Upload folder using huggingface_hub
Browse files- adapter_config.json +1 -1
- adapter_model.safetensors +2 -2
- optimizer.pt +2 -2
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +1293 -186
- training_args.bin +2 -2
adapter_config.json
CHANGED
|
@@ -24,8 +24,8 @@
|
|
| 24 |
"revision": null,
|
| 25 |
"target_modules": [
|
| 26 |
"qkv_proj",
|
| 27 |
-
"o_proj",
|
| 28 |
"gate_up_proj",
|
|
|
|
| 29 |
"down_proj"
|
| 30 |
],
|
| 31 |
"task_type": "CAUSAL_LM",
|
|
|
|
| 24 |
"revision": null,
|
| 25 |
"target_modules": [
|
| 26 |
"qkv_proj",
|
|
|
|
| 27 |
"gate_up_proj",
|
| 28 |
+
"o_proj",
|
| 29 |
"down_proj"
|
| 30 |
],
|
| 31 |
"task_type": "CAUSAL_LM",
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1b0a59517e4a01479864ee5d89d301a8a00f989766f263f6776c0bd7ac658eef
|
| 3 |
+
size 161515608
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:777aa2f3c1c30caa4a52015147ada8dfe521150c7bcc0e8604bc28f2deed2ef2
|
| 3 |
+
size 323181259
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e5b517d1b8e2b0f837c8b00170b154961d4d989feba4326ac25583df7a55c57a
|
| 3 |
size 14645
|
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 sha256:66b0cc75c21e30eb630a9fff29b7a6e24e8501ef27a94592e859fd7084f223ef
|
| 3 |
size 1465
|
trainer_state.json
CHANGED
|
@@ -2,362 +2,1469 @@
|
|
| 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 |
-
"
|
| 14 |
-
"
|
|
|
|
| 15 |
"learning_rate": 0.0,
|
| 16 |
-
"loss": 2.
|
|
|
|
|
|
|
| 17 |
"step": 1
|
| 18 |
},
|
| 19 |
{
|
| 20 |
-
"
|
| 21 |
-
"
|
|
|
|
| 22 |
"learning_rate": 2e-06,
|
| 23 |
-
"loss": 2.
|
|
|
|
|
|
|
| 24 |
"step": 2
|
| 25 |
},
|
| 26 |
{
|
| 27 |
-
"
|
| 28 |
-
"
|
|
|
|
| 29 |
"learning_rate": 4e-06,
|
| 30 |
-
"loss": 2.
|
|
|
|
|
|
|
| 31 |
"step": 3
|
| 32 |
},
|
| 33 |
{
|
| 34 |
-
"
|
| 35 |
-
"
|
|
|
|
| 36 |
"learning_rate": 6e-06,
|
| 37 |
-
"loss": 2.
|
|
|
|
|
|
|
| 38 |
"step": 4
|
| 39 |
},
|
| 40 |
{
|
| 41 |
-
"
|
| 42 |
-
"
|
|
|
|
| 43 |
"learning_rate": 8e-06,
|
| 44 |
-
"loss": 2.
|
|
|
|
|
|
|
| 45 |
"step": 5
|
| 46 |
},
|
| 47 |
{
|
| 48 |
-
"
|
| 49 |
-
"
|
|
|
|
| 50 |
"learning_rate": 9.999999999999999e-06,
|
| 51 |
-
"loss": 2.
|
|
|
|
|
|
|
| 52 |
"step": 6
|
| 53 |
},
|
| 54 |
{
|
| 55 |
-
"
|
| 56 |
-
"
|
|
|
|
| 57 |
"learning_rate": 1.2e-05,
|
| 58 |
-
"loss": 2.
|
|
|
|
|
|
|
| 59 |
"step": 7
|
| 60 |
},
|
| 61 |
{
|
| 62 |
-
"
|
| 63 |
-
"
|
|
|
|
| 64 |
"learning_rate": 1.4e-05,
|
| 65 |
-
"loss": 2.
|
|
|
|
|
|
|
| 66 |
"step": 8
|
| 67 |
},
|
| 68 |
{
|
| 69 |
-
"
|
| 70 |
-
"
|
|
|
|
| 71 |
"learning_rate": 1.6e-05,
|
| 72 |
-
"loss": 1.
|
|
|
|
|
|
|
| 73 |
"step": 9
|
| 74 |
},
|
| 75 |
{
|
| 76 |
-
"
|
| 77 |
-
"
|
|
|
|
| 78 |
"learning_rate": 1.8e-05,
|
| 79 |
-
"loss": 1.
|
|
|
|
|
|
|
| 80 |
"step": 10
|
| 81 |
},
|
| 82 |
{
|
| 83 |
-
"
|
| 84 |
-
"
|
|
|
|
| 85 |
"learning_rate": 1.9999999999999998e-05,
|
| 86 |
-
"loss": 1.
|
|
|
|
|
|
|
| 87 |
"step": 11
|
| 88 |
},
|
| 89 |
{
|
| 90 |
-
"
|
| 91 |
-
"
|
|
|
|
| 92 |
"learning_rate": 2.2e-05,
|
| 93 |
-
"loss":
|
|
|
|
|
|
|
| 94 |
"step": 12
|
| 95 |
},
|
| 96 |
{
|
| 97 |
-
"
|
| 98 |
-
"
|
|
|
|
| 99 |
"learning_rate": 2.4e-05,
|
| 100 |
-
"loss": 1.
|
|
|
|
|
|
|
| 101 |
"step": 13
|
| 102 |
},
|
| 103 |
{
|
| 104 |
-
"
|
| 105 |
-
"
|
|
|
|
| 106 |
"learning_rate": 2.6000000000000002e-05,
|
| 107 |
-
"loss": 1.
|
|
|
|
|
|
|
| 108 |
"step": 14
|
| 109 |
},
|
| 110 |
{
|
| 111 |
-
"
|
| 112 |
-
"
|
|
|
|
| 113 |
"learning_rate": 2.8e-05,
|
| 114 |
-
"loss": 1.
|
|
|
|
|
|
|
| 115 |
"step": 15
|
| 116 |
},
|
| 117 |
{
|
| 118 |
-
"
|
| 119 |
-
"
|
|
|
|
| 120 |
"learning_rate": 3e-05,
|
| 121 |
-
"loss": 1.
|
|
|
|
|
|
|
| 122 |
"step": 16
|
| 123 |
},
|
| 124 |
{
|
| 125 |
-
"
|
| 126 |
-
"
|
| 127 |
-
"
|
| 128 |
-
"
|
|
|
|
|
|
|
|
|
|
| 129 |
"step": 17
|
| 130 |
},
|
| 131 |
{
|
| 132 |
-
"
|
| 133 |
-
"
|
| 134 |
-
"
|
| 135 |
-
"
|
|
|
|
|
|
|
|
|
|
| 136 |
"step": 18
|
| 137 |
},
|
| 138 |
{
|
| 139 |
-
"
|
| 140 |
-
"
|
| 141 |
-
"
|
| 142 |
-
"
|
|
|
|
|
|
|
|
|
|
| 143 |
"step": 19
|
| 144 |
},
|
| 145 |
{
|
| 146 |
-
"
|
| 147 |
-
"
|
| 148 |
-
"
|
| 149 |
-
"
|
|
|
|
|
|
|
|
|
|
| 150 |
"step": 20
|
| 151 |
},
|
| 152 |
{
|
| 153 |
-
"
|
| 154 |
-
"
|
| 155 |
-
"
|
| 156 |
-
"
|
|
|
|
|
|
|
|
|
|
| 157 |
"step": 21
|
| 158 |
},
|
| 159 |
{
|
| 160 |
-
"
|
| 161 |
-
"
|
| 162 |
-
"
|
| 163 |
-
"
|
|
|
|
|
|
|
|
|
|
| 164 |
"step": 22
|
| 165 |
},
|
| 166 |
{
|
| 167 |
-
"
|
| 168 |
-
"
|
| 169 |
-
"
|
| 170 |
-
"
|
|
|
|
|
|
|
|
|
|
| 171 |
"step": 23
|
| 172 |
},
|
| 173 |
{
|
| 174 |
-
"
|
| 175 |
-
"
|
| 176 |
-
"
|
| 177 |
-
"
|
|
|
|
|
|
|
|
|
|
| 178 |
"step": 24
|
| 179 |
},
|
| 180 |
{
|
| 181 |
-
"
|
| 182 |
-
"
|
| 183 |
-
"
|
| 184 |
-
"
|
|
|
|
|
|
|
|
|
|
| 185 |
"step": 25
|
| 186 |
},
|
| 187 |
{
|
| 188 |
-
"
|
| 189 |
-
"
|
| 190 |
-
"
|
| 191 |
-
"
|
|
|
|
|
|
|
|
|
|
| 192 |
"step": 26
|
| 193 |
},
|
| 194 |
{
|
| 195 |
-
"
|
| 196 |
-
"
|
| 197 |
-
"
|
| 198 |
-
"
|
|
|
|
|
|
|
|
|
|
| 199 |
"step": 27
|
| 200 |
},
|
| 201 |
{
|
| 202 |
-
"
|
| 203 |
-
"
|
| 204 |
-
"
|
| 205 |
-
"
|
|
|
|
|
|
|
|
|
|
| 206 |
"step": 28
|
| 207 |
},
|
| 208 |
{
|
| 209 |
-
"
|
| 210 |
-
"
|
| 211 |
-
"
|
| 212 |
-
"
|
|
|
|
|
|
|
|
|
|
| 213 |
"step": 29
|
| 214 |
},
|
| 215 |
{
|
| 216 |
-
"
|
| 217 |
-
"
|
| 218 |
-
"
|
| 219 |
-
"
|
|
|
|
|
|
|
|
|
|
| 220 |
"step": 30
|
| 221 |
},
|
| 222 |
{
|
| 223 |
-
"
|
| 224 |
-
"
|
| 225 |
-
"
|
| 226 |
-
"
|
|
|
|
|
|
|
|
|
|
| 227 |
"step": 31
|
| 228 |
},
|
| 229 |
{
|
| 230 |
-
"
|
| 231 |
-
"
|
| 232 |
-
"
|
| 233 |
-
"
|
|
|
|
|
|
|
|
|
|
| 234 |
"step": 32
|
| 235 |
},
|
| 236 |
{
|
| 237 |
-
"
|
| 238 |
-
"
|
| 239 |
-
"
|
| 240 |
-
"
|
|
|
|
|
|
|
|
|
|
| 241 |
"step": 33
|
| 242 |
},
|
| 243 |
{
|
| 244 |
-
"
|
| 245 |
-
"
|
| 246 |
-
"
|
| 247 |
-
"
|
|
|
|
|
|
|
|
|
|
| 248 |
"step": 34
|
| 249 |
},
|
| 250 |
{
|
| 251 |
-
"
|
| 252 |
-
"
|
| 253 |
-
"
|
| 254 |
-
"
|
|
|
|
|
|
|
|
|
|
| 255 |
"step": 35
|
| 256 |
},
|
| 257 |
{
|
| 258 |
-
"
|
| 259 |
-
"
|
| 260 |
-
"
|
| 261 |
-
"
|
|
|
|
|
|
|
|
|
|
| 262 |
"step": 36
|
| 263 |
},
|
| 264 |
{
|
| 265 |
-
"
|
| 266 |
-
"
|
| 267 |
-
"
|
| 268 |
-
"
|
|
|
|
|
|
|
|
|
|
| 269 |
"step": 37
|
| 270 |
},
|
| 271 |
{
|
| 272 |
-
"
|
| 273 |
-
"
|
| 274 |
-
"
|
| 275 |
-
"
|
|
|
|
|
|
|
|
|
|
| 276 |
"step": 38
|
| 277 |
},
|
| 278 |
{
|
| 279 |
-
"
|
| 280 |
-
"
|
| 281 |
-
"
|
| 282 |
-
"
|
|
|
|
|
|
|
|
|
|
| 283 |
"step": 39
|
| 284 |
},
|
| 285 |
{
|
| 286 |
-
"
|
| 287 |
-
"
|
| 288 |
-
"
|
| 289 |
-
"
|
|
|
|
|
|
|
|
|
|
| 290 |
"step": 40
|
| 291 |
},
|
| 292 |
{
|
| 293 |
-
"
|
| 294 |
-
"
|
| 295 |
-
"
|
| 296 |
-
"
|
|
|
|
|
|
|
|
|
|
| 297 |
"step": 41
|
| 298 |
},
|
| 299 |
{
|
| 300 |
-
"
|
| 301 |
-
"
|
| 302 |
-
"
|
| 303 |
-
"
|
|
|
|
|
|
|
|
|
|
| 304 |
"step": 42
|
| 305 |
},
|
| 306 |
{
|
| 307 |
-
"
|
| 308 |
-
"
|
| 309 |
-
"
|
| 310 |
-
"
|
|
|
|
|
|
|
|
|
|
| 311 |
"step": 43
|
| 312 |
},
|
| 313 |
{
|
| 314 |
-
"
|
| 315 |
-
"
|
| 316 |
-
"
|
| 317 |
-
"
|
|
|
|
|
|
|
|
|
|
| 318 |
"step": 44
|
| 319 |
},
|
| 320 |
{
|
| 321 |
-
"
|
| 322 |
-
"
|
| 323 |
-
"
|
| 324 |
-
"
|
|
|
|
|
|
|
|
|
|
| 325 |
"step": 45
|
| 326 |
},
|
| 327 |
{
|
| 328 |
-
"
|
| 329 |
-
"
|
| 330 |
-
"
|
| 331 |
-
"
|
|
|
|
|
|
|
|
|
|
| 332 |
"step": 46
|
| 333 |
},
|
| 334 |
{
|
| 335 |
-
"
|
| 336 |
-
"
|
| 337 |
-
"
|
| 338 |
-
"
|
|
|
|
|
|
|
|
