Model save

README.md

@@ -0,0 +1,69 @@
+---
+license: apache-2.0
+base_model: mistralai/Mistral-7B-v0.1
+tags:
+- trl
+- sft
+- generated_from_trainer
+datasets:
+- generator
+model-index:
+- name: zephyr-7b-sft-full_epoch3
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# zephyr-7b-sft-full_epoch3
+
+This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the generator dataset.
+It achieves the following results on the evaluation set:
+- Loss: 3.0449
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 2e-05
+- train_batch_size: 4
+- eval_batch_size: 8
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 8
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 128
+- total_eval_batch_size: 64
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_ratio: 0.1
+- num_epochs: 3
+
+### Training results
+
+| Training Loss | Epoch | Step | Validation Loss |
+|:-------------:|:-----:|:----:|:---------------:|
+| 3.9689        | 1.0   | 1089 | 3.9833          |
+| 3.1562        | 2.0   | 2179 | 3.1740          |
+| 2.99          | 3.0   | 3267 | 3.0449          |
+
+
+### Framework versions
+
+- Transformers 4.36.2
+- Pytorch 2.1.2+cu121
+- Datasets 2.14.6
+- Tokenizers 0.15.1

all_results.json

@@ -0,0 +1,13 @@
+{
+    "epoch": 3.0,
+    "eval_loss": 3.044856548309326,
+    "eval_runtime": 298.6002,
+    "eval_samples": 23110,
+    "eval_samples_per_second": 51.678,
+    "eval_steps_per_second": 0.81,
+    "train_loss": 3.471017555902958,
+    "train_runtime": 35969.3936,
+    "train_samples": 207865,
+    "train_samples_per_second": 11.629,
+    "train_steps_per_second": 0.091
+}

eval_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 3.0,
+    "eval_loss": 3.044856548309326,
+    "eval_runtime": 298.6002,
+    "eval_samples": 23110,
+    "eval_samples_per_second": 51.678,
+    "eval_steps_per_second": 0.81
+}

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "transformers_version": "4.36.2"
+}

model-00001-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc46234ea5b9b1bad46a579bb1efec09e6167a197b03437a4f75d4e6d5452272
+size 4943162336

model-00002-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2ea2ae673d969c38c9ae85994a735e56acfa1034f946358d1a652aafaaca6ad
+size 4999819336

model-00003-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a39bf02568e86442bb7cdb058901318e29862ee6d08eed97b56438d73f7031bd
+size 4540516344

model.safetensors.index.json

@@ -0,0 +1,298 @@
+{
+  "metadata": {
+    "total_size": 14483464192
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00003-of-00003.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00003.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
+    "model.norm.weight": "model-00003-of-00003.safetensors"
+  }
+}

runs/Jan24_07-09-02_dsw-691-5496b4c68c-rxngp/events.out.tfevents.1706080988.dsw-691-5496b4c68c-rxngp.169524.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:81aa7cac6e301b80aaacaa07d352e709be93fd835b0340ba857acc60a13a3708
-size 99289
+oid sha256:1ef0b44bfe848434ed43be3b2f407409db76bc51870922aba7ea235fd802d87e
+size 108235

runs/Jan24_07-09-02_dsw-691-5496b4c68c-rxngp/events.out.tfevents.1706117256.dsw-691-5496b4c68c-rxngp.169524.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:527185dd03e2411903204fee575091ca9e62c8f978a195d55d96b32172c57a22
+size 359

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 3.0,
+    "train_loss": 3.471017555902958,
+    "train_runtime": 35969.3936,
+    "train_samples": 207865,
+    "train_samples_per_second": 11.629,
+    "train_steps_per_second": 0.091
+}

