Upload folder using huggingface_hub
Browse files- math-0.3/q/cold-start/README.md +1 -1
- math-0.3/q/cold-start/adapter_config.json +4 -4
- math-0.3/q/cold-start/adapter_model.safetensors +1 -1
- math-0.3/q/cold-start/checkpoint-500/adapter_config.json +4 -4
- math-0.3/q/cold-start/checkpoint-500/adapter_model.safetensors +1 -1
- math-0.3/q/cold-start/checkpoint-500/optimizer.pt +1 -1
- math-0.3/q/cold-start/checkpoint-500/scheduler.pt +1 -1
- math-0.3/q/cold-start/checkpoint-500/trainer_state.json +302 -302
- math-0.3/q/cold-start/checkpoint-500/training_args.bin +1 -1
- math-0.3/q/cold-start/checkpoint-552/adapter_config.json +4 -4
- math-0.3/q/cold-start/checkpoint-552/adapter_model.safetensors +1 -1
- math-0.3/q/cold-start/checkpoint-552/optimizer.pt +1 -1
- math-0.3/q/cold-start/checkpoint-552/trainer_state.json +332 -332
- math-0.3/q/cold-start/checkpoint-552/training_args.bin +1 -1
math-0.3/q/cold-start/README.md
CHANGED
|
@@ -31,7 +31,7 @@ print(output["generated_text"])
|
|
| 31 |
|
| 32 |
## Training procedure
|
| 33 |
|
| 34 |
-
[<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/kstaron-/huggingface/runs/
|
| 35 |
|
| 36 |
|
| 37 |
This model was trained with SFT.
|
|
|
|
| 31 |
|
| 32 |
## Training procedure
|
| 33 |
|
| 34 |
+
[<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/kstaron-/huggingface/runs/xnqhar1n)
|
| 35 |
|
| 36 |
|
| 37 |
This model was trained with SFT.
|
math-0.3/q/cold-start/adapter_config.json
CHANGED
|
@@ -33,12 +33,12 @@
|
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
| 36 |
-
"up_proj",
|
| 37 |
-
"down_proj",
|
| 38 |
"o_proj",
|
| 39 |
-
"gate_proj",
|
| 40 |
-
"k_proj",
|
| 41 |
"v_proj",
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
"q_proj"
|
| 43 |
],
|
| 44 |
"target_parameters": null,
|
|
|
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
|
|
|
|
|
|
| 36 |
"o_proj",
|
|
|
|
|
|
|
| 37 |
"v_proj",
|
| 38 |
+
"down_proj",
|
| 39 |
+
"up_proj",
|
| 40 |
+
"k_proj",
|
| 41 |
+
"gate_proj",
|
| 42 |
"q_proj"
|
| 43 |
],
|
| 44 |
"target_parameters": null,
|
math-0.3/q/cold-start/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 645975704
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:589167f278372c18db431c80611f980e909def4a36684858409210419c692f2c
|
| 3 |
size 645975704
|
math-0.3/q/cold-start/checkpoint-500/adapter_config.json
CHANGED
|
@@ -33,12 +33,12 @@
|
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
| 36 |
-
"up_proj",
|
| 37 |
-
"down_proj",
|
| 38 |
"o_proj",
|
| 39 |
-
"gate_proj",
|
| 40 |
-
"k_proj",
|
| 41 |
"v_proj",
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
"q_proj"
|
| 43 |
],
|
| 44 |
"target_parameters": null,
|
|
|
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
|
|
|
|
|
|
| 36 |
"o_proj",
|
|
|
|
|
|
|
| 37 |
"v_proj",
|
| 38 |
+
"down_proj",
|
| 39 |
+
"up_proj",
|
| 40 |
+
"k_proj",
|
| 41 |
+
"gate_proj",
|
| 42 |
"q_proj"
|
| 43 |
],
|
| 44 |
"target_parameters": null,
|
math-0.3/q/cold-start/checkpoint-500/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 645975704
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ebc45284ce6528f338047a950bb9a82f3af9b908d1516485e2c4b04a896b0c9
|
| 3 |
size 645975704
|
math-0.3/q/cold-start/checkpoint-500/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1292182139
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e10425f9fa6ff59012168253a5149e52ae21efffc7fd37c26fe9f40d8d6026d5
|
| 3 |
size 1292182139
|
math-0.3/q/cold-start/checkpoint-500/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:12ee79043f31eb49d1a4e02f98b7f34ebd2e507e24da998fcc93e6498acae7f7
|
| 3 |
size 1465
|
math-0.3/q/cold-start/checkpoint-500/trainer_state.json
CHANGED
|
@@ -11,709 +11,709 @@
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.001813236627379873,
|
| 14 |
-
"grad_norm": 0.
|
| 15 |
"learning_rate": 0.0,
|
| 16 |
-
"loss": 1.
|
| 17 |
"step": 1
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.009066183136899365,
|
| 21 |
-
"grad_norm": 0.
|
| 22 |
-
"learning_rate":
|
| 23 |
-
"loss": 1.
|
| 24 |
"step": 5
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.01813236627379873,
|
| 28 |
-
"grad_norm": 0.
|
| 29 |
-
"learning_rate":
|
| 30 |
-
"loss": 1.
|
| 31 |
"step": 10
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.027198549410698096,
|
| 35 |
-
"grad_norm": 0.
|
| 36 |
-
"learning_rate":
|
| 37 |
-
"loss": 1.
|
| 38 |
"step": 15
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.03626473254759746,
|
| 42 |
-
"grad_norm": 0.
|
| 43 |
-
"learning_rate":
|
| 44 |
-
"loss":
|
| 45 |
"step": 20
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.04533091568449683,
|
| 49 |
-
"grad_norm": 0.
|
| 50 |
-
"learning_rate":
|
| 51 |
-
"loss":
|
| 52 |
"step": 25
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 0.05439709882139619,
|
| 56 |
-
"grad_norm": 0.
|
| 57 |
-
"learning_rate":
|
| 58 |
-
"loss":
|
| 59 |
"step": 30
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 0.06346328195829556,
|
| 63 |
-
"grad_norm": 0.
|
| 64 |
-
"learning_rate":
|
| 65 |
-
"loss": 0.
|
| 66 |
"step": 35
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 0.07252946509519492,
|
| 70 |
-
"grad_norm": 0.
|
| 71 |
-
"learning_rate":
|
| 72 |
-
"loss": 0.
|
| 73 |
"step": 40
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 0.0815956482320943,
|
| 77 |
-
"grad_norm": 0.
|
| 78 |
-
"learning_rate":
|
| 79 |
-
"loss": 0.
|
| 80 |
"step": 45
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.09066183136899365,
|
| 84 |
-
"grad_norm": 0.
|
| 85 |
-
"learning_rate":
|
| 86 |
-
"loss": 0.
|
| 87 |
"step": 50
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.09972801450589303,
|
| 91 |
-
"grad_norm": 0.
|
| 92 |
-
"learning_rate":
|
| 93 |
-
"loss": 0.
|
| 94 |
"step": 55
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 0.10879419764279238,
|
| 98 |
-
"grad_norm": 0.
|
| 99 |
-
"learning_rate":
|
| 100 |
-
"loss": 0.
|
| 101 |
"step": 60
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 0.11786038077969176,
|
| 105 |
-
"grad_norm": 0.
|
| 106 |
-
"learning_rate":
|
| 107 |
-
"loss": 0.
|
| 108 |
"step": 65
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 0.12692656391659113,
|
| 112 |
-
"grad_norm": 0.
|
| 113 |
-
"learning_rate":
|
| 114 |
-
"loss": 0.
|
| 115 |
"step": 70
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 0.1359927470534905,
|
| 119 |
-
"grad_norm": 0.
|
| 120 |
-
"learning_rate":
|
| 121 |
-
"loss": 0.
|
| 122 |
"step": 75
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.14505893019038985,
|
| 126 |
-
"grad_norm": 0.
|
| 127 |
-
"learning_rate":
|
| 128 |
-
"loss": 0.
|
| 129 |
"step": 80
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.1541251133272892,
|
| 133 |
-
"grad_norm": 0.
|
| 134 |
-
"learning_rate": 4.
|
| 135 |
-
"loss": 0.
|
| 136 |
"step": 85
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.1631912964641886,
|
| 140 |
-
"grad_norm": 0.
|
| 141 |
-
"learning_rate": 4.
|
| 142 |
-
"loss": 0.
|
| 143 |
"step": 90
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 0.17225747960108795,
|
| 147 |
-
"grad_norm": 0.
|
| 148 |
-
"learning_rate": 4.
|
| 149 |
-
"loss": 0.
|
| 150 |
"step": 95
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 0.1813236627379873,
|
| 154 |
-
"grad_norm": 0.
|
| 155 |
-
"learning_rate": 4.
|
| 156 |
-
"loss": 0.
|
| 157 |
"step": 100
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 0.19038984587488667,
|
| 161 |
-
"grad_norm": 0.
|
| 162 |
-
"learning_rate": 4.
|
| 163 |
-
"loss": 0.
|
| 164 |
"step": 105
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.19945602901178605,
|
| 168 |
-
"grad_norm": 0.
|
| 169 |
-
"learning_rate": 4.
|
| 170 |
-
"loss": 0.
|
| 171 |
"step": 110
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.2085222121486854,
|
| 175 |
-
"grad_norm": 0.
|
| 176 |
-
"learning_rate": 4.
|
| 177 |
-
"loss": 0.
|
| 178 |
"step": 115
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 0.21758839528558477,
|
| 182 |
-
"grad_norm": 0.
|
| 183 |
-
"learning_rate": 4.
|
| 184 |
-
"loss": 0.
|
| 185 |
"step": 120
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 0.22665457842248413,
|
| 189 |
-
"grad_norm": 0.
|
| 190 |
-
"learning_rate": 4.
|
| 191 |
-
"loss": 0.
|
| 192 |
"step": 125
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 0.2357207615593835,
|
| 196 |
-
"grad_norm": 0.
|
| 197 |
-
"learning_rate":
|
| 198 |
-
"loss": 0.
|
| 199 |
"step": 130
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.24478694469628287,
|
| 203 |
-
"grad_norm": 0.
|
| 204 |
-
"learning_rate":
|
| 205 |
-
"loss": 0.
|
| 206 |
"step": 135
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.25385312783318226,
|
| 210 |
-
"grad_norm": 0.
|
| 211 |
-
"learning_rate":
|
| 212 |
-
"loss": 0.
|
| 213 |
"step": 140
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.2629193109700816,
|
| 217 |
-
"grad_norm": 0.
|
| 218 |
-
"learning_rate":
|
| 219 |
-
"loss": 0.
|
| 220 |
"step": 145
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 0.271985494106981,
|
| 224 |
-
"grad_norm": 0.
|
| 225 |
-
"learning_rate":
|
| 226 |
-
"loss": 0.
|
| 227 |
"step": 150
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 0.2810516772438803,
|
| 231 |
-
"grad_norm": 0.
|
| 232 |
-
"learning_rate":
|
| 233 |
-
"loss": 0.
|
| 234 |
"step": 155
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.2901178603807797,
|
| 238 |
-
"grad_norm": 0.
|
| 239 |
-
"learning_rate":
|
| 240 |
-
"loss": 0.
|
| 241 |
"step": 160
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 0.2991840435176791,
|
| 245 |
-
"grad_norm": 0.
|
| 246 |
-
"learning_rate":
|
| 247 |
-
"loss": 0.
|
| 248 |
"step": 165
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 0.3082502266545784,
|
| 252 |
-
"grad_norm": 0.
|
| 253 |
-
"learning_rate":
|
| 254 |
-
"loss": 0.
|
| 255 |
"step": 170
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.3173164097914778,
|
| 259 |
-
"grad_norm": 0.
|
| 260 |
-
"learning_rate":
|
| 261 |
-
"loss": 0.
|
| 262 |
"step": 175
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 0.3263825929283772,
|
| 266 |
-
"grad_norm": 0.
|
| 267 |
-
"learning_rate":
|
| 268 |
-
"loss": 0.
|
| 269 |
"step": 180
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 0.3354487760652765,
|
| 273 |
-
"grad_norm": 0.
|
| 274 |
-
"learning_rate":
|
| 275 |
-
"loss": 0.
|
| 276 |
"step": 185
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 0.3445149592021759,
|
| 280 |
-
"grad_norm": 0.
|
| 281 |
-
"learning_rate":
|
| 282 |
-
"loss": 0.
|
| 283 |
"step": 190
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 0.3535811423390752,
|
| 287 |
-
"grad_norm": 0.
|
| 288 |
-
"learning_rate": 3.
|
| 289 |
-
"loss": 0.
|
| 290 |
"step": 195
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 0.3626473254759746,
|
| 294 |
-
"grad_norm": 0.
|
| 295 |
-
"learning_rate": 3.
|
| 296 |
-
"loss": 0.
|
| 297 |
"step": 200
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.371713508612874,
|
| 301 |
-
"grad_norm": 0.
|
| 302 |
-
"learning_rate": 3.
|
| 303 |
-
"loss": 0.
|
| 304 |
"step": 205
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 0.38077969174977333,
|
| 308 |
-
"grad_norm": 0.
|
| 309 |
-
"learning_rate": 3.
|
| 310 |
-
"loss": 0.
|
| 311 |
"step": 210
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 0.3898458748866727,
|
| 315 |
-
"grad_norm": 0.
|
| 316 |
-
"learning_rate": 3.
|
| 317 |
-
"loss": 0.
|
| 318 |
"step": 215
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 0.3989120580235721,
|
| 322 |
-
"grad_norm": 0.
|
| 323 |
-
"learning_rate": 3.
|
| 324 |
-
"loss": 0.
