diff --git a/README.md b/README.md
index beab0ecbb10af31838126e315e9776fdbe5b7595..a2b7fac6d4fef8f82e38d76642f5bec8b61594d1 100644
--- a/README.md
+++ b/README.md
@@ -1,3 +1,27 @@
 ---
 license: llama2
+tags:
+- vision-language model
+- llama
+- video understanding
 ---
+
+# Flash-VStream Model Card
+<a href='https://invinciblewyq.github.io/vstream-page/'><img src='https://img.shields.io/badge/Project-Page-Green'></a> 
+<a href='https://arxiv.org/abs/2406.08085v1'><img src='https://img.shields.io/badge/Paper-Arxiv-red'></a>
+
+## Model details
+We proposed Flash-VStream, a video-language model that simulates the memory mechanism of human. Our model is able to process extremely long video streams in real-time and respond to user queries simultaneously.
+
+## License
+Llama 2 is licensed under the LLAMA 2 Community License, 
+Copyright (c) Meta Platforms, Inc. All Rights Reserved.
+
+## Training data
+This model is trained based on image data from LLaVA-1.5 dataset, and video data from WebVid and ActivityNet datasets following LLaMA-VID, including
+- 558K filtered image-text pairs from LAION/CC/SBU, captioned by BLIP.
+- 158K GPT-generated multimodal instruction-following data.
+- 450K academic-task-oriented VQA data mixture.
+- 40K ShareGPT data.
+- 232K video-caption pairs sampled from the WebVid 2.5M dataset.
+- 98K videos from ActivityNet with QA pairs from Video-ChatGPT.
diff --git a/config.json b/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..303934819295ca4b9a0cf01585b492490244f322
--- /dev/null
+++ b/config.json
@@ -0,0 +1,56 @@
+{
+  "_name_or_path": "./checkpoints-pretrain/vstream-vicuna-7b-pretrain-weighted_kmeans1*8-25*4-25*1/checkpoint-3000",
+  "architectures": [
+    "VStreamLlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "compress_Turing_memory_size": 1,
+  "compress_Turing_update_ratio": 0.2,
+  "compress_long_memory_size": 4,
+  "compress_size": 8,
+  "compress_type": "mean",
+  "eos_token_id": 2,
+  "freeze_mm_mlp_adapter": false,
+  "freeze_mm_vision_resampler": false,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "image_aspect_ratio": "pad",
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 4096,
+  "mm_hidden_size": 1024,
+  "mm_projector_lr": null,
+  "mm_projector_type": "mlp2x_gelu",
+  "mm_resampler_type": null,
+  "mm_use_4_vision_tokens": false,
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "mm_vision_select_feature": "patch",
+  "mm_vision_select_layer": -2,
+  "mm_vision_tower": "./ckpt/clip-vit-large-patch14",
+  "model_type": "vstream",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": null,
+  "tie_word_embeddings": false,
+  "tokenizer_model_max_length": 2048,
+  "tokenizer_padding_side": "right",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.31.0",
+  "tune_mm_mlp_adapter": false,
+  "tune_mm_vision_resampler": false,
+  "unfreeze_mm_vision_tower": false,
+  "use_cache": false,
+  "use_mm_proj": true,
+  "video_Turing_memory_length": 25,
+  "video_current_memory_length": 1,
+  "video_long_memory_length": 25,
+  "video_max_frames": 26,
+  "video_sample_type": "weighted_kmeans",
+  "video_short_memory_length": 10,
+  "vocab_size": 32000
+}
diff --git a/generation_config.json b/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..57dbab308581e71f2adb7e9335aad7bc02cf3bb4
--- /dev/null
+++ b/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "max_length": 4096,
+  "pad_token_id": 0,
+  "temperature": 0.9,
+  "top_p": 0.6,
+  "transformers_version": "4.31.0"
+}
diff --git a/pytorch_model-00001-of-00002.bin b/pytorch_model-00001-of-00002.bin
new file mode 100644
index 0000000000000000000000000000000000000000..fce6bea119b8656e661a4448840bf68e79fef6f3
--- /dev/null
+++ b/pytorch_model-00001-of-00002.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63302cc0909608c878e1d96dd914ad8850d5e0300e69895f2cbb2893ab966796
+size 9976631486
diff --git a/pytorch_model-00002-of-00002.bin b/pytorch_model-00002-of-00002.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d8199239a4aaf32af61b64595910c52b3c61ecfd
--- /dev/null
+++ b/pytorch_model-00002-of-00002.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f62296b51622a18041cd2f9f923927281894fe59b2cbbcbda533c7eba97a75aa
+size 4149054736
diff --git a/pytorch_model.bin.index.json b/pytorch_model.bin.index.json
new file mode 100644
index 0000000000000000000000000000000000000000..c97d119925c8c42af9fc51ac04a4c348854b01b9
--- /dev/null
+++ b/pytorch_model.bin.index.json
@@ -0,0 +1,735 @@
+{
+  "metadata": {
+    "total_size": 14125422784
+  },
+  "weight_map": {
+    "lm_head.weight": "pytorch_model-00002-of-00002.bin",
+    "model.attention_model.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.attention_model.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.attention_model.out_ln.bias": "pytorch_model-00002-of-00002.bin",
+    "model.attention_model.out_ln.weight": "pytorch_model-00002-of-00002.bin",
+    "model.attention_model.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.attention_model.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.attention_model.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.attention_model.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.attention_model.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.attention_model.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.embed_tokens.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.23.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.23.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.23.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.24.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.3.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.30.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.4.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.mm_projector.0.bias": "pytorch_model-00002-of-00002.bin",
+    "model.mm_projector.0.weight": "pytorch_model-00002-of-00002.bin",
+    "model.mm_projector.2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.mm_projector.2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.norm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.embeddings.class_embedding": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.embeddings.patch_embedding.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.embeddings.position_embedding.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.0.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.1.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.10.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.11.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.12.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.13.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.14.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.15.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.16.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.17.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.18.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.19.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.2.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.20.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.21.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.22.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.23.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.3.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.4.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.5.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.6.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.7.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.8.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.layer_norm1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.layer_norm1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.layer_norm2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.layer_norm2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.mlp.fc1.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.mlp.fc1.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.mlp.fc2.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.mlp.fc2.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.encoder.layers.9.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.post_layernorm.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.post_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.pre_layrnorm.bias": "pytorch_model-00002-of-00002.bin",
+    "model.vision_tower.vision_tower.vision_model.pre_layrnorm.weight": "pytorch_model-00002-of-00002.bin"
+  }
+}
diff --git a/special_tokens_map.json b/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..14761dcf1466dc232bd41de9c21d4c617b15755e
--- /dev/null
+++ b/special_tokens_map.json
@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<unk>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/tokenizer.model b/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899
--- /dev/null
+++ b/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723
diff --git a/tokenizer_config.json b/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..740756b4bef305e27d0bb4d2e1a40dd8847797f7
--- /dev/null
+++ b/tokenizer_config.json
@@ -0,0 +1,35 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "bos_token": {
+    "__type": "AddedToken",
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "clean_up_tokenization_spaces": false,
+  "eos_token": {
+    "__type": "AddedToken",
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "legacy": 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": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/trainer_state.json b/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..7d30923ab9956955ee7cab870ec3da7b52edd1df
--- /dev/null
+++ b/trainer_state.json
@@ -0,0 +1,35416 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.989103101424979,
+  "global_step": 5900,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.11731843575419e-07,
+      "loss": 1.4195,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.23463687150838e-07,
+      "loss": 1.3823,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3519553072625703e-07,
+      "loss": 1.4459,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.46927374301676e-07,
+      "loss": 1.5393,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.58659217877095e-07,
+      "loss": 1.4448,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.703910614525141e-07,
+      "loss": 1.4223,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.82122905027933e-07,
+      "loss": 1.4144,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.93854748603352e-07,
+      "loss": 1.3139,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.005586592178771e-06,
+      "loss": 1.4296,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.11731843575419e-06,
+      "loss": 1.4222,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.229050279329609e-06,
+      "loss": 1.3935,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3407821229050281e-06,
+      "loss": 1.3394,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4525139664804472e-06,
+      "loss": 1.4121,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.564245810055866e-06,
+      "loss": 1.3469,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.675977653631285e-06,
+      "loss": 1.4301,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.787709497206704e-06,
+      "loss": 1.3138,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.899441340782123e-06,
+      "loss": 1.2787,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.011173184357542e-06,
+      "loss": 1.2886,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1229050279329612e-06,
+      "loss": 1.2743,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.23463687150838e-06,
+      "loss": 1.2688,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3463687150837993e-06,
+      "loss": 1.2021,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.458100558659218e-06,
+      "loss": 1.2191,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.569832402234637e-06,
+      "loss": 1.1803,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6815642458100562e-06,
+      "loss": 1.2004,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.793296089385475e-06,
+      "loss": 1.1655,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9050279329608943e-06,
+      "loss": 0.3161,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.016759776536313e-06,
+      "loss": 1.2154,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.128491620111732e-06,
+      "loss": 1.2058,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.240223463687151e-06,
+      "loss": 1.2751,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.35195530726257e-06,
+      "loss": 1.1332,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4636871508379893e-06,
+      "loss": 1.1565,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.575418994413408e-06,
+      "loss": 1.1614,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.687150837988827e-06,
+      "loss": 1.1344,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.798882681564246e-06,
+      "loss": 0.3087,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.910614525139665e-06,
+      "loss": 1.0746,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.022346368715084e-06,
+      "loss": 1.1482,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.134078212290504e-06,
+      "loss": 0.3071,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2458100558659224e-06,
+      "loss": 1.1249,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.357541899441341e-06,
+      "loss": 0.3115,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.46927374301676e-06,
+      "loss": 1.1591,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.581005586592179e-06,
+      "loss": 1.141,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.692737430167599e-06,
+      "loss": 1.1086,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8044692737430175e-06,
+      "loss": 1.085,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.916201117318436e-06,
+      "loss": 1.1062,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.027932960893855e-06,
+      "loss": 1.1468,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.139664804469274e-06,
+      "loss": 1.0687,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.251396648044693e-06,
+      "loss": 1.0904,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3631284916201125e-06,
+      "loss": 1.0541,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.474860335195531e-06,
+      "loss": 0.3017,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.58659217877095e-06,
+      "loss": 1.1023,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.698324022346369e-06,
+      "loss": 1.0348,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.810055865921789e-06,
+      "loss": 1.0711,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9217877094972075e-06,
+      "loss": 1.0878,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.033519553072626e-06,
+      "loss": 1.0898,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.145251396648045e-06,
+      "loss": 1.0799,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.256983240223464e-06,
+      "loss": 1.0543,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.368715083798883e-06,
+      "loss": 1.1395,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.480446927374302e-06,
+      "loss": 1.0945,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.592178770949721e-06,
+      "loss": 1.0429,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.70391061452514e-06,
+      "loss": 1.1208,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.815642458100559e-06,
+      "loss": 1.085,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.927374301675979e-06,
+      "loss": 1.0802,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0391061452513975e-06,
+      "loss": 1.0949,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.150837988826816e-06,
+      "loss": 1.1396,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.262569832402235e-06,
+      "loss": 1.1551,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.374301675977654e-06,
+      "loss": 1.0647,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.486033519553073e-06,
+      "loss": 1.0458,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.597765363128492e-06,
+      "loss": 1.0191,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.709497206703911e-06,
+      "loss": 0.2835,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.82122905027933e-06,
+      "loss": 1.0698,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.932960893854749e-06,
+      "loss": 1.083,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.044692737430168e-06,
+      "loss": 1.0165,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.156424581005588e-06,
+      "loss": 1.1093,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.268156424581007e-06,
+      "loss": 1.0901,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.379888268156426e-06,
+      "loss": 1.1368,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.491620111731845e-06,
+      "loss": 1.0725,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.603351955307264e-06,
+      "loss": 1.0732,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.715083798882683e-06,
+      "loss": 0.9819,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.826815642458101e-06,
+      "loss": 1.0615,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.93854748603352e-06,
+      "loss": 1.0932,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.050279329608939e-06,
+      "loss": 1.0645,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.162011173184358e-06,
+      "loss": 1.0893,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.273743016759777e-06,
+      "loss": 1.0977,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.385474860335197e-06,
+      "loss": 1.0095,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.497206703910616e-06,
+      "loss": 1.0536,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.608938547486035e-06,
+      "loss": 0.3001,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.720670391061454e-06,
+      "loss": 1.1213,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.832402234636873e-06,
+      "loss": 1.0857,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.944134078212291e-06,
+      "loss": 1.1084,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.005586592178771e-05,
+      "loss": 1.0932,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0167597765363129e-05,
+      "loss": 1.1151,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0279329608938548e-05,
+      "loss": 1.1025,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0391061452513967e-05,
+      "loss": 1.0693,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0502793296089386e-05,
+      "loss": 1.0803,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0614525139664806e-05,
+      "loss": 1.0208,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0726256983240225e-05,
+      "loss": 1.0734,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0837988826815644e-05,
+      "loss": 1.0538,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0949720670391063e-05,
+      "loss": 1.0481,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1061452513966481e-05,
+      "loss": 1.0069,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.11731843575419e-05,
+      "loss": 1.0412,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1284916201117319e-05,
+      "loss": 1.0039,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1396648044692738e-05,
+      "loss": 1.0952,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1508379888268157e-05,
+      "loss": 1.1374,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1620111731843577e-05,
+      "loss": 1.0281,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1731843575418994e-05,
+      "loss": 1.0593,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1843575418994415e-05,
+      "loss": 1.0191,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1955307262569834e-05,
+      "loss": 1.0218,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2067039106145253e-05,
+      "loss": 1.0797,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2178770949720671e-05,
+      "loss": 1.0986,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.229050279329609e-05,
+      "loss": 1.0304,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2402234636871509e-05,
+      "loss": 1.0778,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2513966480446928e-05,
+      "loss": 1.134,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2625698324022347e-05,
+      "loss": 1.0307,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2737430167597766e-05,
+      "loss": 0.266,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2849162011173186e-05,
+      "loss": 1.0437,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2960893854748603e-05,
+      "loss": 1.1123,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3072625698324024e-05,
+      "loss": 1.0339,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3184357541899443e-05,
+      "loss": 1.0634,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3296089385474861e-05,
+      "loss": 1.003,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.340782122905028e-05,
+      "loss": 1.0444,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3519553072625699e-05,
+      "loss": 0.9979,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3631284916201118e-05,
+      "loss": 1.0716,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3743016759776537e-05,
+      "loss": 1.0541,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3854748603351957e-05,
+      "loss": 0.9872,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3966480446927374e-05,
+      "loss": 1.0388,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4078212290502795e-05,
+      "loss": 1.0212,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4189944134078212e-05,
+      "loss": 0.985,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4301675977653633e-05,
+      "loss": 0.2777,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4413407821229052e-05,
+      "loss": 1.0523,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.452513966480447e-05,
+      "loss": 0.9581,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.463687150837989e-05,
+      "loss": 1.0563,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4748603351955308e-05,
+      "loss": 1.07,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4860335195530729e-05,
+      "loss": 1.0359,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4972067039106146e-05,
+      "loss": 0.9764,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5083798882681566e-05,
+      "loss": 1.016,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5195530726256983e-05,
+      "loss": 1.0635,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5307262569832404e-05,
+      "loss": 0.965,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5418994413407823e-05,
+      "loss": 1.0639,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.553072625698324e-05,
+      "loss": 1.039,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.564245810055866e-05,
+      "loss": 0.2691,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.575418994413408e-05,
+      "loss": 1.0157,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5865921787709498e-05,
+      "loss": 1.0212,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5977653631284917e-05,
+      "loss": 1.0617,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6089385474860336e-05,
+      "loss": 1.0756,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6201117318435755e-05,
+      "loss": 1.0201,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6312849162011177e-05,
+      "loss": 1.0184,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6424581005586592e-05,
+      "loss": 1.054,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6536312849162014e-05,
+      "loss": 1.1014,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.664804469273743e-05,
+      "loss": 1.0176,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6759776536312852e-05,
+      "loss": 1.0576,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.687150837988827e-05,
+      "loss": 1.0996,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.698324022346369e-05,
+      "loss": 1.0468,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.709497206703911e-05,
+      "loss": 1.0631,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7206703910614527e-05,
+      "loss": 1.0168,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7318435754189946e-05,
+      "loss": 1.068,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7430167597765365e-05,
+      "loss": 0.9837,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7541899441340784e-05,
+      "loss": 0.3001,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7653631284916203e-05,
+      "loss": 1.0957,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.776536312849162e-05,
+      "loss": 1.0985,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.787709497206704e-05,
+      "loss": 1.0373,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.798882681564246e-05,
+      "loss": 1.0332,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8100558659217878e-05,
+      "loss": 1.045,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8212290502793297e-05,
+      "loss": 1.0648,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8324022346368716e-05,
+      "loss": 1.0253,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8435754189944135e-05,
+      "loss": 0.9576,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8547486033519553e-05,
+      "loss": 1.0447,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8659217877094972e-05,
+      "loss": 0.9998,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8770949720670394e-05,
+      "loss": 1.1409,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.888268156424581e-05,
+      "loss": 1.0662,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8994413407821232e-05,
+      "loss": 1.0401,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.910614525139665e-05,
+      "loss": 1.0438,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.921787709497207e-05,
+      "loss": 1.0161,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.932960893854749e-05,
+      "loss": 1.1072,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9441340782122907e-05,
+      "loss": 0.9698,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9553072625698326e-05,
+      "loss": 1.027,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9664804469273745e-05,
+      "loss": 1.0583,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9776536312849164e-05,
+      "loss": 0.994,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9888268156424583e-05,
+      "loss": 1.0902,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.0203,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998525947744e-05,
+      "loss": 1.0104,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999994103791404e-05,
+      "loss": 1.0164,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999986733532287e-05,
+      "loss": 0.9776,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997641517256e-05,
+      "loss": 1.0917,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999963148715276e-05,
+      "loss": 1.0055,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999946934164337e-05,
+      "loss": 1.0454,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999927771524526e-05,
+      "loss": 1.0124,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999990566080149e-05,
+      "loss": 1.0107,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999880602001752e-05,
+      "loss": 1.0456,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999852595132698e-05,
+      "loss": 1.0576,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999821640202584e-05,
+      "loss": 1.0671,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999978773722054e-05,
+      "loss": 0.966,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999975088619655e-05,
+      "loss": 1.1166,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999711087141484e-05,
+      "loss": 1.028,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999668340067083e-05,
+      "loss": 1.0081,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999622644985937e-05,
+      "loss": 1.0438,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999574001911526e-05,
+      "loss": 1.0737,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999522410858187e-05,
+      "loss": 1.0315,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999467871841126e-05,
+      "loss": 1.0126,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999941038487643e-05,
+      "loss": 1.0461,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999934994998104e-05,
+      "loss": 0.9935,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999928656717278e-05,
+      "loss": 1.043,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999220236470326e-05,
+      "loss": 0.9858,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999915095789324e-05,
+      "loss": 1.0101,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999078731461942e-05,
+      "loss": 1.0375,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999900355719773e-05,
+      "loss": 1.041,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998925435122757e-05,
+      "loss": 1.0311,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998844365260068e-05,
+      "loss": 1.0565,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998760347633556e-05,
+      "loss": 0.289,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998673382267983e-05,
+      "loss": 1.0967,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998583469189e-05,
+      "loss": 1.0046,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998490608423106e-05,
+      "loss": 1.005,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998394799997684e-05,
+      "loss": 0.3057,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998296043940973e-05,
+      "loss": 0.9926,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998194340282088e-05,
+      "loss": 1.0529,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998089689051014e-05,
+      "loss": 1.0194,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997982090278606e-05,
+      "loss": 0.2541,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999787154399658e-05,
+      "loss": 1.0362,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997758050237533e-05,
+      "loss": 1.1033,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997641609034916e-05,
+      "loss": 0.9754,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997522220423062e-05,
+      "loss": 1.0493,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997399884437167e-05,
+      "loss": 1.0513,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997274601113294e-05,
+      "loss": 1.0015,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997146370488384e-05,
+      "loss": 0.283,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997015192600235e-05,
+      "loss": 0.9815,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996881067487526e-05,
+      "loss": 1.0113,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996743995189793e-05,
+      "loss": 0.281,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996603975747443e-05,
+      "loss": 1.078,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996461009201767e-05,
+      "loss": 1.016,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996315095594903e-05,
+      "loss": 1.0403,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999616623496987e-05,
+      "loss": 1.0036,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996014427370556e-05,
+      "loss": 1.035,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995859672841713e-05,
+      "loss": 1.0055,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995701971428967e-05,
+      "loss": 0.9931,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995541323178808e-05,
+      "loss": 0.964,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995377728138597e-05,
+      "loss": 0.9457,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995211186356564e-05,
+      "loss": 0.983,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995041697881805e-05,
+      "loss": 1.0148,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999486926276429e-05,
+      "loss": 1.1194,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994693881054853e-05,
+      "loss": 0.9882,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99945155528052e-05,
+      "loss": 0.9949,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994334278067905e-05,
+      "loss": 1.0431,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994150056896405e-05,
+      "loss": 0.9903,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993962889345012e-05,
+      "loss": 1.0833,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993772775468907e-05,
+      "loss": 1.0114,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993579715324135e-05,
+      "loss": 1.0357,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993383708967618e-05,
+      "loss": 0.9422,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993184756457132e-05,
+      "loss": 1.0163,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992982857851334e-05,
+      "loss": 1.0553,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992778013209752e-05,
+      "loss": 0.9633,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992570222592768e-05,
+      "loss": 0.9816,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999235948606164e-05,
+      "loss": 1.0291,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992145803678505e-05,
+      "loss": 0.9977,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999192917550635e-05,
+      "loss": 1.0898,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991709601609042e-05,
+      "loss": 1.0618,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991487082051314e-05,
+      "loss": 1.0597,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991261616898767e-05,
+      "loss": 0.9776,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991033206217868e-05,
+      "loss": 1.0153,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999080185007596e-05,
+      "loss": 1.0944,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990567548541245e-05,
+      "loss": 1.0728,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990330301682798e-05,
+      "loss": 1.0651,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990090109570565e-05,
+      "loss": 1.1126,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989846972275356e-05,
+      "loss": 1.1006,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989600889868846e-05,
+      "loss": 0.9786,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989351862423585e-05,
+      "loss": 0.997,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989099890012993e-05,
+      "loss": 1.0592,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998884497271135e-05,
+      "loss": 1.0076,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998858711059381e-05,
+      "loss": 1.0462,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998832630373639e-05,
+      "loss": 1.0561,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988062552215983e-05,
+      "loss": 0.99,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987795856110347e-05,
+      "loss": 1.0551,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99875262154981e-05,
+      "loss": 1.0713,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987253630458738e-05,
+      "loss": 1.0204,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986978101072627e-05,
+      "loss": 1.0142,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986699627420987e-05,
+      "loss": 1.0011,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998641820958592e-05,
+      "loss": 0.944,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986133847650392e-05,
+      "loss": 0.9923,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985846541698234e-05,
+      "loss": 1.0176,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985556291814147e-05,
+      "loss": 1.0848,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99852630980837e-05,
+      "loss": 1.0405,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998496696059333e-05,
+      "loss": 1.0541,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984667879430336e-05,
+      "loss": 1.0211,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99843658546829e-05,
+      "loss": 1.0575,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998406088644006e-05,
+      "loss": 0.3278,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983752974791715e-05,
+      "loss": 0.9662,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983442119828647e-05,
+      "loss": 1.0602,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99831283216425e-05,
+      "loss": 1.019,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982811580325784e-05,
+      "loss": 0.3145,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982491895971878e-05,
+      "loss": 1.0347,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982169268675024e-05,
+      "loss": 1.0748,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981843698530345e-05,
+      "loss": 0.9802,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981515185633812e-05,
+      "loss": 1.0498,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981183730082283e-05,
+      "loss": 0.9916,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980849331973467e-05,
+      "loss": 1.0553,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980511991405955e-05,
+      "loss": 1.0027,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980171708479194e-05,
+      "loss": 0.9694,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979828483293504e-05,
+      "loss": 1.0036,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997948231595007e-05,
+      "loss": 1.0311,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979133206550953e-05,
+      "loss": 0.3074,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978781155199062e-05,
+      "loss": 1.0511,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978426161998195e-05,
+      "loss": 1.0506,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978068227053002e-05,
+      "loss": 0.9957,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977707350469012e-05,
+      "loss": 1.0066,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977343532352608e-05,
+      "loss": 0.9582,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976976772811055e-05,
+      "loss": 1.023,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997660707195247e-05,
+      "loss": 1.0091,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997623442988585e-05,
+      "loss": 0.2919,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975858846721054e-05,
+      "loss": 0.9865,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975480322568802e-05,
+      "loss": 0.9313,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975098857540695e-05,
+      "loss": 0.9569,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974714451749185e-05,
+      "loss": 1.0099,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974327105307605e-05,
+      "loss": 1.0968,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973936818330145e-05,
+      "loss": 1.0231,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973543590931867e-05,
+      "loss": 1.0604,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973147423228698e-05,
+      "loss": 1.043,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972748315337434e-05,
+      "loss": 0.3334,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972346267375736e-05,
+      "loss": 1.0126,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997194127946213e-05,
+      "loss": 1.0066,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971533351716012e-05,
+      "loss": 1.0794,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971122484257643e-05,
+      "loss": 1.0017,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970708677208147e-05,
+      "loss": 0.945,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970291930689526e-05,
+      "loss": 1.0478,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969872244824638e-05,
+      "loss": 1.0193,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969449619737212e-05,
+      "loss": 1.0245,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969024055551843e-05,
+      "loss": 1.0132,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968595552393983e-05,
+      "loss": 0.9864,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996816411038997e-05,
+      "loss": 1.0231,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967729729666993e-05,
+      "loss": 0.987,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967292410353117e-05,
+      "loss": 1.0116,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996685215257726e-05,
+      "loss": 0.3385,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966408956469215e-05,
+      "loss": 1.0269,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996596282215965e-05,
+      "loss": 1.0321,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996551374978008e-05,
+      "loss": 1.0462,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965061739462903e-05,
+      "loss": 1.1047,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964606791341373e-05,
+      "loss": 0.9737,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964148905549617e-05,
+      "loss": 1.0513,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996368808222262e-05,
+      "loss": 0.9591,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996322432149624e-05,
+      "loss": 1.0149,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962757623507197e-05,
+      "loss": 0.9705,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996228798839308e-05,
+      "loss": 0.9716,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996181541629234e-05,
+      "loss": 1.0156,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99613399073443e-05,
+      "loss": 1.0011,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960861461689146e-05,
+      "loss": 1.0068,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996038007946792e-05,
+      "loss": 1.0131,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995989576082255e-05,
+      "loss": 1.0618,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959408505895807e-05,
+      "loss": 1.0406,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958918314831347e-05,
+      "loss": 1.083,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995842518777368e-05,
+      "loss": 1.0579,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957929124868184e-05,
+      "loss": 1.0104,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995743012626111e-05,
+      "loss": 1.096,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995692819209956e-05,
+      "loss": 1.0562,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956423322531515e-05,
+      "loss": 1.056,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995591551770581e-05,
+      "loss": 1.0834,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955404777772157e-05,
+      "loss": 1.0424,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954891102881126e-05,
+      "loss": 0.3342,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954374493184153e-05,
+      "loss": 0.3423,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953854948833537e-05,
+      "loss": 1.0498,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953332469982453e-05,
+      "loss": 1.0183,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952807056784925e-05,
+      "loss": 1.0293,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952278709395855e-05,
+      "loss": 0.285,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951747427971e-05,
+      "loss": 1.0454,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951213212666993e-05,
+      "loss": 1.0458,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950676063641325e-05,
+      "loss": 1.0579,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950135981052353e-05,
+      "loss": 1.0489,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99495929650593e-05,
+      "loss": 1.0549,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994904701582225e-05,
+      "loss": 1.012,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948498133502155e-05,
+      "loss": 0.9953,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947946318260834e-05,
+      "loss": 0.327,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947391570260964e-05,
+      "loss": 1.0219,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994683388966609e-05,
+      "loss": 1.0092,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994627327664063e-05,
+      "loss": 1.0554,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945709731349852e-05,
+      "loss": 0.9892,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945143253959896e-05,
+      "loss": 0.9698,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944573844637763e-05,
+      "loss": 1.0184,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944001503551326e-05,
+      "loss": 0.9346,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943426230869313e-05,
+      "loss": 0.9874,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942848026761324e-05,
+      "loss": 1.0532,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942266891397817e-05,
+      "loss": 0.337,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941682824950117e-05,
+      "loss": 0.2871,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941095827590415e-05,
+      "loss": 1.0189,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994050589949176e-05,
+      "loss": 1.0422,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939913040828073e-05,
+      "loss": 1.0773,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939317251774134e-05,
+      "loss": 0.987,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938718532505584e-05,
+      "loss": 1.0596,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993811688319894e-05,
+      "loss": 0.9809,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993751230403156e-05,
+      "loss": 1.001,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936904795181696e-05,
+      "loss": 0.9813,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993629435682844e-05,
+      "loss": 1.0309,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993568098915176e-05,
+      "loss": 1.0505,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935064692332476e-05,
+      "loss": 0.9289,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934445466552283e-05,
+      "loss": 1.0396,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993382331199374e-05,
+      "loss": 0.9707,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933198228840254e-05,
+      "loss": 1.0354,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932570217276115e-05,
+      "loss": 1.0367,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931939277486463e-05,
+      "loss": 1.019,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931305409657307e-05,
+      "loss": 0.9466,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993066861397552e-05,
+      "loss": 0.9844,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930028890628832e-05,
+      "loss": 1.1074,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929386239805843e-05,
+      "loss": 0.9911,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992874066169601e-05,
+      "loss": 1.017,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928092156489664e-05,
+      "loss": 1.0471,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992744072437798e-05,
+      "loss": 1.0741,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992678636555301e-05,
+      "loss": 0.3642,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926129080207676e-05,
+      "loss": 0.9898,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925468868535743e-05,
+      "loss": 1.013,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924805730731847e-05,
+      "loss": 1.0581,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992413966699149e-05,
+      "loss": 0.997,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992347067751104e-05,
+      "loss": 1.0381,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922798762487715e-05,
+      "loss": 1.0074,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922123922119606e-05,
+      "loss": 0.9511,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921446156605663e-05,
+      "loss": 1.0221,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99207654661457e-05,
+      "loss": 1.0648,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920081850940382e-05,
+      "loss": 0.9394,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919395311191256e-05,
+      "loss": 1.0119,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991870584710072e-05,
+      "loss": 1.004,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918013458872036e-05,
+      "loss": 0.9808,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991731814670932e-05,
+      "loss": 1.0073,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916619910817564e-05,
+      "loss": 1.0454,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915918751402615e-05,
+      "loss": 1.0041,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991521466867118e-05,
+      "loss": 1.0422,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991450766283083e-05,
+      "loss": 0.9996,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991379773409e-05,
+      "loss": 1.0294,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913084882657978e-05,
+      "loss": 1.0339,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912369108744927e-05,
+      "loss": 1.0093,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911650412561862e-05,
+      "loss": 0.9874,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991092879432066e-05,
+      "loss": 0.9816,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910204254234068e-05,
+      "loss": 0.9916,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990947679251568e-05,
+      "loss": 1.053,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990874640937997e-05,
+      "loss": 1.0531,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990801310504225e-05,
+      "loss": 1.0475,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907276879718715e-05,
+      "loss": 1.0233,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906537733626407e-05,
+      "loss": 1.0297,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905795666983234e-05,
+      "loss": 1.0416,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990505068000797e-05,
+      "loss": 1.0498,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904302772920238e-05,
+      "loss": 1.0084,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903551945940537e-05,
+      "loss": 1.031,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990279819929021e-05,
+      "loss": 0.9891,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990204153319147e-05,
+      "loss": 1.0209,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901281947867402e-05,
+      "loss": 1.0202,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900519443541927e-05,
+      "loss": 0.9495,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899754020439843e-05,
+      "loss": 1.0472,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898985678786802e-05,
+      "loss": 0.9953,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989821441880933e-05,
+      "loss": 0.9958,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989744024073479e-05,
+      "loss": 1.0218,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896663144791423e-05,
+      "loss": 0.9916,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895883131208325e-05,
+      "loss": 1.0705,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989510020021545e-05,
+      "loss": 0.9881,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989431435204362e-05,
+      "loss": 1.0519,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98935255869245e-05,
+      "loss": 1.0659,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892733905090642e-05,
+      "loss": 1.0336,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989193930677543e-05,
+      "loss": 0.9237,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989114179221312e-05,
+      "loss": 0.9928,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890341361638837e-05,
+      "loss": 0.9393,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889538015288545e-05,
+      "loss": 1.0642,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988873175339909e-05,
+      "loss": 1.0665,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887922576208155e-05,
+      "loss": 0.3874,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887110483954303e-05,
+      "loss": 0.9612,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988629547687694e-05,
+      "loss": 1.0675,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885477555216346e-05,
+      "loss": 1.0717,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884656719213645e-05,
+      "loss": 1.0396,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883832969110837e-05,
+      "loss": 1.0611,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883006305150767e-05,
+      "loss": 0.9289,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882176727577143e-05,
+      "loss": 0.973,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881344236634537e-05,
+      "loss": 0.9627,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988050883256837e-05,
+      "loss": 1.0181,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879670515624937e-05,
+      "loss": 1.0471,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878829286051374e-05,
+      "loss": 1.0274,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877985144095687e-05,
+      "loss": 1.0441,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987713809000674e-05,
+      "loss": 1.0032,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876288124034252e-05,
+      "loss": 0.97,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875435246428798e-05,
+      "loss": 0.3541,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874579457441824e-05,
+      "loss": 1.0195,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873720757325613e-05,
+      "loss": 1.0177,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872859146333333e-05,
+      "loss": 1.0531,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871994624718985e-05,
+      "loss": 0.9687,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987112719273745e-05,
+      "loss": 0.9826,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987025685064444e-05,
+      "loss": 1.0312,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869383598696555e-05,
+      "loss": 1.0378,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868507437151235e-05,
+      "loss": 1.024,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867628366266776e-05,
+      "loss": 0.9643,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866746386302345e-05,
+      "loss": 1.0372,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865861497517957e-05,
+      "loss": 1.0114,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864973700174483e-05,
+      "loss": 1.0076,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864082994533658e-05,
+      "loss": 1.0753,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863189380858068e-05,
+      "loss": 1.0451,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862292859411164e-05,
+      "loss": 1.0264,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861393430457248e-05,
+      "loss": 1.0396,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986049109426148e-05,
+      "loss": 0.9938,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985958585108988e-05,
+      "loss": 1.029,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858677701209324e-05,
+      "loss": 1.0887,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985776664488754e-05,
+      "loss": 1.044,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985685268239312e-05,
+      "loss": 0.9561,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855935813995504e-05,
+      "loss": 0.9677,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855016039965006e-05,
+      "loss": 1.0331,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854093360572775e-05,
+      "loss": 0.9753,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853167776090832e-05,
+      "loss": 1.0164,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852239286792047e-05,
+      "loss": 0.8929,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851307892950146e-05,
+      "loss": 1.0278,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850373594839717e-05,
+      "loss": 1.0317,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98494363927362e-05,
+      "loss": 0.9874,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848496286915894e-05,
+      "loss": 0.9883,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847553277655948e-05,
+      "loss": 0.9818,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984660736523437e-05,
+      "loss": 0.9902,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984565854993003e-05,
+      "loss": 1.0752,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844706832022646e-05,
+      "loss": 0.9733,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843752211792792e-05,
+      "loss": 1.0121,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842794689521905e-05,
+      "loss": 1.0301,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984183426549227e-05,
+      "loss": 0.9623,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984087093998703e-05,
+      "loss": 0.9986,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839904713290186e-05,
+      "loss": 0.9726,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838935585686586e-05,
+      "loss": 0.9634,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837963557461945e-05,
+      "loss": 0.939,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983698862890282e-05,
+      "loss": 0.9631,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836010800296642e-05,
+      "loss": 1.0101,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835030071931672e-05,
+      "loss": 1.0479,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834046444097046e-05,
+      "loss": 0.9875,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833059917082744e-05,
+      "loss": 0.9789,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983207049117961e-05,
+      "loss": 0.9597,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831078166679334e-05,
+      "loss": 0.9932,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830082943874464e-05,
+      "loss": 0.9835,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98290848230584e-05,
+      "loss": 0.9366,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98280838045254e-05,
+      "loss": 1.0266,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827079888570573e-05,
+      "loss": 1.0111,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982607307548989e-05,
+      "loss": 1.0051,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825063365580165e-05,
+      "loss": 1.0782,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824050759139068e-05,
+      "loss": 1.016,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823035256465136e-05,
+      "loss": 0.9525,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982201685785774e-05,
+      "loss": 0.9766,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820995563617117e-05,
+      "loss": 0.987,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819971374044356e-05,
+      "loss": 1.0113,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818944289441402e-05,
+      "loss": 1.0075,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817914310111044e-05,
+      "loss": 0.9639,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981688143635694e-05,
+      "loss": 1.0275,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981584566848358e-05,
+      "loss": 1.0182,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981480700679633e-05,
+      "loss": 1.0575,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981376545160139e-05,
+      "loss": 1.0343,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812721003205825e-05,
+      "loss": 1.0823,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981167366191755e-05,
+      "loss": 1.0282,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981062342804533e-05,
+      "loss": 0.9715,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980957030189879e-05,
+      "loss": 1.0787,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808514283788398e-05,
+      "loss": 1.0498,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807455374025477e-05,
+      "loss": 0.932,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980639357292221e-05,
+      "loss": 1.0468,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805328880791626e-05,
+      "loss": 1.0305,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804261297947606e-05,
+      "loss": 1.0071,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980319082470488e-05,
+      "loss": 0.9821,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802117461379047e-05,
+      "loss": 0.9816,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801041208286535e-05,
+      "loss": 0.9703,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799962065744634e-05,
+      "loss": 0.9333,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798880034071493e-05,
+      "loss": 1.0102,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797795113586104e-05,
+      "loss": 1.0175,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796707304608314e-05,
+      "loss": 0.9981,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795616607458817e-05,
+      "loss": 0.9836,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794523022459168e-05,
+      "loss": 0.9897,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793426549931757e-05,
+      "loss": 0.9961,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792327190199843e-05,
+      "loss": 0.9787,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979122494358753e-05,
+      "loss": 0.9851,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979011981041977e-05,
+      "loss": 0.9846,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789011791022365e-05,
+      "loss": 1.1081,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978790088572197e-05,
+      "loss": 1.0245,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786787094846097e-05,
+      "loss": 1.0294,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97856704187231e-05,
+      "loss": 0.9407,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978455085768219e-05,
+      "loss": 0.9938,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783428412053422e-05,
+      "loss": 0.9948,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782303082167705e-05,
+      "loss": 0.9775,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781174868356797e-05,
+      "loss": 1.0009,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780043770953306e-05,
+      "loss": 1.0636,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778909790290697e-05,
+      "loss": 0.9688,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777772926703276e-05,
+      "loss": 1.0079,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97766331805262e-05,
+      "loss": 1.0206,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775490552095485e-05,
+      "loss": 1.0417,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774345041747982e-05,
+      "loss": 1.0096,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773196649821405e-05,
+      "loss": 1.0062,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772045376654308e-05,
+      "loss": 1.0222,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97708912225861e-05,
+      "loss": 0.9609,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976973418795704e-05,
+      "loss": 1.0118,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768574273108228e-05,
+      "loss": 0.9747,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767411478381623e-05,
+      "loss": 1.0491,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766245804120032e-05,
+      "loss": 0.324,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765077250667104e-05,
+      "loss": 1.0357,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976390581836734e-05,
+      "loss": 0.9553,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762731507566094e-05,
+      "loss": 0.9861,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976155431860956e-05,
+      "loss": 0.9994,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760374251844793e-05,
+      "loss": 0.9834,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759191307619677e-05,
+      "loss": 0.961,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975800548628297e-05,
+      "loss": 1.0598,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975681678818426e-05,
+      "loss": 1.022,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755625213673984e-05,
+      "loss": 1.0112,
+      "step": 587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754430763103428e-05,
+      "loss": 1.045,
+      "step": 588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975323343682474e-05,
+      "loss": 0.9917,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752033235190895e-05,
+      "loss": 1.0134,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750830158555728e-05,
+      "loss": 0.9918,
+      "step": 591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749624207273916e-05,
+      "loss": 1.004,
+      "step": 592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748415381700988e-05,
+      "loss": 0.9832,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974720368219332e-05,
+      "loss": 1.0243,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745989109108134e-05,
+      "loss": 1.0264,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744771662803496e-05,
+      "loss": 0.9747,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743551343638324e-05,
+      "loss": 1.0094,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974232815197238e-05,
+      "loss": 1.0321,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974110208816627e-05,
+      "loss": 1.013,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739873152581456e-05,
+      "loss": 0.9432,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738641345580237e-05,
+      "loss": 0.9811,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737406667525766e-05,
+      "loss": 1.0407,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973616911878204e-05,
+      "loss": 1.0158,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734928699713897e-05,
+      "loss": 1.0505,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733685410687027e-05,
+      "loss": 1.0324,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732439252067967e-05,
+      "loss": 1.0363,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973119022422409e-05,
+      "loss": 0.9674,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729938327523635e-05,
+      "loss": 0.9479,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728683562335663e-05,
+      "loss": 0.9959,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97274259290301e-05,
+      "loss": 1.0289,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97261654279777e-05,
+      "loss": 1.007,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972490205955008e-05,
+      "loss": 0.9964,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972363582411969e-05,
+      "loss": 1.0354,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722366722059836e-05,
+      "loss": 1.0724,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721094753744655e-05,
+      "loss": 1.005,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719819919549138e-05,
+      "loss": 0.9833,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718542219849122e-05,
+      "loss": 1.0245,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717261655021286e-05,
+      "loss": 0.9156,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715978225443147e-05,
+      "loss": 1.0126,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714691931493087e-05,
+      "loss": 0.339,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713402773550307e-05,
+      "loss": 0.905,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971211075199487e-05,
+      "loss": 0.3343,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971081586720767e-05,
+      "loss": 0.9834,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709518119570465e-05,
+      "loss": 0.9719,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708217509465837e-05,
+      "loss": 1.0122,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970691403727722e-05,
+      "loss": 0.98,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970560770338889e-05,
+      "loss": 0.9678,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704298508185973e-05,
+      "loss": 1.0243,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702986452054426e-05,
+      "loss": 0.9354,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701671535381064e-05,
+      "loss": 1.0951,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700353758553536e-05,
+      "loss": 1.0916,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699033121960333e-05,
+      "loss": 0.9168,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697709625990793e-05,
+      "loss": 0.9704,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96963832710351e-05,
+      "loss": 1.0101,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969505405748428e-05,
+      "loss": 0.9643,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693721985730186e-05,
+      "loss": 1.0397,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969238705616554e-05,
+      "loss": 1.0173,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969104926918389e-05,
+      "loss": 0.9693,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689708625179627e-05,
+      "loss": 0.3512,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688365124547986e-05,
+      "loss": 0.9983,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687018767685048e-05,
+      "loss": 0.9385,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685669554987732e-05,
+      "loss": 0.9907,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96843174868538e-05,
+      "loss": 0.9852,
+      "step": 643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682962563681857e-05,
+      "loss": 0.8984,
+      "step": 644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681604785871347e-05,
+      "loss": 1.0278,
+      "step": 645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680244153822558e-05,
+      "loss": 0.9556,
+      "step": 646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678880667936617e-05,
+      "loss": 0.9063,
+      "step": 647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677514328615496e-05,
+      "loss": 1.0669,
+      "step": 648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676145136262007e-05,
+      "loss": 0.9915,
+      "step": 649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674773091279794e-05,
+      "loss": 1.0251,
+      "step": 650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967339819407336e-05,
+      "loss": 1.0284,
+      "step": 651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672020445048036e-05,
+      "loss": 0.3437,
+      "step": 652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670639844609997e-05,
+      "loss": 1.0384,
+      "step": 653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669256393166258e-05,
+      "loss": 0.9198,
+      "step": 654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966787009112467e-05,
+      "loss": 0.9665,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966648093889394e-05,
+      "loss": 1.0513,
+      "step": 656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665088936883596e-05,
+      "loss": 0.9794,
+      "step": 657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966369408550402e-05,
+      "loss": 0.976,
+      "step": 658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662296385166422e-05,
+      "loss": 0.975,
+      "step": 659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660895836282866e-05,
+      "loss": 1.02,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965949243926624e-05,
+      "loss": 0.9812,
+      "step": 661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965808619453029e-05,
+      "loss": 0.9644,
+      "step": 662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965667710248959e-05,
+      "loss": 1.0136,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655265163559547e-05,
+      "loss": 1.0403,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965385037815642e-05,
+      "loss": 1.0488,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652432746697306e-05,
+      "loss": 1.0124,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651012269600133e-05,
+      "loss": 0.3474,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649588947283674e-05,
+      "loss": 0.96,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648162780167542e-05,
+      "loss": 0.9739,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964673376867218e-05,
+      "loss": 1.0296,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645301913218878e-05,
+      "loss": 1.0259,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643867214229764e-05,
+      "loss": 1.0523,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642429672127802e-05,
+      "loss": 1.0205,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640989287336795e-05,
+      "loss": 0.961,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963954606028138e-05,
+      "loss": 0.9697,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638099991387038e-05,
+      "loss": 0.9801,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963665108108008e-05,
+      "loss": 0.9982,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635199329787672e-05,
+      "loss": 0.9928,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633744737937796e-05,
+      "loss": 0.9881,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632287305959278e-05,
+      "loss": 1.025,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630827034281794e-05,
+      "loss": 1.0273,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962936392333584e-05,
+      "loss": 0.3631,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962789797355276e-05,
+      "loss": 0.9959,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962642918536473e-05,
+      "loss": 0.9844,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624957559204763e-05,
+      "loss": 1.0361,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962348309550671e-05,
+      "loss": 1.019,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962200579470526e-05,
+      "loss": 0.9731,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620525657235938e-05,
+      "loss": 0.9822,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96190426835351e-05,
+      "loss": 1.039,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617556874039945e-05,
+      "loss": 1.0118,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616068229188507e-05,
+      "loss": 1.0062,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961457674941965e-05,
+      "loss": 1.0064,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613082435173078e-05,
+      "loss": 1.0422,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961158528688933e-05,
+      "loss": 0.9952,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610085305009784e-05,
+      "loss": 0.3218,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960858248997665e-05,
+      "loss": 1.0189,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960707684223297e-05,
+      "loss": 0.9823,
+      "step": 697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605568362222632e-05,
+      "loss": 1.0163,
+      "step": 698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604057050390342e-05,
+      "loss": 0.3396,
+      "step": 699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602542907181657e-05,
+      "loss": 1.0044,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601025933042962e-05,
+      "loss": 1.0067,
+      "step": 701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959950612842147e-05,
+      "loss": 0.3818,
+      "step": 702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959798349376525e-05,
+      "loss": 0.9973,
+      "step": 703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596458029523174e-05,
+      "loss": 0.9744,
+      "step": 704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 1.0472,
+      "step": 705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593398614081206e-05,
+      "loss": 1.068,
+      "step": 706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959186466378326e-05,
+      "loss": 0.9993,
+      "step": 707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959032788570336e-05,
+      "loss": 1.0164,
+      "step": 708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588788280294568e-05,
+      "loss": 1.0012,
+      "step": 709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587245848010766e-05,
+      "loss": 1.0362,
+      "step": 710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585700589306688e-05,
+      "loss": 0.9355,
+      "step": 711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584152504637893e-05,
+      "loss": 0.9769,
+      "step": 712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582601594460767e-05,
+      "loss": 0.9713,
+      "step": 713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958104785923254e-05,
+      "loss": 0.976,
+      "step": 714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579491299411263e-05,
+      "loss": 1.0317,
+      "step": 715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577931915455827e-05,
+      "loss": 0.9741,
+      "step": 716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576369707825962e-05,
+      "loss": 0.9541,
+      "step": 717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574804676982215e-05,
+      "loss": 0.9528,
+      "step": 718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957323682338598e-05,
+      "loss": 0.9474,
+      "step": 719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957166614749947e-05,
+      "loss": 0.9423,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570092649785746e-05,
+      "loss": 0.9924,
+      "step": 721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956851633070868e-05,
+      "loss": 0.9478,
+      "step": 722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566937190732994e-05,
+      "loss": 1.0685,
+      "step": 723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565355230324238e-05,
+      "loss": 1.0454,
+      "step": 724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563770449948782e-05,
+      "loss": 0.3429,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562182850073844e-05,
+      "loss": 1.0326,
+      "step": 726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560592431167456e-05,
+      "loss": 0.9642,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95589991936985e-05,
+      "loss": 0.3438,
+      "step": 728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557403138136672e-05,
+      "loss": 0.9815,
+      "step": 729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955580426495251e-05,
+      "loss": 0.975,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554202574617375e-05,
+      "loss": 0.9863,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552598067603466e-05,
+      "loss": 1.0261,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550990744383805e-05,
+      "loss": 0.9819,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954938060543225e-05,
+      "loss": 0.9842,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547767651223486e-05,
+      "loss": 1.0504,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954615188223303e-05,
+      "loss": 0.9912,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544533298937222e-05,
+      "loss": 0.9793,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542911901813247e-05,
+      "loss": 0.3382,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95412876913391e-05,
+      "loss": 1.0173,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953966066799362e-05,
+      "loss": 1.0028,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538030832256468e-05,
+      "loss": 0.9548,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536398184608143e-05,
+      "loss": 1.0263,
+      "step": 742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953476272552996e-05,
+      "loss": 1.0099,
+      "step": 743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533124455504073e-05,
+      "loss": 0.994,
+      "step": 744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531483375013457e-05,
+      "loss": 1.0214,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529839484541925e-05,
+      "loss": 0.3046,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952819278457411e-05,
+      "loss": 0.9628,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526543275595476e-05,
+      "loss": 1.0409,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952489095809232e-05,
+      "loss": 0.9826,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523235832551752e-05,
+      "loss": 0.3528,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521577899461732e-05,
+      "loss": 1.0095,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951991715931103e-05,
+      "loss": 1.0256,
+      "step": 752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951825361258925e-05,
+      "loss": 0.9709,
+      "step": 753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516587259786824e-05,
+      "loss": 1.0211,
+      "step": 754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514918101395012e-05,
+      "loss": 1.0976,
+      "step": 755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513246137905896e-05,
+      "loss": 1.0101,
+      "step": 756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511571369812388e-05,
+      "loss": 1.0113,
+      "step": 757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509893797608228e-05,
+      "loss": 0.9947,
+      "step": 758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508213421787987e-05,
+      "loss": 0.9578,
+      "step": 759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950653024284705e-05,
+      "loss": 1.0217,
+      "step": 760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504844261281638e-05,
+      "loss": 0.3692,
+      "step": 761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503155477588794e-05,
+      "loss": 0.9471,
+      "step": 762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501463892266396e-05,
+      "loss": 1.0062,
+      "step": 763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949976950581314e-05,
+      "loss": 0.972,
+      "step": 764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949807231872854e-05,
+      "loss": 0.9498,
+      "step": 765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949637233151295e-05,
+      "loss": 0.9654,
+      "step": 766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494669544667545e-05,
+      "loss": 0.337,
+      "step": 767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492963958694326e-05,
+      "loss": 0.9569,
+      "step": 768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949125557409611e-05,
+      "loss": 0.9291,
+      "step": 769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489544391376555e-05,
+      "loss": 1.008,
+      "step": 770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948783041104013e-05,
+      "loss": 1.0072,
+      "step": 771
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486113633592136e-05,
+      "loss": 1.0161,
+      "step": 772
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94843940595387e-05,
