diff --git a/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/README.md b/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..8ebb3aeb247ddc9079538b09a0e21b51130cf665
--- /dev/null
+++ b/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/README.md
@@ -0,0 +1,202 @@
+---
+base_model: microsoft/llava-med-v1.5-mistral-7b
+library_name: peft
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/adapter_config.json b/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..68b342a967bb5a452333cd88a4754e0f11c1b407
--- /dev/null
+++ b/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "microsoft/llava-med-v1.5-mistral-7b",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 256,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 128,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "o_proj",
+    "gate_proj",
+    "q_proj",
+    "up_proj",
+    "v_proj",
+    "down_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/adapter_model.safetensors b/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..5b19076ba613ebe4a6c7f257ddc5509ccfd756d6
--- /dev/null
+++ b/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1a019a180353b9484b619aa874a5c977324e9df8dbfb71f18050ba32975802b
+size 671149616
diff --git a/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/config.json b/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..d04297b02a58aee5ee6810b8a445353dfc1ecaf5
--- /dev/null
+++ b/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/config.json
@@ -0,0 +1,66 @@
+{
+  "_name_or_path": "microsoft/llava-med-v1.5-mistral-7b",
+  "architectures": [
+    "LlavaMistralForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "feature_outs": "encoder+decoder",
+  "freeze_mm_mlp_adapter": false,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "image_aspect_ratio": "pad",
+  "img_size": 640,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 32768,
+  "mm_hidden_size": 1024,
+  "mm_patch_merge_type": "flat",
+  "mm_projector_lr": 2e-05,
+  "mm_projector_type": "mlp2x_gelu",
+  "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": "openai/clip-vit-large-patch14-336",
+  "model_type": "llava_llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "proj_vis_to_txt_tokens": false,
+  "prompt_segtok_w_instruct": false,
+  "quantization_config": {
+    "bnb_4bit_compute_dtype": "float16",
+    "bnb_4bit_quant_type": "nf4",
+    "bnb_4bit_use_double_quant": true,
+    "llm_int8_enable_fp32_cpu_offload": false,
+    "llm_int8_has_fp16_weight": false,
+    "llm_int8_skip_modules": [
+      "mm_projector"
+    ],
+    "llm_int8_threshold": 6.0,
+    "load_in_4bit": true,
+    "load_in_8bit": false,
+    "quant_method": "bitsandbytes"
+  },
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "segtok_posembed": "sincos",
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "tokenizer_model_max_length": 2048,
+  "tokenizer_padding_side": "right",
+  "torch_dtype": "float32",
+  "transformers_version": "4.37.2",
+  "tune_mm_mlp_adapter": false,
+  "tune_vision_tokenizer": "none",
+  "use_cache": true,
+  "use_mm_proj": true,
+  "vision_backbone": "convnextlarge",
+  "vision_tokenizer_lr": null,
+  "vocab_size": 32000
+}
diff --git a/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/non_lora_trainables.bin b/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/non_lora_trainables.bin
new file mode 100644
index 0000000000000000000000000000000000000000..ecbd58777432188fe78d873d2ee51f2dd49ecab6
--- /dev/null
+++ b/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/non_lora_trainables.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9fdcd979b2c0c1a3b8734ee8e11764c6a2e98974b5c3a08077412c8a539f916
+size 41961648
diff --git a/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/trainer_state.json b/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..0c508df708237ecde6be9b57699647992530350b
--- /dev/null
+++ b/train_5k-quantized-4-epoch-5-lr-2e-4-per_device_train_batch_size-4-gradient_accumulation_steps-4/trainer_state.json
@@ -0,0 +1,9390 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.992,
+  "eval_steps": 500,
+  "global_step": 1560,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.255319148936171e-06,
+      "loss": 2.3153,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.510638297872341e-06,
+      "loss": 2.528,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2765957446808511e-05,
+      "loss": 2.3197,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7021276595744682e-05,
+      "loss": 2.1681,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1276595744680852e-05,
+      "loss": 2.0251,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5531914893617022e-05,
+      "loss": 1.7686,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9787234042553192e-05,
+      "loss": 1.9893,
+      "step": 7
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4042553191489365e-05,
+      "loss": 1.9126,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.829787234042553e-05,
+      "loss": 1.8677,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2553191489361704e-05,
+      "loss": 1.8034,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.680851063829788e-05,
+      "loss": 1.5356,
+      "step": 11
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.1063829787234044e-05,
+      "loss": 1.6526,
+      "step": 12
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.531914893617022e-05,
+      "loss": 1.6789,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9574468085106384e-05,
+      "loss": 1.7954,
+      "step": 14
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.382978723404256e-05,
+      "loss": 1.7968,
+      "step": 15
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.808510638297873e-05,
+      "loss": 1.6929,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.23404255319149e-05,
+      "loss": 1.5112,
+      "step": 17
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.659574468085106e-05,
+      "loss": 1.4885,
+      "step": 18
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.085106382978723e-05,
+      "loss": 1.6761,
+      "step": 19
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.510638297872341e-05,
+      "loss": 1.5262,
+      "step": 20
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.936170212765958e-05,
+      "loss": 1.5937,
+      "step": 21
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.361702127659576e-05,
+      "loss": 1.679,
+      "step": 22
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.787234042553192e-05,
+      "loss": 1.5021,
+      "step": 23
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010212765957446809,
+      "loss": 1.3683,
+      "step": 24
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010638297872340425,
+      "loss": 1.4835,
+      "step": 25
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011063829787234043,
+      "loss": 1.6558,
+      "step": 26
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011489361702127661,
+      "loss": 1.5377,
+      "step": 27
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011914893617021277,
+      "loss": 1.5644,
+      "step": 28
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012340425531914893,
+      "loss": 1.3314,
+      "step": 29
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00012765957446808513,
+      "loss": 1.5427,
+      "step": 30
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013191489361702127,
+      "loss": 1.5351,
+      "step": 31
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013617021276595746,
+      "loss": 1.4427,
+      "step": 32
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014042553191489363,
+      "loss": 1.6373,
+      "step": 33
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001446808510638298,
+      "loss": 1.5812,
+      "step": 34
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014893617021276596,
+      "loss": 1.5542,
+      "step": 35
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015319148936170213,
+      "loss": 1.5404,
+      "step": 36
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015744680851063832,
+      "loss": 1.4698,
+      "step": 37
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016170212765957446,
+      "loss": 1.4238,
+      "step": 38
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016595744680851065,
+      "loss": 1.4764,
+      "step": 39
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017021276595744682,
+      "loss": 1.4013,
+      "step": 40
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017446808510638298,
+      "loss": 1.3107,
+      "step": 41
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017872340425531915,
+      "loss": 1.6735,
+      "step": 42
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018297872340425532,
+      "loss": 1.7,
+      "step": 43
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001872340425531915,
+      "loss": 1.4501,
+      "step": 44
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019148936170212768,
+      "loss": 1.4683,
+      "step": 45
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019574468085106384,
+      "loss": 1.5361,
+      "step": 46
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 1.3335,
+      "step": 47
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999978442830445,
+      "loss": 1.4127,
+      "step": 48
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999991377141472,
+      "loss": 1.5012,
+      "step": 49
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999980598603165,
+      "loss": 1.5645,
+      "step": 50
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999655087145948,
+      "loss": 1.4885,
+      "step": 51
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999461075408204,
+      "loss": 1.517,
+      "step": 52
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001999922395165488,
+      "loss": 1.5654,
+      "step": 53
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019998943716908332,
+      "loss": 1.4032,
+      "step": 54
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998620372376765,
+      "loss": 1.4559,
+      "step": 55
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998253919454255,
+      "loss": 1.376,
+      "step": 56
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001999784435972075,
+      "loss": 1.519,
+      "step": 57
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001999739169494203,
+      "loss": 1.4283,
+      "step": 58
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996895927069732,
+      "loss": 1.6864,
+      "step": 59
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996357058241332,
+      "loss": 1.4142,
+      "step": 60
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019996357058241332,
+      "loss": 1.4555,
+      "step": 61
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019996357058241332,
+      "loss": 1.3329,
+      "step": 62
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019996357058241332,
+      "loss": 1.572,
+      "step": 63
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019995775090780117,
+      "loss": 1.3172,
+      "step": 64
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999515002719521,
+      "loss": 1.464,
+      "step": 65
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019994481870181527,
+      "loss": 1.7494,
+      "step": 66
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019993770622619782,
+      "loss": 1.4626,
+      "step": 67
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001999301628757648,
+      "loss": 1.2504,
+      "step": 68
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019992218868303873,
+      "loss": 1.4996,
+      "step": 69
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019991378368239992,
+      "loss": 1.4856,
+      "step": 70
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019990494791008597,
+      "loss": 1.5502,
+      "step": 71
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019989568140419168,
+      "loss": 1.4468,
+      "step": 72
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019988598420466903,
+      "loss": 1.4567,
+      "step": 73
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019987585635332678,
+      "loss": 1.641,
+      "step": 74
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019986529789383056,
+      "loss": 1.5396,
+      "step": 75
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019985430887170245,
+      "loss": 1.5463,
+      "step": 76
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019984288933432088,
+      "loss": 1.4809,
+      "step": 77
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019983103933092047,
+      "loss": 1.5589,
+      "step": 78
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019981875891259167,
+      "loss": 1.309,
+      "step": 79
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019980604813228073,
+      "loss": 1.5261,
+      "step": 80
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019979290704478934,
+      "loss": 1.5019,
+      "step": 81
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019977933570677442,
+      "loss": 1.4316,
+      "step": 82
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019976533417674785,
+      "loss": 1.4378,
+      "step": 83
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019975090251507638,
+      "loss": 1.2686,
+      "step": 84
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019973604078398115,
+      "loss": 1.4022,
+      "step": 85
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001997207490475375,
+      "loss": 1.662,
+      "step": 86
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019970502737167472,
+      "loss": 1.5442,
+      "step": 87
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019968887582417588,
+      "loss": 1.4372,
+      "step": 88
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019967229447467722,
+      "loss": 1.3672,
+      "step": 89
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019965528339466814,
+      "loss": 1.3547,
+      "step": 90
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019963784265749078,
+      "loss": 1.5454,
+      "step": 91
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001996199723383398,
+      "loss": 1.5574,
+      "step": 92
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001996016725142618,
+      "loss": 1.4822,
+      "step": 93
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019958294326415533,
+      "loss": 1.5074,
+      "step": 94
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001995637846687703,
+      "loss": 1.4148,
+      "step": 95
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019954419681070772,
+      "loss": 1.3977,
+      "step": 96
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019952417977441936,
+      "loss": 1.5358,
+      "step": 97
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019950373364620728,
+      "loss": 1.3778,
+      "step": 98
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019948285851422373,
+      "loss": 1.4486,
+      "step": 99
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019946155446847036,
+      "loss": 1.5386,
+      "step": 100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001994398216007982,
+      "loss": 1.4186,
+      "step": 101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019941766000490708,
+      "loss": 1.4685,
+      "step": 102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019939506977634524,
+      "loss": 1.4043,
+      "step": 103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019937205101250896,
+      "loss": 1.4492,
+      "step": 104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001993486038126421,
+      "loss": 1.4213,
+      "step": 105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019932472827783576,
+      "loss": 1.248,
+      "step": 106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019930042451102767,
+      "loss": 1.3245,
+      "step": 107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019927569261700195,
+      "loss": 1.3925,
+      "step": 108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019925053270238854,
+      "loss": 1.4988,
+      "step": 109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019922494487566273,
+      "loss": 1.4231,
+      "step": 110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019919892924714474,
+      "loss": 1.4763,
+      "step": 111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019917248592899925,
+      "loss": 1.5743,
+      "step": 112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019914561503523486,
+      "loss": 1.2998,
+      "step": 113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019911831668170367,
+      "loss": 1.5269,
+      "step": 114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019909059098610068,
+      "loss": 1.4129,
+      "step": 115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019906243806796345,
+      "loss": 1.6105,
+      "step": 116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019903385804867144,
+      "loss": 1.6002,
+      "step": 117
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019900485105144543,
+      "loss": 1.4599,
+      "step": 118
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019897541720134725,
+      "loss": 1.5595,
+      "step": 119
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019894555662527898,
+      "loss": 1.2877,
+      "step": 120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001989152694519825,
+      "loss": 1.4855,
+      "step": 121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000198884555812039,
+      "loss": 1.2825,
+      "step": 122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019885341583786824,
+      "loss": 1.4661,
+      "step": 123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001988218496637282,
+      "loss": 1.2819,
+      "step": 124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019878985742571436,
+      "loss": 1.5566,
+      "step": 125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019878985742571436,
+      "loss": 1.8523,
+      "step": 126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019875743926175914,
+      "loss": 2.054,
+      "step": 127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019872459531163127,
+      "loss": 1.5899,
+      "step": 128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019869132571693533,
+      "loss": 1.4858,
+      "step": 129
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019865763062111097,
+      "loss": 1.4444,
+      "step": 130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019862351016943235,
+      "loss": 1.3197,
+      "step": 131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001985889645090075,
+      "loss": 1.3934,
+      "step": 132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001985889645090075,
+      "loss": 2.9456,
+      "step": 133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001985889645090075,
+      "loss": 3.2847,
+      "step": 134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001985889645090075,
+      "loss": 2.6941,
+      "step": 135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019855399378877783,
+      "loss": 2.2898,
+      "step": 136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019851859815951726,
+      "loss": 2.3054,
+      "step": 137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019848277777383166,
+      "loss": 1.6448,
+      "step": 138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019844653278615833,
+      "loss": 1.6761,
+      "step": 139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019840986335276512,
+      "loss": 2.4803,
+      "step": 140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019837276963174981,
+      "loss": 3.0977,
+      "step": 141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001983352517830396,
+      "loss": 4.6433,
+      "step": 142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001982973099683902,
+      "loss": 3.7847,
+      "step": 143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001982589443513852,
+      "loss": 4.1996,
+      "step": 144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001982589443513852,
+      "loss": 2.4142,
+      "step": 145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019822015509743545,
+      "loss": 2.5,
+      "step": 146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019818094237377825,
+      "loss": 4.9906,
+      "step": 147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019814130634947664,
+      "loss": 2.5206,
+      "step": 148
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019810124719541876,
+      "loss": 3.4555,
+      "step": 149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000198060765084317,
+      "loss": 6.6331,
+      "step": 150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019801986019070728,
+      "loss": 7.0909,
+      "step": 151
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001979785326909484,
+      "loss": 7.5777,
+      "step": 152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019793678276322109,
+      "loss": 7.0538,
+      "step": 153
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001978946105875274,
+      "loss": 7.0473,
+      "step": 154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001978520163456899,
+      "loss": 6.5257,
+      "step": 155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019780900022135087,
+      "loss": 6.5279,
+      "step": 156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019776556239997146,
+      "loss": 6.5669,
+      "step": 157
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019772170306883095,
+      "loss": 6.4102,
+      "step": 158
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019767742241702597,
+      "loss": 6.517,
+      "step": 159
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019763272063546964,
+      "loss": 6.4359,
+      "step": 160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001975875979168907,
+      "loss": 6.3224,
+      "step": 161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019754205445583281,
+      "loss": 6.3795,
+      "step": 162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019749609044865358,
+      "loss": 6.2626,
+      "step": 163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019744970609352378,
+      "loss": 6.1169,
+      "step": 164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019740290159042645,
+      "loss": 6.2857,
+      "step": 165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019735567714115616,
+      "loss": 6.2981,
+      "step": 166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019730803294931804,
+      "loss": 6.4165,
+      "step": 167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001972599692203268,
+      "loss": 6.3353,
+      "step": 168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019721148616140606,
+      "loss": 6.1368,
+      "step": 169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019716258398158734,