trainer_state.json

@@ -0,0 +1,3978 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9986232216613127,
+  "eval_steps": 500,
+  "global_step": 3267,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.116207951070337e-08,
+      "loss": 1.1319,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0581039755351683e-07,
+      "loss": 1.1109,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.116207951070337e-07,
+      "loss": 1.0925,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.174311926605506e-07,
+      "loss": 1.0283,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2232415902140673e-06,
+      "loss": 1.0316,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5290519877675841e-06,
+      "loss": 1.0114,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8348623853211011e-06,
+      "loss": 0.9868,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.140672782874618e-06,
+      "loss": 0.9866,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4464831804281347e-06,
+      "loss": 0.9904,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7522935779816517e-06,
+      "loss": 0.9907,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.0581039755351682e-06,
+      "loss": 0.9807,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.3639143730886852e-06,
+      "loss": 0.9628,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6697247706422022e-06,
+      "loss": 0.9902,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.975535168195719e-06,
+      "loss": 0.9675,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.281345565749236e-06,
+      "loss": 0.9575,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.587155963302753e-06,
+      "loss": 0.9595,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892966360856269e-06,
+      "loss": 0.9518,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.198776758409786e-06,
+      "loss": 0.9744,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.504587155963303e-06,
+      "loss": 0.9569,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.81039755351682e-06,
+      "loss": 0.9728,
+      "step": 95
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.1162079510703365e-06,
+      "loss": 0.9719,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.422018348623854e-06,
+      "loss": 0.9435,
+      "step": 105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.7278287461773705e-06,
+      "loss": 0.9598,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.033639143730887e-06,
+      "loss": 0.968,
+      "step": 115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.3394495412844045e-06,
+      "loss": 0.9591,
+      "step": 120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.645259938837921e-06,
+      "loss": 0.975,
+      "step": 125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.951070336391438e-06,
+      "loss": 0.9792,
+      "step": 130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.256880733944956e-06,
+      "loss": 0.9831,
+      "step": 135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.562691131498472e-06,
+      "loss": 0.9819,
+      "step": 140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.868501529051989e-06,
+      "loss": 0.9605,
+      "step": 145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.174311926605506e-06,
+      "loss": 0.9591,
+      "step": 150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.480122324159022e-06,
+      "loss": 0.9806,
+      "step": 155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.785932721712539e-06,
+      "loss": 0.9699,
+      "step": 160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.0091743119266055e-05,
+      "loss": 0.9733,
+      "step": 165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.0397553516819572e-05,
+      "loss": 0.9585,
+      "step": 170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.070336391437309e-05,
+      "loss": 0.9809,
+      "step": 175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1009174311926607e-05,
+      "loss": 0.9723,
+      "step": 180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1314984709480123e-05,
+      "loss": 0.9623,
+      "step": 185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.162079510703364e-05,
+      "loss": 0.9751,
+      "step": 190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1926605504587156e-05,
+      "loss": 0.9739,
+      "step": 195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.2232415902140673e-05,
+      "loss": 0.9952,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2538226299694191e-05,
+      "loss": 0.959,
+      "step": 205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2844036697247708e-05,
+      "loss": 0.9811,
+      "step": 210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3149847094801224e-05,
+      "loss": 0.9649,
+      "step": 215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3455657492354741e-05,
+      "loss": 0.995,
+      "step": 220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3761467889908258e-05,
+      "loss": 0.9922,
+      "step": 225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.4067278287461774e-05,
+      "loss": 0.9769,
+      "step": 230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.437308868501529e-05,
+      "loss": 0.9736,
+      "step": 235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.4678899082568809e-05,
+      "loss": 0.9979,
+      "step": 240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.4984709480122325e-05,
+      "loss": 1.0041,
+      "step": 245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5290519877675842e-05,
+      "loss": 0.9841,
+      "step": 250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.559633027522936e-05,
+      "loss": 0.9775,
+      "step": 255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5902140672782875e-05,
+      "loss": 0.9917,
+      "step": 260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.6207951070336393e-05,
+      "loss": 1.0037,
+      "step": 265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6513761467889912e-05,
+      "loss": 0.9789,
+      "step": 270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6819571865443427e-05,
+      "loss": 0.991,
+      "step": 275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7125382262996945e-05,
+      "loss": 0.9945,
+      "step": 280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743119266055046e-05,
+      "loss": 1.0008,
+      "step": 285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7737003058103978e-05,
+      "loss": 1.0067,
+      "step": 290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8042813455657493e-05,
+      "loss": 0.9863,
+      "step": 295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.834862385321101e-05,
+      "loss": 0.9825,
+      "step": 300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.865443425076453e-05,
+      "loss": 0.9839,
+      "step": 305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8960244648318044e-05,
+      "loss": 0.9841,
+      "step": 310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9266055045871563e-05,
+      "loss": 0.9876,
+      "step": 315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9571865443425077e-05,
+      "loss": 0.9975,
+      "step": 320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9877675840978596e-05,
+      "loss": 0.9879,
+      "step": 325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.999994861726391e-05,
+      "loss": 1.0118,
+      "step": 330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9999634613566673e-05,
+      "loss": 0.9976,
+      "step": 335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9999035161089365e-05,
+      "loss": 1.0047,
+      "step": 340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.99981502769439e-05,
+      "loss": 1.0151,
+      "step": 345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.999697998639012e-05,
+      "loss": 1.0113,
+      "step": 350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9995524322835035e-05,
+      "loss": 3.6465,
+      "step": 355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.999378332783191e-05,
+      "loss": 10.618,
+      "step": 360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.999175705107905e-05,
+      "loss": 7.8776,
+      "step": 365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9989445550418393e-05,
+      "loss": 7.4236,
+      "step": 370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9986848891833846e-05,
+      "loss": 7.2276,
+      "step": 375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9983967149449426e-05,
+      "loss": 7.1782,
+      "step": 380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9980800405527127e-05,
+      "loss": 7.1329,
+      "step": 385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.997734875046456e-05,
+      "loss": 7.0406,
+      "step": 390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9973612282792413e-05,
+      "loss": 6.9716,
+      "step": 395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9969591109171584e-05,
+      "loss": 6.8977,
+      "step": 400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9965285344390185e-05,
+      "loss": 6.9986,
+      "step": 405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9960695111360235e-05,
+      "loss": 6.9526,
+      "step": 410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.995582054111416e-05,
+      "loss": 6.8724,
+      "step": 415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9950661772801062e-05,
+      "loss": 6.8315,
+      "step": 420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 6.8329,
+      "step": 425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9939492239129462e-05,
+      "loss": 6.7751,
+      "step": 430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9933481792615583e-05,
+      "loss": 6.7403,
+      "step": 435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.992718778571483e-05,
+      "loss": 6.7105,
+      "step": 440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9920610398095427e-05,
+      "loss": 6.6564,
+      "step": 445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9913749817514963e-05,
+      "loss": 6.6235,
+      "step": 450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.990660623981503e-05,
+      "loss": 6.5955,
+      "step": 455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.989917986891563e-05,
+      "loss": 6.5393,
+      "step": 460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9891470916809362e-05,
+      "loss": 6.5081,
+      "step": 465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9883479603555366e-05,
+      "loss": 6.6039,
+      "step": 470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9875206157273038e-05,
+      "loss": 6.5238,
+      "step": 475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.986665081413552e-05,
+      "loss": 6.4699,
+      "step": 480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9857813818362963e-05,
+      "loss": 6.4097,
