Uploaded checkpoint-22500
Browse files- adapter_config.json +2 -2
- adapter_model.safetensors +1 -1
- optimizer.pt +2 -2
- rng_state.pth +2 -2
- scheduler.pt +1 -1
- trainer_state.json +1366 -1040
- training_args.bin +2 -2
adapter_config.json
CHANGED
|
@@ -20,11 +20,11 @@
|
|
| 20 |
"revision": null,
|
| 21 |
"target_modules": [
|
| 22 |
"down_proj",
|
| 23 |
-
"up_proj",
|
| 24 |
"gate_proj",
|
|
|
|
| 25 |
"v_proj",
|
| 26 |
-
"k_proj",
|
| 27 |
"o_proj",
|
|
|
|
| 28 |
"q_proj"
|
| 29 |
],
|
| 30 |
"task_type": "CAUSAL_LM",
|
|
|
|
| 20 |
"revision": null,
|
| 21 |
"target_modules": [
|
| 22 |
"down_proj",
|
|
|
|
| 23 |
"gate_proj",
|
| 24 |
+
"up_proj",
|
| 25 |
"v_proj",
|
|
|
|
| 26 |
"o_proj",
|
| 27 |
+
"k_proj",
|
| 28 |
"q_proj"
|
| 29 |
],
|
| 30 |
"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 119975656
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:242bd1c8837db6c74ea117245bfc1b46592098a098bf72cb4a75b6fa1c50ea96
|
| 3 |
size 119975656
|
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 sha256:39f5976b7e007e60478770ca750c77a010d3dbba2afa30dc3c72a0856e2cd01d
|
| 3 |
+
size 240145026
|
rng_state.pth
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 sha256:763d4f397fbd8e2128612f32f5c273b211bf68ec4372f02c7c91ca944e405a2f
|
| 3 |
+
size 14244
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d71c00e9bfbac2252002b6eca4a38910300bb6c14e6c56273842dfbc024260d9
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,1484 +1,1810 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric":
|
| 3 |
-
"best_model_checkpoint":
|
| 4 |
-
"epoch": 0.
|
| 5 |
-
"eval_steps":
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0,
|
| 13 |
-
"grad_norm":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
"learning_rate": 4.000000000000001e-06,
|
| 15 |
-
"loss":
|
| 16 |
"step": 100
|
| 17 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
{
|
| 19 |
"epoch": 0.01,
|
| 20 |
-
"grad_norm":
|
| 21 |
"learning_rate": 8.000000000000001e-06,
|
| 22 |
-
"loss":
|
| 23 |
"step": 200
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"epoch": 0.01,
|
| 27 |
-
"grad_norm":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28 |
"learning_rate": 1.2e-05,
|
| 29 |
-
"loss":
|
| 30 |
"step": 300
|
| 31 |
},
|
| 32 |
{
|
| 33 |
"epoch": 0.01,
|
| 34 |
-
"grad_norm":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
"learning_rate": 1.6000000000000003e-05,
|
| 36 |
-
"loss":
|
| 37 |
"step": 400
|
| 38 |
},
|
| 39 |
{
|
| 40 |
"epoch": 0.01,
|
| 41 |
-
"grad_norm":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
"learning_rate": 2e-05,
|
| 43 |
-
"loss":
|
| 44 |
"step": 500
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.01,
|
| 48 |
-
"
|
| 49 |
-
"
|
| 50 |
-
"
|
| 51 |
-
"
|
|
|
|
| 52 |
},
|
| 53 |
{
|
| 54 |
-
"epoch": 0.
|
| 55 |
-
"grad_norm":
|
| 56 |
-
"learning_rate": 1.
|
| 57 |
-
"loss":
|
| 58 |
-
"step":
|
| 59 |
},
|
| 60 |
{
|
| 61 |
-
"epoch": 0.
|
| 62 |
-
"grad_norm":
|
| 63 |
-
"learning_rate": 1.
|
| 64 |
-
"loss":
|
| 65 |
-
"step":
|
| 66 |
},
|
| 67 |
{
|
| 68 |
-
"epoch": 0.
|
| 69 |
-
"grad_norm":
|
| 70 |
-
"learning_rate": 1.
|
| 71 |
-
"loss":
|
| 72 |
-
"step":
|
| 73 |
},
|
| 74 |
{
|
| 75 |
-
"epoch": 0.
|
| 76 |
-
"grad_norm":
|
| 77 |
-
"learning_rate": 1.
|
| 78 |
-
"loss":
|
| 79 |
-
"step":
|
| 80 |
},
|
| 81 |
{
|
| 82 |
-
"epoch": 0.
|
| 83 |
-
"grad_norm":
|
| 84 |
-
"learning_rate": 1.
|
| 85 |
-
"loss":
|
| 86 |
-
"step":
|
| 87 |
},
|
| 88 |
{
|
| 89 |
-
"epoch": 0.
|
| 90 |
-
"grad_norm":
|
| 91 |
-
"learning_rate": 1.
|
| 92 |
-
"loss":
|
| 93 |
-
"step":
|
| 94 |
},
|
| 95 |
{
|
| 96 |
-
"epoch": 0.
|
| 97 |
-
"grad_norm":
|
| 98 |
-
"learning_rate": 1.
|
| 99 |
-
"loss":
|
| 100 |
-
"step":
|
| 101 |
},
|
| 102 |
{
|
| 103 |
-
"epoch": 0.
|
| 104 |
-
"grad_norm":
|
| 105 |
-
"learning_rate": 1.
|
| 106 |
-
"loss":
|
| 107 |
-
"step":
|
| 108 |
},
|
| 109 |
{
|
| 110 |
-
"epoch": 0.
|
| 111 |
-
"grad_norm":
|
| 112 |
-
"learning_rate": 1.
|
| 113 |
-
"loss":
|
| 114 |
-
"step":
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm":
|
| 119 |
-
"learning_rate": 1.
|
| 120 |
-
"loss":
|
| 121 |
-
"step":
|
| 122 |
},
|
| 123 |
{
|
| 124 |
-
"epoch": 0.
|
| 125 |
-
"grad_norm": 3.
|
| 126 |
-
"learning_rate": 1.
|
| 127 |
-
"loss":
|
| 128 |
-
"step":
|
| 129 |
},
|
| 130 |
{
|
| 131 |
-
"epoch": 0.
|
| 132 |
-
"grad_norm":
|
| 133 |
-
"learning_rate": 1.
|
| 134 |
-
"loss":
|
| 135 |
-
"step":
|
| 136 |
},
|
| 137 |
{
|
| 138 |
-
"epoch": 0.
|
| 139 |
-
"grad_norm":
|
| 140 |
-
"learning_rate": 1.
|
| 141 |
-
"loss": 0.
|
| 142 |
-
"step":
|
| 143 |
},
|
| 144 |
{
|
| 145 |
-
"epoch": 0.
|
| 146 |
-
"grad_norm":
|
| 147 |
-
"learning_rate": 1.
|
| 148 |
-
"loss": 0.
|
| 149 |
-
"step":
|
| 150 |
},
|
| 151 |
{
|
| 152 |
-
"epoch": 0.
|
| 153 |
-
"grad_norm":
|
| 154 |
-
"learning_rate": 1.
|
| 155 |
-
"loss":
|
| 156 |
-
"step":
|
| 157 |
},
|
| 158 |
{
|
| 159 |
-
"epoch": 0.
|
| 160 |
-
"grad_norm":
|
| 161 |
-
"learning_rate": 1.
|
| 162 |
-
"loss":
|
| 163 |
-
"step":
|
| 164 |
},
|
| 165 |
{
|
| 166 |
-
"epoch": 0.
|
| 167 |
-
"grad_norm":
|
| 168 |
-
"learning_rate": 1.
|
| 169 |
-
"loss": 0.
|
| 170 |
-
"step":
|
| 171 |
},
|
| 172 |
{
|
| 173 |
-
"epoch": 0.
|
| 174 |
-
"grad_norm": 2.
|
| 175 |
-
"learning_rate": 1.
|
| 176 |
-
"loss":
|
| 177 |
-
"step":
|
| 178 |
},
|
| 179 |
{
|
| 180 |
-
"epoch": 0.
|
| 181 |
-
"grad_norm":
|
| 182 |
-
"learning_rate": 1.
|
| 183 |
-
"loss": 0.
|
| 184 |
-
"step":
|
| 185 |
},
|
| 186 |
{
|
| 187 |
-
"epoch": 0.
|
| 188 |
-
"
|
| 189 |
-
"
|
| 190 |
-
"
|
| 191 |
-
"
|
| 192 |
-
"step": 2500
|
| 193 |
},
|
| 194 |
{
|
| 195 |
-
"epoch": 0.
|
| 196 |
-
"grad_norm":
|
| 197 |
-
"learning_rate": 1.
|
| 198 |
-
"loss": 0.
|
| 199 |
-
"step":
|
| 200 |
},
|
| 201 |
{
|
| 202 |
-
"epoch": 0.
|
| 203 |
-
"grad_norm":
|
| 204 |
-
"learning_rate": 1.
|
| 205 |
-
"loss": 0.
|
| 206 |
-
"step":
|
| 207 |
},
|
| 208 |
{
|
| 209 |
-
"epoch": 0.
|
| 210 |
-
"grad_norm": 2.
|
| 211 |
-
"learning_rate": 1.
|
| 212 |
-
"loss": 0.
|
| 213 |
-
"step":
|
| 214 |
},
|
| 215 |
{
|
| 216 |
-
"epoch": 0.
|
| 217 |
-
"grad_norm":
|
| 218 |
-
"learning_rate": 1.
|
| 219 |
-
"loss": 0.
|
| 220 |
-
"step":
|
| 221 |
},
|
| 222 |
{
|
| 223 |
-
"epoch": 0.
|
| 224 |
-
"grad_norm":
|
| 225 |
-
"learning_rate": 1.
|
| 226 |
-
"loss": 0.
|
| 227 |
-
"step":
|
| 228 |
},
|
| 229 |
{
|
| 230 |
-
"epoch": 0.
|
| 231 |
-
"grad_norm":
|
| 232 |
-
"learning_rate": 1.
|
| 233 |
-
"loss": 0.
|
| 234 |
-
"step":
|
| 235 |
},
|
| 236 |
{
|
| 237 |
-
"epoch": 0.
|
| 238 |
-
"grad_norm":
|
| 239 |
-
"learning_rate": 1.
|
| 240 |
-
"loss": 0.
|
| 241 |
-
"step":
|
| 242 |
},
|
| 243 |
{
|
| 244 |
-
"epoch": 0.
|
| 245 |
-
"grad_norm":
|
| 246 |
-
"learning_rate": 1.
|
| 247 |
-
"loss": 0.
|
| 248 |
-
"step":
|
| 249 |
},
|
| 250 |
{
|
| 251 |
-
"epoch": 0.
|
| 252 |
-
"grad_norm":
|
| 253 |
-
"learning_rate": 1.
|
| 254 |
-
"loss": 0.
|
| 255 |
-
"step":
|
| 256 |
},
|
| 257 |
{
|
| 258 |
-
"epoch": 0.
|
| 259 |
-
"grad_norm":
|
| 260 |
-
"learning_rate": 1.
|
| 261 |
-
"loss": 0.
|
| 262 |
-
"step":
|
| 263 |
},
|
| 264 |
{
|
| 265 |
-
"epoch": 0.
|
| 266 |
-
"grad_norm":
|
| 267 |
-
"learning_rate": 1.
|
| 268 |
-
"loss": 0.
|
| 269 |
-
"step":
|
| 270 |
},
|
| 271 |
{
|
| 272 |
-
"epoch": 0.
|
| 273 |
-
"grad_norm":
|
| 274 |
-
"learning_rate": 1.
|
| 275 |
-
"loss": 0.
|
| 276 |
-
"step":
|
| 277 |
},
|
| 278 |
{
|
| 279 |
-
"epoch": 0.
|
| 280 |
-
"grad_norm": 2.
