Upload 16 files

config.json

@@ -0,0 +1,24 @@
+{
+  "_name_or_path": "./WizardLM-Uncensored-SuperCOT-StoryTelling-30b",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 6656,
+  "initializer_range": 0.02,
+  "intermediate_size": 17920,
+  "max_position_embeddings": 2048,
+  "max_sequence_length": 2048,
+  "model_type": "llama",
+  "num_attention_heads": 52,
+  "num_hidden_layers": 60,
+  "pad_token_id": 0,
+  "rms_norm_eps": 1e-06,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.30.2",
+  "use_cache": true,
+  "vocab_size": 32001
+}

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "pad_token_id": 0,
+  "transformers_version": "4.30.2"
+}

pytorch_model-00001-of-00007.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45a142bbf21fe6728a9b4b835af37c05167ef6046da9f16f2b4c3ffc9b472332
+size 9818338426

pytorch_model-00002-of-00007.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2169094a17646878e96cc4b50d8ab80c239b991dcc457a8d723aa130bfda315c
+size 9958103230

pytorch_model-00003-of-00007.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a97bb5dfaf13f5a616369b7c04344cf12bfe48b3d061f3cefd751fc8ebdf47ba
+size 9896735202

pytorch_model-00004-of-00007.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be076d894c22c8d28c3ac6a14e96003ac9858eacd52936112c602e87522bf60d
+size 9869470968

pytorch_model-00005-of-00007.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:028bb9076791a8c844118fabe5416097da476780c6332e5a35471c002d820487
+size 9869470932

pytorch_model-00006-of-00007.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec35cdbf8e5ada0a5f486972f12f80895d565c3e2f9468d639c91e7ea4502354
+size 9958103230

pytorch_model-00007-of-00007.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4866a7c2f50d1240657e420a28705d6f2a1614a2cfd0a829d4c1e44d67f33e51
+size 5687917080

pytorch_model.bin.index.json

@@ -0,0 +1,610 @@
+{
+  "metadata": {
+    "total_size": 65057929216
+  },
+  "weight_map": {
+    "lm_head.weight": "pytorch_model-00007-of-00007.bin",
+    "model.embed_tokens.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
+    "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
+    "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.10.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.10.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.10.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
+    "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.11.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.11.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.11.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
+    "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.12.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.12.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.12.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
+    "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.13.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.13.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.13.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
+    "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.14.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.14.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.14.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
+    "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.15.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.15.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.15.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
+    "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.16.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.16.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.16.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
+    "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.17.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.17.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.17.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
+    "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.18.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.18.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.18.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
+    "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.19.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.19.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.19.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
+    "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
+    "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.20.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.20.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.20.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
+    "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.21.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.21.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.21.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
+    "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.22.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.22.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.22.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
+    "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.23.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.23.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.23.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
+    "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.24.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.24.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.24.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
+    "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.25.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.25.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.25.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
+    "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.26.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.26.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.26.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
+    "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.27.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.27.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.27.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
+    "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
+    "model.layers.28.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.28.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.28.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
+    "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.29.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.29.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.29.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
+    "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.3.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
+    "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.30.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.30.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.30.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
+    "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.31.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.31.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.31.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
+    "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.32.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.32.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.32.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.32.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.32.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.32.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.32.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.32.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.32.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
+    "model.layers.32.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.33.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.33.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.33.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.33.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.33.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.33.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.33.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.33.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.33.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
+    "model.layers.33.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.34.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.34.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.34.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.34.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.34.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.34.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.34.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.34.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.34.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
+    "model.layers.34.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.35.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.35.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.35.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.35.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.35.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.35.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.35.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.35.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.35.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
+    "model.layers.35.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.36.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.36.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.36.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.36.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.36.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.36.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.36.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.36.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.36.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
+    "model.layers.36.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
+    "model.layers.37.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.37.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.37.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.37.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.37.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.37.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.37.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.37.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.37.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
+    "model.layers.37.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.38.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.38.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.38.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.38.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.38.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.38.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.38.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.38.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.38.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
+    "model.layers.38.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.39.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.39.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.39.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.39.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.39.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.39.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.39.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.39.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.39.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
+    "model.layers.39.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.4.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
+    "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.40.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.40.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.40.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.40.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.40.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.40.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.40.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.40.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.40.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
+    "model.layers.40.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.41.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.41.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.41.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.41.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.41.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.41.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.41.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.41.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.41.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
+    "model.layers.41.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.42.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.42.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.42.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.42.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.42.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.42.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.42.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.42.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.42.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
+    "model.layers.42.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.43.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.43.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.43.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.43.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.43.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.43.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.43.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.43.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.43.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
+    "model.layers.43.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.44.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.44.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.44.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.44.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.44.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.44.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.44.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.44.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.44.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
+    "model.layers.44.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.45.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.45.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.45.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.45.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.45.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.45.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.45.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.45.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.45.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
+    "model.layers.45.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
+    "model.layers.46.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.46.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.46.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.46.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.46.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.46.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.46.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.46.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.46.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
+    "model.layers.46.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.47.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.47.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.47.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.47.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.47.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.47.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.47.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.47.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.47.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
+    "model.layers.47.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.48.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.48.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.48.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.48.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.48.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.48.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.48.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.48.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.48.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
+    "model.layers.48.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.49.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.49.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.49.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.49.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.49.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.49.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.49.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.49.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.49.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
+    "model.layers.49.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.5.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
+    "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.50.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.50.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.50.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.50.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.50.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.50.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.50.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.50.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.50.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
+    "model.layers.50.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.51.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.51.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.51.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.51.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.51.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.51.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.51.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.51.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.51.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
+    "model.layers.51.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.52.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.52.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.52.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.52.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.52.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.52.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.52.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.52.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.52.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
+    "model.layers.52.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.53.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.53.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.53.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.53.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.53.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.53.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.53.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.53.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.53.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
+    "model.layers.53.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.54.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.54.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.54.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.54.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.54.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.54.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.54.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.54.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.54.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
+    "model.layers.54.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.55.input_layernorm.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.55.mlp.down_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.55.mlp.gate_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.55.mlp.up_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.55.post_attention_layernorm.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.55.self_attn.k_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.55.self_attn.o_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.55.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
+    "model.layers.55.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00007.bin",
+    "model.layers.55.self_attn.v_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.56.input_layernorm.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.56.mlp.down_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.56.mlp.gate_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.56.mlp.up_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.56.post_attention_layernorm.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.56.self_attn.k_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.56.self_attn.o_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.56.self_attn.q_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.56.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00007.bin",
+    "model.layers.56.self_attn.v_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.57.input_layernorm.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.57.mlp.down_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.57.mlp.gate_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.57.mlp.up_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.57.post_attention_layernorm.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.57.self_attn.k_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.57.self_attn.o_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.57.self_attn.q_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.57.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00007.bin",
+    "model.layers.57.self_attn.v_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.58.input_layernorm.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.58.mlp.down_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.58.mlp.gate_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.58.mlp.up_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.58.post_attention_layernorm.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.58.self_attn.k_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.58.self_attn.o_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.58.self_attn.q_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.58.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00007.bin",
+    "model.layers.58.self_attn.v_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.59.input_layernorm.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.59.mlp.down_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.59.mlp.gate_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.59.mlp.up_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.59.post_attention_layernorm.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.59.self_attn.k_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.59.self_attn.o_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.59.self_attn.q_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.59.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00007.bin",
+    "model.layers.59.self_attn.v_proj.weight": "pytorch_model-00007-of-00007.bin",
+    "model.layers.6.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
+    "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.7.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
+    "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.8.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.8.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
+    "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
+    "model.layers.9.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.9.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.9.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
+    "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
+    "model.norm.weight": "pytorch_model-00007-of-00007.bin"
+  }
+}