|
|
| 339 |
"step": 47
|
| 340 |
},
|
| 341 |
{
|
| 342 |
-
"
|
| 343 |
-
"
|
| 344 |
-
"
|
| 345 |
-
"
|
|
|
|
|
|
|
|
|
|
| 346 |
"step": 48
|
| 347 |
},
|
| 348 |
{
|
| 349 |
-
"
|
| 350 |
-
"
|
| 351 |
-
"
|
| 352 |
-
"
|
|
|
|
|
|
|
|
|
|
| 353 |
"step": 49
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 354 |
}
|
| 355 |
],
|
| 356 |
"logging_steps": 1,
|
| 357 |
-
"max_steps":
|
| 358 |
"num_input_tokens_seen": 0,
|
| 359 |
-
"num_train_epochs":
|
| 360 |
-
"save_steps":
|
| 361 |
"stateful_callbacks": {
|
| 362 |
"TrainerControl": {
|
| 363 |
"args": {
|
|
@@ -370,7 +1477,7 @@
|
|
| 370 |
"attributes": {}
|
| 371 |
}
|
| 372 |
},
|
| 373 |
-
"total_flos":
|
| 374 |
"train_batch_size": 2,
|
| 375 |
"trial_name": null,
|
| 376 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 4.398467432950191,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 145,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
+
"entropy": 2.2323372662067413,
|
| 14 |
+
"epoch": 0.03065134099616858,
|
| 15 |
+
"grad_norm": 23.125,
|
| 16 |
"learning_rate": 0.0,
|
| 17 |
+
"loss": 2.7706,
|
| 18 |
+
"mean_token_accuracy": 0.41634324193000793,
|
| 19 |
+
"num_tokens": 1244.0,
|
| 20 |
"step": 1
|
| 21 |
},
|
| 22 |
{
|
| 23 |
+
"entropy": 2.174584299325943,
|
| 24 |
+
"epoch": 0.06130268199233716,
|
| 25 |
+
"grad_norm": 15.625,
|
| 26 |
"learning_rate": 2e-06,
|
| 27 |
+
"loss": 2.4332,
|
| 28 |
+
"mean_token_accuracy": 0.41893551871180534,
|
| 29 |
+
"num_tokens": 3427.0,
|
| 30 |
"step": 2
|
| 31 |
},
|
| 32 |
{
|
| 33 |
+
"entropy": 2.026109978556633,
|
| 34 |
+
"epoch": 0.09195402298850575,
|
| 35 |
+
"grad_norm": 13.625,
|
| 36 |
"learning_rate": 4e-06,
|
| 37 |
+
"loss": 2.334,
|
| 38 |
+
"mean_token_accuracy": 0.4435151480138302,
|
| 39 |
+
"num_tokens": 5582.0,
|
| 40 |
"step": 3
|
| 41 |
},
|
| 42 |
{
|
| 43 |
+
"entropy": 2.292216345667839,
|
| 44 |
+
"epoch": 0.12260536398467432,
|
| 45 |
+
"grad_norm": 17.125,
|
| 46 |
"learning_rate": 6e-06,
|
| 47 |
+
"loss": 2.4632,
|
| 48 |
+
"mean_token_accuracy": 0.4366639256477356,
|
| 49 |
+
"num_tokens": 7064.0,
|
| 50 |
"step": 4
|
| 51 |
},
|
| 52 |
{
|
| 53 |
+
"entropy": 2.308288350701332,
|
| 54 |
+
"epoch": 0.1532567049808429,
|
| 55 |
+
"grad_norm": 11.5625,
|
| 56 |
"learning_rate": 8e-06,
|
| 57 |
+
"loss": 2.196,
|
| 58 |
+
"mean_token_accuracy": 0.43852633982896805,
|
| 59 |
+
"num_tokens": 9091.0,
|
| 60 |
"step": 5
|
| 61 |
},
|
| 62 |
{
|
| 63 |
+
"entropy": 2.431372880935669,
|
| 64 |
+
"epoch": 0.1839080459770115,
|
| 65 |
+
"grad_norm": 12.5625,
|
| 66 |
"learning_rate": 9.999999999999999e-06,
|
| 67 |
+
"loss": 2.4038,
|
| 68 |
+
"mean_token_accuracy": 0.47505030781030655,
|
| 69 |
+
"num_tokens": 10556.0,
|
| 70 |
"step": 6
|
| 71 |
},
|
| 72 |
{
|
| 73 |
+
"entropy": 2.442482829093933,
|
| 74 |
+
"epoch": 0.21455938697318008,
|
| 75 |
+
"grad_norm": 10.625,
|
| 76 |
"learning_rate": 1.2e-05,
|
| 77 |
+
"loss": 2.2507,
|
| 78 |
+
"mean_token_accuracy": 0.4729406535625458,
|
| 79 |
+
"num_tokens": 12215.0,
|
| 80 |
"step": 7
|
| 81 |
},
|
| 82 |
{
|
| 83 |
+
"entropy": 2.356838285923004,
|
| 84 |
+
"epoch": 0.24521072796934865,
|
| 85 |
+
"grad_norm": 8.6875,
|
| 86 |
"learning_rate": 1.4e-05,
|
| 87 |
+
"loss": 2.003,
|
| 88 |
+
"mean_token_accuracy": 0.5002904869616032,
|
| 89 |
+
"num_tokens": 13939.0,
|
| 90 |
"step": 8
|
| 91 |
},
|
| 92 |
{
|
| 93 |
+
"entropy": 2.3505502343177795,
|
| 94 |
+
"epoch": 0.27586206896551724,
|
| 95 |
+
"grad_norm": 7.0,
|
| 96 |
"learning_rate": 1.6e-05,
|
| 97 |
+
"loss": 1.8798,
|
| 98 |
+
"mean_token_accuracy": 0.4973266273736954,
|
| 99 |
+
"num_tokens": 15986.0,
|
| 100 |
"step": 9
|
| 101 |
},
|
| 102 |
{
|
| 103 |
+
"entropy": 2.3207567632198334,
|
| 104 |
+
"epoch": 0.3065134099616858,
|
| 105 |
+
"grad_norm": 6.78125,
|
| 106 |
"learning_rate": 1.8e-05,
|
| 107 |
+
"loss": 1.8644,
|
| 108 |
+
"mean_token_accuracy": 0.5209791846573353,
|
| 109 |
+
"num_tokens": 18444.0,
|
| 110 |
"step": 10
|
| 111 |
},
|
| 112 |
{
|
| 113 |
+
"entropy": 2.071987748146057,
|
| 114 |
+
"epoch": 0.3371647509578544,
|
| 115 |
+
"grad_norm": 5.34375,
|
| 116 |
"learning_rate": 1.9999999999999998e-05,
|
| 117 |
+
"loss": 1.6628,
|
| 118 |
+
"mean_token_accuracy": 0.56425766274333,
|
| 119 |
+
"num_tokens": 21127.0,
|
| 120 |
"step": 11
|
| 121 |
},
|
| 122 |
{
|
| 123 |
+
"entropy": 2.291467070579529,
|
| 124 |
+
"epoch": 0.367816091954023,
|
| 125 |
+
"grad_norm": 6.40625,
|
| 126 |
"learning_rate": 2.2e-05,
|
| 127 |
+
"loss": 1.9017,
|
| 128 |
+
"mean_token_accuracy": 0.5183870233595371,
|
| 129 |
+
"num_tokens": 23308.0,
|
| 130 |
"step": 12
|
| 131 |
},
|
| 132 |
{
|
| 133 |
+
"entropy": 2.2638841718435287,
|
| 134 |
+
"epoch": 0.39846743295019155,
|
| 135 |
+
"grad_norm": 6.5625,
|
| 136 |
"learning_rate": 2.4e-05,
|
| 137 |
+
"loss": 1.975,
|
| 138 |
+
"mean_token_accuracy": 0.5189315676689148,
|
| 139 |
+
"num_tokens": 25072.0,
|
| 140 |
"step": 13
|
| 141 |
},
|
| 142 |
{
|
| 143 |
+
"entropy": 2.312656208872795,
|
| 144 |
+
"epoch": 0.42911877394636017,
|
| 145 |
+
"grad_norm": 7.53125,
|
| 146 |
"learning_rate": 2.6000000000000002e-05,
|
| 147 |
+
"loss": 1.9894,
|
| 148 |
+
"mean_token_accuracy": 0.5101005285978317,
|
| 149 |
+
"num_tokens": 26450.0,
|
| 150 |
"step": 14
|
| 151 |
},
|
| 152 |
{
|
| 153 |
+
"entropy": 2.285459041595459,
|
| 154 |
+
"epoch": 0.45977011494252873,
|
| 155 |
+
"grad_norm": 6.78125,
|
| 156 |
"learning_rate": 2.8e-05,
|
| 157 |
+
"loss": 1.8031,
|
| 158 |
+
"mean_token_accuracy": 0.5333016850054264,
|
| 159 |
+
"num_tokens": 27912.0,
|
| 160 |
"step": 15
|
| 161 |
},
|
| 162 |
{
|
| 163 |
+
"entropy": 2.1630422174930573,
|
| 164 |
+
"epoch": 0.4904214559386973,
|
| 165 |
+
"grad_norm": 6.4375,
|
| 166 |
"learning_rate": 3e-05,
|
| 167 |
+
"loss": 1.6778,
|
| 168 |
+
"mean_token_accuracy": 0.5712834745645523,
|
| 169 |
+
"num_tokens": 29392.0,
|
| 170 |
"step": 16
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"entropy": 2.0271100252866745,
|
| 174 |
+
"epoch": 0.5210727969348659,
|
| 175 |
+
"grad_norm": 4.84375,
|
| 176 |
+
"learning_rate": 2.9996710252122685e-05,
|
| 177 |
+
"loss": 1.4808,
|
| 178 |
+
"mean_token_accuracy": 0.5967330485582352,
|
| 179 |
+
"num_tokens": 31417.0,
|
| 180 |
"step": 17
|
| 181 |
},
|
| 182 |
{
|
| 183 |
+
"entropy": 2.046408548951149,
|
| 184 |
+
"epoch": 0.5517241379310345,
|
| 185 |
+
"grad_norm": 6.0,
|
| 186 |
+
"learning_rate": 2.9986842451482876e-05,
|
| 187 |
+
"loss": 1.7747,
|
| 188 |
+
"mean_token_accuracy": 0.5322429053485394,
|
| 189 |
+
"num_tokens": 33727.0,
|
| 190 |
"step": 18
|
| 191 |
},
|
| 192 |
{
|
| 193 |
+
"entropy": 1.9437416046857834,
|
| 194 |
+
"epoch": 0.5823754789272031,
|
| 195 |
+
"grad_norm": 5.71875,
|
| 196 |
+
"learning_rate": 2.9970400926424075e-05,
|
| 197 |
+
"loss": 1.5966,
|
| 198 |
+
"mean_token_accuracy": 0.5690519884228706,
|
| 199 |
+
"num_tokens": 35836.0,
|
| 200 |
"step": 19
|
| 201 |
},
|
| 202 |
{
|
| 203 |
+
"entropy": 2.019773244857788,
|
| 204 |
+
"epoch": 0.6130268199233716,
|
| 205 |
+
"grad_norm": 5.8125,
|
| 206 |
+
"learning_rate": 2.9947392888742566e-05,
|
| 207 |
+
"loss": 1.7132,
|
| 208 |
+
"mean_token_accuracy": 0.5585189834237099,
|
| 209 |
+
"num_tokens": 37639.0,
|
| 210 |
"step": 20
|
| 211 |
},
|
| 212 |
{
|
| 213 |
+
"entropy": 2.1920904517173767,
|
| 214 |
+
"epoch": 0.6436781609195402,
|
| 215 |
