Model save
Browse files- README.md +2 -4
- all_results.json +5 -5
- model-00001-of-00002.safetensors +1 -1
- model-00002-of-00002.safetensors +1 -1
- train_results.json +5 -5
- trainer_state.json +1535 -182
README.md
CHANGED
|
@@ -1,11 +1,9 @@
|
|
| 1 |
---
|
| 2 |
base_model: Qwen/Qwen2.5-3B-Instruct
|
| 3 |
-
datasets: kguo2/scaffold_finetune
|
| 4 |
library_name: transformers
|
| 5 |
model_name: finetune_demo
|
| 6 |
tags:
|
| 7 |
- generated_from_trainer
|
| 8 |
-
- open-r1
|
| 9 |
- trl
|
| 10 |
- sft
|
| 11 |
licence: license
|
|
@@ -13,7 +11,7 @@ licence: license
|
|
| 13 |
|
| 14 |
# Model Card for finetune_demo
|
| 15 |
|
| 16 |
-
This model is a fine-tuned version of [Qwen/Qwen2.5-3B-Instruct](https://huggingface.co/Qwen/Qwen2.5-3B-Instruct)
|
| 17 |
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 18 |
|
| 19 |
## Quick start
|
|
@@ -29,7 +27,7 @@ print(output["generated_text"])
|
|
| 29 |
|
| 30 |
## Training procedure
|
| 31 |
|
| 32 |
-
[<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/nd/huggingface/runs/
|
| 33 |
|
| 34 |
|
| 35 |
This model was trained with SFT.
|
|
|
|
| 1 |
---
|
| 2 |
base_model: Qwen/Qwen2.5-3B-Instruct
|
|
|
|
| 3 |
library_name: transformers
|
| 4 |
model_name: finetune_demo
|
| 5 |
tags:
|
| 6 |
- generated_from_trainer
|
|
|
|
| 7 |
- trl
|
| 8 |
- sft
|
| 9 |
licence: license
|
|
|
|
| 11 |
|
| 12 |
# Model Card for finetune_demo
|
| 13 |
|
| 14 |
+
This model is a fine-tuned version of [Qwen/Qwen2.5-3B-Instruct](https://huggingface.co/Qwen/Qwen2.5-3B-Instruct).
|
| 15 |
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 16 |
|
| 17 |
## Quick start
|
|
|
|
| 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/nd/huggingface/runs/mvkr71e8)
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with SFT.
|
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"total_flos":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 29047,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"total_flos": 2.7740847281799168e+17,
|
| 3 |
+
"train_loss": 0.0704414379602571,
|
| 4 |
+
"train_runtime": 3418.2006,
|
| 5 |
"train_samples": 29047,
|
| 6 |
+
"train_samples_per_second": 18.999,
|
| 7 |
+
"train_steps_per_second": 0.298
|
| 8 |
}
|
model-00001-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4957560304
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:85790087dcd5089191b26416bd4153cb60820697feb02b07f4c62255c81e612b
|
| 3 |
size 4957560304
|
model-00002-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1836696752
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:43badf75af60904b4dd685210c63f876f68df9fd045abde724222125c6cec9fa
|
| 3 |
size 1836696752
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"total_flos":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 29047,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"total_flos": 2.7740847281799168e+17,
|
| 3 |
+
"train_loss": 0.0704414379602571,
|
| 4 |
+
"train_runtime": 3418.2006,
|
| 5 |
"train_samples": 29047,
|
| 6 |
+
"train_samples_per_second": 18.999,
|
| 7 |
+
"train_steps_per_second": 0.298
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -2,299 +2,1652 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch":
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
-
"epoch": 0.
|
| 14 |
-
"grad_norm":
|
| 15 |
-
"learning_rate":
|
| 16 |
-
"loss":
|
| 17 |
-
"num_tokens":
|
| 18 |
"step": 5
|
| 19 |
},
|
| 20 |
{
|
| 21 |
-
"epoch": 0.
|
| 22 |
-
"grad_norm":
|
| 23 |
-
"learning_rate":
|
| 24 |
-
"loss": 0.
|
| 25 |
-
"num_tokens":
|
| 26 |
"step": 10
|
| 27 |
},
|
| 28 |
{
|
| 29 |
-
"epoch": 0.
|
| 30 |
-
"grad_norm":
|
| 31 |
-
"learning_rate":
|
| 32 |
-
"loss": 0.
|
| 33 |
-
"num_tokens":
|
| 34 |
"step": 15
|
| 35 |
},
|
| 36 |
{
|
| 37 |
-
"epoch": 0.
|
| 38 |
-
"grad_norm":
|
| 39 |
-
"learning_rate":
|
| 40 |
-
"loss": 0.
|
| 41 |
-
"num_tokens":
|
| 42 |
"step": 20
|
| 43 |
},
|
| 44 |
{
|
| 45 |
-
"epoch": 0.
|
| 46 |
-
"grad_norm":
|
| 47 |
-
"learning_rate":
|
| 48 |
-
"loss": 0.
|
| 49 |
-
"num_tokens":
|
| 50 |
"step": 25
|
| 51 |
},
|
| 52 |
{
|
| 53 |
-
"epoch": 0.
|
| 54 |
-
"grad_norm":
|
| 55 |
-
"learning_rate":
|
| 56 |
-
"loss": 0.
|
| 57 |
-
"num_tokens":
|
| 58 |
"step": 30
|
| 59 |
},
|
| 60 |
{
|
| 61 |
-
"epoch": 0.
|
| 62 |
-
"grad_norm":
|
| 63 |
-
"learning_rate":
|
| 64 |
-
"loss": 0.
|
| 65 |
-
"num_tokens":
|
| 66 |
"step": 35
|
| 67 |
},
|
| 68 |
{
|
| 69 |
-
"epoch": 0.
|
| 70 |
-
"grad_norm":
|
| 71 |
-
"learning_rate":
|
| 72 |
-
"loss": 0.
|
| 73 |
-
"num_tokens":
|
| 74 |
"step": 40
|
| 75 |
},
|
| 76 |
{
|
| 77 |
-
"epoch": 0.
|
| 78 |
-
"grad_norm": 0.
|
| 79 |
-
"learning_rate": 4.
|
| 80 |
-
"loss": 0.
|
| 81 |
-
"num_tokens":
|
| 82 |
"step": 45
|
| 83 |
},
|
| 84 |
{
|
| 85 |
-
"epoch": 0.
|
| 86 |
-
"grad_norm": 0.
|
| 87 |
-
"learning_rate": 4.
|
| 88 |
-
"loss": 0.
|
| 89 |
-
"num_tokens":
|
| 90 |
"step": 50
|
| 91 |
},
|
| 92 |
{
|
| 93 |
-
"epoch": 0.
|
| 94 |
-
"grad_norm":
|
| 95 |
-
"learning_rate": 4.
|
| 96 |
-
"loss": 0.
|
| 97 |
-
"num_tokens":
|
| 98 |
"step": 55
|
| 99 |
},
|
| 100 |
{
|
| 101 |
-
"epoch": 0.
|
| 102 |
-
"grad_norm":
|
| 103 |
-
"learning_rate":
|
| 104 |
-
"loss": 0.
|
| 105 |
-
"num_tokens":
|
| 106 |
"step": 60
|
| 107 |
},
|
| 108 |
{
|
| 109 |
-
"epoch": 0.
|
| 110 |
-
"grad_norm": 0.
|
| 111 |
-
"learning_rate":
|
| 112 |
-
"loss": 0.
|
| 113 |
-
"num_tokens":
|
| 114 |
"step": 65
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm": 0.
|
| 119 |
-
"learning_rate":
|
| 120 |
-
"loss": 0.
|
| 121 |
-
"num_tokens":
|
| 122 |
"step": 70
|
| 123 |
},
|
| 124 |
{
|
| 125 |
-
"epoch": 0.
|
| 126 |
-
"grad_norm": 0.
|
| 127 |
-
"learning_rate":
|
| 128 |
-
"loss": 0.
|
| 129 |
-
"num_tokens":
|
| 130 |
"step": 75
|
| 131 |
},
|
| 132 |
{
|
| 133 |
-
"epoch": 0.
|
| 134 |
-
"grad_norm": 0.
|
| 135 |
-
"learning_rate":
|
| 136 |
-
"loss": 0.
|
| 137 |
-
"num_tokens":
|
| 138 |
"step": 80
|
| 139 |
},
|
| 140 |
{
|
| 141 |
-
"epoch": 0.
|
| 142 |
-
"grad_norm": 0.
|
| 143 |
-
"learning_rate":
|
| 144 |
-
"loss": 0.
|
| 145 |
-
"num_tokens":
|
| 146 |
"step": 85
|
| 147 |
},
|
| 148 |
{
|
| 149 |
-
"epoch": 0.
|
| 150 |
-
"grad_norm": 0.
|
| 151 |
-
"learning_rate":
|
| 152 |
-
"loss": 0.
|
| 153 |
-
"num_tokens":
|
| 154 |
"step": 90
|
| 155 |
},
|
| 156 |
{
|
| 157 |
-
"epoch": 0.
|
| 158 |
-
"grad_norm": 0.
|
| 159 |
-
"learning_rate":
|
| 160 |
-
"loss": 0.
|
| 161 |
-
"num_tokens":
|
| 162 |
"step": 95
|
| 163 |
},
|
| 164 |
{
|
| 165 |
-
"epoch": 0.
|
| 166 |
-
"grad_norm": 0.
|
| 167 |
-
"learning_rate":
|
| 168 |
-
"loss": 0.
|
| 169 |
-
"num_tokens":
|
| 170 |
"step": 100
|
| 171 |
},
|
| 172 |
{
|
| 173 |
-
"epoch": 0.
|
| 174 |
-
"grad_norm": 0.
|
| 175 |
-
"learning_rate":
|
| 176 |
-
"loss": 0.
|
| 177 |
-
"num_tokens":
|
| 178 |
"step": 105
|
| 179 |
},
|
| 180 |
{
|
| 181 |
-
"epoch": 0.
|
| 182 |
-
"grad_norm": 0.
|
| 183 |
-
"learning_rate":
|
| 184 |
-
"loss": 0.
|
| 185 |
-
"num_tokens":
|
| 186 |
"step": 110
|
| 187 |
},
|
| 188 |
{
|
| 189 |
-
"epoch": 0.
|
| 190 |
-
"grad_norm": 0.
|
| 191 |
-
"learning_rate":
|
| 192 |
-
"loss": 0.
|
| 193 |
-
"num_tokens":
|
| 194 |
"step": 115
|
| 195 |
},
|
| 196 |
{
|
| 197 |
-
"epoch": 0.
|
| 198 |
-
"grad_norm": 0.