+      "loss": 6.1055,
+      "step": 170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019711326289170912,
+      "loss": 6.1601,
+      "step": 171
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019706352310441605,
+      "loss": 6.1675,
+      "step": 172
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019701336483415793,
+      "loss": 6.1375,
+      "step": 173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019696278829718883,
+      "loss": 6.2452,
+      "step": 174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019691179371156614,
+      "loss": 6.1768,
+      "step": 175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019686038129714963,
+      "loss": 6.0757,
+      "step": 176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019680855127560056,
+      "loss": 6.0918,
+      "step": 177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019675630387038062,
+      "loss": 5.9786,
+      "step": 178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019670363930675104,
+      "loss": 5.9434,
+      "step": 179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001966505578117716,
+      "loss": 6.1925,
+      "step": 180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019659705961429973,
+      "loss": 6.1387,
+      "step": 181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019654314494498926,
+      "loss": 6.0524,
+      "step": 182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019648881403628983,
+      "loss": 6.0224,
+      "step": 183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019643406712244548,
+      "loss": 6.1134,
+      "step": 184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019637890443949396,
+      "loss": 6.0684,
+      "step": 185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001963233262252655,
+      "loss": 6.1824,
+      "step": 186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001962673327193819,
+      "loss": 6.2283,
+      "step": 187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019621092416325552,
+      "loss": 6.0328,
+      "step": 188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019615410080008804,
+      "loss": 6.1134,
+      "step": 189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019609686287486966,
+      "loss": 6.0952,
+      "step": 190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019603921063437793,
+      "loss": 6.1574,
+      "step": 191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019598114432717667,
+      "loss": 6.1678,
+      "step": 192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019592266420361495,
+      "loss": 6.0322,
+      "step": 193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001958637705158259,
+      "loss": 5.9389,
+      "step": 194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019580446351772577,
+      "loss": 6.1912,
+      "step": 195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001957447434650128,
+      "loss": 6.0267,
+      "step": 196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019568461061516605,
+      "loss": 6.1306,
+      "step": 197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019562406522744433,
+      "loss": 5.9708,
+      "step": 198
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019556310756288503,
+      "loss": 6.0867,
+      "step": 199
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019550173788430314,
+      "loss": 6.1984,
+      "step": 200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019543995645628993,
+      "loss": 6.1766,
+      "step": 201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000195377763545212,
+      "loss": 6.0264,
+      "step": 202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019531515941920993,
+      "loss": 5.872,
+      "step": 203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001952521443481973,
+      "loss": 6.1171,
+      "step": 204
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019518871860385942,
+      "loss": 6.1943,
+      "step": 205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019512488245965214,
+      "loss": 6.0122,
+      "step": 206
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019506063619080085,
+      "loss": 6.0736,
+      "step": 207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001949959800742991,
+      "loss": 6.2523,
+      "step": 208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001949309143889074,
+      "loss": 5.9493,
+      "step": 209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019486543941515217,
+      "loss": 6.2468,
+      "step": 210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001947995554353245,
+      "loss": 6.0421,
+      "step": 211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019473326273347868,
+      "loss": 6.3491,
+      "step": 212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019466656159543145,
+      "loss": 5.8813,
+      "step": 213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001945994523087603,
+      "loss": 5.9992,
+      "step": 214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019453193516280243,
+      "loss": 6.0167,
+      "step": 215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019446401044865365,
+      "loss": 6.0687,
+      "step": 216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019439567845916682,
+      "loss": 6.0302,
+      "step": 217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001943269394889508,
+      "loss": 6.1564,
+      "step": 218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019425779383436918,
+      "loss": 6.0587,
+      "step": 219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019418824179353878,
+      "loss": 6.1082,
+      "step": 220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001941182836663287,
+      "loss": 5.9947,
+      "step": 221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019404791975435873,
+      "loss": 6.088,
+      "step": 222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019397715036099827,
+      "loss": 5.7801,
+      "step": 223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019390597579136486,
+      "loss": 6.2133,
+      "step": 224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019383439635232294,
+      "loss": 6.1426,
+      "step": 225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001937624123524826,
+      "loss": 5.9619,
+      "step": 226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019369002410219802,
+      "loss": 6.008,
+      "step": 227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019361723191356638,
+      "loss": 6.0762,
+      "step": 228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019354403610042636,
+      "loss": 5.9659,
+      "step": 229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019347043697835696,
+      "loss": 5.9773,
+      "step": 230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019339643486467587,
+      "loss": 5.9338,
+      "step": 231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001933220300784383,
+      "loss": 5.9372,
+      "step": 232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019324722294043558,
+      "loss": 5.9502,
+      "step": 233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001931720137731938,
+      "loss": 5.8214,
+      "step": 234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019309640290097218,
+      "loss": 6.1952,
+      "step": 235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019302039064976215,
+      "loss": 5.9402,
+      "step": 236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019294397734728543,
+      "loss": 5.9417,
+      "step": 237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001928671633229929,
+      "loss": 5.8543,
+      "step": 238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019278994890806322,
+      "loss": 5.8047,
+      "step": 239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019271233443540116,
+      "loss": 6.0972,
+      "step": 240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019263432023963644,
+      "loss": 5.9922,
+      "step": 241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019255590665712214,
+      "loss": 6.0486,
+      "step": 242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019247709402593315,
+      "loss": 5.9355,
+      "step": 243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019239788268586496,
+      "loss": 5.8124,
+      "step": 244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001923182729784321,
+      "loss": 6.1039,
+      "step": 245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019223826524686642,
+      "loss": 6.2633,
+      "step": 246
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019215785983611608,
+      "loss": 6.0969,
+      "step": 247
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019207705709284366,
+      "loss": 6.0495,
+      "step": 248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019199585736542484,
+      "loss": 5.97,
+      "step": 249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019191426100394687,
+      "loss": 6.042,
+      "step": 250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001918322683602071,
+      "loss": 5.8558,
+      "step": 251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001917498797877114,
+      "loss": 5.8278,
+      "step": 252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019166709564167258,
+      "loss": 5.9785,
+      "step": 253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001915839162790091,
+      "loss": 5.841,
+      "step": 254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019150034205834326,
+      "loss": 6.302,
+      "step": 255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001914163733399998,
+      "loss": 5.8299,
+      "step": 256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019133201048600424,
+      "loss": 5.756,
+      "step": 257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019124725386008152,
+      "loss": 5.8907,
+      "step": 258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001911621038276542,
+      "loss": 6.1069,
+      "step": 259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019107656075584102,
+      "loss": 5.9859,
+      "step": 260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019099062501345526,
+      "loss": 6.0143,
+      "step": 261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019090429697100325,
+      "loss": 6.1595,
+      "step": 262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019081757700068258,
+      "loss": 6.1954,
+      "step": 263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001907304654763807,
+      "loss": 5.9399,
+      "step": 264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019064296277367319,
+      "loss": 5.9363,
+      "step": 265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019055506926982219,
+      "loss": 6.0472,
+      "step": 266
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019046678534377466,
+      "loss": 5.9901,
+      "step": 267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000190378111376161,
+      "loss": 5.9872,
+      "step": 268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001902890477492931,
+      "loss": 5.9045,
+      "step": 269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00019019959484716295,
+      "loss": 5.9391,
+      "step": 270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00019010975305544075,
+      "loss": 6.1015,
+      "step": 271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00019001952276147355,
+      "loss": 5.8908,
+      "step": 272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001899289043542832,
+      "loss": 6.1077,
+      "step": 273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018983789822456509,
+      "loss": 5.9302,
+      "step": 274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000189746504764686,
+      "loss": 5.8863,
+      "step": 275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018965472436868286,
+      "loss": 5.9842,
+      "step": 276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018956255743226078,
+      "loss": 6.1546,
+      "step": 277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018947000435279144,
+      "loss": 5.8895,
+      "step": 278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018937706552931127,
+      "loss": 5.8953,
+      "step": 279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018928374136251993,
+      "loss": 5.752,
+      "step": 280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001891900322547783,
+      "loss": 6.1521,
+      "step": 281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001890959386101071,
+      "loss": 5.9788,
+      "step": 282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018900146083418483,
+      "loss": 5.9873,
+      "step": 283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018890659933434617,
+      "loss": 6.056,
+      "step": 284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018881135451958022,
+      "loss": 6.0276,
+      "step": 285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018871572680052867,
+      "loss": 5.9809,
+      "step": 286
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018861971658948415,
+      "loss": 5.8589,
+      "step": 287
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018852332430038834,
+      "loss": 5.9384,
+      "step": 288
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018842655034883018,
+      "loss": 5.9936,
+      "step": 289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018832939515204416,
+      "loss": 5.7463,
+      "step": 290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001882318591289086,
+      "loss": 5.8609,
+      "step": 291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018813394269994352,
+      "loss": 6.2196,
+      "step": 292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018803564628730915,
+      "loss": 6.0117,
+      "step": 293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018793697031480397,
+      "loss": 6.1415,
+      "step": 294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018783791520786297,
+      "loss": 6.1155,
+      "step": 295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001877384813935556,
+      "loss": 6.095,
+      "step": 296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018763866930058426,
+      "loss": 5.9662,
+      "step": 297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018753847935928215,
+      "loss": 6.0245,
+      "step": 298
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018743791200161162,
+      "loss": 5.8784,
+      "step": 299
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018733696766116213,
+      "loss": 6.0631,
+      "step": 300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018723564677314858,
+      "loss": 5.9795,
+      "step": 301
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018713394977440927,
+      "loss": 6.1493,
+      "step": 302
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018703187710340409,
+      "loss": 6.0358,
+      "step": 303
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018692942920021257,
+      "loss": 5.9614,
+      "step": 304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018682660650653213,
+      "loss": 6.219,
+      "step": 305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018672340946567607,
+      "loss": 5.9281,
+      "step": 306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018661983852257148,
+      "loss": 6.101,
+      "step": 307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001865158941237577,
+      "loss": 5.9239,
+      "step": 308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018641157671738415,
+      "loss": 5.9057,
+      "step": 309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018630688675320842,
+      "loss": 5.9317,
+      "step": 310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018620182468259436,
+      "loss": 5.7505,
+      "step": 311
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018609639095851018,
+      "loss": 5.8288,
+      "step": 312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001859905860355264,
+      "loss": 5.8722,
+      "step": 313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001858844103698139,
+      "loss": 5.9321,
+      "step": 314
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018577786441914212,
+      "loss": 5.9948,
+      "step": 315
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018567094864287685,
+      "loss": 5.8364,
+      "step": 316
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018556366350197835,
+      "loss": 5.9447,
+      "step": 317
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018545600945899952,
+      "loss": 6.041,
+      "step": 318
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018534798697808356,
+      "loss": 5.905,
+      "step": 319
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001852395965249623,
+      "loss": 5.9312,
+      "step": 320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000185130838566954,
+      "loss": 5.9468,
+      "step": 321
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018502171357296144,
+      "loss": 5.8653,
+      "step": 322
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018491222201346975,
+      "loss": 5.825,
+      "step": 323
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018480236436054464,
+      "loss": 5.7977,
+      "step": 324
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018469214108783003,
+      "loss": 5.9456,
+      "step": 325
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018458155267054633,
+      "loss": 5.9267,
+      "step": 326
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001844705995854882,
+      "loss": 6.0035,
+      "step": 327
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018435928231102253,
+      "loss": 5.8315,
+      "step": 328
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018424760132708636,
+      "loss": 5.9454,
+      "step": 329
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018413555711518492,
+      "loss": 5.7779,
+      "step": 330
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018402315015838934,
+      "loss": 5.741,
+      "step": 331
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001839103809413349,
+      "loss": 5.9734,
+      "step": 332
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001837972499502185,
+      "loss": 6.0573,
+      "step": 333
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018368375767279703,
+      "loss": 5.772,
+      "step": 334
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018356990459838496,
+      "loss": 5.9502,
+      "step": 335
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018345569121785217,
+      "loss": 5.798,
+      "step": 336
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018334111802362224,
+      "loss": 5.8834,
+      "step": 337
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018322618550966984,
+      "loss": 5.6882,
+      "step": 338
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018311089417151895,
+      "loss": 5.867,
+      "step": 339
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018299524450624054,
+      "loss": 5.7369,
+      "step": 340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018287923701245046,
+      "loss": 6.0483,
+      "step": 341
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001827628721903074,
+      "loss": 5.79,
+      "step": 342
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001826461505415106,
+      "loss": 5.7992,
+      "step": 343
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018252907256929775,
+      "loss": 5.7246,
+      "step": 344
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018241163877844274,
+      "loss": 5.6732,
+      "step": 345
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018229384967525362,
+      "loss": 5.5457,
+      "step": 346
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018217570576757033,
+      "loss": 5.6081,
+      "step": 347
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018205720756476253,
+      "loss": 5.5167,
+      "step": 348
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018193835557772737,
+      "loss": 5.7784,
+      "step": 349
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018181915031888733,
+      "loss": 5.6691,
+      "step": 350
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018169959230218802,
+      "loss": 5.6264,
+      "step": 351
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018157968204309596,
+      "loss": 5.573,
+      "step": 352
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018145942005859625,
+      "loss": 5.4522,
+      "step": 353
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018133880686719048,
+      "loss": 5.8369,
+      "step": 354
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018121784298889452,
+      "loss": 5.7463,
+      "step": 355
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001810965289452361,
+      "loss": 5.732,
+      "step": 356
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001809748652592527,
+      "loss": 5.525,
+      "step": 357
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001808528524554893,
+      "loss": 5.4988,
+      "step": 358
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018073049105999598,
+      "loss": 5.5883,
+      "step": 359
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018060778160032583,
+      "loss": 5.6593,
+      "step": 360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018048472460553257,
+      "loss": 5.5211,
+      "step": 361
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018036132060616833,
+      "loss": 5.5428,
+      "step": 362
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001802375701342813,
+      "loss": 5.7429,
+      "step": 363
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018011347372341342,
+      "loss": 5.6294,
+      "step": 364
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017998903190859816,
+      "loss": 5.6162,
+      "step": 365
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017986424522635822,
+      "loss": 5.4475,
+      "step": 366
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017973911421470312,
+      "loss": 5.4124,
+      "step": 367
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017961363941312695,
+      "loss": 5.4522,
+      "step": 368
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017948782136260603,
+      "loss": 5.4575,
+      "step": 369
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017936166060559657,
+      "loss": 5.3347,
+      "step": 370
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017923515768603228,
+      "loss": 5.473,
+      "step": 371
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001791083131493222,
+      "loss": 5.5921,
+      "step": 372
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017898112754234817,
