,
+      "learning_rate": 8.262153341312734e-07,
+      "loss": 0.5039,
+      "step": 4480
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5128622291867768,
+      "learning_rate": 8.124184388719e-07,
+      "loss": 0.5189,
+      "step": 4490
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.49970434341288505,
+      "learning_rate": 7.987225716218441e-07,
+      "loss": 0.5266,
+      "step": 4500
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.4990813361708124,
+      "learning_rate": 7.851282472180222e-07,
+      "loss": 0.5189,
+      "step": 4510
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.5361324180050252,
+      "learning_rate": 7.716359766802858e-07,
+      "loss": 0.5283,
+      "step": 4520
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.49325303865409753,
+      "learning_rate": 7.582462671922154e-07,
+      "loss": 0.5134,
+      "step": 4530
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.5074499214352016,
+      "learning_rate": 7.449596220820492e-07,
+      "loss": 0.5219,
+      "step": 4540
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.48687866167974014,
+      "learning_rate": 7.317765408037668e-07,
+      "loss": 0.5131,
+      "step": 4550
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.5209017279406115,
+      "learning_rate": 7.186975189183119e-07,
+      "loss": 0.5263,
+      "step": 4560
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.5017929271897994,
+      "learning_rate": 7.057230480749689e-07,
+      "loss": 0.5221,
+      "step": 4570
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.4909543768911595,
+      "learning_rate": 6.928536159928746e-07,
+      "loss": 0.5082,
+      "step": 4580
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.5217040631589964,
+      "learning_rate": 6.800897064426877e-07,
+      "loss": 0.5136,
+      "step": 4590
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.5007485735211247,
+      "learning_rate": 6.674317992284038e-07,
+      "loss": 0.5158,
+      "step": 4600
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.495432605404129,
+      "learning_rate": 6.548803701693218e-07,
+      "loss": 0.5191,
+      "step": 4610
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.5457479536125451,
+      "learning_rate": 6.424358910821511e-07,
+      "loss": 0.5144,
+      "step": 4620
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.5106414076169086,
+      "learning_rate": 6.300988297632804e-07,
+      "loss": 0.5288,
+      "step": 4630
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.5211736668510725,
+      "learning_rate": 6.178696499711915e-07,
+      "loss": 0.5218,
+      "step": 4640
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.4891406143758845,
+      "learning_rate": 6.057488114090288e-07,
+      "loss": 0.5107,
+      "step": 4650
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.5178228254981688,
+      "learning_rate": 5.937367697073139e-07,
+      "loss": 0.5004,
+      "step": 4660
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.49831173988741256,
+      "learning_rate": 5.818339764068217e-07,
+      "loss": 0.5167,
+      "step": 4670
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.5445792027132667,
+      "learning_rate": 5.700408789416051e-07,
+      "loss": 0.5251,
+      "step": 4680
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.5412064520692698,
+      "learning_rate": 5.58357920622179e-07,
+      "loss": 0.5185,
+      "step": 4690
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.5194173017222409,
+      "learning_rate": 5.467855406188503e-07,
+      "loss": 0.5213,
+      "step": 4700
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.530585691377951,
+      "learning_rate": 5.353241739452134e-07,
+      "loss": 0.5213,
+      "step": 4710
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.5334266089134705,
+      "learning_rate": 5.239742514417958e-07,
+      "loss": 0.5213,
+      "step": 4720
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.5323190599173516,
+      "learning_rate": 5.127361997598647e-07,
+      "loss": 0.5173,
+      "step": 4730
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.4977075988891876,
+      "learning_rate": 5.016104413453866e-07,
+      "loss": 0.5163,
+      "step": 4740
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.5072133518376746,
+      "learning_rate": 4.905973944231479e-07,
+      "loss": 0.5147,
+      "step": 4750
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.5089446326634548,
+      "learning_rate": 4.796974729810328e-07,
+      "loss": 0.5206,
+      "step": 4760
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.5173579821056443,
+      "learning_rate": 4.6891108675446453e-07,
+      "loss": 0.5233,
+      "step": 4770