|
| 199 |
-
"learning_rate":
|
| 200 |
-
"loss": 0.
|
| 201 |
-
"num_tokens":
|
| 202 |
"step": 120
|
| 203 |
},
|
| 204 |
{
|
| 205 |
-
"epoch": 0.
|
| 206 |
-
"grad_norm": 0.
|
| 207 |
-
"learning_rate":
|
| 208 |
-
"loss": 0.
|
| 209 |
-
"num_tokens":
|
| 210 |
"step": 125
|
| 211 |
},
|
| 212 |
{
|
| 213 |
-
"epoch": 0.
|
| 214 |
-
"grad_norm": 0.
|
| 215 |
-
"learning_rate":
|
| 216 |
-
"loss": 0.
|
| 217 |
-
"num_tokens":
|
| 218 |
"step": 130
|
| 219 |
},
|
| 220 |
{
|
| 221 |
-
"epoch": 0.
|
| 222 |
-
"grad_norm": 0.
|
| 223 |
-
"learning_rate":
|
| 224 |
-
"loss": 0.
|
| 225 |
-
"num_tokens":
|
| 226 |
"step": 135
|
| 227 |
},
|
| 228 |
{
|
| 229 |
-
"epoch": 0.
|
| 230 |
-
"grad_norm": 0.
|
| 231 |
-
"learning_rate":
|
| 232 |
-
"loss": 0.
|
| 233 |
-
"num_tokens":
|
| 234 |
"step": 140
|
| 235 |
},
|
| 236 |
{
|
| 237 |
-
"epoch": 0.
|
| 238 |
-
"grad_norm": 0.
|
| 239 |
-
"learning_rate":
|
| 240 |
-
"loss": 0.
|
| 241 |
-
"num_tokens":
|
| 242 |
"step": 145
|
| 243 |
},
|
| 244 |
{
|
| 245 |
-
"epoch": 0.
|
| 246 |
-
"grad_norm": 0.
|
| 247 |
-
"learning_rate":
|
| 248 |
-
"loss": 0.
|
| 249 |
-
"num_tokens":
|
| 250 |
"step": 150
|
| 251 |
},
|
| 252 |
{
|
| 253 |
-
"epoch": 0.
|
| 254 |
-
"grad_norm": 0.
|
| 255 |
-
"learning_rate":
|
| 256 |
-
"loss": 0.
|
| 257 |
-
"num_tokens":
|
| 258 |
"step": 155
|
| 259 |
},
|
| 260 |
{
|
| 261 |
-
"epoch": 0.
|
| 262 |
-
"grad_norm": 0.
|
| 263 |
-
"learning_rate":
|
| 264 |
-
"loss": 0.
|
| 265 |
-
"num_tokens":
|
| 266 |
"step": 160
|
| 267 |
},
|
| 268 |
{
|
| 269 |
-
"epoch": 0.
|
| 270 |
-
"grad_norm": 0.
|
| 271 |
-
"learning_rate":
|
| 272 |
-
"loss": 0.
|
| 273 |
-
"num_tokens":
|
| 274 |
"step": 165
|
| 275 |
},
|
| 276 |
{
|
| 277 |
-
"epoch":
|
| 278 |
-
"grad_norm": 0.
|
| 279 |
-
"learning_rate":
|
| 280 |
-
"loss": 0.
|
| 281 |
-
"num_tokens":
|
| 282 |
"step": 170
|
| 283 |
},
|
| 284 |
{
|
| 285 |
-
"epoch":
|
| 286 |
-
"
|
| 287 |
-
"
|
| 288 |
-
"
|
| 289 |
-
"
|
| 290 |
-
"
|
| 291 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 292 |
}
|
| 293 |
],
|
| 294 |
"logging_steps": 5,
|
| 295 |
-
"max_steps":
|
| 296 |
"num_input_tokens_seen": 0,
|
| 297 |
-
"num_train_epochs":
|
| 298 |
"save_steps": 100,
|
| 299 |
"stateful_callbacks": {
|
| 300 |
"TrainerControl": {
|
|
@@ -308,7 +1661,7 @@
|
|
| 308 |
"attributes": {}
|
| 309 |
}
|
| 310 |
},
|
| 311 |
-
"total_flos":
|
| 312 |
"train_batch_size": 16,
|
| 313 |
"trial_name": null,
|
| 314 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 3.0,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 1017,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
+
"epoch": 0.014749262536873156,
|
| 14 |
+
"grad_norm": 28.681703567504883,
|
| 15 |
+
"learning_rate": 4.901960784313726e-06,
|
| 16 |
+
"loss": 1.7339,
|
| 17 |
+
"num_tokens": 81920.0,
|
| 18 |
"step": 5
|
| 19 |
},
|
| 20 |
{
|
| 21 |
+
"epoch": 0.029498525073746312,
|
| 22 |
+
"grad_norm": 6.434526443481445,
|
| 23 |
+
"learning_rate": 9.803921568627451e-06,
|
| 24 |
+
"loss": 0.985,
|
| 25 |
+
"num_tokens": 163840.0,
|
| 26 |
"step": 10
|
| 27 |
},
|
| 28 |
{
|
| 29 |
+
"epoch": 0.04424778761061947,
|
| 30 |
+
"grad_norm": 2.7200655937194824,
|
| 31 |
+
"learning_rate": 1.4705882352941177e-05,
|
| 32 |
+
"loss": 0.2836,
|
| 33 |
+
"num_tokens": 245760.0,
|
| 34 |
"step": 15
|
| 35 |
},
|
| 36 |
{
|
| 37 |
+
"epoch": 0.058997050147492625,
|
| 38 |
+
"grad_norm": 1.4244611263275146,
|
| 39 |
+
"learning_rate": 1.9607843137254903e-05,
|
| 40 |
+
"loss": 0.1205,
|
| 41 |
+
"num_tokens": 327680.0,
|
| 42 |
"step": 20
|
| 43 |
},
|
| 44 |
{
|
| 45 |
+
"epoch": 0.07374631268436578,
|
| 46 |
+
"grad_norm": 1.7670924663543701,
|
| 47 |
+
"learning_rate": 2.4509803921568626e-05,
|
| 48 |
+
"loss": 0.0935,
|
| 49 |
+
"num_tokens": 409600.0,
|
| 50 |
"step": 25
|
| 51 |
},
|
| 52 |
{
|
| 53 |
+
"epoch": 0.08849557522123894,
|
| 54 |
+
"grad_norm": 1.4682564735412598,
|
| 55 |
+
"learning_rate": 2.9411764705882354e-05,
|
| 56 |
+
"loss": 0.083,
|
| 57 |
+
"num_tokens": 491479.0,
|
| 58 |
"step": 30
|
| 59 |
},
|
| 60 |
{
|
| 61 |
+
"epoch": 0.10324483775811209,
|
| 62 |
+
"grad_norm": 1.1769920587539673,
|
| 63 |
+
"learning_rate": 3.431372549019608e-05,
|
| 64 |
+
"loss": 0.0815,
|
| 65 |
+
"num_tokens": 573399.0,
|
| 66 |
"step": 35
|
| 67 |
},
|
| 68 |
{
|
| 69 |
+
"epoch": 0.11799410029498525,
|
| 70 |
+
"grad_norm": 1.2460798025131226,
|
| 71 |
+
"learning_rate": 3.9215686274509805e-05,
|
| 72 |
+
"loss": 0.0752,
|
| 73 |
+
"num_tokens": 655143.0,
|
| 74 |
"step": 40
|
| 75 |
},
|
| 76 |
{
|
| 77 |
+
"epoch": 0.13274336283185842,
|
| 78 |
+
"grad_norm": 0.7001373171806335,
|
| 79 |
+
"learning_rate": 4.411764705882353e-05,
|
| 80 |
+
"loss": 0.0739,
|
| 81 |
+
"num_tokens": 737063.0,
|
| 82 |
"step": 45
|
| 83 |
},
|
| 84 |
{
|
| 85 |
+
"epoch": 0.14749262536873156,
|
| 86 |
+
"grad_norm": 0.7255822420120239,
|
| 87 |
+
"learning_rate": 4.901960784313725e-05,
|
| 88 |
+
"loss": 0.0708,
|
| 89 |
+
"num_tokens": 818902.0,
|
| 90 |
"step": 50
|
| 91 |
},
|
| 92 |
{
|
| 93 |
+
"epoch": 0.16224188790560473,
|
| 94 |
+
"grad_norm": 1.1189676523208618,
|
| 95 |
+
"learning_rate": 4.999809624142209e-05,
|
| 96 |
+
"loss": 0.0902,
|
| 97 |
+
"num_tokens": 900822.0,
|
| 98 |
"step": 55
|
| 99 |
},
|
| 100 |
{
|
| 101 |
+
"epoch": 0.17699115044247787,
|
| 102 |
+
"grad_norm": 20.670774459838867,
|
| 103 |
+
"learning_rate": 4.9990362774334994e-05,
|
| 104 |
+
"loss": 0.1074,
|
| 105 |
+
"num_tokens": 982742.0,
|
| 106 |
"step": 60
|
| 107 |
},
|
| 108 |
{
|
| 109 |
+
"epoch": 0.19174041297935104,
|
| 110 |
+
"grad_norm": 0.5863003730773926,
|
| 111 |
+
"learning_rate": 4.997668265705137e-05,
|
| 112 |
+
"loss": 0.0683,
|
| 113 |
+
"num_tokens": 1064662.0,
|
| 114 |
"step": 65
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.20648967551622419,
|
| 118 |
+
"grad_norm": 0.6090629696846008,
|
| 119 |
+
"learning_rate": 4.9957059506714846e-05,
|
| 120 |
+
"loss": 0.0693,
|
| 121 |
+
"num_tokens": 1146582.0,
|
| 122 |
"step": 70
|
| 123 |
},
|
| 124 |
{
|
| 125 |
+
"epoch": 0.22123893805309736,
|
| 126 |
+
"grad_norm": 0.4677298069000244,
|
| 127 |
+
"learning_rate": 4.9931498511859377e-05,
|
| 128 |
+
"loss": 0.0673,
|
| 129 |
+
"num_tokens": 1228502.0,
|
| 130 |
"step": 75
|
| 131 |
},
|
| 132 |
{
|
| 133 |
+
"epoch": 0.2359882005899705,
|
| 134 |
+
"grad_norm": 0.5659052729606628,
|
| 135 |
+
"learning_rate": 4.990000643103734e-05,
|
| 136 |
+
"loss": 0.0685,
|
| 137 |
+
"num_tokens": 1310422.0,
|
| 138 |
"step": 80
|
| 139 |
},
|
| 140 |
{
|
| 141 |
+
"epoch": 0.25073746312684364,
|
| 142 |
+
"grad_norm": 0.39706626534461975,
|
| 143 |
+
"learning_rate": 4.986259159103256e-05,
|
| 144 |
+
"loss": 0.0669,
|
| 145 |
+
"num_tokens": 1392342.0,
|
| 146 |
"step": 85
|
| 147 |
},
|
| 148 |
{
|
| 149 |
+
"epoch": 0.26548672566371684,
|
| 150 |
+
"grad_norm": 0.38469579815864563,
|
| 151 |
+
"learning_rate": 4.981926388465857e-05,
|
| 152 |
+
"loss": 0.0653,
|
| 153 |
+
"num_tokens": 1474262.0,
|
| 154 |
"step": 90
|
| 155 |
},
|
| 156 |
{
|
| 157 |
+
"epoch": 0.28023598820059,
|
| 158 |
+
"grad_norm": 0.3758561313152313,
|
| 159 |
+
"learning_rate": 4.9770034768142934e-05,
|
| 160 |
+
"loss": 0.0635,
|
| 161 |
+
"num_tokens": 1555918.0,
|
| 162 |
"step": 95
|
| 163 |
},
|
| 164 |
{
|
| 165 |
+
"epoch": 0.2949852507374631,
|
| 166 |
+
"grad_norm": 0.439568430185318,
|
| 167 |
+
"learning_rate": 4.971491725809807e-05,
|
| 168 |
+
"loss": 0.0638,
|
| 169 |
+
"num_tokens": 1637838.0,
|
| 170 |
"step": 100
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 0.30973451327433627,
|
| 174 |
+
"grad_norm": 0.3522689938545227,
|
| 175 |
+
"learning_rate": 4.965392592807956e-05,
|
| 176 |
+
"loss": 0.0631,
|
| 177 |
+
"num_tokens": 1719758.0,
|
| 178 |
"step": 105
|
| 179 |
},
|
| 180 |
{
|
| 181 |
+
"epoch": 0.32448377581120946,
|
| 182 |
+
"grad_norm": 0.3989611864089966,
|
| 183 |
+
"learning_rate": 4.9587076904732756e-05,
|
| 184 |
+
"loss": 0.0605,
|
| 185 |
+
"num_tokens": 1801678.0,
|
| 186 |
"step": 110
|
| 187 |
},
|
| 188 |