+
"grad_norm": 5.59375,
|
| 216 |
+
"learning_rate": 2.99178284305241e-05,
|
| 217 |
+
"loss": 1.7894,
|
| 218 |
+
"mean_token_accuracy": 0.5173998288810253,
|
| 219 |
+
"num_tokens": 39351.0,
|
| 220 |
"step": 21
|
| 221 |
},
|
| 222 |
{
|
| 223 |
+
"entropy": 1.9828892648220062,
|
| 224 |
+
"epoch": 0.6743295019157088,
|
| 225 |
+
"grad_norm": 5.96875,
|
| 226 |
+
"learning_rate": 2.988172051971717e-05,
|
| 227 |
+
"loss": 1.5489,
|
| 228 |
+
"mean_token_accuracy": 0.5762875340878963,
|
| 229 |
+
"num_tokens": 41024.0,
|
| 230 |
"step": 22
|
| 231 |
},
|
| 232 |
{
|
| 233 |
+
"entropy": 2.076566159725189,
|
| 234 |
+
"epoch": 0.7049808429118773,
|
| 235 |
+
"grad_norm": 5.96875,
|
| 236 |
+
"learning_rate": 2.983908499444483e-05,
|
| 237 |
+
"loss": 1.7828,
|
| 238 |
+
"mean_token_accuracy": 0.554260715842247,
|
| 239 |
+
"num_tokens": 42624.0,
|
| 240 |
"step": 23
|
| 241 |
},
|
| 242 |
{
|
| 243 |
+
"entropy": 2.1858795881271362,
|
| 244 |
+
"epoch": 0.735632183908046,
|
| 245 |
+
"grad_norm": 7.75,
|
| 246 |
+
"learning_rate": 2.9789940556057574e-05,
|
| 247 |
+
"loss": 1.9101,
|
| 248 |
+
"mean_token_accuracy": 0.5420451276004314,
|
| 249 |
+
"num_tokens": 43836.0,
|
| 250 |
"step": 24
|
| 251 |
},
|
| 252 |
{
|
| 253 |
+
"entropy": 2.0579274892807007,
|
| 254 |
+
"epoch": 0.7662835249042146,
|
| 255 |
+
"grad_norm": 5.0625,
|
| 256 |
+
"learning_rate": 2.9734308760930333e-05,
|
| 257 |
+
"loss": 1.7436,
|
| 258 |
+
"mean_token_accuracy": 0.5261486992239952,
|
| 259 |
+
"num_tokens": 45870.0,
|
| 260 |
"step": 25
|
| 261 |
},
|
| 262 |
{
|
| 263 |
+
"entropy": 2.0463136434555054,
|
| 264 |
+
"epoch": 0.7969348659003831,
|
| 265 |
+
"grad_norm": 5.3125,
|
| 266 |
+
"learning_rate": 2.9672214011007087e-05,
|
| 267 |
+
"loss": 1.6874,
|
| 268 |
+
"mean_token_accuracy": 0.5605301037430763,
|
| 269 |
+
"num_tokens": 47664.0,
|
| 270 |
"step": 26
|
| 271 |
},
|
| 272 |
{
|
| 273 |
+
"entropy": 2.0209241062402725,
|
| 274 |
+
"epoch": 0.8275862068965517,
|
| 275 |
+
"grad_norm": 5.125,
|
| 276 |
+
"learning_rate": 2.9603683543097406e-05,
|
| 277 |
+
"loss": 1.6662,
|
| 278 |
+
"mean_token_accuracy": 0.5574814975261688,
|
| 279 |
+
"num_tokens": 49646.0,
|
| 280 |
"step": 27
|
| 281 |
},
|
| 282 |
{
|
| 283 |
+
"entropy": 2.006110742688179,
|
| 284 |
+
"epoch": 0.8582375478927203,
|
| 285 |
+
"grad_norm": 5.15625,
|
| 286 |
+
"learning_rate": 2.9528747416929467e-05,
|
| 287 |
+
"loss": 1.5072,
|
| 288 |
+
"mean_token_accuracy": 0.5874482207000256,
|
| 289 |
+
"num_tokens": 51235.0,
|
| 290 |
"step": 28
|
| 291 |
},
|
| 292 |
{
|
| 293 |
+
"entropy": 2.0461973696947098,
|
| 294 |
+
"epoch": 0.8888888888888888,
|
| 295 |
+
"grad_norm": 5.40625,
|
| 296 |
+
"learning_rate": 2.9447438501964873e-05,
|
| 297 |
+
"loss": 1.6601,
|
| 298 |
+
"mean_token_accuracy": 0.5673006623983383,
|
| 299 |
+
"num_tokens": 53037.0,
|
| 300 |
"step": 29
|
| 301 |
},
|
| 302 |
{
|
| 303 |
+
"entropy": 2.0841187089681625,
|
| 304 |
+
"epoch": 0.9195402298850575,
|
| 305 |
+
"grad_norm": 4.96875,
|
| 306 |
+
"learning_rate": 2.9359792462981007e-05,
|
| 307 |
+
"loss": 1.6926,
|
| 308 |
+
"mean_token_accuracy": 0.5491850897669792,
|
| 309 |
+
"num_tokens": 54826.0,
|
| 310 |
"step": 30
|
| 311 |
},
|
| 312 |
{
|
| 313 |
+
"entropy": 2.110024333000183,
|
| 314 |
+
"epoch": 0.9501915708812261,
|
| 315 |
+
"grad_norm": 5.0,
|
| 316 |
+
"learning_rate": 2.9265847744427305e-05,
|
| 317 |
+
"loss": 1.6511,
|
| 318 |
+
"mean_token_accuracy": 0.5542715303599834,
|
| 319 |
+
"num_tokens": 56737.0,
|
| 320 |
"step": 31
|
| 321 |
},
|
| 322 |
{
|
| 323 |
+
"entropy": 2.1065904200077057,
|
| 324 |
+
"epoch": 0.9808429118773946,
|
| 325 |
+
"grad_norm": 6.1875,
|
| 326 |
+
"learning_rate": 2.9165645553562215e-05,
|
| 327 |
+
"loss": 1.6936,
|
| 328 |
+
"mean_token_accuracy": 0.5561113506555557,
|
| 329 |
+
"num_tokens": 58084.0,
|
| 330 |
"step": 32
|
| 331 |
},
|
| 332 |
{
|
| 333 |
+
"entropy": 1.8917590856552124,
|
| 334 |
+
"epoch": 1.0,
|
| 335 |
+
"grad_norm": 7.78125,
|
| 336 |
+
"learning_rate": 2.9059229842378373e-05,
|
| 337 |
+
"loss": 1.6156,
|
| 338 |
+
"mean_token_accuracy": 0.5620729327201843,
|
| 339 |
+
"num_tokens": 59142.0,
|
| 340 |
"step": 33
|
| 341 |
},
|
| 342 |
{
|
| 343 |
+
"entropy": 1.823348343372345,
|
| 344 |
+
"epoch": 1.0306513409961686,
|
| 345 |
+
"grad_norm": 4.5,
|
| 346 |
+
"learning_rate": 2.894664728832377e-05,
|
| 347 |
+
"loss": 1.2857,
|
| 348 |
+
"mean_token_accuracy": 0.6347417011857033,
|
| 349 |
+
"num_tokens": 60897.0,
|
| 350 |
"step": 34
|
| 351 |
},
|
| 352 |
{
|
| 353 |
+
"entropy": 1.9907677173614502,
|
| 354 |
+
"epoch": 1.0613026819923372,
|
| 355 |
+
"grad_norm": 5.5,
|
| 356 |
+
"learning_rate": 2.8827947273827508e-05,
|
| 357 |
+
"loss": 1.5295,
|
| 358 |
+
"mean_token_accuracy": 0.6137617453932762,
|
| 359 |
+
"num_tokens": 62325.0,
|
| 360 |
"step": 35
|
| 361 |
},
|
| 362 |
{
|
| 363 |
+
"entropy": 1.802370086312294,
|
| 364 |
+
"epoch": 1.0919540229885056,
|
| 365 |
+
"grad_norm": 4.46875,
|
| 366 |
+
"learning_rate": 2.8703181864639013e-05,
|
| 367 |
+
"loss": 1.2187,
|
| 368 |
+
"mean_token_accuracy": 0.645849384367466,
|
| 369 |
+
"num_tokens": 64163.0,
|
| 370 |
"step": 36
|
| 371 |
},
|
| 372 |
{
|
| 373 |
+
"entropy": 1.6892547607421875,
|
| 374 |
+
"epoch": 1.1226053639846743,
|
| 375 |
+
"grad_norm": 4.375,
|
| 376 |
+
"learning_rate": 2.8572405786990293e-05,
|
| 377 |
+
"loss": 1.2733,
|
| 378 |
+
"mean_token_accuracy": 0.6611653044819832,
|
| 379 |
+
"num_tokens": 66168.0,
|
| 380 |
"step": 37
|
| 381 |
},
|
| 382 |
{
|
| 383 |
+
"entropy": 1.7277430444955826,
|
| 384 |
+
"epoch": 1.1532567049808429,
|
| 385 |
+
"grad_norm": 4.28125,
|
| 386 |
+
"learning_rate": 2.8435676403591193e-05,
|
| 387 |
+
"loss": 1.2477,
|
| 388 |
+
"mean_token_accuracy": 0.6409317329525948,
|
| 389 |
+
"num_tokens": 68398.0,
|
| 390 |
"step": 38
|
| 391 |
},
|
| 392 |
{
|
| 393 |
+
"entropy": 1.6852583438158035,
|
| 394 |
+
"epoch": 1.1839080459770115,
|
| 395 |
+
"grad_norm": 4.8125,
|
| 396 |
+
"learning_rate": 2.829305368846822e-05,
|
| 397 |
+
"loss": 1.2649,
|
| 398 |
+
"mean_token_accuracy": 0.6238643229007721,
|
| 399 |
+
"num_tokens": 70548.0,
|
| 400 |
"step": 39
|
| 401 |
},
|
| 402 |
{
|
| 403 |
+
"entropy": 1.6644471734762192,
|
| 404 |
+
"epoch": 1.21455938697318,
|
| 405 |
+
"grad_norm": 5.40625,
|
| 406 |
+
"learning_rate": 2.8144600200657953e-05,
|
| 407 |
+
"loss": 1.2255,
|
| 408 |
+
"mean_token_accuracy": 0.6601320579648018,
|
| 409 |
+
"num_tokens": 72463.0,
|
| 410 |
"step": 40
|
| 411 |
},
|
| 412 |
{
|
| 413 |
+
"entropy": 1.5217097848653793,
|
| 414 |
+
"epoch": 1.2452107279693487,
|
| 415 |
+
"grad_norm": 5.71875,
|
| 416 |
+
"learning_rate": 2.7990381056766583e-05,
|
| 417 |
+
"loss": 1.1483,
|
| 418 |
+
"mean_token_accuracy": 0.6674150452017784,
|
| 419 |
+
"num_tokens": 74061.0,
|
| 420 |
"step": 41
|
| 421 |
},
|
| 422 |
{
|
| 423 |
+
"entropy": 1.5630977302789688,
|
| 424 |
+
"epoch": 1.2758620689655173,
|
| 425 |
+
"grad_norm": 4.8125,
|
| 426 |
+
"learning_rate": 2.78304639024076e-05,
|
| 427 |
+
"loss": 1.2095,
|
| 428 |
+
"mean_token_accuracy": 0.6475186944007874,
|
| 429 |
+
"num_tokens": 76130.0,
|
| 430 |
"step": 42
|
| 431 |
},
|
| 432 |
{
|
| 433 |
+
"entropy": 1.5357515215873718,
|
| 434 |
+
"epoch": 1.3065134099616857,
|
| 435 |
+
"grad_norm": 5.59375,
|
| 436 |
+