+      "loss": 5.4951,
+      "step": 373
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001788536014134625,
+      "loss": 5.5368,
+      "step": 374
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017872573531248567,
+      "loss": 5.9409,
+      "step": 375
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001785975297907039,
+      "loss": 5.5716,
+      "step": 376
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017846898540086693,
+      "loss": 5.619,
+      "step": 377
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017834010269718526,
+      "loss": 5.6452,
+      "step": 378
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017821088223532824,
+      "loss": 5.6296,
+      "step": 379
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017808132457242134,
+      "loss": 5.5037,
+      "step": 380
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001779514302670438,
+      "loss": 5.508,
+      "step": 381
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001778211998792264,
+      "loss": 5.3682,
+      "step": 382
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017769063397044885,
+      "loss": 5.3168,
+      "step": 383
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017755973310363738,
+      "loss": 5.6438,
+      "step": 384
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017742849784316248,
+      "loss": 5.4295,
+      "step": 385
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017729692875483632,
+      "loss": 5.6247,
+      "step": 386
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017716502640591027,
+      "loss": 5.5573,
+      "step": 387
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001770327913650726,
+      "loss": 5.3483,
+      "step": 388
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017690022420244596,
+      "loss": 5.5242,
+      "step": 389
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017676732548958492,
+      "loss": 5.1811,
+      "step": 390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017663409579947353,
+      "loss": 5.331,
+      "step": 391
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017650053570652271,
+      "loss": 5.3544,
+      "step": 392
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017636664578656803,
+      "loss": 5.3406,
+      "step": 393
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017623242661686704,
+      "loss": 5.6099,
+      "step": 394
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001760978787760968,
+      "loss": 5.4096,
+      "step": 395
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017596300284435142,
+      "loss": 5.4136,
+      "step": 396
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017582779940313956,
+      "loss": 5.3646,
+      "step": 397
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017569226903538197,
+      "loss": 5.2344,
+      "step": 398
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017555641232540884,
+      "loss": 5.0619,
+      "step": 399
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017542022985895739,
+      "loss": 5.4175,
+      "step": 400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017528372222316935,
+      "loss": 5.3692,
+      "step": 401
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017514689000658836,
+      "loss": 5.3651,
+      "step": 402
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017500973379915746,
+      "loss": 5.0469,
+      "step": 403
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017487225419221658,
+      "loss": 5.1517,
+      "step": 404
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017473445177849998,
+      "loss": 5.047,
+      "step": 405
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017459632715213365,
+      "loss": 5.3905,
+      "step": 406
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001744578809086328,
+      "loss": 5.3319,
+      "step": 407
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017431911364489922,
+      "loss": 6.1506,
+      "step": 408
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001741800259592188,
+      "loss": 6.49,
+      "step": 409
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017404061845125896,
+      "loss": 6.6326,
+      "step": 410
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017390089172206592,
+      "loss": 6.3181,
+      "step": 411
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017376084637406222,
+      "loss": 6.0462,
+      "step": 412
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017362048301104417,
+      "loss": 6.0371,
+      "step": 413
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001734798022381791,
+      "loss": 6.0669,
+      "step": 414
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017333880466200284,
+      "loss": 5.9624,
+      "step": 415
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017319749089041723,
+      "loss": 5.8624,
+      "step": 416
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017305586153268713,
+      "loss": 5.8335,
+      "step": 417
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001729139171994382,
+      "loss": 5.9358,
+      "step": 418
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017277165850265407,
+      "loss": 5.8574,
+      "step": 419
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017262908605567367,
+      "loss": 5.8699,
+      "step": 420
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017248620047318874,
+      "loss": 5.9223,
+      "step": 421
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000172343002371241,
+      "loss": 5.8174,
+      "step": 422
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017219949236721958,
+      "loss": 5.467,
+      "step": 423
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017205567107985838,
+      "loss": 5.7541,
+      "step": 424
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001719115391292334,
+      "loss": 5.9228,
+      "step": 425
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017176709713676006,
+      "loss": 5.8562,
+      "step": 426
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017162234572519038,
+      "loss": 5.598,
+      "step": 427
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017147728551861053,
+      "loss": 5.611,
+      "step": 428
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017133191714243805,
+      "loss": 5.7704,
+      "step": 429
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000171186241223419,
+      "loss": 5.7631,
+      "step": 430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017104025838962558,
+      "loss": 5.6173,
+      "step": 431
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017089396927045302,
+      "loss": 5.6654,
+      "step": 432
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001707473744966173,
+      "loss": 5.7368,
+      "step": 433
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017060047470015204,
+      "loss": 5.6807,
+      "step": 434
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017045327051440598,
+      "loss": 5.5626,
+      "step": 435
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001703057625740403,
+      "loss": 5.8536,
+      "step": 436
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001701579515150257,
+      "loss": 5.7999,
+      "step": 437
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017000983797463976,
+      "loss": 5.6055,
+      "step": 438
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016986142259146428,
+      "loss": 5.7887,
+      "step": 439
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001697127060053824,
+      "loss": 5.5493,
+      "step": 440
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016956368885757573,
+      "loss": 5.7334,
+      "step": 441
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000169414371790522,
+      "loss": 5.4013,
+      "step": 442
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016926475544799176,
+      "loss": 5.6401,
+      "step": 443
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016911484047504605,
+      "loss": 5.6064,
+      "step": 444
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016896462751803335,
+      "loss": 5.5257,
+      "step": 445
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016881411722458688,
+      "loss": 5.4696,
+      "step": 446
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016866331024362188,
+      "loss": 5.6903,
+      "step": 447
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001685122072253326,
+      "loss": 5.5332,
+      "step": 448
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016836080882118976,
+      "loss": 5.6702,
+      "step": 449
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016820911568393758,
+      "loss": 5.6372,
+      "step": 450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000168057128467591,
+      "loss": 5.6269,
+      "step": 451
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016790484782743282,
+      "loss": 5.5996,
+      "step": 452
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016775227442001098,
+      "loss": 5.4543,
+      "step": 453
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016759940890313565,
+      "loss": 5.4548,
+      "step": 454
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016744625193587643,
+      "loss": 5.5901,
+      "step": 455
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001672928041785594,
+      "loss": 5.5021,
+      "step": 456
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016713906629276445,
+      "loss": 5.6082,
+      "step": 457
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016698503894132234,
+      "loss": 5.5819,
+      "step": 458
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001668307227883118,
+      "loss": 5.6726,
+      "step": 459
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016667611849905666,
+      "loss": 5.3321,
+      "step": 460
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016652122674012322,
+      "loss": 5.4179,
+      "step": 461
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016636604817931697,
+      "loss": 5.291,
+      "step": 462
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016621058348568007,
+      "loss": 5.2556,
+      "step": 463
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001660548333294882,
+      "loss": 5.4275,
+      "step": 464
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016589879838224794,
+      "loss": 6.5086,
+      "step": 465
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016574247931669365,
+      "loss": 5.3442,
+      "step": 466
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016558587680678462,
+      "loss": 6.0182,
+      "step": 467
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016542899152770222,
+      "loss": 5.6566,
+      "step": 468
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016527182415584699,
+      "loss": 5.6803,
+      "step": 469
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001651143753688356,
+      "loss": 5.569,
+      "step": 470
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016495664584549816,
+      "loss": 5.6771,
+      "step": 471
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001647986362658751,
+      "loss": 5.653,
+      "step": 472
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001646403473112142,
+      "loss": 5.577,
+      "step": 473
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001644817796639679,
+      "loss": 5.4513,
+      "step": 474
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016432293400779009,
+      "loss": 5.7484,
+      "step": 475
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016416381102753332,
+      "loss": 5.521,
+      "step": 476
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016400441140924582,
+      "loss": 5.2917,
+      "step": 477
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016384473584016851,
+      "loss": 5.2958,
+      "step": 478
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016368478500873205,
+      "loss": 5.5453,
+      "step": 479
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016352455960455387,
+      "loss": 5.2458,
+      "step": 480
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001633640603184352,
+      "loss": 5.3683,
+      "step": 481
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016320328784235813,
+      "loss": 5.4871,
+      "step": 482
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016304224286948257,
+      "loss": 5.4113,
+      "step": 483
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016288092609414328,
+      "loss": 5.3319,
+      "step": 484
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016271933821184685,
+      "loss": 5.4383,
+      "step": 485
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016255747991926876,
+      "loss": 5.3262,
+      "step": 486
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001623953519142503,
+      "loss": 5.3852,
+      "step": 487
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016223295489579576,
+      "loss": 5.5641,
+      "step": 488
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016207028956406906,
+      "loss": 5.5448,
+      "step": 489
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016190735662039107,
+      "loss": 5.5126,
+      "step": 490
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001617441567672364,
+      "loss": 5.3183,
+      "step": 491
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001615806907082304,
+      "loss": 5.4636,
+      "step": 492
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001614169591481462,
+      "loss": 5.3112,
+      "step": 493
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001612529627929016,
+      "loss": 5.3977,
+      "step": 494
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016108870234955608,
+      "loss": 5.2278,
+      "step": 495
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016092417852630762,
+      "loss": 5.2556,
+      "step": 496
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001607593920324899,
+      "loss": 5.4058,
+      "step": 497
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016059434357856893,
+      "loss": 5.3785,
+      "step": 498
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00016042903387614022,
+      "loss": 5.2771,
+      "step": 499
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00016026346363792567,
+      "loss": 5.1159,
+      "step": 500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00016009763357777035,
+      "loss": 5.3192,
+      "step": 501
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015993154441063965,
+      "loss": 5.2861,
+      "step": 502
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000159765196852616,
+      "loss": 5.4777,
+      "step": 503
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015959859162089596,
+      "loss": 5.4552,
+      "step": 504
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015943172943378692,
+      "loss": 5.3326,
+      "step": 505
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015926461101070424,
+      "loss": 5.2475,
+      "step": 506
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015909723707216786,
+      "loss": 5.508,
+      "step": 507
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015892960833979955,
+      "loss": 5.244,
+      "step": 508
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015876172553631944,
+      "loss": 5.0781,
+      "step": 509
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001585935893855432,
+      "loss": 5.2627,
+      "step": 510
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015842520061237872,
+      "loss": 5.4044,
+      "step": 511
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000158256559942823,
+      "loss": 5.1111,
+      "step": 512
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015808766810395923,
+      "loss": 5.1239,
+      "step": 513
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015791852582395334,
+      "loss": 5.3154,
+      "step": 514
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015774913383205118,
+      "loss": 5.2221,
+      "step": 515
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015757949285857503,
+      "loss": 5.1607,
+      "step": 516
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015740960363492079,
+      "loss": 5.089,
+      "step": 517
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001572394668935546,
+      "loss": 5.119,
+      "step": 518
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001570690833680098,
+      "loss": 5.3552,
+      "step": 519
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015689845379288368,
+      "loss": 5.2033,
+      "step": 520
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015672757890383435,
+      "loss": 5.161,
+      "step": 521
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001565564594375777,
+      "loss": 5.106,
+      "step": 522
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015638509613188388,
+      "loss": 5.54,
+      "step": 523
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015621348972557454,
+      "loss": 5.1417,
+      "step": 524
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015604164095851933,
+      "loss": 5.5533,
+      "step": 525
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015586955057163285,
+      "loss": 5.2707,
+      "step": 526
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001556972193068714,
+      "loss": 5.3316,
+      "step": 527
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001555246479072299,
+      "loss": 5.2491,
+      "step": 528
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001553518371167385,
+      "loss": 5.2957,
+      "step": 529
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015517878768045952,
+      "loss": 5.1038,
+      "step": 530
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015500550034448413,
+      "loss": 5.198,
+      "step": 531
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015483197585592926,
+      "loss": 5.1818,
+      "step": 532
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001546582149629343,
+      "loss": 5.207,
+      "step": 533
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015448421841465776,
+      "loss": 6.7397,
+      "step": 534
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001543099869612744,
+      "loss": 6.7066,
+      "step": 535
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015413552135397146,
+      "loss": 6.3033,
+      "step": 536
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000153960822344946,
+      "loss": 6.2565,
+      "step": 537
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001537858906874012,
+      "loss": 6.2244,
+      "step": 538
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001536107271355433,
+      "loss": 6.4926,
+      "step": 539
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001534353324445785,
+      "loss": 6.196,
+      "step": 540
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015325970737070928,
+      "loss": 6.1896,
+      "step": 541
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015308385267113166,
+      "loss": 6.0204,
+      "step": 542
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015290776910403144,
+      "loss": 6.1268,
+      "step": 543
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015273145742858137,
+      "loss": 5.95,
+      "step": 544
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015255491840493755,
+      "loss": 5.9416,
+      "step": 545
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015237815279423633,
+      "loss": 5.9554,
+      "step": 546
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015220116135859093,
+      "loss": 5.9128,
+      "step": 547
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001520239448610882,
+      "loss": 5.8347,
+      "step": 548
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001518465040657855,
+      "loss": 5.7894,
+      "step": 549
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001516688397377069,
+      "loss": 5.7632,
+      "step": 550
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015149095264284056,
+      "loss": 5.6543,
+      "step": 551
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015131284354813484,
+      "loss": 5.7345,
+      "step": 552
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015113451322149542,
+      "loss": 5.8094,
+      "step": 553
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001509559624317816,
+      "loss": 5.7153,
+      "step": 554
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015077719194880342,
+      "loss": 5.7997,
+      "step": 555
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015059820254331794,
+      "loss": 5.7398,
+      "step": 556
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00015041899498702616,
+      "loss": 5.5224,
+      "step": 557
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00015023957005256964,
+      "loss": 5.8208,
+      "step": 558
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001500599285135271,
+      "loss": 5.6596,
+      "step": 559
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014988007114441118,
+      "loss": 5.4529,
+      "step": 560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000149699998720665,
+      "loss": 5.5734,
+      "step": 561
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014951971201865901,
+      "loss": 5.6206,
+      "step": 562
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014933921181568732,
+      "loss": 5.6413,
+      "step": 563
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014915849888996464,
+      "loss": 5.5859,
+      "step": 564
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014897757402062284,
+      "loss": 5.3504,
+      "step": 565
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014879643798770752,
+      "loss": 5.4171,
+      "step": 566
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014861509157217473,
+      "loss": 5.4439,
+      "step": 567
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001484335355558875,
+      "loss": 5.3649,
+      "step": 568
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014825177072161266,
+      "loss": 5.3784,
+      "step": 569
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014806979785301726,
+      "loss": 5.3671,
+      "step": 570
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014788761773466526,
+      "loss": 5.4474,
+      "step": 571
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014770523115201426,
+      "loss": 5.2232,
+      "step": 572
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014752263889141194,
+      "loss": 5.1777,
+      "step": 573
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014733984174009271,
+      "loss": 5.3468,
+      "step": 574