+      "step": 485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9848695422215548e-05,
+      "loss": 6.3424,
+      "step": 490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98392958859863e-05,
+      "loss": 6.3099,
+      "step": 495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9829615477993626e-05,
+      "loss": 6.2784,
+      "step": 500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.98196544745737e-05,
+      "loss": 6.2284,
+      "step": 505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.980941316007253e-05,
+      "loss": 6.1791,
+      "step": 510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9798891826837872e-05,
+      "loss": 6.1843,
+      "step": 515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.978809077521088e-05,
+      "loss": 6.1663,
+      "step": 520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9777010313517517e-05,
+      "loss": 6.1206,
+      "step": 525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9765650758059766e-05,
+      "loss": 6.1238,
+      "step": 530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97540124331066e-05,
+      "loss": 6.0433,
+      "step": 535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974209567088473e-05,
+      "loss": 6.0664,
+      "step": 540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9729900811569103e-05,
+      "loss": 6.0446,
+      "step": 545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.971742820327321e-05,
+      "loss": 5.9776,
+      "step": 550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9704678202039148e-05,
+      "loss": 5.9745,
+      "step": 555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9691651171827443e-05,
+      "loss": 5.9384,
+      "step": 560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9678347484506667e-05,
+      "loss": 5.9117,
+      "step": 565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.966476751984283e-05,
+      "loss": 6.119,
+      "step": 570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9650911665488533e-05,
+      "loss": 6.0265,
+      "step": 575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9636780316971902e-05,
+      "loss": 5.9482,
+      "step": 580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.962237387768529e-05,
+      "loss": 5.8934,
+      "step": 585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.960769275887378e-05,
+      "loss": 5.8545,
+      "step": 590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9592737379623427e-05,
+      "loss": 5.8376,
+      "step": 595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9577508166849308e-05,
+      "loss": 5.7886,
+      "step": 600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9562005555283328e-05,
+      "loss": 5.7906,
+      "step": 605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.954622998746181e-05,
+      "loss": 5.8274,
+      "step": 610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9530181913712875e-05,
+      "loss": 5.7876,
+      "step": 615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9513861792143562e-05,
+      "loss": 5.7297,
+      "step": 620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9497270088626773e-05,
+      "loss": 5.6929,
+      "step": 625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9480407276787968e-05,
+      "loss": 5.6823,
+      "step": 630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9463273837991643e-05,
+      "loss": 5.6282,
+      "step": 635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9445870261327592e-05,
+      "loss": 5.6081,
+      "step": 640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.942819704359693e-05,
+      "loss": 5.5536,
+      "step": 645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9410254689297944e-05,
+      "loss": 5.5778,
+      "step": 650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.939204371061166e-05,
+      "loss": 5.5404,
+      "step": 655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9373564627387243e-05,
+      "loss": 5.5092,
+      "step": 660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.935481796712713e-05,
+      "loss": 5.5029,
+      "step": 665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9335804264972018e-05,
+      "loss": 5.4471,
+      "step": 670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9316524063685544e-05,
+      "loss": 5.4672,
+      "step": 675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9296977913638806e-05,
+      "loss": 5.4028,
+      "step": 680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9277166372794663e-05,
+      "loss": 5.3503,
+      "step": 685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.92570900066918e-05,
+      "loss": 5.3415,
+      "step": 690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.923674938842857e-05,
+      "loss": 5.3396,
+      "step": 695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9216145098646667e-05,
+      "loss": 5.295,
+      "step": 700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.919527772551451e-05,
+      "loss": 5.2313,
+      "step": 705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9174147864710486e-05,
+      "loss": 5.202,
+      "step": 710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9152756119405937e-05,
+      "loss": 5.2071,
+      "step": 715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9131103100247934e-05,
+      "loss": 5.176,
+      "step": 720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9109189425341853e-05,
+      "loss": 5.1591,
+      "step": 725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.908701572023372e-05,
+      "loss": 5.0989,
+      "step": 730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9064582617892383e-05,
+      "loss": 5.0671,
+      "step": 735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9041890758691403e-05,
+      "loss": 5.0621,
+      "step": 740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9018940790390798e-05,
+      "loss": 5.0536,
+      "step": 745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8995733368118556e-05,
+      "loss": 5.0143,
+      "step": 750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8972269154351917e-05,
+      "loss": 4.9735,
+      "step": 755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8948548818898478e-05,
+      "loss": 4.9158,
+      "step": 760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8924573038877062e-05,
+      "loss": 4.9633,
+      "step": 765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8900342498698384e-05,
+      "loss": 4.9076,
+      "step": 770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8875857890045544e-05,
+      "loss": 4.8885,
+      "step": 775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8851119911854236e-05,
+      "loss": 4.8311,
+      "step": 780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8826129270292837e-05,
+      "loss": 4.8127,
+      "step": 785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8800886678742225e-05,
+      "loss": 4.8224,
+      "step": 790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.877539285777543e-05,
+      "loss": 4.7722,
+      "step": 795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8749648535137048e-05,
+      "loss": 4.7507,
+      "step": 800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8723654445722485e-05,
+      "loss": 4.7067,
+      "step": 805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8697411331556958e-05,
+      "loss": 4.6938,
+      "step": 810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.867091994177433e-05,
+      "loss": 4.7096,
+      "step": 815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8644181032595737e-05,
+      "loss": 4.6845,
+      "step": 820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.861719536730795e-05,
+      "loss": 4.6439,
+      "step": 825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8589963716241653e-05,
+      "loss": 4.665,
+      "step": 830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8562486856749403e-05,
+      "loss": 4.624,
+      "step": 835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.853476557318346e-05,
+      "loss": 4.5849,
+      "step": 840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8506800656873397e-05,
+      "loss": 4.5517,
+      "step": 845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8478592906103508e-05,
+      "loss": 4.5289,
+      "step": 850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8450143126090015e-05,
+      "loss": 4.5497,
+      "step": 855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.842145212895809e-05,
+      "loss": 4.5517,
+      "step": 860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8392520733718684e-05,
+      "loss": 4.5593,
+      "step": 865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8363349766245108e-05,
+      "loss": 4.5105,
+      "step": 870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.83339400592495e-05,
+      "loss": 4.5006,
+      "step": 875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8304292452259037e-05,
+      "loss": 4.4665,
+      "step": 880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8274407791591966e-05,
+      "loss": 4.4582,
+      "step": 885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.824428693033345e-05,
+      "loss": 4.3962,
+      "step": 890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.821393072831121e-05,
+      "loss": 4.4116,
+      "step": 895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8183340052071e-05,
+      "loss": 4.4362,
+      "step": 900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8152515774851846e-05,
+      "loss": 4.3861,
+      "step": 905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.812145877656113e-05,
+      "loss": 4.3563,
+      "step": 910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 4.3578,
+      "step": 915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.805865016958543e-05,
+      "loss": 4.3163,
+      "step": 920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.802690035382998e-05,
+      "loss": 4.3062,
+      "step": 925
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7994921402810862e-05,
+      "loss": 4.2984,
+      "step": 930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.796271422939668e-05,
+      "loss": 4.3264,
+      "step": 935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7930279752970866e-05,
+      "loss": 4.2954,
+      "step": 940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7897618899405423e-05,
+      "loss": 4.2867,
+      "step": 945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.78647326010345e-05,
+      "loss": 4.307,
+      "step": 950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7831621796627773e-05,
+      "loss": 4.2882,
+      "step": 955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7798287431363644e-05,
+      "loss": 4.2169,
+      "step": 960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7764730456802278e-05,