{
|
| 189 |
+
"epoch": 0.3392330383480826,
|
| 190 |
+
"grad_norm": 0.34883007407188416,
|
| 191 |
+
"learning_rate": 4.951438786352881e-05,
|
| 192 |
+
"loss": 0.0619,
|
| 193 |
+
"num_tokens": 1883598.0,
|
| 194 |
"step": 115
|
| 195 |
},
|
| 196 |
{
|
| 197 |
+
"epoch": 0.35398230088495575,
|
| 198 |
+
"grad_norm": 0.29905784130096436,
|
| 199 |
+
"learning_rate": 4.943587802409103e-05,
|
| 200 |
+
"loss": 0.063,
|
| 201 |
+
"num_tokens": 1965518.0,
|
| 202 |
"step": 120
|
| 203 |
},
|
| 204 |
{
|
| 205 |
+
"epoch": 0.3687315634218289,
|
| 206 |
+
"grad_norm": 0.3990210294723511,
|
| 207 |
+
"learning_rate": 4.935156814511314e-05,
|
| 208 |
+
"loss": 0.0617,
|
| 209 |
+
"num_tokens": 2047438.0,
|
| 210 |
"step": 125
|
| 211 |
},
|
| 212 |
{
|
| 213 |
+
"epoch": 0.3834808259587021,
|
| 214 |
+
"grad_norm": 0.23905648291110992,
|
| 215 |
+
"learning_rate": 4.926148051887042e-05,
|
| 216 |
+
"loss": 0.0595,
|
| 217 |
+
"num_tokens": 2129358.0,
|
| 218 |
"step": 130
|
| 219 |
},
|
| 220 |
{
|
| 221 |
+
"epoch": 0.39823008849557523,
|
| 222 |
+
"grad_norm": 0.2639841139316559,
|
| 223 |
+
"learning_rate": 4.916563896532549e-05,
|
| 224 |
+
"loss": 0.059,
|
| 225 |
+
"num_tokens": 2211230.0,
|
| 226 |
"step": 135
|
| 227 |
},
|
| 228 |
{
|
| 229 |
+
"epoch": 0.41297935103244837,
|
| 230 |
+
"grad_norm": 0.29038557410240173,
|
| 231 |
+
"learning_rate": 4.906406882583004e-05,
|
| 232 |
+
"loss": 0.0568,
|
| 233 |
+
"num_tokens": 2293150.0,
|
| 234 |
"step": 140
|
| 235 |
},
|
| 236 |
{
|
| 237 |
+
"epoch": 0.4277286135693215,
|
| 238 |
+
"grad_norm": 0.30302760004997253,
|
| 239 |
+
"learning_rate": 4.895679695642444e-05,
|
| 240 |
+
"loss": 0.0591,
|
| 241 |
+
"num_tokens": 2374790.0,
|
| 242 |
"step": 145
|
| 243 |
},
|
| 244 |
{
|
| 245 |
+
"epoch": 0.4424778761061947,
|
| 246 |
+
"grad_norm": 0.2980372905731201,
|
| 247 |
+
"learning_rate": 4.884385172073666e-05,
|
| 248 |
+
"loss": 0.0608,
|
| 249 |
+
"num_tokens": 2456682.0,
|
| 250 |
"step": 150
|
| 251 |
},
|
| 252 |
{
|
| 253 |
+
"epoch": 0.45722713864306785,
|
| 254 |
+
"grad_norm": 0.3176118731498718,
|
| 255 |
+
"learning_rate": 4.8725262982482794e-05,
|
| 256 |
+
"loss": 0.058,
|
| 257 |
+
"num_tokens": 2538602.0,
|
| 258 |
"step": 155
|
| 259 |
},
|
| 260 |
{
|
| 261 |
+
"epoch": 0.471976401179941,
|
| 262 |
+
"grad_norm": 0.2424178570508957,
|
| 263 |
+
"learning_rate": 4.860106209757071e-05,
|
| 264 |
+
"loss": 0.0578,
|
| 265 |
+
"num_tokens": 2620522.0,
|
| 266 |
"step": 160
|
| 267 |
},
|
| 268 |
{
|
| 269 |
+
"epoch": 0.48672566371681414,
|
| 270 |
+
"grad_norm": 0.4045540988445282,
|
| 271 |
+
"learning_rate": 4.847128190580936e-05,
|
| 272 |
+
"loss": 0.0577,
|
| 273 |
+
"num_tokens": 2702326.0,
|
| 274 |
"step": 165
|
| 275 |
},
|
| 276 |
{
|
| 277 |
+
"epoch": 0.5014749262536873,
|
| 278 |
+
"grad_norm": 0.248455211520195,
|
| 279 |
+
"learning_rate": 4.8335956722225616e-05,
|
| 280 |
+
"loss": 0.0566,
|
| 281 |
+
"num_tokens": 2784246.0,
|
| 282 |
"step": 170
|
| 283 |
},
|
| 284 |
{
|
| 285 |
+
"epoch": 0.5162241887905604,
|
| 286 |
+
"grad_norm": 0.2201586663722992,
|
| 287 |
+
"learning_rate": 4.819512232799107e-05,
|
| 288 |
+
"loss": 0.0575,
|
| 289 |
+
"num_tokens": 2866166.0,
|
| 290 |
+
"step": 175
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.5309734513274337,
|
| 294 |
+
"grad_norm": 0.23909515142440796,
|
| 295 |
+
"learning_rate": 4.804881596096118e-05,
|
| 296 |
+
"loss": 0.0553,
|
| 297 |
+
"num_tokens": 2948086.0,
|
| 298 |
+
"step": 180
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.5457227138643068,
|
| 302 |
+
"grad_norm": 0.23102901875972748,
|
| 303 |
+
"learning_rate": 4.789707630582923e-05,
|
| 304 |
+
"loss": 0.0591,
|
| 305 |
+
"num_tokens": 3029739.0,
|
| 306 |
+
"step": 185
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.56047197640118,
|
| 310 |
+
"grad_norm": 0.2302148938179016,
|
| 311 |
+
"learning_rate": 4.773994348389782e-05,
|
| 312 |
+
"loss": 0.0587,
|
| 313 |
+
"num_tokens": 3111659.0,
|
| 314 |
+
"step": 190
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.5752212389380531,
|
| 318 |
+
"grad_norm": 0.20451895892620087,
|
| 319 |
+
"learning_rate": 4.757745904247038e-05,
|
| 320 |
+
"loss": 0.0601,
|
| 321 |
+
"num_tokens": 3193453.0,
|
| 322 |
+
"step": 195
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 0.5899705014749262,
|
| 326 |
+
"grad_norm": 0.19253046810626984,
|
| 327 |
+
"learning_rate": 4.7409665943865705e-05,
|
| 328 |
+
"loss": 0.0579,
|
| 329 |
+
"num_tokens": 3275373.0,
|
| 330 |
+
"step": 200
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.6047197640117994,
|
| 334 |
+
"grad_norm": 0.2857654094696045,
|
| 335 |
+
"learning_rate": 4.7236608554058375e-05,
|
| 336 |
+
"loss": 0.0577,
|
| 337 |
+
"num_tokens": 3357228.0,
|
| 338 |
+
"step": 205
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.6194690265486725,
|
| 342 |
+
"grad_norm": 0.21651217341423035,
|
| 343 |
+
"learning_rate": 4.7058332630947935e-05,
|
| 344 |
+
"loss": 0.0602,
|
| 345 |
+
"num_tokens": 3439124.0,
|
| 346 |
+
"step": 210
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.6342182890855457,
|
| 350 |
+
"grad_norm": 0.22674554586410522,
|
| 351 |
+
"learning_rate": 4.6874885312260186e-05,
|
| 352 |
+
"loss": 0.0588,
|
| 353 |
+
"num_tokens": 3520967.0,
|
| 354 |
+
"step": 215
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.6489675516224189,
|
| 358 |
+
"grad_norm": 0.22327347099781036,
|
| 359 |
+
"learning_rate": 4.668631510308349e-05,
|
| 360 |
+
"loss": 0.0585,
|
| 361 |
+
"num_tokens": 3602887.0,
|
| 362 |
+
"step": 220
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.6637168141592921,
|
| 366 |
+
"grad_norm": 0.19307062029838562,
|
| 367 |
+
"learning_rate": 4.649267186304362e-05,
|
| 368 |
+
"loss": 0.0574,
|
| 369 |
+
"num_tokens": 3684556.0,
|
| 370 |
+
"step": 225
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.6784660766961652,
|
| 374 |
+
"grad_norm": 0.2287718951702118,
|
| 375 |
+
"learning_rate": 4.6294006793120436e-05,
|
| 376 |
+
"loss": 0.0576,
|
| 377 |
+
"num_tokens": 3766476.0,
|
| 378 |
+
"step": 230
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.6932153392330384,
|
| 382 |
+
"grad_norm": 0.16587583720684052,
|
| 383 |
+
"learning_rate": 4.609037242210989e-05,
|
| 384 |
+
"loss": 0.056,
|
| 385 |
+
"num_tokens": 3848396.0,
|
| 386 |
+
"step": 235
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.7079646017699115,
|
| 390 |
+
"grad_norm": 0.2796880900859833,
|
| 391 |
+
"learning_rate": 4.5881822592734946e-05,
|
| 392 |
+
"loss": 0.0572,
|
| 393 |
+
"num_tokens": 3930227.0,
|
| 394 |
+
"step": 240
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.7227138643067846,
|
| 398 |
+
"grad_norm": 0.2181931585073471,
|
| 399 |
+
"learning_rate": 4.5668412447409116e-05,
|
| 400 |
+
"loss": 0.057,
|
| 401 |
+
"num_tokens": 4011953.0,
|
| 402 |
+
"step": 245
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.7374631268436578,
|
| 406 |
+
"grad_norm": 0.31474748253822327,
|
| 407 |
+
"learning_rate": 4.545019841365628e-05,
|
| 408 |
+
"loss": 0.0555,
|
| 409 |
+
"num_tokens": 4093782.0,
|
| 410 |
+
"step": 250
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.7522123893805309,
|
| 414 |
+
"grad_norm": 0.19964337348937988,
|
| 415 |
+
"learning_rate": 4.5227238189190775e-05,
|
| 416 |
+
"loss": 0.0559,
|
| 417 |
+
"num_tokens": 4175702.0,
|
| 418 |
+
"step": 255
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.7669616519174042,
|
| 422 |
+
"grad_norm": 0.17513608932495117,
|
| 423 |
+
"learning_rate": 4.4999590726661605e-05,
|
| 424 |
+
"loss": 0.0555,
|
| 425 |
+
"num_tokens": 4257622.0,
|
| 426 |
+
"step": 260
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.7817109144542773,
|
| 430 |
+
"grad_norm": 0.18871212005615234,
|
| 431 |
+
"learning_rate": 4.476731621806485e-05,
|
| 432 |
+
"loss": 0.0572,
|
| 433 |
+
"num_tokens": 4339542.0,
|
| 434 |
+
"step": 265
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.7964601769911505,
|
| 438 |
+
"grad_norm": 0.18672113120555878,
|
| 439 |
+
