Commit
•
62bd1a3
1
Parent(s):
90d8786
Model save
Browse files- README.md +37 -37
- all_results.json +7 -7
- generation_config.json +1 -1
- train_results.json +7 -7
- trainer_state.json +521 -2488
README.md
CHANGED
|
@@ -1,57 +1,57 @@
|
|
| 1 |
---
|
| 2 |
-
license: gemma
|
| 3 |
base_model: google/gemma-2-2b-it
|
|
|
|
|
|
|
| 4 |
tags:
|
|
|
|
| 5 |
- trl
|
| 6 |
- sft
|
| 7 |
-
|
| 8 |
-
model-index:
|
| 9 |
-
- name: gemma-2-2b-it-sft-m
|
| 10 |
-
results: []
|
| 11 |
---
|
| 12 |
|
| 13 |
-
|
| 14 |
-
should probably proofread and complete it, then remove this comment. -->
|
| 15 |
-
|
| 16 |
-
# gemma-2-2b-it-sft-m
|
| 17 |
-
|
| 18 |
-
This model is a fine-tuned version of [google/gemma-2-2b-it](https://huggingface.co/google/gemma-2-2b-it) on an unknown dataset.
|
| 19 |
-
|
| 20 |
-
## Model description
|
| 21 |
|
| 22 |
-
|
|
|
|
| 23 |
|
| 24 |
-
##
|
| 25 |
|
| 26 |
-
|
|
|
|
| 27 |
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
|
|
|
|
|
|
| 31 |
|
| 32 |
## Training procedure
|
| 33 |
|
| 34 |
-
|
|
|
|
|
|
|
| 35 |
|
| 36 |
-
|
| 37 |
-
- learning_rate: 2e-05
|
| 38 |
-
- train_batch_size: 8
|
| 39 |
-
- eval_batch_size: 8
|
| 40 |
-
- seed: 42
|
| 41 |
-
- gradient_accumulation_steps: 4
|
| 42 |
-
- total_train_batch_size: 32
|
| 43 |
-
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 44 |
-
- lr_scheduler_type: cosine
|
| 45 |
-
- lr_scheduler_warmup_ratio: 0.1
|
| 46 |
-
- num_epochs: 6
|
| 47 |
|
| 48 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 49 |
|
|
|
|
| 50 |
|
| 51 |
|
| 52 |
-
### Framework versions
|
| 53 |
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
|
|
|
| 2 |
base_model: google/gemma-2-2b-it
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: gemma-2-2b-it-sft-m
|
| 5 |
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
- trl
|
| 8 |
- sft
|
| 9 |
+
licence: license
|
|
|
|
|
|
|
|
|
|
| 10 |
---
|
| 11 |
|
| 12 |
+
# Model Card for gemma-2-2b-it-sft-m
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
|
| 14 |
+
This model is a fine-tuned version of [google/gemma-2-2b-it](https://huggingface.co/google/gemma-2-2b-it).
|
| 15 |
+
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 16 |
|
| 17 |
+
## Quick start
|
| 18 |
|
| 19 |
+
```python
|
| 20 |
+
from transformers import pipeline
|
| 21 |
|
| 22 |
+
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
|
| 23 |
+
generator = pipeline("text-generation", model="ZhangShenao/gemma-2-2b-it-sft-m", device="cuda")
|
| 24 |
+
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
|
| 25 |
+
print(output["generated_text"])
|
| 26 |
+
```
|
| 27 |
|
| 28 |
## Training procedure
|
| 29 |
|
| 30 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/yutongyin/huggingface/runs/7oikn82h)
|
| 31 |
+
|
| 32 |
+
This model was trained with SFT.
|
| 33 |
|
| 34 |
+
### Framework versions
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
|
| 36 |
+
- TRL: 0.12.0
|
| 37 |
+
- Transformers: 4.46.1
|
| 38 |
+
- Pytorch: 2.4.0
|
| 39 |
+
- Datasets: 3.0.2
|
| 40 |
+
- Tokenizers: 0.20.1
|
| 41 |
|
| 42 |
+
## Citations
|
| 43 |
|
| 44 |
|
|
|
|
| 45 |
|
| 46 |
+
Cite TRL as:
|
| 47 |
+
|
| 48 |
+
```bibtex
|
| 49 |
+
@misc{vonwerra2022trl,
|
| 50 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 51 |
+
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
|
| 52 |
+
year = 2020,
|
| 53 |
+
journal = {GitHub repository},
|
| 54 |
+
publisher = {GitHub},
|
| 55 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 56 |
+
}
|
| 57 |
+
```
|
all_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch": 5.
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples":
|
| 7 |
-
"train_samples_per_second": 8.
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 5.958236658932715,
|
| 3 |
+
"total_flos": 6.526533872789914e+16,
|
| 4 |
+
"train_loss": 0.40725265422435564,
|
| 5 |
+
"train_runtime": 2315.0788,
|
| 6 |
+
"train_samples": 3448,
|
| 7 |
+
"train_samples_per_second": 8.936,
|
| 8 |
+
"train_steps_per_second": 0.277
|
| 9 |
}
|
generation_config.json
CHANGED
|
@@ -7,5 +7,5 @@
|
|
| 7 |
107
|
| 8 |
],
|
| 9 |
"pad_token_id": 0,
|
| 10 |
-
"transformers_version": "4.
|
| 11 |
}
|
|
|
|
| 7 |
107
|
| 8 |
],
|
| 9 |
"pad_token_id": 0,
|
| 10 |
+
"transformers_version": "4.46.1"
|
| 11 |
}
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch": 5.
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples":
|
| 7 |
-
"train_samples_per_second": 8.
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 5.958236658932715,
|
| 3 |
+
"total_flos": 6.526533872789914e+16,
|
| 4 |
+
"train_loss": 0.40725265422435564,
|
| 5 |
+
"train_runtime": 2315.0788,
|
| 6 |
+
"train_samples": 3448,
|
| 7 |
+
"train_samples_per_second": 8.936,
|
| 8 |
+
"train_steps_per_second": 0.277
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -1,2888 +1,921 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 5.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
-
"epoch": 0.
|
| 13 |
-
"grad_norm":
|
| 14 |
-
"learning_rate":
|
| 15 |
-
"loss": 2.
|
| 16 |
"step": 5
|
| 17 |
},
|
| 18 |
{
|
| 19 |
-
"epoch": 0.
|
| 20 |
-
"grad_norm":
|
| 21 |
-
"learning_rate":
|
| 22 |
-
"loss": 2.
|
| 23 |
"step": 10
|
| 24 |
},
|
| 25 |
{
|
| 26 |
-
"epoch": 0.
|
| 27 |
-
"grad_norm":
|
| 28 |
-
"learning_rate":
|
| 29 |
-
"loss":
|
| 30 |
"step": 15
|
| 31 |
},
|
| 32 |
{
|
| 33 |
-
"epoch": 0.
|
| 34 |
-
"grad_norm":
|
| 35 |
-
"learning_rate":
|
| 36 |
-
"loss":
|
| 37 |
"step": 20
|
| 38 |
},
|
| 39 |
{
|
| 40 |
-
"epoch": 0.
|
| 41 |
-
"grad_norm":
|
| 42 |
-
"learning_rate":
|
| 43 |
-
"loss": 1.
|
| 44 |
"step": 25
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 0.
|
| 48 |
-
"grad_norm":
|
| 49 |
-
"learning_rate":
|
| 50 |
-
"loss": 1.
|
| 51 |
"step": 30
|
| 52 |
},
|
| 53 |
{
|
| 54 |
-
"epoch": 0.
|
| 55 |
-
"grad_norm":
|
| 56 |
-
"learning_rate":
|
| 57 |
-
"loss":
|
| 58 |
"step": 35
|
| 59 |
},
|
| 60 |
{
|
| 61 |
-
"epoch": 0.
|
| 62 |
-
"grad_norm":
|
| 63 |
-
"learning_rate":
|
| 64 |
-
"loss":
|
| 65 |
"step": 40
|
| 66 |
},
|
| 67 |
{
|
| 68 |
-
"epoch": 0.
|
| 69 |
-
"grad_norm":
|
| 70 |
-
"learning_rate":
|
| 71 |
-
"loss":
|
| 72 |
"step": 45
|
| 73 |
},
|
| 74 |
{
|
| 75 |
-
"epoch": 0.
|
| 76 |
-
"grad_norm":
|
| 77 |
-
"learning_rate":
|
| 78 |
-
"loss":
|
| 79 |
"step": 50
|
| 80 |
},
|
| 81 |
{
|
| 82 |
-
"epoch": 0.
|
| 83 |
-
"grad_norm":
|
| 84 |
-
"learning_rate":
|
| 85 |
-
"loss":
|
| 86 |
"step": 55
|
| 87 |
},
|
| 88 |
{
|
| 89 |
-
"epoch": 0.
|
| 90 |
-
"grad_norm":
|
| 91 |
-
"learning_rate":
|
| 92 |
-
"loss":
|
| 93 |
"step": 60
|
| 94 |
},
|
| 95 |
{
|
| 96 |
-
"epoch": 0.
|
| 97 |
-
"grad_norm": 3.
|
| 98 |
-
"learning_rate":
|
| 99 |
-
"loss": 0.
|
| 100 |
"step": 65
|
| 101 |
},
|
| 102 |
{
|
| 103 |
-
"epoch": 0.
|
| 104 |
-
"grad_norm": 3.
|
| 105 |
-
"learning_rate":
|
| 106 |
-
"loss": 0.
|
| 107 |
"step": 70
|
| 108 |
},
|
| 109 |
{
|
| 110 |
-
"epoch": 0.
|
| 111 |
-
"grad_norm": 3.
|
| 112 |
-
"learning_rate":
|
| 113 |
-
"loss": 0.
|
| 114 |
"step": 75
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm": 3.
|
| 119 |
-
"learning_rate":
|
| 120 |
-
"loss": 0.
|
| 121 |
"step": 80
|
| 122 |
},
|
| 123 |
{
|
| 124 |
-
"epoch": 0.
|
| 125 |
-
"grad_norm": 3.
|
| 126 |
-
"learning_rate":
|
| 127 |
-
"loss": 0.
|
| 128 |
"step": 85
|
| 129 |
},
|
| 130 |
{
|
| 131 |
-
"epoch": 0.
|
| 132 |
-
"grad_norm": 3.
|
| 133 |
-
"learning_rate":
|
| 134 |
-
"loss": 0.
|
| 135 |
"step": 90
|
| 136 |
},
|
| 137 |
{
|
| 138 |
-
"epoch": 0.
|
| 139 |
-
"grad_norm": 3.
|
| 140 |
-
"learning_rate":
|
| 141 |
-
"loss": 0.
|
| 142 |
"step": 95
|
| 143 |
},
|
| 144 |
{
|
| 145 |
-
"epoch": 0.
|
| 146 |
-
"grad_norm":
|
| 147 |
-
"learning_rate":
|
| 148 |
-
"loss": 0.
|
| 149 |
"step": 100
|
| 150 |
},
|
| 151 |
{
|
| 152 |
-
"epoch": 0.
|
| 153 |
-
"grad_norm": 3.
|
| 154 |
-
"learning_rate": 1.
|
| 155 |
-
"loss": 0.
|
| 156 |
"step": 105
|
| 157 |
},
|
| 158 |
{
|
| 159 |
-
"epoch":
|
| 160 |
-
"grad_norm":
|
| 161 |
-
"learning_rate": 1.
|
| 162 |
-
"loss": 0.
|
| 163 |
"step": 110
|
| 164 |
},
|
| 165 |
{
|
| 166 |
-
"epoch":
|
| 167 |
-
"grad_norm":
|
| 168 |
-
"learning_rate": 1.
|
| 169 |
-
"loss": 0.
|
| 170 |
"step": 115
|
| 171 |
},
|
| 172 |
{
|
| 173 |
-
"epoch":
|
| 174 |
-
"grad_norm":
|
| 175 |
-
"learning_rate": 1.
|
| 176 |
-
"loss": 0.
|
| 177 |
"step": 120
|
| 178 |
},
|
| 179 |
{
|
| 180 |
-
"epoch":
|
| 181 |
-
"grad_norm": 3.
|
| 182 |
-
"learning_rate": 1.
|
| 183 |
-
"loss": 0.
|
| 184 |
"step": 125
|
| 185 |
},
|
| 186 |
{
|
| 187 |
-
"epoch":
|
| 188 |
-
"grad_norm": 3.
|
| 189 |
-
"learning_rate": 1.
|
| 190 |
-
"loss": 0.
|
| 191 |
"step": 130
|
| 192 |
},
|
| 193 |
{
|
| 194 |
-
"epoch":
|
| 195 |
-
"grad_norm":
|
| 196 |
-
"learning_rate": 1.
|
| 197 |
-
"loss": 0.
|
| 198 |
"step": 135
|
| 199 |
},
|
| 200 |
{
|
| 201 |
-
"epoch":
|
| 202 |
-
"grad_norm": 3.
|
| 203 |
-
"learning_rate": 1.
|
| 204 |
-
"loss": 0.
|
| 205 |
"step": 140
|
| 206 |
},
|
| 207 |
{
|
| 208 |
-
"epoch":
|
| 209 |
-
"grad_norm":
|
| 210 |
-
"learning_rate": 1.
|
| 211 |
-
"loss": 0.
|
| 212 |
"step": 145
|
| 213 |
},
|
| 214 |
{
|
| 215 |
-
"epoch":
|
| 216 |
-
"grad_norm":
|
| 217 |
-
"learning_rate": 1.
|
| 218 |
-
"loss": 0.
|
| 219 |
"step": 150
|
| 220 |
},
|
| 221 |
{
|
| 222 |
-
"epoch":
|
| 223 |
-
"grad_norm": 3.
|
| 224 |
-
"learning_rate": 1.
|
| 225 |
-
"loss": 0.
|
| 226 |
"step": 155
|
| 227 |
},
|
| 228 |
{
|
| 229 |
-
"epoch":
|
| 230 |
-
"grad_norm": 3.
|
| 231 |
-
"learning_rate": 1.
|
| 232 |
-
"loss": 0.
|
| 233 |
"step": 160
|
| 234 |
},
|
| 235 |
{
|
| 236 |
-
"epoch":
|
| 237 |
-
"grad_norm": 3.
|
| 238 |
-
"learning_rate": 1.
|
| 239 |
-
"loss": 0.
|
| 240 |
"step": 165
|
| 241 |
},
|
| 242 |
{
|
| 243 |
-
"epoch":
|
| 244 |
-
"grad_norm": 3.
|
| 245 |
-
"learning_rate": 1.
|
| 246 |
-
"loss": 0.
|
| 247 |
"step": 170
|
| 248 |
},
|
| 249 |
{
|
| 250 |
-
"epoch":
|
| 251 |
-
"grad_norm": 3.
|
| 252 |
-
"learning_rate": 1.
|
| 253 |
-
"loss": 0.
|
| 254 |
"step": 175
|
| 255 |
},
|
| 256 |
{
|
| 257 |
-
"epoch":
|
| 258 |
-
"grad_norm": 3.
|
| 259 |
-
"learning_rate": 1.
|
| 260 |
-
"loss": 0.
|
| 261 |
"step": 180
|
| 262 |
},
|
| 263 |
{
|
| 264 |
-
"epoch":
|
| 265 |
-
"grad_norm": 3.
|
| 266 |
-
"learning_rate": 1.
|
| 267 |
-
"loss": 0.
|
| 268 |
"step": 185
|
| 269 |
},
|
| 270 |
{
|
| 271 |
-
"epoch":
|
| 272 |
"grad_norm": 3.3125,
|
| 273 |
-
"learning_rate": 1.
|
| 274 |
-
"loss": 0.
|
| 275 |
"step": 190
|
| 276 |
},
|
| 277 |
{
|
| 278 |
-
"epoch":
|
| 279 |
-
"grad_norm": 2.
|
| 280 |
-
"learning_rate": 1.
|
| 281 |
-
"loss": 0.
|
| 282 |
"step": 195
|
| 283 |
},
|
| 284 |
{
|
| 285 |
-
"epoch":
|
| 286 |
-
"grad_norm": 3.
|
| 287 |
-
"learning_rate": 1.
|
| 288 |
-
"loss": 0.
|
| 289 |
"step": 200
|
| 290 |
},
|
| 291 |
{
|
| 292 |
-
"epoch":
|
| 293 |
-
"grad_norm":
|
| 294 |
-
"learning_rate":
|
| 295 |
-
"loss": 0.
|
| 296 |
"step": 205
|
| 297 |
},
|
| 298 |
{
|
| 299 |
-
"epoch":
|
| 300 |
-
"grad_norm": 2.
|
| 301 |
-
"learning_rate": 1.
|
| 302 |
-
"loss": 0.
|
| 303 |
"step": 210
|
| 304 |
},
|
| 305 |
{
|
| 306 |
-
"epoch":
|
| 307 |
-
"grad_norm":
|
| 308 |
-
"learning_rate": 1.
|
| 309 |
-
"loss": 0.
|
| 310 |
"step": 215
|
| 311 |
},
|
| 312 |
{
|
| 313 |
-
"epoch":
|
| 314 |
-
"grad_norm":
|
| 315 |
-
"learning_rate": 1.
|
| 316 |
-
"loss": 0.
|
| 317 |
"step": 220
|
| 318 |
},
|
| 319 |
{
|
| 320 |
-
"epoch":
|
| 321 |
-
"grad_norm":
|
| 322 |
-
"learning_rate": 1.
|
| 323 |
-
"loss": 0.
|
| 324 |
"step": 225
|
| 325 |
},
|
| 326 |
{
|
| 327 |
-
"epoch":
|
| 328 |
-
"grad_norm":
|
| 329 |
-
"learning_rate": 1.
|
| 330 |
-
"loss": 0.
|
| 331 |
"step": 230
|
| 332 |
},
|
| 333 |
{
|
| 334 |
-
"epoch":
|
| 335 |
-
"grad_norm":
|
| 336 |
-
"learning_rate": 1.
|
| 337 |
-
"loss": 0.
|
| 338 |
"step": 235
|
| 339 |
},
|
| 340 |
{
|
| 341 |
-
"epoch":
|
| 342 |
-
"grad_norm":
|
| 343 |
-
"learning_rate": 1.
|
| 344 |
-
"loss": 0.
|
| 345 |
"step": 240
|
| 346 |
},
|
| 347 |
{
|
| 348 |
-
"epoch":
|
| 349 |
-
"grad_norm":
|
| 350 |
-
"learning_rate": 1.
|
| 351 |
-
"loss": 0.
|
| 352 |
"step": 245
|
| 353 |
},
|
| 354 |
{
|
| 355 |
-
"epoch":
|
| 356 |
-
"grad_norm":
|
| 357 |
-
"learning_rate": 1.
|
| 358 |
-
"loss": 0.
|
| 359 |
"step": 250
|
| 360 |
},
|
| 361 |
{
|
| 362 |
-
"epoch":
|
| 363 |
-
"grad_norm":
|
| 364 |
-
"learning_rate": 1.
|
| 365 |
-
"loss": 0.
|
| 366 |
"step": 255
|
| 367 |
},
|
| 368 |
{
|
| 369 |
-
"epoch":
|
| 370 |
-
"grad_norm":
|
| 371 |
-
"learning_rate": 1.
|
| 372 |
-
"loss": 0.
|
| 373 |
"step": 260
|
| 374 |
},
|
| 375 |
{
|
| 376 |
-
"epoch":
|
| 377 |
-
"grad_norm":
|
| 378 |
-
"learning_rate": 1.
|
| 379 |
-
"loss": 0.
|
| 380 |
"step": 265
|
| 381 |
},
|
| 382 |
{
|
| 383 |
-
"epoch":
|
| 384 |
-
"grad_norm":
|
| 385 |
-
"learning_rate": 1.
|
| 386 |
-
"loss": 0.
|
| 387 |
"step": 270
|
| 388 |
},
|
| 389 |
{
|
| 390 |
-
"epoch":
|
| 391 |
-
"grad_norm":
|
| 392 |
-
"learning_rate": 1.
|
| 393 |
-
"loss": 0.
|
| 394 |
"step": 275
|
| 395 |
},
|
| 396 |
{
|
| 397 |
-
"epoch":
|
| 398 |
-
"grad_norm":
|
| 399 |
-
"learning_rate": 1.
|
| 400 |
-
"loss": 0.
|
| 401 |
"step": 280
|
| 402 |
},
|
| 403 |
{
|
| 404 |
-
"epoch":
|
| 405 |
-
"grad_norm":
|
| 406 |
-
"learning_rate": 1.
|
| 407 |
-
"loss": 0.
|
| 408 |
"step": 285
|
| 409 |
},
|
| 410 |
{
|
| 411 |
-
"epoch":
|
| 412 |
-
"grad_norm":
|
| 413 |
-
"learning_rate": 1.
|
| 414 |
-
"loss": 0.
|
| 415 |
"step": 290
|
| 416 |
},
|
| 417 |
{
|
| 418 |
-
"epoch":
|
| 419 |
-
"grad_norm":
|
| 420 |
-
"learning_rate": 1.
|
| 421 |
-
"loss": 0.
|
| 422 |
"step": 295
|
| 423 |
},
|
| 424 |
{
|
| 425 |
-
"epoch":
|
| 426 |
-
"grad_norm":
|
| 427 |
-
"learning_rate": 1.
|
| 428 |
-
"loss": 0.
|
| 429 |
"step": 300
|
| 430 |
},
|
| 431 |
{
|
| 432 |
-
"epoch":
|
| 433 |
-
"grad_norm":
|
| 434 |
-
"learning_rate": 1.
|
| 435 |
-
"loss": 0.
|
| 436 |
"step": 305
|
| 437 |
},
|
| 438 |
{
|
| 439 |
-
"epoch":
|
| 440 |
-
"grad_norm":
|
| 441 |
-
"learning_rate": 1.
|
| 442 |
-
"loss": 0.
|
| 443 |
"step": 310
|
| 444 |
},
|
| 445 |
{
|
| 446 |
-
"epoch":
|
| 447 |
-
"grad_norm":
|
| 448 |
-
"learning_rate": 1.
|
| 449 |
-
"loss": 0.
|
| 450 |
"step": 315
|
| 451 |
},
|
| 452 |
{
|
| 453 |
-
"epoch":
|
| 454 |
-
"grad_norm":
|
| 455 |
-
"learning_rate": 1.
|
| 456 |
-
"loss": 0.
|
| 457 |
"step": 320
|
| 458 |
},
|
| 459 |
{
|
| 460 |
-
"epoch":
|
| 461 |
-
"grad_norm":
|
| 462 |
-
"learning_rate": 1.
|
| 463 |
-
"loss": 0.
|
| 464 |
"step": 325
|
| 465 |
},
|
| 466 |
{
|
| 467 |
-
"epoch":
|
| 468 |
-
"grad_norm": 3.
|
| 469 |
-
"learning_rate": 1.
|
| 470 |
-
"loss": 0.