+      "loss": 4.2223,
+      "step": 965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7730951830858422e-05,
+      "loss": 4.2085,
+      "step": 970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.769695251777406e-05,
+      "loss": 4.1943,
+      "step": 975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.76627334880909e-05,
+      "loss": 4.1485,
+      "step": 980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7628295718622666e-05,
+      "loss": 4.1761,
+      "step": 985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.75936401924272e-05,
+      "loss": 4.2029,
+      "step": 990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.755876789877842e-05,
+      "loss": 4.1792,
+      "step": 995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7523679833138062e-05,
+      "loss": 4.1495,
+      "step": 1000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.748837699712728e-05,
+      "loss": 4.1342,
+      "step": 1005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7452860398498052e-05,
+      "loss": 4.1056,
+      "step": 1010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7417131051104382e-05,
+      "loss": 4.0872,
+      "step": 1015
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.738118997487341e-05,
+      "loss": 4.0758,
+      "step": 1020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7345038195776255e-05,
+      "loss": 4.0705,
+      "step": 1025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7308676745798748e-05,
+      "loss": 4.0337,
+      "step": 1030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7272106662911972e-05,
+      "loss": 4.0706,
+      "step": 1035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.723532899104263e-05,
+      "loss": 4.051,
+      "step": 1040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.719834478004323e-05,
+      "loss": 4.0815,
+      "step": 1045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7161155085662144e-05,
+      "loss": 4.0477,
+      "step": 1050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.712376096951345e-05,
+      "loss": 4.04,
+      "step": 1055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7086163499046633e-05,
+      "loss": 4.0102,
+      "step": 1060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.704836374751612e-05,
+      "loss": 4.0224,
+      "step": 1065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7010362793950627e-05,
+      "loss": 4.0093,
+      "step": 1070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.697216172312238e-05,
+      "loss": 4.0132,
+      "step": 1075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.693376162551613e-05,
+      "loss": 3.9887,
+      "step": 1080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.689516359729803e-05,
+      "loss": 3.9689,
+      "step": 1085
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 3.9833078384399414,
+      "eval_runtime": 298.6999,
+      "eval_samples_per_second": 51.661,
+      "eval_steps_per_second": 0.81,
+      "step": 1089
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6856368740284342e-05,
+      "loss": 3.994,
+      "step": 1090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6817378161909995e-05,
+      "loss": 3.9327,
+      "step": 1095
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.677819297519696e-05,
+      "loss": 3.9248,
+      "step": 1100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6738814298722484e-05,
+      "loss": 3.9413,
+      "step": 1105
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6699243256587156e-05,
+      "loss": 3.9817,
+      "step": 1110
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6659480978382815e-05,
+      "loss": 3.9372,
+      "step": 1115
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6619528599160318e-05,
+      "loss": 3.9028,
+      "step": 1120
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.657938725939713e-05,
+      "loss": 3.8865,
+      "step": 1125
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6539058104964764e-05,
+      "loss": 3.9158,
+      "step": 1130
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6498542287096074e-05,
+      "loss": 3.8638,
+      "step": 1135
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6457840962352403e-05,
+      "loss": 3.8689,
+      "step": 1140
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6416955292590556e-05,
+      "loss": 3.834,
+      "step": 1145
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6375886444929628e-05,
+      "loss": 3.8792,
+      "step": 1150
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.63346355917177e-05,
+      "loss": 3.8686,
+      "step": 1155
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6293203910498375e-05,
+      "loss": 3.8564,
+      "step": 1160
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6251592583977155e-05,
+      "loss": 3.8589,
+      "step": 1165
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6209802799987674e-05,
+      "loss": 3.8278,
+      "step": 1170
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6167835751457812e-05,
+      "loss": 3.8564,
+      "step": 1175
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6125692636375618e-05,
+      "loss": 3.8202,
+      "step": 1180
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6083374657755132e-05,
+      "loss": 3.7818,
+      "step": 1185
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.604088302360203e-05,
+      "loss": 3.8209,
+      "step": 1190
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.599821894687914e-05,
+      "loss": 3.7893,
+      "step": 1195
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5955383645471828e-05,
+      "loss": 3.8644,
+      "step": 1200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5912378342153233e-05,
+      "loss": 3.8387,
+      "step": 1205
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.586920426454934e-05,
+      "loss": 3.8262,
+      "step": 1210
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5825862645103962e-05,
+      "loss": 3.7695,
+      "step": 1215
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5782354721043546e-05,
+      "loss": 3.7794,
+      "step": 1220
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5738681734341855e-05,
+      "loss": 3.7333,
+      "step": 1225
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.569484493168452e-05,
+      "loss": 3.7563,
+      "step": 1230
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.565084556443345e-05,
+      "loss": 3.7361,
+      "step": 1235
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5606684888591105e-05,
+      "loss": 3.747,
+      "step": 1240
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.556236416476465e-05,
+      "loss": 3.75,
+      "step": 1245
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.551788465812997e-05,
+      "loss": 3.7464,
+      "step": 1250
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5473247638395547e-05,
+      "loss": 3.7593,
+      "step": 1255
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5428454379766223e-05,
+      "loss": 3.7157,
+      "step": 1260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5383506160906826e-05,
+      "loss": 3.7071,
+      "step": 1265
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5338404264905654e-05,
+      "loss": 3.6841,
+      "step": 1270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5293149979237875e-05,
+      "loss": 3.6952,
+      "step": 1275
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5247744595728753e-05,
+      "loss": 3.6776,
+      "step": 1280
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.520218941051678e-05,
+      "loss": 3.6699,
+      "step": 1285
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5156485724016672e-05,
+      "loss": 3.7052,
+      "step": 1290
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5110634840882258e-05,
+      "loss": 3.6975,
+      "step": 1295
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5064638069969228e-05,
+      "loss": 3.6662,
+      "step": 1300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5018496724297778e-05,
+      "loss": 3.6939,
+      "step": 1305
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.497221212101511e-05,
+      "loss": 3.6552,
+      "step": 1310
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4925785581357852e-05,
+      "loss": 3.6815,
+      "step": 1315
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4879218430614346e-05,
+      "loss": 3.6587,
+      "step": 1320
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4832511998086802e-05,
+      "loss": 3.6675,
+      "step": 1325
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4785667617053347e-05,
+      "loss": 3.6285,
+      "step": 1330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4738686624729987e-05,
+      "loss": 3.6395,
+      "step": 1335
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4691570362232422e-05,
+      "loss": 3.6219,
+      "step": 1340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4644320174537754e-05,
+      "loss": 3.5879,
+      "step": 1345
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.4596937410446117e-05,
+      "loss": 3.6236,
+      "step": 1350
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.4549423422542148e-05,
+      "loss": 3.6519,
+      "step": 1355
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4501779567156394e-05,
+      "loss": 3.6416,
+      "step": 1360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4454007204326592e-05,
+      "loss": 3.6103,
+      "step": 1365
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4406107697758838e-05,
+      "loss": 3.5947,
+      "step": 1370
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4358082414788666e-05,
+      "loss": 3.6187,
+      "step": 1375
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4309932726342007e-05,
+      "loss": 3.5991,
+      "step": 1380
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4261660006896066e-05,
+      "loss": 3.5571,
+      "step": 1385
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.421326563444008e-05,
+      "loss": 3.6026,
+      "step": 1390
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4164750990435991e-05,
+      "loss": 3.5359,
+      "step": 1395
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4116117459778991e-05,
+      "loss": 3.5824,
+      "step": 1400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4067366430758004e-05,
+      "loss": 3.5873,
+      "step": 1405
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4018499295016057e-05,
+      "loss": 3.5691,
+      "step": 1410
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3969517447510546e-05,
+      "loss": 3.5361,
+      "step": 1415
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3920422286473426e-05,
+      "loss": 3.5292,
+      "step": 1420
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3871215213371284e-05,