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014715684048617449,
+      "loss": 5.3092,
+      "step": 575
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014697363591865506,
+      "loss": 5.0584,
+      "step": 576
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014679022882740876,
+      "loss": 5.0258,
+      "step": 577
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001466066200031832,
+      "loss": 5.2818,
+      "step": 578
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001464228102375957,
+      "loss": 5.4381,
+      "step": 579
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014623880032312982,
+      "loss": 5.3414,
+      "step": 580
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014605459105313225,
+      "loss": 5.3739,
+      "step": 581
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014587018322180905,
+      "loss": 5.3984,
+      "step": 582
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001456855776242224,
+      "loss": 5.394,
+      "step": 583
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014550077505628712,
+      "loss": 5.1957,
+      "step": 584
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014531577631476733,
+      "loss": 5.2464,
+      "step": 585
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014513058219727275,
+      "loss": 5.2046,
+      "step": 586
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014494519350225567,
+      "loss": 5.0035,
+      "step": 587
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014475961102900718,
+      "loss": 5.4268,
+      "step": 588
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014457383557765386,
+      "loss": 5.281,
+      "step": 589
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014438786794915425,
+      "loss": 5.4001,
+      "step": 590
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014420170894529553,
+      "loss": 5.1665,
+      "step": 591
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014401535936868994,
+      "loss": 5.3536,
+      "step": 592
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014382882002277134,
+      "loss": 5.5685,
+      "step": 593
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014364209171179184,
+      "loss": 5.2668,
+      "step": 594
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014345517524081813,
+      "loss": 5.5668,
+      "step": 595
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001432680714157283,
+      "loss": 5.1136,
+      "step": 596
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014308078104320807,
+      "loss": 5.1413,
+      "step": 597
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014289330493074757,
+      "loss": 5.2598,
+      "step": 598
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001427056438866376,
+      "loss": 5.3748,
+      "step": 599
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001425177987199664,
+      "loss": 5.2582,
+      "step": 600
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014232977024061596,
+      "loss": 5.2296,
+      "step": 601
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014214155925925862,
+      "loss": 5.281,
+      "step": 602
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014195316658735364,
+      "loss": 5.2057,
+      "step": 603
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014176459303714357,
+      "loss": 5.2196,
+      "step": 604
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014157583942165075,
+      "loss": 4.9665,
+      "step": 605
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000141386906554674,
+      "loss": 4.8166,
+      "step": 606
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014119779525078484,
+      "loss": 5.0158,
+      "step": 607
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00014100850632532417,
+      "loss": 5.129,
+      "step": 608
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00014081904059439867,
+      "loss": 4.9277,
+      "step": 609
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00014062939887487735,
+      "loss": 4.8676,
+      "step": 610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00014043958198438792,
+      "loss": 5.1919,
+      "step": 611
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00014024959074131337,
+      "loss": 5.0895,
+      "step": 612
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001400594259647884,
+      "loss": 5.3118,
+      "step": 613
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001398690884746958,
+      "loss": 5.3553,
+      "step": 614
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013967857909166323,
+      "loss": 5.2136,
+      "step": 615
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013948789863705912,
+      "loss": 5.1867,
+      "step": 616
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001392970479329898,
+      "loss": 4.9987,
+      "step": 617
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001391060278022954,
+      "loss": 5.2177,
+      "step": 618
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013891483906854662,
+      "loss": 4.8774,
+      "step": 619
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013872348255604102,
+      "loss": 4.8828,
+      "step": 620
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013853195908979956,
+      "loss": 4.9895,
+      "step": 621
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013834026949556303,
+      "loss": 5.1435,
+      "step": 622
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001381484145997884,
+      "loss": 5.0427,
+      "step": 623
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013795639522964546,
+      "loss": 4.8991,
+      "step": 624
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013776421221301291,
+      "loss": 4.9406,
+      "step": 625
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013757186637847524,
+      "loss": 5.1285,
+      "step": 626
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013737935855531876,
+      "loss": 5.0526,
+      "step": 627
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001371866895735282,
+      "loss": 4.7671,
+      "step": 628
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001369938602637832,
+      "loss": 5.0425,
+      "step": 629
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013680087145745453,
+      "loss": 4.7975,
+      "step": 630
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013660772398660072,
+      "loss": 5.0651,
+      "step": 631
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013641441868396427,
+      "loss": 5.0222,
+      "step": 632
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013622095638296826,
+      "loss": 5.0224,
+      "step": 633
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001360273379177126,
+      "loss": 5.2555,
+      "step": 634
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013583356412297056,
+      "loss": 5.1049,
+      "step": 635
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013563963583418493,
+      "loss": 5.0068,
+      "step": 636
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013544555388746483,
+      "loss": 5.0798,
+      "step": 637
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013525131911958167,
+      "loss": 4.7942,
+      "step": 638
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013505693236796584,
+      "loss": 5.0619,
+      "step": 639
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013486239447070298,
+      "loss": 5.0217,
+      "step": 640
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001346677062665304,
+      "loss": 5.0284,
+      "step": 641
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013447286859483336,
+      "loss": 4.9029,
+      "step": 642
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013427788229564169,
+      "loss": 4.9576,
+      "step": 643
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013408274820962584,
+      "loss": 4.788,
+      "step": 644
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013388746717809362,
+      "loss": 4.9099,
+      "step": 645
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001336920400429862,
+      "loss": 5.0374,
+      "step": 646
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001334964676468748,
+      "loss": 4.8558,
+      "step": 647
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001333007508329569,
+      "loss": 5.0715,
+      "step": 648
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013310489044505262,
+      "loss": 4.7199,
+      "step": 649
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000132908887327601,
+      "loss": 4.708,
+      "step": 650
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001327127423256566,
+      "loss": 4.7287,
+      "step": 651
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001325164562848856,
+      "loss": 5.2607,
+      "step": 652
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001323200300515623,
+      "loss": 5.1174,
+      "step": 653
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013212346447256544,
+      "loss": 4.7678,
+      "step": 654
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001319267603953745,
+      "loss": 5.0974,
+      "step": 655
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013172991866806612,
+      "loss": 4.9867,
+      "step": 656
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013153294013931036,
+      "loss": 5.0064,
+      "step": 657
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00013133582565836723,
+      "loss": 4.8399,
+      "step": 658
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00013113857607508265,
+      "loss": 4.9975,
+      "step": 659
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00013094119223988526,
+      "loss": 4.7927,
+      "step": 660
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001307436750037824,
+      "loss": 4.9378,
+      "step": 661
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00013054602521835658,
+      "loss": 4.9675,
+      "step": 662
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00013034824373576176,
+      "loss": 5.2143,
+      "step": 663
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00013015033140871974,
+      "loss": 5.0071,
+      "step": 664
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012995228909051655,
+      "loss": 4.891,
+      "step": 665
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012975411763499833,
+      "loss": 4.9652,
+      "step": 666
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012955581789656843,
+      "loss": 4.9374,
+      "step": 667
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001293573907301829,
+      "loss": 4.6822,
+      "step": 668
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001291588369913475,
+      "loss": 5.157,
+      "step": 669
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012896015753611345,
+      "loss": 4.7871,
+      "step": 670
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012876135322107413,
+      "loss": 4.7692,
+      "step": 671
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001285624249033612,
+      "loss": 4.7584,
+      "step": 672
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012836337344064095,
+      "loss": 4.8607,
+      "step": 673
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012816419969111065,
+      "loss": 5.046,
+      "step": 674
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012796490451349467,
+      "loss": 4.768,
+      "step": 675
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012776548876704106,
+      "loss": 4.7456,
+      "step": 676
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012756595331151762,
+      "loss": 4.6405,
+      "step": 677
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001273662990072083,
+      "loss": 4.9068,
+      "step": 678
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012716652671490938,
+      "loss": 5.0108,
+      "step": 679
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.000126966637295926,
+      "loss": 4.9143,
+      "step": 680
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012676663161206804,
+      "loss": 5.1764,
+      "step": 681
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012656651052564687,
+      "loss": 4.9519,
+      "step": 682
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012636627489947134,
+      "loss": 4.6917,
+      "step": 683
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001261659255968441,
+      "loss": 5.2901,
+      "step": 684
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001259654634815579,
+      "loss": 4.6892,
+      "step": 685
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012576488941789193,
+      "loss": 5.0569,
+      "step": 686
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.000125564204270608,
+      "loss": 5.042,
+      "step": 687
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012536340890494688,
+      "loss": 4.839,
+      "step": 688
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012516250418662448,
+      "loss": 4.7776,
+      "step": 689
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012496149098182826,
+      "loss": 4.739,
+      "step": 690
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012476037015721333,
+      "loss": 4.5773,
+      "step": 691
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012455914257989887,
+      "loss": 4.8296,
+      "step": 692
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012435780911746426,
+      "loss": 4.952,
+      "step": 693
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012415637063794543,
+      "loss": 4.9343,
+      "step": 694
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012395482800983101,
+      "loss": 4.6233,
+      "step": 695
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012375318210205882,
+      "loss": 4.76,
+      "step": 696
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001235514337840118,
+      "loss": 4.836,
+      "step": 697
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001233495839255145,
+      "loss": 4.6404,
+      "step": 698
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012314763339682928,
+      "loss": 4.8832,
+      "step": 699
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012294558306865242,
+      "loss": 4.9661,
+      "step": 700
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012274343381211066,
+      "loss": 4.5808,
+      "step": 701
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012254118649875705,
+      "loss": 4.7694,
+      "step": 702
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001223388420005676,
+      "loss": 4.6978,
+      "step": 703
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001221364011899372,
+      "loss": 4.801,
+      "step": 704
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012193386493967607,
+      "loss": 4.6989,
+      "step": 705
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012173123412300582,
+      "loss": 4.9173,
+      "step": 706
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001215285096135558,
+      "loss": 5.0214,
+      "step": 707
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00012132569228535944,
+      "loss": 4.7094,
+      "step": 708
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00012112278301285011,
+      "loss": 4.6519,
+      "step": 709
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00012091978267085786,
+      "loss": 4.9717,
+      "step": 710
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00012071669213460515,
+      "loss": 4.4051,
+      "step": 711
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00012051351227970348,
+      "loss": 4.5933,
+      "step": 712
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001203102439821493,
+      "loss": 4.7024,
+      "step": 713
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001201068881183205,
+      "loss": 4.6961,
+      "step": 714
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011990344556497238,
+      "loss": 4.7653,
+      "step": 715
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011969991719923412,
+      "loss": 4.3755,
+      "step": 716
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001194963038986048,
+      "loss": 4.2934,
+      "step": 717
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011929260654094969,
+      "loss": 4.5841,
+      "step": 718
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011908882600449654,
+      "loss": 4.717,
+      "step": 719
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011888496316783158,
+      "loss": 4.4808,
+      "step": 720
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011868101890989605,
+      "loss": 4.8887,
+      "step": 721
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011847699410998206,
+      "loss": 4.6303,
+      "step": 722
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011827288964772911,
+      "loss": 4.8038,
+      "step": 723
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011806870640312009,
+      "loss": 4.5396,
+      "step": 724
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011786444525647753,
+      "loss": 4.4964,
+      "step": 725
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001176601070884599,
+      "loss": 4.4879,
+      "step": 726
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011745569278005768,
+      "loss": 4.5072,
+      "step": 727
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011725120321258967,
+      "loss": 4.6169,
+      "step": 728
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011704663926769912,
+      "loss": 4.6185,
+      "step": 729
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011684200182734996,
+      "loss": 4.5036,
+      "step": 730
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011663729177382299,
+      "loss": 4.6097,
+      "step": 731
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011643250998971207,
+      "loss": 4.5821,
+      "step": 732
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011622765735792033,
+      "loss": 4.9634,
+      "step": 733
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011602273476165637,
+      "loss": 4.5491,
+      "step": 734
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001158177430844304,
+      "loss": 4.6825,
+      "step": 735
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011561268321005052,
+      "loss": 4.5378,
+      "step": 736
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011540755602261879,
+      "loss": 4.5477,
+      "step": 737
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011520236240652753,
+      "loss": 4.5072,
+      "step": 738
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011499710324645545,
+      "loss": 4.5155,
+      "step": 739
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011479177942736387,
+      "loss": 4.1123,
+      "step": 740
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011458639183449285,
+      "loss": 4.7578,
+      "step": 741
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011438094135335749,
+      "loss": 4.3124,
+      "step": 742
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011417542886974382,
+      "loss": 4.4428,
+      "step": 743
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001139698552697055,
+      "loss": 4.7097,
+      "step": 744
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011376422143955936,
+      "loss": 4.2994,
+      "step": 745
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011355852826588218,
+      "loss": 4.5129,
+      "step": 746
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011335277663550643,
+      "loss": 4.4841,
+      "step": 747
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011314696743551671,
+      "loss": 4.4364,
+      "step": 748
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011294110155324575,
+      "loss": 4.288,
+      "step": 749
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001127351798762707,
+      "loss": 4.5818,
+      "step": 750
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011252920329240928,
+      "loss": 4.6583,
+      "step": 751
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011232317268971585,
+      "loss": 4.4559,
+      "step": 752
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011211708895647788,
+      "loss": 4.3824,
+      "step": 753
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011191095298121161,
+      "loss": 4.5408,
+      "step": 754
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001117047656526588,
+      "loss": 4.6453,
+      "step": 755
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011149852785978237,
+      "loss": 4.4686,
+      "step": 756
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011129224049176304,
+      "loss": 4.7117,
+      "step": 757
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00011108590443799509,
+      "loss": 4.203,
+      "step": 758
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00011087952058808283,
+      "loss": 4.2739,
+      "step": 759
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00011067308983183658,
+      "loss": 4.6015,
+      "step": 760
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00011046661305926886,
+      "loss": 4.4285,
+      "step": 761
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00011026009116059068,
+      "loss": 4.6538,
+      "step": 762
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00011005352502620751,
+      "loss": 4.5111,
+      "step": 763
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010984691554671566,
+      "loss": 4.4474,
+      "step": 764
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010964026361289817,
+      "loss": 4.5443,
+      "step": 765
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010943357011572124,
+      "loss": 4.3126,
+      "step": 766
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010922683594633021,
+      "loss": 4.1897,
+      "step": 767
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010902006199604577,
+      "loss": 4.6081,
+      "step": 768
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010881324915636019,
+      "loss": 4.5125,
+      "step": 769
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010860639831893331,
+      "loss": 4.2562,
+      "step": 770
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010839951037558887,
+      "loss": 4.0884,
+      "step": 771
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010819258621831057,
+      "loss": 4.3289,
+      "step": 772
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001079856267392382,
+      "loss": 4.38,
+      "step": 773
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010777863283066391,
+      "loss": 4.3863,
+      "step": 774
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010757160538502826,
+      "loss": 4.5037,
+      "step": 775
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001073645452949164,