special_tokens_map.json

@@ -0,0 +1,6 @@
+{
+  "bos_token": "</s>",
+  "eos_token": "</s>",
+  "pad_token": "[PAD]",
+  "unk_token": "</s>"
+}

tokenizer.model

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

tokenizer_config.json

@@ -0,0 +1,34 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "bos_token": {
+    "__type": "AddedToken",
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "clean_up_tokenization_spaces": false,
+  "eos_token": {
+    "__type": "AddedToken",
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "model_max_length": 2048,
+  "pad_token": null,
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": {
+    "__type": "AddedToken",
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

trainer_state.json

@@ -0,0 +1,3874 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9965075669383,
+  "global_step": 1287,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.905,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999880457421163e-05,
+      "loss": 0.6497,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999521832542736e-05,
+      "loss": 0.6121,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998924133938902e-05,
+      "loss": 0.7236,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998087375899756e-05,
+      "loss": 0.7515,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9997011578430938e-05,
+      "loss": 0.7073,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9995696767253165e-05,
+      "loss": 0.6146,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994142973801627e-05,
+      "loss": 0.5923,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992350235225215e-05,
+      "loss": 0.5629,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999031859438565e-05,
+      "loss": 0.5383,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988048099856443e-05,
+      "loss": 0.516,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9985538805921757e-05,
+      "loss": 0.5035,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998279077257508e-05,
+      "loss": 0.5244,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9979804065517808e-05,
+      "loss": 0.486,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976578756157684e-05,
+      "loss": 0.4945,
+      "step": 30
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9973114921607055e-05,
+      "loss": 0.4966,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9969412644681077e-05,
+      "loss": 0.4935,
+      "step": 34
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9965472013895685e-05,
+      "loss": 0.4739,
+      "step": 36
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.996129312346552e-05,
+      "loss": 0.4913,
+      "step": 38
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9956876073301645e-05,
+      "loss": 0.4641,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9952220969009175e-05,
+      "loss": 0.4691,
+      "step": 42
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9947327921884746e-05,
+      "loss": 0.4666,
+      "step": 44
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.994219704891385e-05,
+      "loss": 0.4501,
+      "step": 46
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9936828472768043e-05,
+      "loss": 0.4558,
+      "step": 48
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9931222321802016e-05,
+      "loss": 0.4712,
+      "step": 50
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9925378730050518e-05,
+      "loss": 0.4661,
+      "step": 52
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9919297837225152e-05,
+      "loss": 0.4735,
+      "step": 54
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9912979788711042e-05,
+      "loss": 0.4526,
+      "step": 56
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990642473556335e-05,
+      "loss": 0.4453,
+      "step": 58
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9899632834503662e-05,
+      "loss": 0.4713,
+      "step": 60
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.989260424791626e-05,
+      "loss": 0.4622,
+      "step": 62
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9885339143844217e-05,
+      "loss": 0.4585,
+      "step": 64
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.987783769598538e-05,
+      "loss": 0.4576,
+      "step": 66
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9870100083688242e-05,
+      "loss": 0.4353,
+      "step": 68
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9862126491947624e-05,
+      "loss": 0.4509,
+      "step": 70
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985391711140027e-05,
+      "loss": 0.4402,
+      "step": 72
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9845472138320282e-05,
+      "loss": 0.437,
+      "step": 74
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9836791774614437e-05,
+      "loss": 0.4613,
+      "step": 76
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982787622781735e-05,
+      "loss": 0.4567,
+      "step": 78
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9818725711086506e-05,
+      "loss": 0.4541,
+      "step": 80
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.980934044319718e-05,
+      "loss": 0.4398,
+      "step": 82
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9799720648537197e-05,
+      "loss": 0.4283,
+      "step": 84
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.978986655710157e-05,
+      "loss": 0.4443,
+      "step": 86
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9779778404487e-05,
+      "loss": 0.4457,
+      "step": 88
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9769456431886244e-05,
+      "loss": 0.4326,
+      "step": 90
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9758900886082343e-05,
+      "loss": 0.4557,
+      "step": 92
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9748112019442734e-05,
+      "loss": 0.4402,
+      "step": 94
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9737090089913205e-05,
+      "loss": 0.465,
+      "step": 96
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9725835361011726e-05,
+      "loss": 0.4387,
+      "step": 98
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.971434810182217e-05,
+      "loss": 0.4479,
+      "step": 100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9702628586987846e-05,
+      "loss": 0.4344,
+      "step": 102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9690677096704964e-05,
+      "loss": 0.4302,
+      "step": 104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9678493916715914e-05,
+      "loss": 0.4331,
+      "step": 106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.966607933830245e-05,
+      "loss": 0.4224,
+      "step": 108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9653433658278717e-05,
+      "loss": 0.4225,
+      "step": 110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9640557178984152e-05,
+      "loss": 0.4177,
+      "step": 112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9627450208276265e-05,
+      "loss": 0.4546,
+      "step": 114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9614113059523273e-05,
+      "loss": 0.4257,
+      "step": 116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9600546051596604e-05,
+      "loss": 0.4453,
+      "step": 118
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9586749508863284e-05,
+      "loss": 0.458,
+      "step": 120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9572723761178168e-05,
+      "loss": 0.4287,
+      "step": 122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.955846914387607e-05,
+      "loss": 0.4581,
+      "step": 124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.954398599776373e-05,
+      "loss": 0.4343,
+      "step": 126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.952927466911168e-05,
+      "loss": 0.4431,
+      "step": 128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9514335509645948e-05,
+      "loss": 0.4332,
+      "step": 130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9499168876539666e-05,
+      "loss": 0.4315,
+      "step": 132
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9483775132404517e-05,
+      "loss": 0.4403,
+      "step": 134
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.946815464528208e-05,
+      "loss": 0.4618,
+      "step": 136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9452307788635015e-05,
+      "loss": 0.4292,
+      "step": 138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9436234941338145e-05,
+      "loss": 0.4333,
+      "step": 140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9419936487669396e-05,
+      "loss": 0.4557,
+      "step": 142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.94034128173006e-05,
+      "loss": 0.4575,
+      "step": 144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.938666432528819e-05,
+      "loss": 0.4012,
+      "step": 146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9369691412063755e-05,
+      "loss": 0.4579,
+      "step": 148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9352494483424456e-05,
+      "loss": 0.4337,
+      "step": 150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9335073950523335e-05,
+      "loss": 0.4142,
+      "step": 152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9317430229859474e-05,
+      "loss": 0.4545,
+      "step": 154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.929956374326805e-05,
+      "loss": 0.4679,
+      "step": 156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.928147491791024e-05,
+      "loss": 0.4178,
+      "step": 158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9263164186263003e-05,
+      "loss": 0.4474,
+      "step": 160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9244631986108768e-05,
+      "loss": 0.4237,
+      "step": 162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.922587876052492e-05,
+      "loss": 0.4456,
+      "step": 164
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.920690495787326e-05,
+      "loss": 0.412,
+      "step": 166
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.918771103178924e-05,
+      "loss": 0.4279,
+      "step": 168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.916829744117115e-05,
+      "loss": 0.413,
+      "step": 170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9148664650169128e-05,
+      "loss": 0.4508,
+      "step": 172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9128813128174063e-05,
+      "loss": 0.4054,
+      "step": 174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9108743349806382e-05,
+      "loss": 0.4021,
+      "step": 176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.90884557949047e-05,
+      "loss": 0.4392,
+      "step": 178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9067950948514343e-05,
+      "loss": 0.4414,
+      "step": 180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.904722930087575e-05,
+      "loss": 0.4327,
+      "step": 182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9026291347412765e-05,