+      "loss": 3.5255,
+      "step": 1425
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3821897632865345e-05,
+      "loss": 3.5514,
+      "step": 1430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3772470952771364e-05,
+      "loss": 3.5652,
+      "step": 1435
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3722936584019453e-05,
+      "loss": 3.5358,
+      "step": 1440
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3673295940613789e-05,
+      "loss": 3.4849,
+      "step": 1445
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.362355043959226e-05,
+      "loss": 3.5337,
+      "step": 1450
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3573701500986012e-05,
+      "loss": 3.5249,
+      "step": 1455
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3523750547778903e-05,
+      "loss": 3.5507,
+      "step": 1460
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3473699005866904e-05,
+      "loss": 3.5388,
+      "step": 1465
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.342354830401738e-05,
+      "loss": 3.5469,
+      "step": 1470
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3373299873828303e-05,
+      "loss": 3.508,
+      "step": 1475
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3322955149687391e-05,
+      "loss": 3.5033,
+      "step": 1480
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.327251556873117e-05,
+      "loss": 3.4788,
+      "step": 1485
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.322198257080393e-05,
+      "loss": 3.4982,
+      "step": 1490
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3171357598416642e-05,
+      "loss": 3.5111,
+      "step": 1495
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3120642096705773e-05,
+      "loss": 3.463,
+      "step": 1500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3069837513392024e-05,
+      "loss": 3.4749,
+      "step": 1505
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3018945298739022e-05,
+      "loss": 3.4663,
+      "step": 1510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2967966905511906e-05,
+      "loss": 3.5109,
+      "step": 1515
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2916903788935869e-05,
+      "loss": 3.4501,
+      "step": 1520
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2865757406654598e-05,
+      "loss": 3.4358,
+      "step": 1525
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2814529218688688e-05,
+      "loss": 3.4623,
+      "step": 1530
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2763220687393942e-05,
+      "loss": 3.4665,
+      "step": 1535
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2711833277419648e-05,
+      "loss": 3.4724,
+      "step": 1540
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2660368455666752e-05,
+      "loss": 3.461,
+      "step": 1545
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2608827691245993e-05,
+      "loss": 3.4771,
+      "step": 1550
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2557212455435958e-05,
+      "loss": 3.4635,
+      "step": 1555
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2505524221641097e-05,
+      "loss": 3.4335,
+      "step": 1560
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.245376446534965e-05,
+      "loss": 3.472,
+      "step": 1565
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2401934664091529e-05,
+      "loss": 3.4799,
+      "step": 1570
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2350036297396153e-05,
+      "loss": 3.4485,
+      "step": 1575
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2298070846750197e-05,
+      "loss": 3.4313,
+      "step": 1580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2246039795555314e-05,
+      "loss": 3.4332,
+      "step": 1585
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2193944629085778e-05,
+      "loss": 3.4413,
+      "step": 1590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2141786834446105e-05,
+      "loss": 3.421,
+      "step": 1595
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2089567900528577e-05,
+      "loss": 3.4325,
+      "step": 1600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2037289317970757e-05,
+      "loss": 3.4061,
+      "step": 1605
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1984952579112938e-05,
+      "loss": 3.4377,
+      "step": 1610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1932559177955533e-05,
+      "loss": 3.4344,
+      "step": 1615
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.1880110610116438e-05,
+      "loss": 3.3839,
+      "step": 1620
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.1827608372788325e-05,
+      "loss": 3.3993,
+      "step": 1625
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.1775053964695913e-05,
+      "loss": 3.389,
+      "step": 1630
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.172244888605319e-05,
+      "loss": 3.4025,
+      "step": 1635
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1669794638520578e-05,
+      "loss": 3.4452,
+      "step": 1640
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1617092725162064e-05,
+      "loss": 3.4183,
+      "step": 1645
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.156434465040231e-05,
+      "loss": 3.4039,
+      "step": 1650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.151155191998369e-05,
+      "loss": 3.3809,
+      "step": 1655
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1458716040923319e-05,
+      "loss": 3.3923,
+      "step": 1660
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.140583852147003e-05,
+      "loss": 3.4024,
+      "step": 1665
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1352920871061318e-05,
+      "loss": 3.376,
+      "step": 1670
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.1299964600280247e-05,
+      "loss": 3.3738,
+      "step": 1675
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.1246971220812348e-05,
+      "loss": 3.4101,
+      "step": 1680
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.1193942245402443e-05,
+      "loss": 3.3748,
+      "step": 1685
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.1140879187811482e-05,
+      "loss": 3.3673,
+      "step": 1690
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.108778356277331e-05,
+      "loss": 3.3682,
+      "step": 1695
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.103465688595145e-05,
+      "loss": 3.3493,
+      "step": 1700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0981500673895825e-05,
+      "loss": 3.3544,
+      "step": 1705
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0928316443999462e-05,
+      "loss": 3.3209,
+      "step": 1710
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0875105714455193e-05,
+      "loss": 3.344,
+      "step": 1715
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0821870004212305e-05,
+      "loss": 3.3518,
+      "step": 1720
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0768610832933169e-05,
+      "loss": 3.3127,
+      "step": 1725
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.0715329720949887e-05,
+      "loss": 3.3628,
+      "step": 1730
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.0662028189220876e-05,
+      "loss": 3.3672,
+      "step": 1735
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.0608707759287454e-05,
+      "loss": 3.3463,
+      "step": 1740
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.0555369953230402e-05,
+      "loss": 3.3627,
+      "step": 1745
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.0502016293626513e-05,
+      "loss": 3.3459,
+      "step": 1750
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.044864830350515e-05,
+      "loss": 3.3362,
+      "step": 1755
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.0395267506304742e-05,
+      "loss": 3.3494,
+      "step": 1760
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.034187542582931e-05,
+      "loss": 3.3633,
+      "step": 1765
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.028847358620497e-05,
+      "loss": 3.3545,
+      "step": 1770
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.0235063511836416e-05,
+      "loss": 3.3147,
+      "step": 1775
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.0181646727363413e-05,
+      "loss": 3.3145,
+      "step": 1780
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.0128224757617272e-05,
+      "loss": 3.33,
+      "step": 1785
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.007479912757733e-05,
+      "loss": 3.2897,
+      "step": 1790
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.0021371362327397e-05,
+      "loss": 3.3371,
+      "step": 1795
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.967942987012243e-06,
+      "loss": 3.2989,
+      "step": 1800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.91451552679405e-06,
+      "loss": 3.2985,
+      "step": 1805
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.861090506808876e-06,
+      "loss": 3.3103,
+      "step": 1810
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.80766945212313e-06,
+      "loss": 3.301,
+      "step": 1815
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.754253887690014e-06,
+      "loss": 3.3136,
+      "step": 1820
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.700845338306018e-06,
+      "loss": 3.3193,
+      "step": 1825
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.647445328567368e-06,
+      "loss": 3.3424,
+      "step": 1830
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.594055382826534e-06,
+      "loss": 3.3016,
+      "step": 1835
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.540677025148686e-06,
+      "loss": 3.2602,
+      "step": 1840
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.48731177926821e-06,
+      "loss": 3.2621,
+      "step": 1845
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.433961168545194e-06,
+      "loss": 3.2633,
+      "step": 1850
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.380626715921972e-06,
+      "loss": 3.2901,
+      "step": 1855
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.327309943879604e-06,
+      "loss": 3.2936,
+      "step": 1860
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.274012374394466e-06,
+      "loss": 3.2913,
+      "step": 1865
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.220735528894755e-06,
+      "loss": 3.263,
+      "step": 1870
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.167480928217108e-06,
+      "loss": 3.2958,
+      "step": 1875
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.114250092563142e-06,
+      "loss": 3.2725,
+      "step": 1880
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.061044541456097e-06,
+      "loss": 3.2369,
+      "step": 1885
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.007865793697425e-06,
+      "loss": 3.2583,
+      "step": 1890
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 3.2728,