"learning_rate": 4.453047607882834e-05,
|
| 440 |
+
"loss": 0.0579,
|
| 441 |
+
"num_tokens": 4421462.0,
|
| 442 |
+
"step": 270
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.8112094395280236,
|
| 446 |
+
"grad_norm": 0.20873971283435822,
|
| 447 |
+
"learning_rate": 4.428913293157293e-05,
|
| 448 |
+
"loss": 0.0552,
|
| 449 |
+
"num_tokens": 4503098.0,
|
| 450 |
+
"step": 275
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.8259587020648967,
|
| 454 |
+
"grad_norm": 0.17426809668540955,
|
| 455 |
+
"learning_rate": 4.404335058955446e-05,
|
| 456 |
+
"loss": 0.0554,
|
| 457 |
+
"num_tokens": 4585018.0,
|
| 458 |
+
"step": 280
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.8407079646017699,
|
| 462 |
+
"grad_norm": 0.18556445837020874,
|
| 463 |
+
"learning_rate": 4.379319403979104e-05,
|
| 464 |
+
"loss": 0.0565,
|
| 465 |
+
"num_tokens": 4666938.0,
|
| 466 |
+
"step": 285
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 0.855457227138643,
|
| 470 |
+
"grad_norm": 0.1989375203847885,
|
| 471 |
+
"learning_rate": 4.353872942587985e-05,
|
| 472 |
+
"loss": 0.0559,
|
| 473 |
+
"num_tokens": 4748858.0,
|
| 474 |
+
"step": 290
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.8702064896755162,
|
| 478 |
+
"grad_norm": 0.2044542133808136,
|
| 479 |
+
"learning_rate": 4.32800240305082e-05,
|
| 480 |
+
"loss": 0.0548,
|
| 481 |
+
"num_tokens": 4830778.0,
|
| 482 |
+
"step": 295
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.8849557522123894,
|
| 486 |
+
"grad_norm": 0.17457793653011322,
|
| 487 |
+
"learning_rate": 4.301714625766342e-05,
|
| 488 |
+
"loss": 0.0563,
|
| 489 |
+
"num_tokens": 4912698.0,
|
| 490 |
+
"step": 300
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.8997050147492626,
|
| 494 |
+
"grad_norm": 0.1791624128818512,
|
| 495 |
+
"learning_rate": 4.275016561454622e-05,
|
| 496 |
+
"loss": 0.0566,
|
| 497 |
+
"num_tokens": 4994553.0,
|
| 498 |
+
"step": 305
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 0.9144542772861357,
|
| 502 |
+
"grad_norm": 0.16299232840538025,
|
| 503 |
+
"learning_rate": 4.247915269319241e-05,
|
| 504 |
+
"loss": 0.0553,
|
| 505 |
+
"num_tokens": 5076473.0,
|
| 506 |
+
"step": 310
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.9292035398230089,
|
| 510 |
+
"grad_norm": 0.1811797171831131,
|
| 511 |
+
"learning_rate": 4.2204179151807685e-05,
|
| 512 |
+
"loss": 0.0554,
|
| 513 |
+
"num_tokens": 5158254.0,
|
| 514 |
+
"step": 315
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.943952802359882,
|
| 518 |
+
"grad_norm": 0.1605992168188095,
|
| 519 |
+
"learning_rate": 4.19253176958206e-05,
|
| 520 |
+
"loss": 0.0558,
|
| 521 |
+
"num_tokens": 5240174.0,
|
| 522 |
+
"step": 320
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.9587020648967551,
|
| 526 |
+
"grad_norm": 0.18501946330070496,
|
| 527 |
+
"learning_rate": 4.1642642058658605e-05,
|
| 528 |
+
"loss": 0.0554,
|
| 529 |
+
"num_tokens": 5322094.0,
|
| 530 |
+
"step": 325
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.9734513274336283,
|
| 534 |
+
"grad_norm": 0.18083244562149048,
|
| 535 |
+
"learning_rate": 4.135622698225229e-05,
|
| 536 |
+
"loss": 0.0562,
|
| 537 |
+
"num_tokens": 5403956.0,
|
| 538 |
+
"step": 330
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.9882005899705014,
|
| 542 |
+
"grad_norm": 0.21333028376102448,
|
| 543 |
+
"learning_rate": 4.1066148197272944e-05,
|
| 544 |
+
"loss": 0.0566,
|
| 545 |
+
"num_tokens": 5485712.0,
|
| 546 |
+
"step": 335
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 1.0029498525073746,
|
| 550 |
+
"grad_norm": 0.19363561272621155,
|
| 551 |
+
"learning_rate": 4.077248240310868e-05,
|
| 552 |
+
"loss": 0.0553,
|
| 553 |
+
"num_tokens": 5567632.0,
|
| 554 |
+
"step": 340
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 1.0176991150442478,
|
| 558 |
+
"grad_norm": 0.1972503960132599,
|
| 559 |
+
"learning_rate": 4.047530724758451e-05,
|
| 560 |
+
"loss": 0.0551,
|
| 561 |
+
"num_tokens": 5649552.0,
|
| 562 |
+
"step": 345
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 1.0324483775811208,
|
| 566 |
+
"grad_norm": 0.16727674007415771,
|
| 567 |
+
"learning_rate": 4.017470130643149e-05,
|
| 568 |
+
"loss": 0.0566,
|
| 569 |
+
"num_tokens": 5731214.0,
|
| 570 |
+
"step": 350
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 1.047197640117994,
|
| 574 |
+
"grad_norm": 0.14662833511829376,
|
| 575 |
+
"learning_rate": 3.987074406251067e-05,
|
| 576 |
+
"loss": 0.0567,
|
| 577 |
+
"num_tokens": 5813045.0,
|
| 578 |
+
"step": 355
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.0619469026548674,
|
| 582 |
+
"grad_norm": 0.16213928163051605,
|
| 583 |
+
"learning_rate": 3.9563515884797074e-05,
|
| 584 |
+
"loss": 0.057,
|
| 585 |
+
"num_tokens": 5894884.0,
|
| 586 |
+
"step": 360
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.0766961651917404,
|
| 590 |
+
"grad_norm": 0.13559935986995697,
|
| 591 |
+
"learning_rate": 3.925309800712945e-05,
|
| 592 |
+
"loss": 0.0545,
|
| 593 |
+
"num_tokens": 5976776.0,
|
| 594 |
+
"step": 365
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.0914454277286136,
|
| 598 |
+
"grad_norm": 0.16717351973056793,
|
| 599 |
+
"learning_rate": 3.8939572506731375e-05,
|
| 600 |
+
"loss": 0.0558,
|
| 601 |
+
"num_tokens": 6058573.0,
|
| 602 |
+
"step": 370
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 1.1061946902654867,
|
| 606 |
+
"grad_norm": 0.11658819019794464,
|
| 607 |
+
"learning_rate": 3.8623022282509245e-05,
|
| 608 |
+
"loss": 0.0555,
|
| 609 |
+
"num_tokens": 6140493.0,
|
| 610 |
+
"step": 375
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 1.12094395280236,
|
| 614 |
+
"grad_norm": 0.14316953718662262,
|
| 615 |
+
"learning_rate": 3.83035310331331e-05,
|
| 616 |
+
"loss": 0.0553,
|
| 617 |
+
"num_tokens": 6222413.0,
|
| 618 |
+
"step": 380
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 1.135693215339233,
|
| 622 |
+
"grad_norm": 0.15932655334472656,
|
| 623 |
+
"learning_rate": 3.798118323490597e-05,
|
| 624 |
+
"loss": 0.0542,
|
| 625 |
+
"num_tokens": 6304333.0,
|
| 626 |
+
"step": 385
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.1504424778761062,
|
| 630 |
+
"grad_norm": 0.11980193108320236,
|
| 631 |
+
"learning_rate": 3.765606411942759e-05,
|
| 632 |
+
"loss": 0.0535,
|
| 633 |
+
"num_tokens": 6386253.0,
|
| 634 |
+
"step": 390
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 1.1651917404129795,
|
| 638 |
+
"grad_norm": 0.1464000791311264,
|
| 639 |
+
"learning_rate": 3.7328259651058384e-05,
|
| 640 |
+
"loss": 0.0547,
|
| 641 |
+
"num_tokens": 6468173.0,
|
| 642 |
+
"step": 395
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 1.1799410029498525,
|
| 646 |
+
"grad_norm": 0.12491417676210403,
|
| 647 |
+
"learning_rate": 3.699785650418977e-05,
|
| 648 |
+
"loss": 0.0536,
|
| 649 |
+
"num_tokens": 6550093.0,
|
| 650 |
+
"step": 400
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.1946902654867257,
|
| 654 |
+
"grad_norm": 0.15376126766204834,
|
| 655 |
+
"learning_rate": 3.666494204032668e-05,
|
| 656 |
+
"loss": 0.0555,
|
| 657 |
+
"num_tokens": 6631948.0,
|
| 658 |
+
"step": 405
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.2094395280235988,
|
| 662 |
+
"grad_norm": 0.16010604798793793,
|
| 663 |
+
"learning_rate": 3.63296042849884e-05,
|
| 664 |
+
"loss": 0.0544,
|
| 665 |
+
"num_tokens": 6713791.0,
|
| 666 |
+
"step": 410
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 1.224188790560472,
|
| 670 |
+
"grad_norm": 0.15632019937038422,
|
| 671 |
+
"learning_rate": 3.5991931904433824e-05,
|
| 672 |
+
"loss": 0.0557,
|
| 673 |
+
"num_tokens": 6795632.0,
|
| 674 |
+
"step": 415
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 1.238938053097345,
|
| 678 |
+
"grad_norm": 0.14752759039402008,
|
| 679 |
+
"learning_rate": 3.5652014182217355e-05,
|
| 680 |
+
"loss": 0.0543,
|
| 681 |
+
"num_tokens": 6877552.0,
|
| 682 |
+
"step": 420
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 1.2536873156342183,
|
| 686 |
+
"grad_norm": 0.21005742251873016,
|
| 687 |
+