|
| 281 |
-
"learning_rate": 1.
|
| 282 |
-
"loss": 0.
|
| 283 |
-
"step":
|
| 284 |
},
|
| 285 |
{
|
| 286 |
-
"epoch": 0.
|
| 287 |
-
"grad_norm":
|
| 288 |
-
"learning_rate": 1.
|
| 289 |
-
"loss": 0.
|
| 290 |
-
"step":
|
| 291 |
},
|
| 292 |
{
|
| 293 |
-
"epoch": 0.
|
| 294 |
-
"grad_norm":
|
| 295 |
-
"learning_rate": 1.
|
| 296 |
-
"loss": 0.
|
| 297 |
-
"step":
|
| 298 |
},
|
| 299 |
{
|
| 300 |
-
"epoch": 0.
|
| 301 |
-
"grad_norm":
|
| 302 |
-
"learning_rate": 1.
|
| 303 |
-
"loss": 0.
|
| 304 |
-
"step":
|
| 305 |
},
|
| 306 |
{
|
| 307 |
-
"epoch": 0.
|
| 308 |
-
"grad_norm":
|
| 309 |
-
"learning_rate": 1.
|
| 310 |
-
"loss": 0.
|
| 311 |
-
"step":
|
| 312 |
},
|
| 313 |
{
|
| 314 |
-
"epoch": 0.
|
| 315 |
-
"grad_norm": 2.
|
| 316 |
-
"learning_rate": 1.
|
| 317 |
-
"loss": 0.
|
| 318 |
-
"step":
|
| 319 |
},
|
| 320 |
{
|
| 321 |
-
"epoch": 0.
|
| 322 |
-
"grad_norm":
|
| 323 |
-
"learning_rate": 1.
|
| 324 |
-
"loss": 0.
|
| 325 |
-
"step":
|
| 326 |
},
|
| 327 |
{
|
| 328 |
-
"epoch": 0.
|
| 329 |
-
"grad_norm":
|
| 330 |
-
"learning_rate": 1.
|
| 331 |
-
"loss": 0.
|
| 332 |
-
"step":
|
| 333 |
},
|
| 334 |
{
|
| 335 |
-
"epoch": 0.
|
| 336 |
-
"grad_norm":
|
| 337 |
-
"learning_rate": 1.
|
| 338 |
-
"loss": 0.
|
| 339 |
-
"step":
|
| 340 |
},
|
| 341 |
{
|
| 342 |
-
"epoch": 0.
|
| 343 |
-
"grad_norm":
|
| 344 |
-
"learning_rate": 1.
|
| 345 |
-
"loss": 0.
|
| 346 |
-
"step":
|
| 347 |
},
|
| 348 |
{
|
| 349 |
-
"epoch": 0.
|
| 350 |
-
"grad_norm":
|
| 351 |
-
"learning_rate": 1.
|
| 352 |
-
"loss": 0.
|
| 353 |
-
"step":
|
| 354 |
},
|
| 355 |
{
|
| 356 |
-
"epoch": 0.
|
| 357 |
-
"grad_norm":
|
| 358 |
-
"learning_rate": 1.
|
| 359 |
-
"loss": 0.
|
| 360 |
-
"step":
|
| 361 |
},
|
| 362 |
{
|
| 363 |
-
"epoch": 0.
|
| 364 |
-
"grad_norm":
|
| 365 |
-
"learning_rate": 1.
|
| 366 |
-
"loss": 0.
|
| 367 |
-
"step":
|
| 368 |
},
|
| 369 |
{
|
| 370 |
-
"epoch": 0.
|
| 371 |
-
"
|
| 372 |
-
"
|
| 373 |
-
"
|
| 374 |
-
"
|
| 375 |
-
"step": 5000
|
| 376 |
},
|
| 377 |
{
|
| 378 |
-
"epoch": 0.
|
| 379 |
-
"grad_norm":
|
| 380 |
-
"learning_rate": 1.
|
| 381 |
-
"loss": 0.
|
| 382 |
-
"step":
|
| 383 |
},
|
| 384 |
{
|
| 385 |
-
"epoch": 0.
|
| 386 |
-
"grad_norm": 5.
|
| 387 |
-
"learning_rate": 1.
|
| 388 |
-
"loss": 0.
|
| 389 |
-
"step":
|
| 390 |
},
|
| 391 |
{
|
| 392 |
-
"epoch": 0.
|
| 393 |
-
"grad_norm":
|
| 394 |
-
"learning_rate": 1.
|
| 395 |
-
"loss": 0.
|
| 396 |
-
"step":
|
| 397 |
},
|
| 398 |
{
|
| 399 |
-
"epoch": 0.
|
| 400 |
-
"grad_norm":
|
| 401 |
-
"learning_rate": 1.
|
| 402 |
-
"loss": 0.
|
| 403 |
-
"step":
|
| 404 |
},
|
| 405 |
{
|
| 406 |
-
"epoch": 0.
|
| 407 |
-
"
|
| 408 |
-
"
|
| 409 |
-
"
|
| 410 |
-
"
|
|
|
|
| 411 |
},
|
| 412 |
{
|
| 413 |
-
"epoch": 0.
|
| 414 |
-
"grad_norm":
|
| 415 |
-
"learning_rate": 1.
|
| 416 |
-
"loss": 0.
|
| 417 |
-
"step":
|
| 418 |
},
|
| 419 |
{
|
| 420 |
-
"epoch": 0.
|
| 421 |
-
"grad_norm":
|
| 422 |
-
"learning_rate": 1.
|
| 423 |
-
"loss": 0.
|
| 424 |
-
"step":
|
| 425 |
},
|
| 426 |
{
|
| 427 |
-
"epoch": 0.
|
| 428 |
-
"grad_norm":
|
| 429 |
-
"learning_rate": 1.
|
| 430 |
-
"loss": 0.
|
| 431 |
-
"step":
|
| 432 |
},
|
| 433 |
{
|
| 434 |
-
"epoch": 0.
|
| 435 |
-
"grad_norm":
|
| 436 |
-
"learning_rate": 1.
|
| 437 |
-
"loss": 0.
|
| 438 |
-
"step":
|
| 439 |
},
|
| 440 |
{
|
| 441 |
-
"epoch": 0.
|
| 442 |
-
"grad_norm":
|
| 443 |
-
"learning_rate": 1.
|
| 444 |
-
"loss": 0.
|
| 445 |
-
"step":
|
| 446 |
},
|
| 447 |
{
|
| 448 |
-
"epoch": 0.
|
| 449 |
-
"grad_norm":
|
| 450 |
-
"learning_rate": 1.
|
| 451 |
-
"loss": 0.
|
| 452 |
-
"step":
|
| 453 |
},
|
| 454 |
{
|
| 455 |
-
"epoch": 0.
|
| 456 |
-
"grad_norm":
|
| 457 |
-
"learning_rate": 1.
|
| 458 |
-
"loss": 0.
|
| 459 |
-
"step":
|
| 460 |
},
|
| 461 |
{
|
| 462 |
-
"epoch": 0.
|
| 463 |
-
"grad_norm":
|
| 464 |
-
"learning_rate": 1.
|
| 465 |
-
"loss": 0.
|
| 466 |
-
"step":
|
| 467 |
},
|
| 468 |
{
|
| 469 |
-
"epoch": 0.
|
| 470 |
-
"grad_norm":
|
| 471 |
-
"learning_rate": 1.
|
| 472 |
-
"loss": 0.
|
| 473 |
-
"step":
|
| 474 |
},
|
| 475 |
{
|
| 476 |
-
"epoch": 0.
|
| 477 |
-
"grad_norm":
|
| 478 |
-
"learning_rate": 1.
|
| 479 |
-
"loss": 0.
|
| 480 |
-
"step":
|
| 481 |
},
|
| 482 |
{
|
| 483 |
-
"epoch": 0.
|
| 484 |
-
"grad_norm":
|
| 485 |
-
"learning_rate": 1.
|
| 486 |
-
"loss": 0.
|
| 487 |
-
"step":
|
| 488 |
},
|
| 489 |
{
|
| 490 |
-
"epoch": 0.
|
| 491 |
-
"grad_norm":
|
| 492 |
-
"learning_rate": 1.
|
| 493 |
-
"loss": 0.
|
| 494 |
-
"step":
|
| 495 |
},
|
| 496 |
{
|
| 497 |
-
"epoch": 0.
|
| 498 |
-
"grad_norm":
|
| 499 |
-
"learning_rate": 1.
|
| 500 |
-
"loss": 0.
|
| 501 |
-
"step":
|
| 502 |
},
|
| 503 |
{
|
| 504 |
-
"epoch": 0.
|
| 505 |
-
"grad_norm":
|
| 506 |
-
"learning_rate": 1.
|
| 507 |
-
"loss": 0.
|
| 508 |
-
"step":
|
| 509 |
},
|
| 510 |
{
|
| 511 |
-
"epoch": 0.
|
| 512 |
-
"grad_norm": 4.
|
| 513 |
-
"learning_rate": 1.
|
| 514 |
-
"loss": 0.
|
| 515 |
-
"step":
|
| 516 |
},
|
| 517 |
{
|
| 518 |
-
"epoch": 0.
|
| 519 |
-
"grad_norm": 3.
|
| 520 |
-
"learning_rate": 1.
|
| 521 |
-
"loss": 0.
|
| 522 |
-
"step":
|
| 523 |
},
|
| 524 |
{
|
| 525 |
-
"epoch": 0.
|
| 526 |
-
"grad_norm":
|
| 527 |
-
"learning_rate": 1.
|
| 528 |
-
"loss": 0.
|
| 529 |
-
"step":
|
| 530 |
},
|
| 531 |
{
|
| 532 |
-
"epoch": 0.
|
| 533 |
-
"grad_norm":
|
| 534 |
-
"learning_rate": 1.
|
| 535 |
-
"loss": 0.
|
| 536 |
-
"step":
|
| 537 |
},
|
| 538 |
{
|
| 539 |
-
"epoch": 0.
|
| 540 |
-
"grad_norm":
|
| 541 |
-
"learning_rate": 1.
|
| 542 |
-
"loss": 0.
|
| 543 |
-
"step":
|
| 544 |
},
|
| 545 |
{
|
| 546 |
-
"epoch": 0.
|
| 547 |
-
"grad_norm":
|
| 548 |
-
"learning_rate": 1.
|
| 549 |
-
"loss": 0.
|
| 550 |
-
"step":
|
| 551 |
},
|
| 552 |
{
|
| 553 |
-
"epoch": 0.
|
| 554 |
-
"
|
| 555 |
-
"
|
| 556 |
-
"
|
| 557 |
-
"
|
| 558 |
-
"step": 7500
|
| 559 |
},
|
| 560 |
{
|
| 561 |
-
"epoch": 0.
|
| 562 |
-
"grad_norm":
|
| 563 |
-
"learning_rate": 1.
|
| 564 |
-
"loss": 0.
|
| 565 |
-
"step":
|
| 566 |
},
|
| 567 |
{
|
| 568 |
-
"epoch": 0.
|
| 569 |
-
"grad_norm":
|
| 570 |
-
"learning_rate": 1.
|
| 571 |
-
"loss": 0.
|
| 572 |
-
"step":
|
| 573 |
},
|
| 574 |
{
|
| 575 |
-
"epoch": 0.
|
| 576 |
-
"grad_norm": 4.
|
| 577 |
-
"learning_rate": 1.
|
| 578 |
-
"loss": 0.
|
| 579 |
-
"step":
|
| 580 |
},
|
| 581 |
{
|
| 582 |
-
"epoch": 0.
|
| 583 |
-
"grad_norm":
|
| 584 |
-
"learning_rate": 1.
|
| 585 |
-
"loss": 0.
|
| 586 |
-
"step":
|
| 587 |
},
|
| 588 |
{
|
| 589 |
-
"epoch": 0.
|
| 590 |
-
"grad_norm":
|
| 591 |
-
"learning_rate": 1.
|
| 592 |
-
"loss": 0.