+      "step": 1895
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.90159477956209e-06,
+      "loss": 3.2607,
+      "step": 1900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.848505546789407e-06,
+      "loss": 3.2824,
+      "step": 1905
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.795449184486457e-06,
+      "loss": 3.2628,
+      "step": 1910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.742427207195975e-06,
+      "loss": 3.2473,
+      "step": 1915
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.689441128479134e-06,
+      "loss": 3.275,
+      "step": 1920
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.636492460872347e-06,
+      "loss": 3.2592,
+      "step": 1925
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.583582715844113e-06,
+      "loss": 3.2729,
+      "step": 1930
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.530713403751822e-06,
+      "loss": 3.254,
+      "step": 1935
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.47788603379869e-06,
+      "loss": 3.2348,
+      "step": 1940
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.425102113990647e-06,
+      "loss": 3.2028,
+      "step": 1945
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.372363151093302e-06,
+      "loss": 3.2566,
+      "step": 1950
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.319670650588916e-06,
+      "loss": 3.2351,
+      "step": 1955
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.26702611663346e-06,
+      "loss": 3.2721,
+      "step": 1960
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.214431052013636e-06,
+      "loss": 3.2191,
+      "step": 1965
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.161886958104007e-06,
+      "loss": 3.2367,
+      "step": 1970
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.109395334824127e-06,
+      "loss": 3.2721,
+      "step": 1975
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.056957680595733e-06,
+      "loss": 3.2054,
+      "step": 1980
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.00457549229996e-06,
+      "loss": 3.215,
+      "step": 1985
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.952250265234618e-06,
+      "loss": 3.2181,
+      "step": 1990
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.899983493071506e-06,
+      "loss": 3.2148,
+      "step": 1995
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.847776667813782e-06,
+      "loss": 3.2244,
+      "step": 2000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.795631279753346e-06,
+      "loss": 3.1918,
+      "step": 2005
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.74354881742834e-06,
+      "loss": 3.2312,
+      "step": 2010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.691530767580613e-06,
+      "loss": 3.2341,
+      "step": 2015
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.639578615113312e-06,
+      "loss": 3.2151,
+      "step": 2020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.587693843048475e-06,
+      "loss": 3.1932,
+      "step": 2025
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.535877932484714e-06,
+      "loss": 3.1933,
+      "step": 2030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.484132362554915e-06,
+      "loss": 3.2302,
+      "step": 2035
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.432458610384037e-06,
+      "loss": 3.1807,
+      "step": 2040
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.380858151046922e-06,
+      "loss": 3.2067,
+      "step": 2045
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.329332457526215e-06,
+      "loss": 3.1698,
+      "step": 2050
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.27788300067029e-06,
+      "loss": 3.2017,
+      "step": 2055
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.2265112491512824e-06,
+      "loss": 3.213,
+      "step": 2060
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.175218669423153e-06,
+      "loss": 3.1922,
+      "step": 2065
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.124006725679828e-06,
+      "loss": 3.2127,
+      "step": 2070
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.0728768798134195e-06,
+      "loss": 3.2197,
+      "step": 2075
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.0218305913724615e-06,
+      "loss": 3.2192,
+      "step": 2080
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.97086931752028e-06,
+      "loss": 3.2106,
+      "step": 2085
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.919994512993369e-06,
+      "loss": 3.219,
+      "step": 2090
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.869207630059885e-06,
+      "loss": 3.184,
+      "step": 2095
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.818510118478172e-06,
+      "loss": 3.1925,
+      "step": 2100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.767903425455402e-06,
+      "loss": 3.1785,
+      "step": 2105
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.7173889956062285e-06,
+      "loss": 3.1938,
+      "step": 2110
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.666968270911585e-06,
+      "loss": 3.1738,
+      "step": 2115
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.616642690677488e-06,
+      "loss": 3.1936,
+      "step": 2120
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.566413691493989e-06,
+      "loss": 3.159,
+      "step": 2125
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.5162827071941194e-06,
+      "loss": 3.1676,
+      "step": 2130
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.46625116881301e-06,
+      "loss": 3.1511,
+      "step": 2135
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.4163205045469975e-06,
+      "loss": 3.1833,
+      "step": 2140
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.366492139712886e-06,
+      "loss": 3.1522,
+      "step": 2145
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.316767496707237e-06,
+      "loss": 3.2052,
+      "step": 2150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.267147994965792e-06,
+      "loss": 3.1991,
+      "step": 2155
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.217635050922923e-06,
+      "loss": 3.1547,
+      "step": 2160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.168230077971224e-06,
+      "loss": 3.152,
+      "step": 2165
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.118934486421149e-06,
+      "loss": 3.1728,
+      "step": 2170
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.069749683460765e-06,
+      "loss": 3.1562,
+      "step": 2175
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 3.173952341079712,
+      "eval_runtime": 298.8936,
+      "eval_samples_per_second": 51.627,
+      "eval_steps_per_second": 0.81,
+      "step": 2179
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.020677073115571e-06,
+      "loss": 3.1367,
+      "step": 2180
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.971718056208431e-06,
+      "loss": 3.1139,
+      "step": 2185
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.922874030319567e-06,
+      "loss": 3.1473,
+      "step": 2190
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.874146389746697e-06,
+      "loss": 3.1541,
+      "step": 2195
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.825536525465192e-06,
+      "loss": 3.1406,
+      "step": 2200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.7770458250884044e-06,
+      "loss": 3.119,
+      "step": 2205
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.728675672828037e-06,
+      "loss": 3.1426,
+      "step": 2210
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.680427449454631e-06,
+      "loss": 3.1518,
+      "step": 2215
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.6323025322581694e-06,
+      "loss": 3.1133,
+      "step": 2220
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.5843022950087325e-06,
+      "loss": 3.1132,
+      "step": 2225
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.5364281079173075e-06,
+      "loss": 3.1484,
+      "step": 2230
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.488681337596653e-06,
+      "loss": 3.1142,
+      "step": 2235
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.441063347022313e-06,
+      "loss": 3.1046,
+      "step": 2240
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.393575495493679e-06,
+      "loss": 3.0933,
+      "step": 2245
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.346219138595215e-06,
+      "loss": 3.1068,
+      "step": 2250
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.298995628157738e-06,
+      "loss": 3.1296,
+      "step": 2255
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.251906312219859e-06,
+      "loss": 3.1175,
+      "step": 2260
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.2049525349894625e-06,
+      "loss": 3.066,
+      "step": 2265
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.158135636805372e-06,
+      "loss": 3.0771,
+      "step": 2270
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.111456954099064e-06,
+      "loss": 3.0995,
+      "step": 2275
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.064917819356532e-06,
+      "loss": 3.07,
+      "step": 2280
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.0185195610802365e-06,
+      "loss": 3.0931,
+      "step": 2285
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.9722635037512006e-06,
+      "loss": 3.0755,
+      "step": 2290
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.92615096779118e-06,
+      "loss": 3.0904,
+      "step": 2295
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.8801832695249864e-06,
+      "loss": 3.0652,
+      "step": 2300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.834361721142901e-06,
+      "loss": 3.066,
+      "step": 2305
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.788687630663232e-06,
+      "loss": 3.1098,
+      "step": 2310
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.743162301894952e-06,
+      "loss": 3.106,
+      "step": 2315
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.697787034400511e-06,
+      "loss": 3.1032,
+      "step": 2320
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.652563123458703e-06,
+      "loss": 3.051,
+      "step": 2325
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.607491860027726e-06,
+      "loss": 3.1206,
+      "step": 2330
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.5625745307083e-06,
+      "loss": 3.0868,
+      "step": 2335
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.517812417706967e-06,
+      "loss": 3.1152,
+      "step": 2340
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.473206798799469e-06,
+      "loss": 3.0738,
+      "step": 2345
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.428758947294278e-06,
+      "loss": 3.0936,
+      "step": 2350
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.3844701319962525e-06,