"learning_rate": 3.530994099558153e-05,
|
| 688 |
+
"loss": 0.0533,
|
| 689 |
+
"num_tokens": 6959472.0,
|
| 690 |
+
"step": 425
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 1.2684365781710913,
|
| 694 |
+
"grad_norm": 0.13375329971313477,
|
| 695 |
+
"learning_rate": 3.496580279169265e-05,
|
| 696 |
+
"loss": 0.0553,
|
| 697 |
+
"num_tokens": 7041392.0,
|
| 698 |
+
"step": 430
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 1.2831858407079646,
|
| 702 |
+
"grad_norm": 0.16405825316905975,
|
| 703 |
+
"learning_rate": 3.461969056372582e-05,
|
| 704 |
+
"loss": 0.0545,
|
| 705 |
+
"num_tokens": 7123312.0,
|
| 706 |
+
"step": 435
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.2979351032448379,
|
| 710 |
+
"grad_norm": 0.1644970029592514,
|
| 711 |
+
"learning_rate": 3.427169582680551e-05,
|
| 712 |
+
"loss": 0.0551,
|
| 713 |
+
"num_tokens": 7205232.0,
|
| 714 |
+
"step": 440
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.3126843657817109,
|
| 718 |
+
"grad_norm": 0.1362873613834381,
|
| 719 |
+
"learning_rate": 3.392191059380812e-05,
|
| 720 |
+
"loss": 0.0544,
|
| 721 |
+
"num_tokens": 7287046.0,
|
| 722 |
+
"step": 445
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 1.3274336283185841,
|
| 726 |
+
"grad_norm": 0.148273766040802,
|
| 727 |
+
"learning_rate": 3.3570427351033046e-05,
|
| 728 |
+
"loss": 0.0546,
|
| 729 |
+
"num_tokens": 7368966.0,
|
| 730 |
+
"step": 450
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.3421828908554572,
|
| 734 |
+
"grad_norm": 0.14546248316764832,
|
| 735 |
+
"learning_rate": 3.321733903374841e-05,
|
| 736 |
+
"loss": 0.0539,
|
| 737 |
+
"num_tokens": 7450886.0,
|
| 738 |
+
"step": 455
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 1.3569321533923304,
|
| 742 |
+
"grad_norm": 0.14057567715644836,
|
| 743 |
+
"learning_rate": 3.286273900161818e-05,
|
| 744 |
+
"loss": 0.0553,
|
| 745 |
+
"num_tokens": 7532806.0,
|
| 746 |
+
"step": 460
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 1.3716814159292037,
|
| 750 |
+
"grad_norm": 0.1333959847688675,
|
| 751 |
+
"learning_rate": 3.250672101401707e-05,
|
| 752 |
+
"loss": 0.0546,
|
| 753 |
+
"num_tokens": 7614726.0,
|
| 754 |
+
"step": 465
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 1.3864306784660767,
|
| 758 |
+
"grad_norm": 0.1370707005262375,
|
| 759 |
+
"learning_rate": 3.214937920523974e-05,
|
| 760 |
+
"loss": 0.0561,
|
| 761 |
+
"num_tokens": 7696402.0,
|
| 762 |
+
"step": 470
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.4011799410029497,
|
| 766 |
+
"grad_norm": 0.1322726011276245,
|
| 767 |
+
"learning_rate": 3.1790808059610786e-05,
|
| 768 |
+
"loss": 0.0536,
|
| 769 |
+
"num_tokens": 7778322.0,
|
| 770 |
+
"step": 475
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 1.415929203539823,
|
| 774 |
+
"grad_norm": 0.14645648002624512,
|
| 775 |
+
"learning_rate": 3.143110238650236e-05,
|
| 776 |
+
"loss": 0.0543,
|
| 777 |
+
"num_tokens": 7860242.0,
|
| 778 |
+
"step": 480
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 1.4306784660766962,
|
| 782 |
+
"grad_norm": 0.1392751932144165,
|
| 783 |
+
"learning_rate": 3.107035729526566e-05,
|
| 784 |
+
"loss": 0.0555,
|
| 785 |
+
"num_tokens": 7942073.0,
|
| 786 |
+
"step": 485
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 1.4454277286135693,
|
| 790 |
+
"grad_norm": 0.15789316594600677,
|
| 791 |
+
"learning_rate": 3.070866817008319e-05,
|
| 792 |
+
"loss": 0.0565,
|
| 793 |
+
"num_tokens": 8023993.0,
|
| 794 |
+
"step": 490
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.4601769911504425,
|
| 798 |
+
"grad_norm": 0.11416062712669373,
|
| 799 |
+
"learning_rate": 3.0346130644748367e-05,
|
| 800 |
+
"loss": 0.0557,
|
| 801 |
+
"num_tokens": 8105787.0,
|
| 802 |
+
"step": 495
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.4749262536873156,
|
| 806 |
+
"grad_norm": 0.14937061071395874,
|
| 807 |
+
"learning_rate": 2.998284057737909e-05,
|
| 808 |
+
"loss": 0.0532,
|
| 809 |
+
"num_tokens": 8187707.0,
|
| 810 |
+
"step": 500
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 1.4896755162241888,
|
| 814 |
+
"grad_norm": 0.11278703063726425,
|
| 815 |
+
"learning_rate": 2.9618894025071984e-05,
|
| 816 |
+
"loss": 0.0534,
|
| 817 |
+
"num_tokens": 8269536.0,
|
| 818 |
+
"step": 505
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.504424778761062,
|
| 822 |
+
"grad_norm": 0.1776408553123474,
|
| 823 |
+
"learning_rate": 2.925438721850412e-05,
|
| 824 |
+
"loss": 0.0545,
|
| 825 |
+
"num_tokens": 8351456.0,
|
| 826 |
+
"step": 510
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 1.519174041297935,
|
| 830 |
+
"grad_norm": 0.1571996957063675,
|
| 831 |
+
"learning_rate": 2.88894165364887e-05,
|
| 832 |
+
"loss": 0.054,
|
| 833 |
+
"num_tokens": 8433240.0,
|
| 834 |
+
"step": 515
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 1.5339233038348081,
|
| 838 |
+
"grad_norm": 0.15761998295783997,
|
| 839 |
+
"learning_rate": 2.8524078480491684e-05,
|
| 840 |
+
"loss": 0.0534,
|
| 841 |
+
"num_tokens": 8515160.0,
|
| 842 |
+
"step": 520
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 1.5486725663716814,
|
| 846 |
+
"grad_norm": 0.12947538495063782,
|
| 847 |
+
"learning_rate": 2.8158469649115978e-05,
|
| 848 |
+
"loss": 0.0539,
|
| 849 |
+
"num_tokens": 8597080.0,
|
| 850 |
+
"step": 525
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 1.5634218289085546,
|
| 854 |
+
"grad_norm": 0.15504850447177887,
|
| 855 |
+
"learning_rate": 2.779268671255985e-05,
|
| 856 |
+
"loss": 0.0539,
|
| 857 |
+
"num_tokens": 8678806.0,
|
| 858 |
+
"step": 530
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 1.5781710914454279,
|
| 862 |
+
"grad_norm": 0.11758331209421158,
|
| 863 |
+
"learning_rate": 2.7426826387056555e-05,
|
| 864 |
+
"loss": 0.055,
|
| 865 |
+
"num_tokens": 8760726.0,
|
| 866 |
+
"step": 535
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.592920353982301,
|
| 870 |
+
"grad_norm": 0.14461639523506165,
|
| 871 |
+
"learning_rate": 2.7060985409301627e-05,
|
| 872 |
+
"loss": 0.0557,
|
| 873 |
+
"num_tokens": 8842646.0,
|
| 874 |
+
"step": 540
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.607669616519174,
|
| 878 |
+
"grad_norm": 0.15372057259082794,
|
| 879 |
+
"learning_rate": 2.6695260510874914e-05,
|
| 880 |
+
"loss": 0.054,
|
| 881 |
+
"num_tokens": 8924450.0,
|
| 882 |
+
"step": 545
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 1.6224188790560472,
|
| 886 |
+
"grad_norm": 0.14337101578712463,
|
| 887 |
+
"learning_rate": 2.632974839266385e-05,
|
| 888 |
+
"loss": 0.0531,
|
| 889 |
+
"num_tokens": 9006370.0,
|
| 890 |
+
"step": 550
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.6371681415929205,
|
| 894 |
+
"grad_norm": 0.12814313173294067,
|
| 895 |
+
"learning_rate": 2.5964545699294906e-05,
|
| 896 |
+
"loss": 0.0549,
|
| 897 |
+
"num_tokens": 9088290.0,
|
| 898 |
+
"step": 555
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 1.6519174041297935,
|
| 902 |
+
"grad_norm": 0.1454104632139206,
|
| 903 |
+
"learning_rate": 2.559974899357991e-05,
|
| 904 |
+
"loss": 0.0541,
|
| 905 |
+
"num_tokens": 9170162.0,
|
| 906 |
+
"step": 560
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 1.6666666666666665,
|
| 910 |
+
"grad_norm": 0.11160258203744888,
|
| 911 |
+
"learning_rate": 2.5235454730983955e-05,
|
| 912 |
+
"loss": 0.0535,
|
| 913 |
+
"num_tokens": 9252082.0,
|
| 914 |
+
"step": 565
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 1.6814159292035398,
|
| 918 |
+
"grad_norm": 0.14262907207012177,
|
| 919 |
+
"learning_rate": 2.487175923412175e-05,
|
| 920 |
+
"loss": 0.0538,
|
| 921 |
+
"num_tokens": 9334002.0,
|
| 922 |
+
"step": 570
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 1.696165191740413,
|
| 926 |
+
"grad_norm": 0.12927588820457458,
|
| 927 |
+
"learning_rate": 2.4508758667289076e-05,
|
| 928 |
+
"loss": 0.054,
|
| 929 |
+
"num_tokens": 9415922.0,
|
| 930 |
+
"step": 575
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 1.7109144542772863,
|
| 934 |
+
"grad_norm": 0.1411578357219696,