|
| 325 |
"step": 220
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.40797824116047143,
|
| 329 |
-
"grad_norm": 0.
|
| 330 |
-
"learning_rate": 3.
|
| 331 |
-
"loss": 0.
|
| 332 |
"step": 225
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 0.4170444242973708,
|
| 336 |
-
"grad_norm": 0.
|
| 337 |
-
"learning_rate": 3.
|
| 338 |
-
"loss": 0.
|
| 339 |
"step": 230
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.42611060743427015,
|
| 343 |
-
"grad_norm": 0.
|
| 344 |
-
"learning_rate":
|
| 345 |
-
"loss": 0.
|
| 346 |
"step": 235
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 0.43517679057116954,
|
| 350 |
-
"grad_norm": 0.
|
| 351 |
-
"learning_rate":
|
| 352 |
-
"loss": 0.
|
| 353 |
"step": 240
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 0.4442429737080689,
|
| 357 |
-
"grad_norm": 0.
|
| 358 |
-
"learning_rate":
|
| 359 |
-
"loss": 0.
|
| 360 |
"step": 245
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 0.45330915684496825,
|
| 364 |
-
"grad_norm": 0.
|
| 365 |
-
"learning_rate":
|
| 366 |
-
"loss": 0.
|
| 367 |
"step": 250
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 0.46237533998186764,
|
| 371 |
-
"grad_norm": 0.
|
| 372 |
-
"learning_rate":
|
| 373 |
-
"loss": 0.
|
| 374 |
"step": 255
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 0.471441523118767,
|
| 378 |
-
"grad_norm": 0.
|
| 379 |
-
"learning_rate":
|
| 380 |
-
"loss": 0.
|
| 381 |
"step": 260
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 0.48050770625566636,
|
| 385 |
-
"grad_norm": 0.
|
| 386 |
-
"learning_rate":
|
| 387 |
-
"loss": 0.
|
| 388 |
"step": 265
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 0.48957388939256574,
|
| 392 |
-
"grad_norm": 0.
|
| 393 |
-
"learning_rate":
|
| 394 |
-
"loss": 0.
|
| 395 |
"step": 270
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 0.4986400725294651,
|
| 399 |
-
"grad_norm": 0.
|
| 400 |
-
"learning_rate":
|
| 401 |
-
"loss": 0.
|
| 402 |
"step": 275
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 0.5077062556663645,
|
| 406 |
-
"grad_norm": 0.
|
| 407 |
-
"learning_rate":
|
| 408 |
-
"loss": 0.
|
| 409 |
"step": 280
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 0.5167724388032638,
|
| 413 |
-
"grad_norm": 0.
|
| 414 |
-
"learning_rate": 2.
|
| 415 |
-
"loss": 0.
|
| 416 |
"step": 285
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 0.5258386219401632,
|
| 420 |
-
"grad_norm": 0.
|
| 421 |
-
"learning_rate": 2.
|
| 422 |
-
"loss": 0.
|
| 423 |
"step": 290
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 0.5349048050770625,
|
| 427 |
-
"grad_norm": 0.
|
| 428 |
-
"learning_rate": 2.
|
| 429 |
-
"loss": 0.
|
| 430 |
"step": 295
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 0.543970988213962,
|
| 434 |
-
"grad_norm": 0.
|
| 435 |
-
"learning_rate": 2.
|
| 436 |
-
"loss": 0.
|
| 437 |
"step": 300
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 0.5530371713508613,
|
| 441 |
-
"grad_norm": 0.
|
| 442 |
-
"learning_rate": 2.
|
| 443 |
-
"loss": 0.
|
| 444 |
"step": 305
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 0.5621033544877606,
|
| 448 |
-
"grad_norm": 0.
|
| 449 |
-
"learning_rate": 2.
|
| 450 |
-
"loss": 0.
|
| 451 |
"step": 310
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 0.57116953762466,
|
| 455 |
-
"grad_norm": 0.
|
| 456 |
-
"learning_rate": 2.
|
| 457 |
-
"loss": 0.
|
| 458 |
"step": 315
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 0.5802357207615594,
|
| 462 |
-
"grad_norm": 0.
|
| 463 |
-
"learning_rate": 2.
|
| 464 |
-
"loss": 0.
|
| 465 |
"step": 320
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 0.5893019038984587,
|
| 469 |
-
"grad_norm": 0.
|
| 470 |
-
"learning_rate": 2.
|
| 471 |
-
"loss": 0.
|
| 472 |
"step": 325
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 0.5983680870353582,
|
| 476 |
-
"grad_norm": 0.
|
| 477 |
-
"learning_rate": 2.
|
| 478 |
-
"loss": 0.
|
| 479 |
"step": 330
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 0.6074342701722575,
|
| 483 |
-
"grad_norm": 0.
|
| 484 |
-
"learning_rate": 2.
|
| 485 |
-
"loss": 0.
|
| 486 |
"step": 335
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 0.6165004533091568,
|
| 490 |
-
"grad_norm": 0.
|
| 491 |
-
"learning_rate": 2.
|
| 492 |
-
"loss": 0.
|
| 493 |
"step": 340
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 0.6255666364460563,
|
| 497 |
-
"grad_norm": 0.
|
| 498 |
-
"learning_rate":
|
| 499 |
-
"loss": 0.
|
| 500 |
"step": 345
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 0.6346328195829556,
|
| 504 |
-
"grad_norm": 0.
|
| 505 |
-
"learning_rate":
|
| 506 |
-
"loss": 0.
|
| 507 |
"step": 350
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 0.6436990027198549,
|
| 511 |
-
"grad_norm": 0.
|
| 512 |
-
"learning_rate":
|
| 513 |
-
"loss": 0.
|
| 514 |
"step": 355
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 0.6527651858567544,
|
| 518 |
-
"grad_norm": 0.
|
| 519 |
-
"learning_rate":
|
| 520 |
-
"loss": 0.
|
| 521 |
"step": 360
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 0.6618313689936537,
|
| 525 |
-
"grad_norm": 0.
|
| 526 |
-
"learning_rate":
|
| 527 |
-
"loss": 0.
|
| 528 |
"step": 365
|
| 529 |
},
|
| 530 |
{
|
| 531 |
"epoch": 0.670897552130553,
|
| 532 |
-
"grad_norm": 0.
|
| 533 |
-
"learning_rate":
|
| 534 |
-
"loss": 0.
|
| 535 |
"step": 370
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"epoch": 0.6799637352674524,
|
| 539 |
-
"grad_norm": 0.
|
| 540 |
-
"learning_rate": 1.
|
| 541 |
-
"loss": 0.
|
| 542 |
"step": 375
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 0.6890299184043518,
|
| 546 |
-
"grad_norm": 0.
|
| 547 |
-
"learning_rate": 1.
|
| 548 |
-
"loss": 0.
|
| 549 |
"step": 380
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.6980961015412511,
|
| 553 |
-
"grad_norm": 0.
|
| 554 |
-
"learning_rate": 1.
|
| 555 |
-
"loss": 0.
|
| 556 |
"step": 385
|
| 557 |
},
|
| 558 |
{
|
| 559 |
"epoch": 0.7071622846781505,
|
| 560 |
-
"grad_norm": 0.
|
| 561 |
-
"learning_rate": 1.
|
| 562 |
-
"loss": 0.
|
| 563 |
"step": 390
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 0.7162284678150499,
|
| 567 |
-
"grad_norm": 0.
|
| 568 |
-
"learning_rate": 1.
|
| 569 |
-
"loss": 0.
|
| 570 |
"step": 395
|
| 571 |
},
|
| 572 |
{
|
| 573 |
"epoch": 0.7252946509519492,
|
| 574 |
-
"grad_norm": 0.
|
| 575 |
-
"learning_rate": 1.
|
| 576 |
-
"loss": 0.
|
| 577 |
"step": 400
|
| 578 |
},
|
| 579 |
{
|
| 580 |
"epoch": 0.7343608340888486,
|
| 581 |
-
"grad_norm": 0.
|
| 582 |
-
"learning_rate": 1.
|
| 583 |
-
"loss": 0.
|
| 584 |
"step": 405
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 0.743427017225748,
|
| 588 |
-
"grad_norm": 0.
|
| 589 |
-
"learning_rate": 1.
|
| 590 |
-
"loss": 0.
|
| 591 |
"step": 410
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 0.7524932003626473,
|
| 595 |
-
"grad_norm": 0.
|
| 596 |
-
"learning_rate": 1.
|
| 597 |
-
"loss": 0.
|
| 598 |
"step": 415
|
| 599 |
},
|
| 600 |
{
|
| 601 |
"epoch": 0.7615593834995467,
|
| 602 |
-
"grad_norm": 0.
|
| 603 |
-
"learning_rate": 1.
|
| 604 |
-
"loss": 0.
|
| 605 |
"step": 420
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"epoch": 0.7706255666364461,
|
| 609 |
-
"grad_norm": 0.
|
| 610 |
-
"learning_rate": 1.
|
| 611 |
-
"loss": 0.
|
| 612 |
"step": 425
|
| 613 |
},
|
| 614 |
{
|
| 615 |
"epoch": 0.7796917497733454,
|
| 616 |
-
"grad_norm": 0.
|
| 617 |
-
"learning_rate": 1.
|
| 618 |
-
"loss": 0.
|
| 619 |
"step": 430
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"epoch": 0.7887579329102448,
|
| 623 |
-
"grad_norm": 0.
|
| 624 |
-
"learning_rate": 1.
|
| 625 |
-
"loss": 0.
|
| 626 |
"step": 435
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 0.7978241160471442,
|
| 630 |
-
"grad_norm": 0.
|
| 631 |
-
"learning_rate": 1.
|
| 632 |
-
"loss": 0.
|
| 633 |
"step": 440
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 0.8068902991840435,
|
| 637 |
-
"grad_norm": 0.
|
| 638 |
-
"learning_rate": 1.
|
| 639 |
-
"loss": 0.
|
| 640 |
"step": 445
|
| 641 |
},
|
| 642 |
{
|
| 643 |
"epoch": 0.8159564823209429,
|
| 644 |
-
"grad_norm": 0.
|
| 645 |
-
"learning_rate":
|
| 646 |
-
"loss": 0.
|
| 647 |
"step": 450
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"epoch": 0.8250226654578422,
|
| 651 |
-
"grad_norm": 0.
|
| 652 |
-
"learning_rate":
|
| 653 |
-
"loss": 0.
|
| 654 |
"step": 455
|
| 655 |
},
|
| 656 |
{
|
| 657 |
"epoch": 0.8340888485947416,
|
| 658 |
-
"grad_norm": 0.
|
| 659 |
-
"learning_rate":
|
| 660 |
-
"loss": 0.
|
| 661 |
"step": 460
|
| 662 |
},
|
| 663 |
{
|
| 664 |
"epoch": 0.843155031731641,
|
| 665 |
-
"grad_norm": 0.
|
| 666 |
-
"learning_rate":
|
| 667 |
-
"loss": 0.
|
| 668 |
"step": 465
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 0.8522212148685403,
|
| 672 |
-
"grad_norm": 0.
|
| 673 |
-
"learning_rate":
|
| 674 |
-
"loss": 0.
|
| 675 |
"step": 470
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 0.8612873980054397,
|
| 679 |
-
"grad_norm": 0.
|
| 680 |
-
"learning_rate":
|
| 681 |
-
"loss": 0.
|
| 682 |
"step": 475
|
| 683 |
},
|
| 684 |
{
|
| 685 |
"epoch": 0.8703535811423391,
|
| 686 |
-
"grad_norm": 0.
|
| 687 |
-
"learning_rate":
|
| 688 |
-
"loss": 0.
|
| 689 |
"step": 480
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"epoch": 0.8794197642792384,
|
| 693 |
-
"grad_norm": 0.
|
| 694 |
-
"learning_rate":
|
| 695 |
-
"loss": 0.
|
| 696 |
"step": 485
|
| 697 |
},
|
| 698 |
{
|
| 699 |
"epoch": 0.8884859474161378,
|
| 700 |
-
"grad_norm": 0.
|
| 701 |
-
"learning_rate":
|
| 702 |
-
"loss": 0.
|
| 703 |
"step": 490
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"epoch": 0.8975521305530372,
|
| 707 |
-
"grad_norm": 0.
|
| 708 |
-
"learning_rate":
|
| 709 |
-
"loss": 0.
|
| 710 |
"step": 495
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 0.9066183136899365,
|
| 714 |
-
"grad_norm": 0.
|
| 715 |
-
"learning_rate":
|
| 716 |
-
"loss": 0.