+      "loss": 0.4081,
+      "step": 184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.900513758872078e-05,
+      "loss": 0.4432,
+      "step": 186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8983768530554765e-05,
+      "loss": 0.4355,
+      "step": 188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8962184683817182e-05,
+      "loss": 0.4292,
+      "step": 190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8940386564545773e-05,
+      "loss": 0.4182,
+      "step": 192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.891837469390122e-05,
+      "loss": 0.4402,
+      "step": 194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8896149598154675e-05,
+      "loss": 0.4377,
+      "step": 196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.887371180867519e-05,
+      "loss": 0.4236,
+      "step": 198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8851061861917013e-05,
+      "loss": 0.4399,
+      "step": 200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8828200299406747e-05,
+      "loss": 0.4285,
+      "step": 202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8805127667730426e-05,
+      "loss": 0.4465,
+      "step": 204
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.878184451852042e-05,
+      "loss": 0.4264,
+      "step": 206
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8758351408442278e-05,
+      "loss": 0.4196,
+      "step": 208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8734648899181388e-05,
+      "loss": 0.4104,
+      "step": 210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.871073755742957e-05,
+      "loss": 0.4188,
+      "step": 212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.868661795487151e-05,
+      "loss": 0.4418,
+      "step": 214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8662290668171107e-05,
+      "loss": 0.4183,
+      "step": 216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8637756278957683e-05,
+      "loss": 0.4076,
+      "step": 218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8613015373812066e-05,
+      "loss": 0.4105,
+      "step": 220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8588068544252572e-05,
+      "loss": 0.4478,
+      "step": 222
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8562916386720883e-05,
+      "loss": 0.4312,
+      "step": 224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.853755950256774e-05,
+      "loss": 0.4044,
+      "step": 226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8511998498038615e-05,
+      "loss": 0.4069,
+      "step": 228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8486233984259186e-05,
+      "loss": 0.4349,
+      "step": 230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8460266577220733e-05,
+      "loss": 0.4039,
+      "step": 232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8434096897765422e-05,
+      "loss": 0.4153,
+      "step": 234
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8407725571571448e-05,
+      "loss": 0.4188,
+      "step": 236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.838115322913807e-05,
+      "loss": 0.4409,
+      "step": 238
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.835438050577057e-05,
+      "loss": 0.4109,
+      "step": 240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8327408041565013e-05,
+      "loss": 0.4247,
+      "step": 242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8300236481392995e-05,
+      "loss": 0.4451,
+      "step": 244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8272866474886185e-05,
+      "loss": 0.4127,
+      "step": 246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8245298676420814e-05,
+      "loss": 0.4346,
+      "step": 248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8217533745102032e-05,
+      "loss": 0.4078,
+      "step": 250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.818957234474813e-05,
+      "loss": 0.4034,
+      "step": 252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.81614151438747e-05,
+      "loss": 0.4355,
+      "step": 254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8133062815678614e-05,
+      "loss": 0.446,
+      "step": 256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.810451603802196e-05,
+      "loss": 0.4329,
+      "step": 258
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.807577549341582e-05,
+      "loss": 0.4387,
+      "step": 260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8046841869003962e-05,
+      "loss": 0.4001,
+      "step": 262
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8017715856546397e-05,
+      "loss": 0.4109,
+      "step": 264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7988398152402857e-05,
+      "loss": 0.4156,
+      "step": 266
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7958889457516134e-05,
+      "loss": 0.4121,
+      "step": 268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7929190477395318e-05,
+      "loss": 0.4187,
+      "step": 270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7899301922098958e-05,
+      "loss": 0.4072,
+      "step": 272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7869224506218034e-05,
+      "loss": 0.4556,
+      "step": 274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7838958948858923e-05,
+      "loss": 0.4135,
+      "step": 276
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7808505973626183e-05,
+      "loss": 0.4384,
+      "step": 278
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.777786630860525e-05,
+      "loss": 0.4226,
+      "step": 280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.774704068634504e-05,
+      "loss": 0.4362,
+      "step": 282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.771602984384043e-05,
+      "loss": 0.4243,
+      "step": 284
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7684834522514632e-05,
+      "loss": 0.4622,
+      "step": 286
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7653455468201483e-05,
+      "loss": 0.448,
+      "step": 288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7621893431127596e-05,
+      "loss": 0.4385,
+      "step": 290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.759014916589443e-05,
+      "loss": 0.4149,
+      "step": 292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7558223431460254e-05,
+      "loss": 0.4229,
+      "step": 294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7526116991121988e-05,
+      "loss": 0.4115,
+      "step": 296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7493830612496975e-05,
+      "loss": 0.4204,
+      "step": 298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7461365067504602e-05,
+      "loss": 0.4171,
+      "step": 300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7428721132347863e-05,
+      "loss": 0.4161,
+      "step": 302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.73958995874948e-05,
+      "loss": 0.4214,
+      "step": 304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7362901217659833e-05,
+      "loss": 0.4175,
+      "step": 306
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7329726811785012e-05,
+      "loss": 0.4105,
+      "step": 308
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7296377163021133e-05,
+      "loss": 0.4354,
+      "step": 310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7262853068708807e-05,
+      "loss": 0.4113,
+      "step": 312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7229155330359368e-05,
+      "loss": 0.452,
+      "step": 314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.719528475363573e-05,
+      "loss": 0.4154,
+      "step": 316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7161242148333107e-05,
+      "loss": 0.4236,
+      "step": 318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.712702832835966e-05,
+      "loss": 0.4146,
+      "step": 320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7092644111717052e-05,
+      "loss": 0.4183,
+      "step": 322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7058090320480866e-05,
+      "loss": 0.4038,
+      "step": 324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.702336778078096e-05,
+      "loss": 0.4135,
+      "step": 326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.698847732278173e-05,
+      "loss": 0.408,
+      "step": 328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6953419780662232e-05,
+      "loss": 0.4003,
+      "step": 330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6918195992596274e-05,
+      "loss": 0.4065,
+      "step": 332
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6882806800732338e-05,
+      "loss": 0.4205,
+      "step": 334
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6847253051173487e-05,
+      "loss": 0.4135,
+      "step": 336
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6811535593957093e-05,
+      "loss": 0.3965,
+      "step": 338
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6775655283034548e-05,
+      "loss": 0.4028,
+      "step": 340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6739612976250836e-05,
+      "loss": 0.4578,
+      "step": 342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.670340953532401e-05,
+      "loss": 0.4298,
+      "step": 344
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6667045825824616e-05,
+      "loss": 0.4221,
+      "step": 346
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.663052271715497e-05,
+      "loss": 0.4062,
+      "step": 348
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6593841082528394e-05,
+      "loss": 0.3934,
+      "step": 350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6557001798948324e-05,
+      "loss": 0.4279,
+      "step": 352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6520005747187358e-05,
+      "loss": 0.3993,
+      "step": 354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.648285381176618e-05,
+      "loss": 0.4191,
+      "step": 356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6445546880932425e-05,
+      "loss": 0.4198,
+      "step": 358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6408085846639435e-05,
+      "loss": 0.4,
+      "step": 360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.637047160452494e-05,
+      "loss": 0.4347,
+      "step": 362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6332705053889643e-05,
+      "loss": 0.4188,
+      "step": 364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6294787097675712e-05,
+      "loss": 0.4052,
+      "step": 366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6256718642445202e-05,
+      "loss": 0.4214,
+      "step": 368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6218500598358376e-05,
+      "loss": 0.4283,
+      "step": 370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6180133879151943e-05,
+      "loss": 0.4188,
+      "step": 372
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6141619402117213e-05,
+      "loss": 0.3989,