|
| 935 |
+
"learning_rate": 2.4146549011036074e-05,
|
| 936 |
+
"loss": 0.0519,
|
| 937 |
+
"num_tokens": 9497842.0,
|
| 938 |
+
"step": 580
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 1.7256637168141593,
|
| 942 |
+
"grad_norm": 0.14883753657341003,
|
| 943 |
+
"learning_rate": 2.378522603678917e-05,
|
| 944 |
+
"loss": 0.0524,
|
| 945 |
+
"num_tokens": 9579762.0,
|
| 946 |
+
"step": 585
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 1.7404129793510323,
|
| 950 |
+
"grad_norm": 0.13972261548042297,
|
| 951 |
+
"learning_rate": 2.3424885281528248e-05,
|
| 952 |
+
"loss": 0.0546,
|
| 953 |
+
"num_tokens": 9661480.0,
|
| 954 |
+
"step": 590
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.7551622418879056,
|
| 958 |
+
"grad_norm": 0.15459661185741425,
|
| 959 |
+
"learning_rate": 2.3065622022525813e-05,
|
| 960 |
+
"loss": 0.0528,
|
| 961 |
+
"num_tokens": 9743400.0,
|
| 962 |
+
"step": 595
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 1.7699115044247788,
|
| 966 |
+
"grad_norm": 0.12449204921722412,
|
| 967 |
+
"learning_rate": 2.2707531252154868e-05,
|
| 968 |
+
"loss": 0.0539,
|
| 969 |
+
"num_tokens": 9825283.0,
|
| 970 |
+
"step": 600
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 1.7846607669616519,
|
| 974 |
+
"grad_norm": 0.14224207401275635,
|
| 975 |
+
"learning_rate": 2.2350707652772102e-05,
|
| 976 |
+
"loss": 0.0552,
|
| 977 |
+
"num_tokens": 9906978.0,
|
| 978 |
+
"step": 605
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 1.799410029498525,
|
| 982 |
+
"grad_norm": 0.11525611579418182,
|
| 983 |
+
"learning_rate": 2.1995245571683016e-05,
|
| 984 |
+
"loss": 0.0555,
|
| 985 |
+
"num_tokens": 9988898.0,
|
| 986 |
+
"step": 610
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 1.8141592920353982,
|
| 990 |
+
"grad_norm": 0.12278515100479126,
|
| 991 |
+
"learning_rate": 2.1641238996195645e-05,
|
| 992 |
+
"loss": 0.0532,
|
| 993 |
+
"num_tokens": 10070818.0,
|
| 994 |
+
"step": 615
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 1.8289085545722714,
|
| 998 |
+
"grad_norm": 0.14640717208385468,
|
| 999 |
+
"learning_rate": 2.1288781528769553e-05,
|
| 1000 |
+
"loss": 0.0545,
|
| 1001 |
+
"num_tokens": 10152738.0,
|
| 1002 |
+
"step": 620
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 1.8436578171091447,
|
| 1006 |
+
"grad_norm": 0.11209894716739655,
|
| 1007 |
+
"learning_rate": 2.0937966362266443e-05,
|
| 1008 |
+
"loss": 0.0522,
|
| 1009 |
+
"num_tokens": 10234658.0,
|
| 1010 |
+
"step": 625
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 1.8584070796460177,
|
| 1014 |
+
"grad_norm": 0.13037630915641785,
|
| 1015 |
+
"learning_rate": 2.0588886255309192e-05,
|
| 1016 |
+
"loss": 0.0533,
|
| 1017 |
+
"num_tokens": 10316562.0,
|
| 1018 |
+
"step": 630
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 1.8731563421828907,
|
| 1022 |
+
"grad_norm": 0.12001070380210876,
|
| 1023 |
+
"learning_rate": 2.0241633507755625e-05,
|
| 1024 |
+
"loss": 0.0559,
|
| 1025 |
+
"num_tokens": 10398482.0,
|
| 1026 |
+
"step": 635
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 1.887905604719764,
|
| 1030 |
+
"grad_norm": 0.13977284729480743,
|
| 1031 |
+
"learning_rate": 1.989629993629364e-05,
|
| 1032 |
+
"loss": 0.0526,
|
| 1033 |
+
"num_tokens": 10480361.0,
|
| 1034 |
+
"step": 640
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 1.9026548672566372,
|
| 1038 |
+
"grad_norm": 0.13830554485321045,
|
| 1039 |
+
"learning_rate": 1.9552976850164047e-05,
|
| 1040 |
+
"loss": 0.0538,
|
| 1041 |
+
"num_tokens": 10562281.0,
|
| 1042 |
+
"step": 645
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 1.9174041297935103,
|
| 1046 |
+
"grad_norm": 0.11689659208059311,
|
| 1047 |
+
"learning_rate": 1.9211755027017625e-05,
|
| 1048 |
+
"loss": 0.0521,
|
| 1049 |
+
"num_tokens": 10644201.0,
|
| 1050 |
+
"step": 650
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 1.9321533923303835,
|
| 1054 |
+
"grad_norm": 0.1302964985370636,
|
| 1055 |
+
"learning_rate": 1.8872724688912684e-05,
|
| 1056 |
+
"loss": 0.0544,
|
| 1057 |
+
"num_tokens": 10726121.0,
|
| 1058 |
+
"step": 655
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.9469026548672566,
|
| 1062 |
+
"grad_norm": 0.11920026689767838,
|
| 1063 |
+
"learning_rate": 1.8535975478459566e-05,
|
| 1064 |
+
"loss": 0.0539,
|
| 1065 |
+
"num_tokens": 10807790.0,
|
| 1066 |
+
"step": 660
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.9616519174041298,
|
| 1070 |
+
"grad_norm": 0.14449501037597656,
|
| 1071 |
+
"learning_rate": 1.8201596435118356e-05,
|
| 1072 |
+
"loss": 0.052,
|
| 1073 |
+
"num_tokens": 10889546.0,
|
| 1074 |
+
"step": 665
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 1.976401179941003,
|
| 1078 |
+
"grad_norm": 0.13730932772159576,
|
| 1079 |
+
"learning_rate": 1.7869675971656062e-05,
|
| 1080 |
+
"loss": 0.0545,
|
| 1081 |
+
"num_tokens": 10971466.0,
|
| 1082 |
+
"step": 670
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 1.991150442477876,
|
| 1086 |
+
"grad_norm": 0.14631158113479614,
|
| 1087 |
+
"learning_rate": 1.7540301850769482e-05,
|
| 1088 |
+
"loss": 0.051,
|
| 1089 |
+
"num_tokens": 11053344.0,
|
| 1090 |
+
"step": 675
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 2.005899705014749,
|
| 1094 |
+
"grad_norm": 0.14203102886676788,
|
| 1095 |
+
"learning_rate": 1.721356116188001e-05,
|
| 1096 |
+
"loss": 0.0551,
|
| 1097 |
+
"num_tokens": 11135264.0,
|
| 1098 |
+
"step": 680
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 2.0206489675516224,
|
| 1102 |
+
"grad_norm": 0.14633382856845856,
|
| 1103 |
+
"learning_rate": 1.688954029810639e-05,
|
| 1104 |
+
"loss": 0.0528,
|
| 1105 |
+
"num_tokens": 11217184.0,
|
| 1106 |
+
"step": 685
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 2.0353982300884956,
|
| 1110 |
+
"grad_norm": 0.12269877642393112,
|
| 1111 |
+
"learning_rate": 1.6568324933421605e-05,
|
| 1112 |
+
"loss": 0.0524,
|
| 1113 |
+
"num_tokens": 11299104.0,
|
| 1114 |
+
"step": 690
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 2.050147492625369,
|
| 1118 |
+
"grad_norm": 0.13254043459892273,
|
| 1119 |
+
"learning_rate": 1.6250000000000005e-05,
|
| 1120 |
+
"loss": 0.0532,
|
| 1121 |
+
"num_tokens": 11381024.0,
|
| 1122 |
+
"step": 695
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 2.0648967551622417,
|
| 1126 |
+
"grad_norm": 0.13670207560062408,
|
| 1127 |
+
"learning_rate": 1.5934649665760377e-05,
|
| 1128 |
+
"loss": 0.0538,
|
| 1129 |
+
"num_tokens": 11462944.0,
|
| 1130 |
+
"step": 700
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 2.079646017699115,
|
| 1134 |
+
"grad_norm": 0.13071012496948242,
|
| 1135 |
+
"learning_rate": 1.5622357312111275e-05,
|
| 1136 |
+
"loss": 0.054,
|
| 1137 |
+
"num_tokens": 11544864.0,
|
| 1138 |
+
"step": 705
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 2.094395280235988,
|
| 1142 |
+
"grad_norm": 0.11913175135850906,
|
| 1143 |
+
"learning_rate": 1.5313205511904228e-05,
|
| 1144 |
+
"loss": 0.052,
|
| 1145 |
+
"num_tokens": 11626784.0,
|
| 1146 |
+
"step": 710
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 2.1091445427728615,
|
| 1150 |
+
"grad_norm": 0.11887428164482117,
|
| 1151 |
+
"learning_rate": 1.5007276007600757e-05,
|
| 1152 |
+
"loss": 0.0526,
|
| 1153 |
+
"num_tokens": 11708704.0,
|
| 1154 |
+
"step": 715
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 2.1238938053097347,
|
| 1158 |
+
"grad_norm": 0.12120334059000015,
|
| 1159 |
+
"learning_rate": 1.4704649689658917e-05,
|
| 1160 |
+
"loss": 0.0521,
|
| 1161 |
+
"num_tokens": 11790559.0,
|
| 1162 |
+
"step": 720
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 2.1386430678466075,
|
| 1166 |
+
"grad_norm": 0.09578295052051544,
|
| 1167 |
+
"learning_rate": 1.4405406575145198e-05,
|
| 1168 |
+
"loss": 0.0549,
|
| 1169 |
+
"num_tokens": 11872285.0,
|
| 1170 |
+
"step": 725
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 2.1533923303834808,
|
| 1174 |
+
"grad_norm": 0.13386505842208862,
|
| 1175 |
+
"learning_rate": 1.4109625786577236e-05,
|
| 1176 |
+