+      "loss": 0.9737,
+      "step": 773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482671689386766e-05,
+      "loss": 1.0253,
+      "step": 774
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948094652364411e-05,
+      "loss": 0.9087,
+      "step": 775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479218562819326e-05,
+      "loss": 1.0134,
+      "step": 776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477487807421837e-05,
+      "loss": 1.0301,
+      "step": 777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475754257961887e-05,
+      "loss": 1.0361,
+      "step": 778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474017914950546e-05,
+      "loss": 0.3615,
+      "step": 779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472278778899704e-05,
+      "loss": 1.0714,
+      "step": 780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470536850322076e-05,
+      "loss": 0.9518,
+      "step": 781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94687921297312e-05,
+      "loss": 0.9659,
+      "step": 782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467044617641445e-05,
+      "loss": 0.9475,
+      "step": 783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465294314567987e-05,
+      "loss": 1.0415,
+      "step": 784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946354122102684e-05,
+      "loss": 0.966,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946178533753483e-05,
+      "loss": 0.9592,
+      "step": 786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946002666460961e-05,
+      "loss": 0.9455,
+      "step": 787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458265202769656e-05,
+      "loss": 0.9784,
+      "step": 788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945650095253427e-05,
+      "loss": 1.0447,
+      "step": 789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945473391442356e-05,
+      "loss": 0.9969,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452964088958483e-05,
+      "loss": 0.9347,
+      "step": 791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945119147666079e-05,
+      "loss": 0.9767,
+      "step": 792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449416078053067e-05,
+      "loss": 1.0086,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447637893658727e-05,
+      "loss": 0.9777,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944585692400199e-05,
+      "loss": 1.009,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444073169607907e-05,
+      "loss": 1.0068,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442286631002348e-05,
+      "loss": 0.9806,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440497308712e-05,
+      "loss": 0.9913,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943870520326438e-05,
+      "loss": 1.0063,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436910315187815e-05,
+      "loss": 1.0104,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435112645011462e-05,
+      "loss": 0.9572,
+      "step": 801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943331219326528e-05,
+      "loss": 1.0414,
+      "step": 802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431508960480075e-05,
+      "loss": 1.0178,
+      "step": 803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429702947187455e-05,
+      "loss": 1.0223,
+      "step": 804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942789415391985e-05,
+      "loss": 0.9504,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942608258121051e-05,
+      "loss": 1.0596,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424268229593507e-05,
+      "loss": 0.9984,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942245109960373e-05,
+      "loss": 0.9768,
+      "step": 808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420631191776892e-05,
+      "loss": 0.9916,
+      "step": 809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418808506649515e-05,
+      "loss": 1.0227,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941698304475895e-05,
+      "loss": 1.0428,
+      "step": 811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415154806643358e-05,
+      "loss": 1.0171,
+      "step": 812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413323792841726e-05,
+      "loss": 0.9998,
+      "step": 813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941149000389386e-05,
+      "loss": 0.9847,
+      "step": 814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940965344034037e-05,
+      "loss": 1.0219,
+      "step": 815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94078141027227e-05,
+      "loss": 0.9924,
+      "step": 816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940597199158311e-05,
+      "loss": 0.9577,
+      "step": 817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404127107464662e-05,
+      "loss": 1.0226,
+      "step": 818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402279450911255e-05,
+      "loss": 0.9443,
+      "step": 819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94004290224676e-05,
+      "loss": 1.0664,
+      "step": 820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939857582267922e-05,
+      "loss": 1.0515,
+      "step": 821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396719852092457e-05,
+      "loss": 1.0169,
+      "step": 822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394861111254467e-05,
+      "loss": 1.018,
+      "step": 823
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939299960071323e-05,
+      "loss": 0.9474,
+      "step": 824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391135321017542e-05,
+      "loss": 0.9505,
+      "step": 825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938926827271701e-05,
+      "loss": 1.035,
+      "step": 826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938739845636205e-05,
+      "loss": 0.9702,
+      "step": 827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938552587250392e-05,
+      "loss": 1.0072,
+      "step": 828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383650521694662e-05,
+      "loss": 1.0857,
+      "step": 829
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938177240448716e-05,
+      "loss": 0.9613,
+      "step": 830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379891521435098e-05,
+      "loss": 1.0067,
+      "step": 831
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378007873092975e-05,
+      "loss": 1.0064,
+      "step": 832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376121460016123e-05,
+      "loss": 1.0331,
+      "step": 833
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374232282760664e-05,
+      "loss": 0.994,
+      "step": 834
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937234034188355e-05,
+      "loss": 1.0052,
+      "step": 835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370445637942552e-05,
+      "loss": 1.0053,
+      "step": 836
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368548171496244e-05,
+      "loss": 0.9884,
+      "step": 837
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366647943104014e-05,
+      "loss": 0.9563,
+      "step": 838
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364744953326077e-05,
+      "loss": 0.9919,
+      "step": 839
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936283920272345e-05,
+      "loss": 1.0225,
+      "step": 840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360930691857966e-05,
+      "loss": 0.9591,
+      "step": 841
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935901942129228e-05,
+      "loss": 0.9704,
+      "step": 842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935710539158985e-05,
+      "loss": 1.0419,
+      "step": 843
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355188603314956e-05,
+      "loss": 0.995,
+      "step": 844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935326905703268e-05,
+      "loss": 1.0158,
+      "step": 845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351346753308933e-05,
+      "loss": 0.9818,
+      "step": 846
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349421692710428e-05,
+      "loss": 0.973,
+      "step": 847
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347493875804686e-05,
+      "loss": 1.0533,
+      "step": 848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345563303160056e-05,
+      "loss": 0.9813,
+      "step": 849
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343629975345687e-05,
+      "loss": 1.0012,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341693892931544e-05,
+      "loss": 0.9952,
+      "step": 851
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339755056488405e-05,
+      "loss": 0.979,
+      "step": 852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337813466587864e-05,
+      "loss": 0.9911,
+      "step": 853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335869123802313e-05,
+      "loss": 0.9049,
+      "step": 854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933392202870497e-05,
+      "loss": 0.9446,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331972181869857e-05,
+      "loss": 0.9499,
+      "step": 856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330019583871813e-05,
+      "loss": 0.3065,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932806423528648e-05,
+      "loss": 1.0105,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326106136690322e-05,
+      "loss": 1.0056,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324145288660595e-05,
+      "loss": 0.9649,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322181691775387e-05,
+      "loss": 0.9757,
+      "step": 861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320215346613586e-05,
+      "loss": 0.9648,
+      "step": 862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318246253754886e-05,
+      "loss": 0.939,
+      "step": 863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93162744137798e-05,
+      "loss": 0.9588,
+      "step": 864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931429982726965e-05,
+      "loss": 0.9781,
+      "step": 865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312322494806556e-05,
+      "loss": 0.9482,
+      "step": 866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310342416973468e-05,
+      "loss": 1.0235,
+      "step": 867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930835959435412e-05,
+      "loss": 0.9763,
+      "step": 868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306374027533078e-05,
+      "loss": 1.0382,
+      "step": 869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304385717095708e-05,
+      "loss": 0.9759,
+      "step": 870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930239466362818e-05,
+      "loss": 1.0406,
+      "step": 871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300400867717484e-05,
+      "loss": 0.9649,
+      "step": 872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298404329951404e-05,
+      "loss": 0.8946,
+      "step": 873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296405050918546e-05,
+      "loss": 0.3279,
+      "step": 874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294403031208317e-05,
+      "loss": 1.0636,
+      "step": 875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929239827141093e-05,
+      "loss": 1.0088,
+      "step": 876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929039077211741e-05,
+      "loss": 0.9748,
+      "step": 877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288380533919597e-05,
+      "loss": 1.0252,
+      "step": 878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928636755741012e-05,
+      "loss": 0.9938,
+      "step": 879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928435184318243e-05,
+      "loss": 0.9625,
+      "step": 880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282333391830777e-05,
+      "loss": 0.336,
+      "step": 881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280312203950228e-05,
+      "loss": 0.9471,
+      "step": 882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278288280136647e-05,
+      "loss": 1.0034,
+      "step": 883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927626162098671e-05,
+      "loss": 0.9751,
+      "step": 884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274232227097885e-05,
+      "loss": 1.0158,
+      "step": 885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927220009906848e-05,
+      "loss": 0.9773,
+      "step": 886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927016523749757e-05,
+      "loss": 0.3307,
+      "step": 887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268127642985058e-05,
+      "loss": 0.9597,
+      "step": 888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266087316131655e-05,
+      "loss": 0.945,
+      "step": 889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264044257538864e-05,
+      "loss": 1.0251,
+      "step": 890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261998467809e-05,
+      "loss": 0.9866,
+      "step": 891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259949947545186e-05,
+      "loss": 0.9298,
+      "step": 892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925789869735134e-05,
+      "loss": 0.9571,
+      "step": 893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255844717832206e-05,
+      "loss": 0.9952,
+      "step": 894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253788009593308e-05,
+      "loss": 0.9588,
+      "step": 895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251728573240983e-05,
+      "loss": 1.0245,
+      "step": 896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249666409382387e-05,
+      "loss": 0.3245,
+      "step": 897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247601518625454e-05,
+      "loss": 1.0946,
+      "step": 898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245533901578943e-05,
+      "loss": 1.0348,
+      "step": 899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243463558852405e-05,
+      "loss": 1.022,
+      "step": 900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92413904910562e-05,
+      "loss": 1.0234,
+      "step": 901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239314698801493e-05,
+      "loss": 0.9742,
+      "step": 902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237236182700244e-05,
+      "loss": 1.041,
+      "step": 903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235154943365224e-05,
+      "loss": 0.9639,
+      "step": 904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233070981410007e-05,
+      "loss": 0.9265,
+      "step": 905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923098429744896e-05,
+      "loss": 0.996,
+      "step": 906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228894892097267e-05,
+      "loss": 0.3218,
+      "step": 907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92268027659709e-05,
+      "loss": 0.9905,
+      "step": 908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224707919686648e-05,
+      "loss": 0.9914,
+      "step": 909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922261035386208e-05,
+      "loss": 1.0077,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220510069115595e-05,
+      "loss": 0.9953,
+      "step": 911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921840706606637e-05,
+      "loss": 0.9565,
+      "step": 912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921630134533439e-05,
+      "loss": 1.0236,
+      "step": 913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214192907540452e-05,
+      "loss": 0.9392,
+      "step": 914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212081753306143e-05,
+      "loss": 0.9978,
+      "step": 915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920996788325385e-05,
+      "loss": 0.3377,
+      "step": 916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207851298006766e-05,
+      "loss": 0.9754,
+      "step": 917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920573199818888e-05,
+      "loss": 0.9965,
+      "step": 918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203609984424993e-05,
+      "loss": 1.0739,
+      "step": 919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201485257340683e-05,
+      "loss": 1.017,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199357817562347e-05,
+      "loss": 0.9816,
+      "step": 921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197227665717183e-05,
+      "loss": 0.9237,
+      "step": 922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195094802433175e-05,
+      "loss": 0.9959,
+      "step": 923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9192959228339115e-05,
+      "loss": 0.9705,
+      "step": 924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190820944064594e-05,
+      "loss": 0.9917,
+      "step": 925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918867995024e-05,
+      "loss": 0.9613,
+      "step": 926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918653624749652e-05,
+      "loss": 1.0215,
+      "step": 927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918438983646614e-05,
+      "loss": 1.0479,
+      "step": 928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182240717781642e-05,
+      "loss": 1.0077,
+      "step": 929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918008889207661e-05,
+      "loss": 0.9867,
+      "step": 930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917793435998543e-05,
+      "loss": 1.016,
+      "step": 931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175777122143273e-05,
+      "loss": 0.9152,
+      "step": 932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917361717918612e-05,
+      "loss": 0.9694,
+      "step": 933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171454531750745e-05,
+      "loss": 0.99,
+      "step": 934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169289180474714e-05,
+      "loss": 0.9906,
+      "step": 935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91671211259964e-05,
+      "loss": 0.9719,
+      "step": 936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164950368954965e-05,
+      "loss": 0.9791,
+      "step": 937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162776909990375e-05,
+      "loss": 0.9445,
+      "step": 938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160600749743384e-05,
+      "loss": 0.9866,
+      "step": 939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158421888855548e-05,
+      "loss": 0.9965,
+      "step": 940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156240327969223e-05,
+      "loss": 0.9864,
+      "step": 941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154056067727548e-05,
+      "loss": 0.3406,
+      "step": 942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915186910877447e-05,
+      "loss": 1.0603,
+      "step": 943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149679451754726e-05,
+      "loss": 0.9508,
+      "step": 944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147487097313854e-05,
+      "loss": 0.9682,
+      "step": 945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145292046098175e-05,
+      "loss": 1.0107,
+      "step": 946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914309429875482e-05,
+      "loss": 0.9738,
+      "step": 947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140893855931705e-05,
+      "loss": 0.9991,
+      "step": 948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138690718277542e-05,
+      "loss": 0.9743,
+      "step": 949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913648488644184e-05,
+      "loss": 0.9799,
+      "step": 950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134276361074907e-05,
+      "loss": 0.9338,
+      "step": 951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132065142827834e-05,
+      "loss": 0.9534,
+      "step": 952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912985123235251e-05,
+      "loss": 1.0146,
+      "step": 953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912763463030162e-05,
+      "loss": 0.9317,
+      "step": 954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125415337328644e-05,
+      "loss": 1.0198,
+      "step": 955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912319335408785e-05,
+      "loss": 1.0304,
+      "step": 956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120968681234303e-05,
+      "loss": 0.3646,
+      "step": 957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118741319423862e-05,
+      "loss": 0.951,
+      "step": 958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9116511269313173e-05,
+      "loss": 0.9696,
+      "step": 959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114278531559677e-05,
+      "loss": 0.992,
+      "step": 960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112043106821612e-05,
+      "loss": 1.0155,
+      "step": 961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109804995758003e-05,
+      "loss": 1.0204,
+      "step": 962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910756419902867e-05,
+      "loss": 1.0649,
+      "step": 963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910532071729422e-05,
+      "loss": 1.0015,
+      "step": 964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103074551216058e-05,
+      "loss": 0.9717,
+      "step": 965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100825701456376e-05,
+      "loss": 0.9782,
+      "step": 966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909857416867816e-05,
+      "loss": 1.0428,
+      "step": 967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096319953545186e-05,
+      "loss": 0.9706,
+      "step": 968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094063056722015e-05,
+      "loss": 1.0082,
+      "step": 969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091803478874008e-05,
+      "loss": 1.0324,
+      "step": 970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089541220667312e-05,
+      "loss": 1.0048,
+      "step": 971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908727628276886e-05,
+      "loss": 0.9049,
+      "step": 972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908500866584639e-05,
+      "loss": 0.9893,
+      "step": 973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082738370568407e-05,
+      "loss": 0.9912,
+      "step": 974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908046539760422e-05,
+      "loss": 0.9915,
+      "step": 975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078189747623934e-05,
+      "loss": 0.9702,
+      "step": 976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075911421298425e-05,
+      "loss": 1.0209,
+      "step": 977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073630419299373e-05,
+      "loss": 0.9733,
+      "step": 978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071346742299238e-05,
+      "loss": 0.9528,
+      "step": 979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069060390971273e-05,
+      "loss": 0.9689,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066771365989515e-05,
+      "loss": 0.3441,
+      "step": 981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90644796680288e-05,
+      "loss": 1.0499,
+      "step": 982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906218529776474e-05,
+      "loss": 1.0037,
+      "step": 983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.905988825587374e-05,
+      "loss": 1.0219,
+      "step": 984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057588543032988e-05,
+      "loss": 0.9733,
+      "step": 985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055286159920466e-05,
+      "loss": 0.3237,
+      "step": 986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052981107214947e-05,
+      "loss": 0.9124,
+      "step": 987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050673385595977e-05,
+      "loss": 0.927,
+      "step": 988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048362995743898e-05,
+      "loss": 0.9781,
+      "step": 989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046049938339837e-05,
+      "loss": 0.9409,
+      "step": 990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904373421406571e-05,
+      "loss": 0.9933,
+      "step": 991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041415823604214e-05,
+      "loss": 1.0016,
+      "step": 992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039094767638834e-05,
+      "loss": 0.959,
+      "step": 993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036771046853845e-05,
+      "loss": 0.99,
+      "step": 994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034444661934302e-05,
+      "loss": 0.965,
+      "step": 995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032115613566047e-05,
+      "loss": 0.3551,
+      "step": 996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902978390243571e-05,
+      "loss": 0.9828,
+      "step": 997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027449529230703e-05,
+      "loss": 0.9484,
+      "step": 998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902511249463922e-05,
+      "loss": 0.9768,
+      "step": 999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022772799350248e-05,
+      "loss": 0.9765,
+      "step": 1000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020430444053554e-05,
+      "loss": 0.9316,
+      "step": 1001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018085429439683e-05,
+      "loss": 0.9784,
+      "step": 1002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015737756199976e-05,
+      "loss": 0.9783,
+      "step": 1003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901338742502655e-05,
+      "loss": 0.9726,
+      "step": 1004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011034436612305e-05,
+      "loss": 0.948,
+      "step": 1005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008678791650927e-05,
+      "loss": 0.9689,
+      "step": 1006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006320490836886e-05,
+      "loss": 0.9365,
+      "step": 1007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003959534865437e-05,
+      "loss": 0.9481,
+      "step": 1008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001595924432606e-05,
+      "loss": 0.9822,
+      "step": 1009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899922966023522e-05,
+      "loss": 0.9999,
+      "step": 1010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996860742970872e-05,
+      "loss": 1.024,
+      "step": 1011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994489173337943e-05,
+      "loss": 0.9749,
+      "step": 1012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992114952035602e-05,
+      "loss": 0.9818,
+      "step": 1013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989738079763788e-05,
+      "loss": 1.0473,
+      "step": 1014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987358557223232e-05,
+      "loss": 0.9532,
+      "step": 1015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898497638511544e-05,
+      "loss": 1.0084,
+      "step": 1016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982591564142702e-05,
+      "loss": 1.0289,
+      "step": 1017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980204095008087e-05,
+      "loss": 0.9786,
+      "step": 1018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897781397841545e-05,
+      "loss": 0.9731,
+      "step": 1019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975421215069416e-05,
+      "loss": 0.9803,
+      "step": 1020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973025805675403e-05,
+      "loss": 0.9652,
+      "step": 1021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970627750939595e-05,
+      "loss": 1.0009,
+      "step": 1022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896822705156897e-05,
+      "loss": 0.9512,
+      "step": 1023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896582370827128e-05,
+      "loss": 0.9756,
+      "step": 1024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963417721755052e-05,
+      "loss": 1.0012,
+      "step": 1025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961009092729598e-05,
+      "loss": 1.0186,
+      "step": 1026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958597821905004e-05,
+      "loss": 0.9967,
+      "step": 1027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956183909992144e-05,
+      "loss": 0.9859,
+      "step": 1028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953767357702655e-05,
+      "loss": 0.3478,
+      "step": 1029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951348165748973e-05,
+      "loss": 0.9693,
+      "step": 1030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894892633484429e-05,
+      "loss": 1.0145,
+      "step": 1031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89465018657026e-05,
+      "loss": 0.9428,
+      "step": 1032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944074759038648e-05,
+      "loss": 1.0153,
+      "step": 1033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894164501556798e-05,
+      "loss": 1.0287,
+      "step": 1034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939212636006902e-05,
+      "loss": 0.9632,
+      "step": 1035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936777621072514e-05,
+      "loss": 0.953,
+      "step": 1036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934339971482676e-05,
+      "loss": 0.9859,
+      "step": 1037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931899687956038e-05,
+      "loss": 0.9483,
+      "step": 1038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892945677121202e-05,
+      "loss": 1.0184,
+      "step": 1039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892701122197082e-05,
+      "loss": 0.9609,
+      "step": 1040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924563040953403e-05,
+      "loss": 1.0094,
+      "step": 1041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8922112228881532e-05,
+      "loss": 0.9793,
+      "step": 1042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891965878647772e-05,
+      "loss": 0.9323,
+      "step": 1043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8917202714465275e-05,
+      "loss": 0.9414,
+      "step": 1044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914744013568273e-05,
+      "loss": 1.0091,
+      "step": 1045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891228268451156e-05,
+      "loss": 1.0217,
+      "step": 1046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909818728020765e-05,
+      "loss": 1.0197,
+      "step": 1047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907352144822285e-05,
+      "loss": 0.9881,
+      "step": 1048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904882935643293e-05,
+      "loss": 0.9708,
+      "step": 1049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902411101211747e-05,
+      "loss": 0.3314,
+      "step": 1050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889993664225636e-05,
+      "loss": 0.9396,
+      "step": 1051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897459559506632e-05,
+      "loss": 0.9854,
+      "step": 1052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894979853692836e-05,
+      "loss": 0.9372,
+      "step": 1053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889249752554601e-05,
+      "loss": 1.0329,
+      "step": 1054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889001257579797e-05,
+      "loss": 0.9695,
+      "step": 1055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888752500518131e-05,
+      "loss": 0.9402,
+      "step": 1056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888503481442939e-05,
+      "loss": 0.9693,
+      "step": 1057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8882542004276343e-05,
+      "loss": 0.3051,
+      "step": 1058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880046575457072e-05,
+      "loss": 1.0506,
+      "step": 1059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877548528707267e-05,
+      "loss": 0.9983,
+      "step": 1060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875047864763366e-05,
+      "loss": 1.0086,
+      "step": 1061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88725445843626e-05,
+      "loss": 0.9607,
+      "step": 1062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887003868824295e-05,
+      "loss": 0.9428,
+      "step": 1063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867530177143192e-05,
+      "loss": 0.9446,
+      "step": 1064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886501905180286e-05,
+      "loss": 0.9358,
+      "step": 1065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862505312962257e-05,
+      "loss": 0.9965,
+      "step": 1066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859988961362455e-05,
+      "loss": 0.9442,
+      "step": 1067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885746999774531e-05,
+      "loss": 0.9833,
+      "step": 1068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854948422853436e-05,
+      "loss": 0.9434,
+      "step": 1069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8852424237430215e-05,
+      "loss": 0.3524,
+      "step": 1070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884989744221981e-05,
+      "loss": 0.9363,
+      "step": 1071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847368037967138e-05,
+      "loss": 0.9963,
+      "step": 1072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8844836025417905e-05,
+      "loss": 1.0056,
+      "step": 1073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842301405318567e-05,
+      "loss": 0.9756,
+      "step": 1074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8839764178416354e-05,
+      "loss": 0.9927,
+      "step": 1075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837224345459276e-05,
+      "loss": 1.0286,
+      "step": 1076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834681907196094e-05,
+      "loss": 0.3742,
+      "step": 1077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883213686437635e-05,
+      "loss": 0.9803,
+      "step": 1078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882958921775035e-05,
+      "loss": 0.939,
+      "step": 1079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827038968069163e-05,
+      "loss": 0.9506,
+      "step": 1080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824486116084633e-05,
+      "loss": 0.9882,
+      "step": 1081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821930662549363e-05,
+      "loss": 0.9504,
+      "step": 1082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881937260821673e-05,
+      "loss": 1.0322,
+      "step": 1083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816811953840877e-05,
+      "loss": 1.0129,
+      "step": 1084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881424870017671e-05,
+      "loss": 1.0303,
+      "step": 1085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811682847979902e-05,
+      "loss": 1.0114,
+      "step": 1086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8809114398006895e-05,
+      "loss": 1.0081,
+      "step": 1087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806543351014893e-05,
+      "loss": 0.3205,
+      "step": 1088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803969707761866e-05,
+      "loss": 1.0217,
+      "step": 1089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801393469006558e-05,
+      "loss": 0.9843,
+      "step": 1090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879881463550846e-05,
+      "loss": 0.9334,
+      "step": 1091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879623320802785e-05,
+      "loss": 1.0211,
+      "step": 1092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793649187325754e-05,
+      "loss": 0.9966,
+      "step": 1093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879106257416397e-05,
+      "loss": 0.9717,
+      "step": 1094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788473369305058e-05,
+      "loss": 0.996,
+      "step": 1095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785881573512345e-05,
+      "loss": 1.0669,
+      "step": 1096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783287187549915e-05,
+      "loss": 0.9869,
+      "step": 1097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878069021218262e-05,
+      "loss": 0.9617,
+      "step": 1098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877809064817608e-05,
+      "loss": 0.9576,
+      "step": 1099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775488496296672e-05,
+      "loss": 0.977,
+      "step": 1100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877288375731154e-05,
+      "loss": 0.9904,
+      "step": 1101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770276431988582e-05,
+      "loss": 1.049,
+      "step": 1102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876766652109647e-05,
+      "loss": 0.963,
+      "step": 1103
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765054025404627e-05,
+      "loss": 0.997,
+      "step": 1104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762438945683252e-05,
+      "loss": 1.0017,
+      "step": 1105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875982128270329e-05,
+      "loss": 0.9832,
+      "step": 1106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875720103723646e-05,
+      "loss": 0.9542,
+      "step": 1107
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875457821005524e-05,
+      "loss": 0.9662,
+      "step": 1108
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751952801932866e-05,
+      "loss": 0.9121,
+      "step": 1109
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749324813643328e-05,
+      "loss": 0.9428,
+      "step": 1110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746694245961395e-05,
+      "loss": 1.0123,
+      "step": 1111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874406109966258e-05,
+      "loss": 1.0193,
+      "step": 1112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874142537552316e-05,
+      "loss": 0.9992,
+      "step": 1113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873878707432018e-05,
+      "loss": 1.0042,
+      "step": 1114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736146196831433e-05,
+      "loss": 1.0293,
+      "step": 1115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733502743835483e-05,
+      "loss": 0.9998,
+      "step": 1116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730856716111642e-05,
+      "loss": 1.0046,
+      "step": 1117
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728208114439992e-05,
+      "loss": 0.3424,
+      "step": 1118
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725556939601365e-05,
+      "loss": 0.9906,
+      "step": 1119
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722903192377355e-05,
+      "loss": 0.9319,
+      "step": 1120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872024687355032e-05,
+      "loss": 0.9328,
+      "step": 1121
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871758798390336e-05,
+      "loss": 0.9335,
+      "step": 1122
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871492652422035e-05,
+      "loss": 0.9186,
+      "step": 1123
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712262495285917e-05,
+      "loss": 1.0427,
+      "step": 1124
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870959589788544e-05,
+      "loss": 0.9567,
+      "step": 1125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706926732805065e-05,
+      "loss": 1.0185,
+      "step": 1126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704255000831688e-05,
+      "loss": 0.9529,
+      "step": 1127
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701580702752963e-05,
+      "loss": 0.9834,
+      "step": 1128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698903839357304e-05,
+      "loss": 0.9333,
+      "step": 1129
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869622441143387e-05,
+      "loss": 0.9091,
+      "step": 1130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693542419772593e-05,
+      "loss": 1.0085,
+      "step": 1131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690857865164145e-05,
+      "loss": 1.0083,
+      "step": 1132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868817074839997e-05,
+      "loss": 1.0375,
+      "step": 1133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868548107027225e-05,
+      "loss": 0.9291,
+      "step": 1134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682788831573932e-05,
+      "loss": 1.0179,
+      "step": 1135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680094033098718e-05,
+      "loss": 1.039,
+      "step": 1136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8677396675641062e-05,
+      "loss": 0.9735,
+      "step": 1137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674696759996173e-05,
+      "loss": 0.8872,
+      "step": 1138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671994286960014e-05,
+      "loss": 1.0014,
+      "step": 1139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669289257329305e-05,
+      "loss": 0.9993,
+      "step": 1140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8666581671901513e-05,
+      "loss": 0.9696,
+      "step": 1141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866387153147486e-05,
+      "loss": 0.9943,
+      "step": 1142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8661158836848333e-05,
+      "loss": 0.99,
+      "step": 1143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658443588821657e-05,
+      "loss": 0.9628,
+      "step": 1144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655725788195315e-05,
+      "loss": 0.9624,
+      "step": 1145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653005435770546e-05,
+      "loss": 0.9835,
+      "step": 1146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650282532349332e-05,
+      "loss": 1.0539,
+      "step": 1147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647557078734423e-05,
+      "loss": 0.9636,
+      "step": 1148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644829075729303e-05,
+      "loss": 0.9944,
+      "step": 1149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864209852413822e-05,
+      "loss": 1.0179,
+      "step": 1150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863936542476617e-05,
+      "loss": 0.9946,
+      "step": 1151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636629778418894e-05,
+      "loss": 0.9957,
+      "step": 1152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863389158590289e-05,
+      "loss": 0.9517,
+      "step": 1153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631150848025414e-05,
+      "loss": 0.9281,
+      "step": 1154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862840756559446e-05,
+      "loss": 0.9617,
+      "step": 1155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625661739418767e-05,
+      "loss": 1.048,
+      "step": 1156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622913370307846e-05,
+      "loss": 1.0628,
+      "step": 1157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8620162459071936e-05,
+      "loss": 0.9671,
+      "step": 1158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617409006522042e-05,
+      "loss": 0.9365,
+      "step": 1159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614653013469905e-05,
+      "loss": 0.9716,
+      "step": 1160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861189448072802e-05,
+      "loss": 0.9755,
+      "step": 1161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609133409109637e-05,
+      "loss": 1.0009,
+      "step": 1162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8606369799428744e-05,
+      "loss": 0.9219,
+      "step": 1163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8603603652500085e-05,
+      "loss": 0.9437,
+      "step": 1164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860083496913915e-05,
+      "loss": 1.0225,
+      "step": 1165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598063750162166e-05,
+      "loss": 0.9976,
+      "step": 1166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859528999638613e-05,
+      "loss": 0.8956,
+      "step": 1167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592513708628767e-05,
+      "loss": 0.9436,
+      "step": 1168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589734887708556e-05,
+      "loss": 0.3826,
+      "step": 1169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8586953534444726e-05,
+      "loss": 0.9986,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584169649657244e-05,
+      "loss": 0.8879,
+      "step": 1171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8581383234166834e-05,
+      "loss": 0.9544,
+      "step": 1172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857859428879495e-05,
+      "loss": 0.9635,
+      "step": 1173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575802814363816e-05,
+      "loss": 0.9867,
+      "step": 1174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573008811696378e-05,
+      "loss": 0.9904,
+      "step": 1175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857021228161634e-05,
+      "loss": 0.979,
+      "step": 1176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856741322494815e-05,
+      "loss": 0.9755,
+      "step": 1177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564611642517e-05,
+      "loss": 0.9955,
+      "step": 1178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561807535148818e-05,
+      "loss": 1.034,
+      "step": 1179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559000903670293e-05,
+      "loss": 1.0103,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556191748908845e-05,
+      "loss": 0.9783,
+      "step": 1181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8553380071692645e-05,
+      "loss": 1.01,
+      "step": 1182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550565872850602e-05,
+      "loss": 1.0146,
+      "step": 1183
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547749153212373e-05,
+      "loss": 0.974,
+      "step": 1184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544929913608353e-05,
+      "loss": 0.9829,
+      "step": 1185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542108154869686e-05,
+      "loss": 0.9724,
+      "step": 1186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539283877828257e-05,
+      "loss": 0.9399,
+      "step": 1187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536457083316692e-05,
+      "loss": 0.9522,
+      "step": 1188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8533627772168362e-05,
+      "loss": 0.9535,
+      "step": 1189
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853079594521737e-05,
+      "loss": 0.9533,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527961603298572e-05,
+      "loss": 0.9894,
+      "step": 1191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525124747247566e-05,
+      "loss": 0.3953,
+      "step": 1192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852228537790068e-05,
+      "loss": 0.9874,
+      "step": 1193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851944349609499e-05,
+      "loss": 1.0302,
+      "step": 1194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516599102668324e-05,
+      "loss": 0.9583,
+      "step": 1195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513752198459224e-05,
+      "loss": 0.9583,
+      "step": 1196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851090278430699e-05,
+      "loss": 0.9739,
+      "step": 1197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850805086105167e-05,
+      "loss": 1.0088,
+      "step": 1198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850519642953403e-05,
+      "loss": 1.0248,
+      "step": 1199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850233949059559e-05,
+      "loss": 0.9533,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499480045078602e-05,
+      "loss": 0.9942,
+      "step": 1201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8496618093826064e-05,
+      "loss": 0.3602,
+      "step": 1202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849375363768171e-05,
+      "loss": 1.0329,
+      "step": 1203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490886677490007e-05,
+      "loss": 0.9469,
+      "step": 1204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488017214096173e-05,
+      "loss": 1.0235,
+      "step": 1205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485145248346147e-05,
+      "loss": 0.9493,
+      "step": 1206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848227078108662e-05,
+      "loss": 0.958,
+      "step": 1207
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847939381316501e-05,
+      "loss": 0.9575,
+      "step": 1208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476514345429485e-05,
+      "loss": 1.0123,
+      "step": 1209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473632378728932e-05,
+      "loss": 1.0039,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470747913912993e-05,
+      "loss": 1.018,
+      "step": 1211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467860951832035e-05,
+      "loss": 0.9605,
+      "step": 1212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464971493337167e-05,
+      "loss": 0.9897,
+      "step": 1213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462079539280233e-05,
+      "loss": 0.951,
+      "step": 1214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459185090513802e-05,
+      "loss": 0.9572,
+      "step": 1215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8456288147891196e-05,
+      "loss": 0.9909,
+      "step": 1216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453388712266464e-05,
+      "loss": 0.9605,
+      "step": 1217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450486784494384e-05,
+      "loss": 0.978,
+      "step": 1218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844758236543048e-05,
+      "loss": 0.9735,
+      "step": 1219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444675455931006e-05,
+      "loss": 0.9986,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441766056852947e-05,
+      "loss": 0.9732,
+      "step": 1221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438854169054022e-05,
+      "loss": 0.9316,
+      "step": 1222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435939793392686e-05,
+      "loss": 0.96,
+      "step": 1223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433022930728132e-05,
+      "loss": 0.9164,
+      "step": 1224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430103581920278e-05,
+      "loss": 0.9345,
+      "step": 1225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842718174782978e-05,
+      "loss": 0.9485,
+      "step": 1226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8424257429318027e-05,
+      "loss": 0.9401,
+      "step": 1227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421330627247137e-05,
+      "loss": 1.0052,
+      "step": 1228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841840134247996e-05,
+      "loss": 0.9294,
+      "step": 1229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8415469575880078e-05,
+      "loss": 0.9974,
+      "step": 1230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.9776,
+      "step": 1231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840959860064021e-05,
+      "loss": 1.0029,
+      "step": 1232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840665939373104e-05,
+      "loss": 0.996,
+      "step": 1233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403717708450823e-05,
+      "loss": 0.9609,
+      "step": 1234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400773545666788e-05,
+      "loss": 0.9621,
+      "step": 1235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397826906246913e-05,
+      "loss": 0.982,
+      "step": 1236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394877791059895e-05,
+      "loss": 0.9833,
+      "step": 1237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391926200975162e-05,
+      "loss": 1.0106,
+      "step": 1238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838897213686288e-05,
+      "loss": 0.93,
+      "step": 1239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838601559959393e-05,
+      "loss": 0.9391,
+      "step": 1240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8383056590039932e-05,
+      "loss": 0.9743,
+      "step": 1241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380095109073236e-05,
+      "loss": 1.0233,
+      "step": 1242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8377131157566917e-05,
+      "loss": 0.996,
+      "step": 1243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374164736394777e-05,
+      "loss": 0.8958,
+      "step": 1244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371195846431355e-05,
+      "loss": 0.9439,
+      "step": 1245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368224488551898e-05,
+      "loss": 0.9664,
+      "step": 1246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83652506636324e-05,
+      "loss": 0.3604,
+      "step": 1247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362274372549577e-05,
+      "loss": 0.9639,
+      "step": 1248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835929561618087e-05,
+      "loss": 0.9559,
+      "step": 1249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835631439540445e-05,
+      "loss": 0.9131,
+      "step": 1250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83533307110992e-05,
+      "loss": 0.9813,
+      "step": 1251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835034456414476e-05,
+      "loss": 0.969,
+      "step": 1252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8347355955421462e-05,
+      "loss": 0.8885,
+      "step": 1253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8344364885810388e-05,
+      "loss": 0.9625,
+      "step": 1254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341371356193334e-05,
+      "loss": 0.9515,
+      "step": 1255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338375367452822e-05,
+      "loss": 0.9895,
+      "step": 1256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8335376920472098e-05,
+      "loss": 1.0174,
+      "step": 1257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332376016135146e-05,
+      "loss": 0.957,
+      "step": 1258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832937265532665e-05,
+      "loss": 1.0354,
+      "step": 1259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8326366838932048e-05,
+      "loss": 1.0098,
+      "step": 1260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8323358567837474e-05,
+      "loss": 0.9979,
+      "step": 1261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83203478429298e-05,
+      "loss": 0.91,
+      "step": 1262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831733466509662e-05,
+      "loss": 0.988,
+      "step": 1263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314319035226254e-05,
+      "loss": 0.9491,
+      "step": 1264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311300954207737e-05,
+      "loss": 0.9889,
+      "step": 1265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308280422930832e-05,
+      "loss": 0.9686,
+      "step": 1266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830525744228602e-05,
+      "loss": 1.0163,
+      "step": 1267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302232013164518e-05,
+      "loss": 1.0825,
+      "step": 1268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829920413645824e-05,
+      "loss": 0.9335,
+      "step": 1269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296173813059844e-05,
+      "loss": 0.8598,
+      "step": 1270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82931410438627e-05,
+      "loss": 0.9452,
+      "step": 1271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290105829760902e-05,
+      "loss": 0.9687,
+      "step": 1272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287068171649257e-05,
+      "loss": 1.0124,
+      "step": 1273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82840280704233e-05,
+      "loss": 0.9469,
+      "step": 1274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828098552697929e-05,
+      "loss": 0.9716,
+      "step": 1275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277940542214195e-05,
+      "loss": 0.9915,
+      "step": 1276
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827489311702571e-05,
+      "loss": 0.9451,
+      "step": 1277
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271843252312248e-05,
+      "loss": 0.9634,
+      "step": 1278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268790948972942e-05,
+      "loss": 0.3493,
+      "step": 1279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265736207907637e-05,
+      "loss": 1.0231,
+      "step": 1280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262679030016913e-05,
+      "loss": 0.9534,
+      "step": 1281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.825961941620205e-05,
+      "loss": 0.9779,
+      "step": 1282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256557367365057e-05,
+      "loss": 0.9432,
+      "step": 1283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253492884408658e-05,
+      "loss": 0.8975,
+      "step": 1284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825042596823629e-05,
+      "loss": 0.9529,
+      "step": 1285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824735661975212e-05,
+      "loss": 0.9617,
+      "step": 1286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824428483986102e-05,
+      "loss": 0.9634,
+      "step": 1287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241210629468586e-05,
+      "loss": 0.9904,
+      "step": 1288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823813398948112e-05,
+      "loss": 1.0188,
+      "step": 1289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235054920805653e-05,
+      "loss": 0.9324,
+      "step": 1290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231973424349924e-05,
+      "loss": 0.9396,
+      "step": 1291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228889501022395e-05,
+      "loss": 0.9442,
+      "step": 1292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225803151732236e-05,
+      "loss": 0.9794,
+      "step": 1293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822271437738933e-05,
+      "loss": 1.0204,
+      "step": 1294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821962317890429e-05,
+      "loss": 0.9436,
+      "step": 1295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8216529557188424e-05,
+      "loss": 0.9828,
+      "step": 1296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821343351315377e-05,
+      "loss": 0.9462,
+      "step": 1297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8210335047713074e-05,
+      "loss": 0.8913,
+      "step": 1298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207234161779793e-05,
+      "loss": 0.9707,
+      "step": 1299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82041308562681e-05,
+      "loss": 0.9658,
+      "step": 1300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820102513209289e-05,
+      "loss": 0.9785,
+      "step": 1301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819791699016975e-05,
+      "loss": 1.0213,
+      "step": 1302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194806431415005e-05,
+      "loss": 1.0035,
+      "step": 1303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8191693456745673e-05,
+      "loss": 0.938,
+      "step": 1304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818857806707949e-05,
+      "loss": 1.029,
+      "step": 1305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185460263334914e-05,
+      "loss": 0.922,
+      "step": 1306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81823400464311e-05,
+      "loss": 0.9415,
+      "step": 1307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817921741728792e-05,
+      "loss": 0.9821,
+      "step": 1308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817609237682596e-05,
+      "loss": 0.3507,
+      "step": 1309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817296492596651e-05,
+      "loss": 0.9967,
+      "step": 1310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816983506563158e-05,
+      "loss": 0.9574,
+      "step": 1311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816670279674389e-05,
+      "loss": 0.9727,
+      "step": 1312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163568120226856e-05,
+      "loss": 0.9329,
+      "step": 1313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160431037004618e-05,
+      "loss": 0.9658,
+      "step": 1314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815729154800202e-05,
+      "loss": 0.3549,
+      "step": 1315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815414965414462e-05,
+      "loss": 0.9622,
+      "step": 1316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151005356358673e-05,
+      "loss": 1.0136,
+      "step": 1317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8147858655571156e-05,
+      "loss": 0.9385,
+      "step": 1318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814470955270975e-05,
+      "loss": 0.918,
+      "step": 1319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8141558048702844e-05,
+      "loss": 0.9565,
+      "step": 1320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813840414447953e-05,
+      "loss": 0.9748,
+      "step": 1321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8135247840969615e-05,
+      "loss": 1.0054,
+      "step": 1322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132089139103612e-05,
+      "loss": 0.961,
+      "step": 1323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8128928039812736e-05,
+      "loss": 0.9038,
+      "step": 1324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125764544028913e-05,
+      "loss": 0.9568,
+      "step": 1325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122598652684774e-05,
+      "loss": 0.9795,
+      "step": 1326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8119430366713658e-05,
+      "loss": 1.002,
+      "step": 1327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116259687049612e-05,
+      "loss": 0.9477,
+      "step": 1328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8113086614627377e-05,
+      "loss": 0.9465,
+      "step": 1329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8109911150382417e-05,
+      "loss": 0.9045,
+      "step": 1330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106733295250885e-05,
+      "loss": 0.9708,
+      "step": 1331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8103553050169652e-05,
+      "loss": 0.9414,
+      "step": 1332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810037041607628e-05,
+      "loss": 0.9644,
+      "step": 1333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809718539390905e-05,
+      "loss": 0.9477,
+      "step": 1334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093997984606936e-05,
+      "loss": 0.97,
+      "step": 1335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809080818910962e-05,
+      "loss": 1.0016,
+      "step": 1336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8087616008357488e-05,
+      "loss": 0.9578,
+      "step": 1337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8084421443291633e-05,
+      "loss": 0.9412,
+      "step": 1338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8081224494853834e-05,
+      "loss": 0.9971,
+      "step": 1339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078025163986595e-05,
+      "loss": 0.9379,
+      "step": 1340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807482345163311e-05,
+      "loss": 0.894,
+      "step": 1341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8071619358737274e-05,
+      "loss": 0.948,
+      "step": 1342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068412886243692e-05,
+      "loss": 1.0351,
+      "step": 1343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806520403509766e-05,
+      "loss": 0.9722,
+      "step": 1344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061992806245186e-05,
+      "loss": 0.9087,
+      "step": 1345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805877920063297e-05,
+      "loss": 0.9323,
+      "step": 1346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805556321920842e-05,
+      "loss": 0.9845,
+      "step": 1347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052344862919637e-05,
+      "loss": 0.9891,
+      "step": 1348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804912413271543e-05,
+      "loss": 0.9483,
+      "step": 1349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045901029545304e-05,
+      "loss": 0.9656,
+      "step": 1350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8042675554359456e-05,
+      "loss": 0.9399,
+      "step": 1351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80394477081088e-05,
+      "loss": 0.9908,
+      "step": 1352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036217491744935e-05,
+      "loss": 0.36,
+      "step": 1353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803298490622016e-05,
+      "loss": 0.9831,
+      "step": 1354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029749952487474e-05,
+      "loss": 0.9578,
+      "step": 1355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8026512631500583e-05,
+      "loss": 0.9787,
+      "step": 1356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8023272944213875e-05,
+      "loss": 0.9668,
+      "step": 1357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802003089158245e-05,
+      "loss": 0.8953,
+      "step": 1358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016786474562093e-05,
+      "loss": 0.9142,
+      "step": 1359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013539694109293e-05,
+      "loss": 0.9472,
+      "step": 1360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8010290551181237e-05,
+      "loss": 0.9636,
+      "step": 1361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007039046735807e-05,
+      "loss": 0.948,
+      "step": 1362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800378518173158e-05,
+      "loss": 0.9708,
+      "step": 1363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8000528957127828e-05,
+      "loss": 0.9859,
+      "step": 1364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997270373884523e-05,
+      "loss": 0.9618,
+      "step": 1365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994009432962326e-05,
+      "loss": 1.0097,
+      "step": 1366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7990746135322593e-05,
+      "loss": 0.9195,
+      "step": 1367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798748048192739e-05,
+      "loss": 1.0004,
+      "step": 1368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798421247373945e-05,
+      "loss": 0.9393,
+      "step": 1369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980942111722234e-05,
+      "loss": 0.9771,
+      "step": 1370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977669396839866e-05,
+      "loss": 0.9909,
+      "step": 1371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797439433005718e-05,
+      "loss": 1.0087,
+      "step": 1372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971116912339698e-05,
+      "loss": 0.9926,
+      "step": 1373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967837144653643e-05,
+      "loss": 0.3581,
+      "step": 1374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796455502796592e-05,
+      "loss": 1.0291,
+      "step": 1375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796127056324413e-05,
+      "loss": 0.9419,
+      "step": 1376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795798375145657e-05,
+      "loss": 0.9027,
+      "step": 1377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7954694593572225e-05,
+      "loss": 0.9643,
+      "step": 1378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795140309056078e-05,
+      "loss": 0.9529,
+      "step": 1379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794810924339259e-05,
+      "loss": 0.3777,
+      "step": 1380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944813053038733e-05,
+      "loss": 0.9297,
+      "step": 1381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941514520470944e-05,
+      "loss": 0.9596,
+      "step": 1382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793821364666168e-05,
+      "loss": 0.9656,
+      "step": 1383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934910432584057e-05,
+      "loss": 0.9716,
+      "step": 1384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793160487921191e-05,
+      "loss": 1.007,
+      "step": 1385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792829698751975e-05,
+      "loss": 0.9547,
+      "step": 1386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7924986758482768e-05,
+      "loss": 0.9876,
+      "step": 1387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7921674193076865e-05,
+      "loss": 0.9858,
+      "step": 1388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918359292278614e-05,
+      "loss": 0.9764,
+      "step": 1389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915042057065283e-05,
+      "loss": 0.9861,
+      "step": 1390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791172248841483e-05,
+      "loss": 0.9239,
+      "step": 1391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908400587305896e-05,
+      "loss": 0.9085,
+      "step": 1392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790507635471781e-05,
+      "loss": 0.9532,
+      "step": 1393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79017497916306e-05,
+      "loss": 0.929,
+      "step": 1394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789842089902496e-05,
+      "loss": 0.9725,
+      "step": 1395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895089677882288e-05,
+      "loss": 0.9653,
+      "step": 1396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7891756129184664e-05,
+      "loss": 0.9707,
+      "step": 1397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7888420253914854e-05,
+      "loss": 0.9656,
+      "step": 1398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.78850820530563e-05,
+      "loss": 0.9246,
+      "step": 1399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788174152759315e-05,
+      "loss": 0.9846,
+      "step": 1400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787839867851022e-05,
+      "loss": 1.0135,
+      "step": 1401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7875053506793015e-05,
+      "loss": 0.9185,
+      "step": 1402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871706013427732e-05,
+      "loss": 0.9743,
+      "step": 1403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868356199401242e-05,
+      "loss": 0.3534,
+      "step": 1404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865004065701107e-05,
+      "loss": 0.9927,
+      "step": 1405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786164961331557e-05,
+      "loss": 0.98,
+      "step": 1406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7858292843233564e-05,
+      "loss": 1.0226,
+      "step": 1407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7854933756444692e-05,
+      "loss": 0.9979,
+      "step": 1408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7851572353939253e-05,
+      "loss": 0.9408,
+      "step": 1409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784820863670822e-05,
+      "loss": 1.0039,
+      "step": 1410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844842605743256e-05,
+      "loss": 0.9722,
+      "step": 1411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78414742620367e-05,
+      "loss": 0.3633,
+      "step": 1412
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783810360658158e-05,
+      "loss": 0.9384,
+      "step": 1413
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783473064037159e-05,
+      "loss": 0.9721,
+      "step": 1414
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831355364401123e-05,
+      "loss": 0.9535,
+      "step": 1415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827977779665245e-05,
+      "loss": 0.9954,
+      "step": 1416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824597887159704e-05,
+      "loss": 0.9534,
+      "step": 1417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782121568788092e-05,
+      "loss": 1.0765,
+      "step": 1418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781783118282601e-05,
+      "loss": 0.8943,
+      "step": 1419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781444437299276e-05,
+      "loss": 0.9462,
+      "step": 1420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811055259379635e-05,
+      "loss": 1.0293,
+      "step": 1421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807663842985777e-05,
+      "loss": 0.3537,
+      "step": 1422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780427012481102e-05,
+      "loss": 0.945,
+      "step": 1423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800874105855862e-05,
+      "loss": 0.9088,
+      "step": 1424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797475787121485e-05,
+      "loss": 1.0192,
+      "step": 1425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7794075169609746e-05,
+      "loss": 0.9397,
+      "step": 1426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779067225432319e-05,
+      "loss": 0.9259,
+      "step": 1427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7787267042265028e-05,
+      "loss": 0.9184,
+      "step": 1428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778385953443915e-05,
+      "loss": 0.9871,
+      "step": 1429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778044973185013e-05,
+      "loss": 0.8936,
+      "step": 1430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777703763550321e-05,
+      "loss": 1.0389,
+      "step": 1431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773623246404312e-05,
+      "loss": 0.9231,
+      "step": 1432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770206565560034e-05,
+      "loss": 0.3514,
+      "step": 1433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7766787593977648e-05,
+      "loss": 0.9413,
+      "step": 1434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7763366332665106e-05,
+      "loss": 0.9253,
+      "step": 1435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7759942782631027e-05,
+      "loss": 0.942,
+      "step": 1436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756516944884713e-05,
+      "loss": 1.0019,
+      "step": 1437
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775308882043613e-05,
+      "loss": 0.9388,
+      "step": 1438
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7749658410295935e-05,
+      "loss": 0.9604,
+      "step": 1439
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746225715475444e-05,
+      "loss": 0.9799,
+      "step": 1440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742790736986652e-05,
+      "loss": 0.9406,
+      "step": 1441
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739353475842225e-05,
+      "loss": 0.9533,
+      "step": 1442
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735913933055506e-05,
+      "loss": 0.9465,
+      "step": 1443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732472109640504e-05,
+      "loss": 0.9853,
+      "step": 1444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729028006611908e-05,
+      "loss": 0.9302,
+      "step": 1445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7725581624985073e-05,
+      "loss": 0.9813,
+      "step": 1446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722132965776035e-05,
+      "loss": 0.9321,
+      "step": 1447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7718682030001485e-05,
+      "loss": 0.978,
+      "step": 1448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77152288186788e-05,
+      "loss": 1.0304,
+      "step": 1449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711773332826023e-05,
+      "loss": 0.945,
+      "step": 1450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770831557346187e-05,
+      "loss": 1.0084,
+      "step": 1451
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7704855541605715e-05,
+      "loss": 0.9685,
+      "step": 1452
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7701393238277626e-05,
+      "loss": 1.027,
+      "step": 1453
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7697928664498308e-05,
+      "loss": 0.9741,
+      "step": 1454
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769446182128917e-05,
+      "loss": 0.9609,
+      "step": 1455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769099270967227e-05,
+      "loss": 0.9568,
+      "step": 1456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687521330670333e-05,
+      "loss": 0.9866,
+      "step": 1457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768404768530676e-05,
+      "loss": 0.971,
+      "step": 1458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768057177460562e-05,
+      "loss": 0.9849,
+      "step": 1459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7677093599591643e-05,
+      "loss": 1.0078,
+      "step": 1460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7673613161290237e-05,
+      "loss": 1.0142,
+      "step": 1461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7670130460727465e-05,
+      "loss": 0.9988,
+      "step": 1462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7666645498930074e-05,
+      "loss": 0.9576,
+      "step": 1463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663158276925458e-05,
+      "loss": 0.995,
+      "step": 1464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765966879574169e-05,
+      "loss": 0.9757,
+      "step": 1465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7656177056407508e-05,
+      "loss": 0.9653,
+      "step": 1466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7652683059952306e-05,
+      "loss": 0.9941,
+      "step": 1467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7649186807406153e-05,
+      "loss": 0.966,
+      "step": 1468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645688299799787e-05,
+      "loss": 0.9669,
+      "step": 1469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642187538164595e-05,
+      "loss": 0.9751,
+      "step": 1470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763868452353265e-05,
+      "loss": 1.008,
+      "step": 1471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763517925693667e-05,
+      "loss": 0.9806,
+      "step": 1472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7631671739410042e-05,
+      "loss": 0.9733,
+      "step": 1473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7628161971986822e-05,
+      "loss": 0.9582,
+      "step": 1474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7624649955701726e-05,
+      "loss": 0.9172,
+      "step": 1475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7621135691590132e-05,
+      "loss": 0.9108,
+      "step": 1476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7617619180688087e-05,
+      "loss": 0.9838,
+      "step": 1477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7614100424032284e-05,
+      "loss": 1.0691,
+      "step": 1478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7610579422660103e-05,
+      "loss": 1.0068,
+      "step": 1479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760705617760956e-05,
+      "loss": 0.9162,
+      "step": 1480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760353068991935e-05,
+      "loss": 1.0248,
+      "step": 1481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7600002960628827e-05,
+      "loss": 1.0123,
+      "step": 1482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7596472990777996e-05,
+      "loss": 0.929,
+      "step": 1483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7592940781407533e-05,
+      "loss": 0.9725,
+      "step": 1484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758940633355877e-05,
+      "loss": 0.9396,
+      "step": 1485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7585869648273697e-05,
+      "loss": 0.9962,
+      "step": 1486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582330726594964e-05,
+      "loss": 1.0007,
+      "step": 1487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757878956956589e-05,
+      "loss": 0.9841,
+      "step": 1488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757524617823044e-05,
+      "loss": 1.0183,
+      "step": 1489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571700553633236e-05,
+      "loss": 0.9749,
+      "step": 1490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756815269681958e-05,
+      "loss": 0.985,
+      "step": 1491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564602608835407e-05,
+      "loss": 0.9864,
+      "step": 1492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7561050290727322e-05,
+      "loss": 0.9793,
+      "step": 1493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7557495743542586e-05,
+      "loss": 1.0371,
+      "step": 1494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553938968329114e-05,
+      "loss": 1.0098,
+      "step": 1495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550379966135486e-05,
+      "loss": 0.9913,
+      "step": 1496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754681873801093e-05,
+      "loss": 0.9899,
+      "step": 1497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543255285005335e-05,
+      "loss": 0.9915,
+      "step": 1498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753968960816924e-05,
+      "loss": 0.3454,
+      "step": 1499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536121708553845e-05,
+      "loss": 0.9184,
+      "step": 1500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7532551587211006e-05,
+      "loss": 0.9724,
+      "step": 1501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528979245193233e-05,
+      "loss": 0.9086,
+      "step": 1502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752540468355369e-05,
+      "loss": 0.9582,
+      "step": 1503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752182790334619e-05,
+      "loss": 0.9828,
+      "step": 1504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518248905625214e-05,
+      "loss": 1.0149,
+      "step": 1505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7514667691445873e-05,
+      "loss": 0.9615,
+      "step": 1506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7511084261863965e-05,
+      "loss": 1.0001,
+      "step": 1507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750749861793591e-05,
+      "loss": 0.9213,
+      "step": 1508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503910760718797e-05,
+      "loss": 0.964,
+      "step": 1509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7500320691270365e-05,
+      "loss": 0.9518,
+      "step": 1510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7496728410649e-05,
+      "loss": 0.9949,
+      "step": 1511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749313391991375e-05,
+      "loss": 0.9772,
+      "step": 1512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74895372201243e-05,
+      "loss": 0.9434,
+      "step": 1513
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485938312341003e-05,
+      "loss": 0.9172,
+      "step": 1514
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482337197624853e-05,
+      "loss": 0.9773,
+      "step": 1515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747873387703749e-05,
+      "loss": 1.0602,
+      "step": 1516