+      "loss": 4.3084,
+      "step": 776
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010715745345305419,
+      "loss": 4.4822,
+      "step": 777
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010695033075230448,
+      "loss": 4.3521,
+      "step": 778
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010674317808566303,
+      "loss": 4.6359,
+      "step": 779
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010653599634625493,
+      "loss": 4.5305,
+      "step": 780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010632878642733055,
+      "loss": 4.1261,
+      "step": 781
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010612154922226171,
+      "loss": 4.0726,
+      "step": 782
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010591428562453805,
+      "loss": 4.5296,
+      "step": 783
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001057069965277627,
+      "loss": 4.2673,
+      "step": 784
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010549968282564904,
+      "loss": 4.3051,
+      "step": 785
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010529234541201631,
+      "loss": 4.5406,
+      "step": 786
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010508498518078615,
+      "loss": 4.5241,
+      "step": 787
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010487760302597837,
+      "loss": 4.1474,
+      "step": 788
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010467019984170757,
+      "loss": 4.5275,
+      "step": 789
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010446277652217876,
+      "loss": 4.5127,
+      "step": 790
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010425533396168392,
+      "loss": 4.3639,
+      "step": 791
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010404787305459795,
+      "loss": 4.4237,
+      "step": 792
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010384039469537478,
+      "loss": 4.5244,
+      "step": 793
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010363289977854376,
+      "loss": 4.2291,
+      "step": 794
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010342538919870538,
+      "loss": 4.3009,
+      "step": 795
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010321786385052789,
+      "loss": 4.4757,
+      "step": 796
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010301032462874303,
+      "loss": 4.3542,
+      "step": 797
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010280277242814254,
+      "loss": 4.4933,
+      "step": 798
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010259520814357389,
+      "loss": 4.1104,
+      "step": 799
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010238763266993687,
+      "loss": 4.5627,
+      "step": 800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010218004690217942,
+      "loss": 4.541,
+      "step": 801
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010197245173529379,
+      "loss": 4.6184,
+      "step": 802
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010176484806431288,
+      "loss": 4.2547,
+      "step": 803
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010155723678430618,
+      "loss": 4.3739,
+      "step": 804
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00010134961879037599,
+      "loss": 4.6108,
+      "step": 805
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00010114199497765356,
+      "loss": 4.3169,
+      "step": 806
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00010093436624129529,
+      "loss": 4.2019,
+      "step": 807
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001007267334764787,
+      "loss": 4.4107,
+      "step": 808
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00010051909757839878,
+      "loss": 4.5106,
+      "step": 809
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00010031145944226392,
+      "loss": 4.5698,
+      "step": 810
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00010010381996329228,
+      "loss": 4.2866,
+      "step": 811
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.98961800367077e-05,
+      "loss": 4.1296,
+      "step": 812
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.96885405577361e-05,
+      "loss": 4.1116,
+      "step": 813
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.948090242160126e-05,
+      "loss": 4.4611,
+      "step": 814
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.927326652352132e-05,
+      "loss": 4.8078,
+      "step": 815
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.906563375870472e-05,
+      "loss": 4.2578,
+      "step": 816
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.885800502234647e-05,
+      "loss": 4.3915,
+      "step": 817
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.865038120962406e-05,
+      "loss": 4.6224,
+      "step": 818
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.844276321569385e-05,
+      "loss": 3.9956,
+      "step": 819
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.823515193568715e-05,
+      "loss": 4.5488,
+      "step": 820
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.802754826470623e-05,
+      "loss": 4.3395,
+      "step": 821
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.781995309782064e-05,
+      "loss": 4.8554,
+      "step": 822
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.761236733006312e-05,
+      "loss": 5.1178,
+      "step": 823
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.740479185642612e-05,
+      "loss": 4.9433,
+      "step": 824
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.719722757185752e-05,
+      "loss": 4.6323,
+      "step": 825
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.698967537125697e-05,
+      "loss": 5.0855,
+      "step": 826
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.678213614947213e-05,
+      "loss": 5.5184,
+      "step": 827
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.657461080129465e-05,
+      "loss": 4.6087,
+      "step": 828
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.636710022145629e-05,
+      "loss": 5.3392,
+      "step": 829
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.615960530462522e-05,
+      "loss": 4.8343,
+      "step": 830
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.595212694540207e-05,
+      "loss": 4.9906,
+      "step": 831
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.57446660383161e-05,
+      "loss": 4.9262,
+      "step": 832
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.553722347782124e-05,
+      "loss": 4.7492,
+      "step": 833
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.532980015829246e-05,
+      "loss": 4.7613,
+      "step": 834
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.512239697402164e-05,
+      "loss": 4.9193,
+      "step": 835
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.49150148192139e-05,
+      "loss": 4.8323,
+      "step": 836
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.470765458798368e-05,
+      "loss": 4.9696,
+      "step": 837
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.450031717435097e-05,
+      "loss": 4.6889,
+      "step": 838
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.429300347223732e-05,
+      "loss": 4.8288,
+      "step": 839
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.4085714375462e-05,
+      "loss": 4.7869,
+      "step": 840
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.387845077773828e-05,
+      "loss": 4.4896,
+      "step": 841
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.367121357266947e-05,
+      "loss": 4.4236,
+      "step": 842
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.34640036537451e-05,
+      "loss": 4.5183,
+      "step": 843
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.325682191433697e-05,
+      "loss": 4.6732,
+      "step": 844
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.304966924769556e-05,
+      "loss": 4.4366,
+      "step": 845
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.284254654694583e-05,
+      "loss": 4.76,
+      "step": 846
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.263545470508364e-05,
+      "loss": 4.7493,
+      "step": 847
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.242839461497175e-05,
+      "loss": 4.7151,
+      "step": 848
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.22213671693361e-05,
+      "loss": 4.4304,
+      "step": 849
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.201437326076185e-05,
+      "loss": 4.6306,
+      "step": 850
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.180741378168946e-05,
+      "loss": 4.4189,
+      "step": 851
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.160048962441114e-05,
+      "loss": 4.5622,
+      "step": 852
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.139360168106672e-05,
+      "loss": 4.2736,
+      "step": 853
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.118675084363986e-05,
+      "loss": 4.1404,
+      "step": 854
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.097993800395421e-05,
+      "loss": 4.2279,
+      "step": 855
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.077316405366981e-05,
+      "loss": 4.2193,
+      "step": 856
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.056642988427878e-05,
+      "loss": 4.4301,
+      "step": 857
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.035973638710183e-05,
+      "loss": 4.4324,
+      "step": 858
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.015308445328435e-05,
+      "loss": 4.5676,
+      "step": 859
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.994647497379251e-05,
+      "loss": 4.2919,
+      "step": 860
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.973990883940936e-05,
+      "loss": 4.5187,
+      "step": 861
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.953338694073115e-05,
+      "loss": 4.4055,
+      "step": 862
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.932691016816346e-05,
+      "loss": 4.54,
+      "step": 863
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.912047941191719e-05,
+      "loss": 4.6749,
+      "step": 864
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.891409556200495e-05,
+      "loss": 4.4885,
+      "step": 865
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.870775950823697e-05,
+      "loss": 4.4983,
+      "step": 866
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.850147214021764e-05,
+      "loss": 4.327,
+      "step": 867
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.829523434734125e-05,
+      "loss": 4.1982,
+      "step": 868
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.808904701878839e-05,
+      "loss": 4.4776,
+      "step": 869
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.788291104352214e-05,
+      "loss": 4.5351,
+      "step": 870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.767682731028415e-05,
+      "loss": 4.3506,
+      "step": 871
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.747079670759076e-05,
+      "loss": 3.9608,
+      "step": 872
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.726482012372931e-05,
+      "loss": 4.2236,
+      "step": 873
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.705889844675427e-05,
+      "loss": 4.2721,
+      "step": 874
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.685303256448331e-05,
+      "loss": 4.0514,
+      "step": 875
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.664722336449355e-05,
+      "loss": 4.1015,
+      "step": 876
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.644147173411785e-05,
+      "loss": 4.2572,
+      "step": 877
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.623577856044066e-05,
+      "loss": 4.5093,
+      "step": 878
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.603014473029456e-05,
+      "loss": 4.3019,
+      "step": 879
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.582457113025618e-05,
+      "loss": 4.5843,
+      "step": 880
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.561905864664255e-05,
+      "loss": 3.8994,
+      "step": 881
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.541360816550716e-05,
+      "loss": 4.4006,
+      "step": 882
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.520822057263617e-05,
+      "loss": 3.985,
+      "step": 883
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.500289675354457e-05,
+      "loss": 4.0522,
+      "step": 884
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.479763759347251e-05,
+      "loss": 3.9589,
+      "step": 885
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.459244397738126e-05,
+      "loss": 4.0963,
+      "step": 886
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.438731678994949e-05,
+      "loss": 4.2374,
+      "step": 887
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.418225691556962e-05,
+      "loss": 4.2755,
+      "step": 888
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.397726523834364e-05,
+      "loss": 4.4675,
+      "step": 889
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.377234264207969e-05,
+      "loss": 5.051,
+      "step": 890
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.356749001028794e-05,
+      "loss": 4.6598,
+      "step": 891
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.336270822617702e-05,
+      "loss": 4.5873,
+      "step": 892
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.315799817265008e-05,
+      "loss": 4.5465,
+      "step": 893
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.295336073230089e-05,
+      "loss": 4.5752,
+      "step": 894
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.274879678741034e-05,
+      "loss": 4.5434,
+      "step": 895
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.254430721994234e-05,
+      "loss": 4.8368,
+      "step": 896
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.233989291154016e-05,
+      "loss": 4.881,
+      "step": 897
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.213555474352247e-05,
+      "loss": 4.6397,
+      "step": 898
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.193129359687995e-05,
+      "loss": 4.6292,
+      "step": 899
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.172711035227091e-05,
+      "loss": 4.5801,
+      "step": 900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.152300589001794e-05,
+      "loss": 4.4911,
+      "step": 901
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.131898109010396e-05,
+      "loss": 4.5491,
+      "step": 902
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.111503683216844e-05,
+      "loss": 5.6175,
+      "step": 903
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.091117399550352e-05,
+      "loss": 5.3736,
+      "step": 904
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.070739345905032e-05,
+      "loss": 5.6802,
+      "step": 905
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.050369610139523e-05,
+      "loss": 5.2427,
+      "step": 906
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.03000828007659e-05,
+      "loss": 5.3023,
+      "step": 907
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.009655443502767e-05,
+      "loss": 5.1946,
+      "step": 908
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.989311188167951e-05,
+      "loss": 5.12,
+      "step": 909
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.968975601785073e-05,
+      "loss": 5.0437,
+      "step": 910
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.948648772029655e-05,
+      "loss": 5.1999,
+      "step": 911
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.928330786539485e-05,
+      "loss": 5.083,
+      "step": 912
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.908021732914215e-05,
+      "loss": 4.9265,
+      "step": 913
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.88772169871499e-05,
+      "loss": 5.0785,
+      "step": 914
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.86743077146406e-05,
+      "loss": 4.8797,
+      "step": 915
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.84714903864442e-05,
+      "loss": 5.0303,
+      "step": 916
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.82687658769942e-05,
+      "loss": 4.8702,
+      "step": 917
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.806613506032395e-05,
+      "loss": 4.7571,
+      "step": 918
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.786359881006277e-05,
+      "loss": 4.8751,
+      "step": 919
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.766115799943241e-05,
+      "loss": 4.7779,
+      "step": 920
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.745881350124296e-05,
+      "loss": 4.9918,
+      "step": 921
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.725656618788937e-05,
+      "loss": 4.988,
+      "step": 922
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.705441693134757e-05,
+      "loss": 4.848,
+      "step": 923
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.685236660317074e-05,
+      "loss": 4.799,
+      "step": 924
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.665041607448553e-05,
+      "loss": 4.6384,
+      "step": 925
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.64485662159882e-05,
+      "loss": 4.6305,
+      "step": 926
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.624681789794119e-05,
+      "loss": 4.8133,
+      "step": 927
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.604517199016898e-05,
+      "loss": 5.1304,
+      "step": 928
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.584362936205463e-05,
+      "loss": 4.9423,
+      "step": 929
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.564219088253574e-05,
+      "loss": 4.456,
+      "step": 930
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.544085742010114e-05,
+      "loss": 4.7106,
+      "step": 931
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.523962984278668e-05,
+      "loss": 4.9547,
+      "step": 932
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.503850901817176e-05,
+      "loss": 4.7379,
+      "step": 933
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.483749581337552e-05,
+      "loss": 4.779,
+      "step": 934
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.463659109505315e-05,
+      "loss": 4.6419,
+      "step": 935
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.443579572939203e-05,
+      "loss": 4.6333,
+      "step": 936
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.42351105821081e-05,
+      "loss": 4.5173,
+      "step": 937
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.403453651844214e-05,
+      "loss": 4.6399,
+      "step": 938
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.383407440315596e-05,
+      "loss": 4.8384,
+      "step": 939
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.363372510052871e-05,
+      "loss": 4.7277,
+      "step": 940
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.343348947435314e-05,
+      "loss": 4.7908,
+      "step": 941
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.3233368387932e-05,
+      "loss": 4.7378,
+      "step": 942
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.303336270407405e-05,
+      "loss": 4.5516,
+      "step": 943
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.283347328509061e-05,
+      "loss": 4.5153,
+      "step": 944
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.263370099279172e-05,
+      "loss": 4.5053,
+      "step": 945
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.243404668848239e-05,
+      "loss": 4.4583,
+      "step": 946
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.223451123295897e-05,
+      "loss": 4.3878,
+      "step": 947
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.203509548650534e-05,
+      "loss": 4.7517,
+      "step": 948
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.183580030888938e-05,
+      "loss": 4.5506,
+      "step": 949
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.163662655935906e-05,
+      "loss": 4.6667,
+      "step": 950
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.14375750966388e-05,
+      "loss": 4.4696,
+      "step": 951
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.123864677892587e-05,
+      "loss": 4.3273,
+      "step": 952
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.103984246388657e-05,
+      "loss": 4.5906,
+      "step": 953
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.084116300865254e-05,
+      "loss": 4.705,
+      "step": 954
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.06426092698171e-05,
+      "loss": 4.538,
+      "step": 955
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.04441821034316e-05,
+      "loss": 4.5403,
+      "step": 956
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.024588236500169e-05,
+      "loss": 4.5133,
+      "step": 957
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.004771090948352e-05,
+      "loss": 4.5308,
+      "step": 958
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.984966859128024e-05,
+      "loss": 4.3205,
+      "step": 959
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.965175626423825e-05,
+      "loss": 4.3989,
+      "step": 960
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.945397478164344e-05,
+      "loss": 4.5382,
+      "step": 961
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.925632499621759e-05,
+      "loss": 4.783,
+      "step": 962
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.905880776011475e-05,
+      "loss": 4.4458,
+      "step": 963
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.886142392491736e-05,
+      "loss": 4.3811,
+      "step": 964
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.866417434163282e-05,
+      "loss": 4.6029,
+      "step": 965
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.846705986068964e-05,
+      "loss": 4.5219,
+      "step": 966
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.827008133193391e-05,
+      "loss": 4.5445,
+      "step": 967
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.807323960462555e-05,
+      "loss": 4.7661,
+      "step": 968
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.787653552743456e-05,
+      "loss": 4.2728,
+      "step": 969
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.76799699484377e-05,
+      "loss": 4.4742,
+      "step": 970
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.748354371511441e-05,
+      "loss": 4.3505,
+      "step": 971
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.728725767434345e-05,
+      "loss": 4.4343,
+      "step": 972
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.7091112672399e-05,
+      "loss": 4.3051,
+      "step": 973
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.689510955494742e-05,
+      "loss": 4.4094,
+      "step": 974
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.669924916704311e-05,
+      "loss": 4.3454,