|
| 471 |
"step": 330
|
| 472 |
},
|
| 473 |
{
|
| 474 |
-
"epoch":
|
| 475 |
-
"grad_norm":
|
| 476 |
-
"learning_rate": 1.
|
| 477 |
-
"loss": 0.
|
| 478 |
"step": 335
|
| 479 |
},
|
| 480 |
{
|
| 481 |
-
"epoch":
|
| 482 |
-
"grad_norm":
|
| 483 |
-
"learning_rate": 1.
|
| 484 |
-
"loss": 0.
|
| 485 |
"step": 340
|
| 486 |
},
|
| 487 |
{
|
| 488 |
-
"epoch":
|
| 489 |
-
"grad_norm":
|
| 490 |
-
"learning_rate": 1.
|
| 491 |
-
"loss": 0.
|
| 492 |
"step": 345
|
| 493 |
},
|
| 494 |
{
|
| 495 |
-
"epoch":
|
| 496 |
-
"grad_norm":
|
| 497 |
-
"learning_rate": 1.
|
| 498 |
-
"loss": 0.
|
| 499 |
"step": 350
|
| 500 |
},
|
| 501 |
{
|
| 502 |
-
"epoch":
|
| 503 |
-
"grad_norm":
|
| 504 |
-
"learning_rate":
|
| 505 |
-
"loss": 0.
|
| 506 |
"step": 355
|
| 507 |
},
|
| 508 |
{
|
| 509 |
-
"epoch":
|
| 510 |
-
"grad_norm":
|
| 511 |
-
"learning_rate":
|
| 512 |
-
"loss": 0.
|
| 513 |
"step": 360
|
| 514 |
},
|
| 515 |
{
|
| 516 |
-
"epoch":
|
| 517 |
-
"grad_norm":
|
| 518 |
-
"learning_rate":
|
| 519 |
-
"loss": 0.
|
| 520 |
"step": 365
|
| 521 |
},
|
| 522 |
{
|
| 523 |
-
"epoch":
|
| 524 |
-
"grad_norm": 3.
|
| 525 |
-
"learning_rate":
|
| 526 |
-
"loss": 0.
|
| 527 |
"step": 370
|
| 528 |
},
|
| 529 |
{
|
| 530 |
-
"epoch":
|
| 531 |
-
"grad_norm":
|
| 532 |
-
"learning_rate":
|
| 533 |
-
"loss": 0.
|
| 534 |
"step": 375
|
| 535 |
},
|
| 536 |
{
|
| 537 |
-
"epoch":
|
| 538 |
-
"grad_norm":
|
| 539 |
-
"learning_rate":
|
| 540 |
-
"loss": 0.
|
| 541 |
"step": 380
|
| 542 |
},
|
| 543 |
{
|
| 544 |
-
"epoch":
|
| 545 |
-
"grad_norm":
|
| 546 |
-
"learning_rate":
|
| 547 |
-
"loss": 0.
|
| 548 |
"step": 385
|
| 549 |
},
|
| 550 |
{
|
| 551 |
-
"epoch":
|
| 552 |
-
"grad_norm":
|
| 553 |
-
"learning_rate":
|
| 554 |
-
"loss": 0.
|
| 555 |
"step": 390
|
| 556 |
},
|
| 557 |
{
|
| 558 |
-
"epoch":
|
| 559 |
-
"grad_norm": 3.
|
| 560 |
-
"learning_rate":
|
| 561 |
-
"loss": 0.
|
| 562 |
"step": 395
|
| 563 |
},
|
| 564 |
{
|
| 565 |
-
"epoch":
|
| 566 |
-
"grad_norm": 3.
|
| 567 |
-
"learning_rate":
|
| 568 |
-
"loss": 0.
|
| 569 |
"step": 400
|
| 570 |
},
|
| 571 |
{
|
| 572 |
-
"epoch":
|
| 573 |
"grad_norm": 3.78125,
|
| 574 |
-
"learning_rate":
|
| 575 |
-
"loss": 0.
|
| 576 |
"step": 405
|
| 577 |
},
|
| 578 |
{
|
| 579 |
-
"epoch":
|
| 580 |
-
"grad_norm":
|
| 581 |
-
"learning_rate":
|
| 582 |
-
"loss": 0.
|
| 583 |
"step": 410
|
| 584 |
},
|
| 585 |
{
|
| 586 |
-
"epoch":
|
| 587 |
-
"grad_norm": 3.
|
| 588 |
-
"learning_rate":
|
| 589 |
-
"loss": 0.
|
| 590 |
"step": 415
|
| 591 |
},
|
| 592 |
{
|
| 593 |
-
"epoch":
|
| 594 |
-
"grad_norm":
|
| 595 |
-
"learning_rate":
|
| 596 |
-
"loss": 0.
|
| 597 |
"step": 420
|
| 598 |
},
|
| 599 |
{
|
| 600 |
-
"epoch":
|
| 601 |
-
"grad_norm":
|
| 602 |
-
"learning_rate":
|
| 603 |
-
"loss": 0.
|
| 604 |
"step": 425
|
| 605 |
},
|
| 606 |
{
|
| 607 |
-
"epoch":
|
| 608 |
-
"grad_norm": 3.
|
| 609 |
-
"learning_rate":
|
| 610 |
-
"loss": 0.
|
| 611 |
"step": 430
|
| 612 |
},
|
| 613 |
{
|
| 614 |
-
"epoch":
|
| 615 |
-
"grad_norm":
|
| 616 |
-
"learning_rate":
|
| 617 |
-
"loss": 0.
|
| 618 |
"step": 435
|
| 619 |
},
|
| 620 |
{
|
| 621 |
-
"epoch":
|
| 622 |
-
"grad_norm": 2.
|
| 623 |
-
"learning_rate":
|
| 624 |
-
"loss": 0.
|
| 625 |
"step": 440
|
| 626 |
},
|
| 627 |
{
|
| 628 |
-
"epoch":
|
| 629 |
-
"grad_norm": 3.
|
| 630 |
-
"learning_rate":
|
| 631 |
-
"loss": 0.
|
| 632 |
"step": 445
|
| 633 |
},
|
| 634 |
{
|
| 635 |
-
"epoch":
|
| 636 |
-
"grad_norm": 3.
|
| 637 |
-
"learning_rate":
|
| 638 |
-
"loss": 0.
|
| 639 |
"step": 450
|
| 640 |
},
|
| 641 |
{
|
| 642 |
-
"epoch":
|
| 643 |
-
"grad_norm":
|
| 644 |
-
"learning_rate":
|
| 645 |
-
"loss": 0.
|
| 646 |
"step": 455
|
| 647 |
},
|
| 648 |
{
|
| 649 |
-
"epoch":
|
| 650 |
-
"grad_norm":
|
| 651 |
-
"learning_rate":
|
| 652 |
-
"loss": 0.
|
| 653 |
"step": 460
|
| 654 |
},
|
| 655 |
{
|
| 656 |
-
"epoch":
|
| 657 |
-
"grad_norm":
|
| 658 |
-
"learning_rate":
|
| 659 |
-
"loss": 0.
|
| 660 |
"step": 465
|
| 661 |
},
|
| 662 |
{
|
| 663 |
-
"epoch":
|
| 664 |
-
"grad_norm": 3.
|
| 665 |
-
"learning_rate":
|
| 666 |
-
"loss": 0.
|
| 667 |
"step": 470
|
| 668 |
},
|
| 669 |
{
|
| 670 |
-
"epoch":
|
| 671 |
-
"grad_norm":
|
| 672 |
-
"learning_rate":
|
| 673 |
-
"loss": 0.
|
| 674 |
"step": 475
|
| 675 |
},
|
| 676 |
{
|
| 677 |
-
"epoch":
|
| 678 |
-
"grad_norm":
|
| 679 |
-
"learning_rate":
|
| 680 |
-
"loss": 0.
|
| 681 |
"step": 480
|
| 682 |
},
|
| 683 |
{
|
| 684 |
-
"epoch":
|
| 685 |
-
"grad_norm":
|
| 686 |
-
"learning_rate":
|
| 687 |
-
"loss": 0.
|
| 688 |
"step": 485
|
| 689 |
},
|
| 690 |
{
|
| 691 |
-
"epoch":
|
| 692 |
-
"grad_norm": 3.
|
| 693 |
-
"learning_rate":
|
| 694 |
-
"loss": 0.
|
| 695 |
"step": 490
|
| 696 |
},
|
| 697 |
{
|
| 698 |
-
"epoch":
|
| 699 |
-
"grad_norm":
|
| 700 |
-
"learning_rate":
|
| 701 |
-
"loss": 0.
|
| 702 |
"step": 495
|
| 703 |
},
|
| 704 |
{
|
| 705 |
-
"epoch":
|
| 706 |
-
"grad_norm":
|
| 707 |
-
"learning_rate":
|
| 708 |
-
"loss": 0.
|
| 709 |
"step": 500
|
| 710 |
},
|
| 711 |
{
|
| 712 |
-
"epoch":
|
| 713 |
-
"grad_norm":
|
| 714 |
-
"learning_rate":
|
| 715 |
-
"loss": 0.
|
| 716 |
"step": 505
|
| 717 |
},
|
| 718 |
{
|
| 719 |
-
"epoch":
|
| 720 |
-
"grad_norm": 3.
|
| 721 |
-
"learning_rate":
|
| 722 |
-
"loss": 0.
|
| 723 |
"step": 510
|
| 724 |
},
|
| 725 |
{
|
| 726 |
-
"epoch":
|
| 727 |
-
"grad_norm":
|
| 728 |
-
"learning_rate":
|
| 729 |
-
"loss": 0.
|
| 730 |
"step": 515
|
| 731 |
},
|
| 732 |
{
|
| 733 |
-
"epoch":
|
| 734 |
-
"grad_norm": 3.
|
| 735 |
-
"learning_rate":
|
| 736 |
-
"loss": 0.
|
| 737 |
"step": 520
|
| 738 |
},
|
| 739 |
{
|
| 740 |
-
"epoch":
|
| 741 |
-
"grad_norm":
|
| 742 |
-
"learning_rate": 1.
|
| 743 |
-
"loss": 0.
|
| 744 |
"step": 525
|
| 745 |
},
|
| 746 |
{
|
| 747 |
-
"epoch":
|
| 748 |
-
"grad_norm":
|
| 749 |
-
"learning_rate": 1.
|
| 750 |
-
"loss": 0.
|
| 751 |
"step": 530
|
| 752 |
},
|
| 753 |
{
|
| 754 |
-
"epoch":
|
| 755 |
-
"grad_norm":
|
| 756 |
-
"learning_rate": 1.
|
| 757 |
-
"loss": 0.
|
| 758 |
"step": 535
|
| 759 |
},
|
| 760 |
{
|
| 761 |
-
"epoch":
|
| 762 |
-
"grad_norm":
|
| 763 |
-
"learning_rate": 1.
|
| 764 |
-
"loss": 0.
|
| 765 |
"step": 540
|
| 766 |
},
|
| 767 |
{
|
| 768 |
-
"epoch":
|
| 769 |
-
"grad_norm":
|
| 770 |
-
"learning_rate": 1.
|
| 771 |
-
"loss": 0.
|
| 772 |
"step": 545
|
| 773 |
},
|
| 774 |
{
|
| 775 |
-
"epoch":
|
| 776 |
-
"grad_norm":
|
| 777 |
-
"learning_rate": 1.
|
| 778 |
-
"loss": 0.
|
| 779 |
"step": 550
|
| 780 |
},
|
| 781 |
{
|
| 782 |
-
"epoch":
|
| 783 |
-
"grad_norm":
|
| 784 |
-
"learning_rate": 1.
|
| 785 |
-
"loss": 0.
|
| 786 |
"step": 555
|
| 787 |
},
|
| 788 |
{
|
| 789 |
-
"epoch":
|
| 790 |
-
"grad_norm":
|
| 791 |
-
"learning_rate":
|
| 792 |
-
"loss": 0.
|
| 793 |
"step": 560
|
| 794 |
},
|
| 795 |
{
|
| 796 |
-
"epoch":
|
| 797 |
-
"grad_norm":
|
| 798 |
-
"learning_rate":
|
| 799 |
-
"loss": 0.
|
| 800 |
"step": 565
|
| 801 |
},
|
| 802 |
{
|
| 803 |
-
"epoch":
|
| 804 |
-
"grad_norm":
|
| 805 |
-
"learning_rate":
|
| 806 |
-
"loss": 0.
|
| 807 |
"step": 570
|
| 808 |
},
|
| 809 |
{
|
| 810 |
-
"epoch":
|
| 811 |
-
"grad_norm":
|
| 812 |
-
"learning_rate":
|
| 813 |
-
"loss": 0.
|
| 814 |
"step": 575
|
| 815 |
},
|
| 816 |
{
|
| 817 |
-
"epoch":
|
| 818 |
-
"grad_norm":
|
| 819 |
-
"learning_rate":
|
| 820 |
-
"loss": 0.
|
| 821 |
"step": 580
|
| 822 |
},
|
| 823 |
{
|
| 824 |
-
"epoch":
|
| 825 |
-
"grad_norm":
|
| 826 |
-
"learning_rate":
|
| 827 |
-
"loss": 0.
|
| 828 |
"step": 585
|
| 829 |
},
|
| 830 |
{
|
| 831 |
-
"epoch":
|
| 832 |
-
"grad_norm":
|
| 833 |
-
"learning_rate":
|
| 834 |
-
"loss": 0.
|
| 835 |
"step": 590
|
| 836 |
},
|
| 837 |
{
|
| 838 |
-
"epoch":
|
| 839 |
-
"grad_norm":
|
| 840 |
-
"learning_rate":
|
| 841 |
-
"loss": 0.
|
| 842 |
"step": 595
|
| 843 |
},
|
| 844 |
{
|
| 845 |
-
"epoch":
|
| 846 |
-
"grad_norm":
|
| 847 |
-
"learning_rate":
|
| 848 |
-
"loss": 0.
|
| 849 |
"step": 600
|
| 850 |
},
|
| 851 |
{
|
| 852 |
-
"epoch":
|
| 853 |
-
"grad_norm":
|
| 854 |
-
"learning_rate":
|
| 855 |
-
"loss": 0.
|
| 856 |
"step": 605
|
| 857 |
},
|
| 858 |
{
|
| 859 |
-
"epoch":
|
| 860 |
-
"grad_norm":
|
| 861 |
-
"learning_rate": 1.
|
| 862 |
-
"loss": 0.
|
| 863 |
"step": 610
|
| 864 |
},
|
| 865 |
{
|
| 866 |
-
"epoch":
|
| 867 |
-
"grad_norm":
|
| 868 |
-
"learning_rate": 1.
|
| 869 |
-
"loss": 0.
|
| 870 |
"step": 615
|
| 871 |
},
|
| 872 |
{
|
| 873 |
-
"epoch":
|
| 874 |
-
"grad_norm":
|
| 875 |
-
"learning_rate":
|
| 876 |
-
"loss": 0.
|
| 877 |
"step": 620
|
| 878 |
},
|
| 879 |
{
|
| 880 |
-
"epoch":
|
| 881 |
-
"grad_norm":
|
| 882 |
-
"learning_rate":
|
| 883 |
-
"loss": 0.
|
| 884 |
"step": 625
|
| 885 |
},
|
| 886 |
{
|
| 887 |
-
"epoch":
|
| 888 |
-
"grad_norm":
|
| 889 |
-
"learning_rate":
|
| 890 |
-
"loss": 0.
|
| 891 |
"step": 630
|
| 892 |
},
|
| 893 |
{
|
| 894 |
-
"epoch":
|
| 895 |
-
"grad_norm":
|
| 896 |
-
"learning_rate":
|
| 897 |
-
"loss": 0.
|
| 898 |
"step": 635
|
| 899 |
},
|
| 900 |
{
|
| 901 |
-
"epoch":
|
| 902 |
-
"grad_norm":
|
| 903 |
-
"learning_rate":
|
| 904 |
-
"loss": 0.
|
| 905 |
"step": 640
|
| 906 |
},
|
| 907 |
{
|
| 908 |
-
"epoch":
|
| 909 |
-
"
|
| 910 |
-
"
|
| 911 |
-
"
|
| 912 |
-
"
|
| 913 |
-
|
| 914 |
-
|
| 915 |
-
"epoch": 1.903367496339678,
|
| 916 |
-
"grad_norm": 3.4375,
|
| 917 |
-
"learning_rate": 1.7252666859365525e-05,
|
| 918 |
-
"loss": 0.5579,
|
| 919 |
-
"step": 650
|
| 920 |
-
},
|
| 921 |
-
{
|
| 922 |
-
"epoch": 1.91800878477306,
|
| 923 |
-
"grad_norm": 4.09375,
|
| 924 |
-
"learning_rate": 1.719366142191318e-05,
|
| 925 |
-
"loss": 0.5111,
|
| 926 |
-
"step": 655
|
| 927 |
-
},
|
| 928 |
-
{
|
| 929 |
-
"epoch": 1.9326500732064422,
|
| 930 |
-
"grad_norm": 3.8125,
|
| 931 |
-
"learning_rate": 1.7134132288124464e-05,
|
| 932 |
-
"loss": 0.5121,
|
| 933 |
-
"step": 660
|
| 934 |
-
},
|
| 935 |
-
{
|
| 936 |
-
"epoch": 1.9472913616398242,
|
| 937 |
-
"grad_norm": 3.5,
|
| 938 |
-
"learning_rate": 1.707408379170199e-05,
|
| 939 |
-
"loss": 0.4807,
|
| 940 |
-
"step": 665
|
| 941 |
-
},
|
| 942 |
-
{
|
| 943 |
-
"epoch": 1.9619326500732064,
|
| 944 |
-
"grad_norm": 3.8125,
|
| 945 |
-
"learning_rate": 1.7013520304157802e-05,
|
| 946 |
-
"loss": 0.5231,
|
| 947 |
-
"step": 670
|
| 948 |
-
},
|
| 949 |
-
{
|
| 950 |
-
"epoch": 1.9765739385065886,
|
| 951 |
-
"grad_norm": 4.125,
|
| 952 |
-
"learning_rate": 1.6952446234495145e-05,
|
| 953 |
-
"loss": 0.5067,
|
| 954 |
-
"step": 675
|
| 955 |
-
},
|
| 956 |
-
{
|
| 957 |
-
"epoch": 1.9912152269399708,
|
| 958 |
-
"grad_norm": 3.984375,
|
| 959 |
-
"learning_rate": 1.6890866028887484e-05,
|
| 960 |
-
"loss": 0.5486,
|
| 961 |
-
"step": 680
|
| 962 |
-
},
|
| 963 |
-
{
|
| 964 |
-
"epoch": 2.005856515373353,
|
| 965 |
-
"grad_norm": 3.390625,
|
| 966 |
-
"learning_rate": 1.682878417035483e-05,
|
| 967 |
-
"loss": 0.4237,
|
| 968 |
-
"step": 685
|
| 969 |
-
},
|
| 970 |
-
{
|
| 971 |
-
"epoch": 2.020497803806735,
|
| 972 |
-
"grad_norm": 4.5625,
|
| 973 |
-
"learning_rate": 1.676620517843736e-05,
|
| 974 |
-
"loss": 0.2785,
|
| 975 |
-
"step": 690
|
| 976 |
-
},
|
| 977 |
-
{
|
| 978 |
-
"epoch": 2.035139092240117,
|
| 979 |
-
"grad_norm": 3.703125,
|
| 980 |
-
"learning_rate": 1.6703133608866415e-05,
|
| 981 |
-
"loss": 0.2647,
|
| 982 |
-
"step": 695
|
| 983 |
-
},
|
| 984 |
-
{
|
| 985 |
-
"epoch": 2.049780380673499,
|
| 986 |
-
"grad_norm": 3.375,
|
| 987 |
-
"learning_rate": 1.6639574053232844e-05,
|
| 988 |
-
"loss": 0.2726,
|
| 989 |
-
"step": 700
|
| 990 |
-
},
|
| 991 |
-
{
|
| 992 |
-
"epoch": 2.0644216691068813,
|
| 993 |
-
"grad_norm": 3.859375,
|
| 994 |
-
"learning_rate": 1.6575531138652726e-05,
|
| 995 |
-
"loss": 0.2848,
|
| 996 |
-
"step": 705
|
| 997 |
-
},
|
| 998 |
-
{
|
| 999 |
-
"epoch": 2.0790629575402635,
|
| 1000 |
-
"grad_norm": 3.65625,
|
| 1001 |
-
"learning_rate": 1.6511009527430516e-05,
|
| 1002 |
-
"loss": 0.2691,
|
| 1003 |
-
"step": 710
|
| 1004 |
-
},
|
| 1005 |
-
{
|
| 1006 |
-
"epoch": 2.0937042459736457,
|
| 1007 |
-
"grad_norm": 3.640625,
|
| 1008 |
-
"learning_rate": 1.6446013916719646e-05,
|
| 1009 |
-
"loss": 0.2744,
|
| 1010 |
-
"step": 715
|
| 1011 |
-
},
|
| 1012 |
-
{
|
| 1013 |
-
"epoch": 2.108345534407028,
|
| 1014 |
-
"grad_norm": 3.78125,
|
| 1015 |
-
"learning_rate": 1.638054903818056e-05,
|
| 1016 |
-
"loss": 0.2755,
|
| 1017 |
-
"step": 720
|
| 1018 |
-
},
|
| 1019 |
-
{
|
| 1020 |
-
"epoch": 2.12298682284041,
|
| 1021 |
-
"grad_norm": 3.78125,
|
| 1022 |
-
"learning_rate": 1.6314619657636258e-05,
|
| 1023 |
-
"loss": 0.2661,
|
| 1024 |
-
"step": 725
|
| 1025 |
-
},
|
| 1026 |
-
{
|
| 1027 |
-
"epoch": 2.1376281112737923,
|
| 1028 |
-
"grad_norm": 4.46875,
|
| 1029 |
-
"learning_rate": 1.624823057472534e-05,
|
| 1030 |
-
"loss": 0.2548,
|
| 1031 |
-
"step": 730
|
| 1032 |
-
},
|
| 1033 |
-
{
|
| 1034 |
-
"epoch": 2.1522693997071745,
|
| 1035 |
-
"grad_norm": 4.28125,
|
| 1036 |
-
"learning_rate": 1.6181386622552607e-05,
|
| 1037 |
-
"loss": 0.2666,
|
| 1038 |
-
"step": 735
|
| 1039 |
-
},
|
| 1040 |
-
{
|
| 1041 |
-
"epoch": 2.166910688140556,
|
| 1042 |
-
"grad_norm": 3.921875,
|
| 1043 |
-
"learning_rate": 1.611409266733718e-05,
|
| 1044 |
-
"loss": 0.2717,
|
| 1045 |
-
"step": 740
|
| 1046 |
-
},
|
| 1047 |
-
{
|
| 1048 |
-
"epoch": 2.1815519765739384,
|
| 1049 |
-
"grad_norm": 4.3125,
|
| 1050 |
-
"learning_rate": 1.604635360805829e-05,
|
| 1051 |
-
"loss": 0.2657,
|
| 1052 |
-
"step": 745
|
| 1053 |
-
},
|
| 1054 |
-
{
|
| 1055 |
-
"epoch": 2.1961932650073206,
|
| 1056 |
-
"grad_norm": 4.46875,
|
| 1057 |
-
"learning_rate": 1.5978174376098588e-05,
|
| 1058 |
-
"loss": 0.2823,
|
| 1059 |
-
"step": 750
|
| 1060 |
-
},
|
| 1061 |
-
{
|
| 1062 |
-
"epoch": 2.210834553440703,
|
| 1063 |
-
"grad_norm": 3.8125,
|
| 1064 |
-
"learning_rate": 1.590955993488516e-05,
|
| 1065 |
-
"loss": 0.2748,
|
| 1066 |
-
"step": 755