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7475128351641216e-05,
+      "loss": 0.9993,
+      "step": 1517
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7471520622498982e-05,
+      "loss": 0.374,
+      "step": 1518
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467910690674372e-05,
+      "loss": 1.0166,
+      "step": 1519
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7464298557231642e-05,
+      "loss": 0.9271,
+      "step": 1520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746068422323568e-05,
+      "loss": 0.3588,
+      "step": 1521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457067689752033e-05,
+      "loss": 0.3371,
+      "step": 1522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7453448957846896e-05,
+      "loss": 0.9935,
+      "step": 1523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7449828028587105e-05,
+      "loss": 0.9906,
+      "step": 1524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7446204903040148e-05,
+      "loss": 0.9644,
+      "step": 1525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744257958227416e-05,
+      "loss": 0.9089,
+      "step": 1526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743895206735792e-05,
+      "loss": 0.3412,
+      "step": 1527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7435322359360866e-05,
+      "loss": 0.9591,
+      "step": 1528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743169045935307e-05,
+      "loss": 0.9573,
+      "step": 1529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428056368405247e-05,
+      "loss": 0.9361,
+      "step": 1530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7424420087588777e-05,
+      "loss": 1.0099,
+      "step": 1531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7420781617975667e-05,
+      "loss": 0.9612,
+      "step": 1532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417140960638574e-05,
+      "loss": 0.9525,
+      "step": 1533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7413498116650806e-05,
+      "loss": 0.9851,
+      "step": 1534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7409853087086312e-05,
+      "loss": 0.9694,
+      "step": 1535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406205873019684e-05,
+      "loss": 0.9144,
+      "step": 1536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402556475526155e-05,
+      "loss": 0.9844,
+      "step": 1537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739890489568161e-05,
+      "loss": 0.9599,
+      "step": 1538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7395251134562566e-05,
+      "loss": 0.9297,
+      "step": 1539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7391595193246197e-05,
+      "loss": 0.9972,
+      "step": 1540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7387937072810313e-05,
+      "loss": 0.8822,
+      "step": 1541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384276774333363e-05,
+      "loss": 0.9451,
+      "step": 1542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7380614298894443e-05,
+      "loss": 0.3437,
+      "step": 1543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7376949647573287e-05,
+      "loss": 0.9017,
+      "step": 1544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737328282145027e-05,
+      "loss": 0.975,
+      "step": 1545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736961382160642e-05,
+      "loss": 0.3593,
+      "step": 1546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736594264912339e-05,
+      "loss": 0.9139,
+      "step": 1547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736226930508348e-05,
+      "loss": 0.9732,
+      "step": 1548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735859379056963e-05,
+      "loss": 0.9338,
+      "step": 1549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7354916106665422e-05,
+      "loss": 0.9211,
+      "step": 1550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7351236254455077e-05,
+      "loss": 0.9658,
+      "step": 1551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7347554235023447e-05,
+      "loss": 0.9745,
+      "step": 1552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734387004945604e-05,
+      "loss": 0.9691,
+      "step": 1553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734018369883898e-05,
+      "loss": 0.8937,
+      "step": 1554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336495184259057e-05,
+      "loss": 0.9597,
+      "step": 1555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733280450680367e-05,
+      "loss": 1.003,
+      "step": 1556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329111667560875e-05,
+      "loss": 0.9012,
+      "step": 1557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732541666761936e-05,
+      "loss": 0.9645,
+      "step": 1558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732171950806845e-05,
+      "loss": 0.9214,
+      "step": 1559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7318020189998103e-05,
+      "loss": 0.9487,
+      "step": 1560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314318714498922e-05,
+      "loss": 0.9985,
+      "step": 1561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7310615082662133e-05,
+      "loss": 0.9517,
+      "step": 1562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730690929557961e-05,
+      "loss": 0.8971,
+      "step": 1563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730320135434386e-05,
+      "loss": 0.9942,
+      "step": 1564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729949126004802e-05,
+      "loss": 0.9574,
+      "step": 1565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7295779013785865e-05,
+      "loss": 0.9676,
+      "step": 1566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72920646166518e-05,
+      "loss": 0.997,
+      "step": 1567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7288348069740878e-05,
+      "loss": 1.0261,
+      "step": 1568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7284629374148764e-05,
+      "loss": 0.9811,
+      "step": 1569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728090853097178e-05,
+      "loss": 0.942,
+      "step": 1570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727718554130686e-05,
+      "loss": 0.9684,
+      "step": 1571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273460406251584e-05,
+      "loss": 0.9522,
+      "step": 1572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7269733126904162e-05,
+      "loss": 0.9383,
+      "step": 1573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266003704363432e-05,
+      "loss": 0.9605,
+      "step": 1574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726227213972887e-05,
+      "loss": 0.9818,
+      "step": 1575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725853843410058e-05,
+      "loss": 0.984,
+      "step": 1576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725480258857929e-05,
+      "loss": 0.9654,
+      "step": 1577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7251064604266376e-05,
+      "loss": 0.9579,
+      "step": 1578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7247324482263832e-05,
+      "loss": 0.8993,
+      "step": 1579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243582223674286e-05,
+      "loss": 0.9519,
+      "step": 1580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723983782960099e-05,
+      "loss": 0.9682,
+      "step": 1581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236091301147834e-05,
+      "loss": 0.9424,
+      "step": 1582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7232342639419333e-05,
+      "loss": 0.9582,
+      "step": 1583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7228591845520633e-05,
+      "loss": 0.9727,
+      "step": 1584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7224838920557506e-05,
+      "loss": 0.3763,
+      "step": 1585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7221083865636356e-05,
+      "loss": 0.3552,
+      "step": 1586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721732668186421e-05,
+      "loss": 0.9481,
+      "step": 1587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721356737034872e-05,
+      "loss": 0.9683,
+      "step": 1588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720980593219818e-05,
+      "loss": 0.9672,
+      "step": 1589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7206042368521493e-05,
+      "loss": 1.0087,
+      "step": 1590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7202276680428208e-05,
+      "loss": 0.9538,
+      "step": 1591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7198508869028475e-05,
+      "loss": 0.9774,
+      "step": 1592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7194738935433094e-05,
+      "loss": 0.9867,
+      "step": 1593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719096688075348e-05,
+      "loss": 0.9636,
+      "step": 1594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718719270610167e-05,
+      "loss": 0.8982,
+      "step": 1595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183416412590334e-05,
+      "loss": 0.9768,
+      "step": 1596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717963800133276e-05,
+      "loss": 0.9742,
+      "step": 1597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175857473442864e-05,
+      "loss": 0.9961,
+      "step": 1598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7172074830035185e-05,
+      "loss": 0.3765,
+      "step": 1599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7168290072224886e-05,
+      "loss": 0.9647,
+      "step": 1600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7164503201127753e-05,
+      "loss": 0.9212,
+      "step": 1601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160714217860196e-05,
+      "loss": 0.9336,
+      "step": 1602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156923123539245e-05,
+      "loss": 0.9579,
+      "step": 1603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715312991928256e-05,
+      "loss": 0.9582,
+      "step": 1604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714933460620841e-05,
+      "loss": 0.9404,
+      "step": 1605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145537185435694e-05,
+      "loss": 0.9191,
+      "step": 1606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141737658083936e-05,
+      "loss": 1.0616,
+      "step": 1607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713793602527327e-05,
+      "loss": 0.977,
+      "step": 1608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7134132288124464e-05,
+      "loss": 0.8857,
+      "step": 1609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7130326447758898e-05,
+      "loss": 0.9351,
+      "step": 1610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7126518505298567e-05,
+      "loss": 0.9587,
+      "step": 1611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71227084618661e-05,
+      "loss": 0.9684,
+      "step": 1612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118896318584733e-05,
+      "loss": 0.9479,
+      "step": 1613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7115082076578327e-05,
+      "loss": 0.942,
+      "step": 1614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711126573697136e-05,
+      "loss": 0.9773,
+      "step": 1615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7107447300888932e-05,
+      "loss": 0.9203,
+      "step": 1616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710362676945675e-05,
+      "loss": 0.3345,
+      "step": 1617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709980414380116e-05,
+      "loss": 1.036,
+      "step": 1618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7095979425049098e-05,
+      "loss": 0.9632,
+      "step": 1619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7092152614328135e-05,
+      "loss": 0.9161,
+      "step": 1620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708832371276646e-05,
+      "loss": 0.9847,
+      "step": 1621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708449272149287e-05,
+      "loss": 0.9604,
+      "step": 1622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7080659641636784e-05,
+      "loss": 0.92,
+      "step": 1623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7076824474328226e-05,
+      "loss": 0.8939,
+      "step": 1624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707298722069785e-05,
+      "loss": 0.3272,
+      "step": 1625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706914788187692e-05,
+      "loss": 0.9382,
+      "step": 1626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065306458997305e-05,
+      "loss": 0.9698,
+      "step": 1627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061462953191504e-05,
+      "loss": 0.9096,
+      "step": 1628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7057617365592624e-05,
+      "loss": 1.0126,
+      "step": 1629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705376969733438e-05,
+      "loss": 0.9658,
+      "step": 1630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7049919949551103e-05,
+      "loss": 0.9435,
+      "step": 1631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704606812337774e-05,
+      "loss": 0.3357,
+      "step": 1632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7042214219949854e-05,
+      "loss": 0.9163,
+      "step": 1633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7038358240403615e-05,
+      "loss": 0.9625,
+      "step": 1634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70345001858758e-05,
+      "loss": 0.9129,
+      "step": 1635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7030640057503812e-05,
+      "loss": 0.9909,
+      "step": 1636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7026777856425653e-05,
+      "loss": 0.9609,
+      "step": 1637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702291358377994e-05,
+      "loss": 1.0422,
+      "step": 1638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019047240705902e-05,
+      "loss": 0.9601,
+      "step": 1639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701517882834337e-05,
+      "loss": 0.907,
+      "step": 1640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701130834783281e-05,
+      "loss": 0.9069,
+      "step": 1641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7007435800315263e-05,
+      "loss": 0.9344,
+      "step": 1642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7003561186932403e-05,
+      "loss": 1.017,
+      "step": 1643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699968450882651e-05,
+      "loss": 0.9432,
+      "step": 1644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6995805767140465e-05,
+      "loss": 0.9888,
+      "step": 1645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699192496301776e-05,
+      "loss": 0.8697,
+      "step": 1646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69880420976025e-05,
+      "loss": 0.9514,
+      "step": 1647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984157172039393e-05,
+      "loss": 1.0071,
+      "step": 1648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6980270187473757e-05,
+      "loss": 0.966,
+      "step": 1649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6976381145051513e-05,
+      "loss": 0.9326,
+      "step": 1650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697249004591919e-05,
+      "loss": 0.9445,
+      "step": 1651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696859689122393e-05,
+      "loss": 1.0231,
+      "step": 1652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6964701682113477e-05,
+      "loss": 0.9867,
+      "step": 1653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960804419736172e-05,
+      "loss": 0.949,
+      "step": 1654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695690510524097e-05,
+      "loss": 0.9522,
+      "step": 1655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6953003739777438e-05,
+      "loss": 0.9175,
+      "step": 1656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949100324495727e-05,
+      "loss": 0.9586,
+      "step": 1657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6945194860546614e-05,
+      "loss": 0.9428,
+      "step": 1658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941287349081466e-05,
+      "loss": 1.0129,
+      "step": 1659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6937377791252262e-05,
+      "loss": 1.0099,
+      "step": 1660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6933466188211575e-05,
+      "loss": 0.9204,
+      "step": 1661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6929552541112592e-05,
+      "loss": 1.0058,
+      "step": 1662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6925636851109095e-05,
+      "loss": 0.9402,
+      "step": 1663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692171911935547e-05,
+      "loss": 0.9328,
+      "step": 1664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69177993470067e-05,
+      "loss": 0.9079,
+      "step": 1665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6913877535218386e-05,
+      "loss": 0.9453,
+      "step": 1666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6909953685146713e-05,
+      "loss": 0.94,
+      "step": 1667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6906027797948473e-05,
+      "loss": 1.0224,
+      "step": 1668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902099874781058e-05,
+      "loss": 0.9641,
+      "step": 1669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689816991680246e-05,
+      "loss": 0.9681,
+      "step": 1670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689423792517128e-05,
+      "loss": 0.9808,
+      "step": 1671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6890303901046695e-05,
+      "loss": 0.9642,
+      "step": 1672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6886367845588507e-05,
+      "loss": 0.9921,
+      "step": 1673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688242975995711e-05,
+      "loss": 0.956,
+      "step": 1674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687848964531348e-05,
+      "loss": 0.9801,
+      "step": 1675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874547502819213e-05,
+      "loss": 0.9585,
+      "step": 1676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6870603333636495e-05,
+      "loss": 0.8713,
+      "step": 1677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6866657138928106e-05,
+      "loss": 0.9747,
+      "step": 1678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6862708919857417e-05,
+      "loss": 0.9787,
+      "step": 1679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685875867758842e-05,
+      "loss": 0.9887,
+      "step": 1680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6854806413285674e-05,
+      "loss": 0.9577,
+      "step": 1681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6850852128114357e-05,
+      "loss": 0.8925,
+      "step": 1682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6846895823240228e-05,
+      "loss": 0.9502,
+      "step": 1683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6842937499829652e-05,
+      "loss": 0.9605,
+      "step": 1684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838977159049575e-05,
+      "loss": 0.9004,
+      "step": 1685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683501480206756e-05,
+      "loss": 1.0072,
+      "step": 1686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683105043005174e-05,
+      "loss": 0.9037,
+      "step": 1687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827084044170857e-05,
+      "loss": 0.8911,
+      "step": 1688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823115645594244e-05,
+      "loss": 0.9011,
+      "step": 1689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681914523549183e-05,
+      "loss": 0.9616,
+      "step": 1690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6815172815034128e-05,
+      "loss": 0.9921,
+      "step": 1691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6811198385392246e-05,
+      "loss": 0.9484,
+      "step": 1692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6807221947737895e-05,
+      "loss": 0.8953,
+      "step": 1693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803243503243368e-05,
+      "loss": 0.9667,
+      "step": 1694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6799263053081548e-05,
+      "loss": 0.9856,
+      "step": 1695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6795280598425918e-05,
+      "loss": 0.9723,
+      "step": 1696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6791296140450547e-05,
+      "loss": 0.9898,
+      "step": 1697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6787309680330093e-05,
+      "loss": 0.9839,
+      "step": 1698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6783321219239808e-05,
+      "loss": 0.9403,
+      "step": 1699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.677933075835553e-05,
+      "loss": 0.9561,
+      "step": 1700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6775338298853687e-05,
+      "loss": 0.9461,
+      "step": 1701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6771343841911302e-05,
+      "loss": 0.9229,
+      "step": 1702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676734738870598e-05,
+      "loss": 0.9584,
+      "step": 1703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676334894041592e-05,
+      "loss": 0.9683,
+      "step": 1704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67593484982199e-05,
+      "loss": 0.9516,
+      "step": 1705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755346063297303e-05,
+      "loss": 1.0207,
+      "step": 1706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6751341636828076e-05,
+      "loss": 1.0289,
+      "step": 1707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6747335219992777e-05,
+      "loss": 0.9187,
+      "step": 1708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674332681397253e-05,
+      "loss": 1.0174,
+      "step": 1709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673931641994906e-05,
+      "loss": 0.8875,
+      "step": 1710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6735304039104675e-05,
+      "loss": 0.8931,
+      "step": 1711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731289672622263e-05,
+      "loss": 0.9773,
+      "step": 1712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6727273321685303e-05,
+      "loss": 0.9347,
+      "step": 1713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6723254987477858e-05,
+      "loss": 1.0346,
+      "step": 1714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671923467118457e-05,
+      "loss": 1.017,
+      "step": 1715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6715212373990676e-05,
+      "loss": 0.9319,
+      "step": 1716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711188097081987e-05,
+      "loss": 0.9164,
+      "step": 1717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6707161841644908e-05,
+      "loss": 0.966,
+      "step": 1718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703133608866415e-05,
+      "loss": 0.9559,
+      "step": 1719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699103399934076e-05,
+      "loss": 0.9051,
+      "step": 1720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6695071216036037e-05,
+      "loss": 0.9851,
+      "step": 1721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691037058361032e-05,
+      "loss": 0.9496,
+      "step": 1722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687000928098366e-05,
+      "loss": 0.994,
+      "step": 1723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668296282643794e-05,
+      "loss": 0.9455,
+      "step": 1724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667892275457022e-05,
+      "loss": 0.9694,
+      "step": 1725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667488071368627e-05,
+      "loss": 0.9679,
+      "step": 1726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667083670497772e-05,
+      "loss": 1.0566,
+      "step": 1727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666679072963679e-05,
+      "loss": 0.9542,
+      "step": 1728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6662742788856275e-05,
+      "loss": 0.898,
+      "step": 1729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6658692883829548e-05,
+      "loss": 0.9792,
+      "step": 1730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6654641015750564e-05,
+      "loss": 0.9149,
+      "step": 1731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665058718581386e-05,
+      "loss": 0.9495,
+      "step": 1732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6646531395214537e-05,
+      "loss": 0.9579,
+      "step": 1733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6642473645148297e-05,
+      "loss": 0.9213,
+      "step": 1734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6638413936811398e-05,
+      "loss": 0.3398,
+      "step": 1735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6634352271400693e-05,
+      "loss": 0.9423,
+      "step": 1736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66302886501136e-05,
+      "loss": 0.8848,
+      "step": 1737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6626223074148105e-05,
+      "loss": 0.985,
+      "step": 1738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6622155544702804e-05,
+      "loss": 0.3724,
+      "step": 1739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661808606297683e-05,
+      "loss": 0.9327,
+      "step": 1740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6614014630169916e-05,
+      "loss": 0.9966,
+      "step": 1741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660994124748236e-05,
+      "loss": 0.9721,
+      "step": 1742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6605865916115044e-05,
+      "loss": 0.9238,
+      "step": 1743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660178863726941e-05,
+      "loss": 0.9772,
+      "step": 1744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659770941214749e-05,
+      "loss": 1.0167,
+      "step": 1745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6593628241951875e-05,
+      "loss": 0.9085,
+      "step": 1746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658954512788574e-05,
+      "loss": 0.9217,
+      "step": 1747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658546007115283e-05,
+      "loss": 0.9958,
+      "step": 1748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658137307295746e-05,
+      "loss": 0.3533,
+      "step": 1749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6577284134504527e-05,
+      "loss": 0.9719,
+      "step": 1750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6573193256999486e-05,
+      "loss": 0.9305,
+      "step": 1751
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569100441648373e-05,
+      "loss": 0.9284,
+      "step": 1752
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6565005689657792e-05,
+      "loss": 1.0139,
+      "step": 1753
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6560909002234917e-05,
+      "loss": 0.8721,
+      "step": 1754
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6556810380587497e-05,
+      "loss": 1.0082,
+      "step": 1755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6552709825923846e-05,
+      "loss": 0.9091,
+      "step": 1756
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.9714,
+      "step": 1757
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654450292238397e-05,
+      "loss": 1.0059,
+      "step": 1758
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654039657592723e-05,
+      "loss": 0.9055,
+      "step": 1759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6536288301293218e-05,
+      "loss": 0.981,
+      "step": 1760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65321780996931e-05,
+      "loss": 0.928,
+      "step": 1761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528065972338607e-05,
+      "loss": 0.9819,
+      "step": 1762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6523951920442032e-05,
+      "loss": 0.9029,
+      "step": 1763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6519835945216253e-05,
+      "loss": 0.9378,
+      "step": 1764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651571804787469e-05,
+      "loss": 0.9439,
+      "step": 1765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511598229631344e-05,
+      "loss": 0.3229,
+      "step": 1766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6507476491700788e-05,
+      "loss": 0.3442,
+      "step": 1767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503352835298147e-05,
+      "loss": 0.9555,
+      "step": 1768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499227261639116e-05,
+      "loss": 0.948,
+      "step": 1769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649509977193996e-05,
+      "loss": 1.0448,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6490970367417515e-05,
+      "loss": 0.9558,
+      "step": 1771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648683904928916e-05,
+      "loss": 0.8792,
+      "step": 1772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482705818772853e-05,
+      "loss": 1.0149,
+      "step": 1773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647857067708712e-05,
+      "loss": 0.9475,
+      "step": 1774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6474433625451035e-05,
+      "loss": 0.9052,
+      "step": 1775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6470294665084254e-05,
+      "loss": 0.9054,
+      "step": 1776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6466153797206973e-05,
+      "loss": 0.9744,
+      "step": 1777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6462011023039977e-05,
+      "loss": 0.9955,
+      "step": 1778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457866343804592e-05,
+      "loss": 1.0248,
+      "step": 1779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645371976072271e-05,
+      "loss": 0.9848,
+      "step": 1780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6449571275016795e-05,
+      "loss": 0.9869,
+      "step": 1781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6445420887909858e-05,
+      "loss": 0.9375,
+      "step": 1782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6441268600625476e-05,
+      "loss": 0.9852,
+      "step": 1783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643711441438779e-05,
+      "loss": 1.0049,
+      "step": 1784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432958330421497e-05,
+      "loss": 0.9388,
+      "step": 1785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6428800349951853e-05,
+      "loss": 1.0089,
+      "step": 1786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6424640474204675e-05,
+      "loss": 0.9816,
+      "step": 1787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6420478704406337e-05,
+      "loss": 1.0043,
+      "step": 1788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641631504178377e-05,
+      "loss": 0.9601,
+      "step": 1789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6412149487564473e-05,
+      "loss": 0.9055,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407982042976483e-05,
+      "loss": 0.9125,
+      "step": 1791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640381270924842e-05,
+      "loss": 1.0183,
+      "step": 1792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639964148760943e-05,
+      "loss": 0.948,
+      "step": 1793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395468379289253e-05,
+      "loss": 0.943,
+      "step": 1794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6391293385518146e-05,
+      "loss": 1.0017,
+      "step": 1795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6387116507526958e-05,
+      "loss": 1.0416,
+      "step": 1796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6382937746547064e-05,
+      "loss": 0.9828,
+      "step": 1797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637875710381041e-05,
+      "loss": 0.9286,
+      "step": 1798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637457458054949e-05,
+      "loss": 0.9831,
+      "step": 1799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637039017799736e-05,
+      "loss": 0.9895,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636620389738763e-05,
+      "loss": 0.9481,
+      "step": 1801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6362015739954452e-05,
+      "loss": 0.9323,
+      "step": 1802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635782570693254e-05,
+      "loss": 0.9506,
+      "step": 1803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353633799557153e-05,
+      "loss": 0.9827,
+      "step": 1804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6349440019064126e-05,
+      "loss": 0.3667,
+      "step": 1805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6345244366689813e-05,
+      "loss": 0.979,
+      "step": 1806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6341046843671145e-05,
+      "loss": 0.9373,
+      "step": 1807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6336847451245592e-05,
+      "loss": 0.9483,
+      "step": 1808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633264619065118e-05,
+      "loss": 0.9368,
+      "step": 1809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328443063126488e-05,
+      "loss": 0.9613,
+      "step": 1810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324238069910633e-05,
+      "loss": 0.9062,
+      "step": 1811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63200312122433e-05,
+      "loss": 0.9418,
+      "step": 1812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6315822491364708e-05,
+      "loss": 0.8971,
+      "step": 1813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311611908515635e-05,
+      "loss": 0.9685,
+      "step": 1814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6307399464937404e-05,
+      "loss": 0.9008,
+      "step": 1815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6303185161871895e-05,
+      "loss": 0.9469,
+      "step": 1816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6298969000561515e-05,
+      "loss": 0.9318,
+      "step": 1817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629475098224924e-05,
+      "loss": 0.961,
+      "step": 1818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6290531108178587e-05,
+      "loss": 1.0046,
+      "step": 1819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6286309379593616e-05,
+      "loss": 0.9328,
+      "step": 1820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6282085797738937e-05,
+      "loss": 0.9382,
+      "step": 1821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627786036385971e-05,
+      "loss": 0.9779,
+      "step": 1822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273633079201626e-05,
+      "loss": 0.9118,
+      "step": 1823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6269403945010948e-05,
+      "loss": 1.0058,
+      "step": 1824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626517296253446e-05,
+      "loss": 0.991,
+      "step": 1825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62609401330195e-05,
+      "loss": 0.9844,
+      "step": 1826
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256705457713952e-05,
+      "loss": 0.9602,
+      "step": 1827
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6252468937866243e-05,
+      "loss": 0.3697,
+      "step": 1828
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624823057472534e-05,
+      "loss": 0.9498,
+      "step": 1829
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6243990369540764e-05,
+      "loss": 1.0282,
+      "step": 1830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6239748323562562e-05,
+      "loss": 0.9485,
+      "step": 1831
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6235504438041342e-05,
+      "loss": 0.9553,
+      "step": 1832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623125871422824e-05,
+      "loss": 0.9321,
+      "step": 1833
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6227011153374945e-05,
+      "loss": 0.9608,
+      "step": 1834
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622276175673368e-05,
+      "loss": 1.0294,
+      "step": 1835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218510525557206e-05,
+      "loss": 0.9577,
+      "step": 1836
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621425746109884e-05,
+      "loss": 0.9702,
+      "step": 1837
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210002564612425e-05,
+      "loss": 0.9359,
+      "step": 1838
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6205745837352346e-05,
+      "loss": 0.9087,
+      "step": 1839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6201487280573536e-05,
+      "loss": 0.8731,
+      "step": 1840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619722689553146e-05,
+      "loss": 0.3885,
+      "step": 1841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6192964683482127e-05,
+      "loss": 0.9777,
+      "step": 1842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6188700645682075e-05,
+      "loss": 0.8994,
+      "step": 1843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618443478338839e-05,
+      "loss": 0.9851,
+      "step": 1844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6180167097858697e-05,
+      "loss": 0.982,
+      "step": 1845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175897590351146e-05,
+      "loss": 0.3823,
+      "step": 1846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617162626212444e-05,
+      "loss": 0.9506,
+      "step": 1847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616735311443781e-05,
+      "loss": 0.9131,
+      "step": 1848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616307814855102e-05,
+      "loss": 0.9313,
+      "step": 1849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6158801365724376e-05,
+      "loss": 0.944,
+      "step": 1850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6154522767218726e-05,
+      "loss": 0.9729,
+      "step": 1851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6150242354295435e-05,
+      "loss": 0.9464,
+      "step": 1852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614596012821642e-05,
+      "loss": 0.9784,
+      "step": 1853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614167609024412e-05,
+      "loss": 0.977,
+      "step": 1854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613739024164152e-05,
+      "loss": 0.9471,
+      "step": 1855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613310258367213e-05,
+      "loss": 0.9577,
+      "step": 1856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61288131176e-05,
+      "loss": 0.9028,
+      "step": 1857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6124521844689707e-05,
+      "loss": 0.9266,
+      "step": 1858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612022876620636e-05,
+      "loss": 0.9107,
+      "step": 1859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6115933883415607e-05,
+      "loss": 0.373,
+      "step": 1860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6111637197583623e-05,
+      "loss": 0.9469,
+      "step": 1861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610733870997712e-05,
+      "loss": 0.9323,
+      "step": 1862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6103038421863332e-05,
+      "loss": 0.9121,
+      "step": 1863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609873633451003e-05,
+      "loss": 0.9606,
+      "step": 1864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6094432449185513e-05,
+      "loss": 0.9333,
+      "step": 1865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090126767158616e-05,
+      "loss": 0.9661,
+      "step": 1866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6085819289698695e-05,
+      "loss": 0.9359,
+      "step": 1867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6081510018075638e-05,
+      "loss": 0.9392,
+      "step": 1868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607719895355987e-05,
+      "loss": 0.9575,
+      "step": 1869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6072886097422333e-05,
+      "loss": 0.8808,
+      "step": 1870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60685714509345e-05,
+      "loss": 0.9777,
+      "step": 1871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606425501536838e-05,
+      "loss": 0.8886,
+      "step": 1872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60599367919965e-05,
+      "loss": 0.8994,
+      "step": 1873
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6055616782091917e-05,
+      "loss": 1.0152,
+      "step": 1874
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6051294986928218e-05,
+      "loss": 1.0019,
+      "step": 1875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6046971407779507e-05,
+      "loss": 0.9147,
+      "step": 1876
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604264604592042e-05,
+      "loss": 0.9711,
+      "step": 1877
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603831890262613e-05,
+      "loss": 0.973,
+      "step": 1878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603398997917232e-05,
+      "loss": 0.9246,
+      "step": 1879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6029659276835193e-05,
+      "loss": 0.3349,
+      "step": 1880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025326796891492e-05,
+      "loss": 0.9605,
+      "step": 1881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6020992540618476e-05,
+      "loss": 0.9661,
+      "step": 1882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6016656509293933e-05,
+      "loss": 0.3746,
+      "step": 1883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6012318704196164e-05,
+      "loss": 0.9773,
+      "step": 1884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6007979126604005e-05,
+      "loss": 0.9175,
+      "step": 1885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6003637777796804e-05,
+      "loss": 1.018,
+      "step": 1886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599929465905444e-05,
+      "loss": 0.9275,
+      "step": 1887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5994949771657307e-05,
+      "loss": 0.9665,
+      "step": 1888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5990603116886326e-05,
+      "loss": 0.9449,
+      "step": 1889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986254696022935e-05,
+      "loss": 0.9705,
+      "step": 1890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5981904510349092e-05,
+      "loss": 0.9679,
+      "step": 1891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5977552561147276e-05,
+      "loss": 0.929,
+      "step": 1892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5973198849700494e-05,
+      "loss": 0.9506,
+      "step": 1893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596884337729226e-05,
+      "loss": 0.9344,
+      "step": 1894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5964486145206612e-05,
+      "loss": 0.9805,
+      "step": 1895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5960127154728112e-05,
+      "loss": 0.909,
+      "step": 1896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5955766407141834e-05,
+      "loss": 0.9314,
+      "step": 1897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5951403903733373e-05,
+      "loss": 0.8881,
+      "step": 1898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947039645788834e-05,
+      "loss": 0.9637,
+      "step": 1899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5942673634594853e-05,
+      "loss": 0.9957,
+      "step": 1900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.11731843575419e-07,
+      "loss": 0.9556,
+      "step": 1901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.23463687150838e-07,
+      "loss": 1.0238,
+      "step": 1902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3519553072625703e-07,
+      "loss": 0.9935,
+      "step": 1903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.46927374301676e-07,
+      "loss": 0.9386,
+      "step": 1904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.58659217877095e-07,
+      "loss": 0.9827,
+      "step": 1905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.703910614525141e-07,
+      "loss": 1.0096,
+      "step": 1906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.82122905027933e-07,
+      "loss": 0.9289,
+      "step": 1907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.93854748603352e-07,
+      "loss": 0.8968,
+      "step": 1908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.005586592178771e-06,
+      "loss": 1.0083,
+      "step": 1909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.11731843575419e-06,
+      "loss": 0.9223,
+      "step": 1910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.229050279329609e-06,
+      "loss": 0.9253,
+      "step": 1911
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3407821229050281e-06,
+      "loss": 0.9412,
+      "step": 1912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4525139664804472e-06,
+      "loss": 0.9574,
+      "step": 1913
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.564245810055866e-06,
+      "loss": 0.8988,
+      "step": 1914
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.675977653631285e-06,
+      "loss": 0.9168,
+      "step": 1915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.787709497206704e-06,
+      "loss": 0.9858,
+      "step": 1916
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.899441340782123e-06,
+      "loss": 0.866,
+      "step": 1917
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.011173184357542e-06,
+      "loss": 1.0105,
+      "step": 1918
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.1229050279329612e-06,
+      "loss": 0.9,
+      "step": 1919
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.23463687150838e-06,
+      "loss": 0.9947,
+      "step": 1920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.3463687150837993e-06,
+      "loss": 0.9399,
+      "step": 1921
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.458100558659218e-06,
+      "loss": 0.9611,
+      "step": 1922
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.569832402234637e-06,
+      "loss": 0.9343,
+      "step": 1923
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.6815642458100562e-06,
+      "loss": 0.9352,
+      "step": 1924
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.793296089385475e-06,
+      "loss": 0.9538,
+      "step": 1925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9050279329608943e-06,
+      "loss": 0.3353,
+      "step": 1926
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.016759776536313e-06,
+      "loss": 0.9126,
+      "step": 1927
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.128491620111732e-06,
+      "loss": 0.8986,
+      "step": 1928
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.240223463687151e-06,
+      "loss": 0.9611,
+      "step": 1929
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.35195530726257e-06,
+      "loss": 0.3628,
+      "step": 1930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4636871508379893e-06,
+      "loss": 0.9265,
+      "step": 1931
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.575418994413408e-06,
+      "loss": 0.9461,
+      "step": 1932
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.687150837988827e-06,
+      "loss": 0.9313,
+      "step": 1933
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.798882681564246e-06,
+      "loss": 0.9649,
+      "step": 1934
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.910614525139665e-06,
+      "loss": 0.9071,
+      "step": 1935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.022346368715084e-06,
+      "loss": 0.9716,
+      "step": 1936
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.134078212290504e-06,
+      "loss": 0.9316,
+      "step": 1937
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2458100558659224e-06,
+      "loss": 0.8935,
+      "step": 1938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.357541899441341e-06,
+      "loss": 0.9033,
+      "step": 1939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.46927374301676e-06,
+      "loss": 0.8751,
+      "step": 1940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.581005586592179e-06,
+      "loss": 0.9688,
+      "step": 1941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.692737430167599e-06,
+      "loss": 0.9184,
+      "step": 1942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8044692737430175e-06,
+      "loss": 1.0064,
+      "step": 1943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.916201117318436e-06,
+      "loss": 0.9539,
+      "step": 1944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.027932960893855e-06,
+      "loss": 0.9686,
+      "step": 1945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.139664804469274e-06,
+      "loss": 0.9591,
+      "step": 1946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.251396648044693e-06,
+      "loss": 0.9842,
+      "step": 1947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.3631284916201125e-06,
+      "loss": 0.9152,
+      "step": 1948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.474860335195531e-06,
+      "loss": 0.9581,
+      "step": 1949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.58659217877095e-06,
+      "loss": 0.9359,
+      "step": 1950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.698324022346369e-06,
+      "loss": 0.9513,
+      "step": 1951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.810055865921789e-06,
+      "loss": 0.939,
+      "step": 1952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9217877094972075e-06,
+      "loss": 0.934,
+      "step": 1953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.033519553072626e-06,
+      "loss": 1.0162,
+      "step": 1954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.145251396648045e-06,
+      "loss": 1.0404,
+      "step": 1955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.256983240223464e-06,
+      "loss": 0.9122,
+      "step": 1956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.368715083798883e-06,
+      "loss": 0.9522,
+      "step": 1957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.480446927374302e-06,
+      "loss": 0.9622,
+      "step": 1958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.592178770949721e-06,
+      "loss": 0.935,
+      "step": 1959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.70391061452514e-06,
+      "loss": 0.9248,
+      "step": 1960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.815642458100559e-06,
+      "loss": 0.8945,
+      "step": 1961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.927374301675979e-06,
+      "loss": 0.971,
+      "step": 1962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.0391061452513975e-06,
+      "loss": 0.9063,
+      "step": 1963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.150837988826816e-06,
+      "loss": 0.8926,
+      "step": 1964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.262569832402235e-06,
+      "loss": 0.9231,
+      "step": 1965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.374301675977654e-06,
+      "loss": 0.8559,
+      "step": 1966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.486033519553073e-06,
+      "loss": 0.8943,
+      "step": 1967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.597765363128492e-06,
+      "loss": 0.9218,
+      "step": 1968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.709497206703911e-06,
+      "loss": 0.9532,
+      "step": 1969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82122905027933e-06,
+      "loss": 0.3498,
+      "step": 1970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.932960893854749e-06,
+      "loss": 0.942,
+      "step": 1971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.044692737430168e-06,
+      "loss": 0.9088,
+      "step": 1972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.156424581005588e-06,
+      "loss": 0.9391,
+      "step": 1973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.268156424581007e-06,
+      "loss": 0.8914,
+      "step": 1974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.379888268156426e-06,
+      "loss": 0.9405,
+      "step": 1975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.491620111731845e-06,
+      "loss": 0.96,
+      "step": 1976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.603351955307264e-06,
+      "loss": 0.9487,
+      "step": 1977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.715083798882683e-06,
+      "loss": 0.931,
+      "step": 1978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.826815642458101e-06,
+      "loss": 0.9529,
+      "step": 1979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.93854748603352e-06,
+      "loss": 0.9495,
+      "step": 1980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.050279329608939e-06,
+      "loss": 0.8757,
+      "step": 1981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.162011173184358e-06,
+      "loss": 0.9278,
+      "step": 1982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.273743016759777e-06,
+      "loss": 0.9292,
+      "step": 1983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.385474860335197e-06,
+      "loss": 0.8547,
+      "step": 1984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.497206703910616e-06,
+      "loss": 0.8732,
+      "step": 1985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.608938547486035e-06,
+      "loss": 0.9263,
+      "step": 1986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.720670391061454e-06,
+      "loss": 0.9383,
+      "step": 1987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.832402234636873e-06,
+      "loss": 0.8878,
+      "step": 1988
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.944134078212291e-06,
+      "loss": 0.9788,
+      "step": 1989
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.005586592178771e-05,
+      "loss": 0.9293,
+      "step": 1990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0167597765363129e-05,
+      "loss": 0.9328,
+      "step": 1991
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0279329608938548e-05,
+      "loss": 0.3591,
+      "step": 1992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0391061452513967e-05,
+      "loss": 0.9202,
+      "step": 1993
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0502793296089386e-05,
+      "loss": 0.9627,
+      "step": 1994
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0614525139664806e-05,
+      "loss": 0.9191,
+      "step": 1995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0726256983240225e-05,
+      "loss": 0.8953,
+      "step": 1996
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0837988826815644e-05,
+      "loss": 0.9378,
+      "step": 1997
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.0949720670391063e-05,
+      "loss": 0.933,
+      "step": 1998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1061452513966481e-05,
+      "loss": 0.9743,
+      "step": 1999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.11731843575419e-05,
+      "loss": 0.907,
+      "step": 2000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1284916201117319e-05,
+      "loss": 0.9195,
+      "step": 2001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1396648044692738e-05,
+      "loss": 0.9565,
+      "step": 2002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1508379888268157e-05,
+      "loss": 0.9067,
+      "step": 2003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1620111731843577e-05,
+      "loss": 0.9423,
+      "step": 2004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1731843575418994e-05,
+      "loss": 0.9455,
+      "step": 2005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1843575418994415e-05,
+      "loss": 0.9402,
+      "step": 2006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1955307262569834e-05,
+      "loss": 0.9257,
+      "step": 2007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2067039106145253e-05,
+      "loss": 0.9588,
+      "step": 2008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2178770949720671e-05,
+      "loss": 0.8826,
+      "step": 2009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.229050279329609e-05,
+      "loss": 0.885,
+      "step": 2010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2402234636871509e-05,
+      "loss": 0.3655,
+      "step": 2011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2513966480446928e-05,
+      "loss": 0.9204,
+      "step": 2012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2625698324022347e-05,
+      "loss": 0.9252,
+      "step": 2013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2737430167597766e-05,
+      "loss": 1.0406,
+      "step": 2014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2849162011173186e-05,
+      "loss": 0.9092,
+      "step": 2015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2960893854748603e-05,
+      "loss": 0.9631,
+      "step": 2016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3072625698324024e-05,
+      "loss": 0.9558,
+      "step": 2017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3184357541899443e-05,
+      "loss": 0.9012,
+      "step": 2018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3296089385474861e-05,
+      "loss": 0.8769,
+      "step": 2019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.340782122905028e-05,
+      "loss": 0.8885,
+      "step": 2020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3519553072625699e-05,
+      "loss": 0.343,
+      "step": 2021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3631284916201118e-05,
+      "loss": 0.9886,
+      "step": 2022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3743016759776537e-05,
+      "loss": 0.9512,
+      "step": 2023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3854748603351957e-05,
+      "loss": 0.9012,
+      "step": 2024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3966480446927374e-05,
+      "loss": 0.9985,
+      "step": 2025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4078212290502795e-05,
+      "loss": 0.9283,
+      "step": 2026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4189944134078212e-05,
+      "loss": 0.9153,
+      "step": 2027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4301675977653633e-05,
+      "loss": 0.9614,
+      "step": 2028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4413407821229052e-05,
+      "loss": 0.9418,
+      "step": 2029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.452513966480447e-05,
+      "loss": 0.9835,
+      "step": 2030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.463687150837989e-05,
+      "loss": 0.9929,
+      "step": 2031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4748603351955308e-05,
+      "loss": 0.9346,
+      "step": 2032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4860335195530729e-05,
+      "loss": 0.9297,
+      "step": 2033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4972067039106146e-05,
+      "loss": 0.9329,
+      "step": 2034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5083798882681566e-05,
+      "loss": 0.8894,
+      "step": 2035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5195530726256983e-05,
+      "loss": 0.9348,
+      "step": 2036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5307262569832404e-05,
+      "loss": 0.9124,
+      "step": 2037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5418994413407823e-05,
+      "loss": 0.9486,
+      "step": 2038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.553072625698324e-05,
+      "loss": 1.0199,
+      "step": 2039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.564245810055866e-05,
+      "loss": 0.9374,
+      "step": 2040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.575418994413408e-05,
+      "loss": 0.9037,
+      "step": 2041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5865921787709498e-05,
+      "loss": 0.9518,
+      "step": 2042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5977653631284917e-05,
+      "loss": 0.9505,
+      "step": 2043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6089385474860336e-05,
+      "loss": 0.985,
+      "step": 2044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6201117318435755e-05,
+      "loss": 0.9782,
+      "step": 2045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6312849162011177e-05,
+      "loss": 0.9623,
+      "step": 2046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6424581005586592e-05,
+      "loss": 0.9975,
+      "step": 2047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6536312849162014e-05,
+      "loss": 0.9069,
+      "step": 2048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.664804469273743e-05,
+      "loss": 0.9699,
+      "step": 2049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6759776536312852e-05,
+      "loss": 0.8923,
+      "step": 2050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.687150837988827e-05,
+      "loss": 0.9225,
+      "step": 2051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.698324022346369e-05,
+      "loss": 0.8729,
+      "step": 2052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.709497206703911e-05,
+      "loss": 0.9227,
+      "step": 2053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7206703910614527e-05,
+      "loss": 0.8861,
+      "step": 2054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7318435754189946e-05,
+      "loss": 0.9594,
+      "step": 2055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7430167597765365e-05,
+      "loss": 0.3578,
+      "step": 2056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7541899441340784e-05,
+      "loss": 0.8833,
+      "step": 2057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7653631284916203e-05,
+      "loss": 0.9323,
+      "step": 2058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.776536312849162e-05,
+      "loss": 0.9532,
+      "step": 2059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.787709497206704e-05,
+      "loss": 0.3462,
+      "step": 2060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.798882681564246e-05,
+      "loss": 0.8868,
+      "step": 2061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8100558659217878e-05,
+      "loss": 0.9124,
+      "step": 2062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8212290502793297e-05,
+      "loss": 0.9805,
+      "step": 2063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8324022346368716e-05,
+      "loss": 0.8954,
+      "step": 2064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8435754189944135e-05,
+      "loss": 0.982,
+      "step": 2065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8547486033519553e-05,
+      "loss": 0.8863,
+      "step": 2066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8659217877094972e-05,
+      "loss": 1.008,
+      "step": 2067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8770949720670394e-05,
+      "loss": 0.9123,
+      "step": 2068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.888268156424581e-05,
+      "loss": 0.8969,
+      "step": 2069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8994413407821232e-05,
+      "loss": 0.9238,
+      "step": 2070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.910614525139665e-05,
+      "loss": 0.8787,
+      "step": 2071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.921787709497207e-05,
+      "loss": 0.9531,
+      "step": 2072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.932960893854749e-05,
+      "loss": 0.9578,
+      "step": 2073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9441340782122907e-05,
+      "loss": 0.9244,
+      "step": 2074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9553072625698326e-05,
+      "loss": 0.9653,
+      "step": 2075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9664804469273745e-05,
+      "loss": 0.379,
+      "step": 2076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9776536312849164e-05,
+      "loss": 0.8898,
+      "step": 2077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9888268156424583e-05,
+      "loss": 0.9587,
+      "step": 2078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.8987,
+      "step": 2079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999998525947744e-05,
+      "loss": 0.995,
+      "step": 2080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999994103791404e-05,
+      "loss": 0.9864,
+      "step": 2081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999986733532287e-05,
+      "loss": 0.8768,
+      "step": 2082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.999997641517256e-05,
+      "loss": 0.945,
+      "step": 2083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999963148715276e-05,
+      "loss": 0.9197,
+      "step": 2084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999946934164337e-05,
+      "loss": 0.9735,
+      "step": 2085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999927771524526e-05,
+      "loss": 0.9903,
+      "step": 2086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.999990566080149e-05,
+      "loss": 0.95,
+      "step": 2087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999880602001752e-05,
+      "loss": 0.9967,
+      "step": 2088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999852595132698e-05,
+      "loss": 0.9844,
+      "step": 2089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999821640202584e-05,
+      "loss": 0.961,
+      "step": 2090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.999978773722054e-05,
+      "loss": 0.9694,
+      "step": 2091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.999975088619655e-05,
+      "loss": 0.9558,
+      "step": 2092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999711087141484e-05,
+      "loss": 0.9955,
+      "step": 2093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999668340067083e-05,
+      "loss": 0.9372,
+      "step": 2094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999622644985937e-05,
+      "loss": 0.9977,
+      "step": 2095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999574001911526e-05,
+      "loss": 0.9133,
+      "step": 2096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999522410858187e-05,
+      "loss": 0.9336,
+      "step": 2097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999467871841126e-05,
+      "loss": 0.9781,
+      "step": 2098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.999941038487643e-05,
+      "loss": 0.9462,
+      "step": 2099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.999934994998104e-05,
+      "loss": 0.8491,
+      "step": 2100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.999928656717278e-05,
+      "loss": 0.9412,
+      "step": 2101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999220236470326e-05,
+      "loss": 0.9719,
+      "step": 2102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.999915095789324e-05,
+      "loss": 0.9759,
+      "step": 2103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9999078731461942e-05,
+      "loss": 0.951,
+      "step": 2104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.999900355719773e-05,
+      "loss": 0.9807,
+      "step": 2105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9998925435122757e-05,
+      "loss": 0.9845,
+      "step": 2106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9998844365260068e-05,
+      "loss": 0.9743,
+      "step": 2107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9998760347633556e-05,
+      "loss": 1.013,
+      "step": 2108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9998673382267983e-05,
+      "loss": 0.967,
+      "step": 2109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9998583469189e-05,
+      "loss": 0.9648,
+      "step": 2110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9998490608423106e-05,
+      "loss": 0.9696,
+      "step": 2111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9998394799997684e-05,
+      "loss": 0.9583,
+      "step": 2112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9998296043940973e-05,
+      "loss": 0.9451,
+      "step": 2113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9998194340282088e-05,
+      "loss": 0.9213,
+      "step": 2114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9998089689051014e-05,
+      "loss": 0.9282,
+      "step": 2115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9997982090278606e-05,
+      "loss": 1.0565,
+      "step": 2116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.999787154399658e-05,
+      "loss": 0.9466,
+      "step": 2117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9997758050237533e-05,
+      "loss": 0.988,
+      "step": 2118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9997641609034916e-05,
+      "loss": 0.9047,
+      "step": 2119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9997522220423062e-05,
+      "loss": 0.9399,
+      "step": 2120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9997399884437167e-05,
+      "loss": 0.94,
+      "step": 2121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9997274601113294e-05,
+      "loss": 0.9678,
+      "step": 2122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9997146370488384e-05,
+      "loss": 0.9633,
+      "step": 2123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9997015192600235e-05,
+      "loss": 0.9667,
+      "step": 2124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9996881067487526e-05,
+      "loss": 0.964,
+      "step": 2125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9996743995189793e-05,
+      "loss": 0.3857,
+      "step": 2126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9996603975747443e-05,
+      "loss": 0.9595,
+      "step": 2127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9996461009201767e-05,
+      "loss": 0.3784,
+      "step": 2128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9996315095594903e-05,
+      "loss": 0.9302,
+      "step": 2129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.999616623496987e-05,
+      "loss": 0.3875,
+      "step": 2130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9996014427370556e-05,
+      "loss": 0.9582,
+      "step": 2131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9995859672841713e-05,
+      "loss": 1.0083,
+      "step": 2132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9995701971428967e-05,
+      "loss": 0.9549,
+      "step": 2133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9995541323178808e-05,
+      "loss": 0.9243,
+      "step": 2134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9995377728138597e-05,
+      "loss": 0.9543,
+      "step": 2135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9995211186356564e-05,
+      "loss": 0.9756,
+      "step": 2136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9995041697881805e-05,
+      "loss": 0.9326,
+      "step": 2137
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.999486926276429e-05,
+      "loss": 0.9087,
+      "step": 2138
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9994693881054853e-05,
+      "loss": 0.9643,
+      "step": 2139
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.99945155528052e-05,
+      "loss": 1.0203,
+      "step": 2140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9994334278067905e-05,
+      "loss": 1.0366,
+      "step": 2141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9994150056896405e-05,
+      "loss": 0.9883,
+      "step": 2142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9993962889345012e-05,
+      "loss": 0.971,
+      "step": 2143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9993772775468907e-05,
+      "loss": 0.3625,
+      "step": 2144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9993579715324135e-05,
+      "loss": 0.9937,
+      "step": 2145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9993383708967618e-05,
+      "loss": 0.9292,
+      "step": 2146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9993184756457132e-05,
+      "loss": 0.9348,
+      "step": 2147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9992982857851334e-05,
+      "loss": 1.0105,
+      "step": 2148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9992778013209752e-05,
+      "loss": 0.9786,
+      "step": 2149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9992570222592768e-05,
+      "loss": 1.021,
+      "step": 2150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.999235948606164e-05,
+      "loss": 1.0123,
+      "step": 2151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9992145803678505e-05,
+      "loss": 1.0262,
+      "step": 2152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.999192917550635e-05,
+      "loss": 0.959,
+      "step": 2153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9991709601609042e-05,
+      "loss": 0.9799,
+      "step": 2154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9991487082051314e-05,
+      "loss": 0.9666,
+      "step": 2155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9991261616898767e-05,
+      "loss": 0.9313,
+      "step": 2156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9991033206217868e-05,
+      "loss": 0.9498,
+      "step": 2157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.999080185007596e-05,
+      "loss": 0.9167,
+      "step": 2158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9990567548541245e-05,
+      "loss": 0.3667,
+      "step": 2159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9990330301682798e-05,
+      "loss": 0.9086,
+      "step": 2160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9990090109570565e-05,
+      "loss": 0.9641,
+      "step": 2161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9989846972275356e-05,
+      "loss": 0.9684,
+      "step": 2162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9989600889868846e-05,
+      "loss": 0.9063,
+      "step": 2163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9989351862423585e-05,
+      "loss": 0.9301,
+      "step": 2164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9989099890012993e-05,
+      "loss": 0.9582,
+      "step": 2165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.998884497271135e-05,
+      "loss": 0.9469,
+      "step": 2166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.998858711059381e-05,
+      "loss": 0.9516,
+      "step": 2167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.998832630373639e-05,
+      "loss": 0.922,
+      "step": 2168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9988062552215983e-05,
+      "loss": 0.8531,
+      "step": 2169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9987795856110347e-05,
+      "loss": 0.9349,
+      "step": 2170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.99875262154981e-05,
+      "loss": 0.9598,
+      "step": 2171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9987253630458738e-05,
+      "loss": 0.9098,
+      "step": 2172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9986978101072627e-05,
+      "loss": 0.9451,
+      "step": 2173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9986699627420987e-05,
+      "loss": 0.982,
+      "step": 2174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.998641820958592e-05,
+      "loss": 0.9374,
+      "step": 2175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9986133847650392e-05,
+      "loss": 0.9544,
+      "step": 2176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9985846541698234e-05,
+      "loss": 0.9654,
+      "step": 2177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9985556291814147e-05,
+      "loss": 0.9463,
+      "step": 2178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.99852630980837e-05,
+      "loss": 1.0037,
+      "step": 2179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.998496696059333e-05,
+      "loss": 0.9683,
+      "step": 2180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9984667879430336e-05,
+      "loss": 0.9171,
+      "step": 2181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.99843658546829e-05,
+      "loss": 0.9104,
+      "step": 2182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.998406088644006e-05,
+      "loss": 0.9568,
+      "step": 2183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9983752974791715e-05,
+      "loss": 0.9796,
+      "step": 2184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9983442119828647e-05,
+      "loss": 0.9713,
+      "step": 2185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.99831283216425e-05,
+      "loss": 1.0112,
+      "step": 2186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9982811580325784e-05,
+      "loss": 0.9947,
+      "step": 2187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9982491895971878e-05,
+      "loss": 0.902,
+      "step": 2188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9982169268675024e-05,
+      "loss": 0.9699,
+      "step": 2189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9981843698530345e-05,
+      "loss": 1.0531,
+      "step": 2190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9981515185633812e-05,
+      "loss": 0.9512,
+      "step": 2191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9981183730082283e-05,
+      "loss": 0.9653,
+      "step": 2192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9980849331973467e-05,
+      "loss": 0.9082,
+      "step": 2193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9980511991405955e-05,
+      "loss": 0.8944,
+      "step": 2194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9980171708479194e-05,
+      "loss": 0.9982,
+      "step": 2195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9979828483293504e-05,
+      "loss": 0.9731,
+      "step": 2196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.997948231595007e-05,
+      "loss": 0.9695,
+      "step": 2197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9979133206550953e-05,
+      "loss": 0.9372,
+      "step": 2198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9978781155199062e-05,
+      "loss": 1.0128,
+      "step": 2199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9978426161998195e-05,
+      "loss": 1.0321,
+      "step": 2200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9978068227053002e-05,
+      "loss": 0.9499,
+      "step": 2201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9977707350469012e-05,
+      "loss": 1.0056,
+      "step": 2202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9977343532352608e-05,
+      "loss": 0.9772,
+      "step": 2203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9976976772811055e-05,
+      "loss": 0.9922,
+      "step": 2204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.997660707195247e-05,
+      "loss": 0.9724,
+      "step": 2205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.997623442988585e-05,
+      "loss": 0.9293,
+      "step": 2206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9975858846721054e-05,
+      "loss": 0.883,
+      "step": 2207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9975480322568802e-05,
+      "loss": 0.922,
+      "step": 2208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9975098857540695e-05,
+      "loss": 0.9172,
+      "step": 2209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9974714451749185e-05,
+      "loss": 0.9544,
+      "step": 2210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9974327105307605e-05,
+      "loss": 0.9932,
+      "step": 2211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9973936818330145e-05,
+      "loss": 0.9194,
+      "step": 2212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9973543590931867e-05,
+      "loss": 0.9448,
+      "step": 2213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9973147423228698e-05,
+      "loss": 0.9984,
+      "step": 2214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9972748315337434e-05,
+      "loss": 0.9747,
+      "step": 2215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9972346267375736e-05,
+      "loss": 0.971,
+      "step": 2216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.997194127946213e-05,
+      "loss": 0.9676,
+      "step": 2217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9971533351716012e-05,
+      "loss": 0.9679,
+      "step": 2218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9971122484257643e-05,
+      "loss": 0.9395,
+      "step": 2219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9970708677208147e-05,
+      "loss": 0.9105,
+      "step": 2220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9970291930689526e-05,
+      "loss": 0.9887,
+      "step": 2221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9969872244824638e-05,
+      "loss": 0.959,
+      "step": 2222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9969449619737212e-05,
+      "loss": 0.9753,
+      "step": 2223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9969024055551843e-05,
+      "loss": 0.9526,
+      "step": 2224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9968595552393983e-05,
+      "loss": 0.9303,
+      "step": 2225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.996816411038997e-05,
+      "loss": 0.9165,
+      "step": 2226
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9967729729666993e-05,
+      "loss": 1.0088,
+      "step": 2227
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9967292410353117e-05,
+      "loss": 0.9459,
+      "step": 2228
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.996685215257726e-05,
+      "loss": 0.9475,
+      "step": 2229
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9966408956469215e-05,
+      "loss": 0.9664,
+      "step": 2230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.996596282215965e-05,
+      "loss": 0.9786,
+      "step": 2231
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.996551374978008e-05,
+      "loss": 1.0069,
+      "step": 2232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9965061739462903e-05,
+      "loss": 0.3851,
+      "step": 2233
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9964606791341373e-05,
+      "loss": 0.8833,
+      "step": 2234
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9964148905549617e-05,
+      "loss": 0.9302,
+      "step": 2235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.996368808222262e-05,
+      "loss": 0.9493,
+      "step": 2236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.996322432149624e-05,
+      "loss": 0.9756,
+      "step": 2237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9962757623507197e-05,
+      "loss": 0.986,
+      "step": 2238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.996228798839308e-05,
+      "loss": 0.9261,
+      "step": 2239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.996181541629234e-05,
+      "loss": 0.9511,
+      "step": 2240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.99613399073443e-05,
+      "loss": 0.3786,
+      "step": 2241
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9960861461689146e-05,
+      "loss": 0.9235,
+      "step": 2242
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.996038007946792e-05,
+      "loss": 1.0645,
+      "step": 2243
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.995989576082255e-05,
+      "loss": 0.9461,
+      "step": 2244
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9959408505895807e-05,
+      "loss": 0.9143,
+      "step": 2245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9958918314831347e-05,
+      "loss": 0.913,
+      "step": 2246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.995842518777368e-05,
+      "loss": 0.9848,
+      "step": 2247
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9957929124868184e-05,
+      "loss": 0.8998,
+      "step": 2248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.995743012626111e-05,
+      "loss": 0.938,
+      "step": 2249
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.995692819209956e-05,