"loss": 0.0542,
|
| 1177 |
+
"num_tokens": 11954205.0,
|
| 1178 |
+
"step": 730
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 2.168141592920354,
|
| 1182 |
+
"grad_norm": 0.1328478455543518,
|
| 1183 |
+
"learning_rate": 1.3817385531003186e-05,
|
| 1184 |
+
"loss": 0.0518,
|
| 1185 |
+
"num_tokens": 12036002.0,
|
| 1186 |
+
"step": 735
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.1828908554572273,
|
| 1190 |
+
"grad_norm": 0.13573037087917328,
|
| 1191 |
+
"learning_rate": 1.3528763079323076e-05,
|
| 1192 |
+
"loss": 0.053,
|
| 1193 |
+
"num_tokens": 12117880.0,
|
| 1194 |
+
"step": 740
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 2.1976401179941005,
|
| 1198 |
+
"grad_norm": 0.11889643222093582,
|
| 1199 |
+
"learning_rate": 1.3243834745857667e-05,
|
| 1200 |
+
"loss": 0.0518,
|
| 1201 |
+
"num_tokens": 12199800.0,
|
| 1202 |
+
"step": 745
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 2.2123893805309733,
|
| 1206 |
+
"grad_norm": 0.13203181326389313,
|
| 1207 |
+
"learning_rate": 1.29626758681703e-05,
|
| 1208 |
+
"loss": 0.0538,
|
| 1209 |
+
"num_tokens": 12281720.0,
|
| 1210 |
+
"step": 750
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 2.2271386430678466,
|
| 1214 |
+
"grad_norm": 0.15322619676589966,
|
| 1215 |
+
"learning_rate": 1.2685360787146994e-05,
|
| 1216 |
+
"loss": 0.0521,
|
| 1217 |
+
"num_tokens": 12363640.0,
|
| 1218 |
+
"step": 755
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 2.24188790560472,
|
| 1222 |
+
"grad_norm": 0.1316722333431244,
|
| 1223 |
+
"learning_rate": 1.2411962827340023e-05,
|
| 1224 |
+
"loss": 0.0509,
|
| 1225 |
+
"num_tokens": 12445316.0,
|
| 1226 |
+
"step": 760
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 2.256637168141593,
|
| 1230 |
+
"grad_norm": 0.12419988960027695,
|
| 1231 |
+
"learning_rate": 1.2142554277580288e-05,
|
| 1232 |
+
"loss": 0.0523,
|
| 1233 |
+
"num_tokens": 12527159.0,
|
| 1234 |
+
"step": 765
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 2.271386430678466,
|
| 1238 |
+
"grad_norm": 0.16085049510002136,
|
| 1239 |
+
"learning_rate": 1.187720637186349e-05,
|
| 1240 |
+
"loss": 0.0535,
|
| 1241 |
+
"num_tokens": 12609079.0,
|
| 1242 |
+
"step": 770
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 2.286135693215339,
|
| 1246 |
+
"grad_norm": 0.1237749382853508,
|
| 1247 |
+
"learning_rate": 1.1615989270515268e-05,
|
| 1248 |
+
"loss": 0.0537,
|
| 1249 |
+
"num_tokens": 12690873.0,
|
| 1250 |
+
"step": 775
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 2.3008849557522124,
|
| 1254 |
+
"grad_norm": 0.10973517596721649,
|
| 1255 |
+
"learning_rate": 1.1358972041640139e-05,
|
| 1256 |
+
"loss": 0.0528,
|
| 1257 |
+
"num_tokens": 12772682.0,
|
| 1258 |
+
"step": 780
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 2.3156342182890857,
|
| 1262 |
+
"grad_norm": 0.12612909078598022,
|
| 1263 |
+
"learning_rate": 1.110622264285934e-05,
|
| 1264 |
+
"loss": 0.0527,
|
| 1265 |
+
"num_tokens": 12854602.0,
|
| 1266 |
+
"step": 785
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 2.330383480825959,
|
| 1270 |
+
"grad_norm": 0.14214757084846497,
|
| 1271 |
+
"learning_rate": 1.085780790334219e-05,
|
| 1272 |
+
"loss": 0.0513,
|
| 1273 |
+
"num_tokens": 12936522.0,
|
| 1274 |
+
"step": 790
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 2.3451327433628317,
|
| 1278 |
+
"grad_norm": 0.12092974781990051,
|
| 1279 |
+
"learning_rate": 1.0613793506135872e-05,
|
| 1280 |
+
"loss": 0.05,
|
| 1281 |
+
"num_tokens": 13018442.0,
|
| 1282 |
+
"step": 795
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 2.359882005899705,
|
| 1286 |
+
"grad_norm": 0.11506114155054092,
|
| 1287 |
+
"learning_rate": 1.0374243970798297e-05,
|
| 1288 |
+
"loss": 0.052,
|
| 1289 |
+
"num_tokens": 13100341.0,
|
| 1290 |
+
"step": 800
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 2.3746312684365782,
|
| 1294 |
+
"grad_norm": 0.13261866569519043,
|
| 1295 |
+
"learning_rate": 1.0139222636338505e-05,
|
| 1296 |
+
"loss": 0.0521,
|
| 1297 |
+
"num_tokens": 13182261.0,
|
| 1298 |
+
"step": 805
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 2.3893805309734515,
|
| 1302 |
+
"grad_norm": 0.13615797460079193,
|
| 1303 |
+
"learning_rate": 9.90879164446933e-06,
|
| 1304 |
+
"loss": 0.0531,
|
| 1305 |
+
"num_tokens": 13264181.0,
|
| 1306 |
+
"step": 810
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 2.4041297935103243,
|
| 1310 |
+
"grad_norm": 0.1656993329524994,
|
| 1311 |
+
"learning_rate": 9.683011923176537e-06,
|
| 1312 |
+
"loss": 0.0534,
|
| 1313 |
+
"num_tokens": 13346101.0,
|
| 1314 |
+
"step": 815
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 2.4188790560471976,
|
| 1318 |
+
"grad_norm": 0.12165559083223343,
|
| 1319 |
+
"learning_rate": 9.461943170608942e-06,
|
| 1320 |
+
"loss": 0.0523,
|
| 1321 |
+
"num_tokens": 13427963.0,
|
| 1322 |
+
"step": 820
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 2.433628318584071,
|
| 1326 |
+
"grad_norm": 0.12782715260982513,
|
| 1327 |
+
"learning_rate": 9.245643839293718e-06,
|
| 1328 |
+
"loss": 0.0517,
|
| 1329 |
+
"num_tokens": 13509883.0,
|
| 1330 |
+
"step": 825
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 2.448377581120944,
|
| 1334 |
+
"grad_norm": 0.11893630027770996,
|
| 1335 |
+
"learning_rate": 9.034171120680993e-06,
|
| 1336 |
+
"loss": 0.0543,
|
| 1337 |
+
"num_tokens": 13591803.0,
|
| 1338 |
+
"step": 830
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 2.4631268436578173,
|
| 1342 |
+
"grad_norm": 0.1121756061911583,
|
| 1343 |
+
"learning_rate": 8.827580930021936e-06,
|
| 1344 |
+
"loss": 0.0508,
|
| 1345 |
+
"num_tokens": 13673723.0,
|
| 1346 |
+
"step": 835
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 2.47787610619469,
|
| 1350 |
+
"grad_norm": 0.14923910796642303,
|
| 1351 |
+
"learning_rate": 8.625927891584307e-06,
|
| 1352 |
+
"loss": 0.052,
|
| 1353 |
+
"num_tokens": 13755595.0,
|
| 1354 |
+
"step": 840
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 2.4926253687315634,
|
| 1358 |
+
"grad_norm": 0.12945568561553955,
|
| 1359 |
+
"learning_rate": 8.429265324209275e-06,
|
| 1360 |
+
"loss": 0.0541,
|
| 1361 |
+
"num_tokens": 13837426.0,
|
| 1362 |
+
"step": 845
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 2.5073746312684366,
|
| 1366 |
+
"grad_norm": 0.11035799235105515,
|
| 1367 |
+
"learning_rate": 8.237645227213465e-06,
|
| 1368 |
+
"loss": 0.0535,
|
| 1369 |
+
"num_tokens": 13919095.0,
|
| 1370 |
+
"step": 850
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 2.52212389380531,
|
| 1374 |
+
"grad_norm": 0.12474622577428818,
|
| 1375 |
+
"learning_rate": 8.051118266639879e-06,
|
| 1376 |
+
"loss": 0.0527,
|
| 1377 |
+
"num_tokens": 14001015.0,
|
| 1378 |
+
"step": 855
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 2.5368731563421827,
|
| 1382 |
+
"grad_norm": 0.12021201103925705,
|
| 1383 |
+
"learning_rate": 7.869733761861347e-06,
|
| 1384 |
+
"loss": 0.0519,
|
| 1385 |
+
"num_tokens": 14082733.0,
|
| 1386 |
+
"step": 860
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 2.551622418879056,
|
| 1390 |
+
"grad_norm": 0.15574879944324493,
|
| 1391 |
+
"learning_rate": 7.693539672540045e-06,
|
| 1392 |
+
"loss": 0.0524,
|
| 1393 |
+
"num_tokens": 14164653.0,
|
| 1394 |
+
"step": 865
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 2.566371681415929,
|
| 1398 |
+
"grad_norm": 0.12805156409740448,
|
| 1399 |
+
"learning_rate": 7.522582585946558e-06,
|
| 1400 |
+
"loss": 0.0527,
|
| 1401 |
+
"num_tokens": 14246457.0,
|
| 1402 |
+
"step": 870
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 2.5811209439528024,
|
| 1406 |
+
"grad_norm": 0.13377341628074646,
|
| 1407 |
+
"learning_rate": 7.356907704641764e-06,
|
| 1408 |
+
"loss": 0.0521,
|
| 1409 |
+
"num_tokens": 14328377.0,
|
| 1410 |
+
"step": 875
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 2.5958702064896757,
|
| 1414 |
+
"grad_norm": 0.1164890006184578,
|
| 1415 |
+
"learning_rate": 7.196558834524891e-06,
|
| 1416 |
+
"loss": 0.0531,
|
| 1417 |
+
"num_tokens": 14410297.0,
|