+      "step": 374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6102958088078172e-05,
+      "loss": 0.4126,
+      "step": 376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.606415086136945e-05,
+      "loss": 0.4148,
+      "step": 378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6025198649814243e-05,
+      "loss": 0.42,
+      "step": 380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5986102384702112e-05,
+      "loss": 0.4398,
+      "step": 382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.594686300076673e-05,
+      "loss": 0.3987,
+      "step": 384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.590748143616353e-05,
+      "loss": 0.4313,
+      "step": 386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5867958632447263e-05,
+      "loss": 0.4214,
+      "step": 388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.582829553454951e-05,
+      "loss": 0.4066,
+      "step": 390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5788493090756074e-05,
+      "loss": 0.4064,
+      "step": 392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5748552252684303e-05,
+      "loss": 0.4109,
+      "step": 394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5708473975260356e-05,
+      "loss": 0.4282,
+      "step": 396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5668259216696366e-05,
+      "loss": 0.4358,
+      "step": 398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5627908938467516e-05,
+      "loss": 0.4303,
+      "step": 400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.558742410528907e-05,
+      "loss": 0.4082,
+      "step": 402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5546805685093308e-05,
+      "loss": 0.4041,
+      "step": 404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.550605464900636e-05,
+      "loss": 0.4148,
+      "step": 406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.546517197132502e-05,
+      "loss": 0.386,
+      "step": 408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.542415862949343e-05,
+      "loss": 0.4227,
+      "step": 410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5383015604079723e-05,
+      "loss": 0.4174,
+      "step": 412
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5341743878752563e-05,
+      "loss": 0.4302,
+      "step": 414
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5300344440257657e-05,
+      "loss": 0.4076,
+      "step": 416
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5258818278394125e-05,
+      "loss": 0.4047,
+      "step": 418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5217166385990865e-05,
+      "loss": 0.4242,
+      "step": 420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5175389758882803e-05,
+      "loss": 0.4032,
+      "step": 422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5133489395887089e-05,
+      "loss": 0.4268,
+      "step": 424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.509146629877921e-05,
+      "loss": 0.4132,
+      "step": 426
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5049321472269043e-05,
+      "loss": 0.4031,
+      "step": 428
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5007055923976843e-05,
+      "loss": 0.3714,
+      "step": 430
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4964670664409136e-05,
+      "loss": 0.235,
+      "step": 432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4922166706934566e-05,
+      "loss": 0.2015,
+      "step": 434
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4879545067759673e-05,
+      "loss": 0.2057,
+      "step": 436
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4836806765904587e-05,
+      "loss": 0.1876,
+      "step": 438
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4793952823178676e-05,
+      "loss": 0.1879,
+      "step": 440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4750984264156103e-05,
+      "loss": 0.1897,
+      "step": 442
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4707902116151338e-05,
+      "loss": 0.2166,
+      "step": 444
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4664707409194598e-05,
+      "loss": 0.1852,
+      "step": 446
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.462140117600721e-05,
+      "loss": 0.1909,
+      "step": 448
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.457798445197694e-05,
+      "loss": 0.1845,
+      "step": 450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4534458275133214e-05,
+      "loss": 0.1772,
+      "step": 452
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.449082368612232e-05,
+      "loss": 0.1873,
+      "step": 454
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4447081728182518e-05,
+      "loss": 0.1983,
+      "step": 456
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4403233447119096e-05,
+      "loss": 0.192,
+      "step": 458
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4359279891279376e-05,
+      "loss": 0.1808,
+      "step": 460
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.431522211152764e-05,
+      "loss": 0.1893,
+      "step": 462
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4271061161220007e-05,
+      "loss": 0.186,
+      "step": 464
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4226798096179262e-05,
+      "loss": 0.1854,
+      "step": 466
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4182433974669584e-05,
+      "loss": 0.1736,
+      "step": 468
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4137969857371277e-05,
+      "loss": 0.1876,
+      "step": 470
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4093406807355389e-05,
+      "loss": 0.1904,
+      "step": 472
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4048745890058304e-05,
+      "loss": 0.1829,
+      "step": 474
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4003988173256267e-05,
+      "loss": 0.1835,
+      "step": 476
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3959134727039854e-05,
+      "loss": 0.1829,
+      "step": 478
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3914186623788398e-05,
+      "loss": 0.1907,
+      "step": 480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3869144938144325e-05,
+      "loss": 0.1842,
+      "step": 482
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3824010746987495e-05,
+      "loss": 0.1929,
+      "step": 484
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3778785129409424e-05,
+      "loss": 0.1824,
+      "step": 486
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3733469166687505e-05,
+      "loss": 0.1867,
+      "step": 488
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3688063942259141e-05,
+      "loss": 0.1842,
+      "step": 490
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3642570541695867e-05,
+      "loss": 0.1874,
+      "step": 492
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.359699005267736e-05,
+      "loss": 0.1985,
+      "step": 494
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3551323564965465e-05,
+      "loss": 0.1671,
+      "step": 496
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3505572170378118e-05,
+      "loss": 0.1861,
+      "step": 498
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3459736962763263e-05,
+      "loss": 0.1873,
+      "step": 500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3413819037972682e-05,
+      "loss": 0.1946,
+      "step": 502
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.33678194938358e-05,
+      "loss": 0.1769,
+      "step": 504
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.332173943013345e-05,
+      "loss": 0.205,
+      "step": 506
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.327557994857156e-05,
+      "loss": 0.1899,
+      "step": 508
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.322934215275482e-05,
+      "loss": 0.1746,
+      "step": 510
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3183027148160304e-05,
+      "loss": 0.1843,
+      "step": 512
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3136636042111025e-05,
+      "loss": 0.165,
+      "step": 514
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.189,
+      "step": 516
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3043629964011104e-05,
+      "loss": 0.1745,
+      "step": 518
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2997017215597743e-05,
+      "loss": 0.1829,
+      "step": 520
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.295033281295103e-05,
+      "loss": 0.1871,
+      "step": 522
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2903577872225737e-05,
+      "loss": 0.1786,
+      "step": 524
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2856753511263105e-05,
+      "loss": 0.1759,
+      "step": 526
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2809860849564103e-05,
+      "loss": 0.2027,
+      "step": 528
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2762901008262678e-05,
+      "loss": 0.1824,
+      "step": 530
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.271587511009893e-05,
+      "loss": 0.1805,
+      "step": 532
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2668784279392287e-05,
+      "loss": 0.1777,
+      "step": 534
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2621629642014623e-05,
+      "loss": 0.1873,
+      "step": 536
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2574412325363326e-05,
+      "loss": 0.184,
+      "step": 538
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2527133458334353e-05,
+      "loss": 0.1932,
+      "step": 540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2479794171295248e-05,
+      "loss": 0.1875,
+      "step": 542
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2432395596058097e-05,
+      "loss": 0.1853,
+      "step": 544
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2384938865852482e-05,
+      "loss": 0.1815,
+      "step": 546
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2337425115298389e-05,
+      "loss": 0.1845,
+      "step": 548
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2289855480379074e-05,
+      "loss": 0.1953,
+      "step": 550
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.22422310984139e-05,
+      "loss": 0.1886,
+      "step": 552
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2194553108031153e-05,
+      "loss": 0.1875,
+      "step": 554
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.214682264914082e-05,
+      "loss": 0.1829,
+      "step": 556
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2099040862907332e-05,
+      "loss": 0.1935,
+      "step": 558
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2051208891722274e-05,
+      "loss": 0.1851,
+      "step": 560
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2003327879177085e-05,
+      "loss": 0.1991,
+      "step": 562
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.195539897003571e-05,
+      "loss": 0.1927,
+      "step": 564
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.190742331020723e-05,