|
| 593 |
-
"step":
|
| 594 |
},
|
| 595 |
{
|
| 596 |
-
"epoch": 0.
|
| 597 |
-
"grad_norm":
|
| 598 |
-
"learning_rate": 1.
|
| 599 |
-
"loss": 0.
|
| 600 |
-
"step":
|
| 601 |
},
|
| 602 |
{
|
| 603 |
-
"epoch": 0.
|
| 604 |
-
"grad_norm":
|
| 605 |
-
"learning_rate": 1.
|
| 606 |
-
"loss": 0.
|
| 607 |
-
"step":
|
| 608 |
},
|
| 609 |
{
|
| 610 |
-
"epoch": 0.
|
| 611 |
-
"grad_norm":
|
| 612 |
-
"learning_rate": 1.
|
| 613 |
-
"loss": 0.
|
| 614 |
-
"step":
|
| 615 |
},
|
| 616 |
{
|
| 617 |
-
"epoch": 0.
|
| 618 |
-
"grad_norm":
|
| 619 |
-
"learning_rate": 1.
|
| 620 |
-
"loss": 0.
|
| 621 |
-
"step":
|
| 622 |
},
|
| 623 |
{
|
| 624 |
-
"epoch": 0.
|
| 625 |
-
"grad_norm":
|
| 626 |
-
"learning_rate": 1.
|
| 627 |
-
"loss": 0.
|
| 628 |
-
"step":
|
| 629 |
},
|
| 630 |
{
|
| 631 |
-
"epoch": 0.
|
| 632 |
-
"grad_norm":
|
| 633 |
-
"learning_rate": 1.
|
| 634 |
-
"loss": 0.
|
| 635 |
-
"step":
|
| 636 |
},
|
| 637 |
{
|
| 638 |
-
"epoch": 0.
|
| 639 |
-
"grad_norm":
|
| 640 |
-
"learning_rate": 1.
|
| 641 |
-
"loss": 0.
|
| 642 |
-
"step":
|
| 643 |
},
|
| 644 |
{
|
| 645 |
-
"epoch": 0.
|
| 646 |
-
"grad_norm":
|
| 647 |
-
"learning_rate": 1.
|
| 648 |
-
"loss": 0.
|
| 649 |
-
"step":
|
| 650 |
},
|
| 651 |
{
|
| 652 |
-
"epoch": 0.
|
| 653 |
-
"grad_norm":
|
| 654 |
-
"learning_rate": 1.
|
| 655 |
-
"loss": 0.
|
| 656 |
-
"step":
|
| 657 |
},
|
| 658 |
{
|
| 659 |
-
"epoch": 0.
|
| 660 |
-
"grad_norm":
|
| 661 |
-
"learning_rate": 1.
|
| 662 |
-
"loss": 0.
|
| 663 |
-
"step":
|
| 664 |
},
|
| 665 |
{
|
| 666 |
-
"epoch": 0.
|
| 667 |
-
"grad_norm":
|
| 668 |
-
"learning_rate": 1.
|
| 669 |
-
"loss": 0.
|
| 670 |
-
"step":
|
| 671 |
},
|
| 672 |
{
|
| 673 |
-
"epoch": 0.
|
| 674 |
-
"grad_norm": 4.
|
| 675 |
-
"learning_rate": 1.
|
| 676 |
-
"loss": 0.
|
| 677 |
-
"step":
|
| 678 |
},
|
| 679 |
{
|
| 680 |
-
"epoch": 0.
|
| 681 |
-
"grad_norm":
|
| 682 |
-
"learning_rate": 1.
|
| 683 |
-
"loss": 0.
|
| 684 |
-
"step":
|
| 685 |
},
|
| 686 |
{
|
| 687 |
-
"epoch": 0.
|
| 688 |
-
"grad_norm":
|
| 689 |
-
"learning_rate": 1.
|
| 690 |
-
"loss": 0.
|
| 691 |
-
"step":
|
| 692 |
},
|
| 693 |
{
|
| 694 |
-
"epoch": 0.
|
| 695 |
-
"grad_norm": 2.
|
| 696 |
-
"learning_rate": 1.
|
| 697 |
-
"loss": 0.
|
| 698 |
-
"step":
|
| 699 |
},
|
| 700 |
{
|
| 701 |
-
"epoch": 0.
|
| 702 |
-
"grad_norm":
|
| 703 |
-
"learning_rate": 1.
|
| 704 |
-
"loss": 0.
|
| 705 |
-
"step":
|
| 706 |
},
|
| 707 |
{
|
| 708 |
-
"epoch": 0.
|
| 709 |
-
"grad_norm":
|
| 710 |
-
"learning_rate": 1.
|
| 711 |
-
"loss": 0.
|
| 712 |
-
"step":
|
| 713 |
},
|
| 714 |
{
|
| 715 |
-
"epoch": 0.
|
| 716 |
-
"grad_norm":
|
| 717 |
-
"learning_rate": 1.
|
| 718 |
-
"loss": 0.
|
| 719 |
-
"step":
|
| 720 |
},
|
| 721 |
{
|
| 722 |
-
"epoch": 0.
|
| 723 |
-
"grad_norm": 2.
|
| 724 |
-
"learning_rate": 1.
|
| 725 |
-
"loss": 0.
|
| 726 |
-
"step":
|
| 727 |
},
|
| 728 |
{
|
| 729 |
-
"epoch": 0.
|
| 730 |
-
"grad_norm":
|
| 731 |
-
"learning_rate": 1.
|
| 732 |
-
"loss": 0.
|
| 733 |
-
"step":
|
| 734 |
},
|
| 735 |
{
|
| 736 |
-
"epoch": 0.
|
| 737 |
-
"
|
| 738 |
-
"
|
| 739 |
-
"
|
| 740 |
-
"
|
| 741 |
-
"step": 10000
|
| 742 |
},
|
| 743 |
{
|
| 744 |
-
"epoch": 0.
|
| 745 |
-
"grad_norm":
|
| 746 |
-
"learning_rate": 1.
|
| 747 |
-
"loss": 0.
|
| 748 |
-
"step":
|
| 749 |
},
|
| 750 |
{
|
| 751 |
-
"epoch": 0.
|
| 752 |
-
"grad_norm":
|
| 753 |
-
"learning_rate": 1.
|
| 754 |
-
"loss": 0.
|
| 755 |
-
"step":
|
| 756 |
},
|
| 757 |
{
|
| 758 |
-
"epoch": 0.
|
| 759 |
-
"grad_norm":
|
| 760 |
-
"learning_rate":
|
| 761 |
-
"loss": 0.
|
| 762 |
-
"step":
|
| 763 |
},
|
| 764 |
{
|
| 765 |
-
"epoch": 0.
|
| 766 |
-
"
|
| 767 |
-
"
|
| 768 |
-
"
|
| 769 |
-
"
|
|
|
|
| 770 |
},
|
| 771 |
{
|
| 772 |
-
"epoch": 0.
|
| 773 |
-
"grad_norm":
|
| 774 |
-
"learning_rate":
|
| 775 |
-
"loss": 0.
|
| 776 |
-
"step":
|
| 777 |
},
|
| 778 |
{
|
| 779 |
-
"epoch": 0.
|
| 780 |
-
"grad_norm":
|
| 781 |
-
"learning_rate":
|
| 782 |
-
"loss": 0.
|
| 783 |
-
"step":
|
| 784 |
},
|
| 785 |
{
|
| 786 |
-
"epoch": 0.
|
| 787 |
-
"grad_norm":
|
| 788 |
-
"learning_rate":
|
| 789 |
-
"loss": 0.
|
| 790 |
-
"step":
|
| 791 |
},
|
| 792 |
{
|
| 793 |
-
"epoch": 0.
|
| 794 |
-
"grad_norm":
|
| 795 |
-
"learning_rate":
|
| 796 |
-
"loss": 0.
|
| 797 |
-
"step":
|
| 798 |
},
|
| 799 |
{
|
| 800 |
-
"epoch": 0.
|
| 801 |
-
"grad_norm": 4.
|
| 802 |
-
"learning_rate":
|
| 803 |
-
"loss": 0.
|
| 804 |
-
"step":
|
| 805 |
},
|
| 806 |
{
|
| 807 |
-
"epoch": 0.
|
| 808 |
-
"grad_norm":
|
| 809 |
-
"learning_rate":
|
| 810 |
-
"loss": 0.
|
| 811 |
-
"step":
|
| 812 |
},
|
| 813 |
{
|
| 814 |
-
"epoch": 0.
|
| 815 |
-
"grad_norm":
|
| 816 |
-
"learning_rate":
|
| 817 |
-
"loss": 0.
|
| 818 |
-
"step":
|
| 819 |
},
|
| 820 |
{
|
| 821 |
-
"epoch": 0.
|
| 822 |
-
"grad_norm":
|
| 823 |
-
"learning_rate":
|
| 824 |
-
"loss": 0.
|
| 825 |
-
"step":
|
| 826 |
},
|
| 827 |
{
|
| 828 |
-
"epoch": 0.
|
| 829 |
-
"grad_norm":
|
| 830 |
-
"learning_rate":
|
| 831 |
-
"loss": 0.
|
| 832 |
-
"step":
|
| 833 |
},
|
| 834 |
{
|
| 835 |
-
"epoch": 0.
|
| 836 |
-
"grad_norm":
|
| 837 |
-
"learning_rate":
|
| 838 |
-
"loss": 0.
|
| 839 |
-
"step":
|
| 840 |
},
|
| 841 |
{
|
| 842 |
-
"epoch": 0.
|
| 843 |
-
"grad_norm":
|
| 844 |
-
"learning_rate":
|
| 845 |
-
"loss": 0.
|
| 846 |
-
"step":
|
| 847 |
},
|
| 848 |
{
|
| 849 |
-
"epoch": 0.
|
| 850 |
-
"grad_norm":
|
| 851 |
-
"learning_rate":
|
| 852 |
-
"loss": 0.
|
| 853 |
-
"step":
|
| 854 |
},
|
| 855 |
{
|
| 856 |
-
"epoch": 0.
|
| 857 |
-
"grad_norm":
|
| 858 |
-
"learning_rate":
|
| 859 |
-
"loss": 0.
|
| 860 |
-
"step":
|
| 861 |
},
|
| 862 |
{
|
| 863 |
-
"epoch": 0.
|
| 864 |
-
"grad_norm":
|
| 865 |
-
"learning_rate":
|
| 866 |
-
"loss": 0.
|
| 867 |
-
"step":
|
| 868 |
},
|
| 869 |
{
|
| 870 |
-
"epoch": 0.
|
| 871 |
-
"grad_norm":
|
| 872 |
-
"learning_rate":
|
| 873 |
-
"loss": 0.
|
| 874 |
-
"step":
|
| 875 |
},
|
| 876 |
{
|
| 877 |
-
"epoch": 0.
|
| 878 |
-
"grad_norm":
|
| 879 |
-
"learning_rate":
|
| 880 |
-
"loss": 0.
|
| 881 |
-
"step":
|
| 882 |
},
|
| 883 |
{
|
| 884 |
-
"epoch": 0.
|
| 885 |
-
"grad_norm":
|
| 886 |
-
"learning_rate":
|
| 887 |
-
"loss": 0.
|
| 888 |
-
"step":
|
| 889 |
},
|
| 890 |
{
|
| 891 |
-
"epoch": 0.
|
| 892 |
-
"grad_norm":
|
| 893 |
-
"learning_rate":
|
| 894 |
-
"loss": 0.
|
| 895 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 896 |
},
|
| 897 |
{
|
| 898 |
-
"epoch": 0.
|
| 899 |
-
"grad_norm": 3.
|
| 900 |
-
"learning_rate":
|
| 901 |
-
"loss": 0.
|
| 902 |
-
"step":
|
| 903 |
},
|
| 904 |
{
|
| 905 |
-
"epoch": 0.
|
| 906 |
-
"grad_norm": 5.
|
| 907 |
-
"learning_rate":
|
| 908 |
-
"loss": 0.