+      "step": 975
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.65035323531252e-05,
+      "loss": 4.7379,
+      "step": 976
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.630795995701381e-05,
+      "loss": 4.4643,
+      "step": 977
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.611253282190641e-05,
+      "loss": 4.2021,
+      "step": 978
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.591725179037418e-05,
+      "loss": 4.5317,
+      "step": 979
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.572211770435835e-05,
+      "loss": 4.1355,
+      "step": 980
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.552713140516664e-05,
+      "loss": 4.3072,
+      "step": 981
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.533229373346963e-05,
+      "loss": 4.6673,
+      "step": 982
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.513760552929706e-05,
+      "loss": 4.5009,
+      "step": 983
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.494306763203416e-05,
+      "loss": 4.4305,
+      "step": 984
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.474868088041837e-05,
+      "loss": 4.5538,
+      "step": 985
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.455444611253521e-05,
+      "loss": 4.6243,
+      "step": 986
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.436036416581508e-05,
+      "loss": 4.4192,
+      "step": 987
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.416643587702948e-05,
+      "loss": 4.2811,
+      "step": 988
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.397266208228742e-05,
+      "loss": 4.3847,
+      "step": 989
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.377904361703178e-05,
+      "loss": 4.4137,
+      "step": 990
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.358558131603575e-05,
+      "loss": 4.3813,
+      "step": 991
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.339227601339932e-05,
+      "loss": 3.9793,
+      "step": 992
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.319912854254548e-05,
+      "loss": 4.3141,
+      "step": 993
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.30061397362168e-05,
+      "loss": 4.2746,
+      "step": 994
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.281331042647178e-05,
+      "loss": 4.4593,
+      "step": 995
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.262064144468127e-05,
+      "loss": 4.3952,
+      "step": 996
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.242813362152478e-05,
+      "loss": 4.5454,
+      "step": 997
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.223578778698708e-05,
+      "loss": 4.605,
+      "step": 998
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.204360477035458e-05,
+      "loss": 4.6417,
+      "step": 999
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.185158540021164e-05,
+      "loss": 4.1553,
+      "step": 1000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.165973050443702e-05,
+      "loss": 4.2343,
+      "step": 1001
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.146804091020047e-05,
+      "loss": 4.6799,
+      "step": 1002
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.1276517443959e-05,
+      "loss": 4.2626,
+      "step": 1003
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.10851609314534e-05,
+      "loss": 4.4228,
+      "step": 1004
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.0893972197704585e-05,
+      "loss": 4.2244,
+      "step": 1005
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.070295206701022e-05,
+      "loss": 4.3773,
+      "step": 1006
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.051210136294089e-05,
+      "loss": 4.3031,
+      "step": 1007
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.032142090833683e-05,
+      "loss": 4.6857,
+      "step": 1008
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.013091152530419e-05,
+      "loss": 4.4934,
+      "step": 1009
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9940574035211636e-05,
+      "loss": 4.679,
+      "step": 1010
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.975040925868667e-05,
+      "loss": 4.4058,
+      "step": 1011
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.956041801561207e-05,
+      "loss": 4.2721,
+      "step": 1012
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.937060112512267e-05,
+      "loss": 4.2014,
+      "step": 1013
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.918095940560133e-05,
+      "loss": 4.2335,
+      "step": 1014
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.899149367467587e-05,
+      "loss": 4.3446,
+      "step": 1015
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.880220474921516e-05,
+      "loss": 4.3621,
+      "step": 1016
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.861309344532602e-05,
+      "loss": 4.5432,
+      "step": 1017
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.842416057834925e-05,
+      "loss": 4.2953,
+      "step": 1018
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.8235406962856434e-05,
+      "loss": 4.2915,
+      "step": 1019
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.804683341264638e-05,
+      "loss": 4.6229,
+      "step": 1020
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.78584407407414e-05,
+      "loss": 4.2636,
+      "step": 1021
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.76702297593841e-05,
+      "loss": 4.085,
+      "step": 1022
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.748220128003361e-05,
+      "loss": 4.4141,
+      "step": 1023
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.729435611336239e-05,
+      "loss": 4.2933,
+      "step": 1024
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.710669506925247e-05,
+      "loss": 4.1267,
+      "step": 1025
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6919218956791956e-05,
+      "loss": 3.9119,
+      "step": 1026
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.67319285842717e-05,
+      "loss": 4.1521,
+      "step": 1027
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.654482475918187e-05,
+      "loss": 4.4233,
+      "step": 1028
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.635790828820822e-05,
+      "loss": 4.2104,
+      "step": 1029
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.617117997722866e-05,
+      "loss": 4.2418,
+      "step": 1030
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.598464063131008e-05,
+      "loss": 4.4702,
+      "step": 1031
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.579829105470449e-05,
+      "loss": 4.0894,
+      "step": 1032
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.5612132050845765e-05,
+      "loss": 4.3346,
+      "step": 1033
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.542616442234618e-05,
+      "loss": 4.1152,
+      "step": 1034
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.524038897099285e-05,
+      "loss": 4.5168,
+      "step": 1035
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.505480649774437e-05,
+      "loss": 4.3579,
+      "step": 1036
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.4869417802727297e-05,
+      "loss": 4.1647,
+      "step": 1037
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.468422368523274e-05,
+      "loss": 4.1308,
+      "step": 1038
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.449922494371289e-05,
+      "loss": 3.9377,
+      "step": 1039
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.431442237577762e-05,
+      "loss": 3.9156,
+      "step": 1040
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.4129816778190936e-05,
+      "loss": 3.9804,
+      "step": 1041
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.394540894686777e-05,
+      "loss": 4.0973,
+      "step": 1042
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.37611996768702e-05,
+      "loss": 3.9278,
+      "step": 1043
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.357718976240432e-05,
+      "loss": 4.3412,
+      "step": 1044
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.33933799968168e-05,
+      "loss": 4.4828,
+      "step": 1045
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.320977117259127e-05,
+      "loss": 4.0695,
+      "step": 1046
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.302636408134499e-05,
+      "loss": 4.0163,
+      "step": 1047
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.2843159513825504e-05,
+      "loss": 3.9765,
+      "step": 1048
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.2660158259907275e-05,
+      "loss": 4.4086,
+      "step": 1049
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.247736110858807e-05,
+      "loss": 4.2488,
+      "step": 1050
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.229476884798578e-05,
+      "loss": 4.2214,
+      "step": 1051
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.2112382265334734e-05,
+      "loss": 4.2582,
+      "step": 1052
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.1930202146982766e-05,
+      "loss": 3.9493,
+      "step": 1053
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.174822927838736e-05,
+      "loss": 4.0882,
+      "step": 1054
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.15664644441125e-05,
+      "loss": 4.3226,
+      "step": 1055
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.138490842782531e-05,
+      "loss": 4.0226,
+      "step": 1056
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.1203562012292496e-05,
+      "loss": 3.8473,
+      "step": 1057
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.102242597937717e-05,
+      "loss": 4.6226,
+      "step": 1058
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.0841501110035374e-05,
+      "loss": 4.6324,
+      "step": 1059
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.066078818431271e-05,
+      "loss": 4.3309,
+      "step": 1060
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.048028798134102e-05,
+      "loss": 4.3762,
+      "step": 1061
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.0300001279334965e-05,
+      "loss": 4.2912,
+      "step": 1062
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.011992885558886e-05,
+      "loss": 4.3362,
+      "step": 1063
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9940071486472937e-05,
+      "loss": 3.9657,
+      "step": 1064
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.97604299474304e-05,
+      "loss": 4.0431,
+      "step": 1065
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.958100501297383e-05,
+      "loss": 4.0797,
+      "step": 1066
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.9401797456682066e-05,
+      "loss": 4.0275,
+      "step": 1067
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.9222808051196624e-05,
+      "loss": 3.9974,
+      "step": 1068
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.904403756821843e-05,
+      "loss": 4.0945,
+      "step": 1069
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.88654867785046e-05,
+      "loss": 4.1384,
+      "step": 1070
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.8687156451865145e-05,
+      "loss": 4.333,
+      "step": 1071
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.850904735715949e-05,
+      "loss": 4.1276,
+      "step": 1072
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.8331160262293096e-05,
+      "loss": 4.2125,
+      "step": 1073
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.815349593421453e-05,
+      "loss": 4.4889,
+      "step": 1074
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.797605513891179e-05,
+      "loss": 4.5545,
+      "step": 1075
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7798838641409104e-05,
+      "loss": 4.0797,
+      "step": 1076
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.7621847205763704e-05,
+      "loss": 3.9371,
+      "step": 1077
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.744508159506247e-05,
+      "loss": 4.3169,
+      "step": 1078
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.726854257141865e-05,
+      "loss": 4.6066,
+      "step": 1079
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.709223089596857e-05,
+      "loss": 4.1596,
+      "step": 1080
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.691614732886837e-05,
+      "loss": 4.1358,
+      "step": 1081
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.674029262929073e-05,
+      "loss": 4.2432,
+      "step": 1082
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.6564667555421534e-05,
+      "loss": 4.4686,
+      "step": 1083
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.6389272864456655e-05,
+      "loss": 4.4383,
+      "step": 1084
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.6214109312598844e-05,
+      "loss": 4.2541,
+      "step": 1085
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.6039177655054036e-05,
+      "loss": 4.1354,
+      "step": 1086
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.586447864602853e-05,
+      "loss": 4.2534,
+      "step": 1087
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.569001303872562e-05,
+      "loss": 4.105,
+      "step": 1088
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.551578158534226e-05,
+      "loss": 4.4312,
+      "step": 1089
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.5341785037065754e-05,
+      "loss": 4.6001,
+      "step": 1090
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.5168024144070734e-05,
+      "loss": 4.1275,
+      "step": 1091
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4994499655515865e-05,
+      "loss": 4.1648,
+      "step": 1092
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.482121231954049e-05,
+      "loss": 4.2065,
+      "step": 1093
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.4648162883261536e-05,
+      "loss": 4.4431,
+      "step": 1094
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.4475352092770104e-05,
+      "loss": 4.1755,
+      "step": 1095
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.4302780693128596e-05,
+      "loss": 4.3427,
+      "step": 1096
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.413044942836717e-05,
+      "loss": 4.2142,
+      "step": 1097
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.395835904148069e-05,
+      "loss": 4.2044,
+      "step": 1098
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.3786510274425474e-05,
+      "loss": 3.891,
+      "step": 1099
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.361490386811613e-05,
+      "loss": 4.5586,
+      "step": 1100
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.344354056242232e-05,
+      "loss": 4.3173,
+      "step": 1101
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.327242109616565e-05,
+      "loss": 4.4929,
+      "step": 1102
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.310154620711635e-05,
+      "loss": 4.1077,
+      "step": 1103
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.293091663199024e-05,
+      "loss": 3.7611,
+      "step": 1104
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.27605331064454e-05,
+      "loss": 3.9731,
+      "step": 1105
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.259039636507926e-05,
+      "loss": 3.8743,
+      "step": 1106
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.2420507141425016e-05,
+      "loss": 4.1652,
+      "step": 1107
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.225086616794888e-05,
+      "loss": 4.2562,
+      "step": 1108
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.2081474176046646e-05,
+      "loss": 4.0944,
+      "step": 1109
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.191233189604078e-05,
+      "loss": 4.2361,
+      "step": 1110
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.174344005717703e-05,
+      "loss": 4.1533,
+      "step": 1111
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.15747993876213e-05,
+      "loss": 4.1922,
+      "step": 1112
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.14064106144568e-05,
+      "loss": 4.2253,
+      "step": 1113
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.123827446368055e-05,
+      "loss": 3.9697,
+      "step": 1114
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.107039166020049e-05,
+      "loss": 3.905,
+      "step": 1115
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.090276292783214e-05,
+      "loss": 4.2954,
+      "step": 1116
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.073538898929579e-05,
+      "loss": 4.1561,
+      "step": 1117
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.0568270566213086e-05,
+      "loss": 4.3843,
+      "step": 1118
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.040140837910407e-05,
+      "loss": 4.5443,
+      "step": 1119
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.0234803147384015e-05,
+      "loss": 4.3035,
+      "step": 1120
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.006845558936039e-05,
+      "loss": 4.1945,
+      "step": 1121
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.990236642222969e-05,
+      "loss": 4.0099,
+      "step": 1122
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.973653636207437e-05,
+      "loss": 4.1149,
+      "step": 1123
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.9570966123859796e-05,
+      "loss": 4.2051,
+      "step": 1124
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.940565642143109e-05,
+      "loss": 4.3868,
+      "step": 1125
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.924060796751012e-05,
+      "loss": 4.2942,
+      "step": 1126
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.907582147369234e-05,
+      "loss": 3.8923,
+      "step": 1127
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.891129765044396e-05,
+      "loss": 4.5074,
+      "step": 1128
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8747037207098425e-05,
+      "loss": 4.2157,
+      "step": 1129
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.8583040851853814e-05,
+      "loss": 4.165,
+      "step": 1130
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.8419309291769624e-05,
+      "loss": 4.1808,
+      "step": 1131
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.825584323276367e-05,
+      "loss": 4.5356,
+      "step": 1132
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.809264337960898e-05,
+      "loss": 4.1523,
+      "step": 1133
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.792971043593095e-05,
+      "loss": 3.9827,
+      "step": 1134
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.776704510420426e-05,
+      "loss": 4.4603,
+      "step": 1135
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.760464808574969e-05,
+      "loss": 3.8881,
+      "step": 1136
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.744252008073127e-05,
+      "loss": 3.8104,
+      "step": 1137
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.7280661788153184e-05,
+      "loss": 4.2503,
+      "step": 1138
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.711907390585674e-05,
+      "loss": 4.4002,
+      "step": 1139
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.695775713051743e-05,
+      "loss": 4.3407,
+      "step": 1140
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.679671215764186e-05,
+      "loss": 3.8113,
+      "step": 1141
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.66359396815648e-05,
+      "loss": 4.1197,
+      "step": 1142
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.647544039544615e-05,
+      "loss": 3.9943,
+      "step": 1143
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.631521499126797e-05,
+      "loss": 4.0301,
+      "step": 1144
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.61552641598315e-05,
+      "loss": 4.1275,
+      "step": 1145
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.599558859075419e-05,
+      "loss": 3.8279,
+      "step": 1146
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.58361889724667e-05,
+      "loss": 4.229,
+      "step": 1147
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.567706599220991e-05,
+      "loss": 3.9386,
+      "step": 1148
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.5518220336032136e-05,
+      "loss": 3.8372,
+      "step": 1149
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.5359652688785815e-05,
+      "loss": 3.9506,
+      "step": 1150
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.5201363734124936e-05,
+      "loss": 3.9758,
+      "step": 1151
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.5043354154501816e-05,
+      "loss": 3.9625,
+      "step": 1152
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4885624631164394e-05,
+      "loss": 4.1781,
+      "step": 1153
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4728175844153065e-05,
+      "loss": 4.0935,
+      "step": 1154
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.4571008472297786e-05,
+      "loss": 4.1654,
+      "step": 1155
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.441412319321539e-05,
+      "loss": 4.2227,
+      "step": 1156
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.4257520683306354e-05,
+      "loss": 4.1956,
+      "step": 1157
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.4101201617752086e-05,
+      "loss": 4.0562,
+      "step": 1158
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.39451666705118e-05,
+      "loss": 3.7323,
+      "step": 1159
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.378941651431996e-05,
+      "loss": 4.074,
+      "step": 1160
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.363395182068305e-05,
+      "loss": 4.0819,
+      "step": 1161
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.3478773259876805e-05,
+      "loss": 4.104,
+      "step": 1162
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.3323881500943345e-05,
+      "loss": 4.0208,
+      "step": 1163
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.3169277211688256e-05,
+      "loss": 3.6551,
+      "step": 1164
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.3014961058677694e-05,
+      "loss": 4.0149,
+      "step": 1165
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.286093370723557e-05,
+      "loss": 4.0859,
+      "step": 1166
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.270719582144064e-05,
+      "loss": 4.1167,
+      "step": 1167
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.255374806412361e-05,
+      "loss": 4.1787,
+      "step": 1168
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.2400591096864375e-05,
+      "loss": 4.1422,
+      "step": 1169
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.224772557998902e-05,
+      "loss": 4.0303,
+      "step": 1170
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.209515217256722e-05,
+      "loss": 4.1838,
+      "step": 1171
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.194287153240906e-05,