|
| 1067 |
-
},
|
| 1068 |
-
{
|
| 1069 |
-
"epoch": 2.225475841874085,
|
| 1070 |
-
"grad_norm": 3.765625,
|
| 1071 |
-
"learning_rate": 1.584051527952821e-05,
|
| 1072 |
-
"loss": 0.2524,
|
| 1073 |
-
"step": 760
|
| 1074 |
-
},
|
| 1075 |
-
{
|
| 1076 |
-
"epoch": 2.240117130307467,
|
| 1077 |
-
"grad_norm": 3.890625,
|
| 1078 |
-
"learning_rate": 1.577104543645738e-05,
|
| 1079 |
-
"loss": 0.2782,
|
| 1080 |
-
"step": 765
|
| 1081 |
-
},
|
| 1082 |
-
{
|
| 1083 |
-
"epoch": 2.2547584187408494,
|
| 1084 |
-
"grad_norm": 4.125,
|
| 1085 |
-
"learning_rate": 1.5701155463055858e-05,
|
| 1086 |
-
"loss": 0.2663,
|
| 1087 |
-
"step": 770
|
| 1088 |
-
},
|
| 1089 |
-
{
|
| 1090 |
-
"epoch": 2.269399707174231,
|
| 1091 |
-
"grad_norm": 4.03125,
|
| 1092 |
-
"learning_rate": 1.563085044729218e-05,
|
| 1093 |
-
"loss": 0.2968,
|
| 1094 |
-
"step": 775
|
| 1095 |
-
},
|
| 1096 |
-
{
|
| 1097 |
-
"epoch": 2.2840409956076133,
|
| 1098 |
-
"grad_norm": 4.28125,
|
| 1099 |
-
"learning_rate": 1.5560135507349848e-05,
|
| 1100 |
-
"loss": 0.2751,
|
| 1101 |
-
"step": 780
|
| 1102 |
-
},
|
| 1103 |
-
{
|
| 1104 |
-
"epoch": 2.2986822840409955,
|
| 1105 |
-
"grad_norm": 4.03125,
|
| 1106 |
-
"learning_rate": 1.5489015791254714e-05,
|
| 1107 |
-
"loss": 0.2626,
|
| 1108 |
-
"step": 785
|
| 1109 |
-
},
|
| 1110 |
-
{
|
| 1111 |
-
"epoch": 2.3133235724743777,
|
| 1112 |
-
"grad_norm": 4.90625,
|
| 1113 |
-
"learning_rate": 1.5417496476500212e-05,
|
| 1114 |
-
"loss": 0.2656,
|
| 1115 |
-
"step": 790
|
| 1116 |
-
},
|
| 1117 |
-
{
|
| 1118 |
-
"epoch": 2.32796486090776,
|
| 1119 |
-
"grad_norm": 4.375,
|
| 1120 |
-
"learning_rate": 1.5345582769670428e-05,
|
| 1121 |
-
"loss": 0.2692,
|
| 1122 |
-
"step": 795
|
| 1123 |
-
},
|
| 1124 |
-
{
|
| 1125 |
-
"epoch": 2.342606149341142,
|
| 1126 |
-
"grad_norm": 3.703125,
|
| 1127 |
-
"learning_rate": 1.5273279906061082e-05,
|
| 1128 |
-
"loss": 0.2497,
|
| 1129 |
-
"step": 800
|
| 1130 |
-
},
|
| 1131 |
-
{
|
| 1132 |
-
"epoch": 2.3572474377745243,
|
| 1133 |
-
"grad_norm": 4.125,
|
| 1134 |
-
"learning_rate": 1.5200593149298375e-05,
|
| 1135 |
-
"loss": 0.2608,
|
| 1136 |
-
"step": 805
|
| 1137 |
-
},
|
| 1138 |
-
{
|
| 1139 |
-
"epoch": 2.3718887262079065,
|
| 1140 |
-
"grad_norm": 3.625,
|
| 1141 |
-
"learning_rate": 1.512752779095582e-05,
|
| 1142 |
-
"loss": 0.2757,
|
| 1143 |
-
"step": 810
|
| 1144 |
-
},
|
| 1145 |
-
{
|
| 1146 |
-
"epoch": 2.3865300146412887,
|
| 1147 |
-
"grad_norm": 4.03125,
|
| 1148 |
-
"learning_rate": 1.5054089150169003e-05,
|
| 1149 |
-
"loss": 0.2609,
|
| 1150 |
-
"step": 815
|
| 1151 |
-
},
|
| 1152 |
-
{
|
| 1153 |
-
"epoch": 2.4011713030746704,
|
| 1154 |
-
"grad_norm": 3.765625,
|
| 1155 |
-
"learning_rate": 1.498028257324836e-05,
|
| 1156 |
-
"loss": 0.2484,
|
| 1157 |
-
"step": 820
|
| 1158 |
-
},
|
| 1159 |
-
{
|
| 1160 |
-
"epoch": 2.4158125915080526,
|
| 1161 |
-
"grad_norm": 3.671875,
|
| 1162 |
-
"learning_rate": 1.4906113433289963e-05,
|
| 1163 |
-
"loss": 0.2557,
|
| 1164 |
-
"step": 825
|
| 1165 |
-
},
|
| 1166 |
-
{
|
| 1167 |
-
"epoch": 2.430453879941435,
|
| 1168 |
-
"grad_norm": 4.375,
|
| 1169 |
-
"learning_rate": 1.4831587129784363e-05,
|
| 1170 |
-
"loss": 0.2729,
|
| 1171 |
-
"step": 830
|
| 1172 |
-
},
|
| 1173 |
-
{
|
| 1174 |
-
"epoch": 2.445095168374817,
|
| 1175 |
-
"grad_norm": 3.875,
|
| 1176 |
-
"learning_rate": 1.4756709088223508e-05,
|
| 1177 |
-
"loss": 0.2686,
|
| 1178 |
-
"step": 835
|
| 1179 |
-
},
|
| 1180 |
-
{
|
| 1181 |
-
"epoch": 2.459736456808199,
|
| 1182 |
-
"grad_norm": 3.84375,
|
| 1183 |
-
"learning_rate": 1.4681484759705764e-05,
|
| 1184 |
-
"loss": 0.269,
|
| 1185 |
-
"step": 840
|
| 1186 |
-
},
|
| 1187 |
-
{
|
| 1188 |
-
"epoch": 2.4743777452415814,
|
| 1189 |
-
"grad_norm": 4.03125,
|
| 1190 |
-
"learning_rate": 1.4605919620539082e-05,
|
| 1191 |
-
"loss": 0.2565,
|
| 1192 |
-
"step": 845
|
| 1193 |
-
},
|
| 1194 |
-
{
|
| 1195 |
-
"epoch": 2.4890190336749636,
|
| 1196 |
-
"grad_norm": 4.375,
|
| 1197 |
-
"learning_rate": 1.453001917184233e-05,
|
| 1198 |
-
"loss": 0.2632,
|
| 1199 |
-
"step": 850
|
| 1200 |
-
},
|
| 1201 |
-
{
|
| 1202 |
-
"epoch": 2.5036603221083453,
|
| 1203 |
-
"grad_norm": 3.78125,
|
| 1204 |
-
"learning_rate": 1.4453788939144793e-05,
|
| 1205 |
-
"loss": 0.2623,
|
| 1206 |
-
"step": 855
|
| 1207 |
-
},
|
| 1208 |
-
{
|
| 1209 |
-
"epoch": 2.5183016105417275,
|
| 1210 |
-
"grad_norm": 4.1875,
|
| 1211 |
-
"learning_rate": 1.4377234471983944e-05,
|
| 1212 |
-
"loss": 0.2606,
|
| 1213 |
-
"step": 860
|
| 1214 |
-
},
|
| 1215 |
-
{
|
| 1216 |
-
"epoch": 2.5329428989751097,
|
| 1217 |
-
"grad_norm": 4.40625,
|
| 1218 |
-
"learning_rate": 1.430036134350142e-05,
|
| 1219 |
-
"loss": 0.2519,
|
| 1220 |
-
"step": 865
|
| 1221 |
-
},
|
| 1222 |
-
{
|
| 1223 |
-
"epoch": 2.547584187408492,
|
| 1224 |
-
"grad_norm": 4.15625,
|
| 1225 |
-
"learning_rate": 1.4223175150037297e-05,
|
| 1226 |
-
"loss": 0.2638,
|
| 1227 |
-
"step": 870
|
| 1228 |
-
},
|
| 1229 |
-
{
|
| 1230 |
-
"epoch": 2.562225475841874,
|
| 1231 |
-
"grad_norm": 3.9375,
|
| 1232 |
-
"learning_rate": 1.4145681510722694e-05,
|
| 1233 |
-
"loss": 0.2616,
|
| 1234 |
-
"step": 875
|
| 1235 |
-
},
|
| 1236 |
-
{
|
| 1237 |
-
"epoch": 2.5768667642752563,
|
| 1238 |
-
"grad_norm": 4.34375,
|
| 1239 |
-
"learning_rate": 1.406788606707069e-05,
|
| 1240 |
-
"loss": 0.2691,
|
| 1241 |
-
"step": 880
|
| 1242 |
-
},
|
| 1243 |
-
{
|
| 1244 |
-
"epoch": 2.5915080527086385,
|
| 1245 |
-
"grad_norm": 3.71875,
|
| 1246 |
-
"learning_rate": 1.398979448256563e-05,
|
| 1247 |
-
"loss": 0.2695,
|
| 1248 |
-
"step": 885
|
| 1249 |
-
},
|
| 1250 |
-
{
|
| 1251 |
-
"epoch": 2.6061493411420207,
|
| 1252 |
-
"grad_norm": 3.890625,
|
| 1253 |
-
"learning_rate": 1.3911412442250818e-05,
|
| 1254 |
-
"loss": 0.2566,
|
| 1255 |
-
"step": 890
|
| 1256 |
-
},
|
| 1257 |
-
{
|
| 1258 |
-
"epoch": 2.620790629575403,
|
| 1259 |
-
"grad_norm": 4.03125,
|
| 1260 |
-
"learning_rate": 1.3832745652314652e-05,
|
| 1261 |
-
"loss": 0.2865,
|
| 1262 |
-
"step": 895
|
| 1263 |
-
},
|
| 1264 |
-
{
|
| 1265 |
-
"epoch": 2.6354319180087846,
|
| 1266 |
-
"grad_norm": 4.21875,
|
| 1267 |
-
"learning_rate": 1.3753799839675215e-05,
|
| 1268 |
-
"loss": 0.2768,
|
| 1269 |
-
"step": 900
|
| 1270 |
-
},
|
| 1271 |
-
{
|
| 1272 |
-
"epoch": 2.650073206442167,
|
| 1273 |
-
"grad_norm": 3.421875,
|
| 1274 |
-
"learning_rate": 1.3674580751563357e-05,
|
| 1275 |
-
"loss": 0.2611,
|
| 1276 |
-
"step": 905
|
| 1277 |
-
},
|
| 1278 |
-
{
|
| 1279 |
-
"epoch": 2.664714494875549,
|
| 1280 |
-
"grad_norm": 4.5625,
|
| 1281 |
-
"learning_rate": 1.3595094155104297e-05,
|
| 1282 |
-
"loss": 0.2618,
|
| 1283 |
-
"step": 910
|
| 1284 |
-
},
|
| 1285 |
-
{
|
| 1286 |
-
"epoch": 2.679355783308931,
|
| 1287 |
-
"grad_norm": 4.125,
|
| 1288 |
-
"learning_rate": 1.3515345836897789e-05,
|
| 1289 |
-
"loss": 0.2693,
|
| 1290 |
-
"step": 915
|
| 1291 |
-
},
|
| 1292 |
-
{
|
| 1293 |
-
"epoch": 2.6939970717423134,
|
| 1294 |
-
"grad_norm": 4.3125,
|
| 1295 |
-
"learning_rate": 1.3435341602596834e-05,
|
| 1296 |
-
"loss": 0.2566,
|
| 1297 |
-
"step": 920
|
| 1298 |
-
},
|
| 1299 |
-
{
|
| 1300 |
-
"epoch": 2.7086383601756956,
|
| 1301 |
-
"grad_norm": 3.953125,
|
| 1302 |
-
"learning_rate": 1.3355087276485055e-05,
|
| 1303 |
-
"loss": 0.2821,
|
| 1304 |
-
"step": 925
|
| 1305 |
-
},
|
| 1306 |
-
{
|
| 1307 |
-
"epoch": 2.7232796486090773,
|
| 1308 |
-
"grad_norm": 3.859375,
|
| 1309 |
-
"learning_rate": 1.3274588701052679e-05,
|
| 1310 |
-
"loss": 0.2684,
|
| 1311 |
-
"step": 930
|
| 1312 |
-
},
|
| 1313 |
-
{
|
| 1314 |
-
"epoch": 2.7379209370424595,
|
| 1315 |
-
"grad_norm": 4.46875,
|
| 1316 |
-
"learning_rate": 1.3193851736571213e-05,
|
| 1317 |
-
"loss": 0.2677,
|
| 1318 |
-
"step": 935
|
| 1319 |
-
},
|
| 1320 |
-
{
|
| 1321 |
-
"epoch": 2.7525622254758417,
|
| 1322 |
-
"grad_norm": 3.984375,
|
| 1323 |
-
"learning_rate": 1.3112882260666805e-05,
|
| 1324 |
-
"loss": 0.2551,
|
| 1325 |
-
"step": 940
|
| 1326 |
-
},
|
| 1327 |
-
{
|
| 1328 |
-
"epoch": 2.767203513909224,
|
| 1329 |
-
"grad_norm": 3.734375,
|
| 1330 |
-
"learning_rate": 1.3031686167892375e-05,
|
| 1331 |
-
"loss": 0.2522,
|
| 1332 |
-
"step": 945
|
| 1333 |
-
},
|
| 1334 |
-
{
|
| 1335 |
-
"epoch": 2.781844802342606,
|
| 1336 |
-
"grad_norm": 3.984375,
|
| 1337 |
-
"learning_rate": 1.2950269369298468e-05,
|
| 1338 |
-
"loss": 0.2594,
|
| 1339 |
-
"step": 950
|
| 1340 |
-
},
|
| 1341 |
-
{
|
| 1342 |
-
"epoch": 2.7964860907759883,
|
| 1343 |
-
"grad_norm": 4.34375,
|
| 1344 |
-
"learning_rate": 1.2868637792002952e-05,
|
| 1345 |
-
"loss": 0.2512,
|
| 1346 |
-
"step": 955
|
| 1347 |
-
},
|
| 1348 |
-
{
|
| 1349 |
-
"epoch": 2.8111273792093705,
|
| 1350 |
-
"grad_norm": 4.09375,
|
| 1351 |
-
"learning_rate": 1.278679737875952e-05,
|
| 1352 |
-
"loss": 0.2649,
|
| 1353 |
-
"step": 960
|
| 1354 |
-
},
|
| 1355 |
-
{
|
| 1356 |
-
"epoch": 2.8257686676427527,
|
| 1357 |
-
"grad_norm": 4.0,
|
| 1358 |
-
"learning_rate": 1.2704754087525051e-05,
|
| 1359 |
-
"loss": 0.2579,
|
| 1360 |
-
"step": 965
|
| 1361 |
-
},
|
| 1362 |
-
{
|
| 1363 |
-
"epoch": 2.840409956076135,
|
| 1364 |
-
"grad_norm": 4.3125,
|
| 1365 |
-
"learning_rate": 1.2622513891025889e-05,
|
| 1366 |
-
"loss": 0.2583,
|
| 1367 |
-
"step": 970
|
| 1368 |
-
},
|
| 1369 |
-
{
|
| 1370 |
-
"epoch": 2.855051244509517,
|
| 1371 |
-
"grad_norm": 4.46875,
|
| 1372 |
-
"learning_rate": 1.2540082776323009e-05,
|
| 1373 |
-
"loss": 0.2682,
|
| 1374 |
-
"step": 975
|
| 1375 |
-
},
|
| 1376 |
-
{
|
| 1377 |
-
"epoch": 2.869692532942899,
|
| 1378 |
-
"grad_norm": 4.125,
|
| 1379 |
-
"learning_rate": 1.2457466744376184e-05,
|
| 1380 |
-
"loss": 0.2644,
|
| 1381 |
-
"step": 980
|
| 1382 |
-
},
|
| 1383 |
-
{
|
| 1384 |
-
"epoch": 2.884333821376281,
|
| 1385 |
-
"grad_norm": 4.125,
|
| 1386 |
-
"learning_rate": 1.237467180960709e-05,
|
| 1387 |
-
"loss": 0.2643,
|
| 1388 |
-
"step": 985
|
| 1389 |
-
},
|
| 1390 |
-
{
|
| 1391 |
-
"epoch": 2.898975109809663,
|
| 1392 |
-
"grad_norm": 4.5,
|
| 1393 |
-
"learning_rate": 1.2291703999461498e-05,
|
| 1394 |
-
"loss": 0.26,
|
| 1395 |
-
"step": 990
|
| 1396 |
-
},
|
| 1397 |
-
{
|
| 1398 |
-
"epoch": 2.9136163982430454,
|
| 1399 |
-
"grad_norm": 3.921875,
|
| 1400 |
-
"learning_rate": 1.2208569353970422e-05,
|
| 1401 |
-
"loss": 0.2484,
|
| 1402 |
-
"step": 995
|
| 1403 |
-
},
|
| 1404 |
-
{
|
| 1405 |
-
"epoch": 2.9282576866764276,
|
| 1406 |
-
"grad_norm": 4.125,
|
| 1407 |
-
"learning_rate": 1.2125273925310465e-05,
|
| 1408 |
-
"loss": 0.2638,
|
| 1409 |
-
"step": 1000
|
| 1410 |
-
},
|
| 1411 |
-
{
|
| 1412 |
-
"epoch": 2.94289897510981,
|
| 1413 |
-
"grad_norm": 3.4375,
|
| 1414 |
-
"learning_rate": 1.2041823777363185e-05,
|
| 1415 |
-
"loss": 0.2515,
|
| 1416 |
-
"step": 1005
|
| 1417 |
-
},
|
| 1418 |
-
{
|
| 1419 |
-
"epoch": 2.9575402635431915,
|
| 1420 |
-
"grad_norm": 4.21875,
|
| 1421 |
-
"learning_rate": 1.1958224985273648e-05,
|
| 1422 |
-
"loss": 0.2432,
|
| 1423 |
-
"step": 1010
|
| 1424 |
-
},
|
| 1425 |
-
{
|
| 1426 |
-
"epoch": 2.9721815519765737,
|
| 1427 |
-
"grad_norm": 4.34375,
|
| 1428 |
-
"learning_rate": 1.1874483635008183e-05,
|
| 1429 |
-
"loss": 0.2532,
|
| 1430 |
-
"step": 1015
|
| 1431 |
-
},
|
| 1432 |
-
{
|
| 1433 |
-
"epoch": 2.986822840409956,
|
| 1434 |
-
"grad_norm": 4.03125,
|
| 1435 |
-
"learning_rate": 1.1790605822911294e-05,
|
| 1436 |
-
"loss": 0.2567,
|
| 1437 |
-
"step": 1020
|
| 1438 |
-
},
|
| 1439 |
-
{
|
| 1440 |
-
"epoch": 3.001464128843338,
|
| 1441 |
-
"grad_norm": 4.71875,
|
| 1442 |
-
"learning_rate": 1.1706597655261883e-05,
|
| 1443 |
-
"loss": 0.2133,
|
| 1444 |
-
"step": 1025
|
| 1445 |
-
},
|
| 1446 |
-
{
|
| 1447 |
-
"epoch": 3.0161054172767203,
|
| 1448 |
-
"grad_norm": 2.578125,
|
| 1449 |
-
"learning_rate": 1.1622465247828681e-05,
|
| 1450 |
-
"loss": 0.1427,
|
| 1451 |
-
"step": 1030
|
| 1452 |
-
},
|
| 1453 |
-
{
|
| 1454 |
-
"epoch": 3.0307467057101025,
|
| 1455 |
-
"grad_norm": 5.125,
|
| 1456 |
-
"learning_rate": 1.1538214725425046e-05,
|
| 1457 |
-
"loss": 0.1203,
|
| 1458 |
-
"step": 1035
|
| 1459 |
-
},
|
| 1460 |
-
{
|
| 1461 |
-
"epoch": 3.0453879941434847,
|
| 1462 |
-
"grad_norm": 4.59375,
|
| 1463 |
-
"learning_rate": 1.1453852221463058e-05,
|
| 1464 |
-
"loss": 0.1325,
|
| 1465 |
-
"step": 1040
|
| 1466 |
-
},
|
| 1467 |
-
{
|
| 1468 |
-
"epoch": 3.060029282576867,
|
| 1469 |
-
"grad_norm": 4.0625,
|
| 1470 |
-
"learning_rate": 1.1369383877507035e-05,
|
| 1471 |
-
"loss": 0.129,
|
| 1472 |
-
"step": 1045
|
| 1473 |
-
},
|
| 1474 |
-
{
|
| 1475 |
-
"epoch": 3.074670571010249,
|
| 1476 |
-
"grad_norm": 3.109375,
|
| 1477 |
-
"learning_rate": 1.1284815842826402e-05,
|
| 1478 |
-
"loss": 0.1179,
|
| 1479 |
-
"step": 1050
|
| 1480 |
-
},
|
| 1481 |
-
{
|
| 1482 |
-
"epoch": 3.089311859443631,
|
| 1483 |
-
"grad_norm": 3.609375,
|
| 1484 |
-
"learning_rate": 1.1200154273948047e-05,
|
| 1485 |
-
"loss": 0.1366,
|
| 1486 |
-
"step": 1055
|
| 1487 |
-
},
|
| 1488 |
-
{
|
| 1489 |
-
"epoch": 3.103953147877013,
|
| 1490 |
-
"grad_norm": 3.34375,
|
| 1491 |
-
"learning_rate": 1.1115405334208112e-05,
|
| 1492 |
-
"loss": 0.1242,
|
| 1493 |
-
"step": 1060
|
| 1494 |
-
},
|
| 1495 |
-
{
|
| 1496 |
-
"epoch": 3.1185944363103952,
|
| 1497 |
-
"grad_norm": 3.390625,
|
| 1498 |
-
"learning_rate": 1.1030575193303312e-05,
|
| 1499 |
-
"loss": 0.1238,
|
| 1500 |
-
"step": 1065
|
| 1501 |
-
},
|
| 1502 |
-
{
|
| 1503 |
-
"epoch": 3.1332357247437774,
|
| 1504 |
-
"grad_norm": 4.03125,
|
| 1505 |
-
"learning_rate": 1.0945670026841785e-05,
|
| 1506 |
-
"loss": 0.1238,
|
| 1507 |
-
"step": 1070
|
| 1508 |
-
},
|
| 1509 |
-
{
|
| 1510 |
-
"epoch": 3.1478770131771596,
|
| 1511 |
-
"grad_norm": 3.09375,
|
| 1512 |
-
"learning_rate": 1.0860696015893506e-05,
|
| 1513 |
-
"loss": 0.1212,
|
| 1514 |
-
"step": 1075
|
| 1515 |
-
},
|
| 1516 |
-
{
|
| 1517 |
-