+      "loss": 3.0962,
+      "step": 2355
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.3403416171704225e-06,
+      "loss": 3.1008,
+      "step": 2360
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.296374662505885e-06,
+      "loss": 3.0757,
+      "step": 2365
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.252570523079852e-06,
+      "loss": 3.1204,
+      "step": 2370
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.2089304493218355e-06,
+      "loss": 3.0742,
+      "step": 2375
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.165455686977927e-06,
+      "loss": 3.0537,
+      "step": 2380
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 3.0725,
+      "step": 2385
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.079007055886598e-06,
+      "loss": 3.0505,
+      "step": 2390
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.036035654894967e-06,
+      "loss": 3.075,
+      "step": 2395
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.993234500758597e-06,
+      "loss": 3.0824,
+      "step": 2400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9506048152758584e-06,
+      "loss": 3.1028,
+      "step": 2405
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.90814781535038e-06,
+      "loss": 3.0949,
+      "step": 2410
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.865864712956336e-06,
+      "loss": 3.1145,
+      "step": 2415
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.823756715103822e-06,
+      "loss": 3.0736,
+      "step": 2420
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.7818250238044274e-06,
+      "loss": 3.0743,
+      "step": 2425
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.740070836036893e-06,
+      "loss": 3.0886,
+      "step": 2430
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6984953437129734e-06,
+      "loss": 3.0525,
+      "step": 2435
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6570997336433854e-06,
+      "loss": 3.0829,
+      "step": 2440
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6158851875039458e-06,
+      "loss": 3.0823,
+      "step": 2445
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.57485288180183e-06,
+      "loss": 3.0612,
+      "step": 2450
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.534003987842005e-06,
+      "loss": 3.0736,
+      "step": 2455
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.493339671693765e-06,
+      "loss": 3.092,
+      "step": 2460
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4528610941574724e-06,
+      "loss": 3.0988,
+      "step": 2465
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.412569410731401e-06,
+      "loss": 3.0658,
+      "step": 2470
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.372465771578771e-06,
+      "loss": 3.0688,
+      "step": 2475
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.33255132149489e-06,
+      "loss": 3.1096,
+      "step": 2480
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.2928271998745074e-06,
+      "loss": 3.0544,
+      "step": 2485
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.2532945406792573e-06,
+      "loss": 3.0951,
+      "step": 2490
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.2139544724053083e-06,
+      "loss": 3.0645,
+      "step": 2495
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.1748081180511393e-06,
+      "loss": 3.0489,
+      "step": 2500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.135856595085498e-06,
+      "loss": 3.0727,
+      "step": 2505
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.097101015415478e-06,
+      "loss": 3.0441,
+      "step": 2510
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0585424853547953e-06,
+      "loss": 3.0358,
+      "step": 2515
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.02018210559221e-06,
+      "loss": 3.0223,
+      "step": 2520
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9820209711600858e-06,
+      "loss": 3.0548,
+      "step": 2525
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9440601714031614e-06,
+      "loss": 3.0687,
+      "step": 2530
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9063007899474214e-06,
+      "loss": 3.0546,
+      "step": 2535
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8687439046691956e-06,
+      "loss": 3.043,
+      "step": 2540
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.83139058766436e-06,
+      "loss": 3.0779,
+      "step": 2545
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.794241905217753e-06,
+      "loss": 3.0609,
+      "step": 2550
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.757298917772727e-06,
+      "loss": 3.0404,
+      "step": 2555
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.720562679900892e-06,
+      "loss": 3.0568,
+      "step": 2560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6840342402719867e-06,
+      "loss": 3.033,
+      "step": 2565
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6477146416239695e-06,
+      "loss": 3.0252,
+      "step": 2570
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6116049207332304e-06,
+      "loss": 3.0511,
+      "step": 2575
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.5757061083850153e-06,
+      "loss": 3.0329,
+      "step": 2580
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.5400192293439864e-06,
+      "loss": 3.038,
+      "step": 2585
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.5045453023249724e-06,
+      "loss": 3.0511,
+      "step": 2590
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.469285339963892e-06,
+      "loss": 3.0528,
+      "step": 2595
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.43424034878885e-06,
+      "loss": 3.0811,
+      "step": 2600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.399411329191393e-06,
+      "loss": 3.079,
+      "step": 2605
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3647992753979698e-06,
+      "loss": 3.0413,
+      "step": 2610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.330405175441529e-06,
+      "loss": 3.0101,
+      "step": 2615
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.296230011133337e-06,
+      "loss": 3.0165,
+      "step": 2620
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2622747580349313e-06,
+      "loss": 3.0654,
+      "step": 2625
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2285403854302912e-06,
+      "loss": 3.0348,
+      "step": 2630
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1950278562981497e-06,
+      "loss": 3.0505,
+      "step": 2635
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1617381272845174e-06,
+      "loss": 3.0139,
+      "step": 2640
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.128672148675366e-06,
+      "loss": 3.0148,
+      "step": 2645
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.095830864369518e-06,
+      "loss": 3.0709,
+      "step": 2650
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.063215211851678e-06,
+      "loss": 3.0484,
+      "step": 2655
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0308261221656845e-06,
+      "loss": 3.051,
+      "step": 2660
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9986645198879385e-06,
+      "loss": 3.0223,
+      "step": 2665
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9667313231009955e-06,
+      "loss": 3.012,
+      "step": 2670
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9350274433673745e-06,
+      "loss": 3.0439,
+      "step": 2675
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9035537857035157e-06,
+      "loss": 3.0163,
+      "step": 2680
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.8723112485539741e-06,
+      "loss": 3.0249,
+      "step": 2685
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8413007237657422e-06,
+      "loss": 3.0288,
+      "step": 2690
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.810523096562814e-06,
+      "loss": 3.0737,
+      "step": 2695
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7799792455209019e-06,
+      "loss": 3.0578,
+      "step": 2700
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7496700425423708e-06,
+      "loss": 3.0571,
+      "step": 2705
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.719596352831332e-06,
+      "loss": 3.0531,
+      "step": 2710
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6897590348689607e-06,
+      "loss": 3.0621,
+      "step": 2715
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.660158940388975e-06,
+      "loss": 3.0343,
+      "step": 2720
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6307969143533397e-06,
+      "loss": 2.9669,
+      "step": 2725
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.601673794928127e-06,
+      "loss": 2.9813,
+      "step": 2730
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5727904134596084e-06,
+      "loss": 3.0041,
+      "step": 2735
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.544147594450508e-06,
+      "loss": 3.0225,
+      "step": 2740
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5157461555364772e-06,
+      "loss": 3.0281,
+      "step": 2745
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4875869074627436e-06,
+      "loss": 3.0106,
+      "step": 2750
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4596706540609862e-06,
+      "loss": 3.041,
+      "step": 2755
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4319981922263636e-06,
+      "loss": 3.0251,
+      "step": 2760
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.404570311894793e-06,
+      "loss": 3.0577,
+      "step": 2765
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.377387796020374e-06,
+      "loss": 3.04,
+      "step": 2770
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.350451420553065e-06,
+      "loss": 3.0483,
+      "step": 2775
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3237619544165081e-06,
+      "loss": 3.0236,
+      "step": 2780
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2973201594860985e-06,
+      "loss": 3.0387,
+      "step": 2785
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.271126790567223e-06,
+      "loss": 3.0403,
+      "step": 2790
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2451825953737273e-06,
+      "loss": 3.0073,
+      "step": 2795
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.219488314506556e-06,
+      "loss": 3.0374,
+      "step": 2800
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.19404468143262e-06,
+      "loss": 3.0193,
+      "step": 2805
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1688524224638652e-06,
+      "loss": 3.0517,
+      "step": 2810
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1439122567365214e-06,
+      "loss": 3.0039,
+      "step": 2815
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.119224896190595e-06,
+      "loss": 3.0071,
+      "step": 2820
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0947910455495248e-06,