"learning_rate": 2.7664918882530227e-05,
|
| 437 |
+
"loss": 1.2226,
|
| 438 |
+
"mean_token_accuracy": 0.651666060090065,
|
| 439 |
+
"num_tokens": 77912.0,
|
| 440 |
"step": 43
|
| 441 |
},
|
| 442 |
{
|
| 443 |
+
"entropy": 1.447281613945961,
|
| 444 |
+
"epoch": 1.3371647509578544,
|
| 445 |
+
"grad_norm": 4.8125,
|
| 446 |
+
"learning_rate": 2.7493818610651493e-05,
|
| 447 |
+
"loss": 1.1148,
|
| 448 |
+
"mean_token_accuracy": 0.6739721521735191,
|
| 449 |
+
"num_tokens": 79873.0,
|
| 450 |
"step": 44
|
| 451 |
},
|
| 452 |
{
|
| 453 |
+
"entropy": 1.5578174591064453,
|
| 454 |
+
"epoch": 1.367816091954023,
|
| 455 |
+
"grad_norm": 5.46875,
|
| 456 |
+
"learning_rate": 2.731723813700556e-05,
|
| 457 |
+
"loss": 1.1376,
|
| 458 |
+
"mean_token_accuracy": 0.652525044977665,
|
| 459 |
+
"num_tokens": 81708.0,
|
| 460 |
"step": 45
|
| 461 |
},
|
| 462 |
{
|
| 463 |
+
"entropy": 1.509346380829811,
|
| 464 |
+
"epoch": 1.3984674329501916,
|
| 465 |
+
"grad_norm": 6.09375,
|
| 466 |
+
"learning_rate": 2.7135254915624213e-05,
|
| 467 |
+
"loss": 1.2312,
|
| 468 |
+
"mean_token_accuracy": 0.6451430022716522,
|
| 469 |
+
"num_tokens": 83479.0,
|
| 470 |
"step": 46
|
| 471 |
},
|
| 472 |
{
|
| 473 |
+
"entropy": 1.4848076701164246,
|
| 474 |
+
"epoch": 1.4291187739463602,
|
| 475 |
+
"grad_norm": 6.125,
|
| 476 |
+
"learning_rate": 2.6947948770362945e-05,
|
| 477 |
+
"loss": 1.2792,
|
| 478 |
+
"mean_token_accuracy": 0.6499253809452057,
|
| 479 |
+
"num_tokens": 85091.0,
|
| 480 |
"step": 47
|
| 481 |
},
|
| 482 |
{
|
| 483 |
+
"entropy": 1.5493847280740738,
|
| 484 |
+
"epoch": 1.4597701149425286,
|
| 485 |
+
"grad_norm": 5.03125,
|
| 486 |
+
"learning_rate": 2.6755401859887598e-05,
|
| 487 |
+
"loss": 1.152,
|
| 488 |
+
"mean_token_accuracy": 0.6553128436207771,
|
| 489 |
+
"num_tokens": 87067.0,
|
| 490 |
"step": 48
|
| 491 |
},
|
| 492 |
{
|
| 493 |
+
"entropy": 1.4846443086862564,
|
| 494 |
+
"epoch": 1.4904214559386972,
|
| 495 |
+
"grad_norm": 4.4375,
|
| 496 |
+
"learning_rate": 2.655769864163684e-05,
|
| 497 |
+
"loss": 1.0865,
|
| 498 |
+
"mean_token_accuracy": 0.6880389750003815,
|
| 499 |
+
"num_tokens": 89257.0,
|
| 500 |
"step": 49
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"entropy": 1.5293567329645157,
|
| 504 |
+
"epoch": 1.5210727969348659,
|
| 505 |
+
"grad_norm": 5.5,
|
| 506 |
+
"learning_rate": 2.6354925834776346e-05,
|
| 507 |
+
"loss": 1.2067,
|
| 508 |
+
"mean_token_accuracy": 0.6626974493265152,
|
| 509 |
+
"num_tokens": 91129.0,
|
| 510 |
+
"step": 50
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"entropy": 1.55410997569561,
|
| 514 |
+
"epoch": 1.5517241379310345,
|
| 515 |
+
"grad_norm": 5.15625,
|
| 516 |
+
"learning_rate": 2.6147172382160913e-05,
|
| 517 |
+
"loss": 1.1197,
|
| 518 |
+
"mean_token_accuracy": 0.6779765337705612,
|
| 519 |
+
"num_tokens": 93217.0,
|
| 520 |
+
"step": 51
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"entropy": 1.5660214275121689,
|
| 524 |
+
"epoch": 1.582375478927203,
|
| 525 |
+
"grad_norm": 5.59375,
|
| 526 |
+
"learning_rate": 2.5934529411321174e-05,
|
| 527 |
+
"loss": 1.0945,
|
| 528 |
+
"mean_token_accuracy": 0.660072922706604,
|
| 529 |
+
"num_tokens": 94882.0,
|
| 530 |
+
"step": 52
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"entropy": 1.6198417395353317,
|
| 534 |
+
"epoch": 1.6130268199233715,
|
| 535 |
+
"grad_norm": 6.34375,
|
| 536 |
+
"learning_rate": 2.571709019449205e-05,
|
| 537 |
+
"loss": 1.1716,
|
| 538 |
+
"mean_token_accuracy": 0.6578115820884705,
|
| 539 |
+
"num_tokens": 96226.0,
|
| 540 |
+
"step": 53
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"entropy": 1.5572579056024551,
|
| 544 |
+
"epoch": 1.6436781609195403,
|
| 545 |
+
"grad_norm": 5.5625,
|
| 546 |
+
"learning_rate": 2.5494950107700482e-05,
|
| 547 |
+
"loss": 1.138,
|
| 548 |
+
"mean_token_accuracy": 0.642712339758873,
|
| 549 |
+
"num_tokens": 98064.0,
|
| 550 |
+
"step": 54
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"entropy": 1.6491648256778717,
|
| 554 |
+
"epoch": 1.6743295019157087,
|
| 555 |
+
"grad_norm": 5.125,
|
| 556 |
+
"learning_rate": 2.5268206588930332e-05,
|
| 557 |
+
"loss": 1.1748,
|
| 558 |
+
"mean_token_accuracy": 0.6431401446461678,
|
| 559 |
+
"num_tokens": 99875.0,
|
| 560 |
+
"step": 55
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"entropy": 1.45476695895195,
|
| 564 |
+
"epoch": 1.7049808429118773,
|
| 565 |
+
"grad_norm": 4.625,
|
| 566 |
+
"learning_rate": 2.5036959095382875e-05,
|
| 567 |
+
"loss": 1.1288,
|
| 568 |
+
"mean_token_accuracy": 0.6704870611429214,
|
| 569 |
+
"num_tokens": 102303.0,
|
| 570 |
+
"step": 56
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"entropy": 1.5417988002300262,
|
| 574 |
+
"epoch": 1.735632183908046,
|
| 575 |
+
"grad_norm": 5.875,
|
| 576 |
+
"learning_rate": 2.4801309059851586e-05,
|
| 577 |
+
"loss": 1.1504,
|
| 578 |
+
"mean_token_accuracy": 0.6553002744913101,
|
| 579 |
+
"num_tokens": 103806.0,
|
| 580 |
+
"step": 57
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"entropy": 1.611013650894165,
|
| 584 |
+
"epoch": 1.7662835249042146,
|
| 585 |
+
"grad_norm": 7.0,
|
| 586 |
+
"learning_rate": 2.4561359846230346e-05,
|
| 587 |
+
"loss": 1.2064,
|
| 588 |
+
"mean_token_accuracy": 0.6713609769940376,
|
| 589 |
+
"num_tokens": 105012.0,
|
| 590 |
+
"step": 58
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"entropy": 1.683053657412529,
|
| 594 |
+
"epoch": 1.7969348659003832,
|
| 595 |
+
"grad_norm": 6.15625,
|
| 596 |
+
"learning_rate": 2.4317216704174656e-05,
|
| 597 |
+
"loss": 1.2315,
|
| 598 |
+
"mean_token_accuracy": 0.6650961637496948,
|
| 599 |
+
"num_tokens": 106373.0,
|
| 600 |
+
"step": 59
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"entropy": 1.6171066462993622,
|
| 604 |
+
"epoch": 1.8275862068965516,
|
| 605 |
+
"grad_norm": 5.5625,
|
| 606 |
+
"learning_rate": 2.4068986722935625e-05,
|
| 607 |
+
"loss": 1.1285,
|
| 608 |
+
"mean_token_accuracy": 0.6854713410139084,
|
| 609 |
+
"num_tokens": 107938.0,
|
| 610 |
+
"step": 60
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"entropy": 1.5850916504859924,
|
| 614 |
+
"epoch": 1.8582375478927204,
|
| 615 |
+
"grad_norm": 5.625,
|
| 616 |
+
"learning_rate": 2.3816778784387097e-05,
|
| 617 |
+
"loss": 1.1496,
|
| 618 |
+
"mean_token_accuracy": 0.6636858507990837,
|
| 619 |
+
"num_tokens": 109587.0,
|
| 620 |
+
"step": 61
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"entropy": 1.6505294740200043,
|
| 624 |
+
"epoch": 1.8888888888888888,
|
| 625 |
+
"grad_norm": 6.46875,
|
| 626 |
+
"learning_rate": 2.356070351526648e-05,
|
| 627 |
+
"loss": 1.4317,
|
| 628 |
+
"mean_token_accuracy": 0.6071530058979988,
|
| 629 |
+
"num_tokens": 111253.0,
|
| 630 |
+
"step": 62
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"entropy": 1.5908723175525665,
|
| 634 |
+
"epoch": 1.9195402298850575,
|
| 635 |
+
"grad_norm": 6.34375,
|
| 636 |
+
"learning_rate": 2.3300873238650163e-05,
|
| 637 |
+
"loss": 1.2333,
|
| 638 |
+
"mean_token_accuracy": 0.6693138256669044,
|
| 639 |
+
"num_tokens": 112833.0,
|
| 640 |
+
"step": 63
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"entropy": 1.5132576674222946,
|
| 644 |
+
"epoch": 1.950191570881226,
|
| 645 |
+
"grad_norm": 5.84375,
|
| 646 |
+
"learning_rate": 2.303740192468495e-05,
|
| 647 |
+
"loss": 1.2285,
|
| 648 |
+
"mean_token_accuracy": 0.651925578713417,