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.49509093398735665,
+      "learning_rate": 4.5823864121099967e-07,
+      "loss": 0.5143,
+      "step": 4780
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.510739525920679,
+      "learning_rate": 4.476805375350865e-07,
+      "loss": 0.5204,
+      "step": 4790
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.5285640385275354,
+      "learning_rate": 4.372371726129854e-07,
+      "loss": 0.5226,
+      "step": 4800
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.49804779846917624,
+      "learning_rate": 4.269089390178512e-07,
+      "loss": 0.5257,
+      "step": 4810
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.4960403523798791,
+      "learning_rate": 4.1669622499497205e-07,
+      "loss": 0.5224,
+      "step": 4820
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.509776799973484,
+      "learning_rate": 4.0659941444717833e-07,
+      "loss": 0.5153,
+      "step": 4830
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.48108044641737857,
+      "learning_rate": 3.966188869204094e-07,
+      "loss": 0.5175,
+      "step": 4840
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.5141883943099625,
+      "learning_rate": 3.8675501758944926e-07,
+      "loss": 0.5147,
+      "step": 4850
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.5086149236998669,
+      "learning_rate": 3.7700817724381983e-07,
+      "loss": 0.5128,
+      "step": 4860
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.5107670739104685,
+      "learning_rate": 3.6737873227384263e-07,
+      "loss": 0.5162,
+      "step": 4870
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.48090817905611477,
+      "learning_rate": 3.578670446568711e-07,
+      "loss": 0.5289,
+      "step": 4880
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.5149098967385166,
+      "learning_rate": 3.484734719436782e-07,
+      "loss": 0.5224,
+      "step": 4890
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.4967090096149114,
+      "learning_rate": 3.3919836724501743e-07,
+      "loss": 0.5064,
+      "step": 4900
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.49198009223776107,
+      "learning_rate": 3.3004207921835004e-07,
+      "loss": 0.526,
+      "step": 4910
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.5260886992405347,
+      "learning_rate": 3.210049520547388e-07,
+      "loss": 0.5278,
+      "step": 4920
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.49827609520509064,
+      "learning_rate": 3.1208732546590843e-07,
+      "loss": 0.5269,
+      "step": 4930
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.5199185251610714,
+      "learning_rate": 3.0328953467147543e-07,
+      "loss": 0.5125,
+      "step": 4940
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.5165139482645277,
+      "learning_rate": 2.946119103863483e-07,
+      "loss": 0.5095,
+      "step": 4950
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.48760733590102007,
+      "learning_rate": 2.86054778808296e-07,
+      "loss": 0.5262,
+      "step": 4960
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.49481920675979196,
+      "learning_rate": 2.7761846160568403e-07,
+      "loss": 0.5209,
+      "step": 4970
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.5017608349952136,
+      "learning_rate": 2.69303275905384e-07,
+      "loss": 0.5137,
+      "step": 4980
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.5222144874040826,
+      "learning_rate": 2.611095342808526e-07,
+      "loss": 0.5162,
+      "step": 4990
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.4928255848647095,
+      "learning_rate": 2.530375447403815e-07,
+      "loss": 0.5176,
+      "step": 5000
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.530457616289496,
+      "learning_rate": 2.4508761071551906e-07,
+      "loss": 0.5181,
+      "step": 5010
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.5147706319208548,
+      "learning_rate": 2.3726003104966393e-07,
+      "loss": 0.5095,
+      "step": 5020
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.523763253857449,
+      "learning_rate": 2.2955509998683214e-07,
+      "loss": 0.5108,
+      "step": 5030
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.5323084690421006,
+      "learning_rate": 2.2197310716059603e-07,
+      "loss": 0.511,
+      "step": 5040
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.5088461348117514,
+      "learning_rate": 2.1451433758319543e-07,
+      "loss": 0.5265,
+      "step": 5050
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.5478220673331649,
+      "learning_rate": 2.0717907163482507e-07,
+      "loss": 0.5112,
+      "step": 5060
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.5414027895276027,
+      "learning_rate": 1.9996758505309593e-07,