| 1418 |
+
"step": 880
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 2.6106194690265485,
|
| 1422 |
+
"grad_norm": 0.1588173806667328,
|
| 1423 |
+
"learning_rate": 7.041578373250875e-06,
|
| 1424 |
+
"loss": 0.0513,
|
| 1425 |
+
"num_tokens": 14491946.0,
|
| 1426 |
+
"step": 885
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 2.6253687315634218,
|
| 1430 |
+
"grad_norm": 0.12378757447004318,
|
| 1431 |
+
"learning_rate": 6.892007299020003e-06,
|
| 1432 |
+
"loss": 0.0523,
|
| 1433 |
+
"num_tokens": 14573866.0,
|
| 1434 |
+
"step": 890
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 2.640117994100295,
|
| 1438 |
+
"grad_norm": 0.12086950987577438,
|
| 1439 |
+
"learning_rate": 6.747885159742945e-06,
|
| 1440 |
+
"loss": 0.0528,
|
| 1441 |
+
"num_tokens": 14655786.0,
|
| 1442 |
+
"step": 895
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 2.6548672566371683,
|
| 1446 |
+
"grad_norm": 0.13173659145832062,
|
| 1447 |
+
"learning_rate": 6.609250062583937e-06,
|
| 1448 |
+
"loss": 0.0513,
|
| 1449 |
+
"num_tokens": 14737706.0,
|
| 1450 |
+
"step": 900
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 2.669616519174041,
|
| 1454 |
+
"grad_norm": 0.11908340454101562,
|
| 1455 |
+
"learning_rate": 6.476138663884902e-06,
|
| 1456 |
+
"loss": 0.0508,
|
| 1457 |
+
"num_tokens": 14819610.0,
|
| 1458 |
+
"step": 905
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 2.6843657817109143,
|
| 1462 |
+
"grad_norm": 0.16239605844020844,
|
| 1463 |
+
"learning_rate": 6.34858615947318e-06,
|
| 1464 |
+
"loss": 0.0536,
|
| 1465 |
+
"num_tokens": 14901366.0,
|
| 1466 |
+
"step": 910
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 2.6991150442477876,
|
| 1470 |
+
"grad_norm": 0.14603886008262634,
|
| 1471 |
+
"learning_rate": 6.226626275355474e-06,
|
| 1472 |
+
"loss": 0.0506,
|
| 1473 |
+
"num_tokens": 14983286.0,
|
| 1474 |
+
"step": 915
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 2.713864306784661,
|
| 1478 |
+
"grad_norm": 0.13718026876449585,
|
| 1479 |
+
"learning_rate": 6.110291258800356e-06,
|
| 1480 |
+
"loss": 0.0519,
|
| 1481 |
+
"num_tokens": 15065206.0,
|
| 1482 |
+
"step": 920
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 2.728613569321534,
|
| 1486 |
+
"grad_norm": 0.12413609027862549,
|
| 1487 |
+
"learning_rate": 5.9996118698118335e-06,
|
| 1488 |
+
"loss": 0.052,
|
| 1489 |
+
"num_tokens": 15147126.0,
|
| 1490 |
+
"step": 925
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 2.7433628318584073,
|
| 1494 |
+
"grad_norm": 0.14036346971988678,
|
| 1495 |
+
"learning_rate": 5.89461737299613e-06,
|
| 1496 |
+
"loss": 0.0515,
|
| 1497 |
+
"num_tokens": 15229046.0,
|
| 1498 |
+
"step": 930
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 2.75811209439528,
|
| 1502 |
+
"grad_norm": 0.1299838423728943,
|
| 1503 |
+
"learning_rate": 5.795335529823848e-06,
|
| 1504 |
+
"loss": 0.0523,
|
| 1505 |
+
"num_tokens": 15310966.0,
|
| 1506 |
+
"step": 935
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 2.7728613569321534,
|
| 1510 |
+
"grad_norm": 0.12340506166219711,
|
| 1511 |
+
"learning_rate": 5.701792591289609e-06,
|
| 1512 |
+
"loss": 0.0524,
|
| 1513 |
+
"num_tokens": 15392621.0,
|
| 1514 |
+
"step": 940
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 2.7876106194690267,
|
| 1518 |
+
"grad_norm": 0.11809273809194565,
|
| 1519 |
+
"learning_rate": 5.614013290971055e-06,
|
| 1520 |
+
"loss": 0.0512,
|
| 1521 |
+
"num_tokens": 15474541.0,
|
| 1522 |
+
"step": 945
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 2.8023598820058995,
|
| 1526 |
+
"grad_norm": 0.11352474987506866,
|
| 1527 |
+
"learning_rate": 5.532020838489065e-06,
|
| 1528 |
+
"loss": 0.052,
|
| 1529 |
+
"num_tokens": 15556461.0,
|
| 1530 |
+
"step": 950
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 2.8171091445427727,
|
| 1534 |
+
"grad_norm": 0.12270710617303848,
|
| 1535 |
+
"learning_rate": 5.455836913370934e-06,
|
| 1536 |
+
"loss": 0.0533,
|
| 1537 |
+
"num_tokens": 15638316.0,
|
| 1538 |
+
"step": 955
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 2.831858407079646,
|
| 1542 |
+
"grad_norm": 0.1239713653922081,
|
| 1543 |
+
"learning_rate": 5.38548165931812e-06,
|
| 1544 |
+
"loss": 0.0523,
|
| 1545 |
+
"num_tokens": 15720236.0,
|
| 1546 |
+
"step": 960
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 2.8466076696165192,
|
| 1550 |
+
"grad_norm": 0.13654908537864685,
|
| 1551 |
+
"learning_rate": 5.3209736788800545e-06,
|
| 1552 |
+
"loss": 0.0505,
|
| 1553 |
+
"num_tokens": 15802115.0,
|
| 1554 |
+
"step": 965
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 2.8613569321533925,
|
| 1558 |
+
"grad_norm": 0.12271752953529358,
|
| 1559 |
+
"learning_rate": 5.262330028535478e-06,
|
| 1560 |
+
"loss": 0.0527,
|
| 1561 |
+
"num_tokens": 15884035.0,
|
| 1562 |
+
"step": 970
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 2.8761061946902657,
|
| 1566 |
+
"grad_norm": 0.11097793281078339,
|
| 1567 |
+
"learning_rate": 5.209566214182558e-06,
|
| 1568 |
+
"loss": 0.0503,
|
| 1569 |
+
"num_tokens": 15965955.0,
|
| 1570 |
+
"step": 975
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 2.8908554572271385,
|
| 1574 |
+
"grad_norm": 0.12811797857284546,
|
| 1575 |
+
"learning_rate": 5.162696187038983e-06,
|
| 1576 |
+
"loss": 0.0521,
|
| 1577 |
+
"num_tokens": 16047875.0,
|
| 1578 |
+
"step": 980
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 2.905604719764012,
|
| 1582 |
+
"grad_norm": 0.12047919631004333,
|
| 1583 |
+
"learning_rate": 5.121732339953144e-06,
|
| 1584 |
+
"loss": 0.0526,
|
| 1585 |
+
"num_tokens": 16129795.0,
|
| 1586 |
+
"step": 985
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 2.920353982300885,
|
| 1590 |
+
"grad_norm": 0.1275618076324463,
|
| 1591 |
+
"learning_rate": 5.086685504127355e-06,
|
| 1592 |
+
"loss": 0.0525,
|
| 1593 |
+
"num_tokens": 16211715.0,
|
| 1594 |
+
"step": 990
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 2.935103244837758,
|
| 1598 |
+
"grad_norm": 0.13480406999588013,
|
| 1599 |
+
"learning_rate": 5.057564946253982e-06,
|
| 1600 |
+
"loss": 0.054,
|
| 1601 |
+
"num_tokens": 16293598.0,
|
| 1602 |
+
"step": 995
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 2.949852507374631,
|
| 1606 |
+
"grad_norm": 0.17191572487354279,
|
| 1607 |
+
"learning_rate": 5.03437836606527e-06,
|
| 1608 |
+
"loss": 0.0514,
|
| 1609 |
+
"num_tokens": 16375518.0,
|
| 1610 |
+
"step": 1000
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 2.9646017699115044,
|
| 1614 |
+
"grad_norm": 0.1515762060880661,
|
| 1615 |
+
"learning_rate": 5.0171318942974285e-06,
|
| 1616 |
+
"loss": 0.054,
|
| 1617 |
+
"num_tokens": 16457438.0,
|
| 1618 |
+
"step": 1005
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 2.9793510324483776,
|
| 1622 |
+
"grad_norm": 0.15460149943828583,
|
| 1623 |
+
"learning_rate": 5.005830091069644e-06,
|
| 1624 |
+
"loss": 0.053,
|
| 1625 |
+
"num_tokens": 16539277.0,
|
| 1626 |
+
"step": 1010
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 2.994100294985251,
|
| 1630 |
+
"grad_norm": 0.11925122141838074,
|
| 1631 |
+
"learning_rate": 5.000475944678329e-06,
|
| 1632 |
+
"loss": 0.0522,
|
| 1633 |
+
"num_tokens": 16620976.0,
|
| 1634 |
+
"step": 1015
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 3.0,
|
| 1638 |
+
"num_tokens": 16653744.0,
|
| 1639 |
+
"step": 1017,
|
| 1640 |
+
"total_flos": 2.7740847281799168e+17,
|
| 1641 |
+
"train_loss": 0.0704414379602571,
|
| 1642 |
+
"train_runtime": 3418.2006,
|
| 1643 |
+
"train_samples_per_second": 18.999,
|
| 1644 |
+
"train_steps_per_second": 0.298
|
| 1645 |
}
|
| 1646 |
],
|
| 1647 |
"logging_steps": 5,
|
| 1648 |
+
"max_steps": 1017,
|
| 1649 |
"num_input_tokens_seen": 0,
|
| 1650 |
+
"num_train_epochs": 3,
|
| 1651 |
"save_steps": 100,
|
| 1652 |
"stateful_callbacks": {
|
| 1653 |
"TrainerControl": {
|
|
|
|
| 1661 |
"attributes": {}
|
| 1662 |
}
|
| 1663 |
},
|
| 1664 |
+
"total_flos": 2.7740847281799168e+17,
|
| 1665 |
"train_batch_size": 16,
|
| 1666 |
"trial_name": null,
|
| 1667 |
"trial_params": null
|