|
| 717 |
"step": 500
|
| 718 |
}
|
| 719 |
],
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.001813236627379873,
|
| 14 |
+
"grad_norm": 0.330735981464386,
|
| 15 |
"learning_rate": 0.0,
|
| 16 |
+
"loss": 1.1455,
|
| 17 |
"step": 1
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.009066183136899365,
|
| 21 |
+
"grad_norm": 0.2740427255630493,
|
| 22 |
+
"learning_rate": 1e-05,
|
| 23 |
+
"loss": 1.1478,
|
| 24 |
"step": 5
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.01813236627379873,
|
| 28 |
+
"grad_norm": 0.2780371308326721,
|
| 29 |
+
"learning_rate": 2.25e-05,
|
| 30 |
+
"loss": 1.1416,
|
| 31 |
"step": 10
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.027198549410698096,
|
| 35 |
+
"grad_norm": 0.2090819627046585,
|
| 36 |
+
"learning_rate": 3.5e-05,
|
| 37 |
+
"loss": 1.0763,
|
| 38 |
"step": 15
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.03626473254759746,
|
| 42 |
+
"grad_norm": 0.12894907593727112,
|
| 43 |
+
"learning_rate": 4.75e-05,
|
| 44 |
+
"loss": 0.9366,
|
| 45 |
"step": 20
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.04533091568449683,
|
| 49 |
+
"grad_norm": 0.12664766609668732,
|
| 50 |
+
"learning_rate": 4.9624060150375936e-05,
|
| 51 |
+
"loss": 0.8368,
|
| 52 |
"step": 25
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 0.05439709882139619,
|
| 56 |
+
"grad_norm": 0.10315773636102676,
|
| 57 |
+
"learning_rate": 4.915413533834587e-05,
|
| 58 |
+
"loss": 0.7488,
|
| 59 |
"step": 30
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 0.06346328195829556,
|
| 63 |
+
"grad_norm": 0.08049902319908142,
|
| 64 |
+
"learning_rate": 4.868421052631579e-05,
|
| 65 |
+
"loss": 0.6929,
|
| 66 |
"step": 35
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 0.07252946509519492,
|
| 70 |
+
"grad_norm": 0.06981856375932693,
|
| 71 |
+
"learning_rate": 4.8214285714285716e-05,
|
| 72 |
+
"loss": 0.6419,
|
| 73 |
"step": 40
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 0.0815956482320943,
|
| 77 |
+
"grad_norm": 0.05855804681777954,
|
| 78 |
+
"learning_rate": 4.774436090225564e-05,
|
| 79 |
+
"loss": 0.6141,
|
| 80 |
"step": 45
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.09066183136899365,
|
| 84 |
+
"grad_norm": 0.05399257689714432,
|
| 85 |
+
"learning_rate": 4.7274436090225564e-05,
|
| 86 |
+
"loss": 0.59,
|
| 87 |
"step": 50
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.09972801450589303,
|
| 91 |
+
"grad_norm": 0.05272204801440239,
|
| 92 |
+
"learning_rate": 4.680451127819549e-05,
|
| 93 |
+
"loss": 0.5718,
|
| 94 |
"step": 55
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 0.10879419764279238,
|
| 98 |
+
"grad_norm": 0.050426185131073,
|
| 99 |
+
"learning_rate": 4.633458646616541e-05,
|
| 100 |
+
"loss": 0.5574,
|
| 101 |
"step": 60
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 0.11786038077969176,
|
| 105 |
+
"grad_norm": 0.045167580246925354,
|
| 106 |
+
"learning_rate": 4.586466165413534e-05,
|
| 107 |
+
"loss": 0.5399,
|
| 108 |
"step": 65
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 0.12692656391659113,
|
| 112 |
+
"grad_norm": 0.04907410964369774,
|
| 113 |
+
"learning_rate": 4.539473684210527e-05,
|
| 114 |
+
"loss": 0.5335,
|
| 115 |
"step": 70
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 0.1359927470534905,
|
| 119 |
+
"grad_norm": 0.04511063173413277,
|
| 120 |
+
"learning_rate": 4.492481203007519e-05,
|
| 121 |
+
"loss": 0.5231,
|
| 122 |
"step": 75
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.14505893019038985,
|
| 126 |
+
"grad_norm": 0.03908126801252365,
|
| 127 |
+
"learning_rate": 4.4454887218045117e-05,
|
| 128 |
+
"loss": 0.5162,
|
| 129 |
"step": 80
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.1541251133272892,
|
| 133 |
+
"grad_norm": 0.043064184486866,
|
| 134 |
+
"learning_rate": 4.398496240601504e-05,
|
| 135 |
+
"loss": 0.5085,
|
| 136 |
"step": 85
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.1631912964641886,
|
| 140 |
+
"grad_norm": 0.04580013453960419,
|
| 141 |
+
"learning_rate": 4.3515037593984965e-05,
|
| 142 |
+
"loss": 0.5036,
|
| 143 |
"step": 90
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 0.17225747960108795,
|
| 147 |
+
"grad_norm": 0.05117643624544144,
|
| 148 |
+
"learning_rate": 4.304511278195489e-05,
|
| 149 |
+
"loss": 0.501,
|
| 150 |
"step": 95
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 0.1813236627379873,
|
| 154 |
+
"grad_norm": 0.04914732649922371,
|
| 155 |
+
"learning_rate": 4.2575187969924814e-05,
|
| 156 |
+
"loss": 0.5035,
|
| 157 |
"step": 100
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 0.19038984587488667,
|
| 161 |
+
"grad_norm": 0.05879421904683113,
|
| 162 |
+
"learning_rate": 4.210526315789474e-05,
|
| 163 |
+
"loss": 0.4838,
|
| 164 |
"step": 105
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.19945602901178605,
|
| 168 |
+
"grad_norm": 0.05130897834897041,
|
| 169 |
+
"learning_rate": 4.163533834586467e-05,
|
| 170 |
+
"loss": 0.4835,
|
| 171 |
"step": 110
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.2085222121486854,
|
| 175 |
+
"grad_norm": 0.05152720585465431,
|
| 176 |
+
"learning_rate": 4.116541353383459e-05,
|
| 177 |
+
"loss": 0.4832,
|
| 178 |
"step": 115
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 0.21758839528558477,
|
| 182 |
+
"grad_norm": 0.0526948943734169,
|
| 183 |
+
"learning_rate": 4.069548872180452e-05,
|
| 184 |
+
"loss": 0.4811,
|
| 185 |
"step": 120
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 0.22665457842248413,
|
| 189 |
+
"grad_norm": 0.05324966460466385,
|
| 190 |
+
"learning_rate": 4.022556390977444e-05,
|
| 191 |
+
"loss": 0.4776,
|
| 192 |
"step": 125
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 0.2357207615593835,
|
| 196 |
+
"grad_norm": 0.05410118028521538,
|
| 197 |
+
"learning_rate": 3.9755639097744366e-05,
|
| 198 |
+
"loss": 0.4733,
|
| 199 |
"step": 130
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.24478694469628287,
|
| 203 |
+
"grad_norm": 0.058167893439531326,
|
| 204 |
+
"learning_rate": 3.928571428571429e-05,
|
| 205 |
+
"loss": 0.4703,
|
| 206 |
"step": 135
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.25385312783318226,
|
| 210 |
+
"grad_norm": 0.056478351354599,
|
| 211 |
+
"learning_rate": 3.8815789473684214e-05,
|
| 212 |
+
"loss": 0.4657,
|
| 213 |
"step": 140
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.2629193109700816,
|
| 217 |
+
"grad_norm": 0.07135945558547974,
|
| 218 |
+
"learning_rate": 3.834586466165413e-05,
|
| 219 |
+
"loss": 0.4669,
|
| 220 |
"step": 145
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 0.271985494106981,
|
| 224 |
+
"grad_norm": 0.07001485675573349,
|
| 225 |
+
"learning_rate": 3.787593984962406e-05,
|
| 226 |
+
"loss": 0.4645,
|
| 227 |
"step": 150
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 0.2810516772438803,
|
| 231 |
+
"grad_norm": 0.058485981076955795,
|
| 232 |
+
"learning_rate": 3.740601503759399e-05,
|
| 233 |
+
"loss": 0.4542,
|
| 234 |
"step": 155
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.2901178603807797,
|
| 238 |
+
"grad_norm": 0.05596611276268959,
|
| 239 |
+
"learning_rate": 3.693609022556391e-05,
|
| 240 |
+
"loss": 0.4586,
|
| 241 |
"step": 160
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 0.2991840435176791,
|
| 245 |
+
"grad_norm": 0.06255079060792923,
|
| 246 |
+
"learning_rate": 3.6466165413533835e-05,
|
| 247 |
+
"loss": 0.4582,
|
| 248 |
"step": 165
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 0.3082502266545784,
|
| 252 |
+
"grad_norm": 0.0630059689283371,
|
| 253 |
+
"learning_rate": 3.599624060150376e-05,
|
| 254 |
+
"loss": 0.4504,
|
| 255 |
"step": 170
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.3173164097914778,
|
| 259 |
+
"grad_norm": 0.0674496740102768,
|
| 260 |
+
"learning_rate": 3.5526315789473684e-05,
|
| 261 |
+
"loss": 0.4508,
|
| 262 |
"step": 175
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 0.3263825929283772,
|
| 266 |
+
"grad_norm": 0.06430725008249283,
|
| 267 |
+
"learning_rate": 3.505639097744361e-05,
|
| 268 |
+
"loss": 0.4506,
|
| 269 |
"step": 180
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 0.3354487760652765,
|
| 273 |
+
"grad_norm": 0.07139859348535538,
|
| 274 |
+
"learning_rate": 3.458646616541353e-05,
|
| 275 |
+
"loss": 0.4495,
|
| 276 |
"step": 185
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 0.3445149592021759,
|
| 280 |
+
"grad_norm": 0.08027750253677368,
|
| 281 |
+
"learning_rate": 3.411654135338346e-05,
|
| 282 |
+
"loss": 0.4438,
|
| 283 |
"step": 190
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 0.3535811423390752,
|
| 287 |
+
"grad_norm": 0.07266832143068314,
|
| 288 |
+
"learning_rate": 3.364661654135339e-05,
|
| 289 |
+
"loss": 0.4403,
|
| 290 |
"step": 195
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 0.3626473254759746,
|
| 294 |
+
"grad_norm": 0.0780920684337616,
|
| 295 |
+
"learning_rate": 3.317669172932331e-05,
|
| 296 |
+
"loss": 0.4441,
|
| 297 |
"step": 200
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.371713508612874,
|
| 301 |
+
"grad_norm": 0.07039614021778107,
|
| 302 |
+
"learning_rate": 3.2706766917293236e-05,
|
| 303 |
+
"loss": 0.4362,
|
| 304 |
"step": 205
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 0.38077969174977333,
|
| 308 |
+
"grad_norm": 0.06520690023899078,
|
| 309 |
+
"learning_rate": 3.223684210526316e-05,
|
| 310 |
+
"loss": 0.4391,
|
| 311 |
"step": 210
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 0.3898458748866727,
|
| 315 |
+
"grad_norm": 0.08516666293144226,
|
| 316 |
+
"learning_rate": 3.1766917293233085e-05,
|
| 317 |
+
"loss": 0.4353,
|
| 318 |
"step": 215
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 0.3989120580235721,
|
| 322 |
+
"grad_norm": 0.07929079979658127,
|
| 323 |
+
"learning_rate": 3.129699248120301e-05,
|
| 324 |
+
"loss": 0.4355,
|
| 325 |
"step": 220
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.40797824116047143,
|
| 329 |
+
"grad_norm": 0.07418406754732132,
|
| 330 |
+
"learning_rate": 3.082706766917293e-05,
|
| 331 |
+
"loss": 0.4323,
|
| 332 |
"step": 225
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 0.4170444242973708,
|
| 336 |
+
"grad_norm": 0.07323618978261948,
|
| 337 |
+
"learning_rate": 3.0357142857142857e-05,
|
| 338 |
+
"loss": 0.4346,
|
| 339 |
"step": 230
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.42611060743427015,
|
| 343 |
+
"grad_norm": 0.0705682560801506,
|
| 344 |
+
"learning_rate": 2.9887218045112785e-05,
|
| 345 |
+
"loss": 0.4331,
|
| 346 |
"step": 235
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 0.43517679057116954,
|
| 350 |
+
"grad_norm": 0.0874267965555191,
|
| 351 |
+
"learning_rate": 2.941729323308271e-05,
|
| 352 |
+
"loss": 0.4388,
|
| 353 |
"step": 240
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 0.4442429737080689,
|
| 357 |
+
"grad_norm": 0.07446466386318207,
|
| 358 |
+
"learning_rate": 2.8947368421052634e-05,
|
| 359 |
+
"loss": 0.4349,
|
| 360 |
"step": 245
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 0.45330915684496825,
|
| 364 |
+
"grad_norm": 0.08319786190986633,
|
| 365 |
+
"learning_rate": 2.8477443609022558e-05,
|
| 366 |
+
"loss": 0.4315,
|
| 367 |
"step": 250
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 0.46237533998186764,
|
| 371 |
+
"grad_norm": 0.08069171756505966,
|
| 372 |
+
"learning_rate": 2.8007518796992482e-05,
|
| 373 |
+
"loss": 0.4348,
|
| 374 |
"step": 255
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 0.471441523118767,
|
| 378 |
+
"grad_norm": 0.07842331379652023,
|
| 379 |
+
"learning_rate": 2.7537593984962406e-05,
|
| 380 |
+
"loss": 0.4328,
|
| 381 |
"step": 260
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 0.48050770625566636,
|
| 385 |
+
"grad_norm": 0.07334455102682114,
|
| 386 |
+
"learning_rate": 2.706766917293233e-05,
|
| 387 |
+
"loss": 0.4247,
|
| 388 |
"step": 265
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 0.48957388939256574,