|
| 909 |
-
"step":
|
| 910 |
},
|
| 911 |
{
|
| 912 |
-
"epoch": 0.
|
| 913 |
-
"grad_norm":
|
| 914 |
-
"learning_rate":
|
| 915 |
-
"loss": 0.
|
| 916 |
-
"step":
|
| 917 |
},
|
| 918 |
{
|
| 919 |
-
"epoch": 0.
|
| 920 |
-
"
|
| 921 |
-
"
|
| 922 |
-
"
|
| 923 |
-
"
|
| 924 |
-
"step": 12500
|
| 925 |
},
|
| 926 |
{
|
| 927 |
-
"epoch": 0.
|
| 928 |
-
"grad_norm":
|
| 929 |
-
"learning_rate":
|
| 930 |
-
"loss": 0.
|
| 931 |
-
"step":
|
| 932 |
},
|
| 933 |
{
|
| 934 |
-
"epoch": 0.
|
| 935 |
-
"grad_norm":
|
| 936 |
-
"learning_rate":
|
| 937 |
-
"loss": 0.
|
| 938 |
-
"step":
|
| 939 |
},
|
| 940 |
{
|
| 941 |
-
"epoch": 0.
|
| 942 |
-
"grad_norm":
|
| 943 |
-
"learning_rate":
|
| 944 |
-
"loss": 0.
|
| 945 |
-
"step":
|
| 946 |
},
|
| 947 |
{
|
| 948 |
-
"epoch": 0.
|
| 949 |
-
"grad_norm":
|
| 950 |
-
"learning_rate":
|
| 951 |
-
"loss": 0.
|
| 952 |
-
"step":
|
| 953 |
},
|
| 954 |
{
|
| 955 |
-
"epoch": 0.
|
| 956 |
-
"grad_norm":
|
| 957 |
-
"learning_rate":
|
| 958 |
-
"loss": 0.
|
| 959 |
-
"step":
|
| 960 |
},
|
| 961 |
{
|
| 962 |
-
"epoch": 0.
|
| 963 |
-
"grad_norm":
|
| 964 |
-
"learning_rate":
|
| 965 |
-
"loss": 0.
|
| 966 |
-
"step":
|
| 967 |
},
|
| 968 |
{
|
| 969 |
-
"epoch": 0.
|
| 970 |
-
"grad_norm":
|
| 971 |
-
"learning_rate":
|
| 972 |
-
"loss": 0.
|
| 973 |
-
"step":
|
| 974 |
},
|
| 975 |
{
|
| 976 |
-
"epoch": 0.
|
| 977 |
-
"grad_norm":
|
| 978 |
-
"learning_rate":
|
| 979 |
-
"loss": 0.
|
| 980 |
-
"step":
|
| 981 |
},
|
| 982 |
{
|
| 983 |
-
"epoch": 0.
|
| 984 |
-
"grad_norm":
|
| 985 |
-
"learning_rate":
|
| 986 |
-
"loss": 0.
|
| 987 |
-
"step":
|
| 988 |
},
|
| 989 |
{
|
| 990 |
-
"epoch": 0.
|
| 991 |
-
"grad_norm":
|
| 992 |
-
"learning_rate":
|
| 993 |
-
"loss": 0.
|
| 994 |
-
"step":
|
| 995 |
},
|
| 996 |
{
|
| 997 |
-
"epoch": 0.
|
| 998 |
-
"grad_norm":
|
| 999 |
-
"learning_rate":
|
| 1000 |
-
"loss": 0.
|
| 1001 |
-
"step":
|
| 1002 |
},
|
| 1003 |
{
|
| 1004 |
-
"epoch": 0.
|
| 1005 |
-
"grad_norm":
|
| 1006 |
-
"learning_rate":
|
| 1007 |
-
"loss": 0.
|
| 1008 |
-
"step":
|
| 1009 |
},
|
| 1010 |
{
|
| 1011 |
-
"epoch": 0.
|
| 1012 |
-
"grad_norm":
|
| 1013 |
-
"learning_rate":
|
| 1014 |
-
"loss": 0.
|
| 1015 |
-
"step":
|
| 1016 |
},
|
| 1017 |
{
|
| 1018 |
-
"epoch": 0.
|
| 1019 |
-
"grad_norm": 4.
|
| 1020 |
-
"learning_rate":
|
| 1021 |
-
"loss": 0.
|
| 1022 |
-
"step":
|
| 1023 |
},
|
| 1024 |
{
|
| 1025 |
-
"epoch": 0.
|
| 1026 |
-
"grad_norm":
|
| 1027 |
-
"learning_rate":
|
| 1028 |
-
"loss": 0.
|
| 1029 |
-
"step":
|
| 1030 |
},
|
| 1031 |
{
|
| 1032 |
-
"epoch": 0.
|
| 1033 |
-
"grad_norm":
|
| 1034 |
-
"learning_rate":
|
| 1035 |
-
"loss": 0.
|
| 1036 |
-
"step":
|
| 1037 |
},
|
| 1038 |
{
|
| 1039 |
-
"epoch": 0.
|
| 1040 |
-
"grad_norm":
|
| 1041 |
-
"learning_rate":
|
| 1042 |
-
"loss": 0.
|
| 1043 |
-
"step":
|
| 1044 |
},
|
| 1045 |
{
|
| 1046 |
-
"epoch": 0.
|
| 1047 |
-
"grad_norm":
|
| 1048 |
-
"learning_rate":
|
| 1049 |
-
"loss": 0.
|
| 1050 |
-
"step":
|
| 1051 |
},
|
| 1052 |
{
|
| 1053 |
-
"epoch": 0.
|
| 1054 |
-
"grad_norm":
|
| 1055 |
-
"learning_rate":
|
| 1056 |
-
"loss": 0.
|
| 1057 |
-
"step":
|
| 1058 |
},
|
| 1059 |
{
|
| 1060 |
-
"epoch": 0.
|
| 1061 |
-
"grad_norm":
|
| 1062 |
-
"learning_rate":
|
| 1063 |
-
"loss": 0.
|
| 1064 |
-
"step":
|
| 1065 |
},
|
| 1066 |
{
|
| 1067 |
-
"epoch": 0.
|
| 1068 |
-
"grad_norm":
|
| 1069 |
-
"learning_rate":
|
| 1070 |
-
"loss": 0.
|
| 1071 |
-
"step":
|
| 1072 |
},
|
| 1073 |
{
|
| 1074 |
-
"epoch": 0.
|
| 1075 |
-
"grad_norm":
|
| 1076 |
-
"learning_rate":
|
| 1077 |
-
"loss": 0.
|
| 1078 |
-
"step":
|
| 1079 |
},
|
| 1080 |
{
|
| 1081 |
-
"epoch": 0.
|
| 1082 |
-
"grad_norm":
|
| 1083 |
-
"learning_rate":
|
| 1084 |
-
"loss": 0.
|
| 1085 |
-
"step":
|
| 1086 |
},
|
| 1087 |
{
|
| 1088 |
-
"epoch": 0.
|
| 1089 |
-
"grad_norm":
|
| 1090 |
-
"learning_rate":
|
| 1091 |
-
"loss": 0.
|
| 1092 |
-
"step":
|
| 1093 |
},
|
| 1094 |
{
|
| 1095 |
-
"epoch": 0.
|
| 1096 |
-
"grad_norm":
|
| 1097 |
-
"learning_rate":
|
| 1098 |
-
"loss": 0.
|
| 1099 |
-
"step":
|
| 1100 |
},
|
| 1101 |
{
|
| 1102 |
-
"epoch": 0.
|
| 1103 |
-
"
|
| 1104 |
-
"
|
| 1105 |
-
"
|
| 1106 |
-
"
|
| 1107 |
-
"step": 15000
|
| 1108 |
},
|
| 1109 |
{
|
| 1110 |
-
"epoch": 0.
|
| 1111 |
-
"grad_norm":
|
| 1112 |
-
"learning_rate":
|
| 1113 |
-
"loss": 0.
|
| 1114 |
-
"step":
|
| 1115 |
},
|
| 1116 |
{
|
| 1117 |
-
"epoch": 0.
|
| 1118 |
-
"
|
| 1119 |
-
"
|
| 1120 |
-
"
|
| 1121 |
-
"
|
|
|
|
| 1122 |
},
|
| 1123 |
{
|
| 1124 |
-
"epoch": 0.
|
| 1125 |
-
"grad_norm":
|
| 1126 |
-
"learning_rate":
|
| 1127 |
-
"loss": 0.
|
| 1128 |
-
"step":
|
| 1129 |
},
|
| 1130 |
{
|
| 1131 |
-
"epoch": 0.
|
| 1132 |
-
"grad_norm": 4.
|
| 1133 |
-
"learning_rate":
|
| 1134 |
-
"loss": 0.
|
| 1135 |
-
"step":
|
| 1136 |
},
|
| 1137 |
{
|
| 1138 |
-
"epoch": 0.
|
| 1139 |
-
"grad_norm":
|
| 1140 |
-
"learning_rate":
|
| 1141 |
-
"loss": 0.
|
| 1142 |
-
"step":
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
-
"epoch": 0.
|
| 1146 |
-
"grad_norm":
|
| 1147 |
-
"learning_rate":
|
| 1148 |
-
"loss": 0.
|
| 1149 |
-
"step":
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
-
"epoch": 0.
|
| 1153 |
-
"grad_norm":
|
| 1154 |
-
"learning_rate":
|
| 1155 |
-
"loss": 0.
|
| 1156 |
-
"step":
|
| 1157 |
},
|
| 1158 |
{
|
| 1159 |
-
"epoch": 0.
|
| 1160 |
-
"grad_norm":
|
| 1161 |
-
"learning_rate":
|
| 1162 |
-
"loss": 0.
|
| 1163 |
-
"step":
|
| 1164 |
},
|
| 1165 |
{
|
| 1166 |
-
"epoch": 0.
|
| 1167 |
-
"grad_norm":
|
| 1168 |
-
"learning_rate":
|
| 1169 |
-
"loss": 0.
|
| 1170 |
-
"step":
|
| 1171 |
},
|
| 1172 |
{
|
| 1173 |
-
"epoch": 0.
|
| 1174 |
-
"grad_norm":
|
| 1175 |
-
"learning_rate":
|
| 1176 |
-
"loss": 0.
|
| 1177 |
-
"step":
|
| 1178 |
},
|
| 1179 |
{
|
| 1180 |
-
"epoch": 0.
|
| 1181 |
-
"grad_norm":
|
| 1182 |
-
"learning_rate":
|
| 1183 |
-
"loss": 0.
|
| 1184 |
-
"step":
|
| 1185 |
},
|
| 1186 |
{
|
| 1187 |
-
"epoch": 0.
|
| 1188 |
-
"grad_norm":
|
| 1189 |
-
"learning_rate":
|
| 1190 |
-
"loss": 0.
|
| 1191 |
-
"step":
|
| 1192 |
},
|
| 1193 |
{
|
| 1194 |
-
"epoch": 0.
|
| 1195 |
-
"grad_norm": 2.
|
| 1196 |
-
"learning_rate":
|
| 1197 |
-
"loss": 0.
|
| 1198 |
-
"step":
|
| 1199 |
},
|
| 1200 |
{
|
| 1201 |
-
"epoch": 0.
|
| 1202 |
-
"grad_norm": 4.
|
| 1203 |
-
"learning_rate":
|
| 1204 |
-
"loss": 0.
|
| 1205 |
-
"step":
|
| 1206 |
},
|
| 1207 |
{
|
| 1208 |
-
"epoch": 0.
|
| 1209 |
-
"grad_norm":
|
| 1210 |
-
"learning_rate":
|
| 1211 |
-
"loss": 0.
|
| 1212 |
-
"step":
|
| 1213 |
},
|
| 1214 |
{
|
| 1215 |
-
"epoch": 0.
|
| 1216 |
-
"grad_norm":
|
| 1217 |
-
"learning_rate":
|
| 1218 |
-
"loss": 0.