|
| 649 |
+
"num_tokens": 114573.0,
|
| 650 |
+
"step": 64
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"entropy": 1.5014029294252396,
|
| 654 |
+
"epoch": 1.9808429118773945,
|
| 655 |
+
"grad_norm": 4.9375,
|
| 656 |
+
"learning_rate": 2.2770405140596954e-05,
|
| 657 |
+
"loss": 1.1041,
|
| 658 |
+
"mean_token_accuracy": 0.6395103111863136,
|
| 659 |
+
"num_tokens": 116981.0,
|
| 660 |
+
"step": 65
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"entropy": 1.5801513910293579,
|
| 664 |
+
"epoch": 2.0,
|
| 665 |
+
"grad_norm": 7.21875,
|
| 666 |
+
"learning_rate": 2.25e-05,
|
| 667 |
+
"loss": 1.387,
|
| 668 |
+
"mean_token_accuracy": 0.6241851329803467,
|
| 669 |
+
"num_tokens": 118284.0,
|
| 670 |
+
"step": 66
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"entropy": 1.5819022357463837,
|
| 674 |
+
"epoch": 2.0306513409961684,
|
| 675 |
+
"grad_norm": 5.5,
|
| 676 |
+
"learning_rate": 2.222630511152573e-05,
|
| 677 |
+
"loss": 0.8186,
|
| 678 |
+
"mean_token_accuracy": 0.7697352692484856,
|
| 679 |
+
"num_tokens": 119663.0,
|
| 680 |
+
"step": 67
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"entropy": 1.473861888051033,
|
| 684 |
+
"epoch": 2.0613026819923372,
|
| 685 |
+
"grad_norm": 4.46875,
|
| 686 |
+
"learning_rate": 2.1949440526797928e-05,
|
| 687 |
+
"loss": 0.7233,
|
| 688 |
+
"mean_token_accuracy": 0.7909102663397789,
|
| 689 |
+
"num_tokens": 121476.0,
|
| 690 |
+
"step": 68
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"entropy": 1.4712600708007812,
|
| 694 |
+
"epoch": 2.0919540229885056,
|
| 695 |
+
"grad_norm": 4.15625,
|
| 696 |
+
"learning_rate": 2.166952768777391e-05,
|
| 697 |
+
"loss": 0.7592,
|
| 698 |
+
"mean_token_accuracy": 0.7729948312044144,
|
| 699 |
+
"num_tokens": 123792.0,
|
| 700 |
+
"step": 69
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"entropy": 1.4692800045013428,
|
| 704 |
+
"epoch": 2.1226053639846745,
|
| 705 |
+
"grad_norm": 4.96875,
|
| 706 |
+
"learning_rate": 2.138668937347609e-05,
|
| 707 |
+
"loss": 0.7338,
|
| 708 |
+
"mean_token_accuracy": 0.7888433858752251,
|
| 709 |
+
"num_tokens": 125329.0,
|
| 710 |
+
"step": 70
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"entropy": 1.4230495542287827,
|
| 714 |
+
"epoch": 2.153256704980843,
|
| 715 |
+
"grad_norm": 5.4375,
|
| 716 |
+
"learning_rate": 2.1101049646137008e-05,
|
| 717 |
+
"loss": 0.7094,
|
| 718 |
+
"mean_token_accuracy": 0.7864886075258255,
|
| 719 |
+
"num_tokens": 126654.0,
|
| 720 |
+
"step": 71
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"entropy": 1.2409694343805313,
|
| 724 |
+
"epoch": 2.1839080459770113,
|
| 725 |
+
"grad_norm": 4.6875,
|
| 726 |
+
"learning_rate": 2.0812733796781544e-05,
|
| 727 |
+
"loss": 0.6902,
|
| 728 |
+
"mean_token_accuracy": 0.7980181872844696,
|
| 729 |
+
"num_tokens": 128629.0,
|
| 730 |
+
"step": 72
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"entropy": 1.241740956902504,
|
| 734 |
+
"epoch": 2.21455938697318,
|
| 735 |
+
"grad_norm": 5.90625,
|
| 736 |
+
"learning_rate": 2.052186829027017e-05,
|
| 737 |
+
"loss": 0.8967,
|
| 738 |
+
"mean_token_accuracy": 0.7425272911787033,
|
| 739 |
+
"num_tokens": 130344.0,
|
| 740 |
+
"step": 73
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"entropy": 1.15965236723423,
|
| 744 |
+
"epoch": 2.2452107279693485,
|
| 745 |
+
"grad_norm": 5.625,
|
| 746 |
+
"learning_rate": 2.022858070982723e-05,
|
| 747 |
+
"loss": 0.728,
|
| 748 |
+
"mean_token_accuracy": 0.792903944849968,
|
| 749 |
+
"num_tokens": 132152.0,
|
| 750 |
+
"step": 74
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"entropy": 1.1466698423027992,
|
| 754 |
+
"epoch": 2.2758620689655173,
|
| 755 |
+
"grad_norm": 5.28125,
|
| 756 |
+
"learning_rate": 1.993299970107875e-05,
|
| 757 |
+
"loss": 0.6783,
|
| 758 |
+
"mean_token_accuracy": 0.7755344063043594,
|
| 759 |
+
"num_tokens": 134144.0,
|
| 760 |
+
"step": 75
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"entropy": 1.0256517007946968,
|
| 764 |
+
"epoch": 2.3065134099616857,
|
| 765 |
+
"grad_norm": 5.65625,
|
| 766 |
+
"learning_rate": 1.963525491562421e-05,
|
| 767 |
+
"loss": 0.7747,
|
| 768 |
+
"mean_token_accuracy": 0.7711224406957626,
|
| 769 |
+
"num_tokens": 136715.0,
|
| 770 |
+
"step": 76
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"entropy": 1.0880693718791008,
|
| 774 |
+
"epoch": 2.3371647509578546,
|
| 775 |
+
"grad_norm": 5.96875,
|
| 776 |
+
"learning_rate": 1.9335476954167073e-05,
|
| 777 |
+
"loss": 0.7357,
|
| 778 |
+
"mean_token_accuracy": 0.7778737023472786,
|
| 779 |
+
"num_tokens": 138907.0,
|
| 780 |
+
"step": 77
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"entropy": 1.0792594850063324,
|
| 784 |
+
"epoch": 2.367816091954023,
|
| 785 |
+
"grad_norm": 6.875,
|
| 786 |
+
"learning_rate": 1.9033797309228984e-05,
|
| 787 |
+
"loss": 0.622,
|
| 788 |
+
"mean_token_accuracy": 0.8023911938071251,
|
| 789 |
+
"num_tokens": 140536.0,
|
| 790 |
+
"step": 78
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"entropy": 1.047565944492817,
|
| 794 |
+
"epoch": 2.3984674329501914,
|
| 795 |
+
"grad_norm": 7.90625,
|
| 796 |
+
"learning_rate": 1.8730348307472828e-05,
|
| 797 |
+
"loss": 0.6963,
|
| 798 |
+
"mean_token_accuracy": 0.8018106818199158,
|
| 799 |
+
"num_tokens": 142205.0,
|
| 800 |
+
"step": 79
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"entropy": 1.0534571334719658,
|
| 804 |
+
"epoch": 2.42911877394636,
|
| 805 |
+
"grad_norm": 7.09375,
|
| 806 |
+
"learning_rate": 1.8425263051659838e-05,
|
| 807 |
+
"loss": 0.6769,
|
| 808 |
+
"mean_token_accuracy": 0.7829258516430855,
|
| 809 |
+
"num_tokens": 143937.0,
|
| 810 |
+
"step": 80
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"entropy": 1.06615050137043,
|
| 814 |
+
"epoch": 2.4597701149425286,
|
| 815 |
+
"grad_norm": 6.96875,
|
| 816 |
+
"learning_rate": 1.8118675362266388e-05,
|
| 817 |
+
"loss": 0.724,
|
| 818 |
+
"mean_token_accuracy": 0.7941268905997276,
|
| 819 |
+
"num_tokens": 145867.0,
|
| 820 |
+
"step": 81
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"entropy": 1.076167106628418,
|
| 824 |
+
"epoch": 2.4904214559386975,
|
| 825 |
+
"grad_norm": 7.71875,
|
| 826 |
+
"learning_rate": 1.781071971878587e-05,
|
| 827 |
+
"loss": 0.7821,
|
| 828 |
+
"mean_token_accuracy": 0.774849958717823,
|
| 829 |
+
"num_tokens": 147522.0,
|
| 830 |
+
"step": 82
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"entropy": 1.1518089175224304,
|
| 834 |
+
"epoch": 2.521072796934866,
|
| 835 |
+
"grad_norm": 8.3125,
|
| 836 |
+
"learning_rate": 1.7501531200741536e-05,
|
| 837 |
+
"loss": 0.7973,
|
| 838 |
+
"mean_token_accuracy": 0.7875072434544563,
|
| 839 |
+
"num_tokens": 149154.0,
|
| 840 |
+
"step": 83
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"entropy": 1.036544032394886,
|
| 844 |
+
"epoch": 2.5517241379310347,
|
| 845 |
+
"grad_norm": 5.9375,
|
| 846 |
+
"learning_rate": 1.7191245428436175e-05,
|
| 847 |
+
"loss": 0.7165,
|
| 848 |
+
"mean_token_accuracy": 0.8045910075306892,
|
| 849 |
+
"num_tokens": 151443.0,
|
| 850 |
+
"step": 84
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"entropy": 1.1293774545192719,
|
| 854 |
+
"epoch": 2.582375478927203,
|
| 855 |
+
"grad_norm": 6.125,
|
| 856 |
+
"learning_rate": 1.6879998503464565e-05,
|
| 857 |
+
"loss": 0.6904,
|
| 858 |
+
"mean_token_accuracy": 0.8081322461366653,