|
| 392 |
+
"grad_norm": 0.06868273764848709,
|
| 393 |
+
"learning_rate": 2.6597744360902255e-05,
|
| 394 |
+
"loss": 0.4304,
|
| 395 |
"step": 270
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 0.4986400725294651,
|
| 399 |
+
"grad_norm": 0.08919775485992432,
|
| 400 |
+
"learning_rate": 2.6127819548872186e-05,
|
| 401 |
+
"loss": 0.4237,
|
| 402 |
"step": 275
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 0.5077062556663645,
|
| 406 |
+
"grad_norm": 0.09395933896303177,
|
| 407 |
+
"learning_rate": 2.565789473684211e-05,
|
| 408 |
+
"loss": 0.4293,
|
| 409 |
"step": 280
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 0.5167724388032638,
|
| 413 |
+
"grad_norm": 0.07169164717197418,
|
| 414 |
+
"learning_rate": 2.5187969924812034e-05,
|
| 415 |
+
"loss": 0.4226,
|
| 416 |
"step": 285
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 0.5258386219401632,
|
| 420 |
+
"grad_norm": 0.07571956515312195,
|
| 421 |
+
"learning_rate": 2.471804511278196e-05,
|
| 422 |
+
"loss": 0.424,
|
| 423 |
"step": 290
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 0.5349048050770625,
|
| 427 |
+
"grad_norm": 0.07626249641180038,
|
| 428 |
+
"learning_rate": 2.424812030075188e-05,
|
| 429 |
+
"loss": 0.4238,
|
| 430 |
"step": 295
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 0.543970988213962,
|
| 434 |
+
"grad_norm": 0.08185533434152603,
|
| 435 |
+
"learning_rate": 2.3778195488721804e-05,
|
| 436 |
+
"loss": 0.4238,
|
| 437 |
"step": 300
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 0.5530371713508613,
|
| 441 |
+
"grad_norm": 0.07463429868221283,
|
| 442 |
+
"learning_rate": 2.3308270676691728e-05,
|
| 443 |
+
"loss": 0.4163,
|
| 444 |
"step": 305
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 0.5621033544877606,
|
| 448 |
+
"grad_norm": 0.0868794396519661,
|
| 449 |
+
"learning_rate": 2.2838345864661655e-05,
|
| 450 |
+
"loss": 0.4206,
|
| 451 |
"step": 310
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 0.57116953762466,
|
| 455 |
+
"grad_norm": 0.09378495812416077,
|
| 456 |
+
"learning_rate": 2.236842105263158e-05,
|
| 457 |
+
"loss": 0.4239,
|
| 458 |
"step": 315
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 0.5802357207615594,
|
| 462 |
+
"grad_norm": 0.07692472636699677,
|
| 463 |
+
"learning_rate": 2.1898496240601504e-05,
|
| 464 |
+
"loss": 0.4173,
|
| 465 |
"step": 320
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 0.5893019038984587,
|
| 469 |
+
"grad_norm": 0.08751624077558517,
|
| 470 |
+
"learning_rate": 2.1428571428571428e-05,
|
| 471 |
+
"loss": 0.4228,
|
| 472 |
"step": 325
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 0.5983680870353582,
|
| 476 |
+
"grad_norm": 0.0897555872797966,
|
| 477 |
+
"learning_rate": 2.0958646616541356e-05,
|
| 478 |
+
"loss": 0.4163,
|
| 479 |
"step": 330
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 0.6074342701722575,
|
| 483 |
+
"grad_norm": 0.08029606938362122,
|
| 484 |
+
"learning_rate": 2.048872180451128e-05,
|
| 485 |
+
"loss": 0.4254,
|
| 486 |
"step": 335
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 0.6165004533091568,
|
| 490 |
+
"grad_norm": 0.09147002547979355,
|
| 491 |
+
"learning_rate": 2.0018796992481204e-05,
|
| 492 |
+
"loss": 0.422,
|
| 493 |
"step": 340
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 0.6255666364460563,
|
| 497 |
+
"grad_norm": 0.08232133090496063,
|
| 498 |
+
"learning_rate": 1.954887218045113e-05,
|
| 499 |
+
"loss": 0.4167,
|
| 500 |
"step": 345
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 0.6346328195829556,
|
| 504 |
+
"grad_norm": 0.08974982798099518,
|
| 505 |
+
"learning_rate": 1.9078947368421056e-05,
|
| 506 |
+
"loss": 0.418,
|
| 507 |
"step": 350
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 0.6436990027198549,
|
| 511 |
+
"grad_norm": 0.07837100327014923,
|
| 512 |
+
"learning_rate": 1.860902255639098e-05,
|
| 513 |
+
"loss": 0.4214,
|
| 514 |
"step": 355
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 0.6527651858567544,
|
| 518 |
+
"grad_norm": 0.08045557886362076,
|
| 519 |
+
"learning_rate": 1.81390977443609e-05,
|
| 520 |
+
"loss": 0.4191,
|
| 521 |
"step": 360
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 0.6618313689936537,
|
| 525 |
+
"grad_norm": 0.08100251108407974,
|
| 526 |
+
"learning_rate": 1.7669172932330825e-05,
|
| 527 |
+
"loss": 0.4175,
|
| 528 |
"step": 365
|
| 529 |
},
|
| 530 |
{
|
| 531 |
"epoch": 0.670897552130553,
|
| 532 |
+
"grad_norm": 0.10134581476449966,
|
| 533 |
+
"learning_rate": 1.7199248120300753e-05,
|
| 534 |
+
"loss": 0.4128,
|
| 535 |
"step": 370
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"epoch": 0.6799637352674524,
|
| 539 |
+
"grad_norm": 0.08495834469795227,
|
| 540 |
+
"learning_rate": 1.6729323308270677e-05,
|
| 541 |
+
"loss": 0.4164,
|
| 542 |
"step": 375
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 0.6890299184043518,
|
| 546 |
+
"grad_norm": 0.088101826608181,
|
| 547 |
+
"learning_rate": 1.62593984962406e-05,
|
| 548 |
+
"loss": 0.4218,
|
| 549 |
"step": 380
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.6980961015412511,
|
| 553 |
+
"grad_norm": 0.08768700808286667,
|
| 554 |
+
"learning_rate": 1.5789473684210526e-05,
|
| 555 |
+
"loss": 0.4145,
|
| 556 |
"step": 385
|
| 557 |
},
|
| 558 |
{
|
| 559 |
"epoch": 0.7071622846781505,
|
| 560 |
+
"grad_norm": 0.08453679084777832,
|
| 561 |
+
"learning_rate": 1.5319548872180453e-05,
|
| 562 |
+
"loss": 0.419,
|
| 563 |
"step": 390
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 0.7162284678150499,
|
| 567 |
+
"grad_norm": 0.08174845576286316,
|
| 568 |
+
"learning_rate": 1.4849624060150378e-05,
|
| 569 |
+
"loss": 0.4171,
|
| 570 |
"step": 395
|
| 571 |
},
|
| 572 |
{
|
| 573 |
"epoch": 0.7252946509519492,
|
| 574 |
+
"grad_norm": 0.08191751688718796,
|
| 575 |
+
"learning_rate": 1.4379699248120302e-05,
|
| 576 |
+
"loss": 0.419,
|
| 577 |
"step": 400
|
| 578 |
},
|
| 579 |
{
|
| 580 |
"epoch": 0.7343608340888486,
|
| 581 |
+
"grad_norm": 0.0835120901465416,
|
| 582 |
+
"learning_rate": 1.3909774436090226e-05,
|
| 583 |
+
"loss": 0.4137,
|
| 584 |
"step": 405
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 0.743427017225748,
|
| 588 |
+
"grad_norm": 0.0808102935552597,
|
| 589 |
+
"learning_rate": 1.3439849624060152e-05,
|
| 590 |
+
"loss": 0.4103,
|
| 591 |
"step": 410
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 0.7524932003626473,
|
| 595 |
+
"grad_norm": 0.07358790189027786,
|
| 596 |
+
"learning_rate": 1.2969924812030076e-05,
|
| 597 |
+
"loss": 0.4136,
|
| 598 |
"step": 415
|
| 599 |
},
|
| 600 |
{
|
| 601 |
"epoch": 0.7615593834995467,
|
| 602 |
+
"grad_norm": 0.07726311683654785,
|
| 603 |
+
"learning_rate": 1.25e-05,
|
| 604 |
+
"loss": 0.4176,
|
| 605 |
"step": 420
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"epoch": 0.7706255666364461,
|
| 609 |
+
"grad_norm": 0.08744051307439804,
|
| 610 |
+
"learning_rate": 1.2030075187969925e-05,
|
| 611 |
+
"loss": 0.411,
|
| 612 |
"step": 425
|
| 613 |
},
|
| 614 |
{
|
| 615 |
"epoch": 0.7796917497733454,
|
| 616 |
+
"grad_norm": 0.07695311307907104,
|
| 617 |
+
"learning_rate": 1.156015037593985e-05,
|
| 618 |
+
"loss": 0.4107,
|
| 619 |
"step": 430
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"epoch": 0.7887579329102448,
|
| 623 |
+
"grad_norm": 0.07769734412431717,
|
| 624 |
+
"learning_rate": 1.1090225563909775e-05,
|
| 625 |
+
"loss": 0.4147,
|
| 626 |
"step": 435
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 0.7978241160471442,
|
| 630 |
+
"grad_norm": 0.0807860940694809,
|
| 631 |
+
"learning_rate": 1.06203007518797e-05,
|
| 632 |
+
"loss": 0.4153,
|
| 633 |
"step": 440
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 0.8068902991840435,
|
| 637 |
+
"grad_norm": 0.0794687420129776,
|
| 638 |
+
"learning_rate": 1.0150375939849624e-05,
|
| 639 |
+
"loss": 0.4089,
|
| 640 |
"step": 445
|
| 641 |
},
|
| 642 |
{
|
| 643 |
"epoch": 0.8159564823209429,
|
| 644 |
+
"grad_norm": 0.09177932888269424,
|
| 645 |
+
"learning_rate": 9.68045112781955e-06,
|
| 646 |
+
"loss": 0.4135,
|
| 647 |
"step": 450
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"epoch": 0.8250226654578422,
|
| 651 |
+
"grad_norm": 0.08271961659193039,
|
| 652 |
+
"learning_rate": 9.210526315789474e-06,
|
| 653 |
+
"loss": 0.415,
|
| 654 |
"step": 455
|
| 655 |
},
|
| 656 |
{
|
| 657 |
"epoch": 0.8340888485947416,
|
| 658 |
+
"grad_norm": 0.0818314403295517,
|
| 659 |
+
"learning_rate": 8.7406015037594e-06,
|
| 660 |
+
"loss": 0.4107,
|
| 661 |
"step": 460
|
| 662 |
},
|
| 663 |
{
|
| 664 |
"epoch": 0.843155031731641,
|
| 665 |
+
"grad_norm": 0.09098558127880096,
|
| 666 |
+
"learning_rate": 8.270676691729324e-06,
|
| 667 |
+
"loss": 0.4076,
|
| 668 |
"step": 465
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 0.8522212148685403,
|
| 672 |
+
"grad_norm": 0.07806940376758575,
|
| 673 |
+
"learning_rate": 7.800751879699248e-06,
|
| 674 |
+
"loss": 0.409,
|
| 675 |
"step": 470
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 0.8612873980054397,
|
| 679 |
+
"grad_norm": 0.08273617923259735,
|
| 680 |
+
"learning_rate": 7.330827067669173e-06,
|
| 681 |
+
"loss": 0.4081,
|
| 682 |
"step": 475
|
| 683 |
},
|
| 684 |
{
|
| 685 |
"epoch": 0.8703535811423391,
|
| 686 |
+
"grad_norm": 0.08064981549978256,
|
| 687 |
+
"learning_rate": 6.860902255639098e-06,
|
| 688 |
+
"loss": 0.4144,
|
| 689 |
"step": 480
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"epoch": 0.8794197642792384,
|
| 693 |
+
"grad_norm": 0.07812593877315521,
|
| 694 |
+
"learning_rate": 6.3909774436090225e-06,
|
| 695 |
+
"loss": 0.4148,
|
| 696 |
"step": 485
|
| 697 |
},
|
| 698 |
{
|
| 699 |
"epoch": 0.8884859474161378,
|
| 700 |
+
"grad_norm": 0.083773672580719,
|
| 701 |
+
"learning_rate": 5.921052631578948e-06,
|
| 702 |
+
"loss": 0.4185,
|
| 703 |
"step": 490
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"epoch": 0.8975521305530372,
|
| 707 |
+
"grad_norm": 0.07797183096408844,
|
| 708 |
+
"learning_rate": 5.451127819548872e-06,
|
| 709 |
+
"loss": 0.4106,
|
| 710 |
"step": 495
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 0.9066183136899365,
|
| 714 |
+
"grad_norm": 0.07622473686933517,
|
| 715 |
+
"learning_rate": 4.981203007518797e-06,
|
| 716 |
+
"loss": 0.4119,
|
| 717 |
"step": 500
|
| 718 |
}
|
| 719 |
],
|
math-0.3/q/cold-start/checkpoint-500/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 6289
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:417f069ca3f7ca2d3c0eb8d0203990078a027cfcb3923306206e6eaaa2b7e350
|
| 3 |
size 6289
|
math-0.3/q/cold-start/checkpoint-552/adapter_config.json
CHANGED
|
@@ -33,12 +33,12 @@
|
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
| 36 |
-
"up_proj",
|
| 37 |
-
"down_proj",
|
| 38 |
"o_proj",
|
| 39 |
-
"gate_proj",
|
| 40 |
-
"k_proj",
|
| 41 |
"v_proj",
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
"q_proj"
|
| 43 |
],
|
| 44 |
"target_parameters": null,
|
|
|
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
|
|
|
|
|
|
| 36 |
"o_proj",
|
|
|
|
|
|
|
| 37 |
"v_proj",
|
| 38 |
+
"down_proj",
|
| 39 |
+
"up_proj",
|
| 40 |
+
"k_proj",
|
| 41 |
+
"gate_proj",
|
| 42 |
"q_proj"
|
| 43 |
],
|
| 44 |
"target_parameters": null,
|
math-0.3/q/cold-start/checkpoint-552/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 645975704
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:589167f278372c18db431c80611f980e909def4a36684858409210419c692f2c
|
| 3 |
size 645975704
|
math-0.3/q/cold-start/checkpoint-552/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1292182139
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e94339fc8c3cfdcdc1342951f22fb7da5639fc0525dd0c235666c25ef8d079be
|
| 3 |
size 1292182139
|
math-0.3/q/cold-start/checkpoint-552/trainer_state.json
CHANGED
|
@@ -11,779 +11,779 @@
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.001813236627379873,
|
| 14 |
-
"grad_norm": 0.