+      "loss": 0.9419,
+      "step": 2250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9956423322531515e-05,
+      "loss": 0.8603,
+      "step": 2251
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.995591551770581e-05,
+      "loss": 0.9509,
+      "step": 2252
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9955404777772157e-05,
+      "loss": 0.9502,
+      "step": 2253
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9954891102881126e-05,
+      "loss": 0.9731,
+      "step": 2254
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9954374493184153e-05,
+      "loss": 0.3504,
+      "step": 2255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9953854948833537e-05,
+      "loss": 0.9582,
+      "step": 2256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9953332469982453e-05,
+      "loss": 1.0013,
+      "step": 2257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9952807056784925e-05,
+      "loss": 0.9589,
+      "step": 2258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9952278709395855e-05,
+      "loss": 0.9891,
+      "step": 2259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9951747427971e-05,
+      "loss": 0.8324,
+      "step": 2260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9951213212666993e-05,
+      "loss": 0.3682,
+      "step": 2261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9950676063641325e-05,
+      "loss": 0.9716,
+      "step": 2262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9950135981052353e-05,
+      "loss": 0.9935,
+      "step": 2263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.99495929650593e-05,
+      "loss": 0.9983,
+      "step": 2264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.994904701582225e-05,
+      "loss": 0.8988,
+      "step": 2265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9948498133502155e-05,
+      "loss": 1.0271,
+      "step": 2266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9947946318260834e-05,
+      "loss": 0.9087,
+      "step": 2267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9947391570260964e-05,
+      "loss": 0.965,
+      "step": 2268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.994683388966609e-05,
+      "loss": 0.9843,
+      "step": 2269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.994627327664063e-05,
+      "loss": 1.0084,
+      "step": 2270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9945709731349852e-05,
+      "loss": 0.9343,
+      "step": 2271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9945143253959896e-05,
+      "loss": 0.908,
+      "step": 2272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9944573844637763e-05,
+      "loss": 0.9273,
+      "step": 2273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9944001503551326e-05,
+      "loss": 0.9376,
+      "step": 2274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9943426230869313e-05,
+      "loss": 0.9519,
+      "step": 2275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9942848026761324e-05,
+      "loss": 0.9095,
+      "step": 2276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9942266891397817e-05,
+      "loss": 0.9652,
+      "step": 2277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9941682824950117e-05,
+      "loss": 0.8928,
+      "step": 2278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9941095827590415e-05,
+      "loss": 0.964,
+      "step": 2279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.994050589949176e-05,
+      "loss": 0.963,
+      "step": 2280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9939913040828073e-05,
+      "loss": 0.9221,
+      "step": 2281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9939317251774134e-05,
+      "loss": 1.0532,
+      "step": 2282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9938718532505584e-05,
+      "loss": 0.3635,
+      "step": 2283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.993811688319894e-05,
+      "loss": 0.9531,
+      "step": 2284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.993751230403156e-05,
+      "loss": 0.9654,
+      "step": 2285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9936904795181696e-05,
+      "loss": 0.9415,
+      "step": 2286
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.993629435682844e-05,
+      "loss": 0.9963,
+      "step": 2287
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.993568098915176e-05,
+      "loss": 0.9559,
+      "step": 2288
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9935064692332476e-05,
+      "loss": 1.0141,
+      "step": 2289
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9934445466552283e-05,
+      "loss": 1.0247,
+      "step": 2290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.993382331199374e-05,
+      "loss": 0.8953,
+      "step": 2291
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9933198228840254e-05,
+      "loss": 0.9448,
+      "step": 2292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9932570217276115e-05,
+      "loss": 0.9222,
+      "step": 2293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9931939277486463e-05,
+      "loss": 0.9476,
+      "step": 2294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9931305409657307e-05,
+      "loss": 0.9516,
+      "step": 2295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.993066861397552e-05,
+      "loss": 0.9502,
+      "step": 2296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9930028890628832e-05,
+      "loss": 0.9722,
+      "step": 2297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9929386239805843e-05,
+      "loss": 0.9744,
+      "step": 2298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.992874066169601e-05,
+      "loss": 0.9647,
+      "step": 2299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9928092156489664e-05,
+      "loss": 0.9929,
+      "step": 2300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.992744072437798e-05,
+      "loss": 0.9865,
+      "step": 2301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.992678636555301e-05,
+      "loss": 0.9362,
+      "step": 2302
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9926129080207676e-05,
+      "loss": 1.0174,
+      "step": 2303
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9925468868535743e-05,
+      "loss": 0.973,
+      "step": 2304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9924805730731847e-05,
+      "loss": 0.9095,
+      "step": 2305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.992413966699149e-05,
+      "loss": 0.9517,
+      "step": 2306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.992347067751104e-05,
+      "loss": 0.9935,
+      "step": 2307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9922798762487715e-05,
+      "loss": 0.9282,
+      "step": 2308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9922123922119606e-05,
+      "loss": 1.0262,
+      "step": 2309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9921446156605663e-05,
+      "loss": 1.0159,
+      "step": 2310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.99207654661457e-05,
+      "loss": 0.933,
+      "step": 2311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9920081850940382e-05,
+      "loss": 0.9594,
+      "step": 2312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9919395311191256e-05,
+      "loss": 0.9754,
+      "step": 2313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.991870584710072e-05,
+      "loss": 0.9917,
+      "step": 2314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9918013458872036e-05,
+      "loss": 0.9498,
+      "step": 2315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.991731814670932e-05,
+      "loss": 0.9597,
+      "step": 2316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9916619910817564e-05,
+      "loss": 0.9465,
+      "step": 2317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9915918751402615e-05,
+      "loss": 0.9025,
+      "step": 2318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.991521466867118e-05,
+      "loss": 1.0017,
+      "step": 2319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.991450766283083e-05,
+      "loss": 0.9464,
+      "step": 2320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.991379773409e-05,
+      "loss": 0.957,
+      "step": 2321
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9913084882657978e-05,
+      "loss": 0.9471,
+      "step": 2322
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9912369108744927e-05,
+      "loss": 0.9593,
+      "step": 2323
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9911650412561862e-05,
+      "loss": 0.9432,
+      "step": 2324
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.991092879432066e-05,
+      "loss": 0.9745,
+      "step": 2325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9910204254234068e-05,
+      "loss": 0.3915,
+      "step": 2326
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.990947679251568e-05,
+      "loss": 0.9173,
+      "step": 2327
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.990874640937997e-05,
+      "loss": 0.3916,
+      "step": 2328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.990801310504225e-05,
+      "loss": 1.0295,
+      "step": 2329
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9907276879718715e-05,
+      "loss": 0.9363,
+      "step": 2330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9906537733626407e-05,
+      "loss": 0.992,
+      "step": 2331
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9905795666983234e-05,
+      "loss": 1.0234,
+      "step": 2332
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.990505068000797e-05,
+      "loss": 0.9253,
+      "step": 2333
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9904302772920238e-05,
+      "loss": 0.9243,
+      "step": 2334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9903551945940537e-05,
+      "loss": 0.9584,
+      "step": 2335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.990279819929021e-05,
+      "loss": 0.9585,
+      "step": 2336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.990204153319147e-05,
+      "loss": 0.9256,
+      "step": 2337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9901281947867402e-05,
+      "loss": 0.9351,
+      "step": 2338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9900519443541927e-05,
+      "loss": 0.9837,
+      "step": 2339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9899754020439843e-05,
+      "loss": 0.9696,
+      "step": 2340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9898985678786802e-05,
+      "loss": 0.9366,
+      "step": 2341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.989821441880933e-05,
+      "loss": 0.9727,
+      "step": 2342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.989744024073479e-05,
+      "loss": 0.9224,
+      "step": 2343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9896663144791423e-05,
+      "loss": 0.9892,
+      "step": 2344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9895883131208325e-05,
+      "loss": 0.9737,
+      "step": 2345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.989510020021545e-05,
+      "loss": 0.911,
+      "step": 2346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.989431435204362e-05,
+      "loss": 0.9366,
+      "step": 2347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.98935255869245e-05,
+      "loss": 0.9455,
+      "step": 2348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9892733905090642e-05,
+      "loss": 0.929,
+      "step": 2349
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.989193930677543e-05,
+      "loss": 0.96,
+      "step": 2350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.989114179221312e-05,
+      "loss": 0.9422,
+      "step": 2351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9890341361638837e-05,
+      "loss": 0.9305,
+      "step": 2352
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9889538015288545e-05,
+      "loss": 0.9295,
+      "step": 2353
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988873175339909e-05,
+      "loss": 1.0277,
+      "step": 2354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9887922576208155e-05,
+      "loss": 0.8651,
+      "step": 2355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9887110483954303e-05,
+      "loss": 0.9895,
+      "step": 2356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.988629547687694e-05,
+      "loss": 0.9396,
+      "step": 2357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9885477555216346e-05,
+      "loss": 0.4074,
+      "step": 2358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9884656719213645e-05,
+      "loss": 0.9467,
+      "step": 2359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9883832969110837e-05,
+      "loss": 0.9542,
+      "step": 2360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9883006305150767e-05,
+      "loss": 0.9153,
+      "step": 2361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9882176727577143e-05,
+      "loss": 0.986,
+      "step": 2362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9881344236634537e-05,
+      "loss": 0.9601,
+      "step": 2363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.988050883256837e-05,
+      "loss": 1.0071,
+      "step": 2364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9879670515624937e-05,
+      "loss": 0.9565,
+      "step": 2365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9878829286051374e-05,
+      "loss": 0.9445,
+      "step": 2366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9877985144095687e-05,
+      "loss": 0.9301,
+      "step": 2367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987713809000674e-05,
+      "loss": 0.923,
+      "step": 2368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9876288124034252e-05,
+      "loss": 0.936,
+      "step": 2369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9875435246428798e-05,
+      "loss": 0.9398,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9874579457441824e-05,
+      "loss": 0.9691,
+      "step": 2371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873720757325613e-05,
+      "loss": 0.9124,
+      "step": 2372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872859146333333e-05,
+      "loss": 1.0072,
+      "step": 2373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871994624718985e-05,
+      "loss": 0.9989,
+      "step": 2374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987112719273745e-05,
+      "loss": 0.9425,
+      "step": 2375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987025685064444e-05,
+      "loss": 0.4132,
+      "step": 2376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869383598696555e-05,
+      "loss": 0.9367,
+      "step": 2377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868507437151235e-05,
+      "loss": 0.997,
+      "step": 2378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867628366266776e-05,
+      "loss": 0.4003,
+      "step": 2379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866746386302345e-05,
+      "loss": 0.9467,
+      "step": 2380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865861497517957e-05,
+      "loss": 0.934,
+      "step": 2381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864973700174483e-05,
+      "loss": 0.9725,
+      "step": 2382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864082994533658e-05,
+      "loss": 0.3884,
+      "step": 2383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9863189380858068e-05,
+      "loss": 0.3816,
+      "step": 2384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9862292859411164e-05,
+      "loss": 0.9582,
+      "step": 2385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9861393430457248e-05,
+      "loss": 0.9836,
+      "step": 2386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986049109426148e-05,
+      "loss": 0.3854,
+      "step": 2387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.985958585108988e-05,
+      "loss": 0.9439,
+      "step": 2388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9858677701209324e-05,
+      "loss": 0.8711,
+      "step": 2389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.985776664488754e-05,
+      "loss": 0.9656,
+      "step": 2390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.985685268239312e-05,
+      "loss": 1.0065,
+      "step": 2391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9855935813995504e-05,
+      "loss": 0.9616,
+      "step": 2392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9855016039965006e-05,
+      "loss": 0.3499,
+      "step": 2393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9854093360572775e-05,
+      "loss": 0.9801,
+      "step": 2394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9853167776090832e-05,
+      "loss": 0.9592,
+      "step": 2395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9852239286792047e-05,
+      "loss": 0.3963,
+      "step": 2396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9851307892950146e-05,
+      "loss": 0.9769,
+      "step": 2397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9850373594839717e-05,
+      "loss": 0.9566,
+      "step": 2398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.98494363927362e-05,
+      "loss": 0.9542,
+      "step": 2399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9848496286915894e-05,
+      "loss": 0.8902,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9847553277655948e-05,
+      "loss": 0.9706,
+      "step": 2401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.984660736523437e-05,
+      "loss": 0.9829,
+      "step": 2402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.984565854993003e-05,
+      "loss": 0.9643,
+      "step": 2403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9844706832022646e-05,
+      "loss": 0.9744,
+      "step": 2404
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9843752211792792e-05,
+      "loss": 0.9221,
+      "step": 2405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9842794689521905e-05,
+      "loss": 0.9784,
+      "step": 2406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.984183426549227e-05,
+      "loss": 0.9264,
+      "step": 2407
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.984087093998703e-05,
+      "loss": 1.0185,
+      "step": 2408
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9839904713290186e-05,
+      "loss": 0.9137,
+      "step": 2409
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9838935585686586e-05,
+      "loss": 0.9184,
+      "step": 2410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9837963557461945e-05,
+      "loss": 0.9418,
+      "step": 2411
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.983698862890282e-05,
+      "loss": 0.9684,
+      "step": 2412
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9836010800296642e-05,
+      "loss": 0.97,
+      "step": 2413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9835030071931672e-05,
+      "loss": 1.0213,
+      "step": 2414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9834046444097046e-05,
+      "loss": 0.9374,
+      "step": 2415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9833059917082744e-05,
+      "loss": 0.9585,
+      "step": 2416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.983207049117961e-05,
+      "loss": 0.9834,
+      "step": 2417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9831078166679334e-05,
+      "loss": 0.9665,
+      "step": 2418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9830082943874464e-05,
+      "loss": 0.9404,
+      "step": 2419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.98290848230584e-05,
+      "loss": 0.959,
+      "step": 2420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.98280838045254e-05,
+      "loss": 0.4105,
+      "step": 2421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9827079888570573e-05,
+      "loss": 0.9951,
+      "step": 2422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.982607307548989e-05,
+      "loss": 0.9295,
+      "step": 2423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9825063365580165e-05,
+      "loss": 0.9342,
+      "step": 2424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9824050759139068e-05,
+      "loss": 0.9755,
+      "step": 2425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9823035256465136e-05,
+      "loss": 0.3381,
+      "step": 2426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.982201685785774e-05,
+      "loss": 0.8984,
+      "step": 2427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9820995563617117e-05,
+      "loss": 0.374,
+      "step": 2428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9819971374044356e-05,
+      "loss": 1.0316,
+      "step": 2429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9818944289441402e-05,
+      "loss": 0.9626,
+      "step": 2430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9817914310111044e-05,
+      "loss": 0.9471,
+      "step": 2431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.981688143635694e-05,
+      "loss": 0.9814,
+      "step": 2432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.981584566848358e-05,
+      "loss": 0.9779,
+      "step": 2433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.981480700679633e-05,
+      "loss": 0.9415,
+      "step": 2434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.981376545160139e-05,
+      "loss": 0.8894,
+      "step": 2435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9812721003205825e-05,
+      "loss": 0.9091,
+      "step": 2436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.981167366191755e-05,
+      "loss": 0.8746,
+      "step": 2437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.981062342804533e-05,
+      "loss": 0.9332,
+      "step": 2438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.980957030189879e-05,
+      "loss": 0.9291,
+      "step": 2439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9808514283788398e-05,
+      "loss": 0.9427,
+      "step": 2440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9807455374025477e-05,
+      "loss": 0.9238,
+      "step": 2441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.980639357292221e-05,
+      "loss": 0.395,
+      "step": 2442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9805328880791626e-05,
+      "loss": 0.9814,
+      "step": 2443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9804261297947606e-05,
+      "loss": 1.0104,
+      "step": 2444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.980319082470488e-05,
+      "loss": 1.0043,
+      "step": 2445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9802117461379047e-05,
+      "loss": 0.9862,
+      "step": 2446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9801041208286535e-05,
+      "loss": 0.8913,
+      "step": 2447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9799962065744634e-05,
+      "loss": 0.9603,
+      "step": 2448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9798880034071493e-05,
+      "loss": 0.9365,
+      "step": 2449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9797795113586104e-05,
+      "loss": 0.988,
+      "step": 2450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9796707304608314e-05,
+      "loss": 1.0031,
+      "step": 2451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9795616607458817e-05,
+      "loss": 0.9317,
+      "step": 2452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9794523022459168e-05,
+      "loss": 1.0148,
+      "step": 2453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9793426549931757e-05,
+      "loss": 0.923,
+      "step": 2454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9792327190199843e-05,
+      "loss": 0.981,
+      "step": 2455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.979122494358753e-05,
+      "loss": 0.9461,
+      "step": 2456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.979011981041977e-05,
+      "loss": 0.9498,
+      "step": 2457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9789011791022365e-05,
+      "loss": 0.9581,
+      "step": 2458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.978790088572197e-05,
+      "loss": 0.9866,
+      "step": 2459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9786787094846097e-05,
+      "loss": 0.9887,
+      "step": 2460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.97856704187231e-05,
+      "loss": 0.3509,
+      "step": 2461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.978455085768219e-05,
+      "loss": 0.9697,
+      "step": 2462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9783428412053422e-05,
+      "loss": 1.0128,
+      "step": 2463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9782303082167705e-05,
+      "loss": 0.9576,
+      "step": 2464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9781174868356797e-05,
+      "loss": 0.9048,
+      "step": 2465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9780043770953306e-05,
+      "loss": 0.9354,
+      "step": 2466
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9778909790290697e-05,
+      "loss": 0.9295,
+      "step": 2467
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9777772926703276e-05,
+      "loss": 0.9285,
+      "step": 2468
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.97766331805262e-05,
+      "loss": 0.9504,
+      "step": 2469
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9775490552095485e-05,
+      "loss": 0.902,
+      "step": 2470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9774345041747982e-05,
+      "loss": 0.9668,
+      "step": 2471
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9773196649821405e-05,
+      "loss": 0.9486,
+      "step": 2472
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9772045376654308e-05,
+      "loss": 0.3669,
+      "step": 2473
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.97708912225861e-05,
+      "loss": 0.9302,
+      "step": 2474
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.976973418795704e-05,
+      "loss": 0.9274,
+      "step": 2475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9768574273108228e-05,
+      "loss": 0.9719,
+      "step": 2476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9767411478381623e-05,
+      "loss": 0.9209,
+      "step": 2477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9766245804120032e-05,
+      "loss": 0.9466,
+      "step": 2478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9765077250667104e-05,
+      "loss": 0.9783,
+      "step": 2479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.976390581836734e-05,
+      "loss": 0.9468,
+      "step": 2480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9762731507566094e-05,
+      "loss": 0.9075,
+      "step": 2481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.976155431860956e-05,
+      "loss": 0.9629,
+      "step": 2482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9760374251844793e-05,
+      "loss": 0.984,
+      "step": 2483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9759191307619677e-05,
+      "loss": 0.9634,
+      "step": 2484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.975800548628297e-05,
+      "loss": 0.9311,
+      "step": 2485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.975681678818426e-05,
+      "loss": 0.9785,
+      "step": 2486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9755625213673984e-05,
+      "loss": 0.367,
+      "step": 2487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9754430763103428e-05,
+      "loss": 0.9318,
+      "step": 2488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.975323343682474e-05,
+      "loss": 1.0139,
+      "step": 2489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9752033235190895e-05,
+      "loss": 0.9671,
+      "step": 2490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9750830158555728e-05,
+      "loss": 0.3675,
+      "step": 2491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9749624207273916e-05,
+      "loss": 0.9081,
+      "step": 2492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9748415381700988e-05,
+      "loss": 0.9577,
+      "step": 2493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.974720368219332e-05,
+      "loss": 0.8928,
+      "step": 2494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9745989109108134e-05,
+      "loss": 0.9397,
+      "step": 2495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9744771662803496e-05,
+      "loss": 1.007,
+      "step": 2496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9743551343638324e-05,
+      "loss": 0.965,
+      "step": 2497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.974232815197238e-05,
+      "loss": 0.9845,
+      "step": 2498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.974110208816627e-05,
+      "loss": 0.9891,
+      "step": 2499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9739873152581456e-05,
+      "loss": 0.971,
+      "step": 2500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9738641345580237e-05,
+      "loss": 0.9494,
+      "step": 2501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9737406667525766e-05,
+      "loss": 0.9321,
+      "step": 2502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.973616911878204e-05,
+      "loss": 0.935,
+      "step": 2503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9734928699713897e-05,
+      "loss": 0.9318,
+      "step": 2504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9733685410687027e-05,
+      "loss": 0.937,
+      "step": 2505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9732439252067967e-05,
+      "loss": 0.978,
+      "step": 2506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.973119022422409e-05,
+      "loss": 0.9845,
+      "step": 2507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9729938327523635e-05,
+      "loss": 0.9328,
+      "step": 2508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9728683562335663e-05,
+      "loss": 0.8828,
+      "step": 2509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.97274259290301e-05,
+      "loss": 0.3788,
+      "step": 2510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.97261654279777e-05,
+      "loss": 0.9514,
+      "step": 2511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.972490205955008e-05,
+      "loss": 0.9182,
+      "step": 2512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.972363582411969e-05,
+      "loss": 0.9666,
+      "step": 2513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9722366722059836e-05,
+      "loss": 0.9653,
+      "step": 2514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9721094753744655e-05,
+      "loss": 0.8975,
+      "step": 2515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9719819919549138e-05,
+      "loss": 0.9182,
+      "step": 2516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9718542219849122e-05,
+      "loss": 0.9711,
+      "step": 2517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9717261655021286e-05,
+      "loss": 0.9501,
+      "step": 2518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9715978225443147e-05,
+      "loss": 0.9267,
+      "step": 2519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9714691931493087e-05,
+      "loss": 1.0092,
+      "step": 2520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9713402773550307e-05,
+      "loss": 0.9643,
+      "step": 2521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.971211075199487e-05,
+      "loss": 0.9492,
+      "step": 2522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.971081586720767e-05,
+      "loss": 0.9993,
+      "step": 2523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9709518119570465e-05,
+      "loss": 0.967,
+      "step": 2524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9708217509465837e-05,
+      "loss": 0.917,
+      "step": 2525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.970691403727722e-05,
+      "loss": 1.004,
+      "step": 2526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.970560770338889e-05,
+      "loss": 0.8732,
+      "step": 2527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9704298508185973e-05,
+      "loss": 0.968,
+      "step": 2528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9702986452054426e-05,
+      "loss": 0.9081,
+      "step": 2529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9701671535381064e-05,
+      "loss": 0.9975,
+      "step": 2530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9700353758553536e-05,
+      "loss": 0.8707,
+      "step": 2531
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9699033121960333e-05,
+      "loss": 0.9222,
+      "step": 2532
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9697709625990793e-05,
+      "loss": 1.0052,
+      "step": 2533
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.96963832710351e-05,
+      "loss": 0.3556,
+      "step": 2534
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.969505405748428e-05,
+      "loss": 0.9385,
+      "step": 2535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9693721985730186e-05,
+      "loss": 0.9508,
+      "step": 2536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.969238705616554e-05,
+      "loss": 0.9723,
+      "step": 2537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.969104926918389e-05,
+      "loss": 0.9873,
+      "step": 2538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9689708625179627e-05,
+      "loss": 0.9667,
+      "step": 2539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9688365124547986e-05,
+      "loss": 0.9177,
+      "step": 2540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9687018767685048e-05,
+      "loss": 0.9971,
+      "step": 2541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9685669554987732e-05,
+      "loss": 0.8696,
+      "step": 2542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.96843174868538e-05,
+      "loss": 1.0297,
+      "step": 2543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9682962563681857e-05,
+      "loss": 0.9569,
+      "step": 2544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9681604785871347e-05,
+      "loss": 1.0142,
+      "step": 2545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9680244153822558e-05,
+      "loss": 0.9663,
+      "step": 2546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9678880667936617e-05,
+      "loss": 0.9518,
+      "step": 2547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9677514328615496e-05,
+      "loss": 0.9759,
+      "step": 2548
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9676145136262007e-05,
+      "loss": 0.8913,
+      "step": 2549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9674773091279794e-05,
+      "loss": 0.9923,
+      "step": 2550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.967339819407336e-05,
+      "loss": 0.9767,
+      "step": 2551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9672020445048036e-05,
+      "loss": 0.967,
+      "step": 2552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9670639844609997e-05,
+      "loss": 0.401,
+      "step": 2553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9669256393166258e-05,
+      "loss": 0.9252,
+      "step": 2554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.966787009112467e-05,
+      "loss": 0.9228,
+      "step": 2555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.966648093889394e-05,
+      "loss": 0.935,
+      "step": 2556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9665088936883596e-05,
+      "loss": 0.9969,
+      "step": 2557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.966369408550402e-05,
+      "loss": 0.9334,
+      "step": 2558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9662296385166422e-05,
+      "loss": 0.9056,
+      "step": 2559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9660895836282866e-05,
+      "loss": 1.0198,
+      "step": 2560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.965949243926624e-05,
+      "loss": 0.3753,
+      "step": 2561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.965808619453029e-05,
+      "loss": 0.3492,
+      "step": 2562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.965667710248959e-05,
+      "loss": 0.956,
+      "step": 2563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9655265163559547e-05,
+      "loss": 0.9899,
+      "step": 2564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.965385037815642e-05,
+      "loss": 0.9704,
+      "step": 2565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9652432746697306e-05,
+      "loss": 1.012,
+      "step": 2566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9651012269600133e-05,
+      "loss": 0.982,
+      "step": 2567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9649588947283674e-05,
+      "loss": 0.9015,
+      "step": 2568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9648162780167542e-05,
+      "loss": 0.3723,
+      "step": 2569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.964673376867218e-05,
+      "loss": 0.9656,
+      "step": 2570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9645301913218878e-05,
+      "loss": 0.94,
+      "step": 2571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9643867214229764e-05,
+      "loss": 0.8702,
+      "step": 2572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9642429672127802e-05,
+      "loss": 0.9663,
+      "step": 2573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9640989287336795e-05,
+      "loss": 0.9567,
+      "step": 2574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.963954606028138e-05,
+      "loss": 0.9195,
+      "step": 2575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9638099991387038e-05,
+      "loss": 0.9061,
+      "step": 2576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.963665108108008e-05,
+      "loss": 0.946,
+      "step": 2577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9635199329787672e-05,
+      "loss": 0.8839,
+      "step": 2578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9633744737937796e-05,
+      "loss": 0.9867,
+      "step": 2579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9632287305959278e-05,
+      "loss": 0.9047,
+      "step": 2580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9630827034281794e-05,
+      "loss": 1.0044,
+      "step": 2581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.962936392333584e-05,
+      "loss": 0.9764,
+      "step": 2582
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.962789797355276e-05,
+      "loss": 0.9407,
+      "step": 2583
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.962642918536473e-05,
+      "loss": 0.8837,
+      "step": 2584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9624957559204763e-05,
+      "loss": 0.986,
+      "step": 2585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.962348309550671e-05,
+      "loss": 0.9217,
+      "step": 2586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.962200579470526e-05,
+      "loss": 0.9772,
+      "step": 2587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9620525657235938e-05,
+      "loss": 0.9755,
+      "step": 2588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.96190426835351e-05,
+      "loss": 0.9525,
+      "step": 2589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9617556874039945e-05,
+      "loss": 0.9224,
+      "step": 2590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9616068229188507e-05,
+      "loss": 0.9234,
+      "step": 2591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.961457674941965e-05,
+      "loss": 0.9612,
+      "step": 2592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9613082435173078e-05,
+      "loss": 0.9174,
+      "step": 2593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.961158528688933e-05,
+      "loss": 0.3895,
+      "step": 2594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9610085305009784e-05,
+      "loss": 0.9426,
+      "step": 2595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.960858248997665e-05,
+      "loss": 0.9471,
+      "step": 2596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.960707684223297e-05,
+      "loss": 0.8957,
+      "step": 2597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9605568362222632e-05,
+      "loss": 0.9395,
+      "step": 2598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9604057050390342e-05,
+      "loss": 1.0202,
+      "step": 2599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9602542907181657e-05,
+      "loss": 0.9667,
+      "step": 2600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9601025933042962e-05,
+      "loss": 0.3977,
+      "step": 2601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.959950612842147e-05,
+      "loss": 0.3566,
+      "step": 2602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.959798349376525e-05,
+      "loss": 0.8882,
+      "step": 2603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9596458029523174e-05,
+      "loss": 0.9412,
+      "step": 2604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 0.9594,
+      "step": 2605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9593398614081206e-05,
+      "loss": 0.9418,
+      "step": 2606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.959186466378326e-05,
+      "loss": 0.9005,
+      "step": 2607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.959032788570336e-05,
+      "loss": 0.8764,
+      "step": 2608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9588788280294568e-05,
+      "loss": 0.9584,
+      "step": 2609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9587245848010766e-05,
+      "loss": 0.9285,
+      "step": 2610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9585700589306688e-05,
+      "loss": 0.9275,
+      "step": 2611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9584152504637893e-05,
+      "loss": 1.0371,
+      "step": 2612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9582601594460767e-05,
+      "loss": 0.872,
+      "step": 2613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.958104785923254e-05,
+      "loss": 0.9135,
+      "step": 2614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9579491299411263e-05,
+      "loss": 0.9338,
+      "step": 2615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9577931915455827e-05,
+      "loss": 0.9528,
+      "step": 2616
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9576369707825962e-05,
+      "loss": 0.9411,
+      "step": 2617
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9574804676982215e-05,
+      "loss": 0.9629,
+      "step": 2618
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.957323682338598e-05,
+      "loss": 0.9608,
+      "step": 2619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.957166614749947e-05,
+      "loss": 0.965,
+      "step": 2620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9570092649785746e-05,
+      "loss": 1.0813,
+      "step": 2621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.956851633070868e-05,
+      "loss": 0.9416,
+      "step": 2622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9566937190732994e-05,
+      "loss": 0.9794,
+      "step": 2623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9565355230324238e-05,
+      "loss": 0.9312,
+      "step": 2624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9563770449948782e-05,
+      "loss": 0.9261,
+      "step": 2625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9562182850073844e-05,
+      "loss": 0.9316,
+      "step": 2626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9560592431167456e-05,
+      "loss": 0.9132,
+      "step": 2627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.95589991936985e-05,
+      "loss": 0.9181,
+      "step": 2628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9557403138136672e-05,
+      "loss": 0.9343,
+      "step": 2629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.955580426495251e-05,
+      "loss": 0.969,
+      "step": 2630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9554202574617375e-05,
+      "loss": 0.9437,
+      "step": 2631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9552598067603466e-05,
+      "loss": 0.9352,
+      "step": 2632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9550990744383805e-05,
+      "loss": 0.8712,
+      "step": 2633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.954938060543225e-05,
+      "loss": 0.9724,
+      "step": 2634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9547767651223486e-05,
+      "loss": 0.3813,
+      "step": 2635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.954615188223303e-05,
+      "loss": 0.3876,
+      "step": 2636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9544533298937222e-05,
+      "loss": 0.9905,
+      "step": 2637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9542911901813247e-05,
+      "loss": 0.8834,
+      "step": 2638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.95412876913391e-05,
+      "loss": 0.9874,
+      "step": 2639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.953966066799362e-05,
+      "loss": 0.9311,
+      "step": 2640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9538030832256468e-05,
+      "loss": 0.9861,
+      "step": 2641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9536398184608143e-05,
+      "loss": 0.9528,
+      "step": 2642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.953476272552996e-05,
+      "loss": 0.9695,
+      "step": 2643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9533124455504073e-05,
+      "loss": 0.943,
+      "step": 2644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9531483375013457e-05,
+      "loss": 0.413,
+      "step": 2645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9529839484541925e-05,
+      "loss": 0.9668,
+      "step": 2646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.952819278457411e-05,
+      "loss": 0.9493,
+      "step": 2647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9526543275595476e-05,
+      "loss": 0.8851,
+      "step": 2648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.952489095809232e-05,
+      "loss": 0.9738,
+      "step": 2649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9523235832551752e-05,
+      "loss": 0.9229,
+      "step": 2650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9521577899461732e-05,
+      "loss": 0.9545,
+      "step": 2651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.951991715931103e-05,
+      "loss": 1.0013,
+      "step": 2652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.951825361258925e-05,
+      "loss": 0.9736,
+      "step": 2653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9516587259786824e-05,
+      "loss": 0.9226,
+      "step": 2654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9514918101395012e-05,
+      "loss": 0.8739,
+      "step": 2655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9513246137905896e-05,
+      "loss": 0.9813,
+      "step": 2656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9511571369812388e-05,
+      "loss": 0.9201,
+      "step": 2657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9509893797608228e-05,
+      "loss": 0.8617,
+      "step": 2658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9508213421787987e-05,
+      "loss": 0.9511,
+      "step": 2659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.950653024284705e-05,
+      "loss": 0.9142,
+      "step": 2660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9504844261281638e-05,
+      "loss": 0.9426,
+      "step": 2661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9503155477588794e-05,
+      "loss": 0.9584,
+      "step": 2662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9501463892266396e-05,
+      "loss": 0.9256,
+      "step": 2663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.949976950581314e-05,
+      "loss": 1.0026,
+      "step": 2664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.949807231872854e-05,
+      "loss": 0.9367,
+      "step": 2665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.949637233151295e-05,
+      "loss": 0.8926,
+      "step": 2666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9494669544667545e-05,
+      "loss": 0.9203,
+      "step": 2667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9492963958694326e-05,
+      "loss": 0.903,
+      "step": 2668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.949125557409611e-05,
+      "loss": 0.9363,
+      "step": 2669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9489544391376555e-05,
+      "loss": 0.9117,
+      "step": 2670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.948783041104013e-05,
+      "loss": 0.9372,
+      "step": 2671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9486113633592136e-05,
+      "loss": 0.9746,
+      "step": 2672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.94843940595387e-05,
+      "loss": 0.949,
+      "step": 2673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9482671689386766e-05,
+      "loss": 0.8996,
+      "step": 2674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.948094652364411e-05,
+      "loss": 0.9907,
+      "step": 2675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9479218562819326e-05,
+      "loss": 0.9541,
+      "step": 2676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9477487807421837e-05,
+      "loss": 0.8952,
+      "step": 2677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9475754257961887e-05,
+      "loss": 0.9636,
+      "step": 2678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9474017914950546e-05,
+      "loss": 0.9576,
+      "step": 2679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9472278778899704e-05,
+      "loss": 0.9581,
+      "step": 2680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9470536850322076e-05,
+      "loss": 0.9824,
+      "step": 2681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.94687921297312e-05,
+      "loss": 0.9101,
+      "step": 2682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9467044617641445e-05,
+      "loss": 0.9833,
+      "step": 2683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9465294314567987e-05,
+      "loss": 0.9043,
+      "step": 2684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.946354122102684e-05,
+      "loss": 0.9171,
+      "step": 2685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.946178533753483e-05,
+      "loss": 0.9205,
+      "step": 2686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.946002666460961e-05,
+      "loss": 0.9658,
+      "step": 2687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9458265202769656e-05,
+      "loss": 0.8765,
+      "step": 2688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.945650095253427e-05,
+      "loss": 0.9151,
+      "step": 2689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.945473391442356e-05,
+      "loss": 0.9146,
+      "step": 2690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9452964088958483e-05,
+      "loss": 0.9246,
+      "step": 2691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.945119147666079e-05,
+      "loss": 0.9732,
+      "step": 2692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9449416078053067e-05,
+      "loss": 0.9274,
+      "step": 2693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9447637893658727e-05,
+      "loss": 0.9394,
+      "step": 2694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.944585692400199e-05,
+      "loss": 0.9431,
+      "step": 2695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9444073169607907e-05,
+      "loss": 0.9507,
+      "step": 2696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9442286631002348e-05,
+      "loss": 0.9409,
+      "step": 2697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9440497308712e-05,
+      "loss": 0.9612,
+      "step": 2698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.943870520326438e-05,
+      "loss": 0.8676,
+      "step": 2699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9436910315187815e-05,
+      "loss": 0.9805,
+      "step": 2700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9435112645011462e-05,
+      "loss": 0.9083,
+      "step": 2701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.943331219326528e-05,
+      "loss": 0.9918,
+      "step": 2702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9431508960480075e-05,
+      "loss": 0.9995,
+      "step": 2703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9429702947187455e-05,
+      "loss": 0.8881,
+      "step": 2704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.942789415391985e-05,
+      "loss": 0.8812,
+      "step": 2705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.942608258121051e-05,
+      "loss": 0.9597,
+      "step": 2706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9424268229593507e-05,
+      "loss": 0.8788,
+      "step": 2707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.942245109960373e-05,
+      "loss": 0.9262,
+      "step": 2708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9420631191776892e-05,
+      "loss": 0.9958,
+      "step": 2709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9418808506649515e-05,
+      "loss": 0.9153,
+      "step": 2710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.941698304475895e-05,
+      "loss": 0.953,
+      "step": 2711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9415154806643358e-05,
+      "loss": 0.9333,
+      "step": 2712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9413323792841726e-05,
+      "loss": 0.9322,
+      "step": 2713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.941149000389386e-05,
+      "loss": 1.0025,
+      "step": 2714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.940965344034037e-05,
+      "loss": 0.9449,
+      "step": 2715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.94078141027227e-05,
+      "loss": 0.9752,
+      "step": 2716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.940597199158311e-05,
+      "loss": 0.9238,
+      "step": 2717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9404127107464662e-05,
+      "loss": 0.9692,
+      "step": 2718
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9402279450911255e-05,
+      "loss": 0.934,
+      "step": 2719
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.94004290224676e-05,
+      "loss": 0.9964,
+      "step": 2720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.939857582267922e-05,
+      "loss": 0.3924,
+      "step": 2721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9396719852092457e-05,
+      "loss": 0.998,
+      "step": 2722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9394861111254467e-05,
+      "loss": 0.9688,
+      "step": 2723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.939299960071323e-05,
+      "loss": 1.0011,
+      "step": 2724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9391135321017542e-05,
+      "loss": 0.8818,
+      "step": 2725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.938926827271701e-05,
+      "loss": 0.9579,
+      "step": 2726
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.938739845636205e-05,
+      "loss": 0.9208,
+      "step": 2727
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.938552587250392e-05,
+      "loss": 0.9618,
+      "step": 2728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9383650521694662e-05,
+      "loss": 0.9679,
+      "step": 2729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.938177240448716e-05,
+      "loss": 0.9688,
+      "step": 2730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9379891521435098e-05,
+      "loss": 0.9121,
+      "step": 2731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9378007873092975e-05,
+      "loss": 0.3756,
+      "step": 2732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9376121460016123e-05,
+      "loss": 0.9452,
+      "step": 2733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9374232282760664e-05,
+      "loss": 0.9172,
+      "step": 2734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.937234034188355e-05,
+      "loss": 0.9467,
+      "step": 2735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9370445637942552e-05,
+      "loss": 0.9687,
+      "step": 2736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9368548171496244e-05,
+      "loss": 0.8133,
+      "step": 2737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9366647943104014e-05,
+      "loss": 0.8829,
+      "step": 2738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9364744953326077e-05,
+      "loss": 0.9476,
+      "step": 2739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.936283920272345e-05,
+      "loss": 0.8584,
+      "step": 2740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9360930691857966e-05,
+      "loss": 0.8558,
+      "step": 2741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.935901942129228e-05,
+      "loss": 0.9791,
+      "step": 2742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.935710539158985e-05,
+      "loss": 0.9519,
+      "step": 2743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9355188603314956e-05,
+      "loss": 0.9498,
+      "step": 2744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.935326905703268e-05,
+      "loss": 0.9128,
+      "step": 2745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9351346753308933e-05,
+      "loss": 0.9418,
+      "step": 2746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9349421692710428e-05,
+      "loss": 0.9336,
+      "step": 2747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9347493875804686e-05,
+      "loss": 0.967,
+      "step": 2748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9345563303160056e-05,
+      "loss": 0.9686,
+      "step": 2749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9343629975345687e-05,
+      "loss": 0.9127,
+      "step": 2750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9341693892931544e-05,
+      "loss": 0.9319,
+      "step": 2751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9339755056488405e-05,
+      "loss": 0.9479,
+      "step": 2752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9337813466587864e-05,
+      "loss": 0.966,
+      "step": 2753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9335869123802313e-05,
+      "loss": 0.9739,
+      "step": 2754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.933392202870497e-05,
+      "loss": 0.963,
+      "step": 2755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9331972181869857e-05,
+      "loss": 0.9624,
+      "step": 2756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9330019583871813e-05,
+      "loss": 0.9385,
+      "step": 2757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.932806423528648e-05,
+      "loss": 0.8951,
+      "step": 2758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9326106136690322e-05,
+      "loss": 0.9148,
+      "step": 2759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9324145288660595e-05,
+      "loss": 0.9056,
+      "step": 2760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9322181691775387e-05,
+      "loss": 0.9628,
+      "step": 2761
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9320215346613586e-05,
+      "loss": 0.9258,
+      "step": 2762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9318246253754886e-05,
+      "loss": 0.9643,
+      "step": 2763
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.93162744137798e-05,
+      "loss": 0.9188,
+      "step": 2764
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.931429982726965e-05,
+      "loss": 0.9061,
+      "step": 2765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9312322494806556e-05,
+      "loss": 0.9651,
+      "step": 2766
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9310342416973468e-05,
+      "loss": 0.9288,
+      "step": 2767
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.930835959435412e-05,
+      "loss": 0.9844,
+      "step": 2768
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9306374027533078e-05,
+      "loss": 1.0058,
+      "step": 2769
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9304385717095708e-05,
+      "loss": 0.964,
+      "step": 2770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.930239466362818e-05,
+      "loss": 0.9395,
+      "step": 2771
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9300400867717484e-05,
+      "loss": 0.9762,
+      "step": 2772
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9298404329951404e-05,
+      "loss": 0.9673,
+      "step": 2773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9296405050918546e-05,
+      "loss": 0.9442,
+      "step": 2774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9294403031208317e-05,
+      "loss": 0.983,
+      "step": 2775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.929239827141093e-05,
+      "loss": 0.9916,
+      "step": 2776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.929039077211741e-05,
+      "loss": 0.9585,
+      "step": 2777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9288380533919597e-05,
+      "loss": 0.3618,
+      "step": 2778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.928636755741012e-05,
+      "loss": 0.9229,
+      "step": 2779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.928435184318243e-05,
+      "loss": 0.9529,
+      "step": 2780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9282333391830777e-05,
+      "loss": 0.9489,
+      "step": 2781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9280312203950228e-05,
+      "loss": 1.0185,
+      "step": 2782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9278288280136647e-05,
+      "loss": 0.9449,
+      "step": 2783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.927626162098671e-05,
+      "loss": 0.9203,
+      "step": 2784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9274232227097885e-05,
+      "loss": 0.9459,
+      "step": 2785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.927220009906848e-05,
+      "loss": 0.9227,
+      "step": 2786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.927016523749757e-05,
+      "loss": 0.9312,
+      "step": 2787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9268127642985058e-05,
+      "loss": 0.9571,
+      "step": 2788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9266087316131655e-05,
+      "loss": 0.8642,
+      "step": 2789
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9264044257538864e-05,
+      "loss": 0.9786,
+      "step": 2790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9261998467809e-05,
+      "loss": 0.9479,
+      "step": 2791
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9259949947545186e-05,
+      "loss": 0.9755,
+      "step": 2792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.925789869735134e-05,
+      "loss": 0.9397,
+      "step": 2793
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9255844717832206e-05,
+      "loss": 0.9668,
+      "step": 2794
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9253788009593308e-05,
+      "loss": 0.9507,
+      "step": 2795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9251728573240983e-05,
+      "loss": 0.9821,
+      "step": 2796
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9249666409382387e-05,
+      "loss": 0.9102,
+      "step": 2797
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9247601518625454e-05,
+      "loss": 0.9484,
+      "step": 2798
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9245533901578943e-05,
+      "loss": 0.9879,
+      "step": 2799
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9243463558852405e-05,
+      "loss": 0.9688,
+      "step": 2800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.92413904910562e-05,
+      "loss": 0.944,
+      "step": 2801
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9239314698801493e-05,
+      "loss": 0.9258,
+      "step": 2802
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9237236182700244e-05,
+      "loss": 0.9894,
+      "step": 2803
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9235154943365224e-05,
+      "loss": 0.9478,
+      "step": 2804
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9233070981410007e-05,
+      "loss": 0.8373,
+      "step": 2805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.923098429744896e-05,
+      "loss": 0.9401,
+      "step": 2806
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9228894892097267e-05,
+      "loss": 0.9464,
+      "step": 2807
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.92268027659709e-05,
+      "loss": 0.9382,
+      "step": 2808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9224707919686648e-05,
+      "loss": 0.9861,
+      "step": 2809
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.922261035386208e-05,
+      "loss": 0.9838,
+      "step": 2810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9220510069115595e-05,
+      "loss": 0.9371,
+      "step": 2811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.921840706606637e-05,
+      "loss": 0.945,
+      "step": 2812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.921630134533439e-05,
+      "loss": 0.9942,
+      "step": 2813
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9214192907540452e-05,
+      "loss": 0.9535,
+      "step": 2814
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9212081753306143e-05,
+      "loss": 0.9513,
+      "step": 2815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.920996788325385e-05,
+      "loss": 0.9116,
+      "step": 2816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9207851298006766e-05,
+      "loss": 0.8856,
+      "step": 2817
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.920573199818888e-05,
+      "loss": 0.9219,
+      "step": 2818
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9203609984424993e-05,
+      "loss": 0.9478,
+      "step": 2819
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9201485257340683e-05,
+      "loss": 0.9755,
+      "step": 2820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9199357817562347e-05,
+      "loss": 1.0169,
+      "step": 2821
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9197227665717183e-05,
+      "loss": 1.0045,
+      "step": 2822
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9195094802433175e-05,
+      "loss": 0.9257,
+      "step": 2823
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9192959228339115e-05,
+      "loss": 0.9399,
+      "step": 2824
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190820944064594e-05,
+      "loss": 0.8806,
+      "step": 2825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918867995024e-05,
+      "loss": 0.9966,
+      "step": 2826
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918653624749652e-05,
+      "loss": 0.9567,
+      "step": 2827
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918438983646614e-05,
+      "loss": 0.9786,
+      "step": 2828
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9182240717781642e-05,
+      "loss": 0.9672,
+      "step": 2829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918008889207661e-05,
+      "loss": 0.9026,
+      "step": 2830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917793435998543e-05,
+      "loss": 0.9362,
+      "step": 2831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9175777122143273e-05,
+      "loss": 0.9447,
+      "step": 2832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917361717918612e-05,
+      "loss": 0.9201,
+      "step": 2833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9171454531750745e-05,
+      "loss": 0.8951,
+      "step": 2834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9169289180474714e-05,
+      "loss": 0.9039,
+      "step": 2835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91671211259964e-05,
+      "loss": 0.8881,
+      "step": 2836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9164950368954965e-05,
+      "loss": 0.8945,
+      "step": 2837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9162776909990375e-05,
+      "loss": 0.9132,
+      "step": 2838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9160600749743384e-05,
+      "loss": 0.9735,
+      "step": 2839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9158421888855548e-05,
+      "loss": 0.926,
+      "step": 2840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9156240327969223e-05,
+      "loss": 0.951,
+      "step": 2841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9154056067727548e-05,
+      "loss": 0.9155,
+      "step": 2842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.915186910877447e-05,
+      "loss": 0.9118,
+      "step": 2843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9149679451754726e-05,
+      "loss": 0.9679,
+      "step": 2844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9147487097313854e-05,
+      "loss": 0.9458,
+      "step": 2845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145292046098175e-05,
+      "loss": 0.9915,
+      "step": 2846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914309429875482e-05,
+      "loss": 0.8839,
+      "step": 2847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140893855931705e-05,
+      "loss": 0.9345,
+      "step": 2848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138690718277542e-05,
+      "loss": 0.9334,
+      "step": 2849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913648488644184e-05,
+      "loss": 0.9549,
+      "step": 2850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134276361074907e-05,
+      "loss": 0.9275,
+      "step": 2851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132065142827834e-05,
+      "loss": 0.3941,
+      "step": 2852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912985123235251e-05,
+      "loss": 0.9448,
+      "step": 2853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912763463030162e-05,
+      "loss": 0.9497,
+      "step": 2854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125415337328644e-05,
+      "loss": 0.9704,
+      "step": 2855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912319335408785e-05,
+      "loss": 0.9884,
+      "step": 2856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9120968681234303e-05,
+      "loss": 0.9049,
+      "step": 2857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9118741319423862e-05,
+      "loss": 0.9574,
+      "step": 2858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9116511269313173e-05,
+      "loss": 1.0218,
+      "step": 2859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114278531559677e-05,
+      "loss": 0.908,
+      "step": 2860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9112043106821612e-05,
+      "loss": 0.9503,
+      "step": 2861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9109804995758003e-05,
+      "loss": 0.8831,
+      "step": 2862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910756419902867e-05,
+      "loss": 0.9376,
+      "step": 2863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910532071729422e-05,
+      "loss": 0.9306,
+      "step": 2864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9103074551216058e-05,
+      "loss": 0.9332,
+      "step": 2865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9100825701456376e-05,
+      "loss": 0.9612,
+      "step": 2866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.909857416867816e-05,
+      "loss": 0.951,
+      "step": 2867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9096319953545186e-05,
+      "loss": 0.9885,
+      "step": 2868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9094063056722015e-05,
+      "loss": 0.9423,
+      "step": 2869
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9091803478874008e-05,
+      "loss": 1.0025,
+      "step": 2870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9089541220667312e-05,
+      "loss": 0.9652,
+      "step": 2871
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.908727628276886e-05,
+      "loss": 0.9949,
+      "step": 2872
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.908500866584639e-05,
+      "loss": 0.9006,
+      "step": 2873
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9082738370568407e-05,
+      "loss": 0.9715,
+      "step": 2874
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.908046539760422e-05,
+      "loss": 1.0639,
+      "step": 2875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9078189747623934e-05,
+      "loss": 0.944,
+      "step": 2876
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9075911421298425e-05,
+      "loss": 0.92,
+      "step": 2877
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9073630419299373e-05,
+      "loss": 0.9796,
+      "step": 2878
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9071346742299238e-05,
+      "loss": 0.9441,
+      "step": 2879
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9069060390971273e-05,
+      "loss": 0.9185,
+      "step": 2880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9066771365989515e-05,
+      "loss": 1.0064,
+      "step": 2881
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.90644796680288e-05,
+      "loss": 0.9798,
+      "step": 2882
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.906218529776474e-05,
+      "loss": 0.9294,
+      "step": 2883
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.905988825587374e-05,
+      "loss": 0.9472,
+      "step": 2884
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9057588543032988e-05,
+      "loss": 0.9803,
+      "step": 2885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9055286159920466e-05,
+      "loss": 0.9296,
+      "step": 2886
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9052981107214947e-05,
+      "loss": 0.9495,
+      "step": 2887
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9050673385595977e-05,
+      "loss": 0.9411,
+      "step": 2888
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9048362995743898e-05,
+      "loss": 0.9445,
+      "step": 2889
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9046049938339837e-05,
+      "loss": 0.9258,
+      "step": 2890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.904373421406571e-05,
+      "loss": 0.8742,
+      "step": 2891
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9041415823604214e-05,
+      "loss": 0.3869,
+      "step": 2892
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9039094767638834e-05,
+      "loss": 0.9186,
+      "step": 2893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9036771046853845e-05,
+      "loss": 0.9707,
+      "step": 2894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9034444661934302e-05,
+      "loss": 0.9436,
+      "step": 2895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9032115613566047e-05,
+      "loss": 0.9149,
+      "step": 2896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.902978390243571e-05,
+      "loss": 0.9555,
+      "step": 2897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9027449529230703e-05,
+      "loss": 0.9552,
+      "step": 2898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.902511249463922e-05,
+      "loss": 0.9389,
+      "step": 2899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9022772799350248e-05,
+      "loss": 0.9918,
+      "step": 2900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9020430444053554e-05,
+      "loss": 0.9139,
+      "step": 2901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9018085429439683e-05,
+      "loss": 0.9018,
+      "step": 2902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9015737756199976e-05,
+      "loss": 0.9818,
+      "step": 2903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.901338742502655e-05,
+      "loss": 0.9751,
+      "step": 2904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9011034436612305e-05,
+      "loss": 0.9075,
+      "step": 2905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9008678791650927e-05,
+      "loss": 0.9623,
+      "step": 2906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9006320490836886e-05,
+      "loss": 0.3593,
+      "step": 2907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9003959534865437e-05,
+      "loss": 0.932,
+      "step": 2908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9001595924432606e-05,
+      "loss": 0.9624,
+      "step": 2909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.899922966023522e-05,
+      "loss": 0.971,
+      "step": 2910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8996860742970872e-05,
+      "loss": 0.9448,
+      "step": 2911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8994489173337943e-05,
+      "loss": 0.9078,
+      "step": 2912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8992114952035602e-05,
+      "loss": 0.969,
+      "step": 2913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8989738079763788e-05,
+      "loss": 0.9274,
+      "step": 2914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8987358557223232e-05,
+      "loss": 0.9903,
+      "step": 2915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.898497638511544e-05,
+      "loss": 0.9381,
+      "step": 2916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8982591564142702e-05,
+      "loss": 0.8965,
+      "step": 2917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8980204095008087e-05,
+      "loss": 0.9647,
+      "step": 2918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.897781397841545e-05,
+      "loss": 0.9274,
+      "step": 2919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8975421215069416e-05,
+      "loss": 0.9531,
+      "step": 2920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8973025805675403e-05,
+      "loss": 1.0206,
+      "step": 2921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8970627750939595e-05,
+      "loss": 0.9606,
+      "step": 2922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.896822705156897e-05,
+      "loss": 0.862,
+      "step": 2923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.896582370827128e-05,
+      "loss": 0.9857,
+      "step": 2924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8963417721755052e-05,
+      "loss": 0.9206,
+      "step": 2925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8961009092729598e-05,
+      "loss": 0.9038,
+      "step": 2926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8958597821905004e-05,
+      "loss": 0.9726,
+      "step": 2927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8956183909992144e-05,
+      "loss": 0.972,
+      "step": 2928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8953767357702655e-05,
+      "loss": 0.917,
+      "step": 2929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8951348165748973e-05,
+      "loss": 0.9122,
+      "step": 2930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.894892633484429e-05,
+      "loss": 0.9144,
+      "step": 2931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.89465018657026e-05,
+      "loss": 0.9796,
+      "step": 2932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8944074759038648e-05,
+      "loss": 0.96,
+      "step": 2933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.894164501556798e-05,
+      "loss": 0.8908,
+      "step": 2934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8939212636006902e-05,
+      "loss": 0.9185,
+      "step": 2935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8936777621072514e-05,
+      "loss": 0.9764,
+      "step": 2936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8934339971482676e-05,
+      "loss": 0.4364,
+      "step": 2937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8931899687956038e-05,
+      "loss": 1.002,
+      "step": 2938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.892945677121202e-05,
+      "loss": 0.9292,
+      "step": 2939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.892701122197082e-05,
+      "loss": 0.9533,
+      "step": 2940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8924563040953403e-05,
+      "loss": 0.998,
+      "step": 2941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8922112228881532e-05,
+      "loss": 0.3717,
+      "step": 2942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.891965878647772e-05,
+      "loss": 0.9261,
+      "step": 2943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8917202714465275e-05,
+      "loss": 0.9183,
+      "step": 2944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8914744013568273e-05,
+      "loss": 0.9345,
+      "step": 2945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.891228268451156e-05,
+      "loss": 0.9105,
+      "step": 2946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8909818728020765e-05,
+      "loss": 0.9159,
+      "step": 2947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8907352144822285e-05,
+      "loss": 0.9715,
+      "step": 2948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8904882935643293e-05,
+      "loss": 0.947,
+      "step": 2949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8902411101211747e-05,
+      "loss": 0.9807,
+      "step": 2950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.889993664225636e-05,
+      "loss": 0.9759,
+      "step": 2951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8897459559506632e-05,
+      "loss": 0.9605,
+      "step": 2952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8894979853692836e-05,
+      "loss": 0.9495,
+      "step": 2953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.889249752554601e-05,
+      "loss": 0.9349,
+      "step": 2954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.889001257579797e-05,
+      "loss": 0.3604,
+      "step": 2955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.888752500518131e-05,
+      "loss": 0.8997,
+      "step": 2956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.888503481442939e-05,
+      "loss": 0.883,
+      "step": 2957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8882542004276343e-05,
+      "loss": 0.9827,
+      "step": 2958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8880046575457072e-05,
+      "loss": 0.9163,
+      "step": 2959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8877548528707267e-05,
+      "loss": 0.9364,
+      "step": 2960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8875047864763366e-05,
+      "loss": 0.9225,
+      "step": 2961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.88725445843626e-05,
+      "loss": 0.9104,
+      "step": 2962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.887003868824295e-05,
+      "loss": 0.9777,
+      "step": 2963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8867530177143192e-05,
+      "loss": 0.9289,
+      "step": 2964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.886501905180286e-05,
+      "loss": 0.9382,
+      "step": 2965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8862505312962257e-05,
+      "loss": 0.9282,
+      "step": 2966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8859988961362455e-05,
+      "loss": 0.9608,
+      "step": 2967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.885746999774531e-05,
+      "loss": 0.9366,
+      "step": 2968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8854948422853436e-05,
+      "loss": 0.8856,
+      "step": 2969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8852424237430215e-05,
+      "loss": 0.9204,
+      "step": 2970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.884989744221981e-05,
+      "loss": 0.8802,
+      "step": 2971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8847368037967138e-05,
+      "loss": 1.0375,
+      "step": 2972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8844836025417905e-05,
+      "loss": 0.9668,
+      "step": 2973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8842301405318567e-05,
+      "loss": 0.9636,
+      "step": 2974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8839764178416354e-05,
+      "loss": 0.944,
+      "step": 2975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8837224345459276e-05,
+      "loss": 0.9012,
+      "step": 2976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8834681907196094e-05,
+      "loss": 0.4154,
+      "step": 2977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.883213686437635e-05,
+      "loss": 0.9558,
+      "step": 2978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.882958921775035e-05,
+      "loss": 0.9491,
+      "step": 2979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8827038968069163e-05,
+      "loss": 0.8959,
+      "step": 2980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8824486116084633e-05,
+      "loss": 0.9779,