|
| 859 |
+
"num_tokens": 153346.0,
|
| 860 |
+
"step": 85
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"entropy": 1.1728527545928955,
|
| 864 |
+
"epoch": 2.6130268199233715,
|
| 865 |
+
"grad_norm": 6.46875,
|
| 866 |
+
"learning_rate": 1.6567926949014805e-05,
|
| 867 |
+
"loss": 0.6986,
|
| 868 |
+
"mean_token_accuracy": 0.7757782712578773,
|
| 869 |
+
"num_tokens": 155041.0,
|
| 870 |
+
"step": 86
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"entropy": 1.0829114317893982,
|
| 874 |
+
"epoch": 2.6436781609195403,
|
| 875 |
+
"grad_norm": 6.21875,
|
| 876 |
+
"learning_rate": 1.6255167649984738e-05,
|
| 877 |
+
"loss": 0.6176,
|
| 878 |
+
"mean_token_accuracy": 0.8206781521439552,
|
| 879 |
+
"num_tokens": 156508.0,
|
| 880 |
+
"step": 87
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"entropy": 1.1221561282873154,
|
| 884 |
+
"epoch": 2.6743295019157087,
|
| 885 |
+
"grad_norm": 5.71875,
|
| 886 |
+
"learning_rate": 1.5941857792939702e-05,
|
| 887 |
+
"loss": 0.6748,
|
| 888 |
+
"mean_token_accuracy": 0.7972528710961342,
|
| 889 |
+
"num_tokens": 158506.0,
|
| 890 |
+
"step": 88
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"entropy": 1.1622852012515068,
|
| 894 |
+
"epoch": 2.704980842911877,
|
| 895 |
+
"grad_norm": 5.4375,
|
| 896 |
+
"learning_rate": 1.5628134805937995e-05,
|
| 897 |
+
"loss": 0.6891,
|
| 898 |
+
"mean_token_accuracy": 0.7869931757450104,
|
| 899 |
+
"num_tokens": 160511.0,
|
| 900 |
+
"step": 89
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"entropy": 1.1033327355980873,
|
| 904 |
+
"epoch": 2.735632183908046,
|
| 905 |
+
"grad_norm": 17.0,
|
| 906 |
+
"learning_rate": 1.5314136298250355e-05,
|
| 907 |
+
"loss": 0.6848,
|
| 908 |
+
"mean_token_accuracy": 0.7839160785079002,
|
| 909 |
+
"num_tokens": 162548.0,
|
| 910 |
+
"step": 90
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"entropy": 1.0911096036434174,
|
| 914 |
+
"epoch": 2.766283524904215,
|
| 915 |
+
"grad_norm": 5.03125,
|
| 916 |
+
"learning_rate": 1.5e-05,
|
| 917 |
+
"loss": 0.7139,
|
| 918 |
+
"mean_token_accuracy": 0.7690701261162758,
|
| 919 |
+
"num_tokens": 164874.0,
|
| 920 |
+
"step": 91
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"entropy": 1.0845600441098213,
|
| 924 |
+
"epoch": 2.796934865900383,
|
| 925 |
+
"grad_norm": 5.9375,
|
| 926 |
+
"learning_rate": 1.4685863701749648e-05,
|
| 927 |
+
"loss": 0.6626,
|
| 928 |
+
"mean_token_accuracy": 0.8035678789019585,
|
| 929 |
+
"num_tokens": 166614.0,
|
| 930 |
+
"step": 92
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"entropy": 1.1177793815732002,
|
| 934 |
+
"epoch": 2.8275862068965516,
|
| 935 |
+
"grad_norm": 7.0,
|
| 936 |
+
"learning_rate": 1.4371865194062009e-05,
|
| 937 |
+
"loss": 0.776,
|
| 938 |
+
"mean_token_accuracy": 0.7617417871952057,
|
| 939 |
+
"num_tokens": 168220.0,
|
| 940 |
+
"step": 93
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"entropy": 1.1276764497160912,
|
| 944 |
+
"epoch": 2.8582375478927204,
|
| 945 |
+
"grad_norm": 8.8125,
|
| 946 |
+
"learning_rate": 1.40581422070603e-05,
|
| 947 |
+
"loss": 0.6741,
|
| 948 |
+
"mean_token_accuracy": 0.7929504066705704,
|
| 949 |
+
"num_tokens": 169540.0,
|
| 950 |
+
"step": 94
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"entropy": 1.072372280061245,
|
| 954 |
+
"epoch": 2.888888888888889,
|
| 955 |
+
"grad_norm": 4.84375,
|
| 956 |
+
"learning_rate": 1.3744832350015271e-05,
|
| 957 |
+
"loss": 0.6173,
|
| 958 |
+
"mean_token_accuracy": 0.8033151254057884,
|
| 959 |
+
"num_tokens": 171709.0,
|
| 960 |
+
"step": 95
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"entropy": 1.0452390983700752,
|
| 964 |
+
"epoch": 2.9195402298850572,
|
| 965 |
+
"grad_norm": 6.0,
|
| 966 |
+
"learning_rate": 1.3432073050985201e-05,
|
| 967 |
+
"loss": 0.6426,
|
| 968 |
+
"mean_token_accuracy": 0.8162983357906342,
|
| 969 |
+
"num_tokens": 173432.0,
|
| 970 |
+
"step": 96
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"entropy": 1.0664306953549385,
|
| 974 |
+
"epoch": 2.950191570881226,
|
| 975 |
+
"grad_norm": 7.125,
|
| 976 |
+
"learning_rate": 1.3120001496535434e-05,
|
| 977 |
+
"loss": 0.6244,
|
| 978 |
+
"mean_token_accuracy": 0.8141586259007454,
|
| 979 |
+
"num_tokens": 174962.0,
|
| 980 |
+
"step": 97
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"entropy": 1.0960277691483498,
|
| 984 |
+
"epoch": 2.9808429118773945,
|
| 985 |
+
"grad_norm": 6.6875,
|
| 986 |
+
"learning_rate": 1.2808754571563827e-05,
|
| 987 |
+
"loss": 0.7344,
|
| 988 |
+
"mean_token_accuracy": 0.7819336354732513,
|
| 989 |
+
"num_tokens": 176611.0,
|
| 990 |
+
"step": 98
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"entropy": 1.068971049785614,
|
| 994 |
+
"epoch": 3.0,
|
| 995 |
+
"grad_norm": 8.5,
|
| 996 |
+
"learning_rate": 1.2498468799258468e-05,
|
| 997 |
+
"loss": 0.5678,
|
| 998 |
+
"mean_token_accuracy": 0.8390557408332825,
|
| 999 |
+
"num_tokens": 177426.0,
|
| 1000 |
+
"step": 99
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"entropy": 1.0476091727614403,
|
| 1004 |
+
"epoch": 3.0306513409961684,
|
| 1005 |
+
"grad_norm": 5.21875,
|
| 1006 |
+
"learning_rate": 1.2189280281214128e-05,
|
| 1007 |
+
"loss": 0.4454,
|
| 1008 |
+
"mean_token_accuracy": 0.8829296976327896,
|
| 1009 |
+
"num_tokens": 179048.0,
|
| 1010 |
+
"step": 100
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"entropy": 1.0911573767662048,
|
| 1014 |
+
"epoch": 3.0613026819923372,
|
| 1015 |
+
"grad_norm": 4.71875,
|
| 1016 |
+
"learning_rate": 1.1881324637733613e-05,
|
| 1017 |
+
"loss": 0.4992,
|
| 1018 |
+
"mean_token_accuracy": 0.8481416404247284,
|
| 1019 |
+
"num_tokens": 181115.0,
|
| 1020 |
+
"step": 101
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"entropy": 1.1437290906906128,
|
| 1024 |
+
"epoch": 3.0919540229885056,
|
| 1025 |
+
"grad_norm": 5.625,
|
| 1026 |
+
"learning_rate": 1.1574736948340163e-05,
|
| 1027 |
+
"loss": 0.5052,
|
| 1028 |
+
"mean_token_accuracy": 0.8615396842360497,
|
| 1029 |
+
"num_tokens": 182578.0,
|
| 1030 |
+
"step": 102
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"entropy": 1.0376266986131668,
|
| 1034 |
+
"epoch": 3.1226053639846745,
|
| 1035 |
+
"grad_norm": 5.5,
|
| 1036 |
+
"learning_rate": 1.1269651692527181e-05,
|
| 1037 |
+
"loss": 0.4683,
|
| 1038 |
+
"mean_token_accuracy": 0.870830163359642,
|
| 1039 |
+
"num_tokens": 184376.0,
|
| 1040 |
+
"step": 103
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"entropy": 0.9589341431856155,
|
| 1044 |
+
"epoch": 3.153256704980843,
|
| 1045 |
+
"grad_norm": 4.46875,
|
| 1046 |
+
"learning_rate": 1.0966202690771015e-05,
|
| 1047 |
+
"loss": 0.4558,
|
| 1048 |
+
"mean_token_accuracy": 0.8758394122123718,
|
| 1049 |
+
"num_tokens": 186551.0,
|
| 1050 |
+
"step": 104
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"entropy": 0.9614228904247284,
|
| 1054 |
+
"epoch": 3.1839080459770113,
|
| 1055 |
+
"grad_norm": 5.34375,
|
| 1056 |
+
"learning_rate": 1.0664523045832923e-05,
|
| 1057 |
+
"loss": 0.3656,
|
| 1058 |
+
"mean_token_accuracy": 0.9130250364542007,
|
| 1059 |
+
"num_tokens": 187965.0,
|
| 1060 |
+
"step": 105
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"entropy": 0.927314467728138,
|
| 1064 |
+
"epoch": 3.21455938697318,
|
| 1065 |
+
"grad_norm": 4.8125,
|
| 1066 |
+
"learning_rate": 1.036474508437579e-05,
|
| 1067 |
+
"loss": 0.3549,
|