|
| 15 |
"learning_rate": 0.0,
|
| 16 |
-
"loss": 1.
|
| 17 |
"step": 1
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.009066183136899365,
|
| 21 |
-
"grad_norm": 0.
|
| 22 |
-
"learning_rate":
|
| 23 |
-
"loss": 1.
|
| 24 |
"step": 5
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.01813236627379873,
|
| 28 |
-
"grad_norm": 0.
|
| 29 |
-
"learning_rate":
|
| 30 |
-
"loss": 1.
|
| 31 |
"step": 10
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.027198549410698096,
|
| 35 |
-
"grad_norm": 0.
|
| 36 |
-
"learning_rate":
|
| 37 |
-
"loss": 1.
|
| 38 |
"step": 15
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.03626473254759746,
|
| 42 |
-
"grad_norm": 0.
|
| 43 |
-
"learning_rate":
|
| 44 |
-
"loss":
|
| 45 |
"step": 20
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.04533091568449683,
|
| 49 |
-
"grad_norm": 0.
|
| 50 |
-
"learning_rate":
|
| 51 |
-
"loss":
|
| 52 |
"step": 25
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 0.05439709882139619,
|
| 56 |
-
"grad_norm": 0.
|
| 57 |
-
"learning_rate":
|
| 58 |
-
"loss":
|
| 59 |
"step": 30
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 0.06346328195829556,
|
| 63 |
-
"grad_norm": 0.
|
| 64 |
-
"learning_rate":
|
| 65 |
-
"loss": 0.
|
| 66 |
"step": 35
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 0.07252946509519492,
|
| 70 |
-
"grad_norm": 0.
|
| 71 |
-
"learning_rate":
|
| 72 |
-
"loss": 0.
|
| 73 |
"step": 40
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 0.0815956482320943,
|
| 77 |
-
"grad_norm": 0.
|
| 78 |
-
"learning_rate":
|
| 79 |
-
"loss": 0.
|
| 80 |
"step": 45
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.09066183136899365,
|
| 84 |
-
"grad_norm": 0.
|
| 85 |
-
"learning_rate":
|
| 86 |
-
"loss": 0.
|
| 87 |
"step": 50
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.09972801450589303,
|
| 91 |
-
"grad_norm": 0.
|
| 92 |
-
"learning_rate":
|
| 93 |
-
"loss": 0.
|
| 94 |
"step": 55
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 0.10879419764279238,
|
| 98 |
-
"grad_norm": 0.
|
| 99 |
-
"learning_rate":
|
| 100 |
-
"loss": 0.
|
| 101 |
"step": 60
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 0.11786038077969176,
|
| 105 |
-
"grad_norm": 0.
|
| 106 |
-
"learning_rate":
|
| 107 |
-
"loss": 0.
|
| 108 |
"step": 65
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 0.12692656391659113,
|
| 112 |
-
"grad_norm": 0.
|
| 113 |
-
"learning_rate":
|
| 114 |
-
"loss": 0.
|
| 115 |
"step": 70
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 0.1359927470534905,
|
| 119 |
-
"grad_norm": 0.
|
| 120 |
-
"learning_rate":
|
| 121 |
-
"loss": 0.
|
| 122 |
"step": 75
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.14505893019038985,
|
| 126 |
-
"grad_norm": 0.
|
| 127 |
-
"learning_rate":
|
| 128 |
-
"loss": 0.
|
| 129 |
"step": 80
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.1541251133272892,
|
| 133 |
-
"grad_norm": 0.
|
| 134 |
-
"learning_rate": 4.
|
| 135 |
-
"loss": 0.
|
| 136 |
"step": 85
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.1631912964641886,
|
| 140 |
-
"grad_norm": 0.
|
| 141 |
-
"learning_rate": 4.
|
| 142 |
-
"loss": 0.
|
| 143 |
"step": 90
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 0.17225747960108795,
|
| 147 |
-
"grad_norm": 0.
|
| 148 |
-
"learning_rate": 4.
|
| 149 |
-
"loss": 0.
|
| 150 |
"step": 95
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 0.1813236627379873,
|
| 154 |
-
"grad_norm": 0.
|
| 155 |
-
"learning_rate": 4.
|
| 156 |
-
"loss": 0.
|
| 157 |
"step": 100
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 0.19038984587488667,
|
| 161 |
-
"grad_norm": 0.
|
| 162 |
-
"learning_rate": 4.
|
| 163 |
-
"loss": 0.
|
| 164 |
"step": 105
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.19945602901178605,
|
| 168 |
-
"grad_norm": 0.
|
| 169 |
-
"learning_rate": 4.
|
| 170 |
-
"loss": 0.
|
| 171 |
"step": 110
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.2085222121486854,
|
| 175 |
-
"grad_norm": 0.
|
| 176 |
-
"learning_rate": 4.
|
| 177 |
-
"loss": 0.
|
| 178 |
"step": 115
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 0.21758839528558477,
|
| 182 |
-
"grad_norm": 0.
|
| 183 |
-
"learning_rate": 4.
|
| 184 |
-
"loss": 0.
|
| 185 |
"step": 120
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 0.22665457842248413,
|
| 189 |
-
"grad_norm": 0.
|
| 190 |
-
"learning_rate": 4.
|
| 191 |
-
"loss": 0.
|
| 192 |
"step": 125
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 0.2357207615593835,
|
| 196 |
-
"grad_norm": 0.
|
| 197 |
-
"learning_rate":
|
| 198 |
-
"loss": 0.
|
| 199 |
"step": 130
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.24478694469628287,
|
| 203 |
-
"grad_norm": 0.
|
| 204 |
-
"learning_rate":
|
| 205 |
-
"loss": 0.
|
| 206 |
"step": 135
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.25385312783318226,
|
| 210 |
-
"grad_norm": 0.
|
| 211 |
-
"learning_rate":
|
| 212 |
-
"loss": 0.
|
| 213 |
"step": 140
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.2629193109700816,
|
| 217 |
-
"grad_norm": 0.
|
| 218 |
-
"learning_rate":
|
| 219 |
-
"loss": 0.
|
| 220 |
"step": 145
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 0.271985494106981,
|
| 224 |
-
"grad_norm": 0.
|
| 225 |
-
"learning_rate":
|
| 226 |
-
"loss": 0.
|
| 227 |
"step": 150
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 0.2810516772438803,
|
| 231 |
-
"grad_norm": 0.
|
| 232 |
-
"learning_rate":
|
| 233 |
-
"loss": 0.
|
| 234 |
"step": 155
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.2901178603807797,
|
| 238 |
-
"grad_norm": 0.
|
| 239 |
-
"learning_rate":
|
| 240 |
-
"loss": 0.
|
| 241 |
"step": 160
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 0.2991840435176791,
|
| 245 |
-
"grad_norm": 0.
|
| 246 |
-
"learning_rate":
|
| 247 |
-
"loss": 0.
|
| 248 |
"step": 165
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 0.3082502266545784,
|
| 252 |
-
"grad_norm": 0.
|
| 253 |
-
"learning_rate":
|
| 254 |
-
"loss": 0.
|
| 255 |
"step": 170
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.3173164097914778,
|
| 259 |
-
"grad_norm": 0.
|
| 260 |
-
"learning_rate":
|
| 261 |
-
"loss": 0.
|
| 262 |
"step": 175
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 0.3263825929283772,
|
| 266 |
-
"grad_norm": 0.
|
| 267 |
-
"learning_rate":
|
| 268 |
-
"loss": 0.
|
| 269 |
"step": 180
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 0.3354487760652765,
|
| 273 |
-
"grad_norm": 0.
|
| 274 |
-
"learning_rate":
|
| 275 |
-
"loss": 0.
|
| 276 |
"step": 185
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 0.3445149592021759,
|
| 280 |
-
"grad_norm": 0.
|
| 281 |
-
"learning_rate":
|
| 282 |
-
"loss": 0.
|
| 283 |
"step": 190
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 0.3535811423390752,
|
| 287 |
-
"grad_norm": 0.
|
| 288 |
-
"learning_rate": 3.
|
| 289 |
-
"loss": 0.
|
| 290 |
"step": 195
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 0.3626473254759746,
|
| 294 |
-
"grad_norm": 0.
|
| 295 |
-
"learning_rate": 3.
|
| 296 |
-
"loss": 0.
|
| 297 |
"step": 200
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.371713508612874,
|
| 301 |
-
"grad_norm": 0.
|
| 302 |
-
"learning_rate": 3.
|
| 303 |
-
"loss": 0.
|
| 304 |
"step": 205
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 0.38077969174977333,
|
| 308 |
-
"grad_norm": 0.
|
| 309 |
-
"learning_rate": 3.
|
| 310 |
-
"loss": 0.
|
| 311 |
"step": 210
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 0.3898458748866727,
|
| 315 |
-
"grad_norm": 0.
|
| 316 |
-
"learning_rate": 3.
|
| 317 |
-
"loss": 0.
|
| 318 |
"step": 215
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 0.3989120580235721,
|
| 322 |
-
"grad_norm": 0.
|
| 323 |
-
"learning_rate": 3.
|
| 324 |
-
"loss": 0.
|
| 325 |
"step": 220
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.40797824116047143,
|
| 329 |
-
"grad_norm": 0.
|
| 330 |
-
"learning_rate": 3.
|
| 331 |
-
"loss": 0.
|
| 332 |
"step": 225
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 0.4170444242973708,
|
| 336 |
-
"grad_norm": 0.
|
| 337 |
-
"learning_rate": 3.
|
| 338 |
-
"loss": 0.
|
| 339 |
"step": 230
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.42611060743427015,
|
| 343 |
-
"grad_norm": 0.
|
| 344 |
-
"learning_rate":
|
| 345 |
-
"loss": 0.
|
| 346 |
"step": 235
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 0.43517679057116954,
|
| 350 |
-
"grad_norm": 0.
|
| 351 |
-
"learning_rate":
|
| 352 |
-
"loss": 0.
|
| 353 |
"step": 240
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 0.4442429737080689,
|
| 357 |
-
"grad_norm": 0.
|
| 358 |
-
"learning_rate":
|
| 359 |
-
"loss": 0.
|
| 360 |
"step": 245
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 0.45330915684496825,
|
| 364 |
-
"grad_norm": 0.
|
| 365 |
-
"learning_rate":
|
| 366 |
-
"loss": 0.
|
| 367 |
"step": 250
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 0.46237533998186764,
|
| 371 |
-
"grad_norm": 0.
|
| 372 |
-
"learning_rate":
|
| 373 |
-
"loss": 0.
|
| 374 |
"step": 255
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 0.471441523118767,
|
| 378 |
-
"grad_norm": 0.
|
| 379 |
-
"learning_rate":
|
| 380 |
-
"loss": 0.
|
| 381 |
"step": 260
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 0.48050770625566636,
|
| 385 |
-
"grad_norm": 0.
|
| 386 |
-
"learning_rate":
|
| 387 |
-
"loss": 0.
|
| 388 |
"step": 265
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 0.48957388939256574,
|
| 392 |
-
"grad_norm": 0.
|
| 393 |
-
"learning_rate":
|
| 394 |
-
"loss": 0.
|
| 395 |
"step": 270
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 0.4986400725294651,
|
| 399 |
-
"grad_norm": 0.
|
| 400 |
-
"learning_rate":
|
| 401 |
-
"loss": 0.
|
| 402 |
"step": 275
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 0.5077062556663645,
|
| 406 |
-
"grad_norm": 0.
|
| 407 |
-
"learning_rate":
|
| 408 |
-
"loss": 0.
|
| 409 |
"step": 280
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 0.5167724388032638,
|
| 413 |
-
"grad_norm": 0.
|
| 414 |
-
"learning_rate": 2.
|
| 415 |
-
"loss": 0.
|
| 416 |
"step": 285
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 0.5258386219401632,
|
| 420 |
-
"grad_norm": 0.
|
| 421 |
-
"learning_rate": 2.
|
| 422 |
-
"loss": 0.
|
| 423 |
"step": 290
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 0.5349048050770625,
|
| 427 |
-
"grad_norm": 0.
|
| 428 |
-
"learning_rate": 2.
|
| 429 |
-
"loss": 0.
|
| 430 |
"step": 295
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 0.543970988213962,
|
| 434 |
-
"grad_norm": 0.
|
| 435 |
-
"learning_rate": 2.
|
| 436 |
-
"loss": 0.
|
| 437 |
"step": 300
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 0.5530371713508613,
|
| 441 |
-
"grad_norm": 0.
|
| 442 |
-
"learning_rate": 2.
|
| 443 |
-
"loss": 0.
|
| 444 |
"step": 305
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 0.5621033544877606,
|
| 448 |
-
"grad_norm": 0.
|
| 449 |
-
"learning_rate": 2.
|
| 450 |
-
"loss": 0.
|
| 451 |
"step": 310
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 0.57116953762466,
|
| 455 |
-
"grad_norm": 0.
|
| 456 |
-
"learning_rate": 2.
|
| 457 |
-
"loss": 0.
|
| 458 |
"step": 315
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 0.5802357207615594,
|
| 462 |
-
"grad_norm": 0.
|
| 463 |
-
"learning_rate": 2.
|
| 464 |
-
"loss": 0.
|
| 465 |
"step": 320
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 0.5893019038984587,
|
| 469 |
-
"grad_norm": 0.
|
| 470 |
-
"learning_rate": 2.
|
| 471 |
-
"loss": 0.
|
| 472 |
"step": 325
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 0.5983680870353582,
|
| 476 |
-
"grad_norm": 0.