+      "loss": 4.2742,
+      "step": 1172
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.179088431606243e-05,
+      "loss": 3.848,
+      "step": 1173
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.163919117881024e-05,
+      "loss": 3.8024,
+      "step": 1174
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1487792774667445e-05,
+      "loss": 4.0002,
+      "step": 1175
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.133668975637818e-05,
+      "loss": 3.9813,
+      "step": 1176
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.118588277541312e-05,
+      "loss": 4.0288,
+      "step": 1177
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.103537248196666e-05,
+      "loss": 4.1978,
+      "step": 1178
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0885159524953956e-05,
+      "loss": 4.0508,
+      "step": 1179
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0735244552008245e-05,
+      "loss": 4.4225,
+      "step": 1180
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.058562820947802e-05,
+      "loss": 4.0303,
+      "step": 1181
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.043631114242427e-05,
+      "loss": 4.1941,
+      "step": 1182
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.028729399461764e-05,
+      "loss": 3.9692,
+      "step": 1183
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.0138577408535717e-05,
+      "loss": 4.1606,
+      "step": 1184
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9990162025360247e-05,
+      "loss": 4.2109,
+      "step": 1185
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.984204848497434e-05,
+      "loss": 4.0562,
+      "step": 1186
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9694237425959736e-05,
+      "loss": 3.8164,
+      "step": 1187
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.954672948559405e-05,
+      "loss": 4.2995,
+      "step": 1188
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9399525299848007e-05,
+      "loss": 3.7279,
+      "step": 1189
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9252625503382736e-05,
+      "loss": 4.0042,
+      "step": 1190
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.910603072954696e-05,
+      "loss": 3.9864,
+      "step": 1191
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8959741610374468e-05,
+      "loss": 3.9577,
+      "step": 1192
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8813758776581023e-05,
+      "loss": 4.2448,
+      "step": 1193
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8668082857562005e-05,
+      "loss": 3.954,
+      "step": 1194
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8522714481389467e-05,
+      "loss": 3.9794,
+      "step": 1195
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.837765427480963e-05,
+      "loss": 4.0199,
+      "step": 1196
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8232902863239985e-05,
+      "loss": 4.128,
+      "step": 1197
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8088460870766574e-05,
+      "loss": 4.0033,
+      "step": 1198
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7944328920141617e-05,
+      "loss": 4.0961,
+      "step": 1199
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7800507632780437e-05,
+      "loss": 4.302,
+      "step": 1200
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.765699762875905e-05,
+      "loss": 3.9951,
+      "step": 1201
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.751379952681128e-05,
+      "loss": 3.8726,
+      "step": 1202
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.7370913944326338e-05,
+      "loss": 3.781,
+      "step": 1203
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.7228341497345956e-05,
+      "loss": 3.6683,
+      "step": 1204
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.708608280056183e-05,
+      "loss": 4.1482,
+      "step": 1205
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6944138467312907e-05,
+      "loss": 4.3113,
+      "step": 1206
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6802509109582808e-05,
+      "loss": 4.1557,
+      "step": 1207
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.666119533799715e-05,
+      "loss": 4.0159,
+      "step": 1208
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.652019776182092e-05,
+      "loss": 3.9427,
+      "step": 1209
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.6379516988955843e-05,
+      "loss": 4.0831,
+      "step": 1210
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.6239153625937784e-05,
+      "loss": 3.8976,
+      "step": 1211
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.6099108277934103e-05,
+      "loss": 4.1061,
+      "step": 1212
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5959381548741034e-05,
+      "loss": 3.6877,
+      "step": 1213
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5819974040781213e-05,
+      "loss": 4.1066,
+      "step": 1214
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.5680886355100818e-05,
+      "loss": 4.0678,
+      "step": 1215
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.5542119091367224e-05,
+      "loss": 4.2355,
+      "step": 1216
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.5403672847866355e-05,
+      "loss": 3.8093,
+      "step": 1217
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.5265548221500046e-05,
+      "loss": 4.2428,
+      "step": 1218
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.5127745807783443e-05,
+      "loss": 3.9763,
+      "step": 1219
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4990266200842548e-05,
+      "loss": 3.8709,
+      "step": 1220
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.485310999341166e-05,
+      "loss": 4.0597,
+      "step": 1221
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4716277776830653e-05,
+      "loss": 4.065,
+      "step": 1222
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4579770141042613e-05,
+      "loss": 3.876,
+      "step": 1223
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.4443587674591173e-05,
+      "loss": 4.0177,
+      "step": 1224
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.4307730964618047e-05,
+      "loss": 3.9866,
+      "step": 1225
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.417220059686045e-05,
+      "loss": 3.8171,
+      "step": 1226
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.403699715564861e-05,
+      "loss": 4.3799,
+      "step": 1227
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.390212122390323e-05,
+      "loss": 4.0668,
+      "step": 1228
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.376757338313298e-05,
+      "loss": 3.9493,
+      "step": 1229
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.3633354213431947e-05,
+      "loss": 4.1677,
+      "step": 1230
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.34994642934773e-05,
+      "loss": 3.6931,
+      "step": 1231
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.33659042005265e-05,
+      "loss": 4.2559,
+      "step": 1232
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.3232674510415086e-05,
+      "loss": 4.463,
+      "step": 1233
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.3099775797554023e-05,
+      "loss": 4.2102,
+      "step": 1234
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2967208634927428e-05,
+      "loss": 4.1366,
+      "step": 1235
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.283497359408977e-05,
+      "loss": 3.9068,
+      "step": 1236
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2703071245163722e-05,
+      "loss": 4.1732,
+      "step": 1237
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.257150215683751e-05,
+      "loss": 3.9909,
+      "step": 1238
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2440266896362618e-05,
+      "loss": 3.8523,
+      "step": 1239
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.2309366029551195e-05,
+      "loss": 4.1886,
+      "step": 1240
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.2178800120773602e-05,
+      "loss": 3.9774,
+      "step": 1241
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.2048569732956213e-05,
+      "loss": 3.8079,
+      "step": 1242
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1918675427578694e-05,
+      "loss": 3.9253,
+      "step": 1243
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1789117764671808e-05,
+      "loss": 3.5738,
+      "step": 1244
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1659897302814747e-05,
+      "loss": 4.1309,
+      "step": 1245
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.1531014599133105e-05,
+      "loss": 3.8271,
+      "step": 1246
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.1402470209296087e-05,
+      "loss": 4.0998,
+      "step": 1247
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.1274264687514345e-05,
+      "loss": 3.9867,
+      "step": 1248
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.1146398586537518e-05,
+      "loss": 4.1927,
+      "step": 1249
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.101887245765185e-05,
+      "loss": 3.923,
+      "step": 1250
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0891686850677805e-05,
+      "loss": 3.9116,
+      "step": 1251
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0764842313967737e-05,
+      "loss": 4.043,
+      "step": 1252
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0638339394403472e-05,
+      "loss": 4.13,
+      "step": 1253
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0512178637393997e-05,
+      "loss": 4.0342,
+      "step": 1254
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.0386360586873065e-05,
+      "loss": 3.8992,
+      "step": 1255
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.026088578529687e-05,
+      "loss": 3.9232,
+      "step": 1256
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.0135754773641803e-05,
+      "loss": 3.6286,
+      "step": 1257
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.0010968091401873e-05,
+      "loss": 3.8714,
+      "step": 1258
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9886526276586603e-05,
+      "loss": 3.9394,
+      "step": 1259
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9762429865718723e-05,
+      "loss": 4.0907,
+      "step": 1260
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.9638679393831694e-05,
+      "loss": 3.6827,
+      "step": 1261
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.9515275394467446e-05,
+      "loss": 3.8492,
+      "step": 1262
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.9392218399674176e-05,
+      "loss": 3.9574,
+      "step": 1263
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.9269508940004033e-05,
+      "loss": 4.0074,
+      "step": 1264
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.9147147544510712e-05,
+      "loss": 3.812,
+      "step": 1265
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.9025134740747285e-05,
+      "loss": 3.9458,
+      "step": 1266
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.89034710547639e-05,
+      "loss": 3.7145,
+      "step": 1267
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8782157011105473e-05,
+      "loss": 4.0852,
+      "step": 1268
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.866119313280952e-05,
+      "loss": 4.3483,
+      "step": 1269
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.854057994140378e-05,
+      "loss": 4.0049,
+      "step": 1270
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.8420317956904064e-05,
+      "loss": 3.9619,
+      "step": 1271
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.8300407697811984e-05,
+      "loss": 3.9218,
+      "step": 1272
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.8180849681112665e-05,
+      "loss": 4.3077,
+      "step": 1273
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.806164442227266e-05,
+      "loss": 4.2602,
+      "step": 1274
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7942792435237497e-05,
+      "loss": 3.7119,
+      "step": 1275
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7824294232429685e-05,
+      "loss": 3.7763,
+      "step": 1276
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7706150324746374e-05,
+      "loss": 3.8519,
+      "step": 1277
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7588361221557282e-05,
+      "loss": 3.6636,
+      "step": 1278
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7470927430702277e-05,
+      "loss": 3.6719,
+      "step": 1279
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.73538494584894e-05,
+      "loss": 4.1347,
+      "step": 1280
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.723712780969259e-05,
+      "loss": 3.4995,
+      "step": 1281
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7120762987549543e-05,
+      "loss": 3.9382,
+      "step": 1282
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7004755493759507e-05,
+      "loss": 4.0893,
+      "step": 1283
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6889105828481067e-05,
+      "loss": 3.9955,
+      "step": 1284
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6773814490330165e-05,
+      "loss": 3.9834,
+      "step": 1285
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.665888197637777e-05,
+      "loss": 3.8012,
+      "step": 1286
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.6544308782147844e-05,
+      "loss": 3.8928,
+      "step": 1287
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.6430095401615065e-05,
+      "loss": 4.0349,
+      "step": 1288
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.631624232720296e-05,
+      "loss": 4.1261,
+      "step": 1289
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.6202750049781502e-05,
+      "loss": 3.6318,
+      "step": 1290
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.608961905866514e-05,
+      "loss": 3.9984,
+      "step": 1291
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5976849841610676e-05,
+      "loss": 4.3935,
+      "step": 1292
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.586444288481512e-05,
+      "loss": 4.0717,
+      "step": 1293
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.575239867291366e-05,
+      "loss": 3.8665,
+      "step": 1294
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.564071768897749e-05,
+      "loss": 3.7788,
+      "step": 1295
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.5529400414511806e-05,
+      "loss": 3.7291,
+      "step": 1296
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.5418447329453678e-05,
+      "loss": 4.0615,
+      "step": 1297
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.530785891216997e-05,
+      "loss": 3.8269,
+      "step": 1298
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.5197635639455376e-05,
+      "loss": 3.8863,
+      "step": 1299
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.5087777986530272e-05,
+      "loss": 4.0178,
+      "step": 1300
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4978286427038601e-05,
+      "loss": 3.8778,
+      "step": 1301
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4869161433045997e-05,
+      "loss": 4.0489,
+      "step": 1302
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4760403475037698e-05,
+      "loss": 4.2177,
+      "step": 1303
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4652013021916466e-05,
+      "loss": 3.795,
+      "step": 1304
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.4543990541000519e-05,
+      "loss": 4.1198,
+      "step": 1305
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.443633649802164e-05,
+      "loss": 3.937,
+      "step": 1306
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.4329051357123179e-05,
+      "loss": 4.0462,
+      "step": 1307
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.4222135580857898e-05,
+      "loss": 4.0581,
+      "step": 1308
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.411558963018611e-05,
+      "loss": 4.0708,
+      "step": 1309
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.4009413964473627e-05,
+      "loss": 3.807,
+      "step": 1310
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3903609041489828e-05,
+      "loss": 4.0684,
+      "step": 1311
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3798175317405637e-05,
+      "loss": 4.1014,
+      "step": 1312
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3693113246791589e-05,
+      "loss": 3.8529,
+      "step": 1313
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.358842328261587e-05,
+      "loss": 3.9999,
+      "step": 1314
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.3484105876242315e-05,
+      "loss": 4.1461,
+      "step": 1315
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.3380161477428533e-05,
+      "loss": 3.8065,
+      "step": 1316
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.3276590534323974e-05,
+      "loss": 3.4906,
+      "step": 1317
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.3173393493467867e-05,
+      "loss": 3.8956,
+      "step": 1318
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.3070570799787441e-05,
+      "loss": 3.8491,
+      "step": 1319
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2968122896595936e-05,
+      "loss": 3.9298,
+      "step": 1320
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2866050225590765e-05,
+      "loss": 3.9558,
+      "step": 1321
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2764353226851455e-05,
+      "loss": 4.1914,
+      "step": 1322
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2663032338837876e-05,
+      "loss": 3.8689,
+      "step": 1323
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2562087998388405e-05,
+      "loss": 4.127,
+      "step": 1324
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2461520640717862e-05,
+      "loss": 3.8386,
+      "step": 1325
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2361330699415786e-05,
+      "loss": 4.1869,
+      "step": 1326
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2261518606444411e-05,
+      "loss": 3.896,
+      "step": 1327
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2162084792137063e-05,
+      "loss": 3.9722,
+      "step": 1328
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2063029685196036e-05,
+      "loss": 3.8821,
+      "step": 1329
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.196435371269089e-05,
+      "loss": 3.5531,
+      "step": 1330
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1866057300056499e-05,
+      "loss": 3.7629,
+      "step": 1331
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1768140871091415e-05,
+      "loss": 4.2821,
+      "step": 1332
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1670604847955824e-05,
+      "loss": 3.9708,
+      "step": 1333
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1573449651169854e-05,
+      "loss": 3.8452,
+      "step": 1334
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1476675699611706e-05,
+      "loss": 3.4909,
+      "step": 1335
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.138028341051588e-05,
+      "loss": 3.8492,
+      "step": 1336
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.128427319947135e-05,
+      "loss": 3.956,
+      "step": 1337
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1188645480419813e-05,
+      "loss": 3.9421,
+      "step": 1338
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1093400665653853e-05,
+      "loss": 3.9466,
+      "step": 1339
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0998539165815192e-05,
+      "loss": 4.4935,
+      "step": 1340
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0904061389892894e-05,
+      "loss": 3.6112,
+      "step": 1341
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0809967745221694e-05,
+      "loss": 4.2189,
+      "step": 1342
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0716258637480114e-05,
+      "loss": 4.13,
+      "step": 1343
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.062293447068875e-05,
+      "loss": 3.7577,
+      "step": 1344
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0529995647208568e-05,
+      "loss": 3.9787,
+      "step": 1345
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.043744256773922e-05,
+      "loss": 4.2481,
+      "step": 1346
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.0345275631317163e-05,
+      "loss": 4.039,
+      "step": 1347
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.0253495235314037e-05,
+      "loss": 3.8478,
+      "step": 1348
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.016210177543494e-05,
+      "loss": 3.9643,
+      "step": 1349
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.0071095645716787e-05,
+      "loss": 3.8011,
+      "step": 1350
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.980477238526464e-06,
+      "loss": 3.8001,
+      "step": 1351
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.890246944559245e-06,
+      "loss": 4.1122,
+      "step": 1352
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.800405152837077e-06,
+      "loss": 3.8393,
+      "step": 1353
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.710952250706907e-06,
+      "loss": 4.1849,
+      "step": 1354
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.621888623839014e-06,
+      "loss": 3.6933,
+      "step": 1355
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.533214656225342e-06,
+      "loss": 3.8442,
+      "step": 1356
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.444930730177837e-06,
+      "loss": 3.7134,
+      "step": 1357
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 9.357037226326815e-06,
+      "loss": 3.8689,
+      "step": 1358
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 9.269534523619294e-06,
+      "loss": 4.1704,
+      "step": 1359
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 9.18242299931743e-06,
+      "loss": 4.0981,
+      "step": 1360
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 9.095703028996772e-06,
+      "loss": 4.1069,
+      "step": 1361
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 9.00937498654475e-06,
+      "loss": 4.181,
+      "step": 1362
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.923439244158994e-06,
+      "loss": 3.9549,
+      "step": 1363
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.837896172345827e-06,
+      "loss": 3.7659,
+      "step": 1364
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.7527461399185e-06,
+      "loss": 4.2103,
+      "step": 1365
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.667989513995756e-06,
+      "loss": 3.9445,
+      "step": 1366
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.583626660000222e-06,
+      "loss": 3.6077,
+      "step": 1367
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.499657941656736e-06,