+      "loss": 2.98,
+      "step": 2825
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.070611402300089e-06,
+      "loss": 3.0076,
+      "step": 2830
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0466866566724698e-06,
+      "loss": 3.018,
+      "step": 2835
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.0230174916205681e-06,
+      "loss": 3.0337,
+      "step": 2840
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.99604582802498e-07,
+      "loss": 3.018,
+      "step": 2845
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.764485985613092e-07,
+      "loss": 3.0559,
+      "step": 2850
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.535501999058971e-07,
+      "loss": 3.0159,
+      "step": 2855
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.309100404921445e-07,
+      "loss": 3.0024,
+      "step": 2860
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.085287666042508e-07,
+      "loss": 3.0056,
+      "step": 2865
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.86407017136296e-07,
+      "loss": 2.9986,
+      "step": 2870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 3.0271,
+      "step": 2875
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.429446099766614e-07,
+      "loss": 2.9838,
+      "step": 2880
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.216051929594271e-07,
+      "loss": 3.0144,
+      "step": 2885
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.005277816755985e-07,
+      "loss": 3.0081,
+      "step": 2890
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.797129777992951e-07,
+      "loss": 3.0238,
+      "step": 2895
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.591613755082661e-07,
+      "loss": 3.0089,
+      "step": 2900
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.38873561466924e-07,
+      "loss": 3.0092,
+      "step": 2905
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.188501148096117e-07,
+      "loss": 2.9975,
+      "step": 2910
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.990916071240506e-07,
+      "loss": 3.0156,
+      "step": 2915
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.795986024350465e-07,
+      "loss": 3.0238,
+      "step": 2920
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.60371657188369e-07,
+      "loss": 3.0223,
+      "step": 2925
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.414113202348793e-07,
+      "loss": 2.9978,
+      "step": 2930
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.227181328148568e-07,
+      "loss": 2.9838,
+      "step": 2935
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.042926285425577e-07,
+      "loss": 2.9852,
+      "step": 2940
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.861353333909692e-07,
+      "loss": 3.0411,
+      "step": 2945
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.682467656768054e-07,
+      "loss": 3.024,
+      "step": 2950
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.506274360457087e-07,
+      "loss": 3.0024,
+      "step": 2955
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.332778474576694e-07,
+      "loss": 3.0062,
+      "step": 2960
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.161984951726762e-07,
+      "loss": 3.0125,
+      "step": 2965
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.993898667365671e-07,
+      "loss": 3.007,
+      "step": 2970
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.828524419671266e-07,
+      "loss": 3.0326,
+      "step": 2975
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.6658669294037393e-07,
+      "loss": 3.01,
+      "step": 2980
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.505930839770967e-07,
+      "loss": 2.9916,
+      "step": 2985
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.348720716295918e-07,
+      "loss": 3.0423,
+      "step": 2990
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.194241046686398e-07,
+      "loss": 2.9791,
+      "step": 2995
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.0424962407068167e-07,
+      "loss": 3.0081,
+      "step": 3000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.8934906300524654e-07,
+      "loss": 3.0078,
+      "step": 3005
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.7472284682256943e-07,
+      "loss": 3.0088,
+      "step": 3010
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.603713930414676e-07,
+      "loss": 2.988,
+      "step": 3015
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.462951113374036e-07,
+      "loss": 3.0218,
+      "step": 3020
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.324944035308053e-07,
+      "loss": 3.0111,
+      "step": 3025
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.189696635755868e-07,
+      "loss": 3.0035,
+      "step": 3030
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.05721277547909e-07,
+      "loss": 3.0153,
+      "step": 3035
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.927496236351501e-07,
+      "loss": 3.015,
+      "step": 3040
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.8005507212512164e-07,
+      "loss": 3.0029,
+      "step": 3045
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.676379853954858e-07,
+      "loss": 2.9836,
+      "step": 3050
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.554987179034218e-07,
+      "loss": 2.9989,
+      "step": 3055
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.436376161755005e-07,
+      "loss": 3.0051,
+      "step": 3060
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.3205501879779546e-07,
+      "loss": 3.0094,
+      "step": 3065
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.2075125640621531e-07,
+      "loss": 3.0073,
+      "step": 3070
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.097266516770713e-07,
+      "loss": 2.9887,
+      "step": 3075
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9898151931785682e-07,
+      "loss": 3.0277,
+      "step": 3080
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.885161660582746e-07,
+      "loss": 3.025,
+      "step": 3085
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7833089064146825e-07,
+      "loss": 3.0513,
+      "step": 3090
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6842598381551e-07,
+      "loss": 3.0129,
+      "step": 3095
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5880172832508644e-07,
+      "loss": 3.005,
+      "step": 3100
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.494583989034326e-07,
+      "loss": 2.9974,
+      "step": 3105
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.40396262264495e-07,
+      "loss": 3.0213,
+      "step": 3110
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.3161557709530982e-07,
+      "loss": 3.0189,
+      "step": 3115
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.231165940486234e-07,
+      "loss": 3.0322,
+      "step": 3120
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1489955573573241e-07,
+      "loss": 2.9922,
+      "step": 3125
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0696469671956588e-07,
+      "loss": 3.0492,
+      "step": 3130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.931224350798185e-08,
+      "loss": 2.9918,
+      "step": 3135
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.19424145473058e-08,
+      "loss": 3.0176,
+      "step": 3140
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.485542021609117e-08,
+      "loss": 3.045,
+      "step": 3145
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.805146281912202e-08,
+      "loss": 3.0608,
+      "step": 3150
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.153073658162646e-08,
+      "loss": 3.0038,
+      "step": 3155
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.529342764374225e-08,
+      "loss": 3.0085,
+      "step": 3160
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.933971405519656e-08,
+      "loss": 3.0356,
+      "step": 3165
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.3669765770228974e-08,
+      "loss": 2.9979,
+      "step": 3170
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.828374464273422e-08,
+      "loss": 2.9973,
+      "step": 3175
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.318180442164588e-08,
+      "loss": 3.0554,
+      "step": 3180
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8364090746547676e-08,
+      "loss": 2.9875,
+      "step": 3185
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.383074114351237e-08,
+      "loss": 3.02,
+      "step": 3190
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9581885021181534e-08,
+      "loss": 3.0043,
+      "step": 3195
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5617643667065207e-08,
+      "loss": 2.9746,
+      "step": 3200
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1938130244083535e-08,
+      "loss": 3.025,
+      "step": 3205
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.854344978733824e-08,
+      "loss": 2.9946,
+      "step": 3210
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5433699201108377e-08,
+      "loss": 2.9883,
+      "step": 3215
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2608967256088067e-08,
+      "loss": 2.9877,
+      "step": 3220
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0069334586854106e-08,
+      "loss": 3.0237,
+      "step": 3225
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.8148736895578e-09,
+      "loss": 3.0206,
+      "step": 3230
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.845648919863278e-09,
+      "loss": 2.9899,
+      "step": 3235
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.161716491105639e-09,
+      "loss": 2.9984,
+      "step": 3240
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.763124472685563e-09,
+      "loss": 3.0315,
+      "step": 3245
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6499127887026346e-09,
+      "loss": 3.0021,
+      "step": 3250
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.221132168073631e-10,
+      "loss": 3.0235,
+      "step": 3255
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.797493873019086e-10,
+      "loss": 3.0035,
+      "step": 3260
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.283678246284282e-11,
+      "loss": 2.99,
+      "step": 3265
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 3.044856548309326,
+      "eval_runtime": 298.6451,
+      "eval_samples_per_second": 51.67,
+      "eval_steps_per_second": 0.81,
+      "step": 3267
+    },
+    {
+      "epoch": 3.0,
+      "step": 3267,
+      "total_flos": 3.65384602859539e+19,
+      "train_loss": 3.471017555902958,
+      "train_runtime": 35969.3936,
+      "train_samples_per_second": 11.629,
+      "train_steps_per_second": 0.091
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 3267,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 1000,
+  "total_flos": 3.65384602859539e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}