|
| 1219 |
-
"step":
|
| 1220 |
},
|
| 1221 |
{
|
| 1222 |
-
"epoch": 0.
|
| 1223 |
-
"grad_norm":
|
| 1224 |
-
"learning_rate":
|
| 1225 |
-
"loss": 0.
|
| 1226 |
-
"step":
|
| 1227 |
},
|
| 1228 |
{
|
| 1229 |
-
"epoch": 0.
|
| 1230 |
-
"grad_norm":
|
| 1231 |
-
"learning_rate":
|
| 1232 |
-
"loss": 0.
|
| 1233 |
-
"step":
|
| 1234 |
},
|
| 1235 |
{
|
| 1236 |
-
"epoch": 0.
|
| 1237 |
-
"grad_norm":
|
| 1238 |
-
"learning_rate":
|
| 1239 |
-
"loss": 0.
|
| 1240 |
-
"step":
|
| 1241 |
},
|
| 1242 |
{
|
| 1243 |
-
"epoch": 0.
|
| 1244 |
-
"grad_norm": 11.
|
| 1245 |
-
"learning_rate":
|
| 1246 |
-
"loss": 0.
|
| 1247 |
-
"step":
|
| 1248 |
},
|
| 1249 |
{
|
| 1250 |
-
"epoch": 0.
|
| 1251 |
-
"grad_norm": 3.
|
| 1252 |
-
"learning_rate":
|
| 1253 |
-
"loss": 0.
|
| 1254 |
-
"step":
|
| 1255 |
},
|
| 1256 |
{
|
| 1257 |
-
"epoch": 0.
|
| 1258 |
-
"grad_norm":
|
| 1259 |
-
"learning_rate":
|
| 1260 |
-
"loss": 0.
|
| 1261 |
-
"step":
|
| 1262 |
},
|
| 1263 |
{
|
| 1264 |
-
"epoch": 0.
|
| 1265 |
-
"grad_norm":
|
| 1266 |
-
"learning_rate":
|
| 1267 |
-
"loss": 0.
|
| 1268 |
-
"step":
|
| 1269 |
},
|
| 1270 |
{
|
| 1271 |
-
"epoch": 0.
|
| 1272 |
-
"grad_norm":
|
| 1273 |
-
"learning_rate":
|
| 1274 |
-
"loss": 0.
|
| 1275 |
-
"step":
|
| 1276 |
},
|
| 1277 |
{
|
| 1278 |
-
"epoch": 0.
|
| 1279 |
-
"grad_norm": 4.
|
| 1280 |
-
"learning_rate":
|
| 1281 |
-
"loss": 0.
|
| 1282 |
-
"step":
|
| 1283 |
},
|
| 1284 |
{
|
| 1285 |
-
"epoch": 0.
|
| 1286 |
-
"
|
| 1287 |
-
"
|
| 1288 |
-
"
|
| 1289 |
-
"
|
| 1290 |
-
"step": 17500
|
| 1291 |
},
|
| 1292 |
{
|
| 1293 |
-
"epoch": 0.
|
| 1294 |
-
"grad_norm":
|
| 1295 |
-
"learning_rate":
|
| 1296 |
-
"loss": 0.
|
| 1297 |
-
"step":
|
| 1298 |
},
|
| 1299 |
{
|
| 1300 |
-
"epoch": 0.
|
| 1301 |
-
"grad_norm": 4.
|
| 1302 |
-
"learning_rate":
|
| 1303 |
-
"loss": 0.
|
| 1304 |
-
"step":
|
| 1305 |
},
|
| 1306 |
{
|
| 1307 |
-
"epoch": 0.
|
| 1308 |
-
"grad_norm":
|
| 1309 |
-
"learning_rate":
|
| 1310 |
-
"loss": 0.
|
| 1311 |
-
"step":
|
| 1312 |
},
|
| 1313 |
{
|
| 1314 |
-
"epoch": 0.
|
| 1315 |
-
"grad_norm":
|
| 1316 |
-
"learning_rate":
|
| 1317 |
-
"loss": 0.
|
| 1318 |
-
"step":
|
| 1319 |
},
|
| 1320 |
{
|
| 1321 |
-
"epoch": 0.
|
| 1322 |
-
"grad_norm":
|
| 1323 |
-
"learning_rate":
|
| 1324 |
-
"loss": 0.
|
| 1325 |
-
"step":
|
| 1326 |
},
|
| 1327 |
{
|
| 1328 |
-
"epoch": 0.
|
| 1329 |
-
"grad_norm":
|
| 1330 |
-
"learning_rate": 1.
|
| 1331 |
-
"loss": 0.
|
| 1332 |
-
"step":
|
| 1333 |
},
|
| 1334 |
{
|
| 1335 |
-
"epoch": 0.
|
| 1336 |
-
"grad_norm":
|
| 1337 |
-
"learning_rate": 1.
|
| 1338 |
-
"loss": 0.
|
| 1339 |
-
"step":
|
| 1340 |
},
|
| 1341 |
{
|
| 1342 |
-
"epoch": 0.
|
| 1343 |
-
"grad_norm":
|
| 1344 |
-
"learning_rate": 1.
|
| 1345 |
-
"loss": 0.
|
| 1346 |
-
"step":
|
| 1347 |
},
|
| 1348 |
{
|
| 1349 |
-
"epoch": 0.
|
| 1350 |
-
"grad_norm":
|
| 1351 |
-
"learning_rate": 1.
|
| 1352 |
-
"loss": 0.
|
| 1353 |
-
"step":
|
| 1354 |
},
|
| 1355 |
{
|
| 1356 |
-
"epoch": 0.
|
| 1357 |
-
"grad_norm":
|
| 1358 |
-
"learning_rate": 1.
|
| 1359 |
-
"loss": 0.
|
| 1360 |
-
"step":
|
| 1361 |
},
|
| 1362 |
{
|
| 1363 |
-
"epoch": 0.
|
| 1364 |
-
"grad_norm":
|
| 1365 |
-
"learning_rate": 1.
|
| 1366 |
-
"loss": 0.
|
| 1367 |
-
"step":
|
| 1368 |
},
|
| 1369 |
{
|
| 1370 |
-
"epoch": 0.
|
| 1371 |
-
"grad_norm":
|
| 1372 |
-
"learning_rate": 1.
|
| 1373 |
-
"loss": 0.
|
| 1374 |
-
"step":
|
| 1375 |
},
|
| 1376 |
{
|
| 1377 |
-
"epoch": 0.
|
| 1378 |
-
"grad_norm":
|
| 1379 |
-
"learning_rate": 1.
|
| 1380 |
-
"loss": 0.
|
| 1381 |
-
"step":
|
| 1382 |
},
|
| 1383 |
{
|
| 1384 |
-
"epoch": 0.
|
| 1385 |
-
"grad_norm": 3.
|
| 1386 |
-
"learning_rate": 1.
|
| 1387 |
-
"loss": 0.
|
| 1388 |
-
"step":
|
| 1389 |
},
|
| 1390 |
{
|
| 1391 |
-
"epoch": 0.
|
| 1392 |
-
"grad_norm":
|
| 1393 |
-
"learning_rate": 1.
|
| 1394 |
-
"loss": 0.
|
| 1395 |
-
"step":
|
| 1396 |
},
|
| 1397 |
{
|
| 1398 |
-
"epoch": 0.
|
| 1399 |
-
"grad_norm":
|
| 1400 |
-
"learning_rate":
|
| 1401 |
-
"loss": 0.
|
| 1402 |
-
"step":
|
| 1403 |
},
|
| 1404 |
{
|
| 1405 |
-
"epoch": 0.
|
| 1406 |
-
"grad_norm":
|
| 1407 |
-
"learning_rate":
|
| 1408 |
-
"loss": 0.
|
| 1409 |
-
"step":
|
| 1410 |
},
|
| 1411 |
{
|
| 1412 |
-
"epoch": 0.
|
| 1413 |
-
"grad_norm":
|
| 1414 |
-
"learning_rate":
|
| 1415 |
-
"loss": 0.
|
| 1416 |
-
"step":
|
| 1417 |
},
|
| 1418 |
{
|
| 1419 |
-
"epoch": 0.
|
| 1420 |
-
"grad_norm":
|
| 1421 |
-
"learning_rate":
|
| 1422 |
-
"loss": 0.
|
| 1423 |
-
"step":
|
| 1424 |
},
|
| 1425 |
{
|
| 1426 |
-
"epoch": 0.
|
| 1427 |
-
"grad_norm":
|
| 1428 |
-
"learning_rate":
|
| 1429 |
-
"loss": 0.
|
| 1430 |
-
"step":
|
| 1431 |
},
|
| 1432 |
{
|
| 1433 |
-
"epoch": 0.
|
| 1434 |
-
"grad_norm":
|
| 1435 |
-
"learning_rate":
|
| 1436 |
-
"loss": 0.
|
| 1437 |
-
"step":
|
| 1438 |
},
|
| 1439 |
{
|
| 1440 |
-
"epoch": 0.
|
| 1441 |
-
"grad_norm":
|
| 1442 |
-
"learning_rate":
|
| 1443 |
-
"loss": 0.
|
| 1444 |
-
"step":
|
| 1445 |
},
|
| 1446 |
{
|
| 1447 |
-
"epoch": 0.
|
| 1448 |
-
"grad_norm":
|
| 1449 |
-
"learning_rate":
|
| 1450 |
-
"loss": 0.
|
| 1451 |
-
"step":
|
| 1452 |
},
|
| 1453 |
{
|
| 1454 |
-
"epoch": 0.
|
| 1455 |
-
"grad_norm":
|
| 1456 |
-
"learning_rate": 1.
|
| 1457 |
-
"loss": 0.
|
| 1458 |
-
"step":
|
| 1459 |
},
|
| 1460 |
{
|
| 1461 |
-
"epoch": 0.
|
| 1462 |
-
"grad_norm":
|
| 1463 |
-
"learning_rate":
|
| 1464 |
-
"loss": 0.
|
| 1465 |
-
"step":
|
| 1466 |
},
|
| 1467 |
{
|
| 1468 |
-
"epoch": 0.