+      "loss": 3.8851,
+      "step": 1368
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.416083720990908e-06,
+      "loss": 4.1025,
+      "step": 1369
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.33290435832742e-06,
+      "loss": 4.0549,
+      "step": 1370
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.250120212288626e-06,
+      "loss": 3.8152,
+      "step": 1371
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.167731639792907e-06,
+      "loss": 4.1079,
+      "step": 1372
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.08573899605316e-06,
+      "loss": 3.8152,
+      "step": 1373
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.004142634575185e-06,
+      "loss": 3.8927,
+      "step": 1374
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.92294290715636e-06,
+      "loss": 3.5557,
+      "step": 1375
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.842140163883938e-06,
+      "loss": 3.5754,
+      "step": 1376
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.761734753133599e-06,
+      "loss": 3.8007,
+      "step": 1377
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.681727021567953e-06,
+      "loss": 3.9299,
+      "step": 1378
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.602117314135049e-06,
+      "loss": 4.2086,
+      "step": 1379
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.5229059740668735e-06,
+      "loss": 3.9727,
+      "step": 1380
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.4440933428779e-06,
+      "loss": 3.7862,
+      "step": 1381
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.365679760363564e-06,
+      "loss": 3.7659,
+      "step": 1382
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.287665564598855e-06,
+      "loss": 4.1787,
+      "step": 1383
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.2100510919367985e-06,
+      "loss": 3.8642,
+      "step": 1384
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.1328366770071046e-06,
+      "loss": 3.9386,
+      "step": 1385
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.0560226527146115e-06,
+      "loss": 4.1387,
+      "step": 1386
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.979609350237881e-06,
+      "loss": 3.7437,
+      "step": 1387
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.903597099027803e-06,
+      "loss": 4.3164,
+      "step": 1388
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.827986226806238e-06,
+      "loss": 4.0767,
+      "step": 1389
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.75277705956443e-06,
+      "loss": 3.8551,
+      "step": 1390
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.677969921561711e-06,
+      "loss": 3.9759,
+      "step": 1391
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.603565135324152e-06,
+      "loss": 4.1621,
+      "step": 1392
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.5295630216430485e-06,
+      "loss": 3.9509,
+      "step": 1393
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.455963899573625e-06,
+      "loss": 3.9387,
+      "step": 1394
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.382768086433643e-06,
+      "loss": 4.1134,
+      "step": 1395
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.309975897802012e-06,
+      "loss": 3.7363,
+      "step": 1396
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.237587647517429e-06,
+      "loss": 4.1569,
+      "step": 1397
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.165603647677054e-06,
+      "loss": 3.9906,
+      "step": 1398
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.094024208635152e-06,
+      "loss": 4.2447,
+      "step": 1399
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.022849639001749e-06,
+      "loss": 4.4,
+      "step": 1400
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.952080245641278e-06,
+      "loss": 4.0072,
+      "step": 1401
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.881716333671306e-06,
+      "loss": 4.156,
+      "step": 1402
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.811758206461238e-06,
+      "loss": 4.0038,
+      "step": 1403
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.742206165630848e-06,
+      "loss": 3.9742,
+      "step": 1404
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.6730605110491884e-06,
+      "loss": 4.0928,
+      "step": 1405
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.604321540833168e-06,
+      "loss": 3.9146,
+      "step": 1406
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.535989551346365e-06,
+      "loss": 4.2789,
+      "step": 1407
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.468064837197584e-06,
+      "loss": 3.7722,
+      "step": 1408
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.400547691239732e-06,
+      "loss": 3.8211,
+      "step": 1409
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.33343840456857e-06,
+      "loss": 3.8449,
+      "step": 1410
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.266737266521315e-06,
+      "loss": 3.8688,
+      "step": 1411
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.200444564675555e-06,
+      "loss": 4.1986,
+      "step": 1412
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.134560584847825e-06,
+      "loss": 3.9103,
+      "step": 1413
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.069085611092617e-06,
+      "loss": 3.7727,
+      "step": 1414
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.00401992570092e-06,
+      "loss": 3.8498,
+      "step": 1415
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.939363809199138e-06,
+      "loss": 3.8619,
+      "step": 1416
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.875117540347862e-06,
+      "loss": 3.8201,
+      "step": 1417
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.81128139614061e-06,
+      "loss": 4.0612,
+      "step": 1418
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.747855651802724e-06,
+      "loss": 3.9681,
+      "step": 1419
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.684840580790084e-06,
+      "loss": 3.7536,
+      "step": 1420
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.62223645478802e-06,
+      "loss": 3.6761,
+      "step": 1421
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.56004354371008e-06,
+      "loss": 3.6899,
+      "step": 1422
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.498262115696894e-06,
+      "loss": 3.7502,
+      "step": 1423
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.436892437114993e-06,
+      "loss": 4.0006,
+      "step": 1424
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.375934772555701e-06,
+      "loss": 3.8935,
+      "step": 1425
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.315389384833956e-06,
+      "loss": 3.8104,
+      "step": 1426
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.2552565349871955e-06,
+      "loss": 3.6645,
+      "step": 1427
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.195536482274232e-06,
+      "loss": 3.7861,
+      "step": 1428
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.1362294841741365e-06,
+      "loss": 3.9048,
+      "step": 1429
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.077335796385096e-06,
+      "loss": 3.9792,
+      "step": 1430
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.0188556728233295e-06,
+      "loss": 4.1334,
+      "step": 1431
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.960789365622075e-06,
+      "loss": 3.7841,
+      "step": 1432
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.9031371251303624e-06,
+      "loss": 3.8421,
+      "step": 1433
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.845899199911984e-06,
+      "loss": 3.6136,
+      "step": 1434
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.7890758367445046e-06,
+      "loss": 3.9185,
+      "step": 1435
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.732667280618096e-06,
+      "loss": 4.1439,
+      "step": 1436
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.676673774734518e-06,
+      "loss": 3.9542,
+      "step": 1437
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.62109556050606e-06,
+      "loss": 3.9273,
+      "step": 1438
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.565932877554534e-06,
+      "loss": 3.9051,
+      "step": 1439
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.511185963710184e-06,
+      "loss": 3.9889,
+      "step": 1440
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.4568550550107328e-06,
+      "loss": 4.0087,
+      "step": 1441
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.402940385700293e-06,
+      "loss": 3.898,
+      "step": 1442
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.3494421882283912e-06,
+      "loss": 3.6996,
+      "step": 1443
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.296360693248979e-06,
+      "loss": 4.2735,
+      "step": 1444
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.2436961296193892e-06,
+      "loss": 3.9743,
+      "step": 1445
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.1914487243994597e-06,
+      "loss": 4.0263,
+      "step": 1446
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.1396187028503887e-06,
+      "loss": 3.6888,
+      "step": 1447
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.0882062884338815e-06,
+      "loss": 4.4145,
+      "step": 1448
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.0372117028111825e-06,
+      "loss": 3.7128,
+      "step": 1449
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.9866351658420777e-06,
+      "loss": 3.7647,
+      "step": 1450
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.9364768955839618e-06,
+      "loss": 3.8072,
+      "step": 1451
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.8867371082908823e-06,
+      "loss": 3.8483,
+      "step": 1452
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.8374160184126753e-06,
+      "loss": 3.8855,
+      "step": 1453
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.7885138385939425e-06,
+      "loss": 3.7742,
+      "step": 1454
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.7400307796732192e-06,
+      "loss": 3.8658,
+      "step": 1455
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6919670506819762e-06,
+      "loss": 4.2104,
+      "step": 1456
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6443228588438286e-06,
+      "loss": 3.8153,
+      "step": 1457
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.5970984095735507e-06,
+      "loss": 4.06,
+      "step": 1458
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.5502939064762533e-06,
+      "loss": 3.9406,
+      "step": 1459
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.5039095513464283e-06,
+      "loss": 4.1047,
+      "step": 1460
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.457945544167184e-06,
+      "loss": 3.8246,
+      "step": 1461
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4124020831092884e-06,
+      "loss": 4.0315,
+      "step": 1462
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.3672793645303615e-06,
+      "loss": 3.8531,
+      "step": 1463
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.322577582974028e-06,
+      "loss": 3.8602,
+      "step": 1464
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.278296931169055e-06,
+      "loss": 3.8807,
+      "step": 1465
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.2344376000285604e-06,
+      "loss": 3.7595,
+      "step": 1466
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.1909997786491496e-06,
+      "loss": 4.0731,
+      "step": 1467
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.147983654310104e-06,
+      "loss": 3.8268,
+      "step": 1468
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.105389412472614e-06,
+      "loss": 3.7524,
+      "step": 1469
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.063217236778936e-06,
+      "loss": 3.7444,
+      "step": 1470
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.021467309051617e-06,
+      "loss": 4.1415,
+      "step": 1471
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.9801398092927246e-06,
+      "loss": 3.8739,
+      "step": 1472
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.939234915683008e-06,
+      "loss": 3.8374,
+      "step": 1473
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.8987528045812387e-06,
+      "loss": 4.1716,
+      "step": 1474
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.8586936505233687e-06,
+      "loss": 3.7909,
+      "step": 1475
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.8190576262217874e-06,
+      "loss": 3.684,
+      "step": 1476
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.7798449025645758e-06,
+      "loss": 3.9253,
+      "step": 1477
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.74105564861482e-06,
+      "loss": 4.0456,
+      "step": 1478
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.7026900316098215e-06,
+      "loss": 3.8296,
+      "step": 1479
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.664748216960399e-06,
+      "loss": 3.9393,
+      "step": 1480
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.6272303682501878e-06,
+      "loss": 3.8387,
+      "step": 1481
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.5901366472348967e-06,
+      "loss": 3.9522,
+      "step": 1482
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.553467213841664e-06,
+      "loss": 3.5895,
+      "step": 1483
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.5172222261683355e-06,
+      "loss": 3.4149,
+      "step": 1484
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.4814018404827656e-06,
+      "loss": 4.0587,
+      "step": 1485
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.446006211222173e-06,
+      "loss": 4.0462,
+      "step": 1486
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.4110354909924962e-06,
+      "loss": 3.8541,
+      "step": 1487
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.3764898305676733e-06,
+      "loss": 3.8439,
+      "step": 1488
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.3423693788890413e-06,
+      "loss": 4.1222,
+      "step": 1489
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.3086742830646592e-06,
+      "loss": 3.4435,
+      "step": 1490
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.2754046883687198e-06,
+      "loss": 4.0476,
+      "step": 1491
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.242560738240872e-06,
+      "loss": 3.8462,
+      "step": 1492
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.2101425742856554e-06,
+      "loss": 4.2004,
+      "step": 1493
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.1781503362718104e-06,
+      "loss": 3.5463,
+      "step": 1494
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.1465841621317809e-06,
+      "loss": 4.213,
+      "step": 1495
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.1154441879610233e-06,
+      "loss": 3.9828,
+      "step": 1496
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0847305480174985e-06,
+      "loss": 3.6285,
+      "step": 1497
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0544433747210258e-06,
+      "loss": 4.0292,
+      "step": 1498
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0245827986527513e-06,
+      "loss": 3.6993,
+      "step": 1499
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 9.951489485545695e-07,
+      "loss": 3.8975,
+      "step": 1500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 9.661419513285808e-07,
+      "loss": 3.9521,
+      "step": 1501
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 9.375619320365459e-07,
+      "loss": 4.0065,
+      "step": 1502
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 9.094090138993205e-07,
+      "loss": 4.2081,
+      "step": 1503
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.816833182963558e-07,
+      "loss": 3.7562,
+      "step": 1504
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.543849647651536e-07,
+      "loss": 3.82,
+      "step": 1505
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.275140710007679e-07,
+      "loss": 4.1084,
+      "step": 1506
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.010707528552708e-07,
+      "loss": 3.6244,
+      "step": 1507
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.750551243372761e-07,
+      "loss": 4.1272,
+      "step": 1508
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.494672976114614e-07,
+      "loss": 4.3956,
+      "step": 1509
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.243073829980462e-07,
+      "loss": 4.0193,
+      "step": 1510
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.995754889723483e-07,
+      "loss": 3.8676,
+      "step": 1511
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.752717221642723e-07,
+      "loss": 4.4334,
+      "step": 1512
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.513961873579111e-07,
+      "loss": 3.9873,
+      "step": 1513
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.27948987491056e-07,
+      "loss": 4.3133,
+      "step": 1514
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.04930223654776e-07,
+      "loss": 3.8755,
+      "step": 1515
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.823399950929287e-07,
+      "loss": 3.9822,
+      "step": 1516
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.60178399201805e-07,
+      "loss": 4.0234,
+      "step": 1517
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.38445531529641e-07,
+      "loss": 3.9867,
+      "step": 1518
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.171414857762958e-07,
+      "loss": 3.9712,
+      "step": 1519
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.96266353792707e-07,
+      "loss": 4.066,
+      "step": 1520
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.7582022558066984e-07,
+      "loss": 3.8994,
+      "step": 1521
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.5580318929229203e-07,
+      "loss": 4.0342,
+      "step": 1522
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.3621533122970573e-07,
+      "loss": 3.9184,
+      "step": 1523
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.170567358446675e-07,
+      "loss": 3.889,
+      "step": 1524
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.9832748573820354e-07,
+      "loss": 3.7074,
+      "step": 1525
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.800276616602205e-07,
+      "loss": 3.4806,
+      "step": 1526
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.621573425092062e-07,
+      "loss": 4.0203,
+      "step": 1527
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.44716605331874e-07,
+      "loss": 3.9424,
+      "step": 1528
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.2770552532279677e-07,
+      "loss": 3.2905,
+      "step": 1529
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.11124175824129e-07,
+      "loss": 3.7946,
+      "step": 1530
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.9497262832525187e-07,
+      "loss": 3.9732,
+      "step": 1531
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.7925095246251755e-07,
+      "loss": 4.154,
+      "step": 1532
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.639592160188609e-07,
+      "loss": 3.9326,
+      "step": 1533
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.490974849236216e-07,
+      "loss": 3.7632,
+      "step": 1534
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.346658232521448e-07,
+      "loss": 3.642,
+      "step": 1535
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.2066429322561421e-07,
+      "loss": 3.9112,
+      "step": 1536
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0709295521068596e-07,
+      "loss": 3.9144,
+      "step": 1537
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.9395186771927754e-07,
+      "loss": 4.0351,
+      "step": 1538
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8124108740833478e-07,
+      "loss": 3.8774,
+      "step": 1539
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.689606690795431e-07,
+      "loss": 3.9526,
+      "step": 1540
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.571106656791166e-07,
+      "loss": 4.0964,
+      "step": 1541
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.4569112829755372e-07,
+      "loss": 4.0012,
+      "step": 1542
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.3470210616944867e-07,
+      "loss": 3.9106,
+      "step": 1543
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2414364667322487e-07,
+      "loss": 3.9834,
+      "step": 1544
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1401579533100171e-07,
+      "loss": 3.9015,
+      "step": 1545
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0431859580832814e-07,
+      "loss": 4.0445,
+      "step": 1546
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 9.505208991403835e-08,
+      "loss": 3.9171,
+      "step": 1547
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.62163176000741e-08,
+      "loss": 3.6761,
+      "step": 1548
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.781131696127375e-08,
+      "loss": 3.9284,
+      "step": 1549
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.983712423523913e-08,
+      "loss": 4.0142,
+      "step": 1550
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.229377380218005e-08,
+      "loss": 4.0513,
+      "step": 1551
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.518129818474771e-08,
+      "loss": 4.1492,
+      "step": 1552
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.84997280479238e-08,
+      "loss": 3.6455,
+      "step": 1553
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.2249092198842764e-08,
+      "loss": 4.2358,
+      "step": 1554
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.6429417586714145e-08,
+      "loss": 3.985,
+      "step": 1555
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.104072930268931e-08,
+      "loss": 3.5655,
+      "step": 1556
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.608305057971716e-08,
+      "loss": 3.8618,
+      "step": 1557
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.1556402792521914e-08,
+      "loss": 3.797,
+      "step": 1558
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.746080545744766e-08,
+      "loss": 4.0564,
+      "step": 1559
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.3796276232380667e-08,
+      "loss": 4.1757,
+      "step": 1560
+    },
+    {
+      "epoch": 4.99,
+      "step": 1560,
+      "total_flos": 1.808838425449595e+17,
+      "train_loss": 4.468929938780955,
+      "train_runtime": 20275.8964,
+      "train_samples_per_second": 1.233,
+      "train_steps_per_second": 0.077
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1560,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 50000,
+  "total_flos": 1.808838425449595e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}