"epoch": 3.162518301610542,
|
| 1518 |
-
"grad_norm": 4.40625,
|
| 1519 |
-
"learning_rate": 1.0775659346540303e-05,
|
| 1520 |
-
"loss": 0.1249,
|
| 1521 |
-
"step": 1080
|
| 1522 |
-
},
|
| 1523 |
-
{
|
| 1524 |
-
"epoch": 3.177159590043924,
|
| 1525 |
-
"grad_norm": 4.125,
|
| 1526 |
-
"learning_rate": 1.0690566209425521e-05,
|
| 1527 |
-
"loss": 0.1257,
|
| 1528 |
-
"step": 1085
|
| 1529 |
-
},
|
| 1530 |
-
{
|
| 1531 |
-
"epoch": 3.191800878477306,
|
| 1532 |
-
"grad_norm": 4.0,
|
| 1533 |
-
"learning_rate": 1.060542279930334e-05,
|
| 1534 |
-
"loss": 0.1348,
|
| 1535 |
-
"step": 1090
|
| 1536 |
-
},
|
| 1537 |
-
{
|
| 1538 |
-
"epoch": 3.206442166910688,
|
| 1539 |
-
"grad_norm": 3.984375,
|
| 1540 |
-
"learning_rate": 1.0520235314587796e-05,
|
| 1541 |
-
"loss": 0.1277,
|
| 1542 |
-
"step": 1095
|
| 1543 |
-
},
|
| 1544 |
-
{
|
| 1545 |
-
"epoch": 3.22108345534407,
|
| 1546 |
-
"grad_norm": 3.734375,
|
| 1547 |
-
"learning_rate": 1.0435009956901547e-05,
|
| 1548 |
-
"loss": 0.128,
|
| 1549 |
-
"step": 1100
|
| 1550 |
-
},
|
| 1551 |
-
{
|
| 1552 |
-
"epoch": 3.2357247437774523,
|
| 1553 |
-
"grad_norm": 4.65625,
|
| 1554 |
-
"learning_rate": 1.034975293062439e-05,
|
| 1555 |
-
"loss": 0.1258,
|
| 1556 |
-
"step": 1105
|
| 1557 |
-
},
|
| 1558 |
-
{
|
| 1559 |
-
"epoch": 3.2503660322108345,
|
| 1560 |
-
"grad_norm": 3.1875,
|
| 1561 |
-
"learning_rate": 1.026447044244158e-05,
|
| 1562 |
-
"loss": 0.1147,
|
| 1563 |
-
"step": 1110
|
| 1564 |
-
},
|
| 1565 |
-
{
|
| 1566 |
-
"epoch": 3.2650073206442167,
|
| 1567 |
-
"grad_norm": 3.375,
|
| 1568 |
-
"learning_rate": 1.0179168700892001e-05,
|
| 1569 |
-
"loss": 0.1231,
|
| 1570 |
-
"step": 1115
|
| 1571 |
-
},
|
| 1572 |
-
{
|
| 1573 |
-
"epoch": 3.279648609077599,
|
| 1574 |
-
"grad_norm": 3.609375,
|
| 1575 |
-
"learning_rate": 1.0093853915916165e-05,
|
| 1576 |
-
"loss": 0.1211,
|
| 1577 |
-
"step": 1120
|
| 1578 |
-
},
|
| 1579 |
-
{
|
| 1580 |
-
"epoch": 3.294289897510981,
|
| 1581 |
-
"grad_norm": 4.0625,
|
| 1582 |
-
"learning_rate": 1.0008532298404154e-05,
|
| 1583 |
-
"loss": 0.1245,
|
| 1584 |
-
"step": 1125
|
| 1585 |
-
},
|
| 1586 |
-
{
|
| 1587 |
-
"epoch": 3.3089311859443633,
|
| 1588 |
-
"grad_norm": 3.390625,
|
| 1589 |
-
"learning_rate": 9.923210059743447e-06,
|
| 1590 |
-
"loss": 0.1307,
|
| 1591 |
-
"step": 1130
|
| 1592 |
-
},
|
| 1593 |
-
{
|
| 1594 |
-
"epoch": 3.323572474377745,
|
| 1595 |
-
"grad_norm": 4.125,
|
| 1596 |
-
"learning_rate": 9.837893411366743e-06,
|
| 1597 |
-
"loss": 0.1287,
|
| 1598 |
-
"step": 1135
|
| 1599 |
-
},
|
| 1600 |
-
{
|
| 1601 |
-
"epoch": 3.3382137628111272,
|
| 1602 |
-
"grad_norm": 2.984375,
|
| 1603 |
-
"learning_rate": 9.752588564299776e-06,
|
| 1604 |
-
"loss": 0.1177,
|
| 1605 |
-
"step": 1140
|
| 1606 |
-
},
|
| 1607 |
-
{
|
| 1608 |
-
"epoch": 3.3528550512445094,
|
| 1609 |
-
"grad_norm": 3.796875,
|
| 1610 |
-
"learning_rate": 9.66730172870914e-06,
|
| 1611 |
-
"loss": 0.1236,
|
| 1612 |
-
"step": 1145
|
| 1613 |
-
},
|
| 1614 |
-
{
|
| 1615 |
-
"epoch": 3.3674963396778916,
|
| 1616 |
-
"grad_norm": 3.65625,
|
| 1617 |
-
"learning_rate": 9.582039113450208e-06,
|
| 1618 |
-
"loss": 0.1264,
|
| 1619 |
-
"step": 1150
|
| 1620 |
-
},
|
| 1621 |
-
{
|
| 1622 |
-
"epoch": 3.382137628111274,
|
| 1623 |
-
"grad_norm": 4.03125,
|
| 1624 |
-
"learning_rate": 9.496806925615113e-06,
|
| 1625 |
-
"loss": 0.1292,
|
| 1626 |
-
"step": 1155
|
| 1627 |
-
},
|
| 1628 |
-
{
|
| 1629 |
-
"epoch": 3.396778916544656,
|
| 1630 |
-
"grad_norm": 3.84375,
|
| 1631 |
-
"learning_rate": 9.411611370080885e-06,
|
| 1632 |
-
"loss": 0.1177,
|
| 1633 |
-
"step": 1160
|
| 1634 |
-
},
|
| 1635 |
-
{
|
| 1636 |
-
"epoch": 3.411420204978038,
|
| 1637 |
-
"grad_norm": 2.828125,
|
| 1638 |
-
"learning_rate": 9.326458649057732e-06,
|
| 1639 |
-
"loss": 0.126,
|
| 1640 |
-
"step": 1165
|
| 1641 |
-
},
|
| 1642 |
-
{
|
| 1643 |
-
"epoch": 3.42606149341142,
|
| 1644 |
-
"grad_norm": 3.21875,
|
| 1645 |
-
"learning_rate": 9.241354961637525e-06,
|
| 1646 |
-
"loss": 0.117,
|
| 1647 |
-
"step": 1170
|
| 1648 |
-
},
|
| 1649 |
-
{
|
| 1650 |
-
"epoch": 3.440702781844802,
|
| 1651 |
-
"grad_norm": 3.828125,
|
| 1652 |
-
"learning_rate": 9.156306503342499e-06,
|
| 1653 |
-
"loss": 0.1203,
|
| 1654 |
-
"step": 1175
|
| 1655 |
-
},
|
| 1656 |
-
{
|
| 1657 |
-
"epoch": 3.4553440702781844,
|
| 1658 |
-
"grad_norm": 3.640625,
|
| 1659 |
-
"learning_rate": 9.07131946567423e-06,
|
| 1660 |
-
"loss": 0.1278,
|
| 1661 |
-
"step": 1180
|
| 1662 |
-
},
|
| 1663 |
-
{
|
| 1664 |
-
"epoch": 3.4699853587115665,
|
| 1665 |
-
"grad_norm": 3.375,
|
| 1666 |
-
"learning_rate": 8.986400035662897e-06,
|
| 1667 |
-
"loss": 0.124,
|
| 1668 |
-
"step": 1185
|
| 1669 |
-
},
|
| 1670 |
-
{
|
| 1671 |
-
"epoch": 3.4846266471449487,
|
| 1672 |
-
"grad_norm": 3.5625,
|
| 1673 |
-
"learning_rate": 8.901554395416842e-06,
|
| 1674 |
-
"loss": 0.1246,
|
| 1675 |
-
"step": 1190
|
| 1676 |
-
},
|
| 1677 |
-
{
|
| 1678 |
-
"epoch": 3.499267935578331,
|
| 1679 |
-
"grad_norm": 3.5,
|
| 1680 |
-
"learning_rate": 8.816788721672565e-06,
|
| 1681 |
-
"loss": 0.1224,
|
| 1682 |
-
"step": 1195
|
| 1683 |
-
},
|
| 1684 |
-
{
|
| 1685 |
-
"epoch": 3.513909224011713,
|
| 1686 |
-
"grad_norm": 3.796875,
|
| 1687 |
-
"learning_rate": 8.732109185344995e-06,
|
| 1688 |
-
"loss": 0.1263,
|
| 1689 |
-
"step": 1200
|
| 1690 |
-
},
|
| 1691 |
-
{
|
| 1692 |
-
"epoch": 3.5285505124450953,
|
| 1693 |
-
"grad_norm": 3.390625,
|
| 1694 |
-
"learning_rate": 8.647521951078318e-06,
|
| 1695 |
-
"loss": 0.1209,
|
| 1696 |
-
"step": 1205
|
| 1697 |
-
},
|
| 1698 |
-
{
|
| 1699 |
-
"epoch": 3.5431918008784775,
|
| 1700 |
-
"grad_norm": 3.6875,
|
| 1701 |
-
"learning_rate": 8.563033176797126e-06,
|
| 1702 |
-
"loss": 0.1269,
|
| 1703 |
-
"step": 1210
|
| 1704 |
-
},
|
| 1705 |
-
{
|
| 1706 |
-
"epoch": 3.5578330893118597,
|
| 1707 |
-
"grad_norm": 3.765625,
|
| 1708 |
-
"learning_rate": 8.478649013258186e-06,
|
| 1709 |
-
"loss": 0.1332,
|
| 1710 |
-
"step": 1215
|
| 1711 |
-
},
|
| 1712 |
-
{
|
| 1713 |
-
"epoch": 3.5724743777452415,
|
| 1714 |
-
"grad_norm": 3.171875,
|
| 1715 |
-
"learning_rate": 8.394375603602602e-06,
|
| 1716 |
-
"loss": 0.1263,
|
| 1717 |
-
"step": 1220
|
| 1718 |
-
},
|
| 1719 |
-
{
|
| 1720 |
-
"epoch": 3.5871156661786237,
|
| 1721 |
-
"grad_norm": 4.28125,
|
| 1722 |
-
"learning_rate": 8.310219082908663e-06,
|
| 1723 |
-
"loss": 0.1251,
|
| 1724 |
-
"step": 1225
|
| 1725 |
-
},
|
| 1726 |
-
{
|
| 1727 |
-
"epoch": 3.601756954612006,
|
| 1728 |
-
"grad_norm": 3.625,
|
| 1729 |
-
"learning_rate": 8.226185577745149e-06,
|
| 1730 |
-
"loss": 0.125,
|
| 1731 |
-
"step": 1230
|
| 1732 |
-
},
|
| 1733 |
-
{
|
| 1734 |
-
"epoch": 3.616398243045388,
|
| 1735 |
-
"grad_norm": 3.71875,
|
| 1736 |
-
"learning_rate": 8.142281205725368e-06,
|
| 1737 |
-
"loss": 0.1213,
|
| 1738 |
-
"step": 1235
|
| 1739 |
-
},
|
| 1740 |
-
{
|
| 1741 |
-
"epoch": 3.6310395314787702,
|
| 1742 |
-
"grad_norm": 3.59375,
|
| 1743 |
-
"learning_rate": 8.058512075061758e-06,
|
| 1744 |
-
"loss": 0.1284,
|
| 1745 |
-
"step": 1240
|
| 1746 |
-
},
|
| 1747 |
-
{
|
| 1748 |
-
"epoch": 3.6456808199121524,
|
| 1749 |
-
"grad_norm": 4.28125,
|
| 1750 |
-
"learning_rate": 7.974884284121248e-06,
|
| 1751 |
-
"loss": 0.1197,
|
| 1752 |
-
"step": 1245
|
| 1753 |
-
},
|
| 1754 |
-
{
|
| 1755 |
-
"epoch": 3.660322108345534,
|
| 1756 |
-
"grad_norm": 3.828125,
|
| 1757 |
-
"learning_rate": 7.891403920981251e-06,
|
| 1758 |
-
"loss": 0.1318,
|
| 1759 |
-
"step": 1250
|
| 1760 |
-
},
|
| 1761 |
-
{
|
| 1762 |
-
"epoch": 3.6749633967789164,
|
| 1763 |
-
"grad_norm": 4.09375,
|
| 1764 |
-
"learning_rate": 7.808077062986515e-06,
|
| 1765 |
-
"loss": 0.1209,
|
| 1766 |
-
"step": 1255
|
| 1767 |
-
},
|
| 1768 |
-
{
|
| 1769 |
-
"epoch": 3.6896046852122986,
|
| 1770 |
-
"grad_norm": 4.125,
|
| 1771 |
-
"learning_rate": 7.724909776306625e-06,
|
| 1772 |
-
"loss": 0.1264,
|
| 1773 |
-
"step": 1260
|
| 1774 |
-
},
|
| 1775 |
-
{
|
| 1776 |
-
"epoch": 3.7042459736456808,
|
| 1777 |
-
"grad_norm": 3.90625,
|
| 1778 |
-
"learning_rate": 7.64190811549446e-06,
|
| 1779 |
-
"loss": 0.1187,
|
| 1780 |
-
"step": 1265
|
| 1781 |
-
},
|
| 1782 |
-
{
|
| 1783 |
-
"epoch": 3.718887262079063,
|
| 1784 |
-
"grad_norm": 3.65625,
|
| 1785 |
-
"learning_rate": 7.5590781230453515e-06,
|
| 1786 |
-
"loss": 0.1386,
|
| 1787 |
-
"step": 1270
|
| 1788 |
-
},
|
| 1789 |
-
{
|
| 1790 |
-
"epoch": 3.733528550512445,
|
| 1791 |
-
"grad_norm": 3.21875,
|
| 1792 |
-
"learning_rate": 7.4764258289572575e-06,
|
| 1793 |
-
"loss": 0.1183,
|
| 1794 |
-
"step": 1275
|
| 1795 |
-
},
|
| 1796 |
-
{
|
| 1797 |
-
"epoch": 3.7481698389458273,
|
| 1798 |
-
"grad_norm": 3.203125,
|
| 1799 |
-
"learning_rate": 7.393957250291725e-06,
|
| 1800 |
-
"loss": 0.1258,
|
| 1801 |
-
"step": 1280
|
| 1802 |
-
},
|
| 1803 |
-
{
|
| 1804 |
-
"epoch": 3.7628111273792095,
|
| 1805 |
-
"grad_norm": 3.5625,
|
| 1806 |
-
"learning_rate": 7.3116783907358975e-06,
|
| 1807 |
-
"loss": 0.1217,
|
| 1808 |
-
"step": 1285
|
| 1809 |
-
},
|
| 1810 |
-
{
|
| 1811 |
-
"epoch": 3.7774524158125917,
|
| 1812 |
-
"grad_norm": 3.8125,
|
| 1813 |
-
"learning_rate": 7.229595240165406e-06,
|
| 1814 |
-
"loss": 0.1238,
|
| 1815 |
-
"step": 1290
|
| 1816 |
-
},
|
| 1817 |
-
{
|
| 1818 |
-
"epoch": 3.792093704245974,
|
| 1819 |
-
"grad_norm": 4.125,
|
| 1820 |
-
"learning_rate": 7.1477137742083425e-06,
|
| 1821 |
-
"loss": 0.1221,
|
| 1822 |
-
"step": 1295
|
| 1823 |
-
},
|
| 1824 |
-
{
|
| 1825 |
-
"epoch": 3.8067349926793557,
|
| 1826 |
-
"grad_norm": 3.546875,
|
| 1827 |
-
"learning_rate": 7.066039953810208e-06,
|
| 1828 |
-
"loss": 0.1147,
|
| 1829 |
-
"step": 1300
|
| 1830 |
-
},
|
| 1831 |
-
{
|
| 1832 |
-
"epoch": 3.821376281112738,
|
| 1833 |
-
"grad_norm": 3.890625,
|
| 1834 |
-
"learning_rate": 6.984579724799985e-06,
|
| 1835 |
-
"loss": 0.1248,
|
| 1836 |
-
"step": 1305
|
| 1837 |
-
},
|
| 1838 |
-
{
|
| 1839 |
-
"epoch": 3.83601756954612,
|
| 1840 |
-
"grad_norm": 3.09375,
|
| 1841 |
-
"learning_rate": 6.903339017457254e-06,
|
| 1842 |
-
"loss": 0.1205,
|
| 1843 |
-
"step": 1310
|
| 1844 |
-
},
|
| 1845 |
-
{
|
| 1846 |
-
"epoch": 3.8506588579795022,
|
| 1847 |
-
"grad_norm": 3.71875,
|
| 1848 |
-
"learning_rate": 6.822323746080499e-06,
|
| 1849 |
-
"loss": 0.1205,
|
| 1850 |
-
"step": 1315
|
| 1851 |
-
},
|
| 1852 |
-
{
|
| 1853 |
-
"epoch": 3.8653001464128844,
|
| 1854 |
-
"grad_norm": 3.75,
|
| 1855 |
-
"learning_rate": 6.741539808556525e-06,
|
| 1856 |
-
"loss": 0.1258,
|
| 1857 |
-
"step": 1320
|
| 1858 |
-
},
|
| 1859 |
-
{
|
| 1860 |
-
"epoch": 3.8799414348462666,
|
| 1861 |
-
"grad_norm": 4.03125,
|
| 1862 |
-
"learning_rate": 6.660993085931113e-06,
|
| 1863 |
-
"loss": 0.1225,
|
| 1864 |
-
"step": 1325
|
| 1865 |
-
},
|
| 1866 |
-
{
|
| 1867 |
-
"epoch": 3.8945827232796484,
|
| 1868 |
-
"grad_norm": 3.625,
|
| 1869 |
-
"learning_rate": 6.580689441980861e-06,
|
| 1870 |
-
"loss": 0.1187,
|
| 1871 |
-
"step": 1330
|
| 1872 |
-
},
|
| 1873 |
-
{
|
| 1874 |
-
"epoch": 3.9092240117130306,
|
| 1875 |
-
"grad_norm": 3.375,
|
| 1876 |
-
"learning_rate": 6.5006347227863265e-06,
|
| 1877 |
-
"loss": 0.1204,
|
| 1878 |
-
"step": 1335
|
| 1879 |
-
},
|
| 1880 |
-
{
|
| 1881 |
-
"epoch": 3.9238653001464128,
|
| 1882 |
-
"grad_norm": 3.265625,
|
| 1883 |
-
"learning_rate": 6.420834756306411e-06,
|
| 1884 |
-
"loss": 0.1163,
|
| 1885 |
-
"step": 1340
|
| 1886 |
-
},
|
| 1887 |
-
{
|
| 1888 |
-
"epoch": 3.938506588579795,
|
| 1889 |
-
"grad_norm": 3.4375,
|
| 1890 |
-
"learning_rate": 6.341295351954105e-06,
|
| 1891 |
-
"loss": 0.1184,
|
| 1892 |
-
"step": 1345
|
| 1893 |
-
},
|
| 1894 |
-
{
|
| 1895 |
-
"epoch": 3.953147877013177,
|
| 1896 |
-
"grad_norm": 3.546875,
|
| 1897 |
-
"learning_rate": 6.262022300173549e-06,
|
| 1898 |
-
"loss": 0.1159,
|
| 1899 |
-
"step": 1350
|
| 1900 |
-
},
|
| 1901 |
-
{
|
| 1902 |
-
"epoch": 3.9677891654465594,
|
| 1903 |
-
"grad_norm": 4.40625,
|
| 1904 |
-
"learning_rate": 6.183021372018508e-06,
|
| 1905 |
-
"loss": 0.1212,
|
| 1906 |
-
"step": 1355
|
| 1907 |
-
},
|
| 1908 |
-
{
|
| 1909 |
-
"epoch": 3.9824304538799415,
|
| 1910 |
-
"grad_norm": 3.4375,
|
| 1911 |
-
"learning_rate": 6.104298318732218e-06,
|
| 1912 |
-
"loss": 0.1168,
|
| 1913 |
-
"step": 1360
|
| 1914 |
-
},
|
| 1915 |
-
{
|
| 1916 |
-
"epoch": 3.9970717423133237,
|
| 1917 |
-
"grad_norm": 3.8125,
|
| 1918 |
-
"learning_rate": 6.025858871328721e-06,
|
| 1919 |
-
"loss": 0.1169,
|
| 1920 |
-
"step": 1365
|
| 1921 |
-
},
|
| 1922 |
-
{
|
| 1923 |
-
"epoch": 4.011713030746706,
|
| 1924 |
-
"grad_norm": 1.6875,
|
| 1925 |
-
"learning_rate": 5.947708740175633e-06,
|
| 1926 |
-
"loss": 0.0969,
|
| 1927 |
-
"step": 1370
|
| 1928 |
-
},
|
| 1929 |
-
{
|
| 1930 |
-
"epoch": 4.026354319180088,
|
| 1931 |
-
"grad_norm": 2.4375,
|
| 1932 |
-
"learning_rate": 5.869853614578438e-06,
|
| 1933 |
-
"loss": 0.0787,
|
| 1934 |
-
"step": 1375
|
| 1935 |
-
},
|
| 1936 |
-
{
|
| 1937 |
-
"epoch": 4.04099560761347,
|
| 1938 |
-
"grad_norm": 1.9140625,
|
| 1939 |
-
"learning_rate": 5.792299162366304e-06,
|
| 1940 |
-
"loss": 0.0806,
|
| 1941 |
-
"step": 1380
|
| 1942 |
-
},
|
| 1943 |
-
{
|
| 1944 |
-
"epoch": 4.0556368960468525,
|
| 1945 |
-
"grad_norm": 4.59375,
|
| 1946 |
-
"learning_rate": 5.715051029479475e-06,
|
| 1947 |
-
"loss": 0.0794,
|
| 1948 |
-
"step": 1385
|
| 1949 |
-
},
|
| 1950 |
-
{
|
| 1951 |
-
"epoch": 4.070278184480234,
|
| 1952 |
-
"grad_norm": 2.109375,
|
| 1953 |
-
"learning_rate": 5.638114839558233e-06,
|
| 1954 |
-
"loss": 0.0756,
|
| 1955 |
-
"step": 1390
|
| 1956 |
-
},
|
| 1957 |
-
{
|
| 1958 |
-
"epoch": 4.084919472913616,
|
| 1959 |
-
"grad_norm": 2.828125,
|
| 1960 |
-
"learning_rate": 5.561496193533516e-06,
|
| 1961 |
-
"loss": 0.0793,
|
| 1962 |
-
"step": 1395
|
| 1963 |
-
},
|
| 1964 |
-
{
|
| 1965 |
-
"epoch": 4.099560761346998,
|
| 1966 |
-
"grad_norm": 2.578125,