|
| 477 |
-
"learning_rate": 2.
|
| 478 |
-
"loss": 0.
|
| 479 |
"step": 330
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 0.6074342701722575,
|
| 483 |
-
"grad_norm": 0.
|
| 484 |
-
"learning_rate": 2.
|
| 485 |
-
"loss": 0.
|
| 486 |
"step": 335
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 0.6165004533091568,
|
| 490 |
-
"grad_norm": 0.
|
| 491 |
-
"learning_rate": 2.
|
| 492 |
-
"loss": 0.
|
| 493 |
"step": 340
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 0.6255666364460563,
|
| 497 |
-
"grad_norm": 0.
|
| 498 |
-
"learning_rate":
|
| 499 |
-
"loss": 0.
|
| 500 |
"step": 345
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 0.6346328195829556,
|
| 504 |
-
"grad_norm": 0.
|
| 505 |
-
"learning_rate":
|
| 506 |
-
"loss": 0.
|
| 507 |
"step": 350
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 0.6436990027198549,
|
| 511 |
-
"grad_norm": 0.
|
| 512 |
-
"learning_rate":
|
| 513 |
-
"loss": 0.
|
| 514 |
"step": 355
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 0.6527651858567544,
|
| 518 |
-
"grad_norm": 0.
|
| 519 |
-
"learning_rate":
|
| 520 |
-
"loss": 0.
|
| 521 |
"step": 360
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 0.6618313689936537,
|
| 525 |
-
"grad_norm": 0.
|
| 526 |
-
"learning_rate":
|
| 527 |
-
"loss": 0.
|
| 528 |
"step": 365
|
| 529 |
},
|
| 530 |
{
|
| 531 |
"epoch": 0.670897552130553,
|
| 532 |
-
"grad_norm": 0.
|
| 533 |
-
"learning_rate":
|
| 534 |
-
"loss": 0.
|
| 535 |
"step": 370
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"epoch": 0.6799637352674524,
|
| 539 |
-
"grad_norm": 0.
|
| 540 |
-
"learning_rate": 1.
|
| 541 |
-
"loss": 0.
|
| 542 |
"step": 375
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 0.6890299184043518,
|
| 546 |
-
"grad_norm": 0.
|
| 547 |
-
"learning_rate": 1.
|
| 548 |
-
"loss": 0.
|
| 549 |
"step": 380
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.6980961015412511,
|
| 553 |
-
"grad_norm": 0.
|
| 554 |
-
"learning_rate": 1.
|
| 555 |
-
"loss": 0.
|
| 556 |
"step": 385
|
| 557 |
},
|
| 558 |
{
|
| 559 |
"epoch": 0.7071622846781505,
|
| 560 |
-
"grad_norm": 0.
|
| 561 |
-
"learning_rate": 1.
|
| 562 |
-
"loss": 0.
|
| 563 |
"step": 390
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 0.7162284678150499,
|
| 567 |
-
"grad_norm": 0.
|
| 568 |
-
"learning_rate": 1.
|
| 569 |
-
"loss": 0.
|
| 570 |
"step": 395
|
| 571 |
},
|
| 572 |
{
|
| 573 |
"epoch": 0.7252946509519492,
|
| 574 |
-
"grad_norm": 0.
|
| 575 |
-
"learning_rate": 1.
|
| 576 |
-
"loss": 0.
|
| 577 |
"step": 400
|
| 578 |
},
|
| 579 |
{
|
| 580 |
"epoch": 0.7343608340888486,
|
| 581 |
-
"grad_norm": 0.
|
| 582 |
-
"learning_rate": 1.
|
| 583 |
-
"loss": 0.
|
| 584 |
"step": 405
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 0.743427017225748,
|
| 588 |
-
"grad_norm": 0.
|
| 589 |
-
"learning_rate": 1.
|
| 590 |
-
"loss": 0.
|
| 591 |
"step": 410
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 0.7524932003626473,
|
| 595 |
-
"grad_norm": 0.
|
| 596 |
-
"learning_rate": 1.
|
| 597 |
-
"loss": 0.
|
| 598 |
"step": 415
|
| 599 |
},
|
| 600 |
{
|
| 601 |
"epoch": 0.7615593834995467,
|
| 602 |
-
"grad_norm": 0.
|
| 603 |
-
"learning_rate": 1.
|
| 604 |
-
"loss": 0.
|
| 605 |
"step": 420
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"epoch": 0.7706255666364461,
|
| 609 |
-
"grad_norm": 0.
|
| 610 |
-
"learning_rate": 1.
|
| 611 |
-
"loss": 0.
|
| 612 |
"step": 425
|
| 613 |
},
|
| 614 |
{
|
| 615 |
"epoch": 0.7796917497733454,
|
| 616 |
-
"grad_norm": 0.
|
| 617 |
-
"learning_rate": 1.
|
| 618 |
-
"loss": 0.
|
| 619 |
"step": 430
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"epoch": 0.7887579329102448,
|
| 623 |
-
"grad_norm": 0.
|
| 624 |
-
"learning_rate": 1.
|
| 625 |
-
"loss": 0.
|
| 626 |
"step": 435
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 0.7978241160471442,
|
| 630 |
-
"grad_norm": 0.
|
| 631 |
-
"learning_rate": 1.
|
| 632 |
-
"loss": 0.
|
| 633 |
"step": 440
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 0.8068902991840435,
|
| 637 |
-
"grad_norm": 0.
|
| 638 |
-
"learning_rate": 1.
|
| 639 |
-
"loss": 0.
|
| 640 |
"step": 445
|
| 641 |
},
|
| 642 |
{
|
| 643 |
"epoch": 0.8159564823209429,
|
| 644 |
-
"grad_norm": 0.
|
| 645 |
-
"learning_rate":
|
| 646 |
-
"loss": 0.
|
| 647 |
"step": 450
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"epoch": 0.8250226654578422,
|
| 651 |
-
"grad_norm": 0.
|
| 652 |
-
"learning_rate":
|
| 653 |
-
"loss": 0.
|
| 654 |
"step": 455
|
| 655 |
},
|
| 656 |
{
|
| 657 |
"epoch": 0.8340888485947416,
|
| 658 |
-
"grad_norm": 0.
|
| 659 |
-
"learning_rate":
|
| 660 |
-
"loss": 0.
|
| 661 |
"step": 460
|
| 662 |
},
|
| 663 |
{
|
| 664 |
"epoch": 0.843155031731641,
|
| 665 |
-
"grad_norm": 0.
|
| 666 |
-
"learning_rate":
|
| 667 |
-
"loss": 0.
|
| 668 |
"step": 465
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 0.8522212148685403,
|
| 672 |
-
"grad_norm": 0.
|
| 673 |
-
"learning_rate":
|
| 674 |
-
"loss": 0.
|
| 675 |
"step": 470
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 0.8612873980054397,
|
| 679 |
-
"grad_norm": 0.
|
| 680 |
-
"learning_rate":
|
| 681 |
-
"loss": 0.
|
| 682 |
"step": 475
|
| 683 |
},
|
| 684 |
{
|
| 685 |
"epoch": 0.8703535811423391,
|
| 686 |
-
"grad_norm": 0.
|
| 687 |
-
"learning_rate":
|
| 688 |
-
"loss": 0.
|
| 689 |
"step": 480
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"epoch": 0.8794197642792384,
|
| 693 |
-
"grad_norm": 0.
|
| 694 |
-
"learning_rate":
|
| 695 |
-
"loss": 0.
|
| 696 |
"step": 485
|
| 697 |
},
|
| 698 |
{
|
| 699 |
"epoch": 0.8884859474161378,
|
| 700 |
-
"grad_norm": 0.
|
| 701 |
-
"learning_rate":
|
| 702 |
-
"loss": 0.
|
| 703 |
"step": 490
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"epoch": 0.8975521305530372,
|
| 707 |
-
"grad_norm": 0.
|
| 708 |
-
"learning_rate":
|
| 709 |
-
"loss": 0.
|
| 710 |
"step": 495
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 0.9066183136899365,
|
| 714 |
-
"grad_norm": 0.
|
| 715 |
-
"learning_rate":
|
| 716 |
-
"loss": 0.
|
| 717 |
"step": 500
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 0.915684496826836,
|
| 721 |
-
"grad_norm": 0.
|
| 722 |
-
"learning_rate":
|
| 723 |
-
"loss": 0.
|
| 724 |
"step": 505
|
| 725 |
},
|
| 726 |
{
|
| 727 |
"epoch": 0.9247506799637353,
|
| 728 |
-
"grad_norm": 0.
|
| 729 |
-
"learning_rate": 4.
|
| 730 |
-
"loss": 0.
|
| 731 |
"step": 510
|
| 732 |
},
|
| 733 |
{
|
| 734 |
"epoch": 0.9338168631006346,
|
| 735 |
-
"grad_norm": 0.
|
| 736 |
-
"learning_rate":
|
| 737 |
-
"loss": 0.
|
| 738 |
"step": 515
|
| 739 |
},
|
| 740 |
{
|
| 741 |
"epoch": 0.942883046237534,
|
| 742 |
-
"grad_norm": 0.
|
| 743 |
-
"learning_rate": 3.
|
| 744 |
-
"loss": 0.
|
| 745 |
"step": 520
|
| 746 |
},
|
| 747 |
{
|
| 748 |
"epoch": 0.9519492293744334,
|
| 749 |
-
"grad_norm": 0.
|
| 750 |
-
"learning_rate":
|
| 751 |
-
"loss": 0.
|
| 752 |
"step": 525
|
| 753 |
},
|
| 754 |
{
|
| 755 |
"epoch": 0.9610154125113327,
|
| 756 |
-
"grad_norm": 0.
|
| 757 |
-
"learning_rate": 2.
|
| 758 |
-
"loss": 0.
|
| 759 |
"step": 530
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 0.970081595648232,
|
| 763 |
-
"grad_norm": 0.
|
| 764 |
-
"learning_rate": 1.
|
| 765 |
-
"loss": 0.
|
| 766 |
"step": 535
|
| 767 |
},
|
| 768 |
{
|
| 769 |
"epoch": 0.9791477787851315,
|
| 770 |
-
"grad_norm": 0.
|
| 771 |
-
"learning_rate": 1.
|
| 772 |
-
"loss": 0.
|
| 773 |
"step": 540
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"epoch": 0.9882139619220308,
|
| 777 |
-
"grad_norm": 0.
|
| 778 |
-
"learning_rate":
|
| 779 |
-
"loss": 0.
|
| 780 |
"step": 545
|
| 781 |
},
|
| 782 |
{
|
| 783 |
"epoch": 0.9972801450589301,
|
| 784 |
-
"grad_norm": 0.
|
| 785 |
-
"learning_rate":
|
| 786 |
-
"loss": 0.