|
| 1469 |
-
"
|
| 1470 |
-
"
|
| 1471 |
-
"
|
| 1472 |
-
"
|
| 1473 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1474 |
}
|
| 1475 |
],
|
| 1476 |
-
"logging_steps":
|
| 1477 |
-
"max_steps":
|
| 1478 |
"num_input_tokens_seen": 0,
|
| 1479 |
"num_train_epochs": 1,
|
| 1480 |
"save_steps": 2500,
|
| 1481 |
-
"total_flos":
|
| 1482 |
"train_batch_size": 1,
|
| 1483 |
"trial_name": null,
|
| 1484 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.8163847923278809,
|
| 3 |
+
"best_model_checkpoint": "runs/deepseek_lora_20240423-223943/checkpoint-2500",
|
| 4 |
+
"epoch": 0.0625,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2500,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 3.086414337158203,
|
| 14 |
+
"learning_rate": 4.0000000000000003e-07,
|
| 15 |
+
"loss": 0.7892,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.0,
|
| 20 |
+
"grad_norm": 8.478134155273438,
|
| 21 |
+
"learning_rate": 8.000000000000001e-07,
|
| 22 |
+
"loss": 0.7746,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.0,
|
| 27 |
+
"grad_norm": 5.574502468109131,
|
| 28 |
+
"learning_rate": 1.2000000000000002e-06,
|
| 29 |
+
"loss": 0.8222,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.0,
|
| 34 |
+
"grad_norm": 2.6497371196746826,
|
| 35 |
+
"learning_rate": 1.6000000000000001e-06,
|
| 36 |
+
"loss": 0.7423,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.0,
|
| 41 |
+
"grad_norm": 3.116753339767456,
|
| 42 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 43 |
+
"loss": 0.7622,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.0,
|
| 48 |
+
"grad_norm": 3.179832696914673,
|
| 49 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 50 |
+
"loss": 0.8183,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.0,
|
| 55 |
+
"grad_norm": 3.9869463443756104,
|
| 56 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 57 |
+
"loss": 0.822,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.0,
|
| 62 |
+
"grad_norm": 5.093494415283203,
|
| 63 |
+
"learning_rate": 3.2000000000000003e-06,
|
| 64 |
+
"loss": 0.7966,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.0,
|
| 69 |
+
"grad_norm": 5.230633735656738,
|
| 70 |
+
"learning_rate": 3.6000000000000003e-06,
|
| 71 |
+
"loss": 0.8113,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.0,
|
| 76 |
+
"grad_norm": 9.374403953552246,
|
| 77 |
"learning_rate": 4.000000000000001e-06,
|
| 78 |
+
"loss": 0.7582,
|
| 79 |
"step": 100
|
| 80 |
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.0,
|
| 83 |
+
"grad_norm": 6.465492248535156,
|
| 84 |
+
"learning_rate": 4.4e-06,
|
| 85 |
+
"loss": 0.7662,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.0,
|
| 90 |
+
"grad_norm": 6.279934883117676,
|
| 91 |
+
"learning_rate": 4.800000000000001e-06,
|
| 92 |
+
"loss": 0.8376,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.0,
|
| 97 |
+
"grad_norm": 5.799221992492676,
|
| 98 |
+
"learning_rate": 5.2e-06,
|
| 99 |
+
"loss": 0.7965,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.0,
|
| 104 |
+
"grad_norm": 3.222240686416626,
|
| 105 |
+
"learning_rate": 5.600000000000001e-06,
|
| 106 |
+
"loss": 0.8855,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.0,
|
| 111 |
+
"grad_norm": 9.009174346923828,
|
| 112 |
+
"learning_rate": 6e-06,
|
| 113 |
+
"loss": 0.8394,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.0,
|
| 118 |
+
"grad_norm": 8.040350914001465,
|
| 119 |
+
"learning_rate": 6.4000000000000006e-06,
|
| 120 |
+
"loss": 0.8426,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.0,
|
| 125 |
+
"grad_norm": 4.131030559539795,
|
| 126 |
+
"learning_rate": 6.800000000000001e-06,
|
| 127 |
+
"loss": 0.7747,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.0,
|
| 132 |
+
"grad_norm": 3.31986927986145,
|
| 133 |
+
"learning_rate": 7.2000000000000005e-06,
|
| 134 |
+
"loss": 0.7125,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.0,
|
| 139 |
+
"grad_norm": 5.7623395919799805,
|
| 140 |
+
"learning_rate": 7.600000000000001e-06,
|
| 141 |
+
"loss": 0.7854,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
{
|
| 145 |
"epoch": 0.01,
|
| 146 |
+
"grad_norm": 10.848206520080566,
|
| 147 |
"learning_rate": 8.000000000000001e-06,
|
| 148 |
+
"loss": 0.7756,
|
| 149 |
"step": 200
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 0.01,
|
| 153 |
+
"grad_norm": 13.455166816711426,
|
| 154 |
+
"learning_rate": 8.400000000000001e-06,
|
| 155 |
+
"loss": 0.7894,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.01,
|
| 160 |
+
"grad_norm": 12.759767532348633,
|
| 161 |
+
"learning_rate": 8.8e-06,
|
| 162 |
+
"loss": 0.7454,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.01,
|
| 167 |
+
"grad_norm": 4.262899875640869,
|
| 168 |
+
"learning_rate": 9.200000000000002e-06,
|
| 169 |
+
"loss": 0.8555,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.01,
|
| 174 |
+
"grad_norm": 4.28985071182251,
|
| 175 |
+
"learning_rate": 9.600000000000001e-06,
|
| 176 |
+
"loss": 0.6845,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.01,
|
| 181 |
+
"grad_norm": 4.174241542816162,
|
| 182 |
+
"learning_rate": 1e-05,
|
| 183 |
+
"loss": 0.7983,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.01,
|
| 188 |
+
"grad_norm": 12.931599617004395,
|
| 189 |
+
"learning_rate": 1.04e-05,
|
| 190 |
+
"loss": 0.9041,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.01,
|
| 195 |
+
"grad_norm": 7.004627227783203,
|
| 196 |
+
"learning_rate": 1.0800000000000002e-05,
|
| 197 |
+
"loss": 0.817,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.01,
|
| 202 |
+
"grad_norm": 3.6102757453918457,
|
| 203 |
+
"learning_rate": 1.1200000000000001e-05,
|
| 204 |
+
"loss": 0.7292,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.01,
|
| 209 |
+
"grad_norm": 2.764902353286743,
|
| 210 |
+
"learning_rate": 1.16e-05,
|
| 211 |
+
"loss": 0.9042,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.01,
|
| 216 |
+
"grad_norm": 3.958317995071411,
|
| 217 |
"learning_rate": 1.2e-05,
|
| 218 |
+
"loss": 0.7539,
|
| 219 |
"step": 300
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.01,
|
| 223 |
+
"grad_norm": 7.098923683166504,
|
| 224 |
+
"learning_rate": 1.2400000000000002e-05,
|
| 225 |
+
"loss": 0.7955,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.01,
|
| 230 |
+
"grad_norm": 12.129098892211914,
|
| 231 |
+
"learning_rate": 1.2800000000000001e-05,
|
| 232 |
+
"loss": 0.849,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.01,
|
| 237 |
+
"grad_norm": 2.054119825363159,
|
| 238 |
+
"learning_rate": 1.3200000000000002e-05,
|
| 239 |
+
"loss": 0.8645,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.01,
|
| 244 |
+
"grad_norm": 5.205028057098389,
|
| 245 |
+
"learning_rate": 1.3600000000000002e-05,
|
| 246 |
+
"loss": 0.8175,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.01,
|
| 251 |
+
"grad_norm": 2.614790439605713,
|
| 252 |
+
"learning_rate": 1.4e-05,
|
| 253 |
+
"loss": 0.8998,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.01,
|
| 258 |
+
"grad_norm": 2.9891204833984375,
|
| 259 |
+
"learning_rate": 1.4400000000000001e-05,
|
| 260 |
+
"loss": 0.8108,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.01,
|
| 265 |
+
"grad_norm": 4.152099609375,
|
| 266 |
+
"learning_rate": 1.48e-05,
|
| 267 |
+
"loss": 0.7855,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.01,
|
| 272 |
+
"grad_norm": 9.833850860595703,
|
| 273 |
+
"learning_rate": 1.5200000000000002e-05,
|
| 274 |
+
"loss": 0.7736,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.01,
|
| 279 |
+
"grad_norm": 3.849621295928955,
|
| 280 |
+
"learning_rate": 1.5600000000000003e-05,
|
| 281 |
+
"loss": 0.7668,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.01,
|
| 286 |
+
"grad_norm": 5.4542975425720215,
|
| 287 |
"learning_rate": 1.6000000000000003e-05,
|
| 288 |
+
"loss": 0.7781,
|
| 289 |
"step": 400
|
| 290 |
},
|
| 291 |
{
|
| 292 |
"epoch": 0.01,
|
| 293 |
+
"grad_norm": 6.197661876678467,
|
| 294 |
+
"learning_rate": 1.64e-05,
|
| 295 |
+
"loss": 0.8654,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.01,
|
| 300 |
+
"grad_norm": 3.2606770992279053,
|
| 301 |
+
"learning_rate": 1.6800000000000002e-05,
|
| 302 |
+
"loss": 0.7565,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.01,
|
| 307 |
+
"grad_norm": 3.9680209159851074,
|
| 308 |
+
"learning_rate": 1.72e-05,
|
| 309 |
+
"loss": 0.7886,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.01,
|
| 314 |
+
"grad_norm": 18.749984741210938,
|
| 315 |
+
"learning_rate": 1.76e-05,
|
| 316 |
+
"loss": 0.7305,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.01,
|