|
| 1967 |
-
"learning_rate": 5.485200669219155e-06,
|
| 1968 |
-
"loss": 0.0749,
|
| 1969 |
-
"step": 1400
|
| 1970 |
-
},
|
| 1971 |
-
{
|
| 1972 |
-
"epoch": 4.11420204978038,
|
| 1973 |
-
"grad_norm": 2.078125,
|
| 1974 |
-
"learning_rate": 5.4092338209058375e-06,
|
| 1975 |
-
"loss": 0.0752,
|
| 1976 |
-
"step": 1405
|
| 1977 |
-
},
|
| 1978 |
-
{
|
| 1979 |
-
"epoch": 4.128843338213763,
|
| 1980 |
-
"grad_norm": 2.6875,
|
| 1981 |
-
"learning_rate": 5.333601178956718e-06,
|
| 1982 |
-
"loss": 0.0749,
|
| 1983 |
-
"step": 1410
|
| 1984 |
-
},
|
| 1985 |
-
{
|
| 1986 |
-
"epoch": 4.143484626647145,
|
| 1987 |
-
"grad_norm": 2.1875,
|
| 1988 |
-
"learning_rate": 5.258308249404853e-06,
|
| 1989 |
-
"loss": 0.0764,
|
| 1990 |
-
"step": 1415
|
| 1991 |
-
},
|
| 1992 |
-
{
|
| 1993 |
-
"epoch": 4.158125915080527,
|
| 1994 |
-
"grad_norm": 2.234375,
|
| 1995 |
-
"learning_rate": 5.183360513552313e-06,
|
| 1996 |
-
"loss": 0.0774,
|
| 1997 |
-
"step": 1420
|
| 1998 |
-
},
|
| 1999 |
-
{
|
| 2000 |
-
"epoch": 4.172767203513909,
|
| 2001 |
-
"grad_norm": 2.1875,
|
| 2002 |
-
"learning_rate": 5.108763427571203e-06,
|
| 2003 |
-
"loss": 0.0769,
|
| 2004 |
-
"step": 1425
|
| 2005 |
-
},
|
| 2006 |
-
{
|
| 2007 |
-
"epoch": 4.187408491947291,
|
| 2008 |
-
"grad_norm": 2.375,
|
| 2009 |
-
"learning_rate": 5.034522422106403e-06,
|
| 2010 |
-
"loss": 0.0808,
|
| 2011 |
-
"step": 1430
|
| 2012 |
-
},
|
| 2013 |
-
{
|
| 2014 |
-
"epoch": 4.202049780380674,
|
| 2015 |
-
"grad_norm": 2.796875,
|
| 2016 |
-
"learning_rate": 4.96064290188026e-06,
|
| 2017 |
-
"loss": 0.0831,
|
| 2018 |
-
"step": 1435
|
| 2019 |
-
},
|
| 2020 |
-
{
|
| 2021 |
-
"epoch": 4.216691068814056,
|
| 2022 |
-
"grad_norm": 2.703125,
|
| 2023 |
-
"learning_rate": 4.887130245299082e-06,
|
| 2024 |
-
"loss": 0.0774,
|
| 2025 |
-
"step": 1440
|
| 2026 |
-
},
|
| 2027 |
-
{
|
| 2028 |
-
"epoch": 4.231332357247438,
|
| 2029 |
-
"grad_norm": 2.265625,
|
| 2030 |
-
"learning_rate": 4.813989804061644e-06,
|
| 2031 |
-
"loss": 0.0782,
|
| 2032 |
-
"step": 1445
|
| 2033 |
-
},
|
| 2034 |
-
{
|
| 2035 |
-
"epoch": 4.24597364568082,
|
| 2036 |
-
"grad_norm": 2.28125,
|
| 2037 |
-
"learning_rate": 4.741226902769537e-06,
|
| 2038 |
-
"loss": 0.0801,
|
| 2039 |
-
"step": 1450
|
| 2040 |
-
},
|
| 2041 |
-
{
|
| 2042 |
-
"epoch": 4.260614934114202,
|
| 2043 |
-
"grad_norm": 2.546875,
|
| 2044 |
-
"learning_rate": 4.668846838539581e-06,
|
| 2045 |
-
"loss": 0.0787,
|
| 2046 |
-
"step": 1455
|
| 2047 |
-
},
|
| 2048 |
-
{
|
| 2049 |
-
"epoch": 4.2752562225475845,
|
| 2050 |
-
"grad_norm": 3.03125,
|
| 2051 |
-
"learning_rate": 4.596854880618149e-06,
|
| 2052 |
-
"loss": 0.0782,
|
| 2053 |
-
"step": 1460
|
| 2054 |
-
},
|
| 2055 |
-
{
|
| 2056 |
-
"epoch": 4.289897510980967,
|
| 2057 |
-
"grad_norm": 2.46875,
|
| 2058 |
-
"learning_rate": 4.525256269997621e-06,
|
| 2059 |
-
"loss": 0.0797,
|
| 2060 |
-
"step": 1465
|
| 2061 |
-
},
|
| 2062 |
-
{
|
| 2063 |
-
"epoch": 4.304538799414349,
|
| 2064 |
-
"grad_norm": 3.171875,
|
| 2065 |
-
"learning_rate": 4.4540562190347935e-06,
|
| 2066 |
-
"loss": 0.0788,
|
| 2067 |
-
"step": 1470
|
| 2068 |
-
},
|
| 2069 |
-
{
|
| 2070 |
-
"epoch": 4.31918008784773,
|
| 2071 |
-
"grad_norm": 2.578125,
|
| 2072 |
-
"learning_rate": 4.383259911071465e-06,
|
| 2073 |
-
"loss": 0.0797,
|
| 2074 |
-
"step": 1475
|
| 2075 |
-
},
|
| 2076 |
-
{
|
| 2077 |
-
"epoch": 4.333821376281112,
|
| 2078 |
-
"grad_norm": 2.078125,
|
| 2079 |
-
"learning_rate": 4.31287250005704e-06,
|
| 2080 |
-
"loss": 0.0794,
|
| 2081 |
-
"step": 1480
|
| 2082 |
-
},
|
| 2083 |
-
{
|
| 2084 |
-
"epoch": 4.348462664714495,
|
| 2085 |
-
"grad_norm": 2.375,
|
| 2086 |
-
"learning_rate": 4.242899110173375e-06,
|
| 2087 |
-
"loss": 0.0785,
|
| 2088 |
-
"step": 1485
|
| 2089 |
-
},
|
| 2090 |
-
{
|
| 2091 |
-
"epoch": 4.363103953147877,
|
| 2092 |
-
"grad_norm": 2.125,
|
| 2093 |
-
"learning_rate": 4.173344835461701e-06,
|
| 2094 |
-
"loss": 0.0752,
|
| 2095 |
-
"step": 1490
|
| 2096 |
-
},
|
| 2097 |
-
{
|
| 2098 |
-
"epoch": 4.377745241581259,
|
| 2099 |
-
"grad_norm": 3.796875,
|
| 2100 |
-
"learning_rate": 4.1042147394518106e-06,
|
| 2101 |
-
"loss": 0.0789,
|
| 2102 |
-
"step": 1495
|
| 2103 |
-
},
|
| 2104 |
-
{
|
| 2105 |
-
"epoch": 4.392386530014641,
|
| 2106 |
-
"grad_norm": 2.203125,
|
| 2107 |
-
"learning_rate": 4.035513854793389e-06,
|
| 2108 |
-
"loss": 0.0754,
|
| 2109 |
-
"step": 1500
|
| 2110 |
-
},
|
| 2111 |
-
{
|
| 2112 |
-
"epoch": 4.407027818448023,
|
| 2113 |
-
"grad_norm": 2.390625,
|
| 2114 |
-
"learning_rate": 3.967247182889698e-06,
|
| 2115 |
-
"loss": 0.0793,
|
| 2116 |
-
"step": 1505
|
| 2117 |
-
},
|
| 2118 |
-
{
|
| 2119 |
-
"epoch": 4.421669106881406,
|
| 2120 |
-
"grad_norm": 2.359375,
|
| 2121 |
-
"learning_rate": 3.899419693533423e-06,
|
| 2122 |
-
"loss": 0.0789,
|
| 2123 |
-
"step": 1510
|
| 2124 |
-
},
|
| 2125 |
-
{
|
| 2126 |
-
"epoch": 4.436310395314788,
|
| 2127 |
-
"grad_norm": 2.515625,
|
| 2128 |
-
"learning_rate": 3.832036324544912e-06,
|
| 2129 |
-
"loss": 0.0769,
|
| 2130 |
-
"step": 1515
|
| 2131 |
-
},
|
| 2132 |
-
{
|
| 2133 |
-
"epoch": 4.45095168374817,
|
| 2134 |
-
"grad_norm": 2.015625,
|
| 2135 |
-
"learning_rate": 3.7651019814126656e-06,
|
| 2136 |
-
"loss": 0.0772,
|
| 2137 |
-
"step": 1520
|
| 2138 |
-
},
|
| 2139 |
-
{
|
| 2140 |
-
"epoch": 4.465592972181552,
|
| 2141 |
-
"grad_norm": 2.390625,
|
| 2142 |
-
"learning_rate": 3.698621536936262e-06,
|
| 2143 |
-
"loss": 0.0794,
|
| 2144 |
-
"step": 1525
|
| 2145 |
-
},
|
| 2146 |
-
{
|
| 2147 |
-
"epoch": 4.480234260614934,
|
| 2148 |
-
"grad_norm": 2.40625,
|
| 2149 |
-
"learning_rate": 3.6325998308715827e-06,
|
| 2150 |
-
"loss": 0.0787,
|
| 2151 |
-
"step": 1530
|
| 2152 |
-
},
|
| 2153 |
-
{
|
| 2154 |
-
"epoch": 4.4948755490483165,
|
| 2155 |
-
"grad_norm": 2.140625,
|
| 2156 |
-
"learning_rate": 3.567041669578507e-06,
|
| 2157 |
-
"loss": 0.0777,
|
| 2158 |
-
"step": 1535
|
| 2159 |
-
},
|
| 2160 |
-
{
|
| 2161 |
-
"epoch": 4.509516837481699,
|
| 2162 |
-
"grad_norm": 1.859375,
|
| 2163 |
-
"learning_rate": 3.5019518256709773e-06,
|
| 2164 |
-
"loss": 0.0792,
|
| 2165 |
-
"step": 1540
|
| 2166 |
-
},
|
| 2167 |
-
{
|
| 2168 |
-
"epoch": 4.524158125915081,
|
| 2169 |
-
"grad_norm": 2.296875,
|
| 2170 |
-
"learning_rate": 3.437335037669598e-06,
|
| 2171 |
-
"loss": 0.0787,
|
| 2172 |
-
"step": 1545
|
| 2173 |
-
},
|
| 2174 |
-
{
|
| 2175 |
-
"epoch": 4.538799414348462,
|
| 2176 |
-
"grad_norm": 2.265625,
|
| 2177 |
-
"learning_rate": 3.3731960096566297e-06,
|
| 2178 |
-
"loss": 0.0774,
|
| 2179 |
-
"step": 1550
|
| 2180 |
-
},
|
| 2181 |
-
{
|
| 2182 |
-
"epoch": 4.553440702781844,
|
| 2183 |
-
"grad_norm": 2.21875,
|
| 2184 |
-
"learning_rate": 3.3095394109335686e-06,
|
| 2185 |
-
"loss": 0.0796,
|
| 2186 |
-
"step": 1555
|
| 2187 |
-
},
|
| 2188 |
-
{
|
| 2189 |
-
"epoch": 4.568081991215227,
|
| 2190 |
-
"grad_norm": 2.296875,
|
| 2191 |
-
"learning_rate": 3.2463698756811966e-06,
|
| 2192 |
-
"loss": 0.0768,
|
| 2193 |
-
"step": 1560
|
| 2194 |
-
},
|
| 2195 |
-
{
|
| 2196 |
-
"epoch": 4.582723279648609,
|
| 2197 |
-
"grad_norm": 2.90625,
|
| 2198 |
-
"learning_rate": 3.1836920026222297e-06,
|
| 2199 |
-
"loss": 0.0784,
|
| 2200 |
-
"step": 1565
|
| 2201 |
-
},
|
| 2202 |
-
{
|
| 2203 |
-
"epoch": 4.597364568081991,
|
| 2204 |
-
"grad_norm": 2.171875,
|
| 2205 |
-
"learning_rate": 3.12151035468652e-06,
|
| 2206 |
-
"loss": 0.078,
|
| 2207 |
-
"step": 1570
|
| 2208 |
-
},
|
| 2209 |
-
{
|
| 2210 |
-
"epoch": 4.612005856515373,
|
| 2211 |
-
"grad_norm": 3.015625,
|
| 2212 |
-
"learning_rate": 3.059829458678899e-06,
|
| 2213 |
-
"loss": 0.0778,
|
| 2214 |
-
"step": 1575
|
| 2215 |
-
},
|
| 2216 |
-
{
|
| 2217 |
-
"epoch": 4.626647144948755,
|
| 2218 |
-
"grad_norm": 2.421875,
|
| 2219 |
-
"learning_rate": 2.9986538049495984e-06,
|
| 2220 |
-
"loss": 0.0768,
|
| 2221 |
-
"step": 1580
|
| 2222 |
-
},
|
| 2223 |
-
{
|
| 2224 |
-
"epoch": 4.641288433382138,
|
| 2225 |
-
"grad_norm": 2.09375,
|
| 2226 |
-
"learning_rate": 2.937987847067372e-06,
|
| 2227 |
-
"loss": 0.0755,
|
| 2228 |
-
"step": 1585
|
| 2229 |
-
},
|
| 2230 |
-
{
|
| 2231 |
-
"epoch": 4.65592972181552,
|
| 2232 |
-
"grad_norm": 2.265625,
|
| 2233 |
-
"learning_rate": 2.877836001495269e-06,
|
| 2234 |
-
"loss": 0.0792,
|
| 2235 |
-
"step": 1590
|
| 2236 |
-
},
|
| 2237 |
-
{
|
| 2238 |
-
"epoch": 4.670571010248902,
|
| 2239 |
-
"grad_norm": 2.140625,
|
| 2240 |
-
"learning_rate": 2.8182026472691303e-06,
|
| 2241 |
-
"loss": 0.0778,
|
| 2242 |
-
"step": 1595
|
| 2243 |
-
},
|
| 2244 |
-
{
|
| 2245 |
-
"epoch": 4.685212298682284,
|
| 2246 |
-
"grad_norm": 3.3125,
|
| 2247 |
-
"learning_rate": 2.7590921256787797e-06,
|
| 2248 |
-
"loss": 0.0783,
|
| 2249 |
-
"step": 1600
|
| 2250 |
-
},
|
| 2251 |
-
{
|
| 2252 |
-
"epoch": 4.699853587115666,
|
| 2253 |
-
"grad_norm": 3.0625,
|
| 2254 |
-
"learning_rate": 2.7005087399519836e-06,
|
| 2255 |
-
"loss": 0.0813,
|
| 2256 |
-
"step": 1605
|
| 2257 |
-
},
|
| 2258 |
-
{
|
| 2259 |
-
"epoch": 4.714494875549049,
|
| 2260 |
-
"grad_norm": 2.375,
|
| 2261 |
-
"learning_rate": 2.6424567549411838e-06,
|
| 2262 |
-
"loss": 0.082,
|
| 2263 |
-
"step": 1610
|
| 2264 |
-
},
|
| 2265 |
-
{
|
| 2266 |
-
"epoch": 4.729136163982431,
|
| 2267 |
-
"grad_norm": 2.359375,
|
| 2268 |
-
"learning_rate": 2.5849403968130182e-06,
|
| 2269 |
-
"loss": 0.0721,
|
| 2270 |
-
"step": 1615
|
| 2271 |
-
},
|
| 2272 |
-
{
|
| 2273 |
-
"epoch": 4.743777452415813,
|
| 2274 |
-
"grad_norm": 2.15625,
|
| 2275 |
-
"learning_rate": 2.5279638527406426e-06,
|
| 2276 |
-
"loss": 0.0774,
|
| 2277 |
-
"step": 1620
|
| 2278 |
-
},
|
| 2279 |
-
{
|
| 2280 |
-
"epoch": 4.758418740849194,
|
| 2281 |
-
"grad_norm": 2.46875,
|
| 2282 |
-
"learning_rate": 2.4715312705989236e-06,
|
| 2283 |
-
"loss": 0.0816,
|
| 2284 |
-
"step": 1625
|
| 2285 |
-
},
|
| 2286 |
-
{
|
| 2287 |
-
"epoch": 4.773060029282577,
|
| 2288 |
-
"grad_norm": 2.484375,
|
| 2289 |
-
"learning_rate": 2.4156467586624588e-06,
|
| 2290 |
-
"loss": 0.0793,
|
| 2291 |
-
"step": 1630
|
| 2292 |
-
},
|
| 2293 |
-
{
|
| 2294 |
-
"epoch": 4.787701317715959,
|
| 2295 |
-
"grad_norm": 1.9609375,
|
| 2296 |
-
"learning_rate": 2.3603143853065146e-06,
|
| 2297 |
-
"loss": 0.0755,
|
| 2298 |
-
"step": 1635
|
| 2299 |
-
},
|
| 2300 |
-
{
|
| 2301 |
-
"epoch": 4.802342606149341,
|
| 2302 |
-
"grad_norm": 2.25,
|
| 2303 |
-
"learning_rate": 2.305538178710831e-06,
|
| 2304 |
-
"loss": 0.081,
|
| 2305 |
-
"step": 1640
|
| 2306 |
-
},
|
| 2307 |
-
{
|
| 2308 |
-
"epoch": 4.816983894582723,
|
| 2309 |
-
"grad_norm": 2.84375,
|
| 2310 |
-
"learning_rate": 2.25132212656638e-06,
|
| 2311 |
-
"loss": 0.0824,
|
| 2312 |
-
"step": 1645
|
| 2313 |
-
},
|
| 2314 |
-
{
|
| 2315 |
-
"epoch": 4.831625183016105,
|
| 2316 |
-
"grad_norm": 2.921875,
|
| 2317 |
-
"learning_rate": 2.1976701757850603e-06,
|
| 2318 |
-
"loss": 0.0757,
|
| 2319 |
-
"step": 1650
|
| 2320 |
-
},
|
| 2321 |
-
{
|
| 2322 |
-
"epoch": 4.846266471449487,
|
| 2323 |
-
"grad_norm": 2.5,
|
| 2324 |
-
"learning_rate": 2.1445862322123734e-06,
|
| 2325 |
-
"loss": 0.0787,
|
| 2326 |
-
"step": 1655
|
| 2327 |
-
},
|
| 2328 |
-
{
|
| 2329 |
-
"epoch": 4.86090775988287,
|
| 2330 |
-
"grad_norm": 2.171875,
|
| 2331 |
-
"learning_rate": 2.092074160343063e-06,
|
| 2332 |
-
"loss": 0.0751,
|
| 2333 |
-
"step": 1660
|
| 2334 |
-
},
|
| 2335 |
-
{
|
| 2336 |
-
"epoch": 4.875549048316252,
|
| 2337 |
-
"grad_norm": 2.265625,
|
| 2338 |
-
"learning_rate": 2.0401377830397874e-06,
|
| 2339 |
-
"loss": 0.0797,
|
| 2340 |
-
"step": 1665
|
| 2341 |
-
},
|
| 2342 |
-
{
|
| 2343 |
-
"epoch": 4.890190336749634,
|
| 2344 |
-
"grad_norm": 2.40625,
|
| 2345 |
-
"learning_rate": 1.9887808812548272e-06,
|
| 2346 |
-
"loss": 0.0764,
|
| 2347 |
-
"step": 1670
|
| 2348 |
-
},
|
| 2349 |
-
{
|
| 2350 |
-
"epoch": 4.904831625183016,
|
| 2351 |
-
"grad_norm": 2.109375,
|
| 2352 |
-
"learning_rate": 1.938007193754816e-06,
|
| 2353 |
-
"loss": 0.0809,
|
| 2354 |
-
"step": 1675
|
| 2355 |
-
},
|
| 2356 |
-
{
|
| 2357 |
-
"epoch": 4.919472913616398,
|
| 2358 |
-
"grad_norm": 2.3125,
|
| 2359 |
-
"learning_rate": 1.8878204168485691e-06,
|
| 2360 |
-
"loss": 0.0792,
|
| 2361 |
-
"step": 1680
|
| 2362 |
-
},
|
| 2363 |
-
{
|
| 2364 |
-
"epoch": 4.934114202049781,
|
| 2365 |
-
"grad_norm": 2.34375,
|
| 2366 |
-
"learning_rate": 1.8382242041179876e-06,
|
| 2367 |
-
"loss": 0.0775,
|
| 2368 |
-
"step": 1685
|
| 2369 |
-
},
|
| 2370 |
-
{
|
| 2371 |
-
"epoch": 4.948755490483163,
|
| 2372 |
-
"grad_norm": 2.328125,
|
| 2373 |
-
"learning_rate": 1.7892221661520925e-06,
|
| 2374 |
-
"loss": 0.0836,
|
| 2375 |
-
"step": 1690
|
| 2376 |
-
},
|
| 2377 |
-
{
|
| 2378 |
-
"epoch": 4.963396778916545,
|
| 2379 |
-
"grad_norm": 2.453125,
|
| 2380 |
-
"learning_rate": 1.740817870284155e-06,
|
| 2381 |
-
"loss": 0.0768,
|
| 2382 |
-
"step": 1695
|
| 2383 |
-
},
|
| 2384 |
-
{
|
| 2385 |
-
"epoch": 4.978038067349927,
|
| 2386 |
-
"grad_norm": 3.09375,
|
| 2387 |
-
"learning_rate": 1.693014840332009e-06,
|
| 2388 |
-
"loss": 0.0746,
|
| 2389 |
-
"step": 1700
|
| 2390 |
-
},
|
| 2391 |
-
{
|
| 2392 |
-
"epoch": 4.992679355783309,
|
| 2393 |
-
"grad_norm": 2.25,
|
| 2394 |
-
"learning_rate": 1.645816556341513e-06,
|
| 2395 |
-
"loss": 0.0769,
|
| 2396 |
-
"step": 1705
|
| 2397 |
-
},
|
| 2398 |
-
{
|
| 2399 |
-
"epoch": 5.007320644216691,
|
| 2400 |
-
"grad_norm": 2.0625,
|
| 2401 |
-
"learning_rate": 1.5992264543332125e-06,
|
| 2402 |
-
"loss": 0.0747,
|
| 2403 |
-
"step": 1710
|
| 2404 |
-
},
|
| 2405 |
-
{
|
| 2406 |
-
"epoch": 5.021961932650073,
|
| 2407 |
-
"grad_norm": 2.109375,
|
| 2408 |
-
"learning_rate": 1.5532479260521849e-06,
|
| 2409 |
-
"loss": 0.0711,
|
| 2410 |
-
"step": 1715
|
| 2411 |
-
},
|
| 2412 |
-
{
|
| 2413 |
-
"epoch": 5.036603221083455,
|
| 2414 |
-
"grad_norm": 1.4375,
|
| 2415 |
-
"learning_rate": 1.507884318721131e-06,