| 1068 |
+
"mean_token_accuracy": 0.9026911705732346,
|
| 1069 |
+
"num_tokens": 189763.0,
|
| 1070 |
+
"step": 106
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"entropy": 0.8468578010797501,
|
| 1074 |
+
"epoch": 3.2452107279693485,
|
| 1075 |
+
"grad_norm": 5.375,
|
| 1076 |
+
"learning_rate": 1.0067000298921251e-05,
|
| 1077 |
+
"loss": 0.3705,
|
| 1078 |
+
"mean_token_accuracy": 0.8972815871238708,
|
| 1079 |
+
"num_tokens": 191586.0,
|
| 1080 |
+
"step": 107
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"entropy": 0.847766101360321,
|
| 1084 |
+
"epoch": 3.2758620689655173,
|
| 1085 |
+
"grad_norm": 5.53125,
|
| 1086 |
+
"learning_rate": 9.771419290172776e-06,
|
| 1087 |
+
"loss": 0.3128,
|
| 1088 |
+
"mean_token_accuracy": 0.9080770388245583,
|
| 1089 |
+
"num_tokens": 193154.0,
|
| 1090 |
+
"step": 108
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"entropy": 0.8512594848871231,
|
| 1094 |
+
"epoch": 3.3065134099616857,
|
| 1095 |
+
"grad_norm": 6.59375,
|
| 1096 |
+
"learning_rate": 9.478131709729831e-06,
|
| 1097 |
+
"loss": 0.4468,
|
| 1098 |
+
"mean_token_accuracy": 0.8701968789100647,
|
| 1099 |
+
"num_tokens": 194813.0,
|
| 1100 |
+
"step": 109
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"entropy": 0.85344647616148,
|
| 1104 |
+
"epoch": 3.3371647509578546,
|
| 1105 |
+
"grad_norm": 5.125,
|
| 1106 |
+
"learning_rate": 9.187266203218457e-06,
|
| 1107 |
+
"loss": 0.4125,
|
| 1108 |
+
"mean_token_accuracy": 0.8799988329410553,
|
| 1109 |
+
"num_tokens": 196957.0,
|
| 1110 |
+
"step": 110
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"entropy": 0.8800590857863426,
|
| 1114 |
+
"epoch": 3.367816091954023,
|
| 1115 |
+
"grad_norm": 7.53125,
|
| 1116 |
+
"learning_rate": 8.898950353863e-06,
|
| 1117 |
+
"loss": 0.3593,
|
| 1118 |
+
"mean_token_accuracy": 0.8787510544061661,
|
| 1119 |
+
"num_tokens": 198398.0,
|
| 1120 |
+
"step": 111
|
| 1121 |
+
},
|
| 1122 |
+
{
|
| 1123 |
+
"entropy": 0.7779126092791557,
|
| 1124 |
+
"epoch": 3.3984674329501914,
|
| 1125 |
+
"grad_norm": 6.21875,
|
| 1126 |
+
"learning_rate": 8.61331062652391e-06,
|
| 1127 |
+
"loss": 0.3267,
|
| 1128 |
+
"mean_token_accuracy": 0.9116306900978088,
|
| 1129 |
+
"num_tokens": 200136.0,
|
| 1130 |
+
"step": 112
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"entropy": 0.7932204827666283,
|
| 1134 |
+
"epoch": 3.42911877394636,
|
| 1135 |
+
"grad_norm": 6.96875,
|
| 1136 |
+
"learning_rate": 8.330472312226091e-06,
|
| 1137 |
+
"loss": 0.3873,
|
| 1138 |
+
"mean_token_accuracy": 0.887348584830761,
|
| 1139 |
+
"num_tokens": 201989.0,
|
| 1140 |
+
"step": 113
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"entropy": 0.6965708956122398,
|
| 1144 |
+
"epoch": 3.4597701149425286,
|
| 1145 |
+
"grad_norm": 13.0,
|
| 1146 |
+
"learning_rate": 8.050559473202078e-06,
|
| 1147 |
+
"loss": 0.3972,
|
| 1148 |
+
"mean_token_accuracy": 0.890875369310379,
|
| 1149 |
+
"num_tokens": 203992.0,
|
| 1150 |
+
"step": 114
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"entropy": 0.7952002212405205,
|
| 1154 |
+
"epoch": 3.4904214559386975,
|
| 1155 |
+
"grad_norm": 10.0,
|
| 1156 |
+
"learning_rate": 7.773694888474268e-06,
|
| 1157 |
+
"loss": 0.4792,
|
| 1158 |
+
"mean_token_accuracy": 0.8768180310726166,
|
| 1159 |
+
"num_tokens": 205854.0,
|
| 1160 |
+
"step": 115
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"entropy": 0.7920583859086037,
|
| 1164 |
+
"epoch": 3.521072796934866,
|
| 1165 |
+
"grad_norm": 7.25,
|
| 1166 |
+
"learning_rate": 7.500000000000004e-06,
|
| 1167 |
+
"loss": 0.3583,
|
| 1168 |
+
"mean_token_accuracy": 0.8890090510249138,
|
| 1169 |
+
"num_tokens": 207815.0,
|
| 1170 |
+
"step": 116
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"entropy": 0.7490416169166565,
|
| 1174 |
+
"epoch": 3.5517241379310347,
|
| 1175 |
+
"grad_norm": 6.09375,
|
| 1176 |
+
"learning_rate": 7.229594859403049e-06,
|
| 1177 |
+
"loss": 0.3908,
|
| 1178 |
+
"mean_token_accuracy": 0.8729168474674225,
|
| 1179 |
+
"num_tokens": 209918.0,
|
| 1180 |
+
"step": 117
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"entropy": 0.6982008591294289,
|
| 1184 |
+
"epoch": 3.582375478927203,
|
| 1185 |
+
"grad_norm": 7.9375,
|
| 1186 |
+
"learning_rate": 6.962598075315047e-06,
|
| 1187 |
+
"loss": 0.3882,
|
| 1188 |
+
"mean_token_accuracy": 0.8720036372542381,
|
| 1189 |
+
"num_tokens": 211772.0,
|
| 1190 |
+
"step": 118
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"entropy": 0.7215415760874748,
|
| 1194 |
+
"epoch": 3.6130268199233715,
|
| 1195 |
+
"grad_norm": 7.3125,
|
| 1196 |
+
"learning_rate": 6.699126761349841e-06,
|
| 1197 |
+
"loss": 0.3885,
|
| 1198 |
+
"mean_token_accuracy": 0.8832474499940872,
|
| 1199 |
+
"num_tokens": 213561.0,
|
| 1200 |
+
"step": 119
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"entropy": 0.7576938644051552,
|
| 1204 |
+
"epoch": 3.6436781609195403,
|
| 1205 |
+
"grad_norm": 6.4375,
|
| 1206 |
+
"learning_rate": 6.439296484733526e-06,
|
| 1207 |
+
"loss": 0.3824,
|
| 1208 |
+
"mean_token_accuracy": 0.8854040205478668,
|
| 1209 |
+
"num_tokens": 215335.0,
|
| 1210 |
+
"step": 120
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"entropy": 0.6329802498221397,
|
| 1214 |
+
"epoch": 3.6743295019157087,
|
| 1215 |
+
"grad_norm": 4.5,
|
| 1216 |
+
"learning_rate": 6.1832212156129045e-06,
|
| 1217 |
+
"loss": 0.3047,
|
| 1218 |
+
"mean_token_accuracy": 0.9093970954418182,
|
| 1219 |
+
"num_tokens": 218204.0,
|
| 1220 |
+
"step": 121
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"entropy": 0.7503926008939743,
|
| 1224 |
+
"epoch": 3.704980842911877,
|
| 1225 |
+
"grad_norm": 6.28125,
|
| 1226 |
+
"learning_rate": 5.931013277064377e-06,
|
| 1227 |
+
"loss": 0.3702,
|
| 1228 |
+
"mean_token_accuracy": 0.8855115100741386,
|
| 1229 |
+
"num_tokens": 220090.0,
|
| 1230 |
+
"step": 122
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"entropy": 0.7402322217822075,
|
| 1234 |
+
"epoch": 3.735632183908046,
|
| 1235 |
+
"grad_norm": 9.125,
|
| 1236 |
+
"learning_rate": 5.682783295825346e-06,
|
| 1237 |
+
"loss": 0.3685,
|
| 1238 |
+
"mean_token_accuracy": 0.8846122622489929,
|
| 1239 |
+
"num_tokens": 221667.0,
|
| 1240 |
+
"step": 123
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"entropy": 0.8042828142642975,
|
| 1244 |
+
"epoch": 3.766283524904215,
|
| 1245 |
+
"grad_norm": 8.375,
|
| 1246 |
+
"learning_rate": 5.438640153769654e-06,
|
| 1247 |
+
"loss": 0.3649,
|
| 1248 |
+
"mean_token_accuracy": 0.9067123532295227,
|
| 1249 |
+
"num_tokens": 222962.0,
|
| 1250 |
+
"step": 124
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"entropy": 0.7551866173744202,
|
| 1254 |
+
"epoch": 3.796934865900383,
|
| 1255 |
+
"grad_norm": 5.875,
|
| 1256 |
+
"learning_rate": 5.198690940148418e-06,
|
| 1257 |
+
"loss": 0.3998,
|
| 1258 |
+
"mean_token_accuracy": 0.8821066841483116,
|
| 1259 |
+
"num_tokens": 224965.0,
|
| 1260 |
+
"step": 125
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"entropy": 0.8327518254518509,
|
| 1264 |
+
"epoch": 3.8275862068965516,
|
| 1265 |
+
"grad_norm": 6.78125,
|
| 1266 |
+
"learning_rate": 4.963040904617131e-06,
|
| 1267 |
+
"loss": 0.3778,
|
| 1268 |
+
"mean_token_accuracy": 0.8923891186714172,
|
| 1269 |
+
"num_tokens": 226522.0,
|
| 1270 |
+
"step": 126
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"entropy": 0.7820432782173157,