+      "loss": 0.5231,
+      "step": 5070
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.4983898932091525,
+      "learning_rate": 1.9288014892266753e-07,
+      "loss": 0.5105,
+      "step": 5080
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.5093531347734784,
+      "learning_rate": 1.8591702966505952e-07,
+      "loss": 0.5127,
+      "step": 5090
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.677948629367298,
+      "learning_rate": 1.790784890286352e-07,
+      "loss": 0.5219,
+      "step": 5100
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.5010683504531009,
+      "learning_rate": 1.7236478407876555e-07,
+      "loss": 0.5054,
+      "step": 5110
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.5179768835662841,
+      "learning_rate": 1.6577616718816123e-07,
+      "loss": 0.5251,
+      "step": 5120
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.5087954420227027,
+      "learning_rate": 1.5931288602738958e-07,
+      "loss": 0.5137,
+      "step": 5130
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.5083448366233918,
+      "learning_rate": 1.5297518355556132e-07,
+      "loss": 0.5059,
+      "step": 5140
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.5170972166302202,
+      "learning_rate": 1.467632980112023e-07,
+      "loss": 0.5214,
+      "step": 5150
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.5145933451855358,
+      "learning_rate": 1.406774629032923e-07,
+      "loss": 0.511,
+      "step": 5160
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.5012480980422283,
+      "learning_rate": 1.347179070024903e-07,
+      "loss": 0.5179,
+      "step": 5170
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.5157422802936725,
+      "learning_rate": 1.2888485433253521e-07,
+      "loss": 0.5193,
+      "step": 5180
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.5104197669978088,
+      "learning_rate": 1.2317852416182378e-07,
+      "loss": 0.5221,
+      "step": 5190
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.48689303934415246,
+      "learning_rate": 1.1759913099516816e-07,
+      "loss": 0.5118,
+      "step": 5200
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.5105879788600957,
+      "learning_rate": 1.1214688456573247e-07,
+      "loss": 0.5178,
+      "step": 5210
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.4742285263986786,
+      "learning_rate": 1.0682198982714814e-07,
+      "loss": 0.534,
+      "step": 5220
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.5096564376650945,
+      "learning_rate": 1.0162464694581235e-07,
+      "loss": 0.5272,
+      "step": 5230
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.5068212494030221,
+      "learning_rate": 9.65550512933605e-08,
+      "loss": 0.5252,
+      "step": 5240
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.5132873703711879,
+      "learning_rate": 9.16133934393224e-08,
+      "loss": 0.5161,
+      "step": 5250
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.496214740845792,
+      "learning_rate": 8.67998591439612e-08,
+      "loss": 0.518,
+      "step": 5260
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.5257117991696062,
+      "learning_rate": 8.21146293512876e-08,
+      "loss": 0.5201,
+      "step": 5270
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.5038613162646833,
+      "learning_rate": 7.755788018225961e-08,
+      "loss": 0.5439,
+      "step": 5280
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.5108263338716986,
+      "learning_rate": 7.31297829281617e-08,
+      "loss": 0.5132,
+      "step": 5290
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.5149498952477054,
+      "learning_rate": 6.883050404416552e-08,
+      "loss": 0.5111,
+      "step": 5300
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.5047749285108949,
+      "learning_rate": 6.46602051430732e-08,
+      "loss": 0.5307,
+      "step": 5310
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.5156795764243357,
+      "learning_rate": 6.061904298924253e-08,
+      "loss": 0.5285,
+      "step": 5320
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.5144201053701509,
+      "learning_rate": 5.670716949269278e-08,
+      "loss": 0.5148,
+      "step": 5330
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.507394331507882,
+      "learning_rate": 5.2924731703395564e-08,
+      "loss": 0.5206,
+      "step": 5340
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.48368946217469994,
+      "learning_rate": 4.927187180574666e-08,
+      "loss": 0.526,
+      "step": 5350
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.5047554925764675,
+      "learning_rate": 4.574872711322103e-08,
+      "loss": 0.5126,
+      "step": 5360