+      "loss": 0.183,
+      "step": 566
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.185940204671846e-05,
+      "loss": 0.201,
+      "step": 568
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1811336327686537e-05,
+      "loss": 0.198,
+      "step": 570
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1763227302291464e-05,
+      "loss": 0.185,
+      "step": 572
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1715076120748631e-05,
+      "loss": 0.179,
+      "step": 574
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1666883934281324e-05,
+      "loss": 0.1934,
+      "step": 576
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1618651895093192e-05,
+      "loss": 0.1996,
+      "step": 578
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1570381156340701e-05,
+      "loss": 0.1813,
+      "step": 580
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1522072872105576e-05,
+      "loss": 0.1874,
+      "step": 582
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.147372819736719e-05,
+      "loss": 0.1773,
+      "step": 584
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1425348287974956e-05,
+      "loss": 0.1912,
+      "step": 586
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1376934300620706e-05,
+      "loss": 0.1949,
+      "step": 588
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1328487392811019e-05,
+      "loss": 0.1883,
+      "step": 590
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1280008722839552e-05,
+      "loss": 0.1766,
+      "step": 592
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1231499449759355e-05,
+      "loss": 0.1987,
+      "step": 594
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1182960733355142e-05,
+      "loss": 0.1785,
+      "step": 596
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1134393734115587e-05,
+      "loss": 0.1961,
+      "step": 598
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1085799613205552e-05,
+      "loss": 0.1805,
+      "step": 600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1037179532438345e-05,
+      "loss": 0.1745,
+      "step": 602
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.098853465424793e-05,
+      "loss": 0.1904,
+      "step": 604
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0939866141661148e-05,
+      "loss": 0.1858,
+      "step": 606
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.08911751582699e-05,
+      "loss": 0.2095,
+      "step": 608
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0842462868203329e-05,
+      "loss": 0.1935,
+      "step": 610
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.079373043609999e-05,
+      "loss": 0.1807,
+      "step": 612
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0744979027080003e-05,
+      "loss": 0.194,
+      "step": 614
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.06962098067172e-05,
+      "loss": 0.196,
+      "step": 616
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0647423941011255e-05,
+      "loss": 0.1916,
+      "step": 618
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0598622596359808e-05,
+      "loss": 0.1904,
+      "step": 620
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.054980693953058e-05,
+      "loss": 0.1766,
+      "step": 622
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0500978137633469e-05,
+      "loss": 0.1946,
+      "step": 624
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0452137358092654e-05,
+      "loss": 0.1918,
+      "step": 626
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0403285768618682e-05,
+      "loss": 0.1813,
+      "step": 628
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0354424537180554e-05,
+      "loss": 0.1879,
+      "step": 630
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0305554831977788e-05,
+      "loss": 0.1857,
+      "step": 632
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0256677821412508e-05,
+      "loss": 0.1949,
+      "step": 634
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0207794674061483e-05,
+      "loss": 0.209,
+      "step": 636
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.015890655864822e-05,
+      "loss": 0.2652,
+      "step": 638
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0110014644014994e-05,
+      "loss": 0.263,
+      "step": 640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0061120099094917e-05,
+      "loss": 0.2231,
+      "step": 642
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0012224092883986e-05,
+      "loss": 0.2141,
+      "step": 644
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.963327794413137e-06,
+      "loss": 0.2057,
+      "step": 646
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.914432372720294e-06,
+      "loss": 0.2352,
+      "step": 648
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.865538996822418e-06,
+      "loss": 0.2138,
+      "step": 650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.816648835687557e-06,
+      "loss": 0.2054,
+      "step": 652
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.767763058206897e-06,
+      "loss": 0.2073,
+      "step": 654
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.718882833166823e-06,
+      "loss": 0.2001,
+      "step": 656
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.670009329220963e-06,
+      "loss": 0.1985,
+      "step": 658
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.62114371486226e-06,
+      "loss": 0.2006,
+      "step": 660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.572287158395025e-06,
+      "loss": 0.2005,
+      "step": 662
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.523440827907006e-06,
+      "loss": 0.1974,
+      "step": 664
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.474605891241465e-06,
+      "loss": 0.207,
+      "step": 666
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.425783515969258e-06,
+      "loss": 0.1863,
+      "step": 668
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.376974869360918e-06,
+      "loss": 0.2004,
+      "step": 670
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.328181118358734e-06,
+      "loss": 0.1884,
+      "step": 672
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.279403429548877e-06,
+      "loss": 0.1884,
+      "step": 674
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.230642969133483e-06,
+      "loss": 0.1939,
+      "step": 676
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.181900902902794e-06,
+      "loss": 0.2122,
+      "step": 678
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.13317839620727e-06,
+      "loss": 0.197,
+      "step": 680
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.084476613929726e-06,
+      "loss": 0.1765,
+      "step": 682
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.035796720457495e-06,
+      "loss": 0.1879,
+      "step": 684
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.987139879654575e-06,
+      "loss": 0.189,
+      "step": 686
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.938507254833811e-06,
+      "loss": 0.1925,
+      "step": 688
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.889900008729084e-06,
+      "loss": 0.197,
+      "step": 690
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.841319303467502e-06,
+      "loss": 0.1954,
+      "step": 692
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.792766300541622e-06,
+      "loss": 0.1815,
+      "step": 694
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.744242160781682e-06,
+      "loss": 0.1914,
+      "step": 696
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.69574804432784e-06,
+      "loss": 0.186,
+      "step": 698
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.647285110602443e-06,
+      "loss": 0.1937,
+      "step": 700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.59885451828231e-06,
+      "loss": 0.198,
+      "step": 702
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.550457425271022e-06,
+      "loss": 0.1819,
+      "step": 704
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.502094988671232e-06,
+      "loss": 0.2001,
+      "step": 706
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.453768364757027e-06,
+      "loss": 0.1704,
+      "step": 708
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.405478708946254e-06,
+      "loss": 0.1873,
+      "step": 710
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.35722717577291e-06,
+      "loss": 0.1771,
+      "step": 712
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.309014918859538e-06,
+      "loss": 0.1843,
+      "step": 714
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.26084309088964e-06,
+      "loss": 0.1808,
+      "step": 716
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.212712843580124e-06,
+      "loss": 0.2045,
+      "step": 718
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.164625327653772e-06,
+      "loss": 0.1799,
+      "step": 720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.116581692811711e-06,
+      "loss": 0.1838,
+      "step": 722
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.068583087705946e-06,
+      "loss": 0.1923,
+      "step": 724
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.020630659911881e-06,
+      "loss": 0.1827,
+      "step": 726
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.972725555900895e-06,
+      "loss": 0.1819,
+      "step": 728
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.924868921012918e-06,
+      "loss": 0.1824,
+      "step": 730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.877061899429067e-06,
+      "loss": 0.1973,
+      "step": 732
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.829305634144264e-06,
+      "loss": 0.183,
+      "step": 734
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.781601266939936e-06,
+      "loss": 0.1652,
+      "step": 736
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.733949938356695e-06,
+      "loss": 0.1895,
+      "step": 738
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.686352787667083e-06,
+      "loss": 0.1845,
+      "step": 740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.638810952848328e-06,
+      "loss": 0.1894,
+      "step": 742
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.591325570555136e-06,
+      "loss": 0.1707,
+      "step": 744
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.543897776092519e-06,
+      "loss": 0.1776,
+      "step": 746
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.496528703388648e-06,
+      "loss": 0.1788,
+      "step": 748
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.449219484967749e-06,
+      "loss": 0.1777,
+      "step": 750
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.401971251923015e-06,
+      "loss": 0.183,
+      "step": 752
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.354785133889566e-06,
+      "loss": 0.1857,
+      "step": 754
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.307662259017454e-06,