|
| 787 |
"step": 550
|
| 788 |
}
|
| 789 |
],
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.001813236627379873,
|
| 14 |
+
"grad_norm": 0.330735981464386,
|
| 15 |
"learning_rate": 0.0,
|
| 16 |
+
"loss": 1.1455,
|
| 17 |
"step": 1
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.009066183136899365,
|
| 21 |
+
"grad_norm": 0.2740427255630493,
|
| 22 |
+
"learning_rate": 1e-05,
|
| 23 |
+
"loss": 1.1478,
|
| 24 |
"step": 5
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.01813236627379873,
|
| 28 |
+
"grad_norm": 0.2780371308326721,
|
| 29 |
+
"learning_rate": 2.25e-05,
|
| 30 |
+
"loss": 1.1416,
|
| 31 |
"step": 10
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.027198549410698096,
|
| 35 |
+
"grad_norm": 0.2090819627046585,
|
| 36 |
+
"learning_rate": 3.5e-05,
|
| 37 |
+
"loss": 1.0763,
|
| 38 |
"step": 15
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.03626473254759746,
|
| 42 |
+
"grad_norm": 0.12894907593727112,
|
| 43 |
+
"learning_rate": 4.75e-05,
|
| 44 |
+
"loss": 0.9366,
|
| 45 |
"step": 20
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.04533091568449683,
|
| 49 |
+
"grad_norm": 0.12664766609668732,
|
| 50 |
+
"learning_rate": 4.9624060150375936e-05,
|
| 51 |
+
"loss": 0.8368,
|
| 52 |
"step": 25
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 0.05439709882139619,
|
| 56 |
+
"grad_norm": 0.10315773636102676,
|
| 57 |
+
"learning_rate": 4.915413533834587e-05,
|
| 58 |
+
"loss": 0.7488,
|
| 59 |
"step": 30
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 0.06346328195829556,
|
| 63 |
+
"grad_norm": 0.08049902319908142,
|
| 64 |
+
"learning_rate": 4.868421052631579e-05,
|
| 65 |
+
"loss": 0.6929,
|
| 66 |
"step": 35
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 0.07252946509519492,
|
| 70 |
+
"grad_norm": 0.06981856375932693,
|
| 71 |
+
"learning_rate": 4.8214285714285716e-05,
|
| 72 |
+
"loss": 0.6419,
|
| 73 |
"step": 40
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 0.0815956482320943,
|
| 77 |
+
"grad_norm": 0.05855804681777954,
|
| 78 |
+
"learning_rate": 4.774436090225564e-05,
|
| 79 |
+
"loss": 0.6141,
|
| 80 |
"step": 45
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.09066183136899365,
|
| 84 |
+
"grad_norm": 0.05399257689714432,
|
| 85 |
+
"learning_rate": 4.7274436090225564e-05,
|
| 86 |
+
"loss": 0.59,
|
| 87 |
"step": 50
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.09972801450589303,
|
| 91 |
+
"grad_norm": 0.05272204801440239,
|
| 92 |
+
"learning_rate": 4.680451127819549e-05,
|
| 93 |
+
"loss": 0.5718,
|
| 94 |
"step": 55
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 0.10879419764279238,
|
| 98 |
+
"grad_norm": 0.050426185131073,
|
| 99 |
+
"learning_rate": 4.633458646616541e-05,
|
| 100 |
+
"loss": 0.5574,
|
| 101 |
"step": 60
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 0.11786038077969176,
|
| 105 |
+
"grad_norm": 0.045167580246925354,
|
| 106 |
+
"learning_rate": 4.586466165413534e-05,
|
| 107 |
+
"loss": 0.5399,
|
| 108 |
"step": 65
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 0.12692656391659113,
|
| 112 |
+
"grad_norm": 0.04907410964369774,
|
| 113 |
+
"learning_rate": 4.539473684210527e-05,
|
| 114 |
+
"loss": 0.5335,
|
| 115 |
"step": 70
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 0.1359927470534905,
|
| 119 |
+
"grad_norm": 0.04511063173413277,
|
| 120 |
+
"learning_rate": 4.492481203007519e-05,
|
| 121 |
+
"loss": 0.5231,
|
| 122 |
"step": 75
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.14505893019038985,
|
| 126 |
+
"grad_norm": 0.03908126801252365,
|
| 127 |
+
"learning_rate": 4.4454887218045117e-05,
|
| 128 |
+
"loss": 0.5162,
|
| 129 |
"step": 80
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.1541251133272892,
|
| 133 |
+
"grad_norm": 0.043064184486866,
|
| 134 |
+
"learning_rate": 4.398496240601504e-05,
|
| 135 |
+
"loss": 0.5085,
|
| 136 |
"step": 85
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.1631912964641886,
|
| 140 |
+
"grad_norm": 0.04580013453960419,
|
| 141 |
+
"learning_rate": 4.3515037593984965e-05,
|
| 142 |
+
"loss": 0.5036,
|
| 143 |
"step": 90
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 0.17225747960108795,
|
| 147 |
+
"grad_norm": 0.05117643624544144,
|
| 148 |
+
"learning_rate": 4.304511278195489e-05,
|
| 149 |
+
"loss": 0.501,
|
| 150 |
"step": 95
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 0.1813236627379873,
|
| 154 |
+
"grad_norm": 0.04914732649922371,
|
| 155 |
+
"learning_rate": 4.2575187969924814e-05,
|
| 156 |
+
"loss": 0.5035,
|
| 157 |
"step": 100
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 0.19038984587488667,
|
| 161 |
+
"grad_norm": 0.05879421904683113,
|
| 162 |
+
"learning_rate": 4.210526315789474e-05,
|
| 163 |
+
"loss": 0.4838,
|
| 164 |
"step": 105
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.19945602901178605,
|
| 168 |
+
"grad_norm": 0.05130897834897041,
|
| 169 |
+
"learning_rate": 4.163533834586467e-05,
|
| 170 |
+
"loss": 0.4835,
|
| 171 |
"step": 110
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.2085222121486854,
|
| 175 |
+
"grad_norm": 0.05152720585465431,
|
| 176 |
+
"learning_rate": 4.116541353383459e-05,
|
| 177 |
+
"loss": 0.4832,
|
| 178 |
"step": 115
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 0.21758839528558477,
|
| 182 |
+
"grad_norm": 0.0526948943734169,
|
| 183 |
+
"learning_rate": 4.069548872180452e-05,
|
| 184 |
+
"loss": 0.4811,
|
| 185 |
"step": 120
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 0.22665457842248413,
|
| 189 |
+
"grad_norm": 0.05324966460466385,
|
| 190 |
+
"learning_rate": 4.022556390977444e-05,
|
| 191 |
+
"loss": 0.4776,
|
| 192 |
"step": 125
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 0.2357207615593835,
|
| 196 |
+
"grad_norm": 0.05410118028521538,
|
| 197 |
+
"learning_rate": 3.9755639097744366e-05,
|
| 198 |
+
"loss": 0.4733,
|
| 199 |
"step": 130
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.24478694469628287,
|
| 203 |
+
"grad_norm": 0.058167893439531326,
|
| 204 |
+
"learning_rate": 3.928571428571429e-05,
|
| 205 |
+
"loss": 0.4703,
|
| 206 |
"step": 135
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.25385312783318226,
|
| 210 |
+
"grad_norm": 0.056478351354599,
|
| 211 |
+
"learning_rate": 3.8815789473684214e-05,
|
| 212 |
+
"loss": 0.4657,
|
| 213 |
"step": 140
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.2629193109700816,
|
| 217 |
+
"grad_norm": 0.07135945558547974,
|
| 218 |
+
"learning_rate": 3.834586466165413e-05,
|
| 219 |
+
"loss": 0.4669,
|
| 220 |
"step": 145
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 0.271985494106981,
|
| 224 |
+
"grad_norm": 0.07001485675573349,
|
| 225 |
+
"learning_rate": 3.787593984962406e-05,
|
| 226 |
+
"loss": 0.4645,
|
| 227 |
"step": 150
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 0.2810516772438803,
|
| 231 |
+
"grad_norm": 0.058485981076955795,
|
| 232 |
+
"learning_rate": 3.740601503759399e-05,
|
| 233 |
+
"loss": 0.4542,
|
| 234 |
"step": 155
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.2901178603807797,
|
| 238 |
+
"grad_norm": 0.05596611276268959,
|
| 239 |
+
"learning_rate": 3.693609022556391e-05,
|
| 240 |
+
"loss": 0.4586,
|
| 241 |
"step": 160
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 0.2991840435176791,
|
| 245 |
+
"grad_norm": 0.06255079060792923,
|
| 246 |
+
"learning_rate": 3.6466165413533835e-05,
|
| 247 |
+
"loss": 0.4582,
|
| 248 |
"step": 165
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 0.3082502266545784,
|
| 252 |
+
"grad_norm": 0.0630059689283371,
|
| 253 |
+
"learning_rate": 3.599624060150376e-05,
|
| 254 |
+
"loss": 0.4504,
|
| 255 |
"step": 170
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.3173164097914778,
|
| 259 |
+
"grad_norm": 0.0674496740102768,
|
| 260 |
+
"learning_rate": 3.5526315789473684e-05,
|
| 261 |
+
"loss": 0.4508,
|
| 262 |
"step": 175
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 0.3263825929283772,
|
| 266 |
+
"grad_norm": 0.06430725008249283,
|
| 267 |
+
"learning_rate": 3.505639097744361e-05,
|
| 268 |
+
"loss": 0.4506,
|
| 269 |
"step": 180
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 0.3354487760652765,
|
| 273 |
+
"grad_norm": 0.07139859348535538,
|
| 274 |
+
"learning_rate": 3.458646616541353e-05,
|
| 275 |
+
"loss": 0.4495,
|
| 276 |
"step": 185
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 0.3445149592021759,
|
| 280 |
+
"grad_norm": 0.08027750253677368,
|
| 281 |
+
"learning_rate": 3.411654135338346e-05,
|
| 282 |
+
"loss": 0.4438,
|
| 283 |
"step": 190
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 0.3535811423390752,
|
| 287 |
+
"grad_norm": 0.07266832143068314,
|
| 288 |
+
"learning_rate": 3.364661654135339e-05,
|
| 289 |
+
"loss": 0.4403,
|
| 290 |
"step": 195
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 0.3626473254759746,
|
| 294 |
+
"grad_norm": 0.0780920684337616,
|
| 295 |
+
"learning_rate": 3.317669172932331e-05,
|
| 296 |
+
"loss": 0.4441,
|
| 297 |
"step": 200
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.371713508612874,
|
| 301 |
+
"grad_norm": 0.07039614021778107,
|
| 302 |
+
"learning_rate": 3.2706766917293236e-05,
|
| 303 |
+
"loss": 0.4362,
|
| 304 |
"step": 205
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 0.38077969174977333,
|
| 308 |
+
"grad_norm": 0.06520690023899078,
|
| 309 |
+
"learning_rate": 3.223684210526316e-05,
|
| 310 |
+
"loss": 0.4391,
|
| 311 |
"step": 210
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 0.3898458748866727,
|
| 315 |
+
"grad_norm": 0.08516666293144226,
|
| 316 |
+
"learning_rate": 3.1766917293233085e-05,
|
| 317 |
+
"loss": 0.4353,
|
| 318 |
"step": 215
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 0.3989120580235721,
|
| 322 |
+
"grad_norm": 0.07929079979658127,
|
| 323 |
+
"learning_rate": 3.129699248120301e-05,
|
| 324 |
+
"loss": 0.4355,
|
| 325 |
"step": 220
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.40797824116047143,
|
| 329 |
+
"grad_norm": 0.07418406754732132,
|
| 330 |
+
"learning_rate": 3.082706766917293e-05,
|
| 331 |
+
"loss": 0.4323,
|
| 332 |
"step": 225
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 0.4170444242973708,
|
| 336 |
+
"grad_norm": 0.07323618978261948,
|
| 337 |
+
"learning_rate": 3.0357142857142857e-05,
|
| 338 |
+
"loss": 0.4346,
|
| 339 |
"step": 230
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.42611060743427015,
|
| 343 |
+
"grad_norm": 0.0705682560801506,
|
| 344 |
+
"learning_rate": 2.9887218045112785e-05,
|
| 345 |
+
"loss": 0.4331,
|
| 346 |
"step": 235
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 0.43517679057116954,
|
| 350 |
+
"grad_norm": 0.0874267965555191,
|
| 351 |
+
"learning_rate": 2.941729323308271e-05,
|
| 352 |
+
"loss": 0.4388,
|
| 353 |
"step": 240
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 0.4442429737080689,
|
| 357 |
+
"grad_norm": 0.07446466386318207,
|
| 358 |
+
"learning_rate": 2.8947368421052634e-05,
|
| 359 |
+
"loss": 0.4349,
|
| 360 |
"step": 245
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 0.45330915684496825,
|
| 364 |
+
"grad_norm": 0.08319786190986633,
|
| 365 |
+
"learning_rate": 2.8477443609022558e-05,
|
| 366 |
+
"loss": 0.4315,
|
| 367 |
"step": 250
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 0.46237533998186764,
|
| 371 |
+
"grad_norm": 0.08069171756505966,
|
| 372 |
+
"learning_rate": 2.8007518796992482e-05,
|
| 373 |
+
"loss": 0.4348,
|
| 374 |
"step": 255
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 0.471441523118767,
|
| 378 |
+
"grad_norm": 0.07842331379652023,
|
| 379 |
+
"learning_rate": 2.7537593984962406e-05,
|
| 380 |
+
"loss": 0.4328,
|
| 381 |
"step": 260
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 0.48050770625566636,
|
| 385 |
+
"grad_norm": 0.07334455102682114,
|
| 386 |
+
"learning_rate": 2.706766917293233e-05,
|
| 387 |
+
"loss": 0.4247,
|
| 388 |
"step": 265
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 0.48957388939256574,
|
| 392 |
+
"grad_norm": 0.06868273764848709,
|
| 393 |
+
"learning_rate": 2.6597744360902255e-05,
|
| 394 |
+
"loss": 0.4304,
|
| 395 |
"step": 270
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 0.4986400725294651,
|
| 399 |
+
"grad_norm": 0.08919775485992432,
|
| 400 |
+
"learning_rate": 2.6127819548872186e-05,
|
| 401 |
+
"loss": 0.4237,
|
| 402 |
"step": 275