+      "step": 2981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8821930662549363e-05,
+      "loss": 0.9519,
+      "step": 2982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.881937260821673e-05,
+      "loss": 0.8863,
+      "step": 2983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8816811953840877e-05,
+      "loss": 0.9392,
+      "step": 2984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.881424870017671e-05,
+      "loss": 1.0071,
+      "step": 2985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8811682847979902e-05,
+      "loss": 0.891,
+      "step": 2986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8809114398006895e-05,
+      "loss": 1.0081,
+      "step": 2987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8806543351014893e-05,
+      "loss": 0.9813,
+      "step": 2988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8803969707761866e-05,
+      "loss": 0.8979,
+      "step": 2989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8801393469006558e-05,
+      "loss": 0.9345,
+      "step": 2990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.879881463550846e-05,
+      "loss": 0.9184,
+      "step": 2991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.879623320802785e-05,
+      "loss": 0.942,
+      "step": 2992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8793649187325754e-05,
+      "loss": 0.9375,
+      "step": 2993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.879106257416397e-05,
+      "loss": 0.929,
+      "step": 2994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8788473369305058e-05,
+      "loss": 0.394,
+      "step": 2995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8785881573512345e-05,
+      "loss": 1.0571,
+      "step": 2996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8783287187549915e-05,
+      "loss": 0.9497,
+      "step": 2997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.878069021218262e-05,
+      "loss": 0.9243,
+      "step": 2998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.877809064817608e-05,
+      "loss": 0.945,
+      "step": 2999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8775488496296672e-05,
+      "loss": 0.9662,
+      "step": 3000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.877288375731154e-05,
+      "loss": 0.9581,
+      "step": 3001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8770276431988582e-05,
+      "loss": 0.9373,
+      "step": 3002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.876766652109647e-05,
+      "loss": 0.3832,
+      "step": 3003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8765054025404627e-05,
+      "loss": 0.9375,
+      "step": 3004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8762438945683252e-05,
+      "loss": 0.4045,
+      "step": 3005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.875982128270329e-05,
+      "loss": 0.9953,
+      "step": 3006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.875720103723646e-05,
+      "loss": 0.9678,
+      "step": 3007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.875457821005524e-05,
+      "loss": 0.9366,
+      "step": 3008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8751952801932866e-05,
+      "loss": 0.9529,
+      "step": 3009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8749324813643328e-05,
+      "loss": 0.9903,
+      "step": 3010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8746694245961395e-05,
+      "loss": 0.924,
+      "step": 3011
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.874406109966258e-05,
+      "loss": 0.9782,
+      "step": 3012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.874142537552316e-05,
+      "loss": 0.9762,
+      "step": 3013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.873878707432018e-05,
+      "loss": 0.952,
+      "step": 3014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8736146196831433e-05,
+      "loss": 0.9629,
+      "step": 3015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8733502743835483e-05,
+      "loss": 0.8768,
+      "step": 3016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8730856716111642e-05,
+      "loss": 0.907,
+      "step": 3017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8728208114439992e-05,
+      "loss": 0.3812,
+      "step": 3018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8725556939601365e-05,
+      "loss": 0.9751,
+      "step": 3019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8722903192377355e-05,
+      "loss": 0.9683,
+      "step": 3020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.872024687355032e-05,
+      "loss": 0.4154,
+      "step": 3021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.871758798390336e-05,
+      "loss": 0.9448,
+      "step": 3022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.871492652422035e-05,
+      "loss": 0.9886,
+      "step": 3023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8712262495285917e-05,
+      "loss": 0.9956,
+      "step": 3024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.870959589788544e-05,
+      "loss": 0.9772,
+      "step": 3025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8706926732805065e-05,
+      "loss": 0.9371,
+      "step": 3026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8704255000831688e-05,
+      "loss": 0.893,
+      "step": 3027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8701580702752963e-05,
+      "loss": 0.971,
+      "step": 3028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8698903839357304e-05,
+      "loss": 0.9539,
+      "step": 3029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.869622441143387e-05,
+      "loss": 0.9183,
+      "step": 3030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8693542419772593e-05,
+      "loss": 0.923,
+      "step": 3031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8690857865164145e-05,
+      "loss": 0.9386,
+      "step": 3032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.868817074839997e-05,
+      "loss": 0.9586,
+      "step": 3033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.868548107027225e-05,
+      "loss": 0.9956,
+      "step": 3034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8682788831573932e-05,
+      "loss": 0.9066,
+      "step": 3035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8680094033098718e-05,
+      "loss": 0.9208,
+      "step": 3036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8677396675641062e-05,
+      "loss": 1.001,
+      "step": 3037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8674696759996173e-05,
+      "loss": 0.8942,
+      "step": 3038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8671994286960014e-05,
+      "loss": 1.022,
+      "step": 3039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8669289257329305e-05,
+      "loss": 0.9191,
+      "step": 3040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8666581671901513e-05,
+      "loss": 0.9252,
+      "step": 3041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.866387153147486e-05,
+      "loss": 0.9399,
+      "step": 3042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8661158836848333e-05,
+      "loss": 0.9905,
+      "step": 3043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8658443588821657e-05,
+      "loss": 0.9309,
+      "step": 3044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8655725788195315e-05,
+      "loss": 0.8677,
+      "step": 3045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8653005435770546e-05,
+      "loss": 0.8909,
+      "step": 3046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8650282532349332e-05,
+      "loss": 0.9567,
+      "step": 3047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8647557078734423e-05,
+      "loss": 0.9256,
+      "step": 3048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8644829075729303e-05,
+      "loss": 0.9665,
+      "step": 3049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.864209852413822e-05,
+      "loss": 0.9289,
+      "step": 3050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.863936542476617e-05,
+      "loss": 0.8635,
+      "step": 3051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8636629778418894e-05,
+      "loss": 0.9377,
+      "step": 3052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.863389158590289e-05,
+      "loss": 0.9185,
+      "step": 3053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8631150848025414e-05,
+      "loss": 0.9575,
+      "step": 3054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.862840756559446e-05,
+      "loss": 0.9247,
+      "step": 3055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8625661739418767e-05,
+      "loss": 0.9012,
+      "step": 3056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8622913370307846e-05,
+      "loss": 0.9645,
+      "step": 3057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8620162459071936e-05,
+      "loss": 0.9357,
+      "step": 3058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8617409006522042e-05,
+      "loss": 0.9827,
+      "step": 3059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8614653013469905e-05,
+      "loss": 0.9806,
+      "step": 3060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.861189448072802e-05,
+      "loss": 0.9456,
+      "step": 3061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8609133409109637e-05,
+      "loss": 0.8895,
+      "step": 3062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8606369799428744e-05,
+      "loss": 0.9519,
+      "step": 3063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8603603652500085e-05,
+      "loss": 0.8763,
+      "step": 3064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.860083496913915e-05,
+      "loss": 0.8641,
+      "step": 3065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8598063750162166e-05,
+      "loss": 1.0189,
+      "step": 3066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.859528999638613e-05,
+      "loss": 0.884,
+      "step": 3067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8592513708628767e-05,
+      "loss": 0.9525,
+      "step": 3068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8589734887708556e-05,
+      "loss": 0.944,
+      "step": 3069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8586953534444726e-05,
+      "loss": 0.983,
+      "step": 3070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8584169649657244e-05,
+      "loss": 0.9097,
+      "step": 3071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8581383234166834e-05,
+      "loss": 0.9313,
+      "step": 3072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.857859428879495e-05,
+      "loss": 0.9421,
+      "step": 3073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8575802814363816e-05,
+      "loss": 0.9268,
+      "step": 3074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8573008811696378e-05,
+      "loss": 0.9749,
+      "step": 3075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.857021228161634e-05,
+      "loss": 0.9314,
+      "step": 3076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.856741322494815e-05,
+      "loss": 0.9186,
+      "step": 3077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8564611642517e-05,
+      "loss": 0.9698,
+      "step": 3078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8561807535148818e-05,
+      "loss": 0.8907,
+      "step": 3079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8559000903670293e-05,
+      "loss": 0.973,
+      "step": 3080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8556191748908845e-05,
+      "loss": 0.9685,
+      "step": 3081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8553380071692645e-05,
+      "loss": 0.9351,
+      "step": 3082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8550565872850602e-05,
+      "loss": 0.9727,
+      "step": 3083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8547749153212373e-05,
+      "loss": 0.9473,
+      "step": 3084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8544929913608353e-05,
+      "loss": 1.0132,
+      "step": 3085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8542108154869686e-05,
+      "loss": 0.9785,
+      "step": 3086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8539283877828257e-05,
+      "loss": 0.9323,
+      "step": 3087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8536457083316692e-05,
+      "loss": 0.9503,
+      "step": 3088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8533627772168362e-05,
+      "loss": 0.8878,
+      "step": 3089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.853079594521737e-05,
+      "loss": 0.9094,
+      "step": 3090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8527961603298572e-05,
+      "loss": 0.9277,
+      "step": 3091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8525124747247566e-05,
+      "loss": 0.9887,
+      "step": 3092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.852228537790068e-05,
+      "loss": 0.9656,
+      "step": 3093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.851944349609499e-05,
+      "loss": 0.3616,
+      "step": 3094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8516599102668324e-05,
+      "loss": 0.9622,
+      "step": 3095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8513752198459224e-05,
+      "loss": 0.8929,
+      "step": 3096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.851090278430699e-05,
+      "loss": 0.8792,
+      "step": 3097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.850805086105167e-05,
+      "loss": 0.979,
+      "step": 3098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.850519642953403e-05,
+      "loss": 0.9619,
+      "step": 3099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.850233949059559e-05,
+      "loss": 0.9006,
+      "step": 3100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8499480045078602e-05,
+      "loss": 0.9821,
+      "step": 3101
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8496618093826064e-05,
+      "loss": 0.9036,
+      "step": 3102
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.849375363768171e-05,
+      "loss": 0.913,
+      "step": 3103
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8490886677490007e-05,
+      "loss": 0.8896,
+      "step": 3104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8488017214096173e-05,
+      "loss": 0.946,
+      "step": 3105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8485145248346147e-05,
+      "loss": 0.4072,
+      "step": 3106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.848227078108662e-05,
+      "loss": 0.9403,
+      "step": 3107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.847939381316501e-05,
+      "loss": 0.9093,
+      "step": 3108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8476514345429485e-05,
+      "loss": 0.9413,
+      "step": 3109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8473632378728932e-05,
+      "loss": 0.9562,
+      "step": 3110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8470747913912993e-05,
+      "loss": 0.9385,
+      "step": 3111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8467860951832035e-05,
+      "loss": 0.9462,
+      "step": 3112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8464971493337167e-05,
+      "loss": 0.9438,
+      "step": 3113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8462079539280233e-05,
+      "loss": 0.9891,
+      "step": 3114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8459185090513802e-05,
+      "loss": 0.9912,
+      "step": 3115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8456288147891196e-05,
+      "loss": 0.9036,
+      "step": 3116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8453388712266464e-05,
+      "loss": 0.9844,
+      "step": 3117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8450486784494384e-05,
+      "loss": 0.9761,
+      "step": 3118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.844758236543048e-05,
+      "loss": 1.026,
+      "step": 3119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8444675455931006e-05,
+      "loss": 0.9613,
+      "step": 3120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8441766056852947e-05,
+      "loss": 0.9278,
+      "step": 3121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8438854169054022e-05,
+      "loss": 0.9483,
+      "step": 3122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8435939793392686e-05,
+      "loss": 0.9573,
+      "step": 3123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8433022930728132e-05,
+      "loss": 0.9068,
+      "step": 3124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8430103581920278e-05,
+      "loss": 0.9619,
+      "step": 3125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.842718174782978e-05,
+      "loss": 0.8825,
+      "step": 3126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8424257429318027e-05,
+      "loss": 1.0043,
+      "step": 3127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8421330627247137e-05,
+      "loss": 0.9756,
+      "step": 3128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.841840134247996e-05,
+      "loss": 0.9176,
+      "step": 3129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8415469575880078e-05,
+      "loss": 0.9428,
+      "step": 3130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.8248,
+      "step": 3131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.840959860064021e-05,
+      "loss": 0.9625,
+      "step": 3132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.840665939373104e-05,
+      "loss": 1.0249,
+      "step": 3133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8403717708450823e-05,
+      "loss": 0.9361,
+      "step": 3134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8400773545666788e-05,
+      "loss": 0.9221,
+      "step": 3135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8397826906246913e-05,
+      "loss": 0.9559,
+      "step": 3136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8394877791059895e-05,
+      "loss": 0.9647,
+      "step": 3137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8391926200975162e-05,
+      "loss": 0.4195,
+      "step": 3138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.838897213686288e-05,
+      "loss": 0.8897,
+      "step": 3139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.838601559959393e-05,
+      "loss": 0.9025,
+      "step": 3140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8383056590039932e-05,
+      "loss": 0.8827,
+      "step": 3141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8380095109073236e-05,
+      "loss": 0.9844,
+      "step": 3142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8377131157566917e-05,
+      "loss": 0.9528,
+      "step": 3143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8374164736394777e-05,
+      "loss": 0.9798,
+      "step": 3144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8371195846431355e-05,
+      "loss": 0.9385,
+      "step": 3145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8368224488551898e-05,
+      "loss": 1.0039,
+      "step": 3146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.83652506636324e-05,
+      "loss": 0.9326,
+      "step": 3147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8362274372549577e-05,
+      "loss": 0.8871,
+      "step": 3148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.835929561618087e-05,
+      "loss": 0.9088,
+      "step": 3149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.835631439540445e-05,
+      "loss": 0.9632,
+      "step": 3150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.83533307110992e-05,
+      "loss": 0.902,
+      "step": 3151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.835034456414476e-05,
+      "loss": 0.8714,
+      "step": 3152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8347355955421462e-05,
+      "loss": 0.8602,
+      "step": 3153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8344364885810388e-05,
+      "loss": 0.9519,
+      "step": 3154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8341371356193334e-05,
+      "loss": 1.0344,
+      "step": 3155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8338375367452822e-05,
+      "loss": 0.9471,
+      "step": 3156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8335376920472098e-05,
+      "loss": 0.9648,
+      "step": 3157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8332376016135146e-05,
+      "loss": 0.8887,
+      "step": 3158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.832937265532665e-05,
+      "loss": 0.9448,
+      "step": 3159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8326366838932048e-05,
+      "loss": 0.9927,
+      "step": 3160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8323358567837474e-05,
+      "loss": 0.3924,
+      "step": 3161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.83203478429298e-05,
+      "loss": 0.9464,
+      "step": 3162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.831733466509662e-05,
+      "loss": 0.9716,
+      "step": 3163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8314319035226254e-05,
+      "loss": 0.9175,
+      "step": 3164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8311300954207737e-05,
+      "loss": 0.9347,
+      "step": 3165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8308280422930832e-05,
+      "loss": 0.8972,
+      "step": 3166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.830525744228602e-05,
+      "loss": 0.9615,
+      "step": 3167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8302232013164518e-05,
+      "loss": 0.9039,
+      "step": 3168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.829920413645824e-05,
+      "loss": 0.9887,
+      "step": 3169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8296173813059844e-05,
+      "loss": 0.8777,
+      "step": 3170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.82931410438627e-05,
+      "loss": 1.0049,
+      "step": 3171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8290105829760902e-05,
+      "loss": 0.9121,
+      "step": 3172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8287068171649257e-05,
+      "loss": 0.895,
+      "step": 3173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.82840280704233e-05,
+      "loss": 0.9177,
+      "step": 3174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.828098552697929e-05,
+      "loss": 0.8369,
+      "step": 3175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8277940542214195e-05,
+      "loss": 0.9514,
+      "step": 3176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.827489311702571e-05,
+      "loss": 0.9095,
+      "step": 3177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8271843252312248e-05,
+      "loss": 0.9269,
+      "step": 3178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8268790948972942e-05,
+      "loss": 0.9461,
+      "step": 3179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8265736207907637e-05,
+      "loss": 0.9459,
+      "step": 3180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8262679030016913e-05,
+      "loss": 0.9366,
+      "step": 3181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.825961941620205e-05,
+      "loss": 0.9501,
+      "step": 3182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8256557367365057e-05,
+      "loss": 0.9227,
+      "step": 3183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8253492884408658e-05,
+      "loss": 0.9327,
+      "step": 3184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.825042596823629e-05,
+      "loss": 0.3881,
+      "step": 3185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.824735661975212e-05,
+      "loss": 0.9978,
+      "step": 3186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.824428483986102e-05,
+      "loss": 0.9381,
+      "step": 3187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8241210629468586e-05,
+      "loss": 0.9968,
+      "step": 3188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.823813398948112e-05,
+      "loss": 0.3967,
+      "step": 3189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8235054920805653e-05,
+      "loss": 0.9824,
+      "step": 3190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8231973424349924e-05,
+      "loss": 0.9434,
+      "step": 3191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8228889501022395e-05,
+      "loss": 0.9479,
+      "step": 3192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8225803151732236e-05,
+      "loss": 0.9862,
+      "step": 3193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.822271437738933e-05,
+      "loss": 0.3556,
+      "step": 3194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.821962317890429e-05,
+      "loss": 0.9262,
+      "step": 3195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8216529557188424e-05,
+      "loss": 0.8225,
+      "step": 3196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.821343351315377e-05,
+      "loss": 0.9448,
+      "step": 3197
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8210335047713074e-05,
+      "loss": 0.8588,
+      "step": 3198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8207234161779793e-05,
+      "loss": 0.8855,
+      "step": 3199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.82041308562681e-05,
+      "loss": 0.947,
+      "step": 3200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.820102513209289e-05,
+      "loss": 0.9701,
+      "step": 3201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.819791699016975e-05,
+      "loss": 0.9959,
+      "step": 3202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8194806431415005e-05,
+      "loss": 0.891,
+      "step": 3203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8191693456745673e-05,
+      "loss": 0.972,
+      "step": 3204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.818857806707949e-05,
+      "loss": 0.9288,
+      "step": 3205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8185460263334914e-05,
+      "loss": 0.9651,
+      "step": 3206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.81823400464311e-05,
+      "loss": 0.911,
+      "step": 3207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.817921741728792e-05,
+      "loss": 0.9132,
+      "step": 3208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.817609237682596e-05,
+      "loss": 0.9288,
+      "step": 3209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.817296492596651e-05,
+      "loss": 0.9253,
+      "step": 3210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.816983506563158e-05,
+      "loss": 0.9335,
+      "step": 3211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.816670279674389e-05,
+      "loss": 0.9411,
+      "step": 3212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8163568120226856e-05,
+      "loss": 0.9531,
+      "step": 3213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8160431037004618e-05,
+      "loss": 0.8717,
+      "step": 3214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.815729154800202e-05,
+      "loss": 0.9671,
+      "step": 3215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.815414965414462e-05,
+      "loss": 0.9379,
+      "step": 3216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8151005356358673e-05,
+      "loss": 0.9828,
+      "step": 3217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8147858655571156e-05,
+      "loss": 0.9244,
+      "step": 3218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.814470955270975e-05,
+      "loss": 0.9413,
+      "step": 3219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8141558048702844e-05,
+      "loss": 0.4005,
+      "step": 3220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.813840414447953e-05,
+      "loss": 0.927,
+      "step": 3221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8135247840969615e-05,
+      "loss": 0.9265,
+      "step": 3222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8132089139103612e-05,
+      "loss": 0.8911,
+      "step": 3223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8128928039812736e-05,
+      "loss": 0.3681,
+      "step": 3224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8125764544028913e-05,
+      "loss": 0.883,
+      "step": 3225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8122598652684774e-05,
+      "loss": 0.8744,
+      "step": 3226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8119430366713658e-05,
+      "loss": 0.9763,
+      "step": 3227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8116259687049612e-05,
+      "loss": 0.9339,
+      "step": 3228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8113086614627377e-05,
+      "loss": 0.9451,
+      "step": 3229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8109911150382417e-05,
+      "loss": 0.9649,
+      "step": 3230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8106733295250885e-05,
+      "loss": 0.9246,
+      "step": 3231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8103553050169652e-05,
+      "loss": 0.9334,
+      "step": 3232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.810037041607628e-05,
+      "loss": 0.9467,
+      "step": 3233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.809718539390905e-05,
+      "loss": 0.966,
+      "step": 3234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8093997984606936e-05,
+      "loss": 0.9202,
+      "step": 3235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.809080818910962e-05,
+      "loss": 0.9534,
+      "step": 3236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8087616008357488e-05,
+      "loss": 0.9689,
+      "step": 3237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8084421443291633e-05,
+      "loss": 0.9143,
+      "step": 3238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8081224494853834e-05,
+      "loss": 0.931,
+      "step": 3239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8078025163986595e-05,
+      "loss": 0.9295,
+      "step": 3240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.807482345163311e-05,
+      "loss": 1.0268,
+      "step": 3241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8071619358737274e-05,
+      "loss": 0.9188,
+      "step": 3242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8068412886243692e-05,
+      "loss": 0.9359,
+      "step": 3243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.806520403509766e-05,
+      "loss": 0.8977,
+      "step": 3244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8061992806245186e-05,
+      "loss": 0.8984,
+      "step": 3245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.805877920063297e-05,
+      "loss": 0.9241,
+      "step": 3246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.805556321920842e-05,
+      "loss": 0.9044,
+      "step": 3247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8052344862919637e-05,
+      "loss": 0.9095,
+      "step": 3248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.804912413271543e-05,
+      "loss": 0.949,
+      "step": 3249
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8045901029545304e-05,
+      "loss": 0.8965,
+      "step": 3250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8042675554359456e-05,
+      "loss": 0.907,
+      "step": 3251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.80394477081088e-05,
+      "loss": 0.9232,
+      "step": 3252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8036217491744935e-05,
+      "loss": 0.858,
+      "step": 3253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.803298490622016e-05,
+      "loss": 0.9244,
+      "step": 3254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8029749952487474e-05,
+      "loss": 1.0402,
+      "step": 3255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8026512631500583e-05,
+      "loss": 0.9724,
+      "step": 3256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8023272944213875e-05,
+      "loss": 0.928,
+      "step": 3257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.802003089158245e-05,
+      "loss": 0.9451,
+      "step": 3258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8016786474562093e-05,
+      "loss": 0.9084,
+      "step": 3259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8013539694109293e-05,
+      "loss": 0.9432,
+      "step": 3260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8010290551181237e-05,
+      "loss": 0.9816,
+      "step": 3261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8007039046735807e-05,
+      "loss": 0.9397,
+      "step": 3262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.800378518173158e-05,
+      "loss": 0.9472,
+      "step": 3263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8000528957127828e-05,
+      "loss": 0.9552,
+      "step": 3264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7997270373884523e-05,
+      "loss": 1.0117,
+      "step": 3265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7994009432962326e-05,
+      "loss": 0.9324,
+      "step": 3266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7990746135322593e-05,
+      "loss": 0.9229,
+      "step": 3267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.798748048192739e-05,
+      "loss": 0.8987,
+      "step": 3268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.798421247373945e-05,
+      "loss": 0.891,
+      "step": 3269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7980942111722234e-05,
+      "loss": 0.4003,
+      "step": 3270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7977669396839866e-05,
+      "loss": 0.94,
+      "step": 3271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.797439433005718e-05,
+      "loss": 0.3928,
+      "step": 3272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7971116912339698e-05,
+      "loss": 0.9597,
+      "step": 3273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7967837144653643e-05,
+      "loss": 0.9384,
+      "step": 3274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.796455502796592e-05,
+      "loss": 0.9491,
+      "step": 3275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.796127056324413e-05,
+      "loss": 0.9271,
+      "step": 3276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.795798375145657e-05,
+      "loss": 0.9599,
+      "step": 3277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7954694593572225e-05,
+      "loss": 0.9302,
+      "step": 3278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.795140309056078e-05,
+      "loss": 0.9182,
+      "step": 3279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.794810924339259e-05,
+      "loss": 0.9389,
+      "step": 3280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7944813053038733e-05,
+      "loss": 0.9993,
+      "step": 3281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7941514520470944e-05,
+      "loss": 0.8798,
+      "step": 3282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.793821364666168e-05,
+      "loss": 0.9094,
+      "step": 3283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7934910432584057e-05,
+      "loss": 0.9394,
+      "step": 3284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.793160487921191e-05,
+      "loss": 0.927,
+      "step": 3285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.792829698751975e-05,
+      "loss": 0.906,
+      "step": 3286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7924986758482768e-05,
+      "loss": 0.9546,
+      "step": 3287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7921674193076865e-05,
+      "loss": 0.9154,
+      "step": 3288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7918359292278614e-05,
+      "loss": 0.9171,
+      "step": 3289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7915042057065283e-05,
+      "loss": 0.9949,
+      "step": 3290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.791172248841483e-05,
+      "loss": 0.9509,
+      "step": 3291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7908400587305896e-05,
+      "loss": 0.9383,
+      "step": 3292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.790507635471781e-05,
+      "loss": 0.9283,
+      "step": 3293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.79017497916306e-05,
+      "loss": 0.9678,
+      "step": 3294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.789842089902496e-05,
+      "loss": 0.4103,
+      "step": 3295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7895089677882288e-05,
+      "loss": 0.8767,
+      "step": 3296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7891756129184664e-05,
+      "loss": 0.9398,
+      "step": 3297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7888420253914854e-05,
+      "loss": 0.9166,
+      "step": 3298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.78850820530563e-05,
+      "loss": 0.9058,
+      "step": 3299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.788174152759315e-05,
+      "loss": 0.9282,
+      "step": 3300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.787839867851022e-05,
+      "loss": 0.9334,
+      "step": 3301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7875053506793015e-05,
+      "loss": 0.9496,
+      "step": 3302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7871706013427732e-05,
+      "loss": 0.9312,
+      "step": 3303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7868356199401242e-05,
+      "loss": 0.9189,
+      "step": 3304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7865004065701107e-05,
+      "loss": 0.9421,
+      "step": 3305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.786164961331557e-05,
+      "loss": 0.9385,
+      "step": 3306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7858292843233564e-05,
+      "loss": 0.8707,
+      "step": 3307
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7854933756444692e-05,
+      "loss": 0.8497,
+      "step": 3308
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7851572353939253e-05,
+      "loss": 0.9461,
+      "step": 3309
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.784820863670822e-05,
+      "loss": 0.9536,
+      "step": 3310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7844842605743256e-05,
+      "loss": 1.0062,
+      "step": 3311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.78414742620367e-05,
+      "loss": 0.97,
+      "step": 3312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.783810360658158e-05,
+      "loss": 0.9199,
+      "step": 3313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.783473064037159e-05,
+      "loss": 0.9454,
+      "step": 3314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7831355364401123e-05,
+      "loss": 0.8879,
+      "step": 3315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7827977779665245e-05,
+      "loss": 0.9621,
+      "step": 3316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7824597887159704e-05,
+      "loss": 0.9271,
+      "step": 3317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.782121568788092e-05,
+      "loss": 0.9202,
+      "step": 3318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.781783118282601e-05,
+      "loss": 0.923,
+      "step": 3319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.781444437299276e-05,
+      "loss": 0.903,
+      "step": 3320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7811055259379635e-05,
+      "loss": 0.9226,
+      "step": 3321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7807663842985777e-05,
+      "loss": 0.9806,
+      "step": 3322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.780427012481102e-05,
+      "loss": 0.9211,
+      "step": 3323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7800874105855862e-05,
+      "loss": 0.9632,
+      "step": 3324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7797475787121485e-05,
+      "loss": 0.9256,
+      "step": 3325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7794075169609746e-05,
+      "loss": 0.9261,
+      "step": 3326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.779067225432319e-05,
+      "loss": 0.9214,
+      "step": 3327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7787267042265028e-05,
+      "loss": 0.9071,
+      "step": 3328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.778385953443915e-05,
+      "loss": 0.8785,
+      "step": 3329
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.778044973185013e-05,
+      "loss": 0.9364,
+      "step": 3330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.777703763550321e-05,
+      "loss": 0.9221,
+      "step": 3331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7773623246404312e-05,
+      "loss": 0.9186,
+      "step": 3332
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7770206565560034e-05,
+      "loss": 0.9018,
+      "step": 3333
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7766787593977648e-05,
+      "loss": 0.9435,
+      "step": 3334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7763366332665106e-05,
+      "loss": 0.8644,
+      "step": 3335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7759942782631027e-05,
+      "loss": 0.9303,
+      "step": 3336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7756516944884713e-05,
+      "loss": 0.9567,
+      "step": 3337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.775308882043613e-05,
+      "loss": 0.8615,
+      "step": 3338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7749658410295935e-05,
+      "loss": 0.9457,
+      "step": 3339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7746225715475444e-05,
+      "loss": 0.9244,
+      "step": 3340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7742790736986652e-05,
+      "loss": 0.9557,
+      "step": 3341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7739353475842225e-05,
+      "loss": 0.9152,
+      "step": 3342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7735913933055506e-05,
+      "loss": 0.945,
+      "step": 3343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7732472109640504e-05,
+      "loss": 0.8866,
+      "step": 3344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7729028006611908e-05,
+      "loss": 0.9368,
+      "step": 3345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7725581624985073e-05,
+      "loss": 0.9361,
+      "step": 3346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7722132965776035e-05,
+      "loss": 0.9801,
+      "step": 3347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7718682030001485e-05,
+      "loss": 0.9397,
+      "step": 3348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.77152288186788e-05,
+      "loss": 0.9191,
+      "step": 3349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7711773332826023e-05,
+      "loss": 0.8549,
+      "step": 3350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.770831557346187e-05,
+      "loss": 0.8748,
+      "step": 3351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7704855541605715e-05,
+      "loss": 0.9398,
+      "step": 3352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7701393238277626e-05,
+      "loss": 0.8899,
+      "step": 3353
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7697928664498308e-05,
+      "loss": 0.9712,
+      "step": 3354
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.769446182128917e-05,
+      "loss": 0.9596,
+      "step": 3355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.769099270967227e-05,
+      "loss": 0.9079,
+      "step": 3356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7687521330670333e-05,
+      "loss": 0.9224,
+      "step": 3357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.768404768530676e-05,
+      "loss": 0.9388,
+      "step": 3358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.768057177460562e-05,
+      "loss": 0.9207,
+      "step": 3359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7677093599591643e-05,
+      "loss": 0.9219,
+      "step": 3360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7673613161290237e-05,
+      "loss": 0.8982,
+      "step": 3361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7670130460727465e-05,
+      "loss": 0.9219,
+      "step": 3362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7666645498930074e-05,
+      "loss": 0.8917,
+      "step": 3363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7663158276925458e-05,
+      "loss": 0.9593,
+      "step": 3364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.765966879574169e-05,
+      "loss": 0.426,
+      "step": 3365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7656177056407508e-05,
+      "loss": 0.9458,
+      "step": 3366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7652683059952306e-05,
+      "loss": 0.936,
+      "step": 3367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7649186807406153e-05,
+      "loss": 0.9417,
+      "step": 3368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7645688299799787e-05,
+      "loss": 0.9557,
+      "step": 3369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7642187538164595e-05,
+      "loss": 0.9066,
+      "step": 3370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.763868452353265e-05,
+      "loss": 0.9209,
+      "step": 3371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.763517925693667e-05,
+      "loss": 0.9421,
+      "step": 3372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7631671739410042e-05,
+      "loss": 0.9845,
+      "step": 3373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7628161971986822e-05,
+      "loss": 0.9913,
+      "step": 3374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7624649955701726e-05,
+      "loss": 0.9748,
+      "step": 3375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7621135691590132e-05,
+      "loss": 1.022,
+      "step": 3376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7617619180688087e-05,
+      "loss": 0.9211,
+      "step": 3377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7614100424032284e-05,
+      "loss": 0.8781,
+      "step": 3378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7610579422660103e-05,
+      "loss": 0.9361,
+      "step": 3379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.760705617760956e-05,
+      "loss": 0.9269,
+      "step": 3380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.760353068991935e-05,
+      "loss": 0.9513,
+      "step": 3381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7600002960628827e-05,
+      "loss": 0.9334,
+      "step": 3382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7596472990777996e-05,
+      "loss": 0.9387,
+      "step": 3383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7592940781407533e-05,
+      "loss": 0.9216,
+      "step": 3384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.758940633355877e-05,
+      "loss": 0.9122,
+      "step": 3385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7585869648273697e-05,
+      "loss": 0.9562,
+      "step": 3386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7582330726594964e-05,
+      "loss": 0.9102,
+      "step": 3387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.757878956956589e-05,
+      "loss": 0.8911,
+      "step": 3388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.757524617823044e-05,
+      "loss": 0.9631,
+      "step": 3389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7571700553633236e-05,
+      "loss": 0.9522,
+      "step": 3390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.756815269681958e-05,
+      "loss": 0.9294,
+      "step": 3391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7564602608835407e-05,
+      "loss": 0.9593,
+      "step": 3392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7561050290727322e-05,
+      "loss": 0.938,
+      "step": 3393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7557495743542586e-05,
+      "loss": 0.9188,
+      "step": 3394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7553938968329114e-05,
+      "loss": 0.9021,
+      "step": 3395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7550379966135486e-05,
+      "loss": 0.9704,
+      "step": 3396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.754681873801093e-05,
+      "loss": 0.9323,
+      "step": 3397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7543255285005335e-05,
+      "loss": 0.9334,
+      "step": 3398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.753968960816924e-05,
+      "loss": 1.0382,
+      "step": 3399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7536121708553845e-05,
+      "loss": 0.9471,
+      "step": 3400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7532551587211006e-05,
+      "loss": 0.8923,
+      "step": 3401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7528979245193233e-05,
+      "loss": 0.9571,
+      "step": 3402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.752540468355369e-05,
+      "loss": 0.9602,
+      "step": 3403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.752182790334619e-05,
+      "loss": 0.9552,
+      "step": 3404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7518248905625214e-05,
+      "loss": 0.924,
+      "step": 3405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7514667691445873e-05,
+      "loss": 0.9993,
+      "step": 3406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7511084261863965e-05,
+      "loss": 0.4009,
+      "step": 3407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.750749861793591e-05,
+      "loss": 0.9319,
+      "step": 3408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7503910760718797e-05,
+      "loss": 0.9293,
+      "step": 3409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7500320691270365e-05,
+      "loss": 0.9957,
+      "step": 3410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7496728410649e-05,
+      "loss": 0.8953,
+      "step": 3411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.749313391991375e-05,
+      "loss": 0.8797,
+      "step": 3412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.74895372201243e-05,
+      "loss": 0.9476,
+      "step": 3413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7485938312341003e-05,
+      "loss": 0.9351,
+      "step": 3414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7482337197624853e-05,
+      "loss": 0.9789,
+      "step": 3415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.747873387703749e-05,
+      "loss": 0.9376,
+      "step": 3416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7475128351641216e-05,
+      "loss": 0.9194,
+      "step": 3417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7471520622498982e-05,
+      "loss": 0.9389,
+      "step": 3418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7467910690674372e-05,
+      "loss": 0.9026,
+      "step": 3419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7464298557231642e-05,
+      "loss": 0.9227,
+      "step": 3420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.746068422323568e-05,
+      "loss": 0.3725,
+      "step": 3421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7457067689752033e-05,
+      "loss": 0.99,
+      "step": 3422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7453448957846896e-05,
+      "loss": 0.9522,
+      "step": 3423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7449828028587105e-05,
+      "loss": 0.9141,
+      "step": 3424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7446204903040148e-05,
+      "loss": 0.9128,
+      "step": 3425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.744257958227416e-05,
+      "loss": 0.9486,
+      "step": 3426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.743895206735792e-05,
+      "loss": 0.9267,
+      "step": 3427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7435322359360866e-05,
+      "loss": 0.8818,
+      "step": 3428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.743169045935307e-05,
+      "loss": 0.9356,
+      "step": 3429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7428056368405247e-05,
+      "loss": 0.9144,
+      "step": 3430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7424420087588777e-05,
+      "loss": 0.8787,
+      "step": 3431
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7420781617975667e-05,
+      "loss": 0.9566,
+      "step": 3432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7417140960638574e-05,
+      "loss": 0.9871,
+      "step": 3433
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7413498116650806e-05,
+      "loss": 0.8999,
+      "step": 3434
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7409853087086312e-05,
+      "loss": 1.0207,
+      "step": 3435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7406205873019684e-05,
+      "loss": 0.9479,
+      "step": 3436
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7402556475526155e-05,
+      "loss": 0.9146,
+      "step": 3437
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.739890489568161e-05,
+      "loss": 0.9576,
+      "step": 3438
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7395251134562566e-05,
+      "loss": 0.9733,
+      "step": 3439
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7391595193246197e-05,
+      "loss": 0.9065,
+      "step": 3440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7387937072810313e-05,
+      "loss": 0.8829,
+      "step": 3441
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7384276774333363e-05,
+      "loss": 0.9192,
+      "step": 3442
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7380614298894443e-05,
+      "loss": 0.8616,
+      "step": 3443
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7376949647573287e-05,
+      "loss": 0.9838,
+      "step": 3444
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.737328282145027e-05,
+      "loss": 0.9812,
+      "step": 3445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.736961382160642e-05,
+      "loss": 0.9059,
+      "step": 3446
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.736594264912339e-05,
+      "loss": 0.9639,
+      "step": 3447
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.736226930508348e-05,
+      "loss": 0.9968,
+      "step": 3448
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.735859379056963e-05,
+      "loss": 0.8571,
+      "step": 3449
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7354916106665422e-05,
+      "loss": 0.9024,
+      "step": 3450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7351236254455077e-05,
+      "loss": 0.9467,
+      "step": 3451
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7347554235023447e-05,
+      "loss": 0.8832,
+      "step": 3452
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.734387004945604e-05,
+      "loss": 0.9777,
+      "step": 3453
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.734018369883898e-05,
+      "loss": 0.9455,
+      "step": 3454
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7336495184259057e-05,
+      "loss": 0.8873,
+      "step": 3455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.733280450680367e-05,
+      "loss": 0.9469,
+      "step": 3456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7329111667560875e-05,
+      "loss": 0.8685,
+      "step": 3457
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.732541666761936e-05,
+      "loss": 0.9316,
+      "step": 3458
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.732171950806845e-05,
+      "loss": 0.9201,
+      "step": 3459
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7318020189998103e-05,
+      "loss": 0.8945,
+      "step": 3460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7314318714498922e-05,
+      "loss": 0.9633,
+      "step": 3461
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7310615082662133e-05,
+      "loss": 0.9124,
+      "step": 3462
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.730690929557961e-05,
+      "loss": 0.8791,
+      "step": 3463
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.730320135434386e-05,
+      "loss": 0.9286,
+      "step": 3464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.729949126004802e-05,
+      "loss": 0.9455,
+      "step": 3465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7295779013785865e-05,
+      "loss": 0.8606,
+      "step": 3466
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.72920646166518e-05,
+      "loss": 0.9164,
+      "step": 3467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7288348069740878e-05,
+      "loss": 1.006,
+      "step": 3468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7284629374148764e-05,
+      "loss": 0.9051,
+      "step": 3469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.728090853097178e-05,
+      "loss": 0.9331,
+      "step": 3470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.727718554130686e-05,
+      "loss": 0.9495,
+      "step": 3471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7273460406251584e-05,
+      "loss": 0.9061,
+      "step": 3472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7269733126904162e-05,
+      "loss": 0.9477,
+      "step": 3473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7266003704363432e-05,
+      "loss": 0.9392,
+      "step": 3474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.726227213972887e-05,
+      "loss": 0.979,
+      "step": 3475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.725853843410058e-05,
+      "loss": 0.974,
+      "step": 3476
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.725480258857929e-05,
+      "loss": 0.8817,
+      "step": 3477
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7251064604266376e-05,
+      "loss": 0.9501,
+      "step": 3478
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7247324482263832e-05,
+      "loss": 0.9327,
+      "step": 3479
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7243582223674286e-05,
+      "loss": 0.9361,
+      "step": 3480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.723983782960099e-05,
+      "loss": 0.8766,
+      "step": 3481
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7236091301147834e-05,
+      "loss": 0.9444,
+      "step": 3482
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7232342639419333e-05,
+      "loss": 0.9139,
+      "step": 3483
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7228591845520633e-05,
+      "loss": 0.953,
+      "step": 3484
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7224838920557506e-05,
+      "loss": 0.8984,
+      "step": 3485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7221083865636356e-05,
+      "loss": 0.9619,
+      "step": 3486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.721732668186421e-05,
+      "loss": 0.9392,
+      "step": 3487
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.721356737034872e-05,
+      "loss": 0.8837,
+      "step": 3488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.720980593219818e-05,
+      "loss": 0.9298,
+      "step": 3489
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7206042368521493e-05,
+      "loss": 0.967,
+      "step": 3490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7202276680428208e-05,
+      "loss": 0.8831,
+      "step": 3491
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7198508869028475e-05,
+      "loss": 0.9952,
+      "step": 3492
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7194738935433094e-05,
+      "loss": 0.8982,
+      "step": 3493
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.719096688075348e-05,
+      "loss": 0.9347,
+      "step": 3494
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.718719270610167e-05,
+      "loss": 0.8697,
+      "step": 3495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7183416412590334e-05,
+      "loss": 0.4131,
+      "step": 3496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.717963800133276e-05,
+      "loss": 0.9085,
+      "step": 3497
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7175857473442864e-05,
+      "loss": 0.9731,
+      "step": 3498
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7172074830035185e-05,
+      "loss": 0.8984,
+      "step": 3499
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7168290072224886e-05,
+      "loss": 0.9444,
+      "step": 3500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7164503201127753e-05,
+      "loss": 0.9419,
+      "step": 3501
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7160714217860196e-05,
+      "loss": 0.8905,
+      "step": 3502
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7156923123539245e-05,
+      "loss": 0.9164,
+      "step": 3503
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.715312991928256e-05,
+      "loss": 0.9216,
+      "step": 3504
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.714933460620841e-05,
+      "loss": 0.9114,
+      "step": 3505
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7145537185435694e-05,
+      "loss": 0.9522,
+      "step": 3506
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7141737658083936e-05,
+      "loss": 0.9471,
+      "step": 3507
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.713793602527327e-05,
+      "loss": 0.9175,
+      "step": 3508
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7134132288124464e-05,
+      "loss": 0.9513,
+      "step": 3509
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7130326447758898e-05,
+      "loss": 0.9099,
+      "step": 3510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7126518505298567e-05,
+      "loss": 0.9289,
+      "step": 3511
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.71227084618661e-05,
+      "loss": 0.8793,
+      "step": 3512
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7118896318584733e-05,
+      "loss": 0.9352,
+      "step": 3513
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7115082076578327e-05,
+      "loss": 0.4131,
+      "step": 3514
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.711126573697136e-05,
+      "loss": 0.8913,
+      "step": 3515
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7107447300888932e-05,
+      "loss": 0.9167,
+      "step": 3516
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.710362676945675e-05,
+      "loss": 0.9246,
+      "step": 3517
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.709980414380116e-05,
+      "loss": 0.9189,
+      "step": 3518
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7095979425049098e-05,
+      "loss": 0.9525,
+      "step": 3519
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7092152614328135e-05,
+      "loss": 0.9288,
+      "step": 3520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.708832371276646e-05,
+      "loss": 0.9096,
+      "step": 3521
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.708449272149287e-05,
+      "loss": 0.8257,
+      "step": 3522
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7080659641636784e-05,
+      "loss": 0.9487,
+      "step": 3523
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7076824474328226e-05,
+      "loss": 0.9243,
+      "step": 3524
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.707298722069785e-05,
+      "loss": 0.9071,
+      "step": 3525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.706914788187692e-05,
+      "loss": 0.943,
+      "step": 3526
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7065306458997305e-05,
+      "loss": 0.9461,
+      "step": 3527
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7061462953191504e-05,
+      "loss": 0.9969,
+      "step": 3528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7057617365592624e-05,
+      "loss": 0.975,
+      "step": 3529
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.705376969733438e-05,
+      "loss": 0.9573,
+      "step": 3530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7049919949551103e-05,
+      "loss": 0.9297,
+      "step": 3531
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.704606812337774e-05,
+      "loss": 0.9691,
+      "step": 3532
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7042214219949854e-05,
+      "loss": 0.9223,
+      "step": 3533
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7038358240403615e-05,
+      "loss": 0.9592,
+      "step": 3534
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.70345001858758e-05,
+      "loss": 0.9236,
+      "step": 3535
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7030640057503812e-05,
+      "loss": 0.9,
+      "step": 3536
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7026777856425653e-05,
+      "loss": 0.9199,
+      "step": 3537
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.702291358377994e-05,
+      "loss": 0.8859,
+      "step": 3538
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7019047240705902e-05,
+      "loss": 0.8934,
+      "step": 3539
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.701517882834337e-05,
+      "loss": 0.89,
+      "step": 3540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.701130834783281e-05,
+      "loss": 0.8997,
+      "step": 3541
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7007435800315263e-05,
+      "loss": 0.9134,
+      "step": 3542
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7003561186932403e-05,
+      "loss": 0.9172,
+      "step": 3543
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.699968450882651e-05,
+      "loss": 0.3883,
+      "step": 3544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6995805767140465e-05,
+      "loss": 0.9399,
+      "step": 3545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.699192496301776e-05,
+      "loss": 0.9021,
+      "step": 3546
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.69880420976025e-05,
+      "loss": 0.9369,
+      "step": 3547
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6984157172039393e-05,
+      "loss": 0.9014,
+      "step": 3548
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6980270187473757e-05,
+      "loss": 0.9021,
+      "step": 3549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6976381145051513e-05,
+      "loss": 0.9619,
+      "step": 3550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.697249004591919e-05,
+      "loss": 0.9242,
+      "step": 3551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.696859689122393e-05,
+      "loss": 0.9316,
+      "step": 3552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6964701682113477e-05,
+      "loss": 0.8988,
+      "step": 3553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6960804419736172e-05,
+      "loss": 0.9776,
+      "step": 3554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.695690510524097e-05,
+      "loss": 0.9294,
+      "step": 3555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6953003739777438e-05,
+      "loss": 0.9214,
+      "step": 3556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6949100324495727e-05,
+      "loss": 0.9354,
+      "step": 3557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6945194860546614e-05,
+      "loss": 0.3753,
+      "step": 3558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6941287349081466e-05,
+      "loss": 0.951,
+      "step": 3559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6937377791252262e-05,
+      "loss": 0.9457,
+      "step": 3560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6933466188211575e-05,
+      "loss": 0.3796,
+      "step": 3561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6929552541112592e-05,
+      "loss": 0.951,
+      "step": 3562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6925636851109095e-05,
+      "loss": 0.8927,
+      "step": 3563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.692171911935547e-05,
+      "loss": 0.9184,
+      "step": 3564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.69177993470067e-05,
+      "loss": 0.919,
+      "step": 3565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6913877535218386e-05,
+      "loss": 0.8708,
+      "step": 3566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6909953685146713e-05,
+      "loss": 0.915,
+      "step": 3567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6906027797948473e-05,
+      "loss": 0.9305,
+      "step": 3568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6902099874781058e-05,
+      "loss": 0.3982,
+      "step": 3569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.689816991680246e-05,
+      "loss": 0.8963,
+      "step": 3570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.689423792517128e-05,
+      "loss": 0.9583,
+      "step": 3571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6890303901046695e-05,
+      "loss": 0.9505,
+      "step": 3572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6886367845588507e-05,
+      "loss": 0.9181,
+      "step": 3573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.688242975995711e-05,
+      "loss": 0.9161,
+      "step": 3574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.687848964531348e-05,
+      "loss": 1.0035,
+      "step": 3575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6874547502819213e-05,
+      "loss": 0.8926,
+      "step": 3576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6870603333636495e-05,
+      "loss": 1.0284,
+      "step": 3577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6866657138928106e-05,
+      "loss": 0.9066,
+      "step": 3578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6862708919857417e-05,
+      "loss": 0.8897,
+      "step": 3579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.685875867758842e-05,
+      "loss": 0.9589,
+      "step": 3580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6854806413285674e-05,
+      "loss": 0.8879,
+      "step": 3581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6850852128114357e-05,
+      "loss": 0.9233,
+      "step": 3582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6846895823240228e-05,
+      "loss": 0.898,
+      "step": 3583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6842937499829652e-05,
+      "loss": 0.9547,
+      "step": 3584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6838977159049575e-05,
+      "loss": 0.9429,
+      "step": 3585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.683501480206756e-05,
+      "loss": 0.9378,
+      "step": 3586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.683105043005174e-05,
+      "loss": 0.4002,
+      "step": 3587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6827084044170857e-05,
+      "loss": 0.8459,
+      "step": 3588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6823115645594244e-05,
+      "loss": 0.9679,
+      "step": 3589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.681914523549183e-05,
+      "loss": 0.9499,
+      "step": 3590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6815172815034128e-05,
+      "loss": 0.9532,
+      "step": 3591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6811198385392246e-05,
+      "loss": 0.9415,
+      "step": 3592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6807221947737895e-05,
+      "loss": 0.9216,
+      "step": 3593
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6803243503243368e-05,
+      "loss": 0.8614,
+      "step": 3594
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6799263053081548e-05,
+      "loss": 0.9497,
+      "step": 3595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6795280598425918e-05,
+      "loss": 0.9229,
+      "step": 3596
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6791296140450547e-05,
+      "loss": 0.965,
+      "step": 3597
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6787309680330093e-05,
+      "loss": 0.925,
+      "step": 3598
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6783321219239808e-05,
+      "loss": 0.9822,
+      "step": 3599
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.677933075835553e-05,
+      "loss": 0.916,
+      "step": 3600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6775338298853687e-05,
+      "loss": 0.9624,
+      "step": 3601
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6771343841911302e-05,
+      "loss": 0.9121,
+      "step": 3602
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.676734738870598e-05,
+      "loss": 0.9589,
+      "step": 3603
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.676334894041592e-05,
+      "loss": 0.8506,
+      "step": 3604
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.67593484982199e-05,
+      "loss": 0.9902,
+      "step": 3605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6755346063297303e-05,
+      "loss": 0.9763,
+      "step": 3606
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6751341636828076e-05,
+      "loss": 0.8755,
+      "step": 3607
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6747335219992777e-05,
+      "loss": 0.9356,
+      "step": 3608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.674332681397253e-05,
+      "loss": 0.9117,
+      "step": 3609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.673931641994906e-05,
+      "loss": 0.9541,
+      "step": 3610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6735304039104675e-05,
+      "loss": 0.4044,
+      "step": 3611
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6731289672622263e-05,
+      "loss": 0.9593,
+      "step": 3612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6727273321685303e-05,
+      "loss": 0.9471,
+      "step": 3613
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6723254987477858e-05,
+      "loss": 0.934,
+      "step": 3614
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.671923467118457e-05,
+      "loss": 0.9127,
+      "step": 3615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6715212373990676e-05,
+      "loss": 0.8768,
+      "step": 3616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6711188097081987e-05,
+      "loss": 0.3991,
+      "step": 3617
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6707161841644908e-05,
+      "loss": 0.9134,
+      "step": 3618
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6703133608866415e-05,
+      "loss": 0.9612,
+      "step": 3619
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6699103399934076e-05,
+      "loss": 0.9528,
+      "step": 3620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6695071216036037e-05,
+      "loss": 0.9187,
+      "step": 3621
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6691037058361032e-05,
+      "loss": 0.9754,
+      "step": 3622
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6687000928098366e-05,
+      "loss": 0.9516,
+      "step": 3623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.668296282643794e-05,
+      "loss": 0.9731,
+      "step": 3624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.667892275457022e-05,
+      "loss": 0.9354,
+      "step": 3625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.667488071368627e-05,
+      "loss": 0.9931,
+      "step": 3626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.667083670497772e-05,
+      "loss": 0.9504,
+      "step": 3627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.666679072963679e-05,
+      "loss": 0.9312,
+      "step": 3628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6662742788856275e-05,
+      "loss": 0.9221,
+      "step": 3629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6658692883829548e-05,
+      "loss": 0.9875,
+      "step": 3630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6654641015750564e-05,
+      "loss": 0.9006,
+      "step": 3631
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.665058718581386e-05,
+      "loss": 0.9659,
+      "step": 3632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6646531395214537e-05,
+      "loss": 1.0047,
+      "step": 3633
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6642473645148297e-05,
+      "loss": 0.9258,
+      "step": 3634
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6638413936811398e-05,
+      "loss": 0.9795,
+      "step": 3635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6634352271400693e-05,
+      "loss": 0.9789,
+      "step": 3636
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.66302886501136e-05,
+      "loss": 0.3833,
+      "step": 3637
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6626223074148105e-05,
+      "loss": 0.9269,
+      "step": 3638
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6622155544702804e-05,
+      "loss": 0.9533,
+      "step": 3639
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.661808606297683e-05,
+      "loss": 0.9167,
+      "step": 3640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6614014630169916e-05,
+      "loss": 0.9612,
+      "step": 3641
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.660994124748236e-05,
+      "loss": 0.9546,
+      "step": 3642
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6605865916115044e-05,
+      "loss": 0.9084,
+      "step": 3643
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.660178863726941e-05,
+      "loss": 1.0043,
+      "step": 3644
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.659770941214749e-05,
+      "loss": 0.9405,
+      "step": 3645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6593628241951875e-05,
+      "loss": 0.8695,
+      "step": 3646
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.658954512788574e-05,
+      "loss": 0.9656,
+      "step": 3647
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.658546007115283e-05,
+      "loss": 0.9533,
+      "step": 3648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.658137307295746e-05,
+      "loss": 0.9096,
+      "step": 3649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6577284134504527e-05,
+      "loss": 0.8939,
+      "step": 3650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6573193256999486e-05,
+      "loss": 0.9057,
+      "step": 3651
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6569100441648373e-05,
+      "loss": 0.9167,
+      "step": 3652
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6565005689657792e-05,
+      "loss": 0.9336,
+      "step": 3653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6560909002234917e-05,
+      "loss": 1.0051,
+      "step": 3654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6556810380587497e-05,
+      "loss": 0.99,
+      "step": 3655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6552709825923846e-05,
+      "loss": 0.9451,
+      "step": 3656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.3751,
+      "step": 3657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.654450292238397e-05,
+      "loss": 0.8701,
+      "step": 3658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.654039657592723e-05,
+      "loss": 0.9577,
+      "step": 3659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6536288301293218e-05,
+      "loss": 0.9822,
+      "step": 3660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.65321780996931e-05,
+      "loss": 0.9662,
+      "step": 3661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6528065972338607e-05,
+      "loss": 0.936,
+      "step": 3662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6523951920442032e-05,
+      "loss": 0.9469,
+      "step": 3663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6519835945216253e-05,
+      "loss": 0.9173,
+      "step": 3664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.651571804787469e-05,
+      "loss": 1.0315,
+      "step": 3665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6511598229631344e-05,
+      "loss": 0.9753,
+      "step": 3666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6507476491700788e-05,
+      "loss": 0.9429,
+      "step": 3667
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6503352835298147e-05,
+      "loss": 0.9285,
+      "step": 3668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6499227261639116e-05,
+      "loss": 0.9537,
+      "step": 3669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.649509977193996e-05,
+      "loss": 0.9249,
+      "step": 3670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6490970367417515e-05,
+      "loss": 0.399,
+      "step": 3671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.648683904928916e-05,
+      "loss": 1.0453,
+      "step": 3672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6482705818772853e-05,
+      "loss": 0.9539,
+      "step": 3673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.647857067708712e-05,
+      "loss": 0.9798,
+      "step": 3674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6474433625451035e-05,
+      "loss": 0.9178,
+      "step": 3675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6470294665084254e-05,
+      "loss": 0.9974,
+      "step": 3676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6466153797206973e-05,
+      "loss": 0.9441,
+      "step": 3677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6462011023039977e-05,
+      "loss": 0.9371,
+      "step": 3678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6457866343804592e-05,
+      "loss": 0.9759,
+      "step": 3679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.645371976072271e-05,
+      "loss": 0.8893,
+      "step": 3680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6449571275016795e-05,
+      "loss": 0.8837,
+      "step": 3681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6445420887909858e-05,
+      "loss": 0.9153,
+      "step": 3682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6441268600625476e-05,
+      "loss": 0.9171,
+      "step": 3683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.643711441438779e-05,
+      "loss": 0.9415,
+      "step": 3684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6432958330421497e-05,
+      "loss": 0.9559,
+      "step": 3685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6428800349951853e-05,
+      "loss": 0.8692,
+      "step": 3686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6424640474204675e-05,
+      "loss": 0.9043,
+      "step": 3687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6420478704406337e-05,
+      "loss": 0.8358,
+      "step": 3688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.641631504178377e-05,
+      "loss": 0.879,
+      "step": 3689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6412149487564473e-05,
+      "loss": 0.9829,
+      "step": 3690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6407982042976483e-05,
+      "loss": 0.9904,
+      "step": 3691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.640381270924842e-05,
+      "loss": 0.9208,
+      "step": 3692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.639964148760943e-05,
+      "loss": 0.9325,
+      "step": 3693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6395468379289253e-05,
+      "loss": 0.9228,
+      "step": 3694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6391293385518146e-05,
+      "loss": 0.8298,
+      "step": 3695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6387116507526958e-05,
+      "loss": 0.8562,
+      "step": 3696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6382937746547064e-05,
+      "loss": 0.9471,
+      "step": 3697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.637875710381041e-05,
+      "loss": 0.8117,
+      "step": 3698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.637457458054949e-05,
+      "loss": 0.9826,
+      "step": 3699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.637039017799736e-05,
+      "loss": 0.9711,
+      "step": 3700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.636620389738763e-05,
+      "loss": 0.9426,
+      "step": 3701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6362015739954452e-05,
+      "loss": 0.9077,
+      "step": 3702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.635782570693254e-05,
+      "loss": 0.8598,
+      "step": 3703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6353633799557153e-05,
+      "loss": 0.9995,
+      "step": 3704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6349440019064126e-05,
+      "loss": 0.4036,
+      "step": 3705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6345244366689813e-05,
+      "loss": 0.9843,
+      "step": 3706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6341046843671145e-05,
+      "loss": 0.8571,
+      "step": 3707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6336847451245592e-05,
+      "loss": 0.9359,
+      "step": 3708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.633264619065118e-05,
+      "loss": 0.3881,
+      "step": 3709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6328443063126488e-05,
+      "loss": 0.9395,
+      "step": 3710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6324238069910633e-05,
+      "loss": 0.934,
+      "step": 3711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.63200312122433e-05,
+      "loss": 0.9499,