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.4949463708763226,
+      "learning_rate": 4.2355430063211405e-08,
+      "loss": 0.5204,
+      "step": 5370
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.5079311960306774,
+      "learning_rate": 3.909210821205017e-08,
+      "loss": 0.5189,
+      "step": 5380
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.4902741464423996,
+      "learning_rate": 3.595888423021354e-08,
+      "loss": 0.513,
+      "step": 5390
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.5421885848655773,
+      "learning_rate": 3.295587589771071e-08,
+      "loss": 0.5093,
+      "step": 5400
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.49756539244831294,
+      "learning_rate": 3.008319609965676e-08,
+      "loss": 0.5144,
+      "step": 5410
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.5074328229331989,
+      "learning_rate": 2.734095282202942e-08,
+      "loss": 0.5133,
+      "step": 5420
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.49772891591572227,
+      "learning_rate": 2.4729249147608378e-08,
+      "loss": 0.5251,
+      "step": 5430
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.5088034449477752,
+      "learning_rate": 2.224818325210237e-08,
+      "loss": 0.5175,
+      "step": 5440
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.4826584150965653,
+      "learning_rate": 1.9897848400456496e-08,
+      "loss": 0.5141,
+      "step": 5450
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.5172662799041124,
+      "learning_rate": 1.7678332943348807e-08,
+      "loss": 0.5197,
+      "step": 5460
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.48940063691629393,
+      "learning_rate": 1.5589720313866794e-08,
+      "loss": 0.5059,
+      "step": 5470
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.517098264403305,
+      "learning_rate": 1.3632089024371574e-08,
+      "loss": 0.5141,
+      "step": 5480
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.48979313956431636,
+      "learning_rate": 1.1805512663549345e-08,
+      "loss": 0.5136,
+      "step": 5490
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.48660701715860905,
+      "learning_rate": 1.0110059893640055e-08,
+      "loss": 0.5212,
+      "step": 5500
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.4841422308843411,
+      "learning_rate": 8.54579444786152e-09,
+      "loss": 0.5228,
+      "step": 5510
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.4851052180007293,
+      "learning_rate": 7.112775128009174e-09,
+      "loss": 0.5146,
+      "step": 5520
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.49400550323274894,
+      "learning_rate": 5.811055802249721e-09,
+      "loss": 0.5277,
+      "step": 5530
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.512928633478054,
+      "learning_rate": 4.640685403093147e-09,
+      "loss": 0.5216,
+      "step": 5540
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.48929257944769156,
+      "learning_rate": 3.6017079255547534e-09,
+      "loss": 0.5172,
+      "step": 5550
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.5049424795736568,
+      "learning_rate": 2.6941624255001904e-09,
+      "loss": 0.5147,
+      "step": 5560
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.5046094240590331,
+      "learning_rate": 1.9180830181797505e-09,
+      "loss": 0.5222,
+      "step": 5570
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.5035868811303936,
+      "learning_rate": 1.273498876942558e-09,
+      "loss": 0.511,
+      "step": 5580
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.48822795586505313,
+      "learning_rate": 7.604342321435032e-10,
+      "loss": 0.5222,
+      "step": 5590
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.5199444169728372,
+      "learning_rate": 3.789083702293028e-10,
+      "loss": 0.5236,
+      "step": 5600
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.528390397947652,
+      "learning_rate": 1.2893563301535904e-10,
+      "loss": 0.5187,
+      "step": 5610
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.49922315857737276,
+      "learning_rate": 1.0525417146023396e-11,
+      "loss": 0.5179,
+      "step": 5620
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 5624,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 500,
+  "total_flos": 2354981319475200.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d85d2ff5720e05dbb5441c3d20e7ba4b6a45de57469d8179edcc7769226b30da
+size 7800

zero_to_fp32.py

@@ -0,0 +1,592 @@
+#!/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 _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+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() if _has_callable(shape, 'numel') else math.prod(shape)
+            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)