|
| 1274 |
+
"epoch": 3.8582375478927204,
|
| 1275 |
+
"grad_norm": 5.1875,
|
| 1276 |
+
"learning_rate": 4.731793411069669e-06,
|
| 1277 |
+
"loss": 0.4173,
|
| 1278 |
+
"mean_token_accuracy": 0.8775044828653336,
|
| 1279 |
+
"num_tokens": 229172.0,
|
| 1280 |
+
"step": 127
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"entropy": 0.8462389037013054,
|
| 1284 |
+
"epoch": 3.888888888888889,
|
| 1285 |
+
"grad_norm": 6.53125,
|
| 1286 |
+
"learning_rate": 4.505049892299517e-06,
|
| 1287 |
+
"loss": 0.2995,
|
| 1288 |
+
"mean_token_accuracy": 0.9146698862314224,
|
| 1289 |
+
"num_tokens": 230464.0,
|
| 1290 |
+
"step": 128
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"entropy": 0.8228934556245804,
|
| 1294 |
+
"epoch": 3.9195402298850572,
|
| 1295 |
+
"grad_norm": 6.75,
|
| 1296 |
+
"learning_rate": 4.282909805507953e-06,
|
| 1297 |
+
"loss": 0.3327,
|
| 1298 |
+
"mean_token_accuracy": 0.9099200069904327,
|
| 1299 |
+
"num_tokens": 231921.0,
|
| 1300 |
+
"step": 129
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"entropy": 0.8176594227552414,
|
| 1304 |
+
"epoch": 3.950191570881226,
|
| 1305 |
+
"grad_norm": 8.1875,
|
| 1306 |
+
"learning_rate": 4.06547058867883e-06,
|
| 1307 |
+
"loss": 0.3909,
|
| 1308 |
+
"mean_token_accuracy": 0.8916172161698341,
|
| 1309 |
+
"num_tokens": 233200.0,
|
| 1310 |
+
"step": 130
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"entropy": 0.7947382852435112,
|
| 1314 |
+
"epoch": 3.9808429118773945,
|
| 1315 |
+
"grad_norm": 4.90625,
|
| 1316 |
+
"learning_rate": 3.852827617839085e-06,
|
| 1317 |
+
"loss": 0.3452,
|
| 1318 |
+
"mean_token_accuracy": 0.8946295380592346,
|
| 1319 |
+
"num_tokens": 235153.0,
|
| 1320 |
+
"step": 131
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"entropy": 0.7858668446540833,
|
| 1324 |
+
"epoch": 4.0,
|
| 1325 |
+
"grad_norm": 7.125,
|
| 1326 |
+
"learning_rate": 3.645074165223656e-06,
|
| 1327 |
+
"loss": 0.5253,
|
| 1328 |
+
"mean_token_accuracy": 0.8538177251815796,
|
| 1329 |
+
"num_tokens": 236568.0,
|
| 1330 |
+
"step": 132
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"entropy": 0.8421337753534317,
|
| 1334 |
+
"epoch": 4.030651340996169,
|
| 1335 |
+
"grad_norm": 5.03125,
|
| 1336 |
+
"learning_rate": 3.442301358363163e-06,
|
| 1337 |
+
"loss": 0.2725,
|
| 1338 |
+
"mean_token_accuracy": 0.9301287308335304,
|
| 1339 |
+
"num_tokens": 238229.0,
|
| 1340 |
+
"step": 133
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"entropy": 0.8355702385306358,
|
| 1344 |
+
"epoch": 4.061302681992337,
|
| 1345 |
+
"grad_norm": 5.03125,
|
| 1346 |
+
"learning_rate": 3.244598140112404e-06,
|
| 1347 |
+
"loss": 0.2914,
|
| 1348 |
+
"mean_token_accuracy": 0.9221316650509834,
|
| 1349 |
+
"num_tokens": 239958.0,
|
| 1350 |
+
"step": 134
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"entropy": 0.7881455570459366,
|
| 1354 |
+
"epoch": 4.091954022988506,
|
| 1355 |
+
"grad_norm": 4.84375,
|
| 1356 |
+
"learning_rate": 3.052051229637056e-06,
|
| 1357 |
+
"loss": 0.3885,
|
| 1358 |
+
"mean_token_accuracy": 0.9120407700538635,
|
| 1359 |
+
"num_tokens": 242097.0,
|
| 1360 |
+
"step": 135
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"entropy": 0.8195698857307434,
|
| 1364 |
+
"epoch": 4.1226053639846745,
|
| 1365 |
+
"grad_norm": 5.40625,
|
| 1366 |
+
"learning_rate": 2.86474508437579e-06,
|
| 1367 |
+
"loss": 0.2115,
|
| 1368 |
+
"mean_token_accuracy": 0.9468159675598145,
|
| 1369 |
+
"num_tokens": 243375.0,
|
| 1370 |
+
"step": 136
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"entropy": 0.7908449321985245,
|
| 1374 |
+
"epoch": 4.153256704980843,
|
| 1375 |
+
"grad_norm": 3.890625,
|
| 1376 |
+
"learning_rate": 2.6827618629944394e-06,
|
| 1377 |
+
"loss": 0.2748,
|
| 1378 |
+
"mean_token_accuracy": 0.9350386410951614,
|
| 1379 |
+
"num_tokens": 245859.0,
|
| 1380 |
+
"step": 137
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"entropy": 0.7695152685046196,
|
| 1384 |
+
"epoch": 4.183908045977011,
|
| 1385 |
+
"grad_norm": 4.53125,
|
| 1386 |
+
"learning_rate": 2.5061813893485087e-06,
|
| 1387 |
+
"loss": 0.2897,
|
| 1388 |
+
"mean_token_accuracy": 0.9324667826294899,
|
| 1389 |
+
"num_tokens": 248006.0,
|
| 1390 |
+
"step": 138
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"entropy": 0.816169910132885,
|
| 1394 |
+
"epoch": 4.21455938697318,
|
| 1395 |
+
"grad_norm": 5.125,
|
| 1396 |
+
"learning_rate": 2.335081117469777e-06,
|
| 1397 |
+
"loss": 0.2809,
|
| 1398 |
+
"mean_token_accuracy": 0.9280546009540558,
|
| 1399 |
+
"num_tokens": 249615.0,
|
| 1400 |
+
"step": 139
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"entropy": 0.8056535497307777,
|
| 1404 |
+
"epoch": 4.245210727969349,
|
| 1405 |
+
"grad_norm": 5.375,
|
| 1406 |
+
"learning_rate": 2.169536097592401e-06,
|
| 1407 |
+
"loss": 0.2502,
|
| 1408 |
+
"mean_token_accuracy": 0.9439713805913925,
|
| 1409 |
+
"num_tokens": 251133.0,
|
| 1410 |
+
"step": 140
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"entropy": 0.7664961665868759,
|
| 1414 |
+
"epoch": 4.275862068965517,
|
| 1415 |
+
"grad_norm": 4.1875,
|
| 1416 |
+
"learning_rate": 2.0096189432334194e-06,
|
| 1417 |
+
"loss": 0.251,
|
| 1418 |
+
"mean_token_accuracy": 0.9318085461854935,
|
| 1419 |
+
"num_tokens": 253030.0,
|
| 1420 |
+
"step": 141
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"entropy": 0.7621353343129158,
|
| 1424 |
+
"epoch": 4.306513409961686,
|
| 1425 |
+
"grad_norm": 4.59375,
|
| 1426 |
+
"learning_rate": 1.8553997993420495e-06,
|
| 1427 |
+
"loss": 0.2019,
|
| 1428 |
+
"mean_token_accuracy": 0.9506975933909416,
|
| 1429 |
+
"num_tokens": 254683.0,
|
| 1430 |
+
"step": 142
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"entropy": 0.7942825853824615,
|
| 1434 |
+
"epoch": 4.337164750957855,
|
| 1435 |
+
"grad_norm": 7.5625,
|
| 1436 |
+
"learning_rate": 1.706946311531779e-06,
|
| 1437 |
+
"loss": 0.3202,
|
| 1438 |
+
"mean_token_accuracy": 0.9199245423078537,
|
| 1439 |
+
"num_tokens": 256339.0,
|
| 1440 |
+
"step": 143
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"entropy": 0.7943987771868706,
|
| 1444 |
+
"epoch": 4.3678160919540225,
|
| 1445 |
+
"grad_norm": 5.15625,
|
| 1446 |
+
"learning_rate": 1.5643235964088065e-06,
|
| 1447 |
+
"loss": 0.2463,
|
| 1448 |
+
"mean_token_accuracy": 0.9323071017861366,
|
| 1449 |
+
"num_tokens": 257877.0,
|
| 1450 |
+
"step": 144
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"entropy": 0.7164058461785316,
|
| 1454 |
+
"epoch": 4.398467432950191,
|
| 1455 |
+
"grad_norm": 13.4375,
|
| 1456 |
+
"learning_rate": 1.4275942130097097e-06,
|
| 1457 |
+
"loss": 0.1929,
|
| 1458 |
+
"mean_token_accuracy": 0.9492844417691231,
|
| 1459 |
+
"num_tokens": 259301.0,
|
| 1460 |
+
"step": 145
|
| 1461 |
}
|
| 1462 |
],
|
| 1463 |
"logging_steps": 1,
|
| 1464 |
+
"max_steps": 165,
|
| 1465 |
"num_input_tokens_seen": 0,
|
| 1466 |
+
"num_train_epochs": 5,
|
| 1467 |
+
"save_steps": 5,
|
| 1468 |
"stateful_callbacks": {
|
| 1469 |
"TrainerControl": {
|
| 1470 |
"args": {
|
|
|
|
| 1477 |
"attributes": {}
|
| 1478 |
}
|
| 1479 |
},
|
| 1480 |
+
"total_flos": 6865836098592768.0,
|
| 1481 |
"train_batch_size": 2,
|
| 1482 |
"trial_name": null,
|
| 1483 |
"trial_params": null
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:827ef2be292e1537aee3e07967732029beae9cb85d6335a905d123bebbf9cbca
|
| 3 |
+
size 6353
|