|
| 2416 |
-
"loss": 0.0685,
|
| 2417 |
-
"step": 1720
|
| 2418 |
-
},
|
| 2419 |
-
{
|
| 2420 |
-
"epoch": 5.051244509516837,
|
| 2421 |
-
"grad_norm": 1.8359375,
|
| 2422 |
-
"learning_rate": 1.463138934796694e-06,
|
| 2423 |
-
"loss": 0.0653,
|
| 2424 |
-
"step": 1725
|
| 2425 |
-
},
|
| 2426 |
-
{
|
| 2427 |
-
"epoch": 5.065885797950219,
|
| 2428 |
-
"grad_norm": 1.59375,
|
| 2429 |
-
"learning_rate": 1.4190150317290485e-06,
|
| 2430 |
-
"loss": 0.0697,
|
| 2431 |
-
"step": 1730
|
| 2432 |
-
},
|
| 2433 |
-
{
|
| 2434 |
-
"epoch": 5.080527086383602,
|
| 2435 |
-
"grad_norm": 2.640625,
|
| 2436 |
-
"learning_rate": 1.3755158217247488e-06,
|
| 2437 |
-
"loss": 0.0696,
|
| 2438 |
-
"step": 1735
|
| 2439 |
-
},
|
| 2440 |
-
{
|
| 2441 |
-
"epoch": 5.095168374816984,
|
| 2442 |
-
"grad_norm": 1.6640625,
|
| 2443 |
-
"learning_rate": 1.3326444715128884e-06,
|
| 2444 |
-
"loss": 0.068,
|
| 2445 |
-
"step": 1740
|
| 2446 |
-
},
|
| 2447 |
-
{
|
| 2448 |
-
"epoch": 5.109809663250366,
|
| 2449 |
-
"grad_norm": 1.9765625,
|
| 2450 |
-
"learning_rate": 1.2904041021145597e-06,
|
| 2451 |
-
"loss": 0.0688,
|
| 2452 |
-
"step": 1745
|
| 2453 |
-
},
|
| 2454 |
-
{
|
| 2455 |
-
"epoch": 5.124450951683748,
|
| 2456 |
-
"grad_norm": 1.8203125,
|
| 2457 |
-
"learning_rate": 1.2487977886156522e-06,
|
| 2458 |
-
"loss": 0.0689,
|
| 2459 |
-
"step": 1750
|
| 2460 |
-
},
|
| 2461 |
-
{
|
| 2462 |
-
"epoch": 5.13909224011713,
|
| 2463 |
-
"grad_norm": 1.9765625,
|
| 2464 |
-
"learning_rate": 1.207828559942974e-06,
|
| 2465 |
-
"loss": 0.0688,
|
| 2466 |
-
"step": 1755
|
| 2467 |
-
},
|
| 2468 |
-
{
|
| 2469 |
-
"epoch": 5.153733528550513,
|
| 2470 |
-
"grad_norm": 1.7578125,
|
| 2471 |
-
"learning_rate": 1.1674993986437567e-06,
|
| 2472 |
-
"loss": 0.0684,
|
| 2473 |
-
"step": 1760
|
| 2474 |
-
},
|
| 2475 |
-
{
|
| 2476 |
-
"epoch": 5.168374816983895,
|
| 2477 |
-
"grad_norm": 1.4921875,
|
| 2478 |
-
"learning_rate": 1.1278132406685226e-06,
|
| 2479 |
-
"loss": 0.0688,
|
| 2480 |
-
"step": 1765
|
| 2481 |
-
},
|
| 2482 |
-
{
|
| 2483 |
-
"epoch": 5.183016105417277,
|
| 2484 |
-
"grad_norm": 1.875,
|
| 2485 |
-
"learning_rate": 1.0887729751573562e-06,
|
| 2486 |
-
"loss": 0.0715,
|
| 2487 |
-
"step": 1770
|
| 2488 |
-
},
|
| 2489 |
-
{
|
| 2490 |
-
"epoch": 5.197657393850659,
|
| 2491 |
-
"grad_norm": 1.859375,
|
| 2492 |
-
"learning_rate": 1.0503814442295624e-06,
|
| 2493 |
-
"loss": 0.0712,
|
| 2494 |
-
"step": 1775
|
| 2495 |
-
},
|
| 2496 |
-
{
|
| 2497 |
-
"epoch": 5.212298682284041,
|
| 2498 |
-
"grad_norm": 1.5703125,
|
| 2499 |
-
"learning_rate": 1.0126414427767716e-06,
|
| 2500 |
-
"loss": 0.0677,
|
| 2501 |
-
"step": 1780
|
| 2502 |
-
},
|
| 2503 |
-
{
|
| 2504 |
-
"epoch": 5.2269399707174236,
|
| 2505 |
-
"grad_norm": 1.703125,
|
| 2506 |
-
"learning_rate": 9.755557182594656e-07,
|
| 2507 |
-
"loss": 0.0706,
|
| 2508 |
-
"step": 1785
|
| 2509 |
-
},
|
| 2510 |
-
{
|
| 2511 |
-
"epoch": 5.241581259150805,
|
| 2512 |
-
"grad_norm": 1.59375,
|
| 2513 |
-
"learning_rate": 9.391269705069739e-07,
|
| 2514 |
-
"loss": 0.0729,
|
| 2515 |
-
"step": 1790
|
| 2516 |
-
},
|
| 2517 |
-
{
|
| 2518 |
-
"epoch": 5.256222547584187,
|
| 2519 |
-
"grad_norm": 1.8125,
|
| 2520 |
-
"learning_rate": 9.033578515209108e-07,
|
| 2521 |
-
"loss": 0.0671,
|
| 2522 |
-
"step": 1795
|
| 2523 |
-
},
|
| 2524 |
-
{
|
| 2525 |
-
"epoch": 5.270863836017569,
|
| 2526 |
-
"grad_norm": 1.953125,
|
| 2527 |
-
"learning_rate": 8.682509652821303e-07,
|
| 2528 |
-
"loss": 0.0685,
|
| 2529 |
-
"step": 1800
|
| 2530 |
-
},
|
| 2531 |
-
{
|
| 2532 |
-
"epoch": 5.285505124450951,
|
| 2533 |
-
"grad_norm": 1.5390625,
|
| 2534 |
-
"learning_rate": 8.338088675611323e-07,
|
| 2535 |
-
"loss": 0.0682,
|
| 2536 |
-
"step": 1805
|
| 2537 |
-
},
|
| 2538 |
-
{
|
| 2539 |
-
"epoch": 5.300146412884334,
|
| 2540 |
-
"grad_norm": 1.84375,
|
| 2541 |
-
"learning_rate": 8.000340657320304e-07,
|
| 2542 |
-
"loss": 0.0681,
|
| 2543 |
-
"step": 1810
|
| 2544 |
-
},
|
| 2545 |
-
{
|
| 2546 |
-
"epoch": 5.314787701317716,
|
| 2547 |
-
"grad_norm": 2.234375,
|
| 2548 |
-
"learning_rate": 7.669290185899947e-07,
|
| 2549 |
-
"loss": 0.0683,
|
| 2550 |
-
"step": 1815
|
| 2551 |
-
},
|
| 2552 |
-
{
|
| 2553 |
-
"epoch": 5.329428989751098,
|
| 2554 |
-
"grad_norm": 1.9296875,
|
| 2555 |
-
"learning_rate": 7.34496136172268e-07,
|
| 2556 |
-
"loss": 0.0701,
|
| 2557 |
-
"step": 1820
|
| 2558 |
-
},
|
| 2559 |
-
{
|
| 2560 |
-
"epoch": 5.34407027818448,
|
| 2561 |
-
"grad_norm": 1.7734375,
|
| 2562 |
-
"learning_rate": 7.027377795826962e-07,
|
| 2563 |
-
"loss": 0.0696,
|
| 2564 |
-
"step": 1825
|
| 2565 |
-
},
|
| 2566 |
-
{
|
| 2567 |
-
"epoch": 5.358711566617862,
|
| 2568 |
-
"grad_norm": 1.3984375,
|
| 2569 |
-
"learning_rate": 6.716562608198651e-07,
|
| 2570 |
-
"loss": 0.065,
|
| 2571 |
-
"step": 1830
|
| 2572 |
-
},
|
| 2573 |
-
{
|
| 2574 |
-
"epoch": 5.373352855051245,
|
| 2575 |
-
"grad_norm": 1.6015625,
|
| 2576 |
-
"learning_rate": 6.412538426087667e-07,
|
| 2577 |
-
"loss": 0.0741,
|
| 2578 |
-
"step": 1835
|
| 2579 |
-
},
|
| 2580 |
-
{
|
| 2581 |
-
"epoch": 5.387994143484627,
|
| 2582 |
-
"grad_norm": 2.0,
|
| 2583 |
-
"learning_rate": 6.115327382360892e-07,
|
| 2584 |
-
"loss": 0.0707,
|
| 2585 |
-
"step": 1840
|
| 2586 |
-
},
|
| 2587 |
-
{
|
| 2588 |
-
"epoch": 5.402635431918009,
|
| 2589 |
-
"grad_norm": 1.8828125,
|
| 2590 |
-
"learning_rate": 5.82495111389072e-07,
|
| 2591 |
-
"loss": 0.0675,
|
| 2592 |
-
"step": 1845
|
| 2593 |
-
},
|
| 2594 |
-
{
|
| 2595 |
-
"epoch": 5.417276720351391,
|
| 2596 |
-
"grad_norm": 1.59375,
|
| 2597 |
-
"learning_rate": 5.541430759980137e-07,
|
| 2598 |
-
"loss": 0.0657,
|
| 2599 |
-
"step": 1850
|
| 2600 |
-
},
|
| 2601 |
-
{
|
| 2602 |
-
"epoch": 5.431918008784773,
|
| 2603 |
-
"grad_norm": 2.0,
|
| 2604 |
-
"learning_rate": 5.264786960823565e-07,
|
| 2605 |
-
"loss": 0.0684,
|
| 2606 |
-
"step": 1855
|
| 2607 |
-
},
|
| 2608 |
-
{
|
| 2609 |
-
"epoch": 5.446559297218156,
|
| 2610 |
-
"grad_norm": 1.7109375,
|
| 2611 |
-
"learning_rate": 4.995039856004447e-07,
|
| 2612 |
-
"loss": 0.0687,
|
| 2613 |
-
"step": 1860
|
| 2614 |
-
},
|
| 2615 |
-
{
|
| 2616 |
-
"epoch": 5.461200585651538,
|
| 2617 |
-
"grad_norm": 2.046875,
|
| 2618 |
-
"learning_rate": 4.7322090830288825e-07,
|
| 2619 |
-
"loss": 0.0722,
|
| 2620 |
-
"step": 1865
|
| 2621 |
-
},
|
| 2622 |
-
{
|
| 2623 |
-
"epoch": 5.475841874084919,
|
| 2624 |
-
"grad_norm": 1.5,
|
| 2625 |
-
"learning_rate": 4.4763137758962685e-07,
|
| 2626 |
-
"loss": 0.0672,
|
| 2627 |
-
"step": 1870
|
| 2628 |
-
},
|
| 2629 |
-
{
|
| 2630 |
-
"epoch": 5.490483162518301,
|
| 2631 |
-
"grad_norm": 1.8203125,
|
| 2632 |
-
"learning_rate": 4.227372563706134e-07,
|
| 2633 |
-
"loss": 0.0691,
|
| 2634 |
-
"step": 1875
|
| 2635 |
-
},
|
| 2636 |
-
{
|
| 2637 |
-
"epoch": 5.5051244509516835,
|
| 2638 |
-
"grad_norm": 1.828125,
|
| 2639 |
-
"learning_rate": 3.985403569302093e-07,
|
| 2640 |
-
"loss": 0.0669,
|
| 2641 |
-
"step": 1880
|
| 2642 |
-
},
|
| 2643 |
-
{
|
| 2644 |
-
"epoch": 5.519765739385066,
|
| 2645 |
-
"grad_norm": 1.953125,
|
| 2646 |
-
"learning_rate": 3.7504244079524023e-07,
|
| 2647 |
-
"loss": 0.0656,
|
| 2648 |
-
"step": 1885
|
| 2649 |
-
},
|
| 2650 |
-
{
|
| 2651 |
-
"epoch": 5.534407027818448,
|
| 2652 |
-
"grad_norm": 2.171875,
|
| 2653 |
-
"learning_rate": 3.522452186067671e-07,
|
| 2654 |
-
"loss": 0.0678,
|
| 2655 |
-
"step": 1890
|
| 2656 |
-
},
|
| 2657 |
-
{
|
| 2658 |
-
"epoch": 5.54904831625183,
|
| 2659 |
-
"grad_norm": 1.6484375,
|
| 2660 |
-
"learning_rate": 3.3015034999554273e-07,
|
| 2661 |
-
"loss": 0.0702,
|
| 2662 |
-
"step": 1895
|
| 2663 |
-
},
|
| 2664 |
-
{
|
| 2665 |
-
"epoch": 5.563689604685212,
|
| 2666 |
-
"grad_norm": 1.6484375,
|
| 2667 |
-
"learning_rate": 3.08759443461204e-07,
|
| 2668 |
-
"loss": 0.0668,
|
| 2669 |
-
"step": 1900
|
| 2670 |
-
},
|
| 2671 |
-
{
|
| 2672 |
-
"epoch": 5.578330893118594,
|
| 2673 |
-
"grad_norm": 1.8515625,
|
| 2674 |
-
"learning_rate": 2.8807405625515206e-07,
|
| 2675 |
-
"loss": 0.0692,
|
| 2676 |
-
"step": 1905
|
| 2677 |
-
},
|
| 2678 |
-
{
|
| 2679 |
-
"epoch": 5.592972181551977,
|
| 2680 |
-
"grad_norm": 1.6796875,
|
| 2681 |
-
"learning_rate": 2.680956942672119e-07,
|
| 2682 |
-
"loss": 0.0673,
|
| 2683 |
-
"step": 1910
|
| 2684 |
-
},
|
| 2685 |
-
{
|
| 2686 |
-
"epoch": 5.607613469985359,
|
| 2687 |
-
"grad_norm": 1.6875,
|
| 2688 |
-
"learning_rate": 2.488258119159814e-07,
|
| 2689 |
-
"loss": 0.0714,
|
| 2690 |
-
"step": 1915
|
| 2691 |
-
},
|
| 2692 |
-
{
|
| 2693 |
-
"epoch": 5.622254758418741,
|
| 2694 |
-
"grad_norm": 1.6875,
|
| 2695 |
-
"learning_rate": 2.302658120429635e-07,
|
| 2696 |
-
"loss": 0.0663,
|
| 2697 |
-
"step": 1920
|
| 2698 |
-
},
|
| 2699 |
-
{
|
| 2700 |
-
"epoch": 5.636896046852123,
|
| 2701 |
-
"grad_norm": 1.359375,
|
| 2702 |
-
"learning_rate": 2.1241704581043132e-07,
|
| 2703 |
-
"loss": 0.0626,
|
| 2704 |
-
"step": 1925
|
| 2705 |
-
},
|
| 2706 |
-
{
|
| 2707 |
-
"epoch": 5.651537335285505,
|
| 2708 |
-
"grad_norm": 1.71875,
|
| 2709 |
-
"learning_rate": 1.9528081260307363e-07,
|
| 2710 |
-
"loss": 0.0658,
|
| 2711 |
-
"step": 1930
|
| 2712 |
-
},
|
| 2713 |
-
{
|
| 2714 |
-
"epoch": 5.666178623718888,
|
| 2715 |
-
"grad_norm": 1.8125,
|
| 2716 |
-
"learning_rate": 1.7885835993338817e-07,
|
| 2717 |
-
"loss": 0.0701,
|
| 2718 |
-
"step": 1935
|
| 2719 |
-
},
|
| 2720 |
-
{
|
| 2721 |
-
"epoch": 5.68081991215227,
|
| 2722 |
-
"grad_norm": 1.9921875,
|
| 2723 |
-
"learning_rate": 1.6315088335087549e-07,
|
| 2724 |
-
"loss": 0.0715,
|
| 2725 |
-
"step": 1940
|
| 2726 |
-
},
|
| 2727 |
-
{
|
| 2728 |
-
"epoch": 5.695461200585651,
|
| 2729 |
-
"grad_norm": 1.5234375,
|
| 2730 |
-
"learning_rate": 1.4815952635499063e-07,
|
| 2731 |
-
"loss": 0.0657,
|
| 2732 |
-
"step": 1945
|
| 2733 |
-
},
|
| 2734 |
-
{
|
| 2735 |
-
"epoch": 5.710102489019034,
|
| 2736 |
-
"grad_norm": 1.921875,
|
| 2737 |
-
"learning_rate": 1.3388538031190888e-07,
|
| 2738 |
-
"loss": 0.0682,
|
| 2739 |
-
"step": 1950
|
| 2740 |
-
},
|
| 2741 |
-
{
|
| 2742 |
-
"epoch": 5.7247437774524155,
|
| 2743 |
-
"grad_norm": 1.578125,
|
| 2744 |
-
"learning_rate": 1.2032948437506574e-07,
|
| 2745 |
-
"loss": 0.0667,
|
| 2746 |
-
"step": 1955
|
| 2747 |
-
},
|
| 2748 |
-
{
|
| 2749 |
-
"epoch": 5.739385065885798,
|
| 2750 |
-
"grad_norm": 1.796875,
|
| 2751 |
-
"learning_rate": 1.07492825409512e-07,
|
| 2752 |
-
"loss": 0.071,
|
| 2753 |
-
"step": 1960
|
| 2754 |
-
},
|
| 2755 |
-
{
|
| 2756 |
-
"epoch": 5.75402635431918,
|
| 2757 |
-
"grad_norm": 1.9140625,
|
| 2758 |
-
"learning_rate": 9.537633792006673e-08,
|
| 2759 |
-
"loss": 0.0676,
|
| 2760 |
-
"step": 1965
|
| 2761 |
-
},
|
| 2762 |
-
{
|
| 2763 |
-
"epoch": 5.768667642752562,
|
| 2764 |
-
"grad_norm": 1.53125,
|
| 2765 |
-
"learning_rate": 8.39809039832884e-08,
|
| 2766 |
-
"loss": 0.0722,
|
| 2767 |
-
"step": 1970
|
| 2768 |
-
},
|
| 2769 |
-
{
|
| 2770 |
-
"epoch": 5.783308931185944,
|
| 2771 |
-
"grad_norm": 1.8515625,
|
| 2772 |
-
"learning_rate": 7.330735318325843e-08,
|
| 2773 |
-
"loss": 0.0707,
|
| 2774 |
-
"step": 1975
|
| 2775 |
-
},
|
| 2776 |
-
{
|
| 2777 |
-
"epoch": 5.797950219619326,
|
| 2778 |
-
"grad_norm": 1.640625,
|
| 2779 |
-
"learning_rate": 6.335646255118844e-08,
|
| 2780 |
-
"loss": 0.0697,
|
| 2781 |
-
"step": 1980
|
| 2782 |
-
},
|
| 2783 |
-
{
|
| 2784 |
-
"epoch": 5.812591508052709,
|
| 2785 |
-
"grad_norm": 1.921875,
|
| 2786 |
-
"learning_rate": 5.412895650885319e-08,
|
| 2787 |
-
"loss": 0.0709,
|
| 2788 |
-
"step": 1985
|
| 2789 |
-
},
|
| 2790 |
-
{
|
| 2791 |
-
"epoch": 5.827232796486091,
|
| 2792 |
-
"grad_norm": 1.734375,
|
| 2793 |
-
"learning_rate": 4.562550681584954e-08,
|
| 2794 |
-
"loss": 0.0702,
|
| 2795 |
-
"step": 1990
|
| 2796 |
-
},
|
| 2797 |
-
{
|
| 2798 |
-
"epoch": 5.841874084919473,
|
| 2799 |
-
"grad_norm": 1.828125,
|
| 2800 |
-
"learning_rate": 3.784673252069659e-08,
|
| 2801 |
-
"loss": 0.0713,
|
| 2802 |
-
"step": 1995
|
| 2803 |
-
},
|
| 2804 |
-
{
|
| 2805 |
-
"epoch": 5.856515373352855,
|
| 2806 |
-
"grad_norm": 1.8984375,
|
| 2807 |
-
"learning_rate": 3.079319991576957e-08,
|
| 2808 |
-
"loss": 0.0677,
|
| 2809 |
-
"step": 2000
|
| 2810 |
-
},
|
| 2811 |
-
{
|
| 2812 |
-
"epoch": 5.871156661786237,
|
| 2813 |
-
"grad_norm": 2.03125,
|
| 2814 |
-
"learning_rate": 2.4465422496069425e-08,
|
| 2815 |
-
"loss": 0.0671,
|
| 2816 |
-
"step": 2005
|
| 2817 |
-
},
|
| 2818 |
-
{
|
| 2819 |
-
"epoch": 5.88579795021962,
|
| 2820 |
-
"grad_norm": 1.9921875,
|
| 2821 |
-
"learning_rate": 1.886386092184389e-08,
|
| 2822 |
-
"loss": 0.07,
|
| 2823 |
-
"step": 2010
|
| 2824 |
-
},
|
| 2825 |
-
{
|
| 2826 |
-
"epoch": 5.900439238653002,
|
| 2827 |
-
"grad_norm": 2.046875,
|
| 2828 |
-
"learning_rate": 1.3988922985048724e-08,
|
| 2829 |
-
"loss": 0.0721,
|
| 2830 |
-
"step": 2015
|
| 2831 |
-
},
|
| 2832 |
-
{
|
| 2833 |
-
"epoch": 5.915080527086384,
|
| 2834 |
-
"grad_norm": 1.9375,
|
| 2835 |
-
"learning_rate": 9.840963579667017e-09,
|
| 2836 |
-
"loss": 0.0671,
|
| 2837 |
-
"step": 2020
|
| 2838 |
-
},
|
| 2839 |
-
{
|
| 2840 |
-
"epoch": 5.929721815519766,
|
| 2841 |
-
"grad_norm": 1.765625,
|
| 2842 |
-
"learning_rate": 6.420284675865418e-09,
|
| 2843 |
-
"loss": 0.0665,
|
| 2844 |
-
"step": 2025
|
| 2845 |
-
},
|
| 2846 |
-
{
|
| 2847 |
-
"epoch": 5.9443631039531475,
|
| 2848 |
-
"grad_norm": 1.8984375,
|
| 2849 |
-
"learning_rate": 3.7271352980139395e-09,
|
| 2850 |
-
"loss": 0.0653,
|
| 2851 |
-
"step": 2030
|
| 2852 |
-
},
|
| 2853 |
-
{
|
| 2854 |
-
"epoch": 5.95900439238653,
|
| 2855 |
-
"grad_norm": 1.7265625,
|
| 2856 |
-
"learning_rate": 1.7617115065593493e-09,
|
| 2857 |
-
"loss": 0.0691,
|
| 2858 |
-
"step": 2035
|
| 2859 |
-
},
|
| 2860 |
-
{
|
| 2861 |
-
"epoch": 5.973645680819912,
|
| 2862 |
-
"grad_norm": 1.671875,
|
| 2863 |
-
"learning_rate": 5.241563837476982e-10,