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 0.5077062556663645,
|
| 406 |
+
"grad_norm": 0.09395933896303177,
|
| 407 |
+
"learning_rate": 2.565789473684211e-05,
|
| 408 |
+
"loss": 0.4293,
|
| 409 |
"step": 280
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 0.5167724388032638,
|
| 413 |
+
"grad_norm": 0.07169164717197418,
|
| 414 |
+
"learning_rate": 2.5187969924812034e-05,
|
| 415 |
+
"loss": 0.4226,
|
| 416 |
"step": 285
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 0.5258386219401632,
|
| 420 |
+
"grad_norm": 0.07571956515312195,
|
| 421 |
+
"learning_rate": 2.471804511278196e-05,
|
| 422 |
+
"loss": 0.424,
|
| 423 |
"step": 290
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 0.5349048050770625,
|
| 427 |
+
"grad_norm": 0.07626249641180038,
|
| 428 |
+
"learning_rate": 2.424812030075188e-05,
|
| 429 |
+
"loss": 0.4238,
|
| 430 |
"step": 295
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 0.543970988213962,
|
| 434 |
+
"grad_norm": 0.08185533434152603,
|
| 435 |
+
"learning_rate": 2.3778195488721804e-05,
|
| 436 |
+
"loss": 0.4238,
|
| 437 |
"step": 300
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 0.5530371713508613,
|
| 441 |
+
"grad_norm": 0.07463429868221283,
|
| 442 |
+
"learning_rate": 2.3308270676691728e-05,
|
| 443 |
+
"loss": 0.4163,
|
| 444 |
"step": 305
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 0.5621033544877606,
|
| 448 |
+
"grad_norm": 0.0868794396519661,
|
| 449 |
+
"learning_rate": 2.2838345864661655e-05,
|
| 450 |
+
"loss": 0.4206,
|
| 451 |
"step": 310
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 0.57116953762466,
|
| 455 |
+
"grad_norm": 0.09378495812416077,
|
| 456 |
+
"learning_rate": 2.236842105263158e-05,
|
| 457 |
+
"loss": 0.4239,
|
| 458 |
"step": 315
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 0.5802357207615594,
|
| 462 |
+
"grad_norm": 0.07692472636699677,
|
| 463 |
+
"learning_rate": 2.1898496240601504e-05,
|
| 464 |
+
"loss": 0.4173,
|
| 465 |
"step": 320
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 0.5893019038984587,
|
| 469 |
+
"grad_norm": 0.08751624077558517,
|
| 470 |
+
"learning_rate": 2.1428571428571428e-05,
|
| 471 |
+
"loss": 0.4228,
|
| 472 |
"step": 325
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 0.5983680870353582,
|
| 476 |
+
"grad_norm": 0.0897555872797966,
|
| 477 |
+
"learning_rate": 2.0958646616541356e-05,
|
| 478 |
+
"loss": 0.4163,
|
| 479 |
"step": 330
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 0.6074342701722575,
|
| 483 |
+
"grad_norm": 0.08029606938362122,
|
| 484 |
+
"learning_rate": 2.048872180451128e-05,
|
| 485 |
+
"loss": 0.4254,
|
| 486 |
"step": 335
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 0.6165004533091568,
|
| 490 |
+
"grad_norm": 0.09147002547979355,
|
| 491 |
+
"learning_rate": 2.0018796992481204e-05,
|
| 492 |
+
"loss": 0.422,
|
| 493 |
"step": 340
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 0.6255666364460563,
|
| 497 |
+
"grad_norm": 0.08232133090496063,
|
| 498 |
+
"learning_rate": 1.954887218045113e-05,
|
| 499 |
+
"loss": 0.4167,
|
| 500 |
"step": 345
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 0.6346328195829556,
|
| 504 |
+
"grad_norm": 0.08974982798099518,
|
| 505 |
+
"learning_rate": 1.9078947368421056e-05,
|
| 506 |
+
"loss": 0.418,
|
| 507 |
"step": 350
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 0.6436990027198549,
|
| 511 |
+
"grad_norm": 0.07837100327014923,
|
| 512 |
+
"learning_rate": 1.860902255639098e-05,
|
| 513 |
+
"loss": 0.4214,
|
| 514 |
"step": 355
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 0.6527651858567544,
|
| 518 |
+
"grad_norm": 0.08045557886362076,
|
| 519 |
+
"learning_rate": 1.81390977443609e-05,
|
| 520 |
+
"loss": 0.4191,
|
| 521 |
"step": 360
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 0.6618313689936537,
|
| 525 |
+
"grad_norm": 0.08100251108407974,
|
| 526 |
+
"learning_rate": 1.7669172932330825e-05,
|
| 527 |
+
"loss": 0.4175,
|
| 528 |
"step": 365
|
| 529 |
},
|
| 530 |
{
|
| 531 |
"epoch": 0.670897552130553,
|
| 532 |
+
"grad_norm": 0.10134581476449966,
|
| 533 |
+
"learning_rate": 1.7199248120300753e-05,
|
| 534 |
+
"loss": 0.4128,
|
| 535 |
"step": 370
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"epoch": 0.6799637352674524,
|
| 539 |
+
"grad_norm": 0.08495834469795227,
|
| 540 |
+
"learning_rate": 1.6729323308270677e-05,
|
| 541 |
+
"loss": 0.4164,
|
| 542 |
"step": 375
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 0.6890299184043518,
|
| 546 |
+
"grad_norm": 0.088101826608181,
|
| 547 |
+
"learning_rate": 1.62593984962406e-05,
|
| 548 |
+
"loss": 0.4218,
|
| 549 |
"step": 380
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.6980961015412511,
|
| 553 |
+
"grad_norm": 0.08768700808286667,
|
| 554 |
+
"learning_rate": 1.5789473684210526e-05,
|
| 555 |
+
"loss": 0.4145,
|
| 556 |
"step": 385
|
| 557 |
},
|
| 558 |
{
|
| 559 |
"epoch": 0.7071622846781505,
|
| 560 |
+
"grad_norm": 0.08453679084777832,
|
| 561 |
+
"learning_rate": 1.5319548872180453e-05,
|
| 562 |
+
"loss": 0.419,
|
| 563 |
"step": 390
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 0.7162284678150499,
|
| 567 |
+
"grad_norm": 0.08174845576286316,
|
| 568 |
+
"learning_rate": 1.4849624060150378e-05,
|
| 569 |
+
"loss": 0.4171,
|
| 570 |
"step": 395
|
| 571 |
},
|
| 572 |
{
|
| 573 |
"epoch": 0.7252946509519492,
|
| 574 |
+
"grad_norm": 0.08191751688718796,
|
| 575 |
+
"learning_rate": 1.4379699248120302e-05,
|
| 576 |
+
"loss": 0.419,
|
| 577 |
"step": 400
|
| 578 |
},
|
| 579 |
{
|
| 580 |
"epoch": 0.7343608340888486,
|
| 581 |
+
"grad_norm": 0.0835120901465416,
|
| 582 |
+
"learning_rate": 1.3909774436090226e-05,
|
| 583 |
+
"loss": 0.4137,
|
| 584 |
"step": 405
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 0.743427017225748,
|
| 588 |
+
"grad_norm": 0.0808102935552597,
|
| 589 |
+
"learning_rate": 1.3439849624060152e-05,
|
| 590 |
+
"loss": 0.4103,
|
| 591 |
"step": 410
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 0.7524932003626473,
|
| 595 |
+
"grad_norm": 0.07358790189027786,
|
| 596 |
+
"learning_rate": 1.2969924812030076e-05,
|
| 597 |
+
"loss": 0.4136,
|
| 598 |
"step": 415
|
| 599 |
},
|
| 600 |
{
|
| 601 |
"epoch": 0.7615593834995467,
|
| 602 |
+
"grad_norm": 0.07726311683654785,
|
| 603 |
+
"learning_rate": 1.25e-05,
|
| 604 |
+
"loss": 0.4176,
|
| 605 |
"step": 420
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"epoch": 0.7706255666364461,
|
| 609 |
+
"grad_norm": 0.08744051307439804,
|
| 610 |
+
"learning_rate": 1.2030075187969925e-05,
|
| 611 |
+
"loss": 0.411,
|
| 612 |
"step": 425
|
| 613 |
},
|
| 614 |
{
|
| 615 |
"epoch": 0.7796917497733454,
|
| 616 |
+
"grad_norm": 0.07695311307907104,
|
| 617 |
+
"learning_rate": 1.156015037593985e-05,
|
| 618 |
+
"loss": 0.4107,
|
| 619 |
"step": 430
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"epoch": 0.7887579329102448,
|
| 623 |
+
"grad_norm": 0.07769734412431717,
|
| 624 |
+
"learning_rate": 1.1090225563909775e-05,
|
| 625 |
+
"loss": 0.4147,
|
| 626 |
"step": 435
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 0.7978241160471442,
|
| 630 |
+
"grad_norm": 0.0807860940694809,
|
| 631 |
+
"learning_rate": 1.06203007518797e-05,
|
| 632 |
+
"loss": 0.4153,
|
| 633 |
"step": 440
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 0.8068902991840435,
|
| 637 |
+
"grad_norm": 0.0794687420129776,
|
| 638 |
+
"learning_rate": 1.0150375939849624e-05,
|
| 639 |
+
"loss": 0.4089,
|
| 640 |
"step": 445
|
| 641 |
},
|
| 642 |
{
|
| 643 |
"epoch": 0.8159564823209429,
|
| 644 |
+
"grad_norm": 0.09177932888269424,
|
| 645 |
+
"learning_rate": 9.68045112781955e-06,
|
| 646 |
+
"loss": 0.4135,
|
| 647 |
"step": 450
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"epoch": 0.8250226654578422,
|
| 651 |
+
"grad_norm": 0.08271961659193039,
|
| 652 |
+
"learning_rate": 9.210526315789474e-06,
|
| 653 |
+
"loss": 0.415,
|
| 654 |
"step": 455
|
| 655 |
},
|
| 656 |
{
|
| 657 |
"epoch": 0.8340888485947416,
|
| 658 |
+
"grad_norm": 0.0818314403295517,
|
| 659 |
+
"learning_rate": 8.7406015037594e-06,
|
| 660 |
+
"loss": 0.4107,
|
| 661 |
"step": 460
|
| 662 |
},
|
| 663 |
{
|
| 664 |
"epoch": 0.843155031731641,
|
| 665 |
+
"grad_norm": 0.09098558127880096,
|
| 666 |
+
"learning_rate": 8.270676691729324e-06,
|
| 667 |
+
"loss": 0.4076,
|
| 668 |
"step": 465
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 0.8522212148685403,
|
| 672 |
+
"grad_norm": 0.07806940376758575,
|
| 673 |
+
"learning_rate": 7.800751879699248e-06,
|
| 674 |
+
"loss": 0.409,
|
| 675 |
"step": 470
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 0.8612873980054397,
|
| 679 |
+
"grad_norm": 0.08273617923259735,
|
| 680 |
+
"learning_rate": 7.330827067669173e-06,
|
| 681 |
+
"loss": 0.4081,
|
| 682 |
"step": 475
|
| 683 |
},
|
| 684 |
{
|
| 685 |
"epoch": 0.8703535811423391,
|
| 686 |
+
"grad_norm": 0.08064981549978256,
|
| 687 |
+
"learning_rate": 6.860902255639098e-06,
|
| 688 |
+
"loss": 0.4144,
|
| 689 |
"step": 480
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"epoch": 0.8794197642792384,
|
| 693 |
+
"grad_norm": 0.07812593877315521,
|
| 694 |
+
"learning_rate": 6.3909774436090225e-06,
|
| 695 |
+
"loss": 0.4148,
|
| 696 |
"step": 485
|
| 697 |
},
|
| 698 |
{
|
| 699 |
"epoch": 0.8884859474161378,
|
| 700 |
+
"grad_norm": 0.083773672580719,
|
| 701 |
+
"learning_rate": 5.921052631578948e-06,
|
| 702 |
+
"loss": 0.4185,
|
| 703 |
"step": 490
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"epoch": 0.8975521305530372,
|
| 707 |
+
"grad_norm": 0.07797183096408844,
|
| 708 |
+
"learning_rate": 5.451127819548872e-06,
|
| 709 |
+
"loss": 0.4106,
|
| 710 |
"step": 495
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 0.9066183136899365,
|
| 714 |
+
"grad_norm": 0.07622473686933517,
|
| 715 |
+
"learning_rate": 4.981203007518797e-06,
|
| 716 |
+
"loss": 0.4119,
|
| 717 |
"step": 500
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 0.915684496826836,
|
| 721 |
+
"grad_norm": 0.0780773013830185,
|
| 722 |
+
"learning_rate": 4.511278195488722e-06,
|
| 723 |
+
"loss": 0.4112,
|
| 724 |
"step": 505
|
| 725 |
},
|
| 726 |
{
|
| 727 |
"epoch": 0.9247506799637353,
|
| 728 |
+
"grad_norm": 0.08101441711187363,
|
| 729 |
+
"learning_rate": 4.041353383458646e-06,
|
| 730 |
+
"loss": 0.4089,
|
| 731 |
"step": 510
|
| 732 |
},
|
| 733 |
{
|
| 734 |
"epoch": 0.9338168631006346,
|
| 735 |
+
"grad_norm": 0.08148541301488876,
|
| 736 |
+
"learning_rate": 3.5714285714285714e-06,
|
| 737 |
+
"loss": 0.4154,
|
| 738 |
"step": 515
|
| 739 |
},
|
| 740 |
{
|
| 741 |
"epoch": 0.942883046237534,
|
| 742 |
+
"grad_norm": 0.07560685276985168,
|
| 743 |
+
"learning_rate": 3.101503759398496e-06,
|
| 744 |
+
"loss": 0.4052,
|
| 745 |
"step": 520
|
| 746 |
},
|
| 747 |
{
|
| 748 |
"epoch": 0.9519492293744334,
|
| 749 |
+
"grad_norm": 0.07574689388275146,
|
| 750 |
+
"learning_rate": 2.631578947368421e-06,
|
| 751 |
+
"loss": 0.4142,
|
| 752 |
"step": 525
|
| 753 |
},
|
| 754 |
{
|
| 755 |
"epoch": 0.9610154125113327,
|
| 756 |
+
"grad_norm": 0.07745104283094406,
|
| 757 |
+
"learning_rate": 2.1616541353383458e-06,
|
| 758 |
+
"loss": 0.4104,
|
| 759 |
"step": 530
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 0.970081595648232,
|
| 763 |
+
"grad_norm": 0.07881065458059311,
|
| 764 |
+
"learning_rate": 1.6917293233082707e-06,
|
| 765 |
+
"loss": 0.4151,
|
| 766 |
"step": 535
|
| 767 |
},
|
| 768 |
{
|
| 769 |
"epoch": 0.9791477787851315,
|
| 770 |
+
"grad_norm": 0.07710935920476913,
|
| 771 |
+
"learning_rate": 1.2218045112781955e-06,
|
| 772 |
+
"loss": 0.4095,
|
| 773 |
"step": 540
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"epoch": 0.9882139619220308,
|
| 777 |
+
"grad_norm": 0.07693067193031311,
|
| 778 |
+
"learning_rate": 7.518796992481203e-07,
|
| 779 |
+
"loss": 0.4097,
|
| 780 |
"step": 545
|
| 781 |
},
|
| 782 |
{
|
| 783 |
"epoch": 0.9972801450589301,
|
| 784 |
+
"grad_norm": 0.07568664103746414,
|
| 785 |
+
"learning_rate": 2.8195488721804513e-07,
|
| 786 |
+
"loss": 0.406,
|
| 787 |
"step": 550
|
| 788 |
}
|
| 789 |
],
|
math-0.3/q/cold-start/checkpoint-552/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 6289
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:417f069ca3f7ca2d3c0eb8d0203990078a027cfcb3923306206e6eaaa2b7e350
|
| 3 |
size 6289
|