+      "step": 3712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6315822491364708e-05,
+      "loss": 0.9161,
+      "step": 3713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6311611908515635e-05,
+      "loss": 0.9381,
+      "step": 3714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6307399464937404e-05,
+      "loss": 0.8456,
+      "step": 3715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6303185161871895e-05,
+      "loss": 0.8992,
+      "step": 3716
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6298969000561515e-05,
+      "loss": 0.9738,
+      "step": 3717
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.629475098224924e-05,
+      "loss": 0.9451,
+      "step": 3718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6290531108178587e-05,
+      "loss": 0.8864,
+      "step": 3719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6286309379593616e-05,
+      "loss": 0.9344,
+      "step": 3720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6282085797738937e-05,
+      "loss": 0.9893,
+      "step": 3721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.627786036385971e-05,
+      "loss": 0.9539,
+      "step": 3722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6273633079201626e-05,
+      "loss": 0.9024,
+      "step": 3723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6269403945010948e-05,
+      "loss": 0.8786,
+      "step": 3724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.626517296253446e-05,
+      "loss": 0.9783,
+      "step": 3725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.62609401330195e-05,
+      "loss": 0.9272,
+      "step": 3726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6256705457713952e-05,
+      "loss": 0.8455,
+      "step": 3727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6252468937866243e-05,
+      "loss": 0.9727,
+      "step": 3728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.624823057472534e-05,
+      "loss": 0.9355,
+      "step": 3729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6243990369540764e-05,
+      "loss": 0.9111,
+      "step": 3730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6239748323562562e-05,
+      "loss": 0.9031,
+      "step": 3731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6235504438041342e-05,
+      "loss": 0.934,
+      "step": 3732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.623125871422824e-05,
+      "loss": 0.971,
+      "step": 3733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6227011153374945e-05,
+      "loss": 0.9344,
+      "step": 3734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.622276175673368e-05,
+      "loss": 0.908,
+      "step": 3735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6218510525557206e-05,
+      "loss": 0.9284,
+      "step": 3736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.621425746109884e-05,
+      "loss": 0.8922,
+      "step": 3737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6210002564612425e-05,
+      "loss": 0.9369,
+      "step": 3738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6205745837352346e-05,
+      "loss": 0.9265,
+      "step": 3739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6201487280573536e-05,
+      "loss": 0.9418,
+      "step": 3740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.619722689553146e-05,
+      "loss": 0.924,
+      "step": 3741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6192964683482127e-05,
+      "loss": 0.9582,
+      "step": 3742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6188700645682075e-05,
+      "loss": 0.8807,
+      "step": 3743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.618443478338839e-05,
+      "loss": 0.9822,
+      "step": 3744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6180167097858697e-05,
+      "loss": 0.8762,
+      "step": 3745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6175897590351146e-05,
+      "loss": 0.3972,
+      "step": 3746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.617162626212444e-05,
+      "loss": 0.8786,
+      "step": 3747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.616735311443781e-05,
+      "loss": 0.8856,
+      "step": 3748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.616307814855102e-05,
+      "loss": 0.9055,
+      "step": 3749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6158801365724376e-05,
+      "loss": 0.9103,
+      "step": 3750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6154522767218726e-05,
+      "loss": 0.9748,
+      "step": 3751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6150242354295435e-05,
+      "loss": 0.8972,
+      "step": 3752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.614596012821642e-05,
+      "loss": 0.9068,
+      "step": 3753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.614167609024412e-05,
+      "loss": 0.9479,
+      "step": 3754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.613739024164152e-05,
+      "loss": 0.9811,
+      "step": 3755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.613310258367213e-05,
+      "loss": 0.9103,
+      "step": 3756
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.61288131176e-05,
+      "loss": 0.992,
+      "step": 3757
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6124521844689707e-05,
+      "loss": 0.4026,
+      "step": 3758
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.612022876620636e-05,
+      "loss": 0.9441,
+      "step": 3759
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6115933883415607e-05,
+      "loss": 0.3549,
+      "step": 3760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6111637197583623e-05,
+      "loss": 0.9483,
+      "step": 3761
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.610733870997712e-05,
+      "loss": 0.3973,
+      "step": 3762
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6103038421863332e-05,
+      "loss": 0.9457,
+      "step": 3763
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.609873633451003e-05,
+      "loss": 0.8501,
+      "step": 3764
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6094432449185513e-05,
+      "loss": 0.8994,
+      "step": 3765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6090126767158616e-05,
+      "loss": 0.8855,
+      "step": 3766
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6085819289698695e-05,
+      "loss": 0.913,
+      "step": 3767
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6081510018075638e-05,
+      "loss": 0.8493,
+      "step": 3768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.607719895355987e-05,
+      "loss": 0.978,
+      "step": 3769
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6072886097422333e-05,
+      "loss": 0.9024,
+      "step": 3770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.60685714509345e-05,
+      "loss": 0.9042,
+      "step": 3771
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.606425501536838e-05,
+      "loss": 0.8945,
+      "step": 3772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.60599367919965e-05,
+      "loss": 0.8866,
+      "step": 3773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6055616782091917e-05,
+      "loss": 0.9226,
+      "step": 3774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6051294986928218e-05,
+      "loss": 0.9626,
+      "step": 3775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6046971407779507e-05,
+      "loss": 0.9291,
+      "step": 3776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.604264604592042e-05,
+      "loss": 0.8392,
+      "step": 3777
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.603831890262613e-05,
+      "loss": 0.9341,
+      "step": 3778
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.603398997917232e-05,
+      "loss": 0.9412,
+      "step": 3779
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6029659276835193e-05,
+      "loss": 0.9978,
+      "step": 3780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6025326796891492e-05,
+      "loss": 0.9561,
+      "step": 3781
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6020992540618476e-05,
+      "loss": 0.9709,
+      "step": 3782
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6016656509293933e-05,
+      "loss": 0.9295,
+      "step": 3783
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6012318704196164e-05,
+      "loss": 0.9483,
+      "step": 3784
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6007979126604005e-05,
+      "loss": 0.9806,
+      "step": 3785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6003637777796804e-05,
+      "loss": 0.9104,
+      "step": 3786
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.599929465905444e-05,
+      "loss": 0.9102,
+      "step": 3787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5994949771657307e-05,
+      "loss": 0.9966,
+      "step": 3788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5990603116886326e-05,
+      "loss": 1.002,
+      "step": 3789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5986254696022935e-05,
+      "loss": 0.95,
+      "step": 3790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5981904510349092e-05,
+      "loss": 0.981,
+      "step": 3791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5977552561147276e-05,
+      "loss": 0.8768,
+      "step": 3792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5973198849700494e-05,
+      "loss": 0.8693,
+      "step": 3793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.596884337729226e-05,
+      "loss": 0.8347,
+      "step": 3794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5964486145206612e-05,
+      "loss": 0.3973,
+      "step": 3795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5960127154728112e-05,
+      "loss": 0.8741,
+      "step": 3796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5955766407141834e-05,
+      "loss": 0.9317,
+      "step": 3797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5951403903733373e-05,
+      "loss": 0.9301,
+      "step": 3798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5947039645788834e-05,
+      "loss": 0.9664,
+      "step": 3799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5942673634594853e-05,
+      "loss": 0.8955,
+      "step": 3800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.593830587143858e-05,
+      "loss": 0.95,
+      "step": 3801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5933936357607663e-05,
+      "loss": 0.9392,
+      "step": 3802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5929565094390294e-05,
+      "loss": 0.9241,
+      "step": 3803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5925192083075154e-05,
+      "loss": 0.9562,
+      "step": 3804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5920817324951466e-05,
+      "loss": 0.9263,
+      "step": 3805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5916440821308948e-05,
+      "loss": 0.8999,
+      "step": 3806
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591206257343784e-05,
+      "loss": 0.8597,
+      "step": 3807
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5907682582628893e-05,
+      "loss": 0.8645,
+      "step": 3808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5903300850173377e-05,
+      "loss": 0.9019,
+      "step": 3809
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.589891737736307e-05,
+      "loss": 0.9312,
+      "step": 3810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5894532165490267e-05,
+      "loss": 0.9272,
+      "step": 3811
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5890145215847775e-05,
+      "loss": 0.8988,
+      "step": 3812
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5885756529728912e-05,
+      "loss": 0.8833,
+      "step": 3813
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5881366108427506e-05,
+      "loss": 0.9565,
+      "step": 3814
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5876973953237902e-05,
+      "loss": 0.4044,
+      "step": 3815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5872580065454952e-05,
+      "loss": 0.8769,
+      "step": 3816
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.586818444637402e-05,
+      "loss": 0.9475,
+      "step": 3817
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5863787097290987e-05,
+      "loss": 0.8899,
+      "step": 3818
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5859388019502225e-05,
+      "loss": 0.9556,
+      "step": 3819
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.585498721430464e-05,
+      "loss": 0.9035,
+      "step": 3820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5850584682995625e-05,
+      "loss": 0.9377,
+      "step": 3821
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.58461804268731e-05,
+      "loss": 0.9087,
+      "step": 3822
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5841774447235483e-05,
+      "loss": 0.9132,
+      "step": 3823
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5837366745381698e-05,
+      "loss": 0.9807,
+      "step": 3824
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5832957322611192e-05,
+      "loss": 0.9937,
+      "step": 3825
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.58285461802239e-05,
+      "loss": 0.9438,
+      "step": 3826
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.582413331952028e-05,
+      "loss": 0.9835,
+      "step": 3827
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5819718741801283e-05,
+      "loss": 0.932,
+      "step": 3828
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5815302448368376e-05,
+      "loss": 0.9208,
+      "step": 3829
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5810884440523527e-05,
+      "loss": 0.9065,
+      "step": 3830
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5806464719569214e-05,
+      "loss": 0.8785,
+      "step": 3831
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5802043286808412e-05,
+      "loss": 0.9225,
+      "step": 3832
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.579762014354461e-05,
+      "loss": 0.9491,
+      "step": 3833
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5793195291081794e-05,
+      "loss": 0.873,
+      "step": 3834
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5788768730724457e-05,
+      "loss": 0.967,
+      "step": 3835
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.57843404637776e-05,
+      "loss": 0.9378,
+      "step": 3836
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5779910491546716e-05,
+      "loss": 0.9302,
+      "step": 3837
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.577547881533781e-05,
+      "loss": 0.9696,
+      "step": 3838
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.577104543645738e-05,
+      "loss": 0.9232,
+      "step": 3839
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5766610356212445e-05,
+      "loss": 0.9287,
+      "step": 3840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5762173575910504e-05,
+      "loss": 0.9129,
+      "step": 3841
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5757735096859572e-05,
+      "loss": 0.955,
+      "step": 3842
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5753294920368153e-05,
+      "loss": 0.9303,
+      "step": 3843
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5748853047745264e-05,
+      "loss": 0.8783,
+      "step": 3844
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5744409480300403e-05,
+      "loss": 0.8838,
+      "step": 3845
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5739964219343596e-05,
+      "loss": 0.9127,
+      "step": 3846
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5735517266185344e-05,
+      "loss": 0.4319,
+      "step": 3847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5731068622136655e-05,
+      "loss": 0.3869,
+      "step": 3848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5726618288509036e-05,
+      "loss": 0.9062,
+      "step": 3849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5722166266614496e-05,
+      "loss": 0.9272,
+      "step": 3850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.571771255776553e-05,
+      "loss": 0.8655,
+      "step": 3851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5713257163275148e-05,
+      "loss": 0.9078,
+      "step": 3852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5708800084456836e-05,
+      "loss": 0.9016,
+      "step": 3853
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5704341322624597e-05,
+      "loss": 0.9025,
+      "step": 3854
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5699880879092914e-05,
+      "loss": 0.9159,
+      "step": 3855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.569541875517678e-05,
+      "loss": 0.9376,
+      "step": 3856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5690954952191664e-05,
+      "loss": 0.9039,
+      "step": 3857
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.568648947145355e-05,
+      "loss": 0.9807,
+      "step": 3858
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5682022314278907e-05,
+      "loss": 0.9455,
+      "step": 3859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5677553481984704e-05,
+      "loss": 0.9056,
+      "step": 3860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5673082975888388e-05,
+      "loss": 1.0085,
+      "step": 3861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.566861079730792e-05,
+      "loss": 0.9697,
+      "step": 3862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5664136947561745e-05,
+      "loss": 0.8644,
+      "step": 3863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5659661427968792e-05,
+      "loss": 0.9083,
+      "step": 3864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.56551842398485e-05,
+      "loss": 0.9208,
+      "step": 3865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5650705384520794e-05,
+      "loss": 0.9333,
+      "step": 3866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5646224863306075e-05,
+      "loss": 0.9818,
+      "step": 3867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5641742677525257e-05,
+      "loss": 0.9367,
+      "step": 3868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.563725882849973e-05,
+      "loss": 0.9479,
+      "step": 3869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5632773317551385e-05,
+      "loss": 0.9391,
+      "step": 3870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.562828614600259e-05,
+      "loss": 0.9422,
+      "step": 3871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.562379731517622e-05,
+      "loss": 0.8898,
+      "step": 3872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5619306826395617e-05,
+      "loss": 0.8433,
+      "step": 3873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5614814680984637e-05,
+      "loss": 0.9803,
+      "step": 3874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5610320880267603e-05,
+      "loss": 0.9315,
+      "step": 3875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5605825425569333e-05,
+      "loss": 0.9167,
+      "step": 3876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.560132831821514e-05,
+      "loss": 0.896,
+      "step": 3877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5596829559530814e-05,
+      "loss": 0.8898,
+      "step": 3878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5592329150842642e-05,
+      "loss": 0.9268,
+      "step": 3879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5587827093477384e-05,
+      "loss": 0.9011,
+      "step": 3880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.55833233887623e-05,
+      "loss": 0.3841,
+      "step": 3881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.557881803802512e-05,
+      "loss": 0.851,
+      "step": 3882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.557431104259408e-05,
+      "loss": 0.9543,
+      "step": 3883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5569802403797882e-05,
+      "loss": 0.9674,
+      "step": 3884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5565292122965724e-05,
+      "loss": 0.8878,
+      "step": 3885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5560780201427278e-05,
+      "loss": 0.9161,
+      "step": 3886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.555626664051271e-05,
+      "loss": 0.9754,
+      "step": 3887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5551751441552666e-05,
+      "loss": 0.872,
+      "step": 3888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.554723460587827e-05,
+      "loss": 0.8829,
+      "step": 3889
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.554271613482114e-05,
+      "loss": 0.9849,
+      "step": 3890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.553819602971336e-05,
+      "loss": 0.9475,
+      "step": 3891
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5533674291887505e-05,
+      "loss": 0.9326,
+      "step": 3892
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5529150922676635e-05,
+      "loss": 0.874,
+      "step": 3893
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5524625923414284e-05,
+      "loss": 0.9763,
+      "step": 3894
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5520099295434468e-05,
+      "loss": 0.9333,
+      "step": 3895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5515571040071687e-05,
+      "loss": 0.8543,
+      "step": 3896
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5511041158660916e-05,
+      "loss": 0.8863,
+      "step": 3897
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.550650965253761e-05,
+      "loss": 0.9308,
+      "step": 3898
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5501976523037708e-05,
+      "loss": 0.9423,
+      "step": 3899
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5497441771497623e-05,
+      "loss": 0.9846,
+      "step": 3900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5492905399254247e-05,
+      "loss": 0.9956,
+      "step": 3901
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.548836740764495e-05,
+      "loss": 0.9688,
+      "step": 3902
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5483827798007574e-05,
+      "loss": 0.9678,
+      "step": 3903
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5479286571680454e-05,
+      "loss": 0.9352,
+      "step": 3904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5474743730002383e-05,
+      "loss": 0.93,
+      "step": 3905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.547019927431264e-05,
+      "loss": 0.9027,
+      "step": 3906
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.546565320595098e-05,
+      "loss": 0.9288,
+      "step": 3907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5461105526257627e-05,
+      "loss": 0.9338,
+      "step": 3908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.545655623657329e-05,
+      "loss": 0.9699,
+      "step": 3909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5452005338239142e-05,
+      "loss": 0.4017,
+      "step": 3910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5447452832596842e-05,
+      "loss": 0.9015,
+      "step": 3911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5442898720988507e-05,
+      "loss": 0.9549,
+      "step": 3912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5438343004756746e-05,
+      "loss": 0.9366,
+      "step": 3913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5433785685244623e-05,
+      "loss": 0.8673,
+      "step": 3914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5429226763795693e-05,
+      "loss": 0.8937,
+      "step": 3915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5424666241753966e-05,
+      "loss": 0.9442,
+      "step": 3916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5420104120463934e-05,
+      "loss": 0.9098,
+      "step": 3917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5415540401270557e-05,
+      "loss": 0.8892,
+      "step": 3918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5410975085519272e-05,
+      "loss": 0.9801,
+      "step": 3919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5406408174555978e-05,
+      "loss": 0.9038,
+      "step": 3920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5401839669727046e-05,
+      "loss": 0.415,
+      "step": 3921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5397269572379323e-05,
+      "loss": 0.3546,
+      "step": 3922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5392697883860118e-05,
+      "loss": 0.8981,
+      "step": 3923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5388124605517218e-05,
+      "loss": 0.9329,
+      "step": 3924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5383549738698867e-05,
+      "loss": 0.9014,
+      "step": 3925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5378973284753784e-05,
+      "loss": 0.9062,
+      "step": 3926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.537439524503116e-05,
+      "loss": 0.8784,
+      "step": 3927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5369815620880647e-05,
+      "loss": 0.9033,
+      "step": 3928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5365234413652365e-05,
+      "loss": 0.962,
+      "step": 3929
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.53606516246969e-05,
+      "loss": 0.9354,
+      "step": 3930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5356067255365312e-05,
+      "loss": 0.9507,
+      "step": 3931
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5351481307009116e-05,
+      "loss": 0.8941,
+      "step": 3932
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.53468937809803e-05,
+      "loss": 0.9441,
+      "step": 3933
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5342304678631313e-05,
+      "loss": 0.8877,
+      "step": 3934
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5337714001315068e-05,
+      "loss": 0.9495,
+      "step": 3935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5333121750384955e-05,
+      "loss": 0.8849,
+      "step": 3936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.53285279271948e-05,
+      "loss": 1.0272,
+      "step": 3937
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5323932533098927e-05,
+      "loss": 0.826,
+      "step": 3938
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5319335569452102e-05,
+      "loss": 0.8835,
+      "step": 3939
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.531473703760955e-05,
+      "loss": 0.8808,
+      "step": 3940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.531013693892697e-05,
+      "loss": 0.9019,
+      "step": 3941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5305535274760524e-05,
+      "loss": 0.9218,
+      "step": 3942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5300932046466827e-05,
+      "loss": 0.8994,
+      "step": 3943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.529632725540296e-05,
+      "loss": 0.8747,
+      "step": 3944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5291720902926462e-05,
+      "loss": 0.8893,
+      "step": 3945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5287112990395338e-05,
+      "loss": 0.914,
+      "step": 3946
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5282503519168038e-05,
+      "loss": 0.9258,
+      "step": 3947
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5277892490603495e-05,
+      "loss": 0.9203,
+      "step": 3948
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5273279906061082e-05,
+      "loss": 0.919,
+      "step": 3949
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5268665766900638e-05,
+      "loss": 0.9257,
+      "step": 3950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.526405007448246e-05,
+      "loss": 0.8963,
+      "step": 3951
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.52594328301673e-05,
+      "loss": 0.9333,
+      "step": 3952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5254814035316376e-05,
+      "loss": 0.8829,
+      "step": 3953
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5250193691291353e-05,
+      "loss": 0.9044,
+      "step": 3954
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5245571799454354e-05,
+      "loss": 0.3695,
+      "step": 3955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5240948361167964e-05,
+      "loss": 0.9689,
+      "step": 3956
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5236323377795224e-05,
+      "loss": 0.3789,
+      "step": 3957
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.523169685069962e-05,
+      "loss": 0.948,
+      "step": 3958
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5227068781245104e-05,
+      "loss": 0.9303,
+      "step": 3959
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5222439170796081e-05,
+      "loss": 0.9687,
+      "step": 3960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.521780802071741e-05,
+      "loss": 0.9563,
+      "step": 3961
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5213175332374399e-05,
+      "loss": 0.8961,
+      "step": 3962
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.520854110713281e-05,
+      "loss": 0.3975,
+      "step": 3963
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5203905346358871e-05,
+      "loss": 0.8898,
+      "step": 3964
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5199268051419243e-05,
+      "loss": 0.8822,
+      "step": 3965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.519462922368105e-05,
+      "loss": 0.9181,
+      "step": 3966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5189988864511871e-05,
+      "loss": 0.9617,
+      "step": 3967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5185346975279735e-05,
+      "loss": 0.9209,
+      "step": 3968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5180703557353112e-05,
+      "loss": 0.8899,
+      "step": 3969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5176058612100935e-05,
+      "loss": 0.907,
+      "step": 3970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5171412140892577e-05,
+      "loss": 0.9216,
+      "step": 3971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5166764145097874e-05,
+      "loss": 0.907,
+      "step": 3972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5162114626087098e-05,
+      "loss": 0.9048,
+      "step": 3973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.515746358523098e-05,
+      "loss": 0.9028,
+      "step": 3974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5152811023900688e-05,
+      "loss": 0.8869,
+      "step": 3975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5148156943467855e-05,
+      "loss": 0.9148,
+      "step": 3976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5143501345304547e-05,
+      "loss": 0.9322,
+      "step": 3977
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.513884423078329e-05,
+      "loss": 0.9195,
+      "step": 3978
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5134185601277036e-05,
+      "loss": 0.9023,
+      "step": 3979
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5129525458159211e-05,
+      "loss": 0.9382,
+      "step": 3980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5124863802803667e-05,
+      "loss": 0.8944,
+      "step": 3981
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5120200636584713e-05,
+      "loss": 0.9059,
+      "step": 3982
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5115535960877095e-05,
+      "loss": 0.8238,
+      "step": 3983
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5110869777056008e-05,
+      "loss": 0.9743,
+      "step": 3984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5106202086497095e-05,
+      "loss": 0.9037,
+      "step": 3985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5101532890576443e-05,
+      "loss": 0.97,
+      "step": 3986
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5096862190670571e-05,
+      "loss": 0.3614,
+      "step": 3987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5092189988156455e-05,
+      "loss": 0.9855,
+      "step": 3988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.508751628441151e-05,
+      "loss": 0.8888,
+      "step": 3989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5082841080813594e-05,
+      "loss": 0.9044,
+      "step": 3990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5078164378740998e-05,
+      "loss": 0.9073,
+      "step": 3991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.507348617957247e-05,
+      "loss": 0.9359,
+      "step": 3992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5068806484687191e-05,
+      "loss": 0.8666,
+      "step": 3993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5064125295464783e-05,
+      "loss": 0.9289,
+      "step": 3994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5059442613285308e-05,
+      "loss": 0.8974,
+      "step": 3995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5054758439529271e-05,
+      "loss": 0.9282,
+      "step": 3996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5050072775577617e-05,
+      "loss": 0.949,
+      "step": 3997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5045385622811726e-05,
+      "loss": 0.883,
+      "step": 3998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.504069698261342e-05,
+      "loss": 0.8827,
+      "step": 3999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5036006856364961e-05,
+      "loss": 0.9018,
+      "step": 4000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5031315245449053e-05,
+      "loss": 0.9502,
+      "step": 4001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5026622151248815e-05,
+      "loss": 0.9444,
+      "step": 4002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5021927575147834e-05,
+      "loss": 0.8398,
+      "step": 4003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5017231518530118e-05,
+      "loss": 0.8799,
+      "step": 4004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.501253398278011e-05,
+      "loss": 0.9303,
+      "step": 4005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5007834969282695e-05,
+      "loss": 0.8995,
+      "step": 4006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5003134479423191e-05,
+      "loss": 0.9984,
+      "step": 4007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4998432514587355e-05,
+      "loss": 0.9389,
+      "step": 4008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4993729076161367e-05,
+      "loss": 0.8454,
+      "step": 4009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.498902416553186e-05,
+      "loss": 0.8636,
+      "step": 4010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.498431778408588e-05,
+      "loss": 0.9852,
+      "step": 4011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4979609933210926e-05,
+      "loss": 0.9201,
+      "step": 4012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4974900614294917e-05,
+      "loss": 0.9172,
+      "step": 4013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4970189828726213e-05,
+      "loss": 0.8633,
+      "step": 4014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.49654775778936e-05,
+      "loss": 0.9808,
+      "step": 4015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4960763863186295e-05,
+      "loss": 0.9023,
+      "step": 4016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4956048685993963e-05,
+      "loss": 0.8409,
+      "step": 4017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4951332047706676e-05,
+      "loss": 0.9083,
+      "step": 4018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4946613949714949e-05,
+      "loss": 0.9023,
+      "step": 4019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4941894393409733e-05,
+      "loss": 0.8659,
+      "step": 4020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4937173380182398e-05,
+      "loss": 0.9255,
+      "step": 4021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4932450911424749e-05,
+      "loss": 0.9332,
+      "step": 4022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4927726988529019e-05,
+      "loss": 0.8931,
+      "step": 4023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.492300161288787e-05,
+      "loss": 0.891,
+      "step": 4024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4918274785894392e-05,
+      "loss": 0.9089,
+      "step": 4025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4913546508942106e-05,
+      "loss": 0.9038,
+      "step": 4026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4908816783424948e-05,
+      "loss": 0.8677,
+      "step": 4027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4904085610737302e-05,
+      "loss": 0.8785,
+      "step": 4028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4899352992273958e-05,
+      "loss": 0.918,
+      "step": 4029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4894618929430151e-05,
+      "loss": 0.9599,
+      "step": 4030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4889883423601526e-05,
+      "loss": 0.9703,
+      "step": 4031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4885146476184157e-05,
+      "loss": 0.9541,
+      "step": 4032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4880408088574551e-05,
+      "loss": 0.9069,
+      "step": 4033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4875668262169633e-05,
+      "loss": 0.8635,
+      "step": 4034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4870926998366755e-05,
+      "loss": 0.9574,
+      "step": 4035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4866184298563682e-05,
+      "loss": 0.9268,
+      "step": 4036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4861440164158623e-05,
+      "loss": 0.9348,
+      "step": 4037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4856694596550192e-05,
+      "loss": 0.9073,
+      "step": 4038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4851947597137434e-05,
+      "loss": 0.9216,
+      "step": 4039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4847199167319812e-05,
+      "loss": 0.8919,
+      "step": 4040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4842449308497217e-05,
+      "loss": 0.9217,
+      "step": 4041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4837698022069953e-05,
+      "loss": 0.9265,
+      "step": 4042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.483294530943875e-05,
+      "loss": 0.915,
+      "step": 4043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4828191172004755e-05,
+      "loss": 0.8847,
+      "step": 4044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4823435611169545e-05,
+      "loss": 0.936,
+      "step": 4045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.48186786283351e-05,
+      "loss": 0.9215,
+      "step": 4046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4813920224903838e-05,
+      "loss": 0.9826,
+      "step": 4047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4809160402278574e-05,
+      "loss": 0.8681,
+      "step": 4048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4804399161862564e-05,
+      "loss": 0.9334,
+      "step": 4049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4799636505059463e-05,
+      "loss": 0.8765,
+      "step": 4050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.479487243327336e-05,
+      "loss": 0.9539,
+      "step": 4051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.479010694790875e-05,
+      "loss": 0.956,
+      "step": 4052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4785340050370542e-05,
+      "loss": 0.9364,
+      "step": 4053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4780571742064079e-05,
+      "loss": 0.9191,
+      "step": 4054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4775802024395098e-05,
+      "loss": 0.8921,
+      "step": 4055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4771030898769766e-05,
+      "loss": 0.9164,
+      "step": 4056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4766258366594656e-05,
+      "loss": 0.8761,
+      "step": 4057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.476148442927677e-05,
+      "loss": 0.9354,
+      "step": 4058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4756709088223508e-05,
+      "loss": 0.8773,
+      "step": 4059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4751932344842689e-05,
+      "loss": 0.9596,
+      "step": 4060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.474715420054255e-05,
+      "loss": 0.8895,
+      "step": 4061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4742374656731739e-05,
+      "loss": 0.9299,
+      "step": 4062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4737593714819314e-05,
+      "loss": 0.8541,
+      "step": 4063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4732811376214744e-05,
+      "loss": 0.8445,
+      "step": 4064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4728027642327914e-05,
+      "loss": 1.0096,
+      "step": 4065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4723242514569122e-05,
+      "loss": 0.9714,
+      "step": 4066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4718455994349068e-05,
+      "loss": 0.9607,
+      "step": 4067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4713668083078874e-05,
+      "loss": 0.9482,
+      "step": 4068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4708878782170062e-05,
+      "loss": 0.9209,
+      "step": 4069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4704088093034573e-05,
+      "loss": 0.901,
+      "step": 4070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4699296017084743e-05,
+      "loss": 0.8952,
+      "step": 4071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4694502555733336e-05,
+      "loss": 0.3475,
+      "step": 4072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4689707710393509e-05,
+      "loss": 0.8827,
+      "step": 4073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4684911482478833e-05,
+      "loss": 0.8647,
+      "step": 4074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4680113873403288e-05,
+      "loss": 0.8689,
+      "step": 4075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.467531488458126e-05,
+      "loss": 0.8905,
+      "step": 4076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4670514517427536e-05,
+      "loss": 0.9027,
+      "step": 4077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4665712773357319e-05,
+      "loss": 0.9562,
+      "step": 4078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4660909653786212e-05,
+      "loss": 0.8868,
+      "step": 4079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4656105160130226e-05,
+      "loss": 0.9228,
+      "step": 4080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4651299293805774e-05,
+      "loss": 0.8978,
+      "step": 4081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4646492056229675e-05,
+      "loss": 0.9344,
+      "step": 4082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4641683448819158e-05,
+      "loss": 0.3776,
+      "step": 4083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4636873472991844e-05,
+      "loss": 0.893,
+      "step": 4084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4632062130165767e-05,
+      "loss": 0.9044,
+      "step": 4085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4627249421759362e-05,
+      "loss": 0.4264,
+      "step": 4086
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4622435349191469e-05,
+      "loss": 0.9765,
+      "step": 4087
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4617619913881318e-05,
+      "loss": 0.3816,
+      "step": 4088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4612803117248557e-05,
+      "loss": 0.3401,
+      "step": 4089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4607984960713225e-05,
+      "loss": 0.9598,
+      "step": 4090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4603165445695766e-05,
+      "loss": 0.8826,
+      "step": 4091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4598344573617023e-05,
+      "loss": 0.9394,
+      "step": 4092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4593522345898237e-05,
+      "loss": 0.9289,
+      "step": 4093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4588698763961056e-05,
+      "loss": 0.927,
+      "step": 4094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4583873829227522e-05,
+      "loss": 0.9028,
+      "step": 4095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4579047543120071e-05,
+      "loss": 0.3923,
+      "step": 4096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4574219907061546e-05,
+      "loss": 0.8885,
+      "step": 4097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4569390922475186e-05,
+      "loss": 0.9294,
+      "step": 4098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4564560590784622e-05,
+      "loss": 0.9456,
+      "step": 4099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4559728913413892e-05,
+      "loss": 0.9102,
+      "step": 4100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4554895891787417e-05,
+      "loss": 0.9184,
+      "step": 4101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4550061527330033e-05,
+      "loss": 0.9311,
+      "step": 4102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4545225821466951e-05,
+      "loss": 0.906,
+      "step": 4103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4540388775623793e-05,
+      "loss": 0.8513,
+      "step": 4104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4535550391226566e-05,
+      "loss": 0.9321,
+      "step": 4105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4530710669701684e-05,
+      "loss": 0.8969,
+      "step": 4106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4525869612475942e-05,
+      "loss": 0.8271,
+      "step": 4107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4521027220976538e-05,
+      "loss": 0.9202,
+      "step": 4108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4516183496631052e-05,
+      "loss": 0.8335,
+      "step": 4109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4511338440867473e-05,
+      "loss": 0.9225,
+      "step": 4110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.450649205511417e-05,
+      "loss": 0.917,
+      "step": 4111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.450164434079991e-05,
+      "loss": 0.9074,
+      "step": 4112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4496795299353845e-05,
+      "loss": 0.9366,
+      "step": 4113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4491944932205533e-05,
+      "loss": 0.9437,
+      "step": 4114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4487093240784902e-05,
+      "loss": 0.9086,
+      "step": 4115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4482240226522288e-05,
+      "loss": 0.9083,
+      "step": 4116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4477385890848408e-05,
+      "loss": 0.9686,
+      "step": 4117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.447253023519437e-05,
+      "loss": 0.9068,
+      "step": 4118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4467673260991674e-05,
+      "loss": 0.9098,
+      "step": 4119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.446281496967221e-05,
+      "loss": 0.8584,
+      "step": 4120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4457955362668245e-05,
+      "loss": 0.4081,
+      "step": 4121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4453094441412448e-05,
+      "loss": 0.9029,
+      "step": 4122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4448232207337868e-05,
+      "loss": 0.384,
+      "step": 4123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4443368661877939e-05,
+      "loss": 0.8486,
+      "step": 4124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.443850380646649e-05,
+      "loss": 0.9335,
+      "step": 4125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4433637642537728e-05,
+      "loss": 0.9457,
+      "step": 4126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4428770171526253e-05,
+      "loss": 0.848,
+      "step": 4127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4423901394867041e-05,
+      "loss": 0.8965,
+      "step": 4128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4419031313995463e-05,
+      "loss": 0.8982,
+      "step": 4129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4414159930347265e-05,
+      "loss": 0.3983,
+      "step": 4130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4409287245358586e-05,
+      "loss": 0.8876,
+      "step": 4131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4404413260465941e-05,
+      "loss": 1.0081,
+      "step": 4132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4399537977106235e-05,
+      "loss": 0.9634,
+      "step": 4133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4394661396716749e-05,
+      "loss": 0.9285,
+      "step": 4134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4389783520735154e-05,
+      "loss": 0.9396,
+      "step": 4135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4384904350599498e-05,
+      "loss": 0.8968,
+      "step": 4136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4380023887748209e-05,
+      "loss": 0.9178,
+      "step": 4137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4375142133620097e-05,
+      "loss": 0.9257,
+      "step": 4138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.437025908965436e-05,
+      "loss": 0.9621,
+      "step": 4139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4365374757290562e-05,
+      "loss": 0.9395,
+      "step": 4140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4360489137968668e-05,
+      "loss": 0.8673,
+      "step": 4141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4355602233128996e-05,
+      "loss": 0.9367,
+      "step": 4142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4350714044212262e-05,
+      "loss": 0.9636,
+      "step": 4143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4345824572659556e-05,
+      "loss": 0.9267,
+      "step": 4144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.434093381991235e-05,
+      "loss": 0.9366,
+      "step": 4145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.433604178741248e-05,
+      "loss": 0.8732,
+      "step": 4146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.433114847660217e-05,
+      "loss": 0.898,
+      "step": 4147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4326253888924027e-05,
+      "loss": 0.8883,
+      "step": 4148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.432135802582102e-05,
+      "loss": 0.9366,
+      "step": 4149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4316460888736496e-05,
+      "loss": 0.9198,
+      "step": 4150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4311562479114193e-05,
+      "loss": 0.8998,
+      "step": 4151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.430666279839821e-05,
+      "loss": 0.8368,
+      "step": 4152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4301761848033018e-05,
+      "loss": 0.8923,
+      "step": 4153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4296859629463473e-05,
+      "loss": 0.9292,
+      "step": 4154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4291956144134802e-05,
+      "loss": 0.948,
+      "step": 4155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4287051393492601e-05,
+      "loss": 0.8997,
+      "step": 4156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4282145378982844e-05,
+      "loss": 0.9488,
+      "step": 4157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4277238102051871e-05,
+      "loss": 0.9373,
+      "step": 4158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4272329564146403e-05,
+      "loss": 0.9496,
+      "step": 4159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4267419766713528e-05,
+      "loss": 0.8592,
+      "step": 4160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4262508711200702e-05,
+      "loss": 0.8985,
+      "step": 4161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4257596399055757e-05,
+      "loss": 0.8792,
+      "step": 4162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4252682831726897e-05,
+      "loss": 0.8697,
+      "step": 4163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.424776801066269e-05,
+      "loss": 0.8949,
+      "step": 4164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4242851937312078e-05,
+      "loss": 0.9411,
+      "step": 4165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.423793461312437e-05,
+      "loss": 0.8736,
+      "step": 4166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4233016039549243e-05,
+      "loss": 0.9207,
+      "step": 4167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4228096218036746e-05,
+      "loss": 0.9112,
+      "step": 4168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4223175150037297e-05,
+      "loss": 0.9261,
+      "step": 4169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.421825283700167e-05,
+      "loss": 0.9674,
+      "step": 4170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.421332928038102e-05,
+      "loss": 0.8928,
+      "step": 4171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4208404481626863e-05,
+      "loss": 0.892,
+      "step": 4172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.420347844219108e-05,
+      "loss": 0.9968,
+      "step": 4173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4198551163525918e-05,
+      "loss": 0.9343,
+      "step": 4174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4193622647083994e-05,
+      "loss": 0.9363,
+      "step": 4175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4188692894318279e-05,
+      "loss": 0.9462,
+      "step": 4176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4183761906682125e-05,
+      "loss": 0.8944,
+      "step": 4177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.417882968562923e-05,
+      "loss": 0.933,
+      "step": 4178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.417389623261367e-05,
+      "loss": 0.9053,
+      "step": 4179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4168961549089875e-05,
+      "loss": 0.8624,
+      "step": 4180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4164025636512644e-05,
+      "loss": 0.897,
+      "step": 4181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4159088496337133e-05,
+      "loss": 0.887,
+      "step": 4182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.8922,
+      "step": 4183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.414921053901372e-05,
+      "loss": 0.9152,
+      "step": 4184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4144269724777946e-05,
+      "loss": 0.9279,
+      "step": 4185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4139327688768139e-05,
+      "loss": 0.8919,
+      "step": 4186
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4134384432441266e-05,
+      "loss": 0.7794,
+      "step": 4187
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4129439957254655e-05,
+      "loss": 0.3758,
+      "step": 4188
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4124494264665982e-05,
+      "loss": 0.9297,
+      "step": 4189
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4119547356133291e-05,
+      "loss": 0.8762,
+      "step": 4190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4114599233114988e-05,
+      "loss": 0.8213,
+      "step": 4191
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4109649897069826e-05,
+      "loss": 0.9378,
+      "step": 4192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4104699349456918e-05,
+      "loss": 0.9858,
+      "step": 4193
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4099747591735742e-05,
+      "loss": 0.9069,
+      "step": 4194
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4094794625366129e-05,
+      "loss": 0.8866,
+      "step": 4195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4089840451808261e-05,
+      "loss": 0.9204,
+      "step": 4196
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4084885072522683e-05,
+      "loss": 0.8895,
+      "step": 4197
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4079928488970291e-05,
+      "loss": 0.8837,
+      "step": 4198
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.407497070261234e-05,
+      "loss": 0.8991,
+      "step": 4199
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4070011714910437e-05,
+      "loss": 0.8958,
+      "step": 4200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4065051527326539e-05,
+      "loss": 0.9576,
+      "step": 4201
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4060090141322969e-05,
+      "loss": 0.3944,
+      "step": 4202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4055127558362385e-05,
+      "loss": 1.0009,
+      "step": 4203
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4050163779907817e-05,
+      "loss": 0.922,
+      "step": 4204
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4045198807422632e-05,
+      "loss": 0.8905,
+      "step": 4205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4040232642370569e-05,
+      "loss": 0.9082,
+      "step": 4206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4035265286215687e-05,
+      "loss": 0.9163,
+      "step": 4207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4030296740422427e-05,
+      "loss": 0.9213,
+      "step": 4208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4025327006455563e-05,
+      "loss": 0.9162,
+      "step": 4209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.402035608578023e-05,
+      "loss": 0.9169,
+      "step": 4210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4015383979861898e-05,
+      "loss": 0.9048,
+      "step": 4211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4010410690166406e-05,
+      "loss": 0.8507,
+      "step": 4212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4005436218159927e-05,
+      "loss": 0.8774,
+      "step": 4213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4000460565308984e-05,
+      "loss": 0.8826,
+      "step": 4214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3995483733080457e-05,
+      "loss": 0.9654,
+      "step": 4215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3990505722941565e-05,
+      "loss": 0.9259,
+      "step": 4216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3985526536359881e-05,
+      "loss": 0.8904,
+      "step": 4217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3980546174803318e-05,
+      "loss": 0.9072,
+      "step": 4218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3975564639740138e-05,
+      "loss": 0.8697,
+      "step": 4219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3970581932638951e-05,
+      "loss": 0.9355,
+      "step": 4220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3965598054968715e-05,
+      "loss": 0.9289,
+      "step": 4221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3960613008198722e-05,
+      "loss": 0.9715,
+      "step": 4222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.395562679379862e-05,
+      "loss": 0.9723,
+      "step": 4223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3950639413238394e-05,
+      "loss": 0.898,
+      "step": 4224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3945650867988382e-05,
+      "loss": 0.9348,
+      "step": 4225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3940661159519255e-05,
+      "loss": 0.8832,
+      "step": 4226
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.393567028930203e-05,
+      "loss": 0.4267,
+      "step": 4227
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.393067825880807e-05,
+      "loss": 0.8888,
+      "step": 4228
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3925685069509079e-05,
+      "loss": 0.9216,
+      "step": 4229
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3920690722877099e-05,
+      "loss": 0.9762,
+      "step": 4230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3915695220384514e-05,
+      "loss": 0.8969,
+      "step": 4231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3910698563504057e-05,
+      "loss": 0.9715,
+      "step": 4232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3905700753708786e-05,
+      "loss": 0.9196,
+      "step": 4233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3900701792472113e-05,
+      "loss": 0.952,
+      "step": 4234
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3895701681267784e-05,
+      "loss": 0.8888,
+      "step": 4235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3890700421569882e-05,
+      "loss": 0.8949,
+      "step": 4236
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3885698014852832e-05,
+      "loss": 0.9393,
+      "step": 4237
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3880694462591397e-05,
+      "loss": 0.9682,
+      "step": 4238
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3875689766260671e-05,
+      "loss": 0.9365,
+      "step": 4239
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3870683927336097e-05,
+      "loss": 0.9234,
+      "step": 4240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3865676947293445e-05,
+      "loss": 0.8865,
+      "step": 4241
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3860668827608827e-05,
+      "loss": 0.9325,
+      "step": 4242
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3855659569758689e-05,
+      "loss": 0.9537,
+      "step": 4243
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.385064917521981e-05,
+      "loss": 0.9135,
+      "step": 4244
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.384563764546931e-05,
+      "loss": 0.9359,
+      "step": 4245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.384062498198464e-05,
+      "loss": 0.9415,
+      "step": 4246
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.383561118624358e-05,
+      "loss": 0.8646,
+      "step": 4247
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3830596259724257e-05,
+      "loss": 0.8833,
+      "step": 4248
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3825580203905122e-05,
+      "loss": 0.897,
+      "step": 4249
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3820563020264958e-05,
+      "loss": 0.8507,
+      "step": 4250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3815544710282885e-05,
+      "loss": 0.8571,
+      "step": 4251
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3810525275438353e-05,
+      "loss": 0.8414,
+      "step": 4252
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3805504717211143e-05,
+      "loss": 0.9082,
+      "step": 4253
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3800483037081369e-05,
+      "loss": 0.875,
+      "step": 4254
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3795460236529472e-05,
+      "loss": 0.8988,
+      "step": 4255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.379043631703623e-05,
+      "loss": 0.9047,
+      "step": 4256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3785411280082747e-05,
+      "loss": 0.8672,
+      "step": 4257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3780385127150454e-05,
+      "loss": 0.9226,
+      "step": 4258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3775357859721112e-05,
+      "loss": 0.9621,
+      "step": 4259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3770329479276815e-05,
+      "loss": 0.8995,
+      "step": 4260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3765299987299985e-05,
+      "loss": 0.825,
+      "step": 4261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.376026938527336e-05,
+      "loss": 0.937,
+      "step": 4262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3755237674680024e-05,
+      "loss": 0.9221,
+      "step": 4263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.375020485700337e-05,
+      "loss": 0.418,
+      "step": 4264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.374517093372713e-05,
+      "loss": 0.8512,
+      "step": 4265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3740135906335355e-05,
+      "loss": 0.9363,
+      "step": 4266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3735099776312426e-05,
+      "loss": 0.962,
+      "step": 4267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3730062545143041e-05,
+      "loss": 0.8808,
+      "step": 4268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3725024214312237e-05,
+      "loss": 0.9497,
+      "step": 4269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3719984785305359e-05,
+      "loss": 0.9086,
+      "step": 4270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3714944259608087e-05,
+      "loss": 0.9155,
+      "step": 4271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3709902638706418e-05,
+      "loss": 0.3825,
+      "step": 4272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.370485992408668e-05,
+      "loss": 0.411,
+      "step": 4273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3699816117235512e-05,
+      "loss": 0.3918,
+      "step": 4274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3694771219639886e-05,
+      "loss": 0.9045,
+      "step": 4275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3689725232787087e-05,
+      "loss": 0.8572,
+      "step": 4276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3684678158164725e-05,
+      "loss": 0.9638,
+      "step": 4277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.367962999726073e-05,
+      "loss": 0.993,
+      "step": 4278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3674580751563357e-05,
+      "loss": 0.9382,
+      "step": 4279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.366953042256117e-05,
+      "loss": 0.9281,
+      "step": 4280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3664479011743063e-05,
+      "loss": 0.8248,
+      "step": 4281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3659426520598245e-05,
+      "loss": 0.9354,
+      "step": 4282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.365437295061624e-05,
+      "loss": 0.9389,
+      "step": 4283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3649318303286893e-05,
+      "loss": 0.9343,
+      "step": 4284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3644262580100372e-05,
+      "loss": 0.9138,
+      "step": 4285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3639205782547155e-05,
+      "loss": 0.8894,
+      "step": 4286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3634147912118035e-05,
+      "loss": 0.9582,
+      "step": 4287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3629088970304127e-05,
+      "loss": 0.902,
+      "step": 4288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3624028958596861e-05,
+      "loss": 0.8511,
+      "step": 4289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3618967878487983e-05,
+      "loss": 0.4023,
+      "step": 4290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.361390573146955e-05,
+      "loss": 0.8886,
+      "step": 4291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3608842519033934e-05,
+      "loss": 1.0119,
+      "step": 4292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3603778242673826e-05,
+      "loss": 0.9525,
+      "step": 4293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.359871290388223e-05,
+      "loss": 0.9066,
+      "step": 4294
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.359364650415245e-05,
+      "loss": 0.3854,
+      "step": 4295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3588579044978127e-05,
+      "loss": 0.9167,
+      "step": 4296
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3583510527853193e-05,
+      "loss": 0.9636,
+      "step": 4297
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3578440954271904e-05,
+      "loss": 0.8791,
+      "step": 4298
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3573370325728818e-05,
+      "loss": 0.894,
+      "step": 4299
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3568298643718815e-05,
+      "loss": 0.9524,
+      "step": 4300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3563225909737076e-05,
+      "loss": 0.9032,
+      "step": 4301
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3558152125279099e-05,
+      "loss": 0.9385,
+      "step": 4302
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3553077291840687e-05,
+      "loss": 0.9559,
+      "step": 4303
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3548001410917949e-05,
+      "loss": 0.8525,
+      "step": 4304
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3542924484007317e-05,
+      "loss": 0.9175,
+      "step": 4305
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3537846512605515e-05,
+      "loss": 0.8876,
+      "step": 4306
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3532767498209589e-05,
+      "loss": 0.8536,
+      "step": 4307
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3527687442316877e-05,
+      "loss": 0.9124,
+      "step": 4308
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.352260634642504e-05,
+      "loss": 0.9588,
+      "step": 4309
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3517524212032033e-05,
+      "loss": 0.9044,
+      "step": 4310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3512441040636126e-05,
+      "loss": 0.9083,
+      "step": 4311
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3507356833735887e-05,
+      "loss": 0.9729,
+      "step": 4312
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3502271592830198e-05,
+      "loss": 0.9473,
+      "step": 4313
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3497185319418238e-05,
+      "loss": 0.935,
+      "step": 4314
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3492098014999497e-05,
+      "loss": 0.9204,
+      "step": 4315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3487009681073758e-05,
+      "loss": 0.8797,
+      "step": 4316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3481920319141123e-05,
+      "loss": 0.9686,
+      "step": 4317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3476829930701986e-05,
+      "loss": 0.9989,
+      "step": 4318
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.347173851725705e-05,
+      "loss": 0.8883,
+      "step": 4319
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3466646080307309e-05,
+      "loss": 0.9444,
+      "step": 4320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3461552621354073e-05,
+      "loss": 0.9653,
+      "step": 4321
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3456458141898946e-05,
+      "loss": 0.9293,
+      "step": 4322
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3451362643443832e-05,
+      "loss": 0.9118,
+      "step": 4323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.344626612749094e-05,
+      "loss": 0.9093,
+      "step": 4324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3441168595542773e-05,
+      "loss": 0.9377,
+      "step": 4325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3436070049102139e-05,
+      "loss": 0.9103,
+      "step": 4326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.343097048967214e-05,
+      "loss": 0.8467,
+      "step": 4327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3425869918756182e-05,
+      "loss": 0.8705,
+      "step": 4328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3420768337857967e-05,
+      "loss": 0.9194,
+      "step": 4329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3415665748481494e-05,
+      "loss": 0.8466,
+      "step": 4330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3410562152131059e-05,
+      "loss": 0.9015,
+      "step": 4331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3405457550311256e-05,
+      "loss": 0.9299,
+      "step": 4332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.340035194452697e-05,
+      "loss": 0.8896,
+      "step": 4333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3395245336283398e-05,
+      "loss": 0.9948,
+      "step": 4334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.339013772708601e-05,
+      "loss": 0.9186,
+      "step": 4335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.338502911844059e-05,
+      "loss": 0.9109,
+      "step": 4336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3379919511853205e-05,
+      "loss": 0.8647,
+      "step": 4337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3374808908830226e-05,
+      "loss": 0.9416,
+      "step": 4338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3369697310878305e-05,
+      "loss": 0.9105,
+      "step": 4339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.33645847195044e-05,
+      "loss": 0.9513,
+      "step": 4340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.335947113621575e-05,
+      "loss": 0.9208,
+      "step": 4341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3354356562519901e-05,
+      "loss": 0.8686,
+      "step": 4342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3349240999924678e-05,
+      "loss": 0.8881,
+      "step": 4343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3344124449938205e-05,
+      "loss": 0.9223,
+      "step": 4344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.333900691406889e-05,
+      "loss": 0.8525,
+      "step": 4345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3333888393825438e-05,
+      "loss": 0.916,
+      "step": 4346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3328768890716846e-05,
+      "loss": 0.9328,
+      "step": 4347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3323648406252392e-05,
+      "loss": 0.9196,
+      "step": 4348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.331852694194165e-05,
+      "loss": 0.9868,
+      "step": 4349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3313404499294482e-05,
+      "loss": 0.9141,
+      "step": 4350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3308281079821036e-05,
+      "loss": 0.8962,
+      "step": 4351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3303156685031753e-05,
+      "loss": 0.8645,
+      "step": 4352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3298031316437354e-05,
+      "loss": 0.8806,
+      "step": 4353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3292904975548852e-05,
+      "loss": 0.9202,
+      "step": 4354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3287777663877552e-05,
+      "loss": 0.9579,
+      "step": 4355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3282649382935028e-05,
+      "loss": 0.9433,
+      "step": 4356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3277520134233161e-05,
+      "loss": 0.9513,
+      "step": 4357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.32723899192841e-05,
+      "loss": 0.9468,
+      "step": 4358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3267258739600293e-05,
+      "loss": 0.9247,
+      "step": 4359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3262126596694458e-05,
+      "loss": 0.9044,
+      "step": 4360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3256993492079606e-05,
+      "loss": 0.9517,
+      "step": 4361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3251859427269037e-05,
+      "loss": 0.933,
+      "step": 4362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3246724403776319e-05,
+      "loss": 0.8996,
+      "step": 4363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.324158842311531e-05,
+      "loss": 0.9028,
+      "step": 4364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3236451486800157e-05,
+      "loss": 0.9242,
+      "step": 4365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3231313596345282e-05,
+      "loss": 0.8941,
+      "step": 4366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3226174753265381e-05,
+      "loss": 0.9247,
+      "step": 4367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3221034959075447e-05,
+      "loss": 0.8376,
+      "step": 4368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.321589421529074e-05,
+      "loss": 0.9214,
+      "step": 4369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3210752523426811e-05,
+      "loss": 0.9569,
+      "step": 4370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3205609884999477e-05,
+      "loss": 0.9084,
+      "step": 4371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3200466301524848e-05,
+      "loss": 0.9441,
+      "step": 4372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3195321774519298e-05,
+      "loss": 0.9149,
+      "step": 4373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3190176305499497e-05,
+      "loss": 0.9299,
+      "step": 4374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3185029895982377e-05,
+      "loss": 0.917,
+      "step": 4375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3179882547485156e-05,
+      "loss": 0.9337,
+      "step": 4376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3174734261525321e-05,
+      "loss": 0.9471,
+      "step": 4377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3169585039620651e-05,
+      "loss": 0.9037,
+      "step": 4378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3164434883289178e-05,
+      "loss": 0.8847,
+      "step": 4379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3159283794049234e-05,
+      "loss": 0.8631,
+      "step": 4380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.31541317734194e-05,
+      "loss": 0.9757,
+      "step": 4381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3148978822918562e-05,
+      "loss": 0.8552,
+      "step": 4382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3143824944065848e-05,
+      "loss": 0.9023,
+      "step": 4383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3138670138380687e-05,
+      "loss": 0.8888,
+      "step": 4384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3133514407382763e-05,
+      "loss": 0.4013,
+      "step": 4385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.312835775259204e-05,
+      "loss": 0.36,
+      "step": 4386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3123200175528756e-05,
+      "loss": 0.3889,
+      "step": 4387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.311804167771342e-05,
+      "loss": 0.9698,
+      "step": 4388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3112882260666805e-05,
+      "loss": 0.9062,
+      "step": 4389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3107721925909967e-05,
+      "loss": 0.8346,
+      "step": 4390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3102560674964226e-05,
+      "loss": 0.8887,
+      "step": 4391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.309739850935117e-05,
+      "loss": 0.8882,
+      "step": 4392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3092235430592659e-05,
+      "loss": 0.9403,
+      "step": 4393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3087071440210823e-05,
+      "loss": 0.8771,
+      "step": 4394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3081906539728066e-05,
+      "loss": 0.9275,
+      "step": 4395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3076740730667048e-05,
+      "loss": 0.9417,
+      "step": 4396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3071574014550706e-05,
+      "loss": 0.904,
+      "step": 4397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3066406392902239e-05,
+      "loss": 0.883,
+      "step": 4398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.306123786724512e-05,
+      "loss": 0.8862,
+      "step": 4399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3056068439103084e-05,
+      "loss": 0.9551,
+      "step": 4400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3050898110000132e-05,
+      "loss": 0.8759,
+      "step": 4401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3045726881460526e-05,
+      "loss": 0.8258,
+      "step": 4402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3040554755008805e-05,
+      "loss": 0.8675,
+      "step": 4403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3035381732169762e-05,
+      "loss": 0.3693,
+      "step": 4404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3030207814468466e-05,
+      "loss": 0.8684,
+      "step": 4405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3025033003430226e-05,
+      "loss": 0.9148,
+      "step": 4406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3019857300580645e-05,
+      "loss": 0.9092,
+      "step": 4407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3014680707445571e-05,
+      "loss": 0.8852,
+      "step": 4408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3009503225551114e-05,
+      "loss": 0.9268,
+      "step": 4409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3004324856423652e-05,
+      "loss": 0.9419,
+      "step": 4410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2999145601589822e-05,
+      "loss": 0.9215,
+      "step": 4411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2993965462576523e-05,
+      "loss": 0.9457,
+      "step": 4412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2988784440910913e-05,
+      "loss": 0.9019,
+      "step": 4413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2983602538120415e-05,
+      "loss": 0.9177,
+      "step": 4414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.29784197557327e-05,
+      "loss": 0.8853,
+      "step": 4415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2973236095275716e-05,
+      "loss": 0.8378,
+      "step": 4416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2968051558277655e-05,
+      "loss": 0.8635,
+      "step": 4417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2962866146266975e-05,
+      "loss": 0.8928,
+      "step": 4418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2957679860772383e-05,
+      "loss": 0.8605,
+      "step": 4419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.295249270332286e-05,
+      "loss": 0.8878,
+      "step": 4420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.294730467544763e-05,
+      "loss": 0.9017,
+      "step": 4421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2942115778676176e-05,
+      "loss": 0.3803,
+      "step": 4422
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.293692601453824e-05,
+      "loss": 0.9095,
+      "step": 4423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2931735384563821e-05,
+      "loss": 0.9061,
+      "step": 4424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2926543890283168e-05,
+      "loss": 0.9191,
+      "step": 4425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2921351533226789e-05,
+      "loss": 0.9439,
+      "step": 4426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2916158314925444e-05,
+      "loss": 0.9027,
+      "step": 4427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2910964236910152e-05,
+      "loss": 0.3988,
+      "step": 4428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2905769300712176e-05,
+      "loss": 0.8744,
+      "step": 4429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2900573507863041e-05,
+      "loss": 0.9389,
+      "step": 4430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2895376859894519e-05,
+      "loss": 0.9108,
+      "step": 4431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2890179358338642e-05,
+      "loss": 0.9116,
+      "step": 4432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2884981004727676e-05,
+      "loss": 0.8856,
+      "step": 4433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2879781800594162e-05,
+      "loss": 0.8776,
+      "step": 4434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2874581747470874e-05,
+      "loss": 0.8771,
+      "step": 4435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2869380846890842e-05,
+      "loss": 0.8097,
+      "step": 4436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2864179100387349e-05,
+      "loss": 0.9009,
+      "step": 4437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2858976509493915e-05,
+      "loss": 0.9439,
+      "step": 4438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.285377307574433e-05,
+      "loss": 0.9176,
+      "step": 4439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2848568800672617e-05,
+      "loss": 0.903,
+      "step": 4440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2843363685813048e-05,
+      "loss": 0.9382,
+      "step": 4441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2838157732700143e-05,
+      "loss": 0.9485,