|
| 2864 |
-
"loss": 0.0699,
|
| 2865 |
-
"step": 2040
|
| 2866 |
-
},
|
| 2867 |
-
{
|
| 2868 |
-
"epoch": 5.988286969253294,
|
| 2869 |
-
"grad_norm": 1.671875,
|
| 2870 |
-
"learning_rate": 1.4560023211540598e-11,
|
| 2871 |
-
"loss": 0.0676,
|
| 2872 |
-
"step": 2045
|
| 2873 |
-
},
|
| 2874 |
-
{
|
| 2875 |
-
"epoch": 5.991215226939971,
|
| 2876 |
-
"step": 2046,
|
| 2877 |
-
"total_flos": 2.3380471586755123e+17,
|
| 2878 |
-
"train_loss": 0.343939907065905,
|
| 2879 |
-
"train_runtime": 7450.6058,
|
| 2880 |
-
"train_samples_per_second": 8.795,
|
| 2881 |
-
"train_steps_per_second": 0.275
|
| 2882 |
}
|
| 2883 |
],
|
| 2884 |
"logging_steps": 5,
|
| 2885 |
-
"max_steps":
|
| 2886 |
"num_input_tokens_seen": 0,
|
| 2887 |
"num_train_epochs": 6,
|
| 2888 |
"save_steps": 999999,
|
|
@@ -2898,7 +931,7 @@
|
|
| 2898 |
"attributes": {}
|
| 2899 |
}
|
| 2900 |
},
|
| 2901 |
-
"total_flos":
|
| 2902 |
"train_batch_size": 8,
|
| 2903 |
"trial_name": null,
|
| 2904 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 5.958236658932715,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 642,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.04640371229698376,
|
| 13 |
+
"grad_norm": 304.0,
|
| 14 |
+
"learning_rate": 1.5384615384615387e-06,
|
| 15 |
+
"loss": 2.4879,
|
| 16 |
"step": 5
|
| 17 |
},
|
| 18 |
{
|
| 19 |
+
"epoch": 0.09280742459396751,
|
| 20 |
+
"grad_norm": 58.5,
|
| 21 |
+
"learning_rate": 3.0769230769230774e-06,
|
| 22 |
+
"loss": 2.2869,
|
| 23 |
"step": 10
|
| 24 |
},
|
| 25 |
{
|
| 26 |
+
"epoch": 0.13921113689095127,
|
| 27 |
+
"grad_norm": 208.0,
|
| 28 |
+
"learning_rate": 4.615384615384616e-06,
|
| 29 |
+
"loss": 1.8801,
|
| 30 |
"step": 15
|
| 31 |
},
|
| 32 |
{
|
| 33 |
+
"epoch": 0.18561484918793503,
|
| 34 |
+
"grad_norm": 7.40625,
|
| 35 |
+
"learning_rate": 6.153846153846155e-06,
|
| 36 |
+
"loss": 1.4544,
|
| 37 |
"step": 20
|
| 38 |
},
|
| 39 |
{
|
| 40 |
+
"epoch": 0.23201856148491878,
|
| 41 |
+
"grad_norm": 5.71875,
|
| 42 |
+
"learning_rate": 7.692307692307694e-06,
|
| 43 |
+
"loss": 1.1734,
|
| 44 |
"step": 25
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 0.27842227378190254,
|
| 48 |
+
"grad_norm": 4.65625,
|
| 49 |
+
"learning_rate": 9.230769230769232e-06,
|
| 50 |
+
"loss": 1.006,
|
| 51 |
"step": 30
|
| 52 |
},
|
| 53 |
{
|
| 54 |
+
"epoch": 0.3248259860788863,
|
| 55 |
+
"grad_norm": 3.984375,
|
| 56 |
+
"learning_rate": 1.076923076923077e-05,
|
| 57 |
+
"loss": 0.9744,
|
| 58 |
"step": 35
|
| 59 |
},
|
| 60 |
{
|
| 61 |
+
"epoch": 0.37122969837587005,
|
| 62 |
+
"grad_norm": 4.34375,
|
| 63 |
+
"learning_rate": 1.230769230769231e-05,
|
| 64 |
+
"loss": 0.9372,
|
| 65 |
"step": 40
|
| 66 |
},
|
| 67 |
{
|
| 68 |
+
"epoch": 0.4176334106728538,
|
| 69 |
+
"grad_norm": 3.6875,
|
| 70 |
+
"learning_rate": 1.3846153846153847e-05,
|
| 71 |
+
"loss": 0.8942,
|
| 72 |
"step": 45
|
| 73 |
},
|
| 74 |
{
|
| 75 |
+
"epoch": 0.46403712296983757,
|
| 76 |
+
"grad_norm": 3.6875,
|
| 77 |
+
"learning_rate": 1.5384615384615387e-05,
|
| 78 |
+
"loss": 0.8729,
|
| 79 |
"step": 50
|
| 80 |
},
|
| 81 |
{
|
| 82 |
+
"epoch": 0.5104408352668214,
|
| 83 |
+
"grad_norm": 3.8125,
|
| 84 |
+
"learning_rate": 1.6923076923076924e-05,
|
| 85 |
+
"loss": 0.8978,
|
| 86 |
"step": 55
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"epoch": 0.5568445475638051,
|
| 90 |
+
"grad_norm": 3.546875,
|
| 91 |
+
"learning_rate": 1.8461538461538465e-05,
|
| 92 |
+
"loss": 0.8524,
|
| 93 |
"step": 60
|
| 94 |
},
|
| 95 |
{
|
| 96 |
+
"epoch": 0.6032482598607889,
|
| 97 |
+
"grad_norm": 3.5625,
|
| 98 |
+
"learning_rate": 2e-05,
|
| 99 |
+
"loss": 0.8231,
|
| 100 |
"step": 65
|
| 101 |
},
|
| 102 |
{
|
| 103 |
+
"epoch": 0.6496519721577726,
|
| 104 |
+
"grad_norm": 3.28125,
|
| 105 |
+
"learning_rate": 1.9996294632312766e-05,
|
| 106 |
+
"loss": 0.8315,
|
| 107 |
"step": 70
|
| 108 |
},
|
| 109 |
{
|
| 110 |
+
"epoch": 0.6960556844547564,
|
| 111 |
+
"grad_norm": 3.65625,
|
| 112 |
+
"learning_rate": 1.9985181275201e-05,
|
| 113 |
+
"loss": 0.8262,
|
| 114 |
"step": 75
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.7424593967517401,
|
| 118 |
+
"grad_norm": 3.109375,
|
| 119 |
+
"learning_rate": 1.9966668164479567e-05,
|
| 120 |
+
"loss": 0.8193,
|
| 121 |
"step": 80
|
| 122 |
},
|
| 123 |
{
|
| 124 |
+
"epoch": 0.7888631090487239,
|
| 125 |
+
"grad_norm": 3.28125,
|
| 126 |
+
"learning_rate": 1.9940769019724926e-05,
|
| 127 |
+
"loss": 0.8732,
|
| 128 |
"step": 85
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 0.8352668213457076,
|
| 132 |
+
"grad_norm": 3.03125,
|
| 133 |
+
"learning_rate": 1.9907503034107893e-05,
|
| 134 |
+
"loss": 0.8356,
|
| 135 |
"step": 90
|
| 136 |
},
|
| 137 |
{
|
| 138 |
+
"epoch": 0.8816705336426914,
|
| 139 |
+
"grad_norm": 3.0625,
|
| 140 |
+
"learning_rate": 1.9866894860170104e-05,
|
| 141 |
+
"loss": 0.8692,
|
| 142 |
"step": 95
|
| 143 |
},
|
| 144 |
{
|
| 145 |
+
"epoch": 0.9280742459396751,
|
| 146 |
+
"grad_norm": 2.984375,
|
| 147 |
+
"learning_rate": 1.9818974591554668e-05,
|
| 148 |
+
"loss": 0.8007,
|
| 149 |
"step": 100
|
| 150 |
},
|
| 151 |
{
|
| 152 |
+
"epoch": 0.974477958236659,
|
| 153 |
+
"grad_norm": 3.296875,
|
| 154 |
+
"learning_rate": 1.9763777740704572e-05,
|
| 155 |
+
"loss": 0.8366,
|
| 156 |
"step": 105
|
| 157 |
},
|
| 158 |
{
|
| 159 |
+
"epoch": 1.0208816705336428,
|
| 160 |
+
"grad_norm": 2.625,
|
| 161 |
+
"learning_rate": 1.970134521254532e-05,
|
| 162 |
+
"loss": 0.8049,
|
| 163 |
"step": 110
|
| 164 |
},
|
| 165 |
{
|
| 166 |
+
"epoch": 1.0672853828306264,
|
| 167 |
+
"grad_norm": 2.921875,
|
| 168 |
+
"learning_rate": 1.9631723274171412e-05,
|
| 169 |
+
"loss": 0.6126,
|
| 170 |
"step": 115
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 1.1136890951276102,
|
| 174 |
+
"grad_norm": 2.96875,
|
| 175 |
+
"learning_rate": 1.9554963520559003e-05,
|
| 176 |
+
"loss": 0.6394,
|
| 177 |
"step": 120
|
| 178 |
},
|
| 179 |
{
|
| 180 |
+
"epoch": 1.160092807424594,
|
| 181 |
+
"grad_norm": 3.421875,
|
| 182 |
+
"learning_rate": 1.9471122836330236e-05,
|
| 183 |
+
"loss": 0.6416,
|
| 184 |
"step": 125
|
| 185 |
},
|
| 186 |
{
|
| 187 |
+
"epoch": 1.2064965197215778,
|
| 188 |
+
"grad_norm": 3.40625,
|
| 189 |
+
"learning_rate": 1.9380263353597553e-05,
|
| 190 |
+
"loss": 0.6129,
|
| 191 |
"step": 130
|
| 192 |
},
|
| 193 |
{
|
| 194 |
+
"epoch": 1.2529002320185616,
|
| 195 |
+
"grad_norm": 3.359375,
|
| 196 |
+
"learning_rate": 1.9282452405919235e-05,
|
| 197 |
+
"loss": 0.6533,
|
| 198 |
"step": 135
|
| 199 |
},
|
| 200 |
{
|
| 201 |
+
"epoch": 1.2993039443155452,
|
| 202 |
+
"grad_norm": 3.515625,
|
| 203 |
+
"learning_rate": 1.9177762478400276e-05,
|
| 204 |
+
"loss": 0.6333,
|
| 205 |
"step": 140
|
| 206 |
},
|
| 207 |
{
|
| 208 |
+
"epoch": 1.345707656612529,
|
| 209 |
+
"grad_norm": 2.921875,
|
| 210 |
+
"learning_rate": 1.9066271153975602e-05,
|
| 211 |
+
"loss": 0.6426,
|
| 212 |
"step": 145
|
| 213 |
},
|
| 214 |
{
|
| 215 |
+
"epoch": 1.3921113689095128,
|
| 216 |
+
"grad_norm": 3.421875,
|
| 217 |
+
"learning_rate": 1.8948061055915395e-05,
|
| 218 |
+
"loss": 0.6492,
|
| 219 |
"step": 150
|
| 220 |
},
|
| 221 |
{
|
| 222 |
+
"epoch": 1.4385150812064964,
|
| 223 |
+
"grad_norm": 3.09375,
|
| 224 |
+
"learning_rate": 1.882321978659519e-05,
|
| 225 |
+
"loss": 0.6052,
|
| 226 |
"step": 155
|
| 227 |
},
|
| 228 |
{
|
| 229 |
+
"epoch": 1.4849187935034802,
|
| 230 |
+
"grad_norm": 3.15625,
|
| 231 |
+
"learning_rate": 1.869183986257606e-05,
|
| 232 |
+
"loss": 0.6471,
|
| 233 |
"step": 160
|
| 234 |
},
|
| 235 |
{
|
| 236 |
+
"epoch": 1.531322505800464,
|
| 237 |
+
"grad_norm": 3.15625,
|
| 238 |
+
"learning_rate": 1.8554018646043045e-05,
|
| 239 |
+
"loss": 0.6229,
|
| 240 |
"step": 165
|
| 241 |
},
|
| 242 |
{
|
| 243 |
+
"epoch": 1.5777262180974478,
|
| 244 |
+
"grad_norm": 3.46875,
|
| 245 |
+
"learning_rate": 1.840985827265262e-05,
|
| 246 |
+
"loss": 0.6153,
|
| 247 |
"step": 170
|
| 248 |
},
|
| 249 |
{
|
| 250 |
+
"epoch": 1.6241299303944317,
|
| 251 |
+
"grad_norm": 3.1875,
|
| 252 |
+
"learning_rate": 1.825946557584265e-05,
|
| 253 |
+
"loss": 0.6302,
|
| 254 |
"step": 175
|
| 255 |
},
|
| 256 |
{
|
| 257 |
+
"epoch": 1.6705336426914155,
|
| 258 |
+
"grad_norm": 3.421875,
|
| 259 |
+
"learning_rate": 1.810295200766097e-05,
|
| 260 |
+
"loss": 0.6296,
|
| 261 |
"step": 180
|
| 262 |
},
|
| 263 |
{
|
| 264 |
+
"epoch": 1.716937354988399,
|
| 265 |
+
"grad_norm": 3.328125,
|
| 266 |
+
"learning_rate": 1.794043355617121e-05,
|
| 267 |
+
"loss": 0.6141,
|
| 268 |
"step": 185
|
| 269 |
},
|
| 270 |
{
|
| 271 |
+
"epoch": 1.7633410672853829,
|
| 272 |
"grad_norm": 3.3125,
|
| 273 |
+
"learning_rate": 1.7772030659497112e-05,
|
| 274 |
+
"loss": 0.6288,
|
| 275 |
"step": 190
|
| 276 |
},
|
| 277 |
{
|
| 278 |
+
"epoch": 1.8097447795823665,
|
| 279 |
+
"grad_norm": 2.984375,
|
| 280 |
+
"learning_rate": 1.7597868116569036e-05,
|
| 281 |
+
"loss": 0.6376,
|
| 282 |
"step": 195
|
| 283 |
},
|
| 284 |
{
|
| 285 |
+
"epoch": 1.8561484918793503,
|
| 286 |
+
"grad_norm": 3.25,
|
| 287 |
+
"learning_rate": 1.7418074994638752e-05,
|
| 288 |
+
"loss": 0.6139,
|
| 289 |
"step": 200
|
| 290 |
},
|
| 291 |
{
|
| 292 |
+
"epoch": 1.902552204176334,
|
| 293 |
+
"grad_norm": 3.609375,
|
| 294 |
+
"learning_rate": 1.7232784533631148e-05,
|
| 295 |
+
"loss": 0.6259,
|
| 296 |
"step": 205
|
| 297 |
},
|
| 298 |
{
|
| 299 |
+
"epoch": 1.948955916473318,
|
| 300 |
+
"grad_norm": 2.9375,
|
| 301 |
+
"learning_rate": 1.7042134047403613e-05,
|
| 302 |
+
"loss": 0.6429,
|
| 303 |
"step": 210
|
| 304 |
},
|
| 305 |
{
|
| 306 |
+
"epoch": 1.9953596287703017,
|
| 307 |
+
"grad_norm": 3.703125,
|
| 308 |
+
"learning_rate": 1.684626482198639e-05,
|
| 309 |
+
"loss": 0.6532,
|
| 310 |
"step": 215
|
| 311 |
},
|
| 312 |
{
|
| 313 |
+
"epoch": 2.0417633410672855,
|
| 314 |
+
"grad_norm": 3.71875,
|
| 315 |
+
"learning_rate": 1.6645322010879242e-05,
|
| 316 |
+
"loss": 0.4533,
|
| 317 |
"step": 220
|
| 318 |
},
|
| 319 |
{
|
| 320 |
+
"epoch": 2.0881670533642693,
|
| 321 |
+
"grad_norm": 4.71875,
|
| 322 |
+
"learning_rate": 1.6439454527482014e-05,
|
| 323 |
+
"loss": 0.3677,
|
| 324 |
"step": 225
|
| 325 |
},
|
| 326 |
{
|
| 327 |
+
"epoch": 2.1345707656612527,
|
| 328 |
+
"grad_norm": 3.703125,
|
| 329 |
+
"learning_rate": 1.6228814934738873e-05,
|
| 330 |
+
"loss": 0.3657,
|
| 331 |
"step": 230
|
| 332 |
},
|
| 333 |
{
|
| 334 |
+
"epoch": 2.1809744779582365,
|
| 335 |
+
"grad_norm": 4.3125,
|
| 336 |
+
"learning_rate": 1.6013559332077945e-05,
|
| 337 |
+
"loss": 0.3359,
|
| 338 |
"step": 235
|
| 339 |
},
|
| 340 |
{
|
| 341 |
+
"epoch": 2.2273781902552203,
|
| 342 |
+
"grad_norm": 5.3125,
|
| 343 |
+
"learning_rate": 1.5793847239730148e-05,
|
| 344 |
+
"loss": 0.3438,
|
| 345 |
"step": 240
|
| 346 |
},
|
| 347 |
{
|
| 348 |
+
"epoch": 2.273781902552204,
|
| 349 |
+
"grad_norm": 4.4375,
|
| 350 |
+
"learning_rate": 1.5569841480512972e-05,
|
| 351 |
+
"loss": 0.3437,
|
| 352 |
"step": 245
|
| 353 |
},
|
| 354 |
{
|
| 355 |
+
"epoch": 2.320185614849188,
|
| 356 |
+
"grad_norm": 4.09375,
|
| 357 |
+
"learning_rate": 1.534170805916681e-05,
|
| 358 |
+
"loss": 0.3474,
|
| 359 |
"step": 250
|
| 360 |
},
|
| 361 |
{
|
| 362 |
+
"epoch": 2.3665893271461718,
|
| 363 |
+
"grad_norm": 4.65625,
|
| 364 |
+
"learning_rate": 1.510961603933324e-05,
|
| 365 |
+
"loss": 0.3529,
|
| 366 |
"step": 255
|
| 367 |
},
|
| 368 |
{
|
| 369 |
+
"epoch": 2.4129930394431556,
|
| 370 |
+
"grad_norm": 3.96875,
|
| 371 |
+
"learning_rate": 1.4873737418266398e-05,
|
| 372 |
+
"loss": 0.3486,
|
| 373 |
"step": 260
|
| 374 |
},
|
| 375 |
{
|
| 376 |
+
"epoch": 2.4593967517401394,
|
| 377 |
+
"grad_norm": 4.15625,
|
| 378 |
+
"learning_rate": 1.4634246999370415e-05,
|
| 379 |
+
"loss": 0.3345,
|
| 380 |
"step": 265
|
| 381 |
},
|
| 382 |
{
|
| 383 |
+
"epoch": 2.505800464037123,
|
| 384 |
+
"grad_norm": 4.125,
|
| 385 |
+
"learning_rate": 1.4391322262657206e-05,
|
| 386 |
+
"loss": 0.3498,
|
| 387 |
"step": 270
|
| 388 |
},
|
| 389 |
{
|
| 390 |
+
"epoch": 2.5522041763341066,
|
| 391 |
+
"grad_norm": 3.71875,
|
| 392 |
+
"learning_rate": 1.4145143233220741e-05,
|
| 393 |
+
"loss": 0.331,
|
| 394 |
"step": 275
|
| 395 |
},
|
| 396 |
{
|
| 397 |
+
"epoch": 2.5986078886310904,
|
| 398 |
+
"grad_norm": 4.28125,
|
| 399 |
+
"learning_rate": 1.3895892347825205e-05,
|
| 400 |
+
"loss": 0.3499,
|
| 401 |
"step": 280
|
| 402 |
},
|
| 403 |
{
|
| 404 |
+
"epoch": 2.645011600928074,
|
| 405 |
+
"grad_norm": 4.78125,
|
| 406 |
+
"learning_rate": 1.3643754319705956e-05,
|
| 407 |
+
"loss": 0.348,
|
| 408 |
"step": 285
|
| 409 |
},
|
| 410 |
{
|
| 411 |
+
"epoch": 2.691415313225058,
|
| 412 |
+
"grad_norm": 4.03125,
|
| 413 |
+
"learning_rate": 1.3388916001683412e-05,
|
| 414 |
+
"loss": 0.3527,
|
| 415 |
"step": 290
|
| 416 |
},
|
| 417 |
{
|
| 418 |
+
"epoch": 2.737819025522042,
|
| 419 |
+
"grad_norm": 4.0625,
|
| 420 |
+
"learning_rate": 1.3131566247691387e-05,
|
| 421 |
+
"loss": 0.3207,
|
| 422 |
"step": 295
|
| 423 |
},
|
| 424 |
{
|
| 425 |
+
"epoch": 2.7842227378190256,
|
| 426 |
+
"grad_norm": 4.25,
|
| 427 |
+
"learning_rate": 1.2871895772822442e-05,
|
| 428 |
+
"loss": 0.3458,
|
| 429 |
"step": 300
|
| 430 |
},
|
| 431 |
{
|
| 432 |
+
"epoch": 2.8306264501160094,
|
| 433 |
+
"grad_norm": 4.0625,
|
| 434 |
+
"learning_rate": 1.261009701199395e-05,
|
| 435 |
+
"loss": 0.3501,
|
| 436 |
"step": 305
|
| 437 |
},
|
| 438 |
{
|
| 439 |
+
"epoch": 2.877030162412993,
|
| 440 |
+
"grad_norm": 3.578125,
|
| 441 |
+
"learning_rate": 1.2346363977339698e-05,
|
| 442 |
+
"loss": 0.3583,
|
| 443 |
"step": 310
|
| 444 |
},
|
| 445 |
{
|
| 446 |
+
"epoch": 2.9234338747099766,
|
| 447 |
+
"grad_norm": 4.03125,
|
| 448 |
+
"learning_rate": 1.208089211443262e-05,
|
| 449 |
+
"loss": 0.3328,
|
| 450 |
"step": 315
|
| 451 |
},
|
| 452 |
{
|