+      "loss": 0.1892,
+      "step": 756
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.260603753944674e-06,
+      "loss": 0.1785,
+      "step": 758
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.213610743770234e-06,
+      "loss": 0.1884,
+      "step": 760
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.166684352027265e-06,
+      "loss": 0.1773,
+      "step": 762
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.119825700656138e-06,
+      "loss": 0.1862,
+      "step": 764
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.073035909977661e-06,
+      "loss": 0.1872,
+      "step": 766
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.026316098666282e-06,
+      "loss": 0.1917,
+      "step": 768
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.979667383723345e-06,
+      "loss": 0.1823,
+      "step": 770
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.9330908804503874e-06,
+      "loss": 0.179,
+      "step": 772
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.886587702422474e-06,
+      "loss": 0.1731,
+      "step": 774
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.840158961461567e-06,
+      "loss": 0.1843,
+      "step": 776
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.793805767609953e-06,
+      "loss": 0.1789,
+      "step": 778
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.7475292291037e-06,
+      "loss": 0.1851,
+      "step": 780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.701330452346156e-06,
+      "loss": 0.1795,
+      "step": 782
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.655210541881502e-06,
+      "loss": 0.1907,
+      "step": 784
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.609170600368346e-06,
+      "loss": 0.1885,
+      "step": 786
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.56321172855336e-06,
+      "loss": 0.1804,
+      "step": 788
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.51733502524495e-06,
+      "loss": 0.184,
+      "step": 790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.471541587287003e-06,
+      "loss": 0.186,
+      "step": 792
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.425832509532652e-06,
+      "loss": 0.167,
+      "step": 794
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.380208884818104e-06,
+      "loss": 0.1728,
+      "step": 796
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.3346718039365076e-06,
+      "loss": 0.1765,
+      "step": 798
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.289222355611881e-06,
+      "loss": 0.1813,
+      "step": 800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.243861626473073e-06,
+      "loss": 0.1875,
+      "step": 802
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.198590701027796e-06,
+      "loss": 0.1829,
+      "step": 804
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.153410661636683e-06,
+      "loss": 0.1803,
+      "step": 806
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.108322588487419e-06,
+      "loss": 0.1768,
+      "step": 808
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.063327559568908e-06,
+      "loss": 0.1764,
+      "step": 810
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.0184266506455125e-06,
+      "loss": 0.1818,
+      "step": 812
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.973620935231318e-06,
+      "loss": 0.1834,
+      "step": 814
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.928911484564481e-06,
+      "loss": 0.1682,
+      "step": 816
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.884299367581607e-06,
+      "loss": 0.1828,
+      "step": 818
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.8397856508922e-06,
+      "loss": 0.1802,
+      "step": 820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.795371398753153e-06,
+      "loss": 0.1949,
+      "step": 822
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.751057673043316e-06,
+      "loss": 0.1777,
+      "step": 824
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.706845533238097e-06,
+      "loss": 0.1728,
+      "step": 826
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.662736036384142e-06,
+      "loss": 0.1701,
+      "step": 828
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.618730237074048e-06,
+      "loss": 0.1667,
+      "step": 830
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.574829187421166e-06,
+      "loss": 0.1746,
+      "step": 832
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.531033937034429e-06,
+      "loss": 0.1827,
+      "step": 834
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.4873455329932736e-06,
+      "loss": 0.1769,
+      "step": 836
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.443765019822593e-06,
+      "loss": 0.1854,
+      "step": 838
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.400293439467781e-06,
+      "loss": 0.1921,
+      "step": 840
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.356931831269798e-06,
+      "loss": 0.1815,
+      "step": 842
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.313681231940338e-06,
+      "loss": 0.1781,
+      "step": 844
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.270542675537034e-06,
+      "loss": 0.2022,
+      "step": 846
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.227517193438746e-06,
+      "loss": 0.1866,
+      "step": 848
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.184605814320889e-06,
+      "loss": 0.1754,
+      "step": 850
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.141809564130847e-06,
+      "loss": 0.1745,
+      "step": 852
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.099129466063444e-06,
+      "loss": 0.1803,
+      "step": 854
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.056566540536476e-06,
+      "loss": 0.1678,
+      "step": 856
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.014121805166321e-06,
+      "loss": 0.1702,
+      "step": 858
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.971796274743601e-06,
+      "loss": 0.1313,
+      "step": 860
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.9295909612089265e-06,
+      "loss": 0.0643,
+      "step": 862
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.887506873628708e-06,
+      "loss": 0.0624,
+      "step": 864
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.845545018171013e-06,
+      "loss": 0.0604,
+      "step": 866
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.80370639808152e-06,
+      "loss": 0.0648,
+      "step": 868
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.7619920136595465e-06,
+      "loss": 0.0731,
+      "step": 870
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.720402862234105e-06,
+      "loss": 0.0582,
+      "step": 872
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.678939938140079e-06,
+      "loss": 0.0601,
+      "step": 874
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.637604232694441e-06,
+      "loss": 0.0527,
+      "step": 876
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.596396734172559e-06,
+      "loss": 0.0575,
+      "step": 878
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.555318427784561e-06,
+      "loss": 0.0578,
+      "step": 880
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.514370295651781e-06,
+      "loss": 0.0543,
+      "step": 882
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.473553316783282e-06,
+      "loss": 0.0547,
+      "step": 884
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.432868467052449e-06,
+      "loss": 0.053,
+      "step": 886
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.392316719173651e-06,
+      "loss": 0.0587,
+      "step": 888
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.351899042678993e-06,
+      "loss": 0.0628,
+      "step": 890
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.311616403895126e-06,
+      "loss": 0.0582,
+      "step": 892
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.271469765920163e-06,
+      "loss": 0.0578,
+      "step": 894
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.231460088600626e-06,
+      "loss": 0.064,
+      "step": 896
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.191588328508518e-06,
+      "loss": 0.0525,
+      "step": 898
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1518554389184416e-06,
+      "loss": 0.0584,
+      "step": 900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1122623697848164e-06,
+      "loss": 0.0621,
+      "step": 902
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.0728100677191585e-06,
+      "loss": 0.0563,
+      "step": 904
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.033499475967451e-06,
+      "loss": 0.0598,
+      "step": 906
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.994331534387602e-06,
+      "loss": 0.0528,
+      "step": 908
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.95530717942696e-06,
+      "loss": 0.0588,
+      "step": 910
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.916427344099928e-06,
+      "loss": 0.0668,
+      "step": 912
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.877692957965663e-06,
+      "loss": 0.0569,
+      "step": 914
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.839104947105847e-06,
+      "loss": 0.0588,
+      "step": 916
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.8006642341025456e-06,
+      "loss": 0.0594,
+      "step": 918
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.762371738016153e-06,
+      "loss": 0.059,
+      "step": 920
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.72422837436341e-06,
+      "loss": 0.0523,
+      "step": 922
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.686235055095536e-06,
+      "loss": 0.0538,
+      "step": 924
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.648392688576401e-06,
+      "loss": 0.0586,
+      "step": 926
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.610702179560821e-06,
+      "loss": 0.055,
+      "step": 928
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.573164429172924e-06,
+      "loss": 0.0524,
+      "step": 930
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.5357803348846087e-06,
+      "loss": 0.0618,
+      "step": 932
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.498550790494083e-06,
+      "loss": 0.0527,
+      "step": 934
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.461476686104495e-06,
+      "loss": 0.0541,
+      "step": 936
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.424558908102653e-06,
+      "loss": 0.0579,
+      "step": 938
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.387798339137837e-06,
+      "loss": 0.0567,
+      "step": 940
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.3511958581006874e-06,
+      "loss": 0.0519,
+      "step": 942
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.314752340102201e-06,
+      "loss": 0.0573,
+      "step": 944
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.278468656452798e-06,