+      "step": 4442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2832950942868681e-05,
+      "loss": 0.8766,
+      "step": 4443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2827743317853667e-05,
+      "loss": 0.9757,
+      "step": 4444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2822534859190371e-05,
+      "loss": 0.9351,
+      "step": 4445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2817325568414299e-05,
+      "loss": 0.9054,
+      "step": 4446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2812115447061203e-05,
+      "loss": 0.8655,
+      "step": 4447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2806904496667081e-05,
+      "loss": 0.867,
+      "step": 4448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2801692718768179e-05,
+      "loss": 0.9141,
+      "step": 4449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2796480114900975e-05,
+      "loss": 0.8845,
+      "step": 4450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2791266686602213e-05,
+      "loss": 0.3999,
+      "step": 4451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2786052435408855e-05,
+      "loss": 0.891,
+      "step": 4452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2780837362858121e-05,
+      "loss": 0.9187,
+      "step": 4453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2775621470487467e-05,
+      "loss": 0.9091,
+      "step": 4454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2770404759834593e-05,
+      "loss": 0.8989,
+      "step": 4455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2765187232437444e-05,
+      "loss": 0.9073,
+      "step": 4456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2759968889834195e-05,
+      "loss": 0.9234,
+      "step": 4457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2754749733563273e-05,
+      "loss": 0.9075,
+      "step": 4458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2749529765163335e-05,
+      "loss": 0.8657,
+      "step": 4459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.274430898617329e-05,
+      "loss": 0.8907,
+      "step": 4460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2739087398132271e-05,
+      "loss": 0.9084,
+      "step": 4461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.273386500257966e-05,
+      "loss": 0.9537,
+      "step": 4462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2728641801055068e-05,
+      "loss": 0.9061,
+      "step": 4463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2723417795098358e-05,
+      "loss": 0.9649,
+      "step": 4464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2718192986249618e-05,
+      "loss": 0.8501,
+      "step": 4465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2712967376049177e-05,
+      "loss": 0.9622,
+      "step": 4466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2707740966037596e-05,
+      "loss": 0.8461,
+      "step": 4467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2702513757755681e-05,
+      "loss": 0.8928,
+      "step": 4468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2697285752744462e-05,
+      "loss": 0.9321,
+      "step": 4469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2692056952545214e-05,
+      "loss": 0.8705,
+      "step": 4470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2686827358699435e-05,
+      "loss": 0.9629,
+      "step": 4471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2681596972748876e-05,
+      "loss": 0.8815,
+      "step": 4472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2676365796235497e-05,
+      "loss": 0.9115,
+      "step": 4473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2671133830701512e-05,
+      "loss": 0.9348,
+      "step": 4474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2665901077689352e-05,
+      "loss": 0.9046,
+      "step": 4475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2660667538741696e-05,
+      "loss": 0.3775,
+      "step": 4476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2655433215401438e-05,
+      "loss": 0.9313,
+      "step": 4477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2650198109211718e-05,
+      "loss": 0.9001,
+      "step": 4478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2644962221715893e-05,
+      "loss": 0.911,
+      "step": 4479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2639725554457563e-05,
+      "loss": 0.892,
+      "step": 4480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.263448810898055e-05,
+      "loss": 0.9388,
+      "step": 4481
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2629249886828906e-05,
+      "loss": 0.9462,
+      "step": 4482
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2624010889546915e-05,
+      "loss": 0.9619,
+      "step": 4483
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.261877111867909e-05,
+      "loss": 0.8769,
+      "step": 4484
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2613530575770169e-05,
+      "loss": 0.8732,
+      "step": 4485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2608289262365119e-05,
+      "loss": 0.9106,
+      "step": 4486
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2603047180009131e-05,
+      "loss": 0.9412,
+      "step": 4487
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.259780433024763e-05,
+      "loss": 0.9199,
+      "step": 4488
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2592560714626262e-05,
+      "loss": 0.9377,
+      "step": 4489
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2587316334690898e-05,
+      "loss": 0.8573,
+      "step": 4490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2582071191987637e-05,
+      "loss": 0.9008,
+      "step": 4491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2576825288062801e-05,
+      "loss": 0.3723,
+      "step": 4492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.257157862446294e-05,
+      "loss": 0.9173,
+      "step": 4493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2566331202734822e-05,
+      "loss": 0.9119,
+      "step": 4494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2561083024425443e-05,
+      "loss": 0.8763,
+      "step": 4495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.255583409108202e-05,
+      "loss": 0.8687,
+      "step": 4496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2550584404251996e-05,
+      "loss": 0.8897,
+      "step": 4497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2545333965483031e-05,
+      "loss": 0.8912,
+      "step": 4498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2540082776323009e-05,
+      "loss": 0.8951,
+      "step": 4499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2534830838320037e-05,
+      "loss": 0.9328,
+      "step": 4500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2529578153022444e-05,
+      "loss": 0.9035,
+      "step": 4501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2524324721978768e-05,
+      "loss": 0.9056,
+      "step": 4502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2519070546737783e-05,
+      "loss": 0.8789,
+      "step": 4503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2513815628848473e-05,
+      "loss": 0.8897,
+      "step": 4504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2508559969860042e-05,
+      "loss": 0.9409,
+      "step": 4505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2503303571321915e-05,
+      "loss": 0.8962,
+      "step": 4506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2498046434783727e-05,
+      "loss": 0.8592,
+      "step": 4507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2492788561795347e-05,
+      "loss": 0.9034,
+      "step": 4508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2487529953906841e-05,
+      "loss": 0.9305,
+      "step": 4509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2482270612668508e-05,
+      "loss": 0.9931,
+      "step": 4510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.247701053963085e-05,
+      "loss": 0.915,
+      "step": 4511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2471749736344601e-05,
+      "loss": 0.3802,
+      "step": 4512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2466488204360694e-05,
+      "loss": 0.9613,
+      "step": 4513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2461225945230289e-05,
+      "loss": 0.9495,
+      "step": 4514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2455962960504744e-05,
+      "loss": 0.8933,
+      "step": 4515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2450699251735656e-05,
+      "loss": 0.9028,
+      "step": 4516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2445434820474813e-05,
+      "loss": 0.9082,
+      "step": 4517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2440169668274226e-05,
+      "loss": 0.3625,
+      "step": 4518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2434903796686117e-05,
+      "loss": 0.8801,
+      "step": 4519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.242963720726292e-05,
+      "loss": 0.9093,
+      "step": 4520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.242436990155728e-05,
+      "loss": 0.8897,
+      "step": 4521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2419101881122055e-05,
+      "loss": 0.9577,
+      "step": 4522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2413833147510312e-05,
+      "loss": 0.93,
+      "step": 4523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2408563702275329e-05,
+      "loss": 0.8499,
+      "step": 4524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2403293546970593e-05,
+      "loss": 0.8867,
+      "step": 4525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.23980226831498e-05,
+      "loss": 0.9001,
+      "step": 4526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2392751112366856e-05,
+      "loss": 0.4242,
+      "step": 4527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2387478836175877e-05,
+      "loss": 0.8685,
+      "step": 4528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2382205856131186e-05,
+      "loss": 0.8602,
+      "step": 4529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2376932173787308e-05,
+      "loss": 0.9015,
+      "step": 4530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2371657790698986e-05,
+      "loss": 0.9444,
+      "step": 4531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2366382708421154e-05,
+      "loss": 0.9032,
+      "step": 4532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2361106928508971e-05,
+      "loss": 0.9215,
+      "step": 4533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2355830452517786e-05,
+      "loss": 0.8758,
+      "step": 4534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2350553282003163e-05,
+      "loss": 0.819,
+      "step": 4535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2345275418520863e-05,
+      "loss": 0.873,
+      "step": 4536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.233999686362686e-05,
+      "loss": 0.8054,
+      "step": 4537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2334717618877324e-05,
+      "loss": 0.9274,
+      "step": 4538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2329437685828634e-05,
+      "loss": 0.9346,
+      "step": 4539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.232415706603736e-05,
+      "loss": 0.8759,
+      "step": 4540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.23188757610603e-05,
+      "loss": 0.9444,
+      "step": 4541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2313593772454428e-05,
+      "loss": 0.8561,
+      "step": 4542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2308311101776933e-05,
+      "loss": 0.8212,
+      "step": 4543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2303027750585195e-05,
+      "loss": 0.9592,
+      "step": 4544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2297743720436807e-05,
+      "loss": 0.9015,
+      "step": 4545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2292459012889555e-05,
+      "loss": 0.8656,
+      "step": 4546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2287173629501428e-05,
+      "loss": 0.853,
+      "step": 4547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2281887571830609e-05,
+      "loss": 0.9062,
+      "step": 4548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2276600841435485e-05,
+      "loss": 0.9392,
+      "step": 4549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.227131343987464e-05,
+      "loss": 0.9194,
+      "step": 4550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2266025368706851e-05,
+      "loss": 0.9974,
+      "step": 4551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.22607366294911e-05,
+      "loss": 0.8732,
+      "step": 4552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.225544722378656e-05,
+      "loss": 0.9026,
+      "step": 4553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.225015715315261e-05,
+      "loss": 0.9079,
+      "step": 4554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2244866419148812e-05,
+      "loss": 0.3811,
+      "step": 4555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2239575023334931e-05,
+      "loss": 1.0109,
+      "step": 4556
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2234282967270926e-05,
+      "loss": 0.8704,
+      "step": 4557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2228990252516951e-05,
+      "loss": 0.8706,
+      "step": 4558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2223696880633349e-05,
+      "loss": 1.0089,
+      "step": 4559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2218402853180669e-05,
+      "loss": 0.8893,
+      "step": 4560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.221310817171964e-05,
+      "loss": 0.8853,
+      "step": 4561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2207812837811194e-05,
+      "loss": 0.8965,
+      "step": 4562
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2202516853016447e-05,
+      "loss": 0.9014,
+      "step": 4563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.219722021889671e-05,
+      "loss": 0.9321,
+      "step": 4564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2191922937013489e-05,
+      "loss": 0.4044,
+      "step": 4565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2186625008928479e-05,
+      "loss": 0.8957,
+      "step": 4566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.218132643620356e-05,
+      "loss": 0.8804,
+      "step": 4567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.217602722040081e-05,
+      "loss": 0.872,
+      "step": 4568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2170727363082495e-05,
+      "loss": 0.9045,
+      "step": 4569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2165426865811061e-05,
+      "loss": 0.9139,
+      "step": 4570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2160125730149157e-05,
+      "loss": 0.9582,
+      "step": 4571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.215482395765961e-05,
+      "loss": 0.9132,
+      "step": 4572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2149521549905441e-05,
+      "loss": 0.8497,
+      "step": 4573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2144218508449852e-05,
+      "loss": 0.91,
+      "step": 4574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2138914834856237e-05,
+      "loss": 0.9404,
+      "step": 4575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2133610530688169e-05,
+      "loss": 0.9019,
+      "step": 4576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2128305597509422e-05,
+      "loss": 0.9181,
+      "step": 4577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2123000036883939e-05,
+      "loss": 0.8318,
+      "step": 4578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2117693850375858e-05,
+      "loss": 0.9282,
+      "step": 4579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2112387039549492e-05,
+      "loss": 0.9182,
+      "step": 4580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2107079605969351e-05,
+      "loss": 0.8782,
+      "step": 4581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2101771551200121e-05,
+      "loss": 0.9212,
+      "step": 4582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2096462876806672e-05,
+      "loss": 0.8954,
+      "step": 4583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.209115358435405e-05,
+      "loss": 0.9608,
+      "step": 4584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.20858436754075e-05,
+      "loss": 0.8937,
+      "step": 4585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2080533151532432e-05,
+      "loss": 0.8779,
+      "step": 4586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2075222014294448e-05,
+      "loss": 0.865,
+      "step": 4587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2069910265259323e-05,
+      "loss": 0.8729,
+      "step": 4588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2064597905993018e-05,
+      "loss": 0.9257,
+      "step": 4589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.205928493806167e-05,
+      "loss": 0.8856,
+      "step": 4590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2053971363031605e-05,
+      "loss": 0.9052,
+      "step": 4591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.204865718246931e-05,
+      "loss": 0.9644,
+      "step": 4592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2043342397941466e-05,
+      "loss": 0.9208,
+      "step": 4593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.203802701101493e-05,
+      "loss": 0.3704,
+      "step": 4594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2032711023256727e-05,
+      "loss": 0.9508,
+      "step": 4595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2027394436234072e-05,
+      "loss": 0.8976,
+      "step": 4596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2022077251514343e-05,
+      "loss": 0.938,
+      "step": 4597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2016759470665112e-05,
+      "loss": 0.8946,
+      "step": 4598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2011441095254109e-05,
+      "loss": 0.9414,
+      "step": 4599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2006122126849247e-05,
+      "loss": 0.9222,
+      "step": 4600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2000802567018613e-05,
+      "loss": 0.9208,
+      "step": 4601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1995482417330473e-05,
+      "loss": 0.952,
+      "step": 4602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.199016167935326e-05,
+      "loss": 0.8737,
+      "step": 4603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1984840354655585e-05,
+      "loss": 0.9234,
+      "step": 4604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1979518444806224e-05,
+      "loss": 0.8674,
+      "step": 4605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1974195951374141e-05,
+      "loss": 0.8813,
+      "step": 4606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1968872875928455e-05,
+      "loss": 0.8982,
+      "step": 4607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.196354922003847e-05,
+      "loss": 0.846,
+      "step": 4608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1958224985273648e-05,
+      "loss": 0.902,
+      "step": 4609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1952900173203639e-05,
+      "loss": 0.9256,
+      "step": 4610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1947574785398244e-05,
+      "loss": 0.9024,
+      "step": 4611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1942248823427449e-05,
+      "loss": 0.8928,
+      "step": 4612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.19369222888614e-05,
+      "loss": 0.873,
+      "step": 4613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1931595183270417e-05,
+      "loss": 0.9067,
+      "step": 4614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1926267508224987e-05,
+      "loss": 0.9096,
+      "step": 4615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1920939265295762e-05,
+      "loss": 0.8701,
+      "step": 4616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1915610456053563e-05,
+      "loss": 0.9078,
+      "step": 4617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1910281082069381e-05,
+      "loss": 0.9014,
+      "step": 4618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1904951144914372e-05,
+      "loss": 0.8905,
+      "step": 4619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1899620646159855e-05,
+      "loss": 0.8883,
+      "step": 4620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1894289587377315e-05,
+      "loss": 0.9536,
+      "step": 4621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1888957970138408e-05,
+      "loss": 0.9444,
+      "step": 4622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1883625796014951e-05,
+      "loss": 0.908,
+      "step": 4623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.187829306657892e-05,
+      "loss": 0.8322,
+      "step": 4624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1872959783402461e-05,
+      "loss": 0.952,
+      "step": 4625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1867625948057882e-05,
+      "loss": 0.8915,
+      "step": 4626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1862291562117656e-05,
+      "loss": 0.4036,
+      "step": 4627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1856956627154412e-05,
+      "loss": 0.8925,
+      "step": 4628
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1851621144740945e-05,
+      "loss": 0.4088,
+      "step": 4629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1846285116450214e-05,
+      "loss": 0.9516,
+      "step": 4630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1840948543855336e-05,
+      "loss": 0.9126,
+      "step": 4631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1835611428529581e-05,
+      "loss": 0.9189,
+      "step": 4632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1830273772046395e-05,
+      "loss": 0.9165,
+      "step": 4633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1824935575979373e-05,
+      "loss": 0.8938,
+      "step": 4634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1819596841902267e-05,
+      "loss": 0.9515,
+      "step": 4635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1814257571388994e-05,
+      "loss": 0.8796,
+      "step": 4636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1808917766013627e-05,
+      "loss": 0.8761,
+      "step": 4637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1803577427350401e-05,
+      "loss": 0.8475,
+      "step": 4638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1798236556973694e-05,
+      "loss": 0.9238,
+      "step": 4639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.179289515645806e-05,
+      "loss": 0.87,
+      "step": 4640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1787553227378189e-05,
+      "loss": 0.8599,
+      "step": 4641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1782210771308948e-05,
+      "loss": 0.9368,
+      "step": 4642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1776867789825344e-05,
+      "loss": 0.927,
+      "step": 4643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1771524284502544e-05,
+      "loss": 0.8764,
+      "step": 4644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1766180256915867e-05,
+      "loss": 0.95,
+      "step": 4645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1760835708640794e-05,
+      "loss": 0.8787,
+      "step": 4646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1755490641252951e-05,
+      "loss": 0.9223,
+      "step": 4647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1750145056328119e-05,
+      "loss": 0.8819,
+      "step": 4648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1744798955442229e-05,
+      "loss": 0.892,
+      "step": 4649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1739452340171376e-05,
+      "loss": 0.8856,
+      "step": 4650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1734105212091789e-05,
+      "loss": 0.9355,
+      "step": 4651
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1728757572779863e-05,
+      "loss": 0.8903,
+      "step": 4652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1723409423812135e-05,
+      "loss": 0.3742,
+      "step": 4653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1718060766765298e-05,
+      "loss": 0.886,
+      "step": 4654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.171271160321619e-05,
+      "loss": 0.9428,
+      "step": 4655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.17073619347418e-05,
+      "loss": 0.9105,
+      "step": 4656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1702011762919268e-05,
+      "loss": 0.9311,
+      "step": 4657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1696661089325878e-05,
+      "loss": 0.924,
+      "step": 4658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1691309915539069e-05,
+      "loss": 0.8195,
+      "step": 4659
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1685958243136418e-05,
+      "loss": 0.9235,
+      "step": 4660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1680606073695652e-05,
+      "loss": 0.8376,
+      "step": 4661
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1675253408794656e-05,
+      "loss": 0.8819,
+      "step": 4662
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1669900250011444e-05,
+      "loss": 0.8692,
+      "step": 4663
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1664546598924184e-05,
+      "loss": 0.9199,
+      "step": 4664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1659192457111193e-05,
+      "loss": 0.8574,
+      "step": 4665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1653837826150918e-05,
+      "loss": 0.9001,
+      "step": 4666
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1648482707621972e-05,
+      "loss": 0.8714,
+      "step": 4667
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1643127103103094e-05,
+      "loss": 0.936,
+      "step": 4668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1637771014173172e-05,
+      "loss": 0.8903,
+      "step": 4669
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1632414442411233e-05,
+      "loss": 0.3717,
+      "step": 4670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1627057389396462e-05,
+      "loss": 0.8767,
+      "step": 4671
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1621699856708164e-05,
+      "loss": 0.3909,
+      "step": 4672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.16163418459258e-05,
+      "loss": 0.9017,
+      "step": 4673
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1610983358628962e-05,
+      "loss": 0.8844,
+      "step": 4674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.16056243963974e-05,
+      "loss": 0.8789,
+      "step": 4675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1600264960810978e-05,
+      "loss": 0.8948,
+      "step": 4676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1594905053449724e-05,
+      "loss": 0.855,
+      "step": 4677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.158954467589379e-05,
+      "loss": 0.89,
+      "step": 4678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.158418382972347e-05,
+      "loss": 0.8833,
+      "step": 4679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.15788225165192e-05,
+      "loss": 0.3668,
+      "step": 4680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1573460737861555e-05,
+      "loss": 0.9074,
+      "step": 4681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1568098495331235e-05,
+      "loss": 0.846,
+      "step": 4682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.156273579050909e-05,
+      "loss": 0.3841,
+      "step": 4683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1557372624976105e-05,
+      "loss": 0.8897,
+      "step": 4684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1552009000313387e-05,
+      "loss": 0.8758,
+      "step": 4685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1546644918102197e-05,
+      "loss": 0.886,
+      "step": 4686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.154128037992392e-05,
+      "loss": 0.8878,
+      "step": 4687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1535915387360081e-05,
+      "loss": 0.8915,
+      "step": 4688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.153054994199233e-05,
+      "loss": 0.8471,
+      "step": 4689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.152518404540246e-05,
+      "loss": 0.8721,
+      "step": 4690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1519817699172392e-05,
+      "loss": 0.9347,
+      "step": 4691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1514450904884182e-05,
+      "loss": 0.9346,
+      "step": 4692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1509083664120017e-05,
+      "loss": 0.8974,
+      "step": 4693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1503715978462216e-05,
+      "loss": 0.8577,
+      "step": 4694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1498347849493227e-05,
+      "loss": 0.3857,
+      "step": 4695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1492979278795635e-05,
+      "loss": 0.3443,
+      "step": 4696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1487610267952143e-05,
+      "loss": 0.923,
+      "step": 4697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1482240818545597e-05,
+      "loss": 0.8674,
+      "step": 4698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1476870932158967e-05,
+      "loss": 0.92,
+      "step": 4699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1471500610375352e-05,
+      "loss": 0.8701,
+      "step": 4700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1466129854777977e-05,
+      "loss": 0.853,
+      "step": 4701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1460758666950196e-05,
+      "loss": 0.9004,
+      "step": 4702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1455387048475492e-05,
+      "loss": 0.8967,
+      "step": 4703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1450015000937476e-05,
+      "loss": 0.8912,
+      "step": 4704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1444642525919883e-05,
+      "loss": 0.9353,
+      "step": 4705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1439269625006573e-05,
+      "loss": 0.8819,
+      "step": 4706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1433896299781538e-05,
+      "loss": 0.9218,
+      "step": 4707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1428522551828885e-05,
+      "loss": 0.8796,
+      "step": 4708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.9304,
+      "step": 4709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1417773794077801e-05,
+      "loss": 0.8921,
+      "step": 4710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1412398787448222e-05,
+      "loss": 0.8793,
+      "step": 4711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1407023364428715e-05,
+      "loss": 0.9542,
+      "step": 4712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1401647526604018e-05,
+      "loss": 0.8246,
+      "step": 4713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1396271275558975e-05,
+      "loss": 0.854,
+      "step": 4714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1390894612878571e-05,
+      "loss": 0.9082,
+      "step": 4715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1385517540147897e-05,
+      "loss": 0.9159,
+      "step": 4716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1380140058952174e-05,
+      "loss": 0.9452,
+      "step": 4717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1374762170876731e-05,
+      "loss": 0.8801,
+      "step": 4718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1369383877507035e-05,
+      "loss": 0.8997,
+      "step": 4719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1364005180428658e-05,
+      "loss": 0.9369,
+      "step": 4720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.13586260812273e-05,
+      "loss": 0.8397,
+      "step": 4721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1353246581488773e-05,
+      "loss": 0.9441,
+      "step": 4722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1347866682799007e-05,
+      "loss": 0.8651,
+      "step": 4723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1342486386744058e-05,
+      "loss": 0.833,
+      "step": 4724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1337105694910093e-05,
+      "loss": 0.9501,
+      "step": 4725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1331724608883389e-05,
+      "loss": 0.3814,
+      "step": 4726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1326343130250357e-05,
+      "loss": 0.9262,
+      "step": 4727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1320961260597505e-05,
+      "loss": 0.8557,
+      "step": 4728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1315579001511467e-05,
+      "loss": 0.9295,
+      "step": 4729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1310196354578993e-05,
+      "loss": 0.9306,
+      "step": 4730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1304813321386934e-05,
+      "loss": 0.8927,
+      "step": 4731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1299429903522274e-05,
+      "loss": 0.8875,
+      "step": 4732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1294046102572097e-05,
+      "loss": 0.9594,
+      "step": 4733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1288661920123604e-05,
+      "loss": 0.907,
+      "step": 4734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1283277357764104e-05,
+      "loss": 0.9007,
+      "step": 4735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1277892417081033e-05,
+      "loss": 0.9847,
+      "step": 4736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1272507099661919e-05,
+      "loss": 0.866,
+      "step": 4737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1267121407094412e-05,
+      "loss": 0.8567,
+      "step": 4738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1261735340966269e-05,
+      "loss": 0.9112,
+      "step": 4739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1256348902865363e-05,
+      "loss": 0.8891,
+      "step": 4740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.125096209437967e-05,
+      "loss": 0.364,
+      "step": 4741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1245574917097275e-05,
+      "loss": 0.9087,
+      "step": 4742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1240187372606376e-05,
+      "loss": 0.8967,
+      "step": 4743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1234799462495278e-05,
+      "loss": 0.8851,
+      "step": 4744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1229411188352392e-05,
+      "loss": 0.8789,
+      "step": 4745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1224022551766242e-05,
+      "loss": 0.8822,
+      "step": 4746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1218633554325449e-05,
+      "loss": 0.9358,
+      "step": 4747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1213244197618745e-05,
+      "loss": 0.8847,
+      "step": 4748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1207854483234974e-05,
+      "loss": 0.9066,
+      "step": 4749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1202464412763075e-05,
+      "loss": 0.8497,
+      "step": 4750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1197073987792099e-05,
+      "loss": 0.9262,
+      "step": 4751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1191683209911202e-05,
+      "loss": 0.8571,
+      "step": 4752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1186292080709639e-05,
+      "loss": 0.9462,
+      "step": 4753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1180900601776775e-05,
+      "loss": 0.8958,
+      "step": 4754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1175508774702066e-05,
+      "loss": 0.8244,
+      "step": 4755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1170116601075085e-05,
+      "loss": 0.9687,
+      "step": 4756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1164724082485502e-05,
+      "loss": 0.925,
+      "step": 4757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1159331220523088e-05,
+      "loss": 0.8972,
+      "step": 4758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.115393801677771e-05,
+      "loss": 0.8719,
+      "step": 4759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1148544472839345e-05,
+      "loss": 0.9579,
+      "step": 4760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1143150590298066e-05,
+      "loss": 0.8357,
+      "step": 4761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1137756370744046e-05,
+      "loss": 0.8945,
+      "step": 4762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1132361815767554e-05,
+      "loss": 0.8924,
+      "step": 4763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1126966926958965e-05,
+      "loss": 0.9052,
+      "step": 4764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1121571705908748e-05,
+      "loss": 0.9258,
+      "step": 4765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1116176154207469e-05,
+      "loss": 0.9537,
+      "step": 4766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1110780273445794e-05,
+      "loss": 0.3696,
+      "step": 4767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1105384065214485e-05,
+      "loss": 0.8359,
+      "step": 4768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1099987531104401e-05,
+      "loss": 0.9418,
+      "step": 4769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1094590672706495e-05,
+      "loss": 0.918,
+      "step": 4770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1089193491611819e-05,
+      "loss": 0.8857,
+      "step": 4771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1083795989411518e-05,
+      "loss": 0.8528,
+      "step": 4772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1078398167696831e-05,
+      "loss": 0.8909,
+      "step": 4773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1073000028059095e-05,
+      "loss": 0.8571,
+      "step": 4774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1067601572089732e-05,
+      "loss": 0.8955,
+      "step": 4775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1062202801380273e-05,
+      "loss": 0.8761,
+      "step": 4776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1056803717522322e-05,
+      "loss": 0.9233,
+      "step": 4777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1051404322107595e-05,
+      "loss": 0.8808,
+      "step": 4778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1046004616727877e-05,
+      "loss": 0.8544,
+      "step": 4779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.104060460297507e-05,
+      "loss": 0.9034,
+      "step": 4780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1035204282441152e-05,
+      "loss": 0.8974,
+      "step": 4781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1029803656718192e-05,
+      "loss": 0.9226,
+      "step": 4782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1024402727398347e-05,
+      "loss": 0.8654,
+      "step": 4783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1019001496073873e-05,
+      "loss": 0.8995,
+      "step": 4784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1013599964337107e-05,
+      "loss": 0.8879,
+      "step": 4785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1008198133780479e-05,
+      "loss": 0.8276,
+      "step": 4786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1002796005996503e-05,
+      "loss": 0.9203,
+      "step": 4787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0997393582577784e-05,
+      "loss": 0.8919,
+      "step": 4788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0991990865117011e-05,
+      "loss": 0.8496,
+      "step": 4789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0986587855206968e-05,
+      "loss": 0.934,
+      "step": 4790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0981184554440507e-05,
+      "loss": 0.8833,
+      "step": 4791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0975780964410587e-05,
+      "loss": 0.9089,
+      "step": 4792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0970377086710244e-05,
+      "loss": 0.9742,
+      "step": 4793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0964972922932589e-05,
+      "loss": 0.9108,
+      "step": 4794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0959568474670831e-05,
+      "loss": 0.9019,
+      "step": 4795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.095416374351826e-05,
+      "loss": 0.9499,
+      "step": 4796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0948758731068246e-05,
+      "loss": 0.8763,
+      "step": 4797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0943353438914237e-05,
+      "loss": 0.9199,
+      "step": 4798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0937947868649781e-05,
+      "loss": 0.9219,
+      "step": 4799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0932542021868482e-05,
+      "loss": 0.8384,
+      "step": 4800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0927135900164056e-05,
+      "loss": 0.8575,
+      "step": 4801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0921729505130272e-05,
+      "loss": 0.9245,
+      "step": 4802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0916322838361e-05,
+      "loss": 0.9773,
+      "step": 4803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0910915901450173e-05,
+      "loss": 0.8873,
+      "step": 4804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0905508695991822e-05,
+      "loss": 0.8876,
+      "step": 4805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0900101223580042e-05,
+      "loss": 0.8987,
+      "step": 4806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0894693485809016e-05,
+      "loss": 0.9601,
+      "step": 4807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0889285484272993e-05,
+      "loss": 0.9447,
+      "step": 4808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0883877220566322e-05,
+      "loss": 0.8519,
+      "step": 4809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0878468696283404e-05,
+      "loss": 1.0192,
+      "step": 4810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0873059913018735e-05,
+      "loss": 0.3849,
+      "step": 4811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0867650872366877e-05,
+      "loss": 0.9456,
+      "step": 4812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0862241575922474e-05,
+      "loss": 0.921,
+      "step": 4813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0856832025280244e-05,
+      "loss": 0.8766,
+      "step": 4814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0851422222034977e-05,
+      "loss": 0.886,
+      "step": 4815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0846012167781538e-05,
+      "loss": 0.8473,
+      "step": 4816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0840601864114873e-05,
+      "loss": 0.8243,
+      "step": 4817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0835191312629994e-05,
+      "loss": 0.8361,
+      "step": 4818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0829780514921982e-05,
+      "loss": 0.9015,
+      "step": 4819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0824369472586005e-05,
+      "loss": 0.8962,
+      "step": 4820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0818958187217288e-05,
+      "loss": 0.8282,
+      "step": 4821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0813546660411143e-05,
+      "loss": 0.8883,
+      "step": 4822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0808134893762935e-05,
+      "loss": 0.9309,
+      "step": 4823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0802722888868116e-05,
+      "loss": 0.9382,
+      "step": 4824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0797310647322199e-05,
+      "loss": 0.954,
+      "step": 4825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0791898170720772e-05,
+      "loss": 0.9377,
+      "step": 4826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0786485460659486e-05,
+      "loss": 0.9616,
+      "step": 4827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0781072518734065e-05,
+      "loss": 0.9151,
+      "step": 4828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0775659346540303e-05,
+      "loss": 0.9416,
+      "step": 4829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0770245945674061e-05,
+      "loss": 0.8666,
+      "step": 4830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0764832317731262e-05,
+      "loss": 0.8911,
+      "step": 4831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.07594184643079e-05,
+      "loss": 0.8723,
+      "step": 4832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.075400438700004e-05,
+      "loss": 1.0159,
+      "step": 4833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0748590087403805e-05,
+      "loss": 0.8967,
+      "step": 4834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.074317556711539e-05,
+      "loss": 0.8606,
+      "step": 4835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0737760827731048e-05,
+      "loss": 0.8944,
+      "step": 4836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0732345870847104e-05,
+      "loss": 0.8757,
+      "step": 4837
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0726930698059944e-05,
+      "loss": 0.9204,
+      "step": 4838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0721515310966017e-05,
+      "loss": 0.9023,
+      "step": 4839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0716099711161833e-05,
+      "loss": 0.8812,
+      "step": 4840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0710683900243973e-05,
+      "loss": 0.815,
+      "step": 4841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0705267879809069e-05,
+      "loss": 0.8954,
+      "step": 4842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0699851651453826e-05,
+      "loss": 0.8781,
+      "step": 4843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0694435216774997e-05,
+      "loss": 0.9039,
+      "step": 4844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.068901857736941e-05,
+      "loss": 0.8911,
+      "step": 4845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0683601734833943e-05,
+      "loss": 0.8887,
+      "step": 4846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0678184690765545e-05,
+      "loss": 0.864,
+      "step": 4847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0672767446761208e-05,
+      "loss": 0.8787,
+      "step": 4848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0667350004417996e-05,
+      "loss": 0.932,
+      "step": 4849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0661932365333028e-05,
+      "loss": 0.4002,
+      "step": 4850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0656514531103484e-05,
+      "loss": 0.8761,
+      "step": 4851
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.065109650332659e-05,
+      "loss": 0.9551,
+      "step": 4852
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0645678283599643e-05,
+      "loss": 0.9198,
+      "step": 4853
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.064025987351999e-05,
+      "loss": 0.9084,
+      "step": 4854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0634841274685035e-05,
+      "loss": 0.855,
+      "step": 4855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0629422488692236e-05,
+      "loss": 0.9365,
+      "step": 4856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.062400351713911e-05,
+      "loss": 0.8903,
+      "step": 4857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0618584361623224e-05,
+      "loss": 0.3702,
+      "step": 4858
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0613165023742206e-05,
+      "loss": 0.8308,
+      "step": 4859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0607745505093727e-05,
+      "loss": 0.9538,
+      "step": 4860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0602325807275521e-05,
+      "loss": 0.899,
+      "step": 4861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0596905931885374e-05,
+      "loss": 0.8709,
+      "step": 4862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0591485880521119e-05,
+      "loss": 0.8553,
+      "step": 4863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0586065654780645e-05,
+      "loss": 0.886,
+      "step": 4864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.058064525626189e-05,
+      "loss": 0.9474,
+      "step": 4865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0575224686562846e-05,
+      "loss": 0.9085,
+      "step": 4866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.056980394728155e-05,
+      "loss": 0.3902,
+      "step": 4867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0564383040016098e-05,
+      "loss": 0.8572,
+      "step": 4868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0558961966364623e-05,
+      "loss": 0.9267,
+      "step": 4869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.055354072792532e-05,
+      "loss": 0.8654,
+      "step": 4870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0548119326296424e-05,
+      "loss": 0.8927,
+      "step": 4871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0542697763076226e-05,
+      "loss": 0.8819,
+      "step": 4872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0537276039863049e-05,
+      "loss": 0.8866,
+      "step": 4873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0531854158255286e-05,
+      "loss": 0.8653,
+      "step": 4874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0526432119851356e-05,
+      "loss": 0.9257,
+      "step": 4875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0521009926249736e-05,
+      "loss": 0.8934,
+      "step": 4876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.051558757904894e-05,
+      "loss": 0.9787,
+      "step": 4877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0510165079847544e-05,
+      "loss": 0.9064,
+      "step": 4878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0504742430244144e-05,
+      "loss": 0.8953,
+      "step": 4879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0499319631837403e-05,
+      "loss": 0.3844,
+      "step": 4880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0493896686226018e-05,
+      "loss": 0.8435,
+      "step": 4881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0488473595008723e-05,
+      "loss": 0.8787,
+      "step": 4882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.048305035978431e-05,
+      "loss": 0.8961,
+      "step": 4883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0477626982151603e-05,
+      "loss": 0.9483,
+      "step": 4884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0472203463709467e-05,
+      "loss": 0.9318,
+      "step": 4885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0466779806056816e-05,
+      "loss": 0.9505,
+      "step": 4886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.04613560107926e-05,
+      "loss": 0.9312,
+      "step": 4887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.045593207951581e-05,
+      "loss": 0.8462,
+      "step": 4888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0450508013825476e-05,
+      "loss": 0.8609,
+      "step": 4889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0445083815320672e-05,
+      "loss": 0.8515,
+      "step": 4890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0439659485600508e-05,
+      "loss": 0.8857,
+      "step": 4891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0434235026264132e-05,
+      "loss": 0.8999,
+      "step": 4892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0428810438910733e-05,
+      "loss": 0.8347,
+      "step": 4893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0423385725139532e-05,
+      "loss": 0.8204,
+      "step": 4894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0417960886549799e-05,
+      "loss": 0.8564,
+      "step": 4895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0412535924740824e-05,
+      "loss": 0.8538,
+      "step": 4896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0407110841311948e-05,
+      "loss": 0.9223,
+      "step": 4897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.040168563786254e-05,
+      "loss": 0.8521,
+      "step": 4898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0396260315992009e-05,
+      "loss": 0.8961,
+      "step": 4899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0390834877299793e-05,
+      "loss": 0.8504,
+      "step": 4900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0385409323385373e-05,
+      "loss": 0.9209,
+      "step": 4901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0379983655848254e-05,
+      "loss": 0.3875,
+      "step": 4902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0374557876287984e-05,
+      "loss": 0.8582,
+      "step": 4903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0369131986304138e-05,
+      "loss": 0.9099,
+      "step": 4904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0363705987496322e-05,
+      "loss": 0.8911,
+      "step": 4905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0358279881464182e-05,
+      "loss": 0.8974,
+      "step": 4906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0352853669807385e-05,
+      "loss": 0.8323,
+      "step": 4907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0347427354125643e-05,
+      "loss": 0.8752,
+      "step": 4908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0342000936018682e-05,
+      "loss": 0.9477,
+      "step": 4909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0336574417086271e-05,
+      "loss": 0.8883,
+      "step": 4910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0331147798928205e-05,
+      "loss": 0.8922,
+      "step": 4911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0325721083144306e-05,
+      "loss": 0.8135,
+      "step": 4912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0320294271334428e-05,
+      "loss": 0.8759,
+      "step": 4913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0314867365098451e-05,
+      "loss": 0.9168,
+      "step": 4914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0309440366036284e-05,
+      "loss": 0.8765,
+      "step": 4915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0304013275747863e-05,
+      "loss": 0.9429,
+      "step": 4916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0298586095833152e-05,
+      "loss": 0.9267,
+      "step": 4917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0293158827892136e-05,
+      "loss": 0.8168,
+      "step": 4918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0287731473524836e-05,
+      "loss": 0.9372,
+      "step": 4919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.028230403433129e-05,
+      "loss": 0.8806,
+      "step": 4920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0276876511911564e-05,
+      "loss": 0.8612,
+      "step": 4921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0271448907865748e-05,
+      "loss": 0.8708,
+      "step": 4922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0266021223793955e-05,
+      "loss": 0.367,
+      "step": 4923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.026059346129633e-05,
+      "loss": 0.9494,
+      "step": 4924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0255165621973026e-05,
+      "loss": 0.8252,
+      "step": 4925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.024973770742423e-05,
+      "loss": 0.8899,
+      "step": 4926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0244309719250149e-05,
+      "loss": 0.8893,
+      "step": 4927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0238881659051007e-05,
+      "loss": 0.9239,
+      "step": 4928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0233453528427055e-05,
+      "loss": 0.3885,
+      "step": 4929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0228025328978563e-05,
+      "loss": 0.8838,
+      "step": 4930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0222597062305824e-05,
+      "loss": 0.8591,
+      "step": 4931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0217168730009142e-05,
+      "loss": 0.8789,
+      "step": 4932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0211740333688852e-05,
+      "loss": 0.8623,
+      "step": 4933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0206311874945291e-05,
+      "loss": 0.9548,
+      "step": 4934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0200883355378839e-05,
+      "loss": 0.9017,
+      "step": 4935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0195454776589872e-05,
+      "loss": 0.8881,
+      "step": 4936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0190026140178796e-05,
+      "loss": 0.8661,
+      "step": 4937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0184597447746025e-05,
+      "loss": 0.9158,
+      "step": 4938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0179168700892001e-05,
+      "loss": 0.9384,
+      "step": 4939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0173739901217168e-05,
+      "loss": 0.8696,
+      "step": 4940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0168311050322e-05,
+      "loss": 0.3787,
+      "step": 4941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0162882149806967e-05,
+      "loss": 0.9069,
+      "step": 4942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0157453201272583e-05,
+      "loss": 0.9417,
+      "step": 4943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0152024206319344e-05,
+      "loss": 0.8846,
+      "step": 4944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0146595166547784e-05,
+      "loss": 0.9362,
+      "step": 4945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0141166083558436e-05,
+      "loss": 0.9034,
+      "step": 4946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0135736958951848e-05,
+      "loss": 0.9494,
+      "step": 4947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0130307794328586e-05,
+      "loss": 0.9104,
+      "step": 4948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0124878591289228e-05,
+      "loss": 0.8694,
+      "step": 4949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0119449351434353e-05,
+      "loss": 0.9209,
+      "step": 4950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0114020076364563e-05,
+      "loss": 0.8016,
+      "step": 4951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0108590767680464e-05,
+      "loss": 0.9233,
+      "step": 4952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0103161426982668e-05,
+      "loss": 0.9105,
+      "step": 4953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0097732055871807e-05,
+      "loss": 0.9471,
+      "step": 4954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0092302655948513e-05,
+      "loss": 0.8771,
+      "step": 4955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0086873228813434e-05,
+      "loss": 0.8689,
+      "step": 4956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0081443776067216e-05,
+      "loss": 0.9537,
+      "step": 4957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0076014299310523e-05,
+      "loss": 0.9365,
+      "step": 4958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0070584800144022e-05,
+      "loss": 0.8282,
+      "step": 4959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0065155280168382e-05,
+      "loss": 0.4214,
+      "step": 4960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0059725740984285e-05,
+      "loss": 0.9032,
+      "step": 4961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0054296184192414e-05,
+      "loss": 0.8511,
+      "step": 4962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0048866611393462e-05,
+      "loss": 0.9012,
+      "step": 4963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0043437024188123e-05,
+      "loss": 0.9001,
+      "step": 4964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0038007424177092e-05,
+      "loss": 0.933,
+      "step": 4965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0032577812961077e-05,
+      "loss": 0.912,
+      "step": 4966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0027148192140783e-05,
+      "loss": 0.9428,
+      "step": 4967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0021718563316917e-05,
+      "loss": 0.9019,
+      "step": 4968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0016288928090194e-05,
+      "loss": 0.892,
+      "step": 4969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0010859288061322e-05,
+      "loss": 0.8984,
+      "step": 4970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0005429644831021e-05,
+      "loss": 0.9004,
+      "step": 4971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1e-05,
+      "loss": 0.9076,
+      "step": 4972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.994570355168982e-06,
+      "loss": 0.8628,
+      "step": 4973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.98914071193868e-06,
+      "loss": 0.8597,
+      "step": 4974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.98371107190981e-06,
+      "loss": 0.9043,
+      "step": 4975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.978281436683086e-06,
+      "loss": 0.8737,
+      "step": 4976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.972851807859218e-06,
+      "loss": 0.922,
+      "step": 4977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.967422187038923e-06,
+      "loss": 0.8744,
+      "step": 4978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.96199257582291e-06,
+      "loss": 0.8336,
+      "step": 4979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.95656297581188e-06,
+      "loss": 0.8826,
+      "step": 4980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.951133388606542e-06,
+      "loss": 0.8656,
+      "step": 4981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.94570381580759e-06,
+      "loss": 0.8141,
+      "step": 4982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.94027425901572e-06,
+      "loss": 0.9004,
+      "step": 4983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.934844719831623e-06,
+      "loss": 0.877,
+      "step": 4984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.92941519985598e-06,
+      "loss": 0.8903,
+      "step": 4985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.923985700689477e-06,
+      "loss": 0.8798,
+      "step": 4986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.918556223932785e-06,
+      "loss": 0.8917,
+      "step": 4987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.913126771186569e-06,
+      "loss": 0.8633,
+      "step": 4988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.90769734405149e-06,
+      "loss": 0.8787,
+      "step": 4989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.902267944128197e-06,
+      "loss": 0.8652,
+      "step": 4990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.896838573017337e-06,
+      "loss": 0.9192,
+      "step": 4991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.891409232319543e-06,
+      "loss": 0.9101,
+      "step": 4992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.885979923635438e-06,
+      "loss": 0.9122,
+      "step": 4993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.880550648565648e-06,
+      "loss": 0.8509,
+      "step": 4994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.875121408710774e-06,
+      "loss": 0.9345,
+      "step": 4995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.869692205671415e-06,
+      "loss": 0.8611,
+      "step": 4996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.864263041048154e-06,
+      "loss": 0.8968,
+      "step": 4997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.858833916441568e-06,
+      "loss": 0.9318,
+      "step": 4998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.85340483345222e-06,
+      "loss": 0.8913,
+      "step": 4999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.847975793680657e-06,
+      "loss": 0.8637,
+      "step": 5000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.842546798727417e-06,
+      "loss": 0.9016,
+      "step": 5001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.837117850193032e-06,
+      "loss": 0.9472,
+      "step": 5002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831688949678004e-06,
+      "loss": 0.9084,
+      "step": 5003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.826260098782836e-06,
+      "loss": 0.8702,
+      "step": 5004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.820831299108002e-06,
+      "loss": 0.9162,
+      "step": 5005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.815402552253977e-06,
+      "loss": 0.9123,
+      "step": 5006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.809973859821207e-06,
+      "loss": 0.8731,
+      "step": 5007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.804545223410133e-06,
+      "loss": 0.9107,
+      "step": 5008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.799116644621163e-06,
+      "loss": 0.8829,
+      "step": 5009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.793688125054709e-06,
+      "loss": 0.9118,
+      "step": 5010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.788259666311152e-06,
+      "loss": 0.9044,
+      "step": 5011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.78283126999086e-06,
+      "loss": 0.9074,
+      "step": 5012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.777402937694178e-06,
+      "loss": 0.8431,
+      "step": 5013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.771974671021438e-06,
+      "loss": 0.8651,
+      "step": 5014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.766546471572947e-06,
+      "loss": 0.9109,
+      "step": 5015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.761118340949e-06,
+      "loss": 0.8936,
+      "step": 5016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.755690280749854e-06,
+      "loss": 0.9143,
+      "step": 5017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.750262292575771e-06,
+      "loss": 0.9021,
+      "step": 5018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.744834378026977e-06,
+      "loss": 0.8852,
+      "step": 5019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.739406538703671e-06,
+      "loss": 0.9263,
+      "step": 5020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.733978776206046e-06,
+      "loss": 0.8914,
+      "step": 5021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.728551092134255e-06,
+      "loss": 0.8884,
+      "step": 5022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.723123488088441e-06,
+      "loss": 0.8532,
+      "step": 5023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.717695965668714e-06,
+      "loss": 0.8916,
+      "step": 5024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.712268526475164e-06,
+      "loss": 0.8899,
+      "step": 5025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.706841172107864e-06,
+      "loss": 0.9273,
+      "step": 5026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.701413904166852e-06,
+      "loss": 0.3896,
+      "step": 5027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.695986724252139e-06,
+      "loss": 0.8244,
+      "step": 5028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.69055963396372e-06,
+      "loss": 0.8836,
+      "step": 5029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.685132634901552e-06,
+      "loss": 0.8949,
+      "step": 5030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.679705728665577e-06,
+      "loss": 0.9178,
+      "step": 5031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.674278916855697e-06,
+      "loss": 0.8692,
+      "step": 5032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.668852201071797e-06,
+      "loss": 0.8455,
+      "step": 5033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.663425582913732e-06,
+      "loss": 0.8898,
+      "step": 5034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.65799906398132e-06,
+      "loss": 0.8989,
+      "step": 5035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.65257264587436e-06,
+      "loss": 0.8344,
+      "step": 5036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.647146330192617e-06,
+      "loss": 0.9024,
+      "step": 5037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.641720118535821e-06,
+      "loss": 0.8706,
+      "step": 5038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.636294012503682e-06,
+      "loss": 0.9068,
+      "step": 5039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.630868013695866e-06,
+      "loss": 0.9211,
+      "step": 5040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.625442123712019e-06,
+      "loss": 0.9585,
+      "step": 5041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.620016344151746e-06,
+      "loss": 0.908,
+      "step": 5042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.614590676614628e-06,
+      "loss": 0.8565,
+      "step": 5043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.609165122700209e-06,
+      "loss": 0.9496,
+      "step": 5044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.603739684007995e-06,
+      "loss": 0.8356,
+      "step": 5045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.598314362137464e-06,
+      "loss": 0.9001,
+      "step": 5046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.592889158688058e-06,
+      "loss": 0.906,
+      "step": 5047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.587464075259181e-06,
+      "loss": 0.4076,
+      "step": 5048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.582039113450208e-06,
+      "loss": 0.8419,
+      "step": 5049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.57661427486047e-06,
+      "loss": 0.8924,
+      "step": 5050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.571189561089268e-06,
+      "loss": 0.9243,
+      "step": 5051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.56576497373587e-06,
+      "loss": 0.3396,
+      "step": 5052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.560340514399494e-06,
+      "loss": 0.8672,
+      "step": 5053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.554916184679331e-06,
+      "loss": 0.8739,
+      "step": 5054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.549491986174525e-06,
+      "loss": 0.8874,
+      "step": 5055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.544067920484196e-06,
+      "loss": 0.8594,
+      "step": 5056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.538643989207405e-06,
+      "loss": 0.9046,
+      "step": 5057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.533220193943186e-06,
+      "loss": 0.9047,
+      "step": 5058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.527796536290536e-06,
+      "loss": 0.3621,
+      "step": 5059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.5223730178484e-06,
+      "loss": 0.368,
+      "step": 5060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.516949640215694e-06,
+      "loss": 0.9186,
+      "step": 5061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.511526404991278e-06,
+      "loss": 0.9323,
+      "step": 5062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.506103313773987e-06,
+      "loss": 0.9257,
+      "step": 5063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.5006803681626e-06,
+      "loss": 0.8909,
+      "step": 5064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.49525756975586e-06,
+      "loss": 0.8815,
+      "step": 5065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.489834920152458e-06,
+      "loss": 0.8907,
+      "step": 5066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.48441242095106e-06,
+      "loss": 0.8961,
+      "step": 5067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.478990073750267e-06,
+      "loss": 0.9271,
+      "step": 5068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.473567880148647e-06,
+      "loss": 0.9397,
+      "step": 5069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.468145841744717e-06,
+      "loss": 0.9102,
+      "step": 5070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.462723960136953e-06,
+      "loss": 0.8694,
+      "step": 5071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.45730223692378e-06,
+      "loss": 0.8979,
+      "step": 5072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.45188067370358e-06,
+      "loss": 0.8972,
+      "step": 5073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.44645927207468e-06,
+      "loss": 0.9378,
+      "step": 5074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.441038033635379e-06,
+      "loss": 0.9362,
+      "step": 5075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.435616959983905e-06,
+      "loss": 0.923,
+      "step": 5076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.430196052718454e-06,
+      "loss": 0.9458,
+      "step": 5077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.424775313437158e-06,
+      "loss": 0.9343,
+      "step": 5078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.419354743738114e-06,
+      "loss": 0.3656,
+      "step": 5079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.413934345219358e-06,
+      "loss": 0.9027,
+      "step": 5080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.408514119478884e-06,
+      "loss": 0.9027,
+      "step": 5081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.403094068114627e-06,
+      "loss": 0.8767,
+      "step": 5082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.397674192724479e-06,
+      "loss": 0.8266,
+      "step": 5083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.392254494906276e-06,
+      "loss": 0.9044,
+      "step": 5084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.386834976257797e-06,
+      "loss": 0.8844,
+      "step": 5085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.381415638376778e-06,
+      "loss": 0.8284,
+      "step": 5086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.375996482860893e-06,
+      "loss": 0.9313,
+      "step": 5087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.370577511307766e-06,
+      "loss": 0.9256,
+      "step": 5088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.365158725314968e-06,
+      "loss": 0.9128,
+      "step": 5089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.35974012648001e-06,
+      "loss": 0.9432,
+      "step": 5090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.354321716400357e-06,
+      "loss": 0.939,
+      "step": 5091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.348903496673413e-06,
+      "loss": 0.9008,
+      "step": 5092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.34348546889652e-06,
+      "loss": 0.9008,
+      "step": 5093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.338067634666975e-06,
+      "loss": 0.9168,
+      "step": 5094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.332649995582008e-06,
+      "loss": 0.8707,
+      "step": 5095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.327232553238797e-06,
+      "loss": 0.8893,
+      "step": 5096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.32181530923446e-06,
+      "loss": 0.8961,
+      "step": 5097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.316398265166057e-06,
+      "loss": 0.9081,
+      "step": 5098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.310981422630592e-06,
+      "loss": 0.8708,
+      "step": 5099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.305564783225006e-06,
+      "loss": 0.9462,
+      "step": 5100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.30014834854618e-06,
+      "loss": 0.8975,
+      "step": 5101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.294732120190935e-06,
+      "loss": 0.8462,
+      "step": 5102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.289316099756029e-06,
+      "loss": 0.8617,
+      "step": 5103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.28390028883817e-06,
+      "loss": 0.9136,
+      "step": 5104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.278484689033986e-06,
+      "loss": 0.9041,
+      "step": 5105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.273069301940059e-06,
+      "loss": 0.9002,
+      "step": 5106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.267654129152897e-06,
+      "loss": 0.8738,
+      "step": 5107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.262239172268952e-06,
+      "loss": 0.9652,
+      "step": 5108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.256824432884614e-06,
+      "loss": 0.8844,
+      "step": 5109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.251409912596196e-06,
+      "loss": 0.8801,
+      "step": 5110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.245995612999964e-06,
+      "loss": 0.8549,
+      "step": 5111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.240581535692102e-06,
+      "loss": 0.8946,
+      "step": 5112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.235167682268743e-06,
+      "loss": 0.8641,
+      "step": 5113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.229754054325944e-06,
+      "loss": 0.8584,
+      "step": 5114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.224340653459698e-06,
+      "loss": 0.8457,
+      "step": 5115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.218927481265936e-06,
+      "loss": 0.3343,
+      "step": 5116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.213514539340517e-06,
+      "loss": 0.9106,
+      "step": 5117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.20810182927923e-06,
+      "loss": 0.8851,
+      "step": 5118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.202689352677805e-06,
+      "loss": 0.8885,
+      "step": 5119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.197277111131888e-06,
+      "loss": 0.8905,
+      "step": 5120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.19186510623707e-06,
+      "loss": 0.8484,
+      "step": 5121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.186453339588862e-06,
+      "loss": 0.8667,
+      "step": 5122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.181041812782712e-06,
+      "loss": 0.8717,
+      "step": 5123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.175630527413999e-06,
+      "loss": 0.911,
+      "step": 5124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.17021948507802e-06,
+      "loss": 0.8561,
+      "step": 5125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.164808687370011e-06,
+      "loss": 0.8833,
+      "step": 5126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.15939813588513e-06,
+      "loss": 0.8921,
+      "step": 5127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.153987832218463e-06,
+      "loss": 0.9465,
+      "step": 5128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.148577777965028e-06,
+      "loss": 0.8999,
+      "step": 5129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.14316797471976e-06,
+      "loss": 0.9338,
+      "step": 5130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.137758424077525e-06,
+      "loss": 0.891,
+      "step": 5131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.132349127633125e-06,
+      "loss": 0.8992,
+      "step": 5132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.126940086981268e-06,
+      "loss": 0.9176,
+      "step": 5133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.121531303716598e-06,
+      "loss": 0.8493,
+      "step": 5134
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.116122779433681e-06,
+      "loss": 0.8325,
+      "step": 5135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.11071451572701e-06,
+      "loss": 0.9265,
+      "step": 5136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.10530651419099e-06,
+      "loss": 0.8466,
+      "step": 5137
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.099898776419963e-06,
+      "loss": 0.8691,
+      "step": 5138
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.094491304008178e-06,
+      "loss": 0.8921,
+      "step": 5139
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.089084098549827e-06,
+      "loss": 0.8928,
+      "step": 5140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.083677161639002e-06,
+      "loss": 0.9,
+      "step": 5141
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.078270494869732e-06,
+      "loss": 0.8727,
+      "step": 5142
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.072864099835947e-06,
+      "loss": 0.9055,
+      "step": 5143
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.06745797813152e-06,
+      "loss": 0.869,
+      "step": 5144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.062052131350225e-06,
+      "loss": 0.9121,
+      "step": 5145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.056646561085766e-06,
+      "loss": 0.8972,
+      "step": 5146
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.051241268931759e-06,
+      "loss": 0.8908,
+      "step": 5147
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.045836256481742e-06,
+      "loss": 0.8877,
+      "step": 5148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.04043152532917e-06,
+      "loss": 0.8853,
+      "step": 5149
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.035027077067413e-06,
+      "loss": 0.8771,
+      "step": 5150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.02962291328976e-06,
+      "loss": 0.4086,
+      "step": 5151
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.024219035589415e-06,
+      "loss": 0.8979,
+      "step": 5152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.018815445559495e-06,
+      "loss": 0.9253,
+      "step": 5153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.013412144793039e-06,
+      "loss": 0.4001,
+      "step": 5154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.008009134882989e-06,
+      "loss": 0.8968,
+      "step": 5155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.002606417422218e-06,
+      "loss": 0.8646,
+      "step": 5156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.9972039940035e-06,
+      "loss": 0.9054,
+      "step": 5157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.991801866219523e-06,
+      "loss": 0.859,
+      "step": 5158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.986400035662897e-06,
+      "loss": 0.8592,
+      "step": 5159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.98099850392613e-06,
+      "loss": 0.9138,
+      "step": 5160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.975597272601658e-06,
+      "loss": 0.8706,
+      "step": 5161
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.970196343281815e-06,
+      "loss": 0.9194,
+      "step": 5162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.96479571755885e-06,
+      "loss": 0.3968,
+      "step": 5163
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.959395397024928e-06,
+      "loss": 0.8306,
+      "step": 5164
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.953995383272123e-06,
+      "loss": 0.8641,
+      "step": 5165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.948595677892408e-06,
+      "loss": 0.8506,
+      "step": 5166
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.94319628247768e-06,
+      "loss": 0.9153,
+      "step": 5167
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.93779719861973e-06,
+      "loss": 0.8656,
+      "step": 5168
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.93239842791027e-06,
+      "loss": 0.8794,
+      "step": 5169
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.92699997194091e-06,
+      "loss": 0.8942,
+      "step": 5170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.921601832303174e-06,
+      "loss": 0.8583,
+      "step": 5171
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.916204010588483e-06,
+      "loss": 0.9031,
+      "step": 5172
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.910806508388183e-06,
+      "loss": 0.8971,
+      "step": 5173
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.905409327293508e-06,