| 453 |
+
"epoch": 2.9698375870069604,
|
| 454 |
+
"grad_norm": 3.890625,
|
| 455 |
+
"learning_rate": 1.1813878157445253e-05,
|
| 456 |
+
"loss": 0.3365,
|
| 457 |
"step": 320
|
| 458 |
},
|
| 459 |
{
|
| 460 |
+
"epoch": 3.0162412993039442,
|
| 461 |
+
"grad_norm": 3.53125,
|
| 462 |
+
"learning_rate": 1.1545519983355255e-05,
|
| 463 |
+
"loss": 0.3156,
|
| 464 |
"step": 325
|
| 465 |
},
|
| 466 |
{
|
| 467 |
+
"epoch": 3.062645011600928,
|
| 468 |
+
"grad_norm": 3.90625,
|
| 469 |
+
"learning_rate": 1.1276016465303989e-05,
|
| 470 |
+
"loss": 0.1769,
|
| 471 |
"step": 330
|
| 472 |
},
|
| 473 |
{
|
| 474 |
+
"epoch": 3.109048723897912,
|
| 475 |
+
"grad_norm": 5.78125,
|
| 476 |
+
"learning_rate": 1.1005567325216946e-05,
|
| 477 |
+
"loss": 0.1737,
|
| 478 |
"step": 335
|
| 479 |
},
|
| 480 |
{
|
| 481 |
+
"epoch": 3.1554524361948957,
|
| 482 |
+
"grad_norm": 4.28125,
|
| 483 |
+
"learning_rate": 1.0734372985795062e-05,
|
| 484 |
+
"loss": 0.165,
|
| 485 |
"step": 340
|
| 486 |
},
|
| 487 |
{
|
| 488 |
+
"epoch": 3.2018561484918795,
|
| 489 |
+
"grad_norm": 3.546875,
|
| 490 |
+
"learning_rate": 1.0462634421986786e-05,
|
| 491 |
+
"loss": 0.1701,
|
| 492 |
"step": 345
|
| 493 |
},
|
| 494 |
{
|
| 495 |
+
"epoch": 3.2482598607888633,
|
| 496 |
+
"grad_norm": 4.46875,
|
| 497 |
+
"learning_rate": 1.0190553012050868e-05,
|
| 498 |
+
"loss": 0.1707,
|
| 499 |
"step": 350
|
| 500 |
},
|
| 501 |
{
|
| 502 |
+
"epoch": 3.2946635730858467,
|
| 503 |
+
"grad_norm": 4.25,
|
| 504 |
+
"learning_rate": 9.918330388320235e-06,
|
| 505 |
+
"loss": 0.1603,
|
| 506 |
"step": 355
|
| 507 |
},
|
| 508 |
{
|
| 509 |
+
"epoch": 3.3410672853828305,
|
| 510 |
+
"grad_norm": 4.0625,
|
| 511 |
+
"learning_rate": 9.646168287777633e-06,
|
| 512 |
+
"loss": 0.1662,
|
| 513 |
"step": 360
|
| 514 |
},
|
| 515 |
{
|
| 516 |
+
"epoch": 3.3874709976798143,
|
| 517 |
+
"grad_norm": 4.46875,
|
| 518 |
+
"learning_rate": 9.374268402553665e-06,
|
| 519 |
+
"loss": 0.159,
|
| 520 |
"step": 365
|
| 521 |
},
|
| 522 |
{
|
| 523 |
+
"epoch": 3.433874709976798,
|
| 524 |
+
"grad_norm": 3.984375,
|
| 525 |
+
"learning_rate": 9.102832230458115e-06,
|
| 526 |
+
"loss": 0.163,
|
| 527 |
"step": 370
|
| 528 |
},
|
| 529 |
{
|
| 530 |
+
"epoch": 3.480278422273782,
|
| 531 |
+
"grad_norm": 4.09375,
|
| 532 |
+
"learning_rate": 8.83206092565522e-06,
|
| 533 |
+
"loss": 0.1591,
|
| 534 |
"step": 375
|
| 535 |
},
|
| 536 |
{
|
| 537 |
+
"epoch": 3.5266821345707657,
|
| 538 |
+
"grad_norm": 4.34375,
|
| 539 |
+
"learning_rate": 8.562155149593673e-06,
|
| 540 |
+
"loss": 0.1706,
|
| 541 |
"step": 380
|
| 542 |
},
|
| 543 |
{
|
| 544 |
+
"epoch": 3.5730858468677495,
|
| 545 |
+
"grad_norm": 4.4375,
|
| 546 |
+
"learning_rate": 8.293314922301715e-06,
|
| 547 |
+
"loss": 0.1595,
|
| 548 |
"step": 385
|
| 549 |
},
|
| 550 |
{
|
| 551 |
+
"epoch": 3.619489559164733,
|
| 552 |
+
"grad_norm": 4.03125,
|
| 553 |
+
"learning_rate": 8.025739474157595e-06,
|
| 554 |
+
"loss": 0.1637,
|
| 555 |
"step": 390
|
| 556 |
},
|
| 557 |
{
|
| 558 |
+
"epoch": 3.6658932714617167,
|
| 559 |
+
"grad_norm": 3.828125,
|
| 560 |
+
"learning_rate": 7.759627098245207e-06,
|
| 561 |
+
"loss": 0.1665,
|
| 562 |
"step": 395
|
| 563 |
},
|
| 564 |
{
|
| 565 |
+
"epoch": 3.7122969837587005,
|
| 566 |
+
"grad_norm": 3.953125,
|
| 567 |
+
"learning_rate": 7.49517500340432e-06,
|
| 568 |
+
"loss": 0.167,
|
| 569 |
"step": 400
|
| 570 |
},
|
| 571 |
{
|
| 572 |
+
"epoch": 3.7587006960556844,
|
| 573 |
"grad_norm": 3.78125,
|
| 574 |
+
"learning_rate": 7.232579168084344e-06,
|
| 575 |
+
"loss": 0.1533,
|
| 576 |
"step": 405
|
| 577 |
},
|
| 578 |
{
|
| 579 |
+
"epoch": 3.805104408352668,
|
| 580 |
+
"grad_norm": 4.3125,
|
| 581 |
+
"learning_rate": 6.972034195109885e-06,
|
| 582 |
+
"loss": 0.1651,
|
| 583 |
"step": 410
|
| 584 |
},
|
| 585 |
{
|
| 586 |
+
"epoch": 3.851508120649652,
|
| 587 |
+
"grad_norm": 3.640625,
|
| 588 |
+
"learning_rate": 6.713733167465723e-06,
|
| 589 |
+
"loss": 0.1673,
|
| 590 |
"step": 415
|
| 591 |
},
|
| 592 |
{
|
| 593 |
+
"epoch": 3.897911832946636,
|
| 594 |
+
"grad_norm": 4.125,
|
| 595 |
+
"learning_rate": 6.4578675052081395e-06,
|
| 596 |
+
"loss": 0.1576,
|
| 597 |
"step": 420
|
| 598 |
},
|
| 599 |
{
|
| 600 |
+
"epoch": 3.9443155452436196,
|
| 601 |
+
"grad_norm": 3.96875,
|
| 602 |
+
"learning_rate": 6.204626823608584e-06,
|
| 603 |
+
"loss": 0.1688,
|
| 604 |
"step": 425
|
| 605 |
},
|
| 606 |
{
|
| 607 |
+
"epoch": 3.9907192575406034,
|
| 608 |
+
"grad_norm": 3.78125,
|
| 609 |
+
"learning_rate": 5.954198792634782e-06,
|
| 610 |
+
"loss": 0.1571,
|
| 611 |
"step": 430
|
| 612 |
},
|
| 613 |
{
|
| 614 |
+
"epoch": 4.037122969837587,
|
| 615 |
+
"grad_norm": 2.546875,
|
| 616 |
+
"learning_rate": 5.706768997873533e-06,
|
| 617 |
+
"loss": 0.1248,
|
| 618 |
"step": 435
|
| 619 |
},
|
| 620 |
{
|
| 621 |
+
"epoch": 4.083526682134571,
|
| 622 |
+
"grad_norm": 2.1875,
|
| 623 |
+
"learning_rate": 5.462520802998108e-06,
|
| 624 |
+
"loss": 0.0991,
|
| 625 |
"step": 440
|
| 626 |
},
|
| 627 |
{
|
| 628 |
+
"epoch": 4.129930394431555,
|
| 629 |
+
"grad_norm": 3.234375,
|
| 630 |
+
"learning_rate": 5.221635213882295e-06,
|
| 631 |
+
"loss": 0.0951,
|
| 632 |
"step": 445
|
| 633 |
},
|
| 634 |
{
|
| 635 |
+
"epoch": 4.176334106728539,
|
| 636 |
+
"grad_norm": 3.46875,
|
| 637 |
+
"learning_rate": 4.9842907444617415e-06,
|
| 638 |
+
"loss": 0.1023,
|
| 639 |
"step": 450
|
| 640 |
},
|
| 641 |
{
|
| 642 |
+
"epoch": 4.222737819025522,
|
| 643 |
+
"grad_norm": 2.90625,
|
| 644 |
+
"learning_rate": 4.750663284442001e-06,
|
| 645 |
+
"loss": 0.0941,
|
| 646 |
"step": 455
|
| 647 |
},
|
| 648 |
{
|
| 649 |
+
"epoch": 4.269141531322505,
|
| 650 |
+
"grad_norm": 2.984375,
|
| 651 |
+
"learning_rate": 4.52092596895131e-06,
|
| 652 |
+
"loss": 0.0942,
|
| 653 |
"step": 460
|
| 654 |
},
|
| 655 |
{
|
| 656 |
+
"epoch": 4.315545243619489,
|
| 657 |
+
"grad_norm": 2.90625,
|
| 658 |
+
"learning_rate": 4.295249050234738e-06,
|
| 659 |
+
"loss": 0.0941,
|
| 660 |
"step": 465
|
| 661 |
},
|
| 662 |
{
|
| 663 |
+
"epoch": 4.361948955916473,
|
| 664 |
+
"grad_norm": 3.046875,
|
| 665 |
+
"learning_rate": 4.073799771484768e-06,
|
| 666 |
+
"loss": 0.0956,
|
| 667 |
"step": 470
|
| 668 |
},
|
| 669 |
{
|
| 670 |
+
"epoch": 4.408352668213457,
|
| 671 |
+
"grad_norm": 2.578125,
|
| 672 |
+
"learning_rate": 3.8567422429017585e-06,
|
| 673 |
+
"loss": 0.097,
|
| 674 |
"step": 475
|
| 675 |
},
|
| 676 |
{
|
| 677 |
+
"epoch": 4.454756380510441,
|
| 678 |
+
"grad_norm": 2.546875,
|
| 679 |
+
"learning_rate": 3.644237320076256e-06,
|
| 680 |
+
"loss": 0.0963,
|
| 681 |
"step": 480
|
| 682 |
},
|
| 683 |
{
|
| 684 |
+
"epoch": 4.5011600928074245,
|
| 685 |
+
"grad_norm": 2.984375,
|
| 686 |
+
"learning_rate": 3.436442484783138e-06,
|
| 687 |
+
"loss": 0.0993,
|
| 688 |
"step": 485
|
| 689 |
},
|
| 690 |
{
|
| 691 |
+
"epoch": 4.547563805104408,
|
| 692 |
+
"grad_norm": 3.015625,
|
| 693 |
+
"learning_rate": 3.2335117282760563e-06,
|
| 694 |
+
"loss": 0.0954,
|
| 695 |
"step": 490
|
| 696 |
},
|
| 697 |
{
|
| 698 |
+
"epoch": 4.593967517401392,
|
| 699 |
+
"grad_norm": 2.890625,
|
| 700 |
+
"learning_rate": 3.0355954371685948e-06,
|
| 701 |
+
"loss": 0.0967,
|
| 702 |
"step": 495
|
| 703 |
},
|
| 704 |
{
|
| 705 |
+
"epoch": 4.640371229698376,
|
| 706 |
+
"grad_norm": 2.40625,
|
| 707 |
+
"learning_rate": 2.842840281986726e-06,
|
| 708 |
+
"loss": 0.0988,
|
| 709 |
"step": 500
|
| 710 |
},
|
| 711 |
{
|
| 712 |
+
"epoch": 4.68677494199536,
|
| 713 |
+
"grad_norm": 2.78125,
|
| 714 |
+
"learning_rate": 2.6553891084751604e-06,
|
| 715 |
+
"loss": 0.0976,
|
| 716 |
"step": 505
|
| 717 |
},
|
| 718 |
{
|
| 719 |
+
"epoch": 4.7331786542923435,
|
| 720 |
+
"grad_norm": 3.578125,
|
| 721 |
+
"learning_rate": 2.473380831738146e-06,
|
| 722 |
+
"loss": 0.094,
|
| 723 |
"step": 510
|
| 724 |
},
|
| 725 |
{
|
| 726 |
+
"epoch": 4.779582366589327,
|
| 727 |
+
"grad_norm": 2.5625,
|
| 728 |
+
"learning_rate": 2.2969503332931754e-06,
|
| 729 |
+
"loss": 0.0956,
|
| 730 |
"step": 515
|
| 731 |
},
|
| 732 |
{
|
| 733 |
+
"epoch": 4.825986078886311,
|
| 734 |
+
"grad_norm": 3.0,
|
| 735 |
+
"learning_rate": 2.126228361113839e-06,
|
| 736 |
+
"loss": 0.0988,
|
| 737 |
"step": 520
|
| 738 |
},
|
| 739 |
{
|
| 740 |
+
"epoch": 4.872389791183295,
|
| 741 |
+
"grad_norm": 2.703125,
|
| 742 |
+
"learning_rate": 1.9613414327359824e-06,
|
| 743 |
+
"loss": 0.098,
|
| 744 |
"step": 525
|
| 745 |
},
|
| 746 |
{
|
| 747 |
+
"epoch": 4.918793503480279,
|
| 748 |
+
"grad_norm": 2.71875,
|
| 749 |
+
"learning_rate": 1.8024117414989007e-06,
|
| 750 |
+
"loss": 0.0968,
|
| 751 |
"step": 530
|
| 752 |
},
|
| 753 |
{
|
| 754 |
+
"epoch": 4.965197215777263,
|
| 755 |
+
"grad_norm": 2.75,
|
| 756 |
+
"learning_rate": 1.649557065991081e-06,
|
| 757 |
+
"loss": 0.0938,
|
| 758 |
"step": 535
|
| 759 |
},
|
| 760 |
{
|
| 761 |
+
"epoch": 5.0116009280742455,
|
| 762 |
+
"grad_norm": 1.75,
|
| 763 |
+
"learning_rate": 1.5028906827676148e-06,
|
| 764 |
+
"loss": 0.0965,
|
| 765 |
"step": 540
|
| 766 |
},
|
| 767 |
{
|
| 768 |
+
"epoch": 5.058004640371229,
|
| 769 |
+
"grad_norm": 1.84375,
|
| 770 |
+
"learning_rate": 1.3625212824039468e-06,
|
| 771 |
+
"loss": 0.0829,
|
| 772 |
"step": 545
|
| 773 |
},
|
| 774 |
{
|
| 775 |
+
"epoch": 5.104408352668213,
|
| 776 |
+
"grad_norm": 2.015625,
|
| 777 |
+
"learning_rate": 1.228552888948149e-06,
|
| 778 |
+
"loss": 0.0819,
|
| 779 |
"step": 550
|
| 780 |
},
|
| 781 |
{
|
| 782 |
+
"epoch": 5.150812064965197,
|
| 783 |
+
"grad_norm": 2.078125,
|
| 784 |
+
"learning_rate": 1.1010847828314708e-06,
|
| 785 |
+
"loss": 0.082,
|
| 786 |
"step": 555
|
| 787 |
},
|
| 788 |
{
|
| 789 |
+
"epoch": 5.197215777262181,
|
| 790 |
+
"grad_norm": 2.109375,
|
| 791 |
+
"learning_rate": 9.80211427294222e-07,
|
| 792 |
+
"loss": 0.0849,
|
| 793 |
"step": 560
|
| 794 |
},
|
| 795 |
{
|
| 796 |
+
"epoch": 5.243619489559165,
|
| 797 |
+
"grad_norm": 1.875,
|
| 798 |
+
"learning_rate": 8.660223983815708e-07,
|
| 799 |
+
"loss": 0.0807,
|
| 800 |
"step": 565
|
| 801 |
},
|
| 802 |
{
|
| 803 |
+
"epoch": 5.290023201856148,
|
| 804 |
+
"grad_norm": 2.359375,
|
| 805 |
+
"learning_rate": 7.586023185611136e-07,
|
| 806 |
+
"loss": 0.0866,
|
| 807 |
"step": 570
|
| 808 |
},
|
| 809 |
{
|
| 810 |
+
"epoch": 5.336426914153132,
|
| 811 |
+
"grad_norm": 2.078125,
|
| 812 |
+
"learning_rate": 6.580307940113972e-07,
|
| 813 |
+
"loss": 0.0844,
|
| 814 |
"step": 575
|
| 815 |
},
|
| 816 |
{
|
| 817 |
+
"epoch": 5.382830626450116,
|
| 818 |
+
"grad_norm": 1.953125,
|
| 819 |
+
"learning_rate": 5.643823556278849e-07,
|
| 820 |
+
"loss": 0.0846,
|
| 821 |
"step": 580
|
| 822 |
},
|
| 823 |
{
|
| 824 |
+
"epoch": 5.4292343387471,
|
| 825 |
+
"grad_norm": 2.15625,
|
| 826 |
+
"learning_rate": 4.777264037900841e-07,
|
| 827 |
+
"loss": 0.0841,
|
| 828 |
"step": 585
|
| 829 |
},
|
| 830 |
{
|
| 831 |
+
"epoch": 5.475638051044084,
|
| 832 |
+
"grad_norm": 1.640625,
|
| 833 |
+
"learning_rate": 3.981271569307654e-07,
|
| 834 |
+
"loss": 0.0816,
|
| 835 |
"step": 590
|
| 836 |
},
|
| 837 |
{
|
| 838 |
+
"epoch": 5.522041763341067,
|
| 839 |
+
"grad_norm": 1.7578125,
|
| 840 |
+
"learning_rate": 3.2564360394537696e-07,
|
| 841 |
+
"loss": 0.0845,
|
| 842 |
"step": 595
|
| 843 |
},
|
| 844 |
{
|
| 845 |
+
"epoch": 5.568445475638051,
|
| 846 |
+
"grad_norm": 2.109375,
|
| 847 |
+
"learning_rate": 2.6032946047693794e-07,
|
| 848 |
+
"loss": 0.0823,
|
| 849 |
"step": 600
|
| 850 |
},
|
| 851 |
{
|
| 852 |
+
"epoch": 5.614849187935035,
|
| 853 |
+
"grad_norm": 2.03125,
|
| 854 |
+
"learning_rate": 2.02233129108792e-07,
|
| 855 |
+
"loss": 0.0829,
|
| 856 |
"step": 605
|
| 857 |
},
|
| 858 |
{
|
| 859 |
+
"epoch": 5.661252900232019,
|
| 860 |
+
"grad_norm": 2.046875,
|
| 861 |
+
"learning_rate": 1.5139766349474004e-07,
|
| 862 |
+
"loss": 0.0829,
|
| 863 |
"step": 610
|
| 864 |
},
|
| 865 |
{
|
| 866 |
+
"epoch": 5.707656612529003,
|
| 867 |
+
"grad_norm": 2.078125,
|
| 868 |
+
"learning_rate": 1.0786073645311035e-07,
|
| 869 |
+
"loss": 0.085,
|
| 870 |
"step": 615
|
| 871 |
},
|
| 872 |
{
|
| 873 |
+
"epoch": 5.754060324825986,
|
| 874 |
+
"grad_norm": 2.078125,
|
| 875 |
+
"learning_rate": 7.165461204843738e-08,
|
| 876 |
+
"loss": 0.0842,
|
| 877 |
"step": 620
|
| 878 |
},
|
| 879 |
{
|
| 880 |
+
"epoch": 5.800464037122969,
|
| 881 |
+
"grad_norm": 2.265625,
|
| 882 |
+
"learning_rate": 4.2806121681409076e-08,
|
| 883 |
+
"loss": 0.0809,
|
| 884 |
"step": 625
|
| 885 |
},
|
| 886 |
{
|
| 887 |
+
"epoch": 5.846867749419953,
|
| 888 |
+
"grad_norm": 1.90625,
|
| 889 |
+
"learning_rate": 2.1336644204834613e-08,
|
| 890 |
+
"loss": 0.08,
|
| 891 |
"step": 630
|
| 892 |
},
|
| 893 |
{
|
| 894 |
+
"epoch": 5.893271461716937,
|
| 895 |
+
"grad_norm": 2.390625,
|
| 896 |
+
"learning_rate": 7.262090080331075e-09,
|
| 897 |
+
"loss": 0.0871,
|
| 898 |
"step": 635
|
| 899 |
},
|
| 900 |
{
|
| 901 |
+
"epoch": 5.939675174013921,
|
| 902 |
+
"grad_norm": 1.921875,
|
| 903 |
+
"learning_rate": 5.92889587515133e-10,
|
| 904 |
+
"loss": 0.0866,
|
| 905 |
"step": 640
|
| 906 |
},
|
| 907 |
{
|
| 908 |
+
"epoch": 5.958236658932715,
|
| 909 |
+
"step": 642,
|
| 910 |
+
"total_flos": 6.526533872789914e+16,
|
| 911 |
+
"train_loss": 0.40725265422435564,
|
| 912 |
+
"train_runtime": 2315.0788,
|
| 913 |
+
"train_samples_per_second": 8.936,
|
| 914 |
+
"train_steps_per_second": 0.277
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 915 |
}
|
| 916 |
],
|
| 917 |
"logging_steps": 5,
|
| 918 |
+
"max_steps": 642,
|
| 919 |
"num_input_tokens_seen": 0,
|
| 920 |
"num_train_epochs": 6,
|
| 921 |
"save_steps": 999999,
|
|
|
|
| 931 |
"attributes": {}
|
| 932 |
}
|
| 933 |
},
|
| 934 |
+
"total_flos": 6.526533872789914e+16,
|
| 935 |
"train_batch_size": 8,
|
| 936 |
"trial_name": null,
|
| 937 |
"trial_params": null
|