+      "loss": 0.061,
+      "step": 946
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.242345674641508e-06,
+      "loss": 0.0611,
+      "step": 948
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.2063842583152095e-06,
+      "loss": 0.0604,
+      "step": 950
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.1705852672579853e-06,
+      "loss": 0.0556,
+      "step": 952
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.134949557370587e-06,
+      "loss": 0.0557,
+      "step": 954
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.099477980649941e-06,
+      "loss": 0.0539,
+      "step": 956
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0641713851687994e-06,
+      "loss": 0.061,
+      "step": 958
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0290306150554573e-06,
+      "loss": 0.0566,
+      "step": 960
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.994056510473571e-06,
+      "loss": 0.0631,
+      "step": 962
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.959249907602071e-06,
+      "loss": 0.052,
+      "step": 964
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.9246116386151704e-06,
+      "loss": 0.0553,
+      "step": 966
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.890142531662471e-06,
+      "loss": 0.0578,
+      "step": 968
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.8558434108491585e-06,
+      "loss": 0.0522,
+      "step": 970
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.8217150962163044e-06,
+      "loss": 0.0575,
+      "step": 972
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7877584037212555e-06,
+      "loss": 0.0615,
+      "step": 974
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.75397414521813e-06,
+      "loss": 0.0512,
+      "step": 976
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.720363128438408e-06,
+      "loss": 0.0595,
+      "step": 978
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6869261569716134e-06,
+      "loss": 0.0557,
+      "step": 980
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6536640302461036e-06,
+      "loss": 0.0605,
+      "step": 982
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6205775435099624e-06,
+      "loss": 0.0548,
+      "step": 984
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5876674878119735e-06,
+      "loss": 0.0544,
+      "step": 986
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.554934649982731e-06,
+      "loss": 0.0566,
+      "step": 988
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5223798126158004e-06,
+      "loss": 0.055,
+      "step": 990
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.490003754049024e-06,
+      "loss": 0.0545,
+      "step": 992
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.457807248345908e-06,
+      "loss": 0.0611,
+      "step": 994
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.425791065277119e-06,
+      "loss": 0.0558,
+      "step": 996
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.393955970302072e-06,
+      "loss": 0.0699,
+      "step": 998
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.362302724550639e-06,
+      "loss": 0.0589,
+      "step": 1000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3308320848049436e-06,
+      "loss": 0.0584,
+      "step": 1002
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.299544803481274e-06,
+      "loss": 0.0524,
+      "step": 1004
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.2684416286120846e-06,
+      "loss": 0.0595,
+      "step": 1006
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.23752330382813e-06,
+      "loss": 0.0551,
+      "step": 1008
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.20679056834066e-06,
+      "loss": 0.0523,
+      "step": 1010
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.176244156923768e-06,
+      "loss": 0.0913,
+      "step": 1012
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1458847998968123e-06,
+      "loss": 0.0561,
+      "step": 1014
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.115713223106959e-06,
+      "loss": 0.0562,
+      "step": 1016
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0857301479118276e-06,
+      "loss": 0.0574,
+      "step": 1018
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0559362911622438e-06,
+      "loss": 0.0641,
+      "step": 1020
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.026332365185102e-06,
+      "loss": 0.0569,
+      "step": 1022
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.996919077766334e-06,
+      "loss": 0.0656,
+      "step": 1024
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.967697132133981e-06,
+      "loss": 0.0508,
+      "step": 1026
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9386672269413976e-06,
+      "loss": 0.0533,
+      "step": 1028
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.0555,
+      "step": 1030
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8811863095153182e-06,
+      "loss": 0.0522,
+      "step": 1032
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.852736671565244e-06,
+      "loss": 0.0541,
+      "step": 1034
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8244818225889183e-06,
+      "loss": 0.053,
+      "step": 1036
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7964224381178474e-06,
+      "loss": 0.0514,
+      "step": 1038
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.768559189010267e-06,
+      "loss": 0.0513,
+      "step": 1040
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7408927414351051e-06,
+      "loss": 0.0546,
+      "step": 1042
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7134237568560619e-06,
+      "loss": 0.0515,
+      "step": 1044
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6861528920157877e-06,
+      "loss": 0.0559,
+      "step": 1046
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6590807989201841e-06,
+      "loss": 0.0594,
+      "step": 1048
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.632208124822815e-06,
+      "loss": 0.0527,
+      "step": 1050
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6055355122094352e-06,
+      "loss": 0.0503,
+      "step": 1052
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.579063598782622e-06,
+      "loss": 0.0534,
+      "step": 1054
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5527930174465356e-06,
+      "loss": 0.0639,
+      "step": 1056
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5267243962917833e-06,
+      "loss": 0.0575,
+      "step": 1058
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5008583585804048e-06,
+      "loss": 0.052,
+      "step": 1060
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.4751955227309722e-06,
+      "loss": 0.0532,
+      "step": 1062
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4497365023038012e-06,
+      "loss": 0.0542,
+      "step": 1064
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4244819059862824e-06,
+      "loss": 0.0525,
+      "step": 1066
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.399432337578327e-06,
+      "loss": 0.0588,
+      "step": 1068
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3745883959779415e-06,
+      "loss": 0.0552,
+      "step": 1070
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3499506751668933e-06,
+      "loss": 0.0535,
+      "step": 1072
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.325519764196519e-06,
+      "loss": 0.0496,
+      "step": 1074
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.301296247173638e-06,
+      "loss": 0.0536,
+      "step": 1076
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.2772807032465895e-06,
+      "loss": 0.0546,
+      "step": 1078
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.2534737065913839e-06,
+      "loss": 0.062,
+      "step": 1080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.229875826397976e-06,
+      "loss": 0.0482,
+      "step": 1082
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2064876268566572e-06,
+      "loss": 0.0526,
+      "step": 1084
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.1833096671445644e-06,
+      "loss": 0.0513,
+      "step": 1086
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.1603425014123126e-06,
+      "loss": 0.06,
+      "step": 1088
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1375866787707435e-06,
+      "loss": 0.0553,
+      "step": 1090
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1150427432778078e-06,
+      "loss": 0.0504,
+      "step": 1092
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0927112339255374e-06,
+      "loss": 0.0512,
+      "step": 1094
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0705926846271787e-06,
+      "loss": 0.05,
+      "step": 1096
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0486876242044153e-06,
+      "loss": 0.0577,
+      "step": 1098
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0269965763747292e-06,
+      "loss": 0.0533,
+      "step": 1100
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0055200597388793e-06,
+      "loss": 0.0556,
+      "step": 1102
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.84258587768504e-07,
+      "loss": 0.0501,
+      "step": 1104
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.632126687938392e-07,
+      "loss": 0.0472,
+      "step": 1106
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.423828059915685e-07,
+      "loss": 0.0637,
+      "step": 1108
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.217694973728009e-07,
+      "loss": 0.0508,
+      "step": 1110
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.013732357711469e-07,
+      "loss": 0.0614,
+      "step": 1112
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.811945088309493e-07,
+      "loss": 0.0534,
+      "step": 1114
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.612337989956199e-07,
+      "loss": 0.0569,
+      "step": 1116
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.414915834961035e-07,
+      "loss": 0.053,
+      "step": 1118
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.219683343394691e-07,
+      "loss": 0.0554,
+      "step": 1120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.0266451829763e-07,
+      "loss": 0.0538,
+      "step": 1122
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.835805968961762e-07,
+      "loss": 0.0529,
+      "step": 1124
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.647170264033422e-07,
+      "loss": 0.0535,
+      "step": 1126
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.460742578191016e-07,
+      "loss": 0.0457,
+      "step": 1128
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.276527368643793e-07,
+      "loss": 0.0531,
+      "step": 1130
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.094529039704013e-07,
+      "loss": 0.052,
+      "step": 1132
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.914751942681585e-07,
+      "loss": 0.0527,
+      "step": 1134