+      "loss": 0.8944,
+      "step": 5174
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.900012468895602e-06,
+      "loss": 0.9024,
+      "step": 5175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.894615934785519e-06,
+      "loss": 0.8666,
+      "step": 5176
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.88921972655421e-06,
+      "loss": 0.864,
+      "step": 5177
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.883823845792534e-06,
+      "loss": 0.8403,
+      "step": 5178
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.878428294091257e-06,
+      "loss": 0.8602,
+      "step": 5179
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.873033073041037e-06,
+      "loss": 0.9309,
+      "step": 5180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.867638184232446e-06,
+      "loss": 0.8742,
+      "step": 5181
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.862243629255957e-06,
+      "loss": 0.9297,
+      "step": 5182
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.856849409701936e-06,
+      "loss": 0.8421,
+      "step": 5183
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.851455527160658e-06,
+      "loss": 0.9038,
+      "step": 5184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.846061983222293e-06,
+      "loss": 0.9235,
+      "step": 5185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.840668779476917e-06,
+      "loss": 0.9555,
+      "step": 5186
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.835275917514501e-06,
+      "loss": 0.8577,
+      "step": 5187
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.829883398924915e-06,
+      "loss": 0.8895,
+      "step": 5188
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.824491225297935e-06,
+      "loss": 0.8828,
+      "step": 5189
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.81909939822323e-06,
+      "loss": 0.8685,
+      "step": 5190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.813707919290362e-06,
+      "loss": 0.8509,
+      "step": 5191
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.808316790088801e-06,
+      "loss": 0.9376,
+      "step": 5192
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.802926012207902e-06,
+      "loss": 0.9086,
+      "step": 5193
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.79753558723693e-06,
+      "loss": 0.8415,
+      "step": 5194
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.792145516765031e-06,
+      "loss": 0.8971,
+      "step": 5195
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.786755802381255e-06,
+      "loss": 0.8252,
+      "step": 5196
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.781366445674555e-06,
+      "loss": 0.8676,
+      "step": 5197
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.77597744823376e-06,
+      "loss": 0.8589,
+      "step": 5198
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.77058881164761e-06,
+      "loss": 0.8617,
+      "step": 5199
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.765200537504725e-06,
+      "loss": 0.8715,
+      "step": 5200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.759812627393628e-06,
+      "loss": 0.962,
+      "step": 5201
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.754425082902728e-06,
+      "loss": 0.955,
+      "step": 5202
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.749037905620334e-06,
+      "loss": 0.8924,
+      "step": 5203
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.743651097134637e-06,
+      "loss": 0.8201,
+      "step": 5204
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.738264659033731e-06,
+      "loss": 0.8991,
+      "step": 5205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.73287859290559e-06,
+      "loss": 0.9271,
+      "step": 5206
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.727492900338083e-06,
+      "loss": 0.9123,
+      "step": 5207
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.72210758291897e-06,
+      "loss": 0.8997,
+      "step": 5208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.716722642235897e-06,
+      "loss": 0.8751,
+      "step": 5209
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.711338079876401e-06,
+      "loss": 0.8737,
+      "step": 5210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.705953897427908e-06,
+      "loss": 0.367,
+      "step": 5211
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.70057009647773e-06,
+      "loss": 0.8668,
+      "step": 5212
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.695186678613068e-06,
+      "loss": 0.8296,
+      "step": 5213
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.68980364542101e-06,
+      "loss": 0.7355,
+      "step": 5214
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.684420998488536e-06,
+      "loss": 0.864,
+      "step": 5215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.679038739402497e-06,
+      "loss": 0.9209,
+      "step": 5216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.673656869749647e-06,
+      "loss": 0.8658,
+      "step": 5217
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.668275391116613e-06,
+      "loss": 0.8781,
+      "step": 5218
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.662894305089912e-06,
+      "loss": 0.8744,
+      "step": 5219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.657513613255942e-06,
+      "loss": 0.9521,
+      "step": 5220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.652133317200993e-06,
+      "loss": 0.9178,
+      "step": 5221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.646753418511232e-06,
+      "loss": 0.8411,
+      "step": 5222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.641373918772702e-06,
+      "loss": 0.8826,
+      "step": 5223
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.635994819571345e-06,
+      "loss": 0.8913,
+      "step": 5224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.630616122492967e-06,
+      "loss": 0.8763,
+      "step": 5225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.625237829123274e-06,
+      "loss": 0.94,
+      "step": 5226
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.619859941047832e-06,
+      "loss": 0.8952,
+      "step": 5227
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.614482459852104e-06,
+      "loss": 0.8718,
+      "step": 5228
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.609105387121429e-06,
+      "loss": 0.9343,
+      "step": 5229
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.603728724441025e-06,
+      "loss": 0.9447,
+      "step": 5230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.598352473395985e-06,
+      "loss": 0.8317,
+      "step": 5231
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.592976635571287e-06,
+      "loss": 0.8757,
+      "step": 5232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.587601212551781e-06,
+      "loss": 0.9226,
+      "step": 5233
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.5822262059222e-06,
+      "loss": 0.8494,
+      "step": 5234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.8202,
+      "step": 5235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.57147744817112e-06,
+      "loss": 0.9269,
+      "step": 5236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.566103700218465e-06,
+      "loss": 0.8688,
+      "step": 5237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.560730374993429e-06,
+      "loss": 0.8646,
+      "step": 5238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.55535747408012e-06,
+      "loss": 0.8907,
+      "step": 5239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.549984999062526e-06,
+      "loss": 0.9021,
+      "step": 5240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.54461295152451e-06,
+      "loss": 0.8291,
+      "step": 5241
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.539241333049807e-06,
+      "loss": 0.9346,
+      "step": 5242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.533870145222028e-06,
+      "loss": 0.8749,
+      "step": 5243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.528499389624653e-06,
+      "loss": 0.8409,
+      "step": 5244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.523129067841033e-06,
+      "loss": 0.8676,
+      "step": 5245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.517759181454403e-06,
+      "loss": 0.9301,
+      "step": 5246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.512389732047859e-06,
+      "loss": 0.8867,
+      "step": 5247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.507020721204368e-06,
+      "loss": 0.9164,
+      "step": 5248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.501652150506776e-06,
+      "loss": 0.8633,
+      "step": 5249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.496284021537788e-06,
+      "loss": 0.8803,
+      "step": 5250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.490916335879988e-06,
+      "loss": 0.8447,
+      "step": 5251
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.485549095115821e-06,
+      "loss": 0.8412,
+      "step": 5252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.48018230082761e-06,
+      "loss": 0.8996,
+      "step": 5253
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.474815954597541e-06,
+      "loss": 0.8497,
+      "step": 5254
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.469450058007673e-06,
+      "loss": 0.9213,
+      "step": 5255
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.464084612639922e-06,
+      "loss": 0.8492,
+      "step": 5256
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.458719620076083e-06,
+      "loss": 0.8508,
+      "step": 5257
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.453355081897804e-06,
+      "loss": 0.9272,
+      "step": 5258
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.447990999686618e-06,
+      "loss": 0.8768,
+      "step": 5259
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.442627375023902e-06,
+      "loss": 0.877,
+      "step": 5260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.43726420949091e-06,
+      "loss": 0.9534,
+      "step": 5261
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.431901504668766e-06,
+      "loss": 0.8909,
+      "step": 5262
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.426539262138448e-06,
+      "loss": 0.8701,
+      "step": 5263
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.421177483480803e-06,
+      "loss": 0.8666,
+      "step": 5264
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.415816170276533e-06,
+      "loss": 0.8215,
+      "step": 5265
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.410455324106214e-06,
+      "loss": 0.9253,
+      "step": 5266
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.405094946550281e-06,
+      "loss": 0.8448,
+      "step": 5267
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.399735039189026e-06,
+      "loss": 0.8743,
+      "step": 5268
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.394375603602602e-06,
+      "loss": 0.9085,
+      "step": 5269
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.389016641371038e-06,
+      "loss": 0.3927,
+      "step": 5270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.383658154074203e-06,
+      "loss": 0.8316,
+      "step": 5271
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.37830014329184e-06,
+      "loss": 0.8733,
+      "step": 5272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.372942610603542e-06,
+      "loss": 0.8807,
+      "step": 5273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.367585557588769e-06,
+      "loss": 0.8614,
+      "step": 5274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.362228985826834e-06,
+      "loss": 0.9227,
+      "step": 5275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.356872896896913e-06,
+      "loss": 0.8956,
+      "step": 5276
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.351517292378031e-06,
+      "loss": 0.9079,
+      "step": 5277
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.346162173849082e-06,
+      "loss": 0.8977,
+      "step": 5278
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.34080754288881e-06,
+      "loss": 0.8618,
+      "step": 5279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.335453401075818e-06,
+      "loss": 0.9034,
+      "step": 5280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.330099749988558e-06,
+      "loss": 0.9036,
+      "step": 5281
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.324746591205349e-06,
+      "loss": 0.8581,
+      "step": 5282
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.31939392630435e-06,
+      "loss": 0.3378,
+      "step": 5283
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.314041756863589e-06,
+      "loss": 0.7986,
+      "step": 5284
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.308690084460935e-06,
+      "loss": 0.9375,
+      "step": 5285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.303338910674124e-06,
+      "loss": 0.8382,
+      "step": 5286
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.297988237080735e-06,
+      "loss": 0.9749,
+      "step": 5287
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.292638065258203e-06,
+      "loss": 0.8621,
+      "step": 5288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.287288396783814e-06,
+      "loss": 0.9283,
+      "step": 5289
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.281939233234706e-06,
+      "loss": 0.8911,
+      "step": 5290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.27659057618787e-06,
+      "loss": 0.9478,
+      "step": 5291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.27124242722014e-06,
+      "loss": 0.8585,
+      "step": 5292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.265894787908213e-06,
+      "loss": 0.8727,
+      "step": 5293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.260547659828625e-06,
+      "loss": 0.8992,
+      "step": 5294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.255201044557773e-06,
+      "loss": 0.8768,
+      "step": 5295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.249854943671884e-06,
+      "loss": 0.941,
+      "step": 5296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.244509358747052e-06,
+      "loss": 0.8202,
+      "step": 5297
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.239164291359207e-06,
+      "loss": 0.8993,
+      "step": 5298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.233819743084137e-06,
+      "loss": 0.8807,
+      "step": 5299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.228475715497461e-06,
+      "loss": 0.9349,
+      "step": 5300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.223132210174663e-06,
+      "loss": 0.8929,
+      "step": 5301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.217789228691055e-06,
+      "loss": 0.8899,
+      "step": 5302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.212446772621813e-06,
+      "loss": 1.0142,
+      "step": 5303
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.207104843541944e-06,
+      "loss": 0.8686,
+      "step": 5304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.20176344302631e-06,
+      "loss": 0.9092,
+      "step": 5305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.196422572649602e-06,
+      "loss": 0.8584,
+      "step": 5306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.191082233986375e-06,
+      "loss": 0.3634,
+      "step": 5307
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.18574242861101e-06,
+      "loss": 0.8787,
+      "step": 5308
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.180403158097738e-06,
+      "loss": 0.8718,
+      "step": 5309
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.17506442402063e-06,
+      "loss": 0.8987,
+      "step": 5310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.169726227953605e-06,
+      "loss": 0.8664,
+      "step": 5311
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.16438857147042e-06,
+      "loss": 0.8728,
+      "step": 5312
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.15905145614467e-06,
+      "loss": 0.9023,
+      "step": 5313
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.153714883549789e-06,
+      "loss": 0.805,
+      "step": 5314
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.148378855259056e-06,
+      "loss": 0.8865,
+      "step": 5315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.143043372845593e-06,
+      "loss": 0.3529,
+      "step": 5316
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.13770843788235e-06,
+      "loss": 0.8822,
+      "step": 5317
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.13237405194212e-06,
+      "loss": 0.9017,
+      "step": 5318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.12704021659754e-06,
+      "loss": 0.8721,
+      "step": 5319
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.121706933421084e-06,
+      "loss": 0.8697,
+      "step": 5320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.116374203985054e-06,
+      "loss": 0.9285,
+      "step": 5321
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.111042029861595e-06,
+      "loss": 0.8626,
+      "step": 5322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.105710412622686e-06,
+      "loss": 0.3665,
+      "step": 5323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.10037935384015e-06,
+      "loss": 0.3747,
+      "step": 5324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.095048855085633e-06,
+      "loss": 0.8964,
+      "step": 5325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.08971891793062e-06,
+      "loss": 0.8367,
+      "step": 5326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.08438954394644e-06,
+      "loss": 0.9298,
+      "step": 5327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.079060734704241e-06,
+      "loss": 0.8759,
+      "step": 5328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.073732491775015e-06,
+      "loss": 0.9289,
+      "step": 5329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.068404816729584e-06,
+      "loss": 0.8623,
+      "step": 5330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.063077711138601e-06,
+      "loss": 0.8686,
+      "step": 5331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.057751176572555e-06,
+      "loss": 0.9178,
+      "step": 5332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.05242521460176e-06,
+      "loss": 0.3408,
+      "step": 5333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.047099826796363e-06,
+      "loss": 0.8558,
+      "step": 5334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.041775014726354e-06,
+      "loss": 0.9252,
+      "step": 5335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.036450779961533e-06,
+      "loss": 0.8631,
+      "step": 5336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.031127124071549e-06,
+      "loss": 0.3767,
+      "step": 5337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.025804048625862e-06,
+      "loss": 0.893,
+      "step": 5338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.02048155519378e-06,
+      "loss": 0.9175,
+      "step": 5339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.015159645344421e-06,
+      "loss": 0.8482,
+      "step": 5340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.009838320646746e-06,
+      "loss": 0.8825,
+      "step": 5341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.004517582669527e-06,
+      "loss": 0.8882,
+      "step": 5342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.999197432981389e-06,
+      "loss": 0.9107,
+      "step": 5343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.993877873150756e-06,
+      "loss": 0.8453,
+      "step": 5344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.988558904745895e-06,
+      "loss": 0.9244,
+      "step": 5345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.983240529334891e-06,
+      "loss": 0.8672,
+      "step": 5346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.977922748485658e-06,
+      "loss": 0.8354,
+      "step": 5347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.972605563765933e-06,
+      "loss": 0.8239,
+      "step": 5348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.967288976743276e-06,
+      "loss": 0.855,
+      "step": 5349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.961972988985072e-06,
+      "loss": 0.952,
+      "step": 5350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.956657602058534e-06,
+      "loss": 0.8862,
+      "step": 5351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.951342817530693e-06,
+      "loss": 0.9398,
+      "step": 5352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.9460286369684e-06,
+      "loss": 0.8501,
+      "step": 5353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.940715061938332e-06,
+      "loss": 0.3672,
+      "step": 5354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.935402094006985e-06,
+      "loss": 0.8541,
+      "step": 5355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.93008973474068e-06,
+      "loss": 0.8454,
+      "step": 5356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.924777985705556e-06,
+      "loss": 0.9319,
+      "step": 5357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.91946684846757e-06,
+      "loss": 0.8934,
+      "step": 5358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.914156324592501e-06,
+      "loss": 0.9132,
+      "step": 5359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.90884641564595e-06,
+      "loss": 0.8816,
+      "step": 5360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.903537123193332e-06,
+      "loss": 0.9099,
+      "step": 5361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.898228448799884e-06,
+      "loss": 0.8667,
+      "step": 5362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.89292039403065e-06,
+      "loss": 0.8919,
+      "step": 5363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.887612960450513e-06,
+      "loss": 0.914,
+      "step": 5364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.882306149624149e-06,
+      "loss": 0.8361,
+      "step": 5365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.876999963116066e-06,
+      "loss": 0.8037,
+      "step": 5366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.871694402490578e-06,
+      "loss": 0.9106,
+      "step": 5367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.866389469311831e-06,
+      "loss": 0.9222,
+      "step": 5368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.861085165143766e-06,
+      "loss": 0.8617,
+      "step": 5369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.855781491550151e-06,
+      "loss": 0.9474,
+      "step": 5370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.85047845009456e-06,
+      "loss": 0.9235,
+      "step": 5371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.845176042340392e-06,
+      "loss": 0.8924,
+      "step": 5372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.839874269850846e-06,
+      "loss": 0.864,
+      "step": 5373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.834573134188942e-06,
+      "loss": 0.8756,
+      "step": 5374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.829272636917509e-06,
+      "loss": 0.8507,
+      "step": 5375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.82397277959919e-06,
+      "loss": 0.8857,
+      "step": 5376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.818673563796442e-06,
+      "loss": 0.8283,
+      "step": 5377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.813374991071524e-06,
+      "loss": 0.8714,
+      "step": 5378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.808077062986515e-06,
+      "loss": 0.8623,
+      "step": 5379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.802779781103293e-06,
+      "loss": 0.9342,
+      "step": 5380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.797483146983557e-06,
+      "loss": 0.8776,
+      "step": 5381
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.79218716218881e-06,
+      "loss": 0.8924,
+      "step": 5382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.786891828280361e-06,
+      "loss": 0.9003,
+      "step": 5383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.781597146819331e-06,
+      "loss": 0.8373,
+      "step": 5384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.776303119366653e-06,
+      "loss": 0.8916,
+      "step": 5385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.771009747483054e-06,
+      "loss": 0.9374,
+      "step": 5386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.76571703272908e-06,
+      "loss": 0.833,
+      "step": 5387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.760424976665074e-06,
+      "loss": 0.9273,
+      "step": 5388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.755133580851193e-06,
+      "loss": 0.8692,
+      "step": 5389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.749842846847394e-06,
+      "loss": 0.8531,
+      "step": 5390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.74455277621344e-06,
+      "loss": 0.8484,
+      "step": 5391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.739263370508902e-06,
+      "loss": 0.9214,
+      "step": 5392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.733974631293152e-06,
+      "loss": 0.8592,
+      "step": 5393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.728686560125364e-06,
+      "loss": 0.8326,
+      "step": 5394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.723399158564519e-06,
+      "loss": 0.8735,
+      "step": 5395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.718112428169394e-06,
+      "loss": 0.921,
+      "step": 5396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.712826370498577e-06,
+      "loss": 0.9355,
+      "step": 5397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.707540987110448e-06,
+      "loss": 0.8233,
+      "step": 5398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.702256279563194e-06,
+      "loss": 0.9144,
+      "step": 5399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.696972249414807e-06,
+      "loss": 0.8897,
+      "step": 5400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.69168889822307e-06,
+      "loss": 0.8731,
+      "step": 5401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.686406227545575e-06,
+      "loss": 0.8877,
+      "step": 5402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.681124238939701e-06,
+      "loss": 0.9052,
+      "step": 5403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.675842933962641e-06,
+      "loss": 0.9524,
+      "step": 5404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.670562314171373e-06,
+      "loss": 0.8696,
+      "step": 5405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.665282381122681e-06,
+      "loss": 0.9285,
+      "step": 5406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.66000313637314e-06,
+      "loss": 0.8992,
+      "step": 5407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.654724581479138e-06,
+      "loss": 0.9149,
+      "step": 5408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.649446717996838e-06,
+      "loss": 0.8776,
+      "step": 5409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.644169547482216e-06,
+      "loss": 0.8927,
+      "step": 5410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.63889307149103e-06,
+      "loss": 0.8591,
+      "step": 5411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.63361729157885e-06,
+      "loss": 0.9171,
+      "step": 5412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.62834220930102e-06,
+      "loss": 0.8637,
+      "step": 5413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.623067826212697e-06,
+      "loss": 0.8605,
+      "step": 5414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.617794143868817e-06,
+      "loss": 0.8781,
+      "step": 5415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.612521163824123e-06,
+      "loss": 0.8993,
+      "step": 5416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.607248887633145e-06,
+      "loss": 0.9094,
+      "step": 5417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.601977316850202e-06,
+      "loss": 0.8909,
+      "step": 5418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.59670645302941e-06,
+      "loss": 0.8706,
+      "step": 5419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.5914362977246745e-06,
+      "loss": 0.8645,
+      "step": 5420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.58616685248969e-06,
+      "loss": 0.9612,
+      "step": 5421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.580898118877949e-06,
+      "loss": 0.8544,
+      "step": 5422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.5756300984427234e-06,
+      "loss": 0.3691,
+      "step": 5423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.5703627927370806e-06,
+      "loss": 0.9043,
+      "step": 5424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.5650962033138855e-06,
+      "loss": 0.8911,
+      "step": 5425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.559830331725776e-06,
+      "loss": 0.799,
+      "step": 5426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.5545651795251906e-06,
+      "loss": 0.8862,
+      "step": 5427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.549300748264347e-06,
+      "loss": 0.8469,
+      "step": 5428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.544037039495258e-06,
+      "loss": 0.8945,
+      "step": 5429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.538774054769717e-06,
+      "loss": 0.9077,
+      "step": 5430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.53351179563931e-06,
+      "loss": 0.9388,
+      "step": 5431
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.528250263655399e-06,
+      "loss": 0.9318,
+      "step": 5432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.52298946036915e-06,
+      "loss": 0.9773,
+      "step": 5433
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.517729387331496e-06,
+      "loss": 0.8482,
+      "step": 5434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.512470046093163e-06,
+      "loss": 0.8508,
+      "step": 5435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.507211438204658e-06,
+      "loss": 0.9042,
+      "step": 5436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.501953565216276e-06,
+      "loss": 0.9234,
+      "step": 5437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.49669642867809e-06,
+      "loss": 0.8528,
+      "step": 5438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.4914400301399626e-06,
+      "loss": 0.8815,
+      "step": 5439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.486184371151528e-06,
+      "loss": 0.8212,
+      "step": 5440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.4809294532622175e-06,
+      "loss": 0.9278,
+      "step": 5441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.475675278021235e-06,
+      "loss": 0.8927,
+      "step": 5442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.47042184697756e-06,
+      "loss": 0.878,
+      "step": 5443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.465169161679964e-06,
+      "loss": 0.3249,
+      "step": 5444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.4599172236769935e-06,
+      "loss": 0.8862,
+      "step": 5445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.454666034516973e-06,
+      "loss": 0.9021,
+      "step": 5446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.449415595748008e-06,
+      "loss": 0.8811,
+      "step": 5447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.4441659089179815e-06,
+      "loss": 0.986,
+      "step": 5448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.43891697557456e-06,
+      "loss": 0.901,
+      "step": 5449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.433668797265181e-06,
+      "loss": 0.9233,
+      "step": 5450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.428421375537063e-06,
+      "loss": 0.8373,
+      "step": 5451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.423174711937202e-06,
+      "loss": 0.8997,
+      "step": 5452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.417928808012366e-06,
+      "loss": 0.8857,
+      "step": 5453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.412683665309106e-06,
+      "loss": 0.3876,
+      "step": 5454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.407439285373741e-06,
+      "loss": 0.3786,
+      "step": 5455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.40219566975237e-06,
+      "loss": 0.8611,
+      "step": 5456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.396952819990868e-06,
+      "loss": 0.8933,
+      "step": 5457
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.391710737634884e-06,
+      "loss": 0.8459,
+      "step": 5458
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.386469424229832e-06,
+      "loss": 0.9005,
+      "step": 5459
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.381228881320912e-06,
+      "loss": 0.8206,
+      "step": 5460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.375989110453087e-06,
+      "loss": 0.8507,
+      "step": 5461
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.370750113171099e-06,
+      "loss": 0.8546,
+      "step": 5462
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.365511891019455e-06,
+      "loss": 0.8996,
+      "step": 5463
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.360274445542439e-06,
+      "loss": 0.9057,
+      "step": 5464
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.355037778284109e-06,
+      "loss": 0.8779,
+      "step": 5465
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.3498018907882855e-06,
+      "loss": 0.8625,
+      "step": 5466
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.344566784598565e-06,
+      "loss": 0.8653,
+      "step": 5467
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.339332461258307e-06,
+      "loss": 0.3838,
+      "step": 5468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.334098922310651e-06,
+      "loss": 0.8996,
+      "step": 5469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.328866169298493e-06,
+      "loss": 0.8785,
+      "step": 5470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.323634203764507e-06,
+      "loss": 0.8689,
+      "step": 5471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.318403027251127e-06,
+      "loss": 0.8709,
+      "step": 5472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.313172641300565e-06,
+      "loss": 0.8368,
+      "step": 5473
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.30794304745479e-06,
+      "loss": 0.8713,
+      "step": 5474
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.302714247255542e-06,
+      "loss": 0.3768,
+      "step": 5475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.297486242244322e-06,
+      "loss": 0.9232,
+      "step": 5476
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.292259033962408e-06,
+      "loss": 0.8631,
+      "step": 5477
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.287032623950827e-06,
+      "loss": 0.8955,
+      "step": 5478
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.281807013750386e-06,
+      "loss": 0.9064,
+      "step": 5479
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.276582204901643e-06,
+      "loss": 0.898,
+      "step": 5480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.271358198944934e-06,
+      "loss": 0.8752,
+      "step": 5481
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.266134997420345e-06,
+      "loss": 0.9178,
+      "step": 5482
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.260912601867733e-06,
+      "loss": 0.8688,
+      "step": 5483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.255691013826714e-06,
+      "loss": 0.9347,
+      "step": 5484
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.250470234836667e-06,
+      "loss": 0.8773,
+      "step": 5485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.24525026643673e-06,
+      "loss": 0.9112,
+      "step": 5486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.240031110165808e-06,
+      "loss": 0.919,
+      "step": 5487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.23481276756256e-06,
+      "loss": 0.8938,
+      "step": 5488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.229595240165406e-06,
+      "loss": 0.9383,
+      "step": 5489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.224378529512535e-06,
+      "loss": 0.9183,
+      "step": 5490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.219162637141881e-06,
+      "loss": 0.38,
+      "step": 5491
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.213947564591148e-06,
+      "loss": 0.895,
+      "step": 5492
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.208733313397791e-06,
+      "loss": 0.9153,
+      "step": 5493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.203519885099026e-06,
+      "loss": 0.8777,
+      "step": 5494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.198307281231827e-06,
+      "loss": 0.9345,
+      "step": 5495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.193095503332923e-06,
+      "loss": 0.8996,
+      "step": 5496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.187884552938799e-06,
+      "loss": 0.8954,
+      "step": 5497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.182674431585703e-06,
+      "loss": 0.8747,
+      "step": 5498
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.177465140809629e-06,
+      "loss": 0.9026,
+      "step": 5499
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.172256682146334e-06,
+      "loss": 0.8702,
+      "step": 5500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.167049057131324e-06,
+      "loss": 0.8609,
+      "step": 5501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.161842267299859e-06,
+      "loss": 0.91,
+      "step": 5502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.1566363141869576e-06,
+      "loss": 0.8708,
+      "step": 5503
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.1514311993273884e-06,
+      "loss": 0.8506,
+      "step": 5504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.146226924255671e-06,
+      "loss": 0.8204,
+      "step": 5505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.141023490506085e-06,
+      "loss": 0.9374,
+      "step": 5506
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.135820899612656e-06,
+      "loss": 0.898,
+      "step": 5507
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.130619153109162e-06,
+      "loss": 0.8585,
+      "step": 5508
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.125418252529128e-06,
+      "loss": 0.8635,
+      "step": 5509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.120218199405842e-06,
+      "loss": 0.3722,
+      "step": 5510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.115018995272326e-06,
+      "loss": 0.8823,
+      "step": 5511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.109820641661365e-06,
+      "loss": 0.892,
+      "step": 5512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.104623140105482e-06,
+      "loss": 0.7964,
+      "step": 5513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.09942649213696e-06,
+      "loss": 0.8775,
+      "step": 5514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.094230699287827e-06,
+      "loss": 0.8614,
+      "step": 5515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.089035763089851e-06,
+      "loss": 0.8348,
+      "step": 5516
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.083841685074559e-06,
+      "loss": 0.9047,
+      "step": 5517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.078648466773215e-06,
+      "loss": 0.8988,
+      "step": 5518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.073456109716836e-06,
+      "loss": 0.4233,
+      "step": 5519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.068264615436184e-06,
+      "loss": 0.8203,
+      "step": 5520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.063073985461761e-06,
+      "loss": 0.8185,
+      "step": 5521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.057884221323825e-06,
+      "loss": 0.8629,
+      "step": 5522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.052695324552372e-06,
+      "loss": 0.8795,
+      "step": 5523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.047507296677141e-06,
+      "loss": 0.8788,
+      "step": 5524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.042320139227619e-06,
+      "loss": 0.9054,
+      "step": 5525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.037133853733029e-06,
+      "loss": 0.8902,
+      "step": 5526
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.031948441722349e-06,
+      "loss": 0.8925,
+      "step": 5527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.0267639047242874e-06,
+      "loss": 0.8858,
+      "step": 5528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.021580244267299e-06,
+      "loss": 0.9061,
+      "step": 5529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.01639746187959e-06,
+      "loss": 0.8071,
+      "step": 5530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.0112155590890885e-06,
+      "loss": 0.9281,
+      "step": 5531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.006034537423481e-06,
+      "loss": 0.8552,
+      "step": 5532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.000854398410182e-06,
+      "loss": 0.8398,
+      "step": 5533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.99567514357635e-06,
+      "loss": 0.9137,
+      "step": 5534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.990496774448891e-06,
+      "loss": 0.8529,
+      "step": 5535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.985319292554433e-06,
+      "loss": 0.848,
+      "step": 5536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.9801426994193544e-06,
+      "loss": 0.9109,
+      "step": 5537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.974966996569774e-06,
+      "loss": 0.8957,
+      "step": 5538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.969792185531539e-06,
+      "loss": 0.8282,
+      "step": 5539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.9646182678302385e-06,
+      "loss": 0.8846,
+      "step": 5540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.959445244991197e-06,
+      "loss": 0.9494,
+      "step": 5541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.954273118539478e-06,
+      "loss": 0.9115,
+      "step": 5542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.949101889999874e-06,
+      "loss": 0.9528,
+      "step": 5543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.943931560896921e-06,
+      "loss": 0.8641,
+      "step": 5544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.938762132754879e-06,
+      "loss": 0.7888,
+      "step": 5545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.933593607097763e-06,
+      "loss": 0.8534,
+      "step": 5546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.928425985449296e-06,
+      "loss": 0.949,
+      "step": 5547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.923259269332956e-06,
+      "loss": 0.9083,
+      "step": 5548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.918093460271938e-06,
+      "loss": 0.881,
+      "step": 5549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.912928559789178e-06,
+      "loss": 0.8975,
+      "step": 5550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.907764569407345e-06,
+      "loss": 0.8693,
+      "step": 5551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.902601490648836e-06,
+      "loss": 0.8831,
+      "step": 5552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.897439325035778e-06,
+      "loss": 0.889,
+      "step": 5553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.892278074090033e-06,
+      "loss": 0.9394,
+      "step": 5554
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.887117739333195e-06,
+      "loss": 0.9175,
+      "step": 5555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.881958322286582e-06,
+      "loss": 0.8472,
+      "step": 5556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.876799824471246e-06,
+      "loss": 0.874,
+      "step": 5557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.871642247407962e-06,
+      "loss": 0.8612,
+      "step": 5558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.86648559261724e-06,
+      "loss": 0.8862,
+      "step": 5559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.861329861619317e-06,
+      "loss": 0.8684,
+      "step": 5560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.856175055934153e-06,
+      "loss": 0.8532,
+      "step": 5561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.851021177081441e-06,
+      "loss": 0.8552,
+      "step": 5562
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.845868226580598e-06,
+      "loss": 0.8596,
+      "step": 5563
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.840716205950768e-06,
+      "loss": 0.8747,
+      "step": 5564
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.835565116710823e-06,
+      "loss": 0.8892,
+      "step": 5565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.830414960379352e-06,
+      "loss": 0.8655,
+      "step": 5566
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.82526573847468e-06,
+      "loss": 0.8699,
+      "step": 5567
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.820117452514847e-06,
+      "loss": 0.9181,
+      "step": 5568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.8149701040176275e-06,
+      "loss": 0.9031,
+      "step": 5569
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.809823694500504e-06,
+      "loss": 0.8947,
+      "step": 5570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.804678225480702e-06,
+      "loss": 0.841,
+      "step": 5571
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.799533698475155e-06,
+      "loss": 0.8419,
+      "step": 5572
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.7943901150005255e-06,
+      "loss": 0.842,
+      "step": 5573
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.789247476573191e-06,
+      "loss": 0.8918,
+      "step": 5574
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.784105784709262e-06,
+      "loss": 0.8292,
+      "step": 5575
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.778965040924555e-06,
+      "loss": 0.8278,
+      "step": 5576
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.773825246734622e-06,
+      "loss": 0.8463,
+      "step": 5577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.768686403654721e-06,
+      "loss": 0.8371,
+      "step": 5578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.763548513199842e-06,
+      "loss": 0.3541,
+      "step": 5579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.758411576884691e-06,
+      "loss": 0.8576,
+      "step": 5580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.7532755962236855e-06,
+      "loss": 0.3684,
+      "step": 5581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.748140572730968e-06,
+      "loss": 0.8637,
+      "step": 5582
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.743006507920396e-06,
+      "loss": 0.8736,
+      "step": 5583
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.737873403305548e-06,
+      "loss": 0.8683,
+      "step": 5584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.732741260399713e-06,
+      "loss": 0.8836,
+      "step": 5585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.7276100807159e-06,
+      "loss": 0.9106,
+      "step": 5586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.72247986576684e-06,
+      "loss": 0.8502,
+      "step": 5587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.717350617064973e-06,
+      "loss": 0.8449,
+      "step": 5588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.712222336122452e-06,
+      "loss": 0.904,
+      "step": 5589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.707095024451149e-06,
+      "loss": 0.8754,
+      "step": 5590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.701968683562649e-06,
+      "loss": 0.9063,
+      "step": 5591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.696843314968251e-06,
+      "loss": 0.9085,
+      "step": 5592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.691718920178967e-06,
+      "loss": 0.8555,
+      "step": 5593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.68659550070552e-06,
+      "loss": 0.4048,
+      "step": 5594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.681473058058352e-06,
+      "loss": 0.9135,
+      "step": 5595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.676351593747611e-06,
+      "loss": 0.8987,
+      "step": 5596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.671231109283156e-06,
+      "loss": 0.9664,
+      "step": 5597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.666111606174564e-06,
+      "loss": 0.883,
+      "step": 5598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.660993085931113e-06,
+      "loss": 0.9381,
+      "step": 5599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.6558755500618e-06,
+      "loss": 0.8672,
+      "step": 5600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.650759000075324e-06,
+      "loss": 0.9144,
+      "step": 5601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.6456434374800986e-06,
+      "loss": 0.8572,
+      "step": 5602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.64052886378425e-06,
+      "loss": 0.8568,
+      "step": 5603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.635415280495603e-06,
+      "loss": 0.9471,
+      "step": 5604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.630302689121698e-06,
+      "loss": 0.9191,
+      "step": 5605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.625191091169778e-06,
+      "loss": 0.8212,
+      "step": 5606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.6200804881467975e-06,
+      "loss": 0.8919,
+      "step": 5607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.614970881559413e-06,
+      "loss": 0.8477,
+      "step": 5608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.609862272913994e-06,
+      "loss": 0.9232,
+      "step": 5609
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.604754663716604e-06,
+      "loss": 0.8577,
+      "step": 5610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.599648055473029e-06,
+      "loss": 0.3778,
+      "step": 5611
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.5945424496887465e-06,
+      "loss": 0.8933,
+      "step": 5612
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.5894378478689446e-06,
+      "loss": 0.8134,
+      "step": 5613
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.5843342515185075e-06,
+      "loss": 0.8729,
+      "step": 5614
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.579231662142036e-06,
+      "loss": 0.9439,
+      "step": 5615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.574130081243821e-06,
+      "loss": 0.8484,
+      "step": 5616
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.569029510327864e-06,
+      "loss": 0.8945,
+      "step": 5617
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.563929950897867e-06,
+      "loss": 0.8616,
+      "step": 5618
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.558831404457229e-06,
+      "loss": 0.8758,
+      "step": 5619
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.553733872509064e-06,
+      "loss": 0.871,
+      "step": 5620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.548637356556171e-06,
+      "loss": 0.9101,
+      "step": 5621
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.543541858101057e-06,
+      "loss": 0.9335,
+      "step": 5622
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.53844737864593e-06,
+      "loss": 0.8451,
+      "step": 5623
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.533353919692693e-06,
+      "loss": 0.9009,
+      "step": 5624
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.528261482742956e-06,
+      "loss": 0.9062,
+      "step": 5625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.523170069298017e-06,
+      "loss": 0.8976,
+      "step": 5626
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.518079680858877e-06,
+      "loss": 0.9505,
+      "step": 5627
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.512990318926243e-06,
+      "loss": 0.8679,
+      "step": 5628
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.507901985000506e-06,
+      "loss": 0.9139,
+      "step": 5629
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.502814680581765e-06,
+      "loss": 0.906,
+      "step": 5630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.497728407169805e-06,
+      "loss": 0.8236,
+      "step": 5631
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.492643166264116e-06,
+      "loss": 0.8794,
+      "step": 5632
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.487558959363879e-06,
+      "loss": 0.8814,
+      "step": 5633
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.482475787967972e-06,
+      "loss": 0.843,
+      "step": 5634
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.4773936535749615e-06,
+      "loss": 0.8814,
+      "step": 5635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.472312557683125e-06,
+      "loss": 0.9117,
+      "step": 5636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.467232501790414e-06,
+      "loss": 0.8266,
+      "step": 5637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.462153487394487e-06,
+      "loss": 0.3364,
+      "step": 5638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.457075515992686e-06,
+      "loss": 0.8903,
+      "step": 5639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.451998589082054e-06,
+      "loss": 0.9185,
+      "step": 5640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.446922708159319e-06,
+      "loss": 0.8701,
+      "step": 5641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.441847874720907e-06,
+      "loss": 0.8712,
+      "step": 5642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.436774090262925e-06,
+      "loss": 0.8352,
+      "step": 5643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.431701356281186e-06,
+      "loss": 0.9118,
+      "step": 5644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.426629674271183e-06,
+      "loss": 0.9322,
+      "step": 5645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.4215590457280986e-06,
+      "loss": 0.8492,
+      "step": 5646
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.41648947214681e-06,
+      "loss": 0.8297,
+      "step": 5647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.4114209550218755e-06,
+      "loss": 0.8626,
+      "step": 5648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.40635349584755e-06,
+      "loss": 0.9214,
+      "step": 5649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.401287096117776e-06,
+      "loss": 0.8538,
+      "step": 5650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.396221757326175e-06,
+      "loss": 0.8495,
+      "step": 5651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.391157480966066e-06,
+      "loss": 0.9119,
+      "step": 5652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.386094268530454e-06,
+      "loss": 0.8461,
+      "step": 5653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.381032121512018e-06,
+      "loss": 0.9249,
+      "step": 5654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.375971041403142e-06,
+      "loss": 0.8807,
+      "step": 5655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.370911029695877e-06,
+      "loss": 0.8934,
+      "step": 5656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.36585208788197e-06,
+      "loss": 0.8581,
+      "step": 5657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.36079421745285e-06,
+      "loss": 0.8184,
+      "step": 5658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.355737419899628e-06,
+      "loss": 0.8845,
+      "step": 5659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.350681696713106e-06,
+      "loss": 0.8886,
+      "step": 5660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.345627049383763e-06,
+      "loss": 0.8638,
+      "step": 5661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.340573479401757e-06,
+      "loss": 0.8571,
+      "step": 5662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.33552098825694e-06,
+      "loss": 0.9168,
+      "step": 5663
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.3304695774388315e-06,
+      "loss": 0.888,
+      "step": 5664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.325419248436648e-06,
+      "loss": 0.8817,
+      "step": 5665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.3203700027392725e-06,
+      "loss": 0.8578,
+      "step": 5666
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.3153218418352755e-06,
+      "loss": 0.8626,
+      "step": 5667
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.310274767212916e-06,
+      "loss": 0.8789,
+      "step": 5668
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.305228780360115e-06,
+      "loss": 0.8433,
+      "step": 5669
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.3001838827644905e-06,
+      "loss": 0.8574,
+      "step": 5670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.295140075913323e-06,
+      "loss": 0.8267,
+      "step": 5671
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.290097361293585e-06,
+      "loss": 0.8204,
+      "step": 5672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.285055740391916e-06,
+      "loss": 0.8596,
+      "step": 5673
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.2800152146946445e-06,
+      "loss": 0.8935,
+      "step": 5674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.274975785687764e-06,
+      "loss": 0.7979,
+      "step": 5675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.269937454856961e-06,
+      "loss": 0.9406,
+      "step": 5676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.264900223687577e-06,
+      "loss": 0.8551,
+      "step": 5677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.259864093664647e-06,
+      "loss": 0.8248,
+      "step": 5678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.254829066272872e-06,
+      "loss": 0.9155,
+      "step": 5679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.249795142996633e-06,
+      "loss": 0.8695,
+      "step": 5680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.24476232531998e-06,
+      "loss": 0.8848,
+      "step": 5681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.2397306147266425e-06,
+      "loss": 0.9263,
+      "step": 5682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.2347000127000205e-06,
+      "loss": 0.9325,
+      "step": 5683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.229670520723185e-06,
+      "loss": 0.9082,
+      "step": 5684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.22464214027889e-06,
+      "loss": 0.8185,
+      "step": 5685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.21961487284955e-06,
+      "loss": 0.3591,
+      "step": 5686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.214588719917256e-06,
+      "loss": 0.8922,
+      "step": 5687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.209563682963773e-06,
+      "loss": 0.8567,
+      "step": 5688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.204539763470531e-06,
+      "loss": 0.8697,
+      "step": 5689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.199516962918637e-06,
+      "loss": 0.8875,
+      "step": 5690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.1944952827888615e-06,
+      "loss": 0.8642,
+      "step": 5691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.189474724561648e-06,
+      "loss": 0.8089,
+      "step": 5692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.184455289717117e-06,
+      "loss": 0.9201,
+      "step": 5693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.1794369797350426e-06,
+      "loss": 0.878,
+      "step": 5694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.17441979609488e-06,
+      "loss": 0.9034,
+      "step": 5695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.169403740275743e-06,
+      "loss": 0.8409,
+      "step": 5696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.164388813756421e-06,
+      "loss": 0.8317,
+      "step": 5697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.159375018015365e-06,
+      "loss": 0.898,
+      "step": 5698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.1543623545306944e-06,
+      "loss": 0.8555,
+      "step": 5699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.149350824780189e-06,
+      "loss": 0.9101,
+      "step": 5700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.144340430241313e-06,
+      "loss": 0.8609,
+      "step": 5701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.139331172391174e-06,
+      "loss": 0.8265,
+      "step": 5702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.134323052706557e-06,
+      "loss": 0.8706,
+      "step": 5703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.129316072663906e-06,
+      "loss": 0.3208,
+      "step": 5704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.124310233739332e-06,
+      "loss": 0.873,
+      "step": 5705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.119305537408609e-06,
+      "loss": 0.8561,
+      "step": 5706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.114301985147173e-06,
+      "loss": 0.8416,
+      "step": 5707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.109299578430119e-06,
+      "loss": 0.8716,
+      "step": 5708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.104298318732218e-06,
+      "loss": 0.8688,
+      "step": 5709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.099298207527888e-06,
+      "loss": 0.9412,
+      "step": 5710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.094299246291215e-06,
+      "loss": 0.9396,
+      "step": 5711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.089301436495947e-06,
+      "loss": 0.8926,
+      "step": 5712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.084304779615489e-06,
+      "loss": 0.8932,
+      "step": 5713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.0793092771229045e-06,
+      "loss": 0.8568,
+      "step": 5714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.074314930490925e-06,
+      "loss": 0.8409,
+      "step": 5715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.0693217411919305e-06,
+      "loss": 0.8984,
+      "step": 5716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.0643297106979705e-06,
+      "loss": 0.9165,
+      "step": 5717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.059338840480749e-06,
+      "loss": 0.8358,
+      "step": 5718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.0543491320116185e-06,
+      "loss": 0.3746,
+      "step": 5719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.049360586761608e-06,
+      "loss": 0.8449,
+      "step": 5720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.0443732062013835e-06,
+      "loss": 0.8981,
+      "step": 5721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.039386991801282e-06,
+      "loss": 0.8083,
+      "step": 5722
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.034401945031289e-06,
+      "loss": 0.3509,
+      "step": 5723
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.0294180673610485e-06,
+      "loss": 0.8659,
+      "step": 5724
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.024435360259862e-06,
+      "loss": 0.8617,
+      "step": 5725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.019453825196684e-06,
+      "loss": 0.9579,
+      "step": 5726
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.01447346364012e-06,
+      "loss": 0.8819,
+      "step": 5727
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.009494277058436e-06,
+      "loss": 0.9217,
+      "step": 5728
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.004516266919546e-06,
+      "loss": 0.8733,
+      "step": 5729
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.999539434691019e-06,
+      "loss": 0.8777,
+      "step": 5730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.99456378184008e-06,
+      "loss": 0.9133,
+      "step": 5731
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.989589309833596e-06,
+      "loss": 0.8446,
+      "step": 5732
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.984616020138103e-06,
+      "loss": 0.8915,
+      "step": 5733
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.979643914219775e-06,
+      "loss": 0.9103,
+      "step": 5734
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.974672993544439e-06,
+      "loss": 0.8681,
+      "step": 5735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.969703259577577e-06,
+      "loss": 0.8791,
+      "step": 5736
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.964734713784315e-06,
+      "loss": 0.8668,
+      "step": 5737
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.959767357629438e-06,
+      "loss": 0.8951,
+      "step": 5738
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.954801192577368e-06,
+      "loss": 0.9333,
+      "step": 5739
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.949836220092184e-06,
+      "loss": 0.8899,
+      "step": 5740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.944872441637617e-06,
+      "loss": 0.3802,
+      "step": 5741
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.939909858677036e-06,
+      "loss": 0.8581,
+      "step": 5742
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.9349484726734625e-06,
+      "loss": 0.8793,
+      "step": 5743
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.929988285089566e-06,
+      "loss": 0.8484,
+      "step": 5744
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.925029297387663e-06,
+      "loss": 0.919,
+      "step": 5745
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.9200715110297105e-06,
+      "loss": 0.8323,
+      "step": 5746
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.915114927477321e-06,
+      "loss": 0.8099,
+      "step": 5747
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.910159548191743e-06,
+      "loss": 0.9028,
+      "step": 5748
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.905205374633874e-06,
+      "loss": 0.8664,
+      "step": 5749
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.900252408264258e-06,
+      "loss": 0.8611,
+      "step": 5750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.895300650543085e-06,
+      "loss": 0.8454,
+      "step": 5751
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.890350102930178e-06,
+      "loss": 0.8532,
+      "step": 5752
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.885400766885016e-06,
+      "loss": 0.8994,
+      "step": 5753
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.88045264386671e-06,
+      "loss": 0.9772,
+      "step": 5754
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8755057353340216e-06,
+      "loss": 0.9019,
+      "step": 5755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.87056004274535e-06,
+      "loss": 0.866,
+      "step": 5756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.865615567558733e-06,
+      "loss": 0.3543,
+      "step": 5757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.860672311231863e-06,
+      "loss": 0.8466,
+      "step": 5758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.855730275222057e-06,
+      "loss": 0.9183,
+      "step": 5759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.850789460986281e-06,
+      "loss": 0.9272,
+      "step": 5760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.857,
+      "step": 5761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.84091150366287e-06,
+      "loss": 0.8719,
+      "step": 5762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.83597436348736e-06,
+      "loss": 0.8862,
+      "step": 5763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.83103845091013e-06,
+      "loss": 0.8729,
+      "step": 5764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.826103767386333e-06,
+      "loss": 0.8607,
+      "step": 5765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.82117031437077e-06,
+      "loss": 0.8802,
+      "step": 5766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.816238093317878e-06,
+      "loss": 0.8898,
+      "step": 5767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.811307105681723e-06,
+      "loss": 0.8294,
+      "step": 5768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.806377352916012e-06,
+      "loss": 0.8447,
+      "step": 5769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.801448836474084e-06,
+      "loss": 0.8393,
+      "step": 5770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.796521557808923e-06,
+      "loss": 0.9062,
+      "step": 5771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.791595518373141e-06,
+      "loss": 0.8578,
+      "step": 5772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.786670719618982e-06,
+      "loss": 0.8511,
+      "step": 5773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.781747162998335e-06,
+      "loss": 0.9021,
+      "step": 5774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.776824849962706e-06,
+      "loss": 0.8987,
+      "step": 5775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.771903781963255e-06,
+      "loss": 0.8909,
+      "step": 5776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.766983960450761e-06,
+      "loss": 0.8616,
+      "step": 5777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.762065386875633e-06,
+      "loss": 0.8854,
+      "step": 5778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.757148062687925e-06,
+      "loss": 0.8731,
+      "step": 5779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7522319893373136e-06,
+      "loss": 0.8896,
+      "step": 5780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.747317168273103e-06,
+      "loss": 0.9134,
+      "step": 5781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.742403600944244e-06,
+      "loss": 0.8365,
+      "step": 5782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.737491288799299e-06,
+      "loss": 0.9008,
+      "step": 5783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.732580233286476e-06,
+      "loss": 0.8787,
+      "step": 5784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.727670435853601e-06,
+      "loss": 0.8701,
+      "step": 5785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.722761897948133e-06,
+      "loss": 0.9042,
+      "step": 5786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7178546210171596e-06,
+      "loss": 0.892,
+      "step": 5787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.712948606507403e-06,
+      "loss": 0.8437,
+      "step": 5788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7080438558652e-06,
+      "loss": 0.8394,
+      "step": 5789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.703140370536529e-06,
+      "loss": 0.8392,
+      "step": 5790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.698238151966983e-06,
+      "loss": 0.8511,
+      "step": 5791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.693337201601794e-06,
+      "loss": 0.8301,
+      "step": 5792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.68843752088581e-06,
+      "loss": 0.9452,
+      "step": 5793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.683539111263508e-06,
+      "loss": 0.9207,
+      "step": 5794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.678641974178984e-06,
+      "loss": 0.886,
+      "step": 5795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.673746111075978e-06,
+      "loss": 0.7983,
+      "step": 5796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.668851523397829e-06,
+      "loss": 0.9028,
+      "step": 5797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.663958212587524e-06,
+      "loss": 0.792,
+      "step": 5798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6590661800876555e-06,
+      "loss": 0.3674,
+      "step": 5799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.654175427340444e-06,
+      "loss": 0.9345,
+      "step": 5800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.64928595578774e-06,
+      "loss": 0.7977,
+      "step": 5801
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.64439776687101e-06,
+      "loss": 0.9264,
+      "step": 5802
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.639510862031337e-06,
+      "loss": 0.8669,
+      "step": 5803
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.634625242709439e-06,
+      "loss": 0.8988,
+      "step": 5804
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.629740910345642e-06,
+      "loss": 0.8643,
+      "step": 5805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.624857866379905e-06,
+      "loss": 0.8671,
+      "step": 5806
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.619976112251796e-06,
+      "loss": 0.8378,
+      "step": 5807
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.615095649400504e-06,
+      "loss": 0.8982,
+      "step": 5808
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.610216479264847e-06,
+      "loss": 0.8502,
+      "step": 5809
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.605338603283253e-06,
+      "loss": 0.8214,
+      "step": 5810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.60046202289377e-06,
+      "loss": 0.3585,
+      "step": 5811
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.595586739534062e-06,
+      "loss": 0.9325,
+      "step": 5812
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.590712754641418e-06,
+      "loss": 0.9018,
+      "step": 5813
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.585840069652737e-06,
+      "loss": 0.9667,
+      "step": 5814
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.580968686004542e-06,
+      "loss": 0.8401,
+      "step": 5815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.576098605132959e-06,
+      "loss": 0.8779,
+      "step": 5816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.57122982847375e-06,
+      "loss": 0.9214,
+      "step": 5817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.566362357462274e-06,
+      "loss": 0.8737,
+      "step": 5818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.561496193533516e-06,
+      "loss": 0.9016,
+      "step": 5819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.556631338122063e-06,
+      "loss": 0.9235,
+      "step": 5820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.5517677926621375e-06,
+      "loss": 0.8959,
+      "step": 5821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.546905558587554e-06,
+      "loss": 0.8982,
+      "step": 5822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.542044637331758e-06,
+      "loss": 0.938,
+      "step": 5823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.537185030327795e-06,
+      "loss": 0.8698,
+      "step": 5824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.5323267390083256e-06,
+      "loss": 0.8499,
+      "step": 5825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.527469764805632e-06,
+      "loss": 0.9063,
+      "step": 5826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.522614109151596e-06,
+      "loss": 0.9022,
+      "step": 5827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.517759773477714e-06,
+      "loss": 0.8641,
+      "step": 5828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.512906759215102e-06,
+      "loss": 0.8819,
+      "step": 5829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.508055067794469e-06,
+      "loss": 0.8996,
+      "step": 5830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.5032047006461555e-06,
+      "loss": 0.8468,
+      "step": 5831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.4983556592000945e-06,
+      "loss": 0.9117,
+      "step": 5832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.4935079448858305e-06,
+      "loss": 0.394,
+      "step": 5833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.488661559132529e-06,
+      "loss": 0.8593,
+      "step": 5834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.483816503368951e-06,
+      "loss": 0.8538,
+      "step": 5835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.47897277902347e-06,
+      "loss": 0.8619,
+      "step": 5836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.47413038752406e-06,
+      "loss": 0.8744,
+      "step": 5837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.469289330298315e-06,
+      "loss": 0.8628,
+      "step": 5838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.464449608773434e-06,
+      "loss": 0.8283,
+      "step": 5839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.459611224376211e-06,
+      "loss": 0.8895,
+      "step": 5840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.454774178533051e-06,
+      "loss": 0.905,
+      "step": 5841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.449938472669971e-06,
+      "loss": 0.908,
+      "step": 5842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.445104108212586e-06,
+      "loss": 0.8826,
+      "step": 5843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.440271086586115e-06,
+      "loss": 0.86,
+      "step": 5844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.435439409215379e-06,
+      "loss": 0.9313,
+      "step": 5845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.430609077524816e-06,
+      "loss": 0.8745,
+      "step": 5846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.425780092938455e-06,
+      "loss": 0.8806,
+      "step": 5847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.420952456879932e-06,
+      "loss": 0.7787,
+      "step": 5848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.416126170772484e-06,
+      "loss": 0.8613,
+      "step": 5849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.411301236038945e-06,
+      "loss": 0.8876,
+      "step": 5850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.406477654101765e-06,
+      "loss": 0.884,
+      "step": 5851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.401655426382984e-06,
+      "loss": 0.8416,
+      "step": 5852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.396834554304237e-06,
+      "loss": 0.86,
+      "step": 5853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.392015039286775e-06,
+      "loss": 0.8382,
+      "step": 5854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.387196882751445e-06,
+      "loss": 0.8296,
+      "step": 5855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.382380086118685e-06,
+      "loss": 0.867,
+      "step": 5856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.3775646508085364e-06,
+      "loss": 0.8876,
+      "step": 5857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.3727505782406375e-06,
+      "loss": 0.8506,
+      "step": 5858
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.367937869834234e-06,
+      "loss": 0.8279,
+      "step": 5859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.36312652700816e-06,
+      "loss": 0.832,
+      "step": 5860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.358316551180849e-06,
+      "loss": 0.8877,
+      "step": 5861
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.3535079437703275e-06,
+      "loss": 0.9017,
+      "step": 5862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.348700706194227e-06,
+      "loss": 0.8615,
+      "step": 5863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.343894839869777e-06,
+      "loss": 0.8507,
+      "step": 5864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.339090346213791e-06,
+      "loss": 0.9372,
+      "step": 5865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.334287226642683e-06,
+      "loss": 0.8798,
+      "step": 5866
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.329485482572467e-06,
+      "loss": 0.8324,
+      "step": 5867
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.324685115418746e-06,
+      "loss": 0.8483,
+      "step": 5868
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.319886126596717e-06,
+      "loss": 0.8633,
+      "step": 5869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.31508851752117e-06,
+      "loss": 0.9368,
+      "step": 5870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.310292289606491e-06,
+      "loss": 0.8587,
+      "step": 5871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.305497444266666e-06,
+      "loss": 0.8539,
+      "step": 5872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.30070398291526e-06,
+      "loss": 0.8668,
+      "step": 5873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.295911906965434e-06,
+      "loss": 0.8342,
+      "step": 5874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.29112121782994e-06,
+      "loss": 0.8557,
+      "step": 5875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2863319169211295e-06,
+      "loss": 0.9036,
+      "step": 5876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.281544005650936e-06,
+      "loss": 0.8626,
+      "step": 5877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.276757485430881e-06,
+      "loss": 0.9092,
+      "step": 5878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.271972357672086e-06,
+      "loss": 0.8716,
+      "step": 5879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.267188623785258e-06,
+      "loss": 0.9078,
+      "step": 5880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.26240628518069e-06,
+      "loss": 0.8853,
+      "step": 5881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.257625343268264e-06,
+      "loss": 0.9027,
+      "step": 5882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.25284579945745e-06,
+      "loss": 0.8513,
+      "step": 5883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.248067655157314e-06,
+      "loss": 0.8635,
+      "step": 5884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.243290911776497e-06,
+      "loss": 0.8535,
+      "step": 5885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.238515570723235e-06,
+      "loss": 0.9532,
+      "step": 5886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.233741633405345e-06,
+      "loss": 0.8643,
+      "step": 5887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.228969101230237e-06,
+      "loss": 0.8687,
+      "step": 5888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.224197975604906e-06,
+      "loss": 0.8835,
+      "step": 5889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2194282579359255e-06,
+      "loss": 0.8968,
+      "step": 5890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.214659949629458e-06,
+      "loss": 0.8889,
+      "step": 5891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.209893052091255e-06,
+      "loss": 0.8537,
+      "step": 5892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.205127566726644e-06,
+      "loss": 0.8737,
+      "step": 5893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.200363494940541e-06,
+      "loss": 0.895,
+      "step": 5894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.19560083813744e-06,
+      "loss": 0.8321,
+      "step": 5895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.190839597721426e-06,
+      "loss": 0.9092,
+      "step": 5896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1860797750961656e-06,
+      "loss": 0.3371,
+      "step": 5897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.181321371664901e-06,
+      "loss": 0.8638,
+      "step": 5898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.17656438883046e-06,
+      "loss": 0.9093,
+      "step": 5899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.171808827995245e-06,
+      "loss": 0.8604,
+      "step": 5900
+    }
+  ],
+  "max_steps": 5965,
+  "num_train_epochs": 1,
+  "total_flos": 2.225173262932692e+19,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/training_args.bin b/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..14ac429c07428bd663e096f7d7bc42b9e6e45897
--- /dev/null
+++ b/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd91608fa1c67901a21f7495e52f38c1f455da7e3bd64c792a078e48b68c9b47
+size 5499