| 321 |
+
"grad_norm": 5.822000503540039,
|
| 322 |
+
"learning_rate": 1.8e-05,
|
| 323 |
+
"loss": 0.7833,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.01,
|
| 328 |
+
"grad_norm": 12.999715805053711,
|
| 329 |
+
"learning_rate": 1.8400000000000003e-05,
|
| 330 |
+
"loss": 0.8483,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.01,
|
| 335 |
+
"grad_norm": 7.193736553192139,
|
| 336 |
+
"learning_rate": 1.88e-05,
|
| 337 |
+
"loss": 0.84,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.01,
|
| 342 |
+
"grad_norm": 12.573124885559082,
|
| 343 |
+
"learning_rate": 1.9200000000000003e-05,
|
| 344 |
+
"loss": 0.8437,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.01,
|
| 349 |
+
"grad_norm": 4.4221601486206055,
|
| 350 |
+
"learning_rate": 1.9600000000000002e-05,
|
| 351 |
+
"loss": 0.6836,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.01,
|
| 356 |
+
"grad_norm": 3.0399410724639893,
|
| 357 |
"learning_rate": 2e-05,
|
| 358 |
+
"loss": 0.8264,
|
| 359 |
"step": 500
|
| 360 |
},
|
| 361 |
{
|
| 362 |
"epoch": 0.01,
|
| 363 |
+
"eval_loss": 0.8175864219665527,
|
| 364 |
+
"eval_runtime": 67.7802,
|
| 365 |
+
"eval_samples_per_second": 14.754,
|
| 366 |
+
"eval_steps_per_second": 14.754,
|
| 367 |
+
"step": 500
|
| 368 |
},
|
| 369 |
{
|
| 370 |
+
"epoch": 0.01,
|
| 371 |
+
"grad_norm": 3.971303701400757,
|
| 372 |
+
"learning_rate": 1.9978947368421054e-05,
|
| 373 |
+
"loss": 0.7385,
|
| 374 |
+
"step": 510
|
| 375 |
},
|
| 376 |
{
|
| 377 |
+
"epoch": 0.01,
|
| 378 |
+
"grad_norm": 3.8043839931488037,
|
| 379 |
+
"learning_rate": 1.9957894736842107e-05,
|
| 380 |
+
"loss": 0.7826,
|
| 381 |
+
"step": 520
|
| 382 |
},
|
| 383 |
{
|
| 384 |
+
"epoch": 0.01,
|
| 385 |
+
"grad_norm": 11.702253341674805,
|
| 386 |
+
"learning_rate": 1.993684210526316e-05,
|
| 387 |
+
"loss": 0.7971,
|
| 388 |
+
"step": 530
|
| 389 |
},
|
| 390 |
{
|
| 391 |
+
"epoch": 0.01,
|
| 392 |
+
"grad_norm": 5.176826000213623,
|
| 393 |
+
"learning_rate": 1.9915789473684212e-05,
|
| 394 |
+
"loss": 0.748,
|
| 395 |
+
"step": 540
|
| 396 |
},
|
| 397 |
{
|
| 398 |
+
"epoch": 0.01,
|
| 399 |
+
"grad_norm": 7.120133876800537,
|
| 400 |
+
"learning_rate": 1.9894736842105265e-05,
|
| 401 |
+
"loss": 0.8461,
|
| 402 |
+
"step": 550
|
| 403 |
},
|
| 404 |
{
|
| 405 |
+
"epoch": 0.01,
|
| 406 |
+
"grad_norm": 12.286151885986328,
|
| 407 |
+
"learning_rate": 1.9873684210526318e-05,
|
| 408 |
+
"loss": 0.8335,
|
| 409 |
+
"step": 560
|
| 410 |
},
|
| 411 |
{
|
| 412 |
+
"epoch": 0.01,
|
| 413 |
+
"grad_norm": 7.857172966003418,
|
| 414 |
+
"learning_rate": 1.985263157894737e-05,
|
| 415 |
+
"loss": 0.7231,
|
| 416 |
+
"step": 570
|
| 417 |
},
|
| 418 |
{
|
| 419 |
+
"epoch": 0.01,
|
| 420 |
+
"grad_norm": 5.327859401702881,
|
| 421 |
+
"learning_rate": 1.9831578947368423e-05,
|
| 422 |
+
"loss": 0.877,
|
| 423 |
+
"step": 580
|
| 424 |
},
|
| 425 |
{
|
| 426 |
+
"epoch": 0.01,
|
| 427 |
+
"grad_norm": 6.9340362548828125,
|
| 428 |
+
"learning_rate": 1.9810526315789476e-05,
|
| 429 |
+
"loss": 0.8984,
|
| 430 |
+
"step": 590
|
| 431 |
},
|
| 432 |
{
|
| 433 |
+
"epoch": 0.01,
|
| 434 |
+
"grad_norm": 2.1034326553344727,
|
| 435 |
+
"learning_rate": 1.9789473684210528e-05,
|
| 436 |
+
"loss": 0.7045,
|
| 437 |
+
"step": 600
|
| 438 |
},
|
| 439 |
{
|
| 440 |
+
"epoch": 0.02,
|
| 441 |
+
"grad_norm": 3.853721857070923,
|
| 442 |
+
"learning_rate": 1.976842105263158e-05,
|
| 443 |
+
"loss": 0.761,
|
| 444 |
+
"step": 610
|
| 445 |
},
|
| 446 |
{
|
| 447 |
+
"epoch": 0.02,
|
| 448 |
+
"grad_norm": 7.6926398277282715,
|
| 449 |
+
"learning_rate": 1.9747368421052633e-05,
|
| 450 |
+
"loss": 0.9493,
|
| 451 |
+
"step": 620
|
| 452 |
},
|
| 453 |
{
|
| 454 |
+
"epoch": 0.02,
|
| 455 |
+
"grad_norm": 6.261799335479736,
|
| 456 |
+
"learning_rate": 1.9726315789473686e-05,
|
| 457 |
+
"loss": 0.7719,
|
| 458 |
+
"step": 630
|
| 459 |
},
|
| 460 |
{
|
| 461 |
+
"epoch": 0.02,
|
| 462 |
+
"grad_norm": 3.864114284515381,
|
| 463 |
+
"learning_rate": 1.970526315789474e-05,
|
| 464 |
+
"loss": 0.9406,
|
| 465 |
+
"step": 640
|
| 466 |
},
|
| 467 |
{
|
| 468 |
+
"epoch": 0.02,
|
| 469 |
+
"grad_norm": 7.093533515930176,
|
| 470 |
+
"learning_rate": 1.968421052631579e-05,
|
| 471 |
+
"loss": 0.7951,
|
| 472 |
+
"step": 650
|
| 473 |
},
|
| 474 |
{
|
| 475 |
+
"epoch": 0.02,
|
| 476 |
+
"grad_norm": 2.3724496364593506,
|
| 477 |
+
"learning_rate": 1.9663157894736844e-05,
|
| 478 |
+
"loss": 0.8648,
|
| 479 |
+
"step": 660
|
| 480 |
},
|
| 481 |
{
|
| 482 |
+
"epoch": 0.02,
|
| 483 |
+
"grad_norm": 10.12341022491455,
|
| 484 |
+
"learning_rate": 1.9642105263157897e-05,
|
| 485 |
+
"loss": 0.7823,
|
| 486 |
+
"step": 670
|
| 487 |
},
|
| 488 |
{
|
| 489 |
+
"epoch": 0.02,
|
| 490 |
+
"grad_norm": 2.80940842628479,
|
| 491 |
+
"learning_rate": 1.962105263157895e-05,
|
| 492 |
+
"loss": 0.706,
|
| 493 |
+
"step": 680
|
| 494 |
},
|
| 495 |
{
|
| 496 |
+
"epoch": 0.02,
|
| 497 |
+
"grad_norm": 8.243487358093262,
|
| 498 |
+
"learning_rate": 1.9600000000000002e-05,
|
| 499 |
+
"loss": 0.8244,
|
| 500 |
+
"step": 690
|
| 501 |
},
|
| 502 |
{
|
| 503 |
+
"epoch": 0.02,
|
| 504 |
+
"grad_norm": 11.420123100280762,
|
| 505 |
+
"learning_rate": 1.9578947368421055e-05,
|
| 506 |
+
"loss": 0.6753,
|
| 507 |
+
"step": 700
|
|
|
|
| 508 |
},
|
| 509 |
{
|
| 510 |
+
"epoch": 0.02,
|
| 511 |
+
"grad_norm": 63.8618278503418,
|
| 512 |
+
"learning_rate": 1.9557894736842107e-05,
|
| 513 |
+
"loss": 0.8309,
|
| 514 |
+
"step": 710
|
| 515 |
},
|
| 516 |
{
|
| 517 |
+
"epoch": 0.02,
|
| 518 |
+
"grad_norm": 4.521258354187012,
|
| 519 |
+
"learning_rate": 1.953684210526316e-05,
|
| 520 |
+
"loss": 0.8101,
|
| 521 |
+
"step": 720
|
| 522 |
},
|
| 523 |
{
|
| 524 |
+
"epoch": 0.02,
|
| 525 |
+
"grad_norm": 2.9532318115234375,
|
| 526 |
+
"learning_rate": 1.9515789473684213e-05,
|
| 527 |
+
"loss": 0.8533,
|
| 528 |
+
"step": 730
|
| 529 |
},
|
| 530 |
{
|
| 531 |
+
"epoch": 0.02,
|
| 532 |
+
"grad_norm": 3.792180061340332,
|
| 533 |
+
"learning_rate": 1.9494736842105265e-05,
|
| 534 |
+
"loss": 0.7573,
|
| 535 |
+
"step": 740
|
| 536 |
},
|
| 537 |
{
|
| 538 |
+
"epoch": 0.02,
|
| 539 |
+
"grad_norm": 5.155513286590576,
|
| 540 |
+
"learning_rate": 1.9473684210526318e-05,
|
| 541 |
+
"loss": 0.8961,
|
| 542 |
+
"step": 750
|
| 543 |
},
|
| 544 |
{
|
| 545 |
+
"epoch": 0.02,
|
| 546 |
+
"grad_norm": 9.195950508117676,
|
| 547 |
+
"learning_rate": 1.945263157894737e-05,
|
| 548 |
+
"loss": 0.8398,
|
| 549 |
+
"step": 760
|
| 550 |
},
|
| 551 |
{
|
| 552 |
+
"epoch": 0.02,
|
| 553 |
+
"grad_norm": 6.699478626251221,
|
| 554 |
+
"learning_rate": 1.9431578947368423e-05,
|
| 555 |
+
"loss": 0.8018,
|
| 556 |
+
"step": 770
|
| 557 |
},
|
| 558 |
{
|
| 559 |
+
"epoch": 0.02,
|
| 560 |
+
"grad_norm": 5.254507541656494,
|
| 561 |
+
"learning_rate": 1.9410526315789476e-05,
|
| 562 |
+
"loss": 0.8408,
|
| 563 |
+
"step": 780
|
| 564 |
},
|
| 565 |
{
|
| 566 |
+
"epoch": 0.02,
|
| 567 |
+
"grad_norm": 4.351966857910156,
|
| 568 |
+
"learning_rate": 1.9389473684210525e-05,
|
| 569 |
+
"loss": 0.7323,
|
| 570 |
+
"step": 790
|
| 571 |
},
|
| 572 |
{
|
| 573 |
+
"epoch": 0.02,
|
| 574 |
+
"grad_norm": 2.361276626586914,
|
| 575 |
+
"learning_rate": 1.936842105263158e-05,
|
| 576 |
+
"loss": 0.8401,
|
| 577 |
+
"step": 800
|
| 578 |
},
|
| 579 |
{
|
| 580 |
+
"epoch": 0.02,
|
| 581 |
+
"grad_norm": 5.449990272521973,
|
| 582 |
+
"learning_rate": 1.9347368421052634e-05,
|
| 583 |
+
"loss": 0.726,
|
| 584 |
+
"step": 810
|
| 585 |
},
|
| 586 |
{
|
| 587 |
+
"epoch": 0.02,
|
| 588 |
+
"grad_norm": 5.375738143920898,
|
| 589 |
+
"learning_rate": 1.9326315789473687e-05,
|
| 590 |
+
"loss": 0.8305,
|
| 591 |
+
"step": 820
|
| 592 |
},
|
| 593 |
{
|
| 594 |
+
"epoch": 0.02,
|
| 595 |
+
"grad_norm": 2.601025342941284,
|
| 596 |
+
"learning_rate": 1.930526315789474e-05,
|
| 597 |
+
"loss": 0.9152,
|
| 598 |
+
"step": 830
|
| 599 |
},
|
| 600 |
{
|
| 601 |
+
"epoch": 0.02,
|
| 602 |
+
"grad_norm": 12.153268814086914,
|
| 603 |
+
"learning_rate": 1.9284210526315792e-05,
|
| 604 |
+
"loss": 0.8423,
|
| 605 |
+
"step": 840
|
| 606 |
},
|
| 607 |
{
|
| 608 |
+
"epoch": 0.02,
|
| 609 |
+
"grad_norm": 3.785663604736328,
|
| 610 |
+
"learning_rate": 1.9263157894736845e-05,
|
| 611 |
+
"loss": 0.7733,
|
| 612 |
+
"step": 850
|
| 613 |
},
|
| 614 |
{
|
| 615 |
+
"epoch": 0.02,
|
| 616 |
+
"grad_norm": 10.162787437438965,
|
| 617 |
+
"learning_rate": 1.9242105263157894e-05,
|
| 618 |
+
"loss": 0.893,
|
| 619 |
+
"step": 860
|
| 620 |
},
|
| 621 |
{
|
| 622 |
+
"epoch": 0.02,
|
| 623 |
+