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.737200375780073e-07,
+      "loss": 0.0555,
+      "step": 1136
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.561878583993897e-07,
+      "loss": 0.0502,
+      "step": 1138
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.388790759006902e-07,
+      "loss": 0.0502,
+      "step": 1140
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.217941039092068e-07,
+      "loss": 0.0602,
+      "step": 1142
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.049333509012611e-07,
+      "loss": 0.0564,
+      "step": 1144
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.882972199924353e-07,
+      "loss": 0.0578,
+      "step": 1146
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.718861089279249e-07,
+      "loss": 0.0553,
+      "step": 1148
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.557004100730357e-07,
+      "loss": 0.0531,
+      "step": 1150
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.39740510403809e-07,
+      "loss": 0.0499,
+      "step": 1152
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.240067914977554e-07,
+      "loss": 0.054,
+      "step": 1154
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.084996295247402e-07,
+      "loss": 0.0512,
+      "step": 1156
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.932193952379915e-07,
+      "loss": 0.0537,
+      "step": 1158
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.781664539652319e-07,
+      "loss": 0.0541,
+      "step": 1160
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.633411655999431e-07,
+      "loss": 0.0515,
+      "step": 1162
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.487438845927683e-07,
+      "loss": 0.054,
+      "step": 1164
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.34374959943028e-07,
+      "loss": 0.0516,
+      "step": 1166
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.202347351903857e-07,
+      "loss": 0.0551,
+      "step": 1168
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.063235484066275e-07,
+      "loss": 0.0498,
+      "step": 1170
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.9264173218758083e-07,
+      "loss": 0.0523,
+      "step": 1172
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.791896136451656e-07,
+      "loss": 0.0535,
+      "step": 1174
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6596751439957003e-07,
+      "loss": 0.0515,
+      "step": 1176
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.5297575057156255e-07,
+      "loss": 0.0642,
+      "step": 1178
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4021463277493337e-07,
+      "loss": 0.0576,
+      "step": 1180
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.2768446610906834e-07,
+      "loss": 0.0512,
+      "step": 1182
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.153855501516545e-07,
+      "loss": 0.0497,
+      "step": 1184
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.0331817895151827e-07,
+      "loss": 0.0493,
+      "step": 1186
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9148264102159316e-07,
+      "loss": 0.0526,
+      "step": 1188
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.7987921933202655e-07,
+      "loss": 0.0616,
+      "step": 1190
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.685081913034082e-07,
+      "loss": 0.0558,
+      "step": 1192
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.573698288001403e-07,
+      "loss": 0.0537,
+      "step": 1194
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.46464398123939e-07,
+      "loss": 0.0549,
+      "step": 1196
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3579216000746418e-07,
+      "loss": 0.0523,
+      "step": 1198
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2535336960809118e-07,
+      "loss": 0.0558,
+      "step": 1200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1514827650180425e-07,
+      "loss": 0.0488,
+      "step": 1202
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.051771246772305e-07,
+      "loss": 0.052,
+      "step": 1204
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.954401525298144e-07,
+      "loss": 0.0518,
+      "step": 1206
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.859375928561058e-07,
+      "loss": 0.0477,
+      "step": 1208
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7666967284820202e-07,
+      "loss": 0.055,
+      "step": 1210
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6763661408831677e-07,
+      "loss": 0.0534,
+      "step": 1212
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5883863254347653e-07,
+      "loss": 0.0522,
+      "step": 1214
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5027593856036137e-07,
+      "loss": 0.0518,
+      "step": 1216
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4194873686027566e-07,
+      "loss": 0.0484,
+      "step": 1218
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3385722653425304e-07,
+      "loss": 0.0571,
+      "step": 1220
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2600160103829584e-07,
+      "loss": 0.048,
+      "step": 1222
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1838204818874877e-07,
+      "loss": 0.0521,
+      "step": 1224
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1099875015781359e-07,
+      "loss": 0.0546,
+      "step": 1226
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0385188346918485e-07,
+      "loss": 0.0492,
+      "step": 1228
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.694161899383992e-08,
+      "loss": 0.0555,
+      "step": 1230
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.026812194594448e-08,
+      "loss": 0.0494,
+      "step": 1232
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.383155187890901e-08,
+      "loss": 0.0527,
+      "step": 1234
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.763206268156964e-08,
+      "loss": 0.0527,
+      "step": 1236
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.166980257451106e-08,
+      "loss": 0.0618,
+      "step": 1238
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.594491410652493e-08,
+      "loss": 0.0506,
+      "step": 1240
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.045753415119593e-08,
+      "loss": 0.0577,
+      "step": 1242
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.520779390363551e-08,
+      "loss": 0.0495,
+      "step": 1244
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.019581887733993e-08,
+      "loss": 0.0563,
+      "step": 1246
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.542172890119267e-08,
+      "loss": 0.0556,
+      "step": 1248
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.0885638116601176e-08,
+      "loss": 0.0518,
+      "step": 1250
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6587654974761246e-08,
+      "loss": 0.0517,
+      "step": 1252
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.252788223407244e-08,
+      "loss": 0.0531,
+      "step": 1254
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.870641695767451e-08,
+      "loss": 0.0525,
+      "step": 1256
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5123350511129242e-08,
+      "loss": 0.0609,
+      "step": 1258
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.177876856023997e-08,
+      "loss": 0.0578,
+      "step": 1260
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8672751068995464e-08,
+      "loss": 0.0544,
+      "step": 1262
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5805372297662546e-08,
+      "loss": 0.0557,
+      "step": 1264
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3176700801014186e-08,
+      "loss": 0.0509,
+      "step": 1266
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0786799426683037e-08,
+      "loss": 0.0564,
+      "step": 1268
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.635725313663745e-09,
+      "loss": 0.055,
+      "step": 1270
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.723529890946268e-09,
+      "loss": 0.0494,
+      "step": 1272
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.05025887628352e-09,
+      "loss": 0.0501,
+      "step": 1274
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.615952275104473e-09,
+      "loss": 0.0505,
+      "step": 1276
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.420644379549364e-09,
+      "loss": 0.0515,
+      "step": 1278
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4643637676559074e-09,
+      "loss": 0.2096,
+      "step": 1280
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.471333026742856e-10,
+      "loss": 0.0539,
+      "step": 1282
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.689701325209182e-10,
+      "loss": 0.0536,
+      "step": 1284
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.988568936768132e-11,
+      "loss": 0.0512,
+      "step": 1286
+    }
+  ],
+  "max_steps": 1287,
+  "num_train_epochs": 3,
+  "total_flos": 999035450556416.0,
+  "trial_name": null,
+  "trial_params": null
+}

zero_to_fp32.py

@@ -0,0 +1,584 @@
+#!/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 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())
+
+        # record shared parameters so that they can be recovered based on partners
+        # this is because such parameters holding reference only are not saved by optimizer
+        shared_params = []
+        for param in state_dict["module"]:
+            if param not in [*param_names, *buffer_names]:
+                for share_param in state_dict["module"]:
+                    if (state_dict["module"][share_param].data_ptr() == state_dict["module"][param].data_ptr()
+                            and share_param != param):
+                        shared_params.append([param, share_param])
+                        break
+
+        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_dicts.append(torch.load(f, map_location=device))
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage == 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    if zero_stage == 2:
+        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    elif zero_stage == 3:
+        # if there is more than one param group, there will be multiple flattened tensors - one
+        # flattened tensor per group - for simplicity merge them into a single tensor
+        #
+        # XXX: could make the script more memory efficient for when there are multiple groups - it
+        # will require matching the sub-lists of param_shapes for each param group flattened tensor
+
+        fp32_flat_groups = [
+            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
+        ]
+
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage == 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel()
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        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:
+        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("-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)