"grad_norm": 3.871621608734131,
|
| 624 |
+
"learning_rate": 1.922105263157895e-05,
|
| 625 |
+
"loss": 0.798,
|
| 626 |
+
"step": 870
|
| 627 |
},
|
| 628 |
{
|
| 629 |
+
"epoch": 0.02,
|
| 630 |
+
"grad_norm": 2.9919800758361816,
|
| 631 |
+
"learning_rate": 1.9200000000000003e-05,
|
| 632 |
+
"loss": 0.8484,
|
| 633 |
+
"step": 880
|
| 634 |
},
|
| 635 |
{
|
| 636 |
+
"epoch": 0.02,
|
| 637 |
+
"grad_norm": 5.40109920501709,
|
| 638 |
+
"learning_rate": 1.9178947368421055e-05,
|
| 639 |
+
"loss": 0.9129,
|
| 640 |
+
"step": 890
|
| 641 |
},
|
| 642 |
{
|
| 643 |
+
"epoch": 0.02,
|
| 644 |
+
"grad_norm": 6.794926643371582,
|
| 645 |
+
"learning_rate": 1.9157894736842108e-05,
|
| 646 |
+
"loss": 0.8687,
|
| 647 |
+
"step": 900
|
| 648 |
},
|
| 649 |
{
|
| 650 |
+
"epoch": 0.02,
|
| 651 |
+
"grad_norm": 5.942440986633301,
|
| 652 |
+
"learning_rate": 1.913684210526316e-05,
|
| 653 |
+
"loss": 0.8564,
|
| 654 |
+
"step": 910
|
| 655 |
},
|
| 656 |
{
|
| 657 |
+
"epoch": 0.02,
|
| 658 |
+
"grad_norm": 5.968307018280029,
|
| 659 |
+
"learning_rate": 1.9115789473684213e-05,
|
| 660 |
+
"loss": 0.8495,
|
| 661 |
+
"step": 920
|
| 662 |
},
|
| 663 |
{
|
| 664 |
+
"epoch": 0.02,
|
| 665 |
+
"grad_norm": 8.425616264343262,
|
| 666 |
+
"learning_rate": 1.9094736842105262e-05,
|
| 667 |
+
"loss": 0.7242,
|
| 668 |
+
"step": 930
|
| 669 |
},
|
| 670 |
{
|
| 671 |
+
"epoch": 0.02,
|
| 672 |
+
"grad_norm": 2.819301128387451,
|
| 673 |
+
"learning_rate": 1.907368421052632e-05,
|
| 674 |
+
"loss": 0.8381,
|
| 675 |
+
"step": 940
|
| 676 |
},
|
| 677 |
{
|
| 678 |
+
"epoch": 0.02,
|
| 679 |
+
"grad_norm": 6.81688117980957,
|
| 680 |
+
"learning_rate": 1.9052631578947368e-05,
|
| 681 |
+
"loss": 0.8817,
|
| 682 |
+
"step": 950
|
| 683 |
},
|
| 684 |
{
|
| 685 |
+
"epoch": 0.02,
|
| 686 |
+
"grad_norm": 5.102423191070557,
|
| 687 |
+
"learning_rate": 1.9031578947368424e-05,
|
| 688 |
+
"loss": 0.8274,
|
| 689 |
+
"step": 960
|
|
|
|
| 690 |
},
|
| 691 |
{
|
| 692 |
+
"epoch": 0.02,
|
| 693 |
+
"grad_norm": 4.12994909286499,
|
| 694 |
+
"learning_rate": 1.9010526315789476e-05,
|
| 695 |
+
"loss": 0.7052,
|
| 696 |
+
"step": 970
|
| 697 |
},
|
| 698 |
{
|
| 699 |
+
"epoch": 0.02,
|
| 700 |
+
"grad_norm": 5.15468692779541,
|
| 701 |
+
"learning_rate": 1.898947368421053e-05,
|
| 702 |
+
"loss": 0.772,
|
| 703 |
+
"step": 980
|
| 704 |
},
|
| 705 |
{
|
| 706 |
+
"epoch": 0.02,
|
| 707 |
+
"grad_norm": 1.62323796749115,
|
| 708 |
+
"learning_rate": 1.8968421052631582e-05,
|
| 709 |
+
"loss": 0.7764,
|
| 710 |
+
"step": 990
|
| 711 |
},
|
| 712 |
{
|
| 713 |
+
"epoch": 0.03,
|
| 714 |
+
"grad_norm": 2.546677589416504,
|
| 715 |
+
"learning_rate": 1.894736842105263e-05,
|
| 716 |
+
"loss": 0.8365,
|
| 717 |
+
"step": 1000
|
| 718 |
},
|
| 719 |
{
|
| 720 |
+
"epoch": 0.03,
|
| 721 |
+
"eval_loss": 0.7952949404716492,
|
| 722 |
+
"eval_runtime": 67.7544,
|
| 723 |
+
"eval_samples_per_second": 14.759,
|
| 724 |
+
"eval_steps_per_second": 14.759,
|
| 725 |
+
"step": 1000
|
| 726 |
},
|
| 727 |
{
|
| 728 |
+
"epoch": 0.03,
|
| 729 |
+
"grad_norm": 9.28386402130127,
|
| 730 |
+
"learning_rate": 1.8926315789473687e-05,
|
| 731 |
+
"loss": 0.8765,
|
| 732 |
+
"step": 1010
|
| 733 |
},
|
| 734 |
{
|
| 735 |
+
"epoch": 0.03,
|
| 736 |
+
"grad_norm": 7.3430304527282715,
|
| 737 |
+
"learning_rate": 1.8905263157894736e-05,
|
| 738 |
+
"loss": 0.8763,
|
| 739 |
+
"step": 1020
|
| 740 |
},
|
| 741 |
{
|
| 742 |
+
"epoch": 0.03,
|
| 743 |
+
"grad_norm": 4.0531206130981445,
|
| 744 |
+
"learning_rate": 1.8884210526315792e-05,
|
| 745 |
+
"loss": 0.7943,
|
| 746 |
+
"step": 1030
|
| 747 |
},
|
| 748 |
{
|
| 749 |
+
"epoch": 0.03,
|
| 750 |
+
"grad_norm": 3.028320074081421,
|
| 751 |
+
"learning_rate": 1.886315789473684e-05,
|
| 752 |
+
"loss": 0.836,
|
| 753 |
+
"step": 1040
|
| 754 |
},
|
| 755 |
{
|
| 756 |
+
"epoch": 0.03,
|
| 757 |
+
"grad_norm": 3.3861188888549805,
|
| 758 |
+
"learning_rate": 1.8842105263157898e-05,
|
| 759 |
+
"loss": 0.7336,
|
| 760 |
+
"step": 1050
|
| 761 |
},
|
| 762 |
{
|
| 763 |
+
"epoch": 0.03,
|
| 764 |
+
"grad_norm": 3.7832908630371094,
|
| 765 |
+
"learning_rate": 1.882105263157895e-05,
|
| 766 |
+
"loss": 0.9283,
|
| 767 |
+
"step": 1060
|
| 768 |
},
|
| 769 |
{
|
| 770 |
+
"epoch": 0.03,
|
| 771 |
+
"grad_norm": 3.8170342445373535,
|
| 772 |
+
"learning_rate": 1.88e-05,
|
| 773 |
+
"loss": 0.7655,
|
| 774 |
+
"step": 1070
|
| 775 |
},
|
| 776 |
{
|
| 777 |
+
"epoch": 0.03,
|
| 778 |
+
"grad_norm": 6.15322732925415,
|
| 779 |
+
"learning_rate": 1.8778947368421056e-05,
|
| 780 |
+
"loss": 0.9341,
|
| 781 |
+
"step": 1080
|
| 782 |
},
|
| 783 |
{
|
| 784 |
+
"epoch": 0.03,
|
| 785 |
+
"grad_norm": 7.066686153411865,
|
| 786 |
+
"learning_rate": 1.8757894736842105e-05,
|
| 787 |
+
"loss": 0.85,
|
| 788 |
+
"step": 1090
|
| 789 |
},
|
| 790 |
{
|
| 791 |
+
"epoch": 0.03,
|
| 792 |
+
"grad_norm": 2.986961603164673,
|
| 793 |
+
"learning_rate": 1.873684210526316e-05,
|
| 794 |
+
"loss": 0.8943,
|
| 795 |
+
"step": 1100
|
| 796 |
},
|
| 797 |
{
|
| 798 |
+
"epoch": 0.03,
|
| 799 |
+
"grad_norm": 2.8456902503967285,
|
| 800 |
+
"learning_rate": 1.871578947368421e-05,
|
| 801 |
+
"loss": 0.8279,
|
| 802 |
+
"step": 1110
|
| 803 |
},
|
| 804 |
{
|
| 805 |
+
"epoch": 0.03,
|
| 806 |
+
"grad_norm": 3.6177377700805664,
|
| 807 |
+
"learning_rate": 1.8694736842105266e-05,
|
| 808 |
+
"loss": 0.8192,
|
| 809 |
+
"step": 1120
|
| 810 |
},
|
| 811 |
{
|
| 812 |
+
"epoch": 0.03,
|
| 813 |
+
"grad_norm": 14.768010139465332,
|
| 814 |
+
"learning_rate": 1.8673684210526316e-05,
|
| 815 |
+
"loss": 0.8005,
|
| 816 |
+
"step": 1130
|
| 817 |
},
|
| 818 |
{
|
| 819 |
+
"epoch": 0.03,
|
| 820 |
+
"grad_norm": 11.347342491149902,
|
| 821 |
+
"learning_rate": 1.8652631578947368e-05,
|
| 822 |
+
"loss": 0.8081,
|
| 823 |
+
"step": 1140
|
| 824 |
},
|
| 825 |
{
|
| 826 |
+
"epoch": 0.03,
|
| 827 |
+
"grad_norm": 4.0560150146484375,
|
| 828 |
+
"learning_rate": 1.8631578947368424e-05,
|
| 829 |
+
"loss": 0.9389,
|
| 830 |
+
"step": 1150
|
| 831 |
},
|
| 832 |
{
|
| 833 |
+
"epoch": 0.03,
|
| 834 |
+
"grad_norm": 3.3164710998535156,
|
| 835 |
+
"learning_rate": 1.8610526315789473e-05,
|
| 836 |
+
"loss": 0.8501,
|
| 837 |
+
"step": 1160
|
| 838 |
},
|
| 839 |
{
|
| 840 |
+
"epoch": 0.03,
|
| 841 |
+
"grad_norm": 11.112225532531738,
|
| 842 |
+
"learning_rate": 1.858947368421053e-05,
|
| 843 |
+
"loss": 0.7162,
|
| 844 |
+
"step": 1170
|
| 845 |
},
|
| 846 |
{
|
| 847 |
+
"epoch": 0.03,
|
| 848 |
+
"grad_norm": 6.200588703155518,
|
| 849 |
+
"learning_rate": 1.856842105263158e-05,
|
| 850 |
+
"loss": 0.7448,
|
| 851 |
+
"step": 1180
|
| 852 |
},
|
| 853 |
{
|
| 854 |
+
"epoch": 0.03,
|
| 855 |
+
"grad_norm": 6.573482513427734,
|
| 856 |
+
"learning_rate": 1.8547368421052635e-05,
|
| 857 |
+
"loss": 0.8071,
|
| 858 |
+
"step": 1190
|
| 859 |
},
|
| 860 |
{
|
| 861 |
+
"epoch": 0.03,
|
| 862 |
+
"grad_norm": 5.153548717498779,
|
| 863 |
+
"learning_rate": 1.8526315789473684e-05,
|
| 864 |
+
"loss": 0.7957,
|
| 865 |
+
"step": 1200
|
| 866 |
},
|
| 867 |
{
|
| 868 |
+
"epoch": 0.03,
|
| 869 |
+
"grad_norm": 5.3308305740356445,
|
| 870 |
+
"learning_rate": 1.8505263157894737e-05,
|
| 871 |
+
"loss": 0.7301,
|
| 872 |
+
"step": 1210
|
|
|
|
| 873 |
},
|
| 874 |
{
|
| 875 |
+
"epoch": 0.03,
|
| 876 |
+
"grad_norm": 5.269808769226074,
|
| 877 |
+
"learning_rate": 1.8484210526315793e-05,
|
| 878 |
+
"loss": 0.8072,
|
| 879 |
+
"step": 1220
|
| 880 |
},
|
| 881 |
{
|
| 882 |
+
"epoch": 0.03,
|
| 883 |
+
"grad_norm": 5.588324546813965,
|
| 884 |
+
"learning_rate": 1.8463157894736842e-05,
|
| 885 |
+
"loss": 0.8587,
|
| 886 |
+
"step": 1230
|
| 887 |
},
|
| 888 |
{
|
| 889 |
+
"epoch": 0.03,
|
| 890 |
+
"grad_norm": 4.593557357788086,
|
| 891 |
+
"learning_rate": 1.8442105263157898e-05,
|
| 892 |
+
"loss": 0.856,
|
| 893 |
+
"step": 1240
|
| 894 |
},
|
| 895 |
{
|
| 896 |
+
"epoch": 0.03,
|
| 897 |
+
"grad_norm": 5.2591094970703125,
|
| 898 |
+
"learning_rate": 1.8421052631578947e-05,
|
| 899 |
+
"loss": 0.7717,
|
| 900 |
+
"step": 1250
|
| 901 |
},
|
| 902 |
{
|
| 903 |
+
"epoch": 0.03,
|
| 904 |
+
"grad_norm": 4.052567958831787,
|
| 905 |
+
"learning_rate": 1.8400000000000003e-05,
|
| 906 |
+
"loss": 0.7823,
|
| 907 |
+
"step": 1260
|
| 908 |
},
|
| 909 |
{
|
| 910 |
+
"epoch": 0.03,
|
| 911 |
+
"grad_norm": 4.447838306427002,
|
| 912 |
+
"learning_rate": 1.8378947368421053e-05,
|
| 913 |
+
"loss": 0.83,
|
| 914 |
+
"step": 1270
|
| 915 |
},
|
| 916 |
{
|
| 917 |
+
"epoch": 0.03,
|
| 918 |
+
"grad_norm": 4.029257774353027,
|
| 919 |
+
"learning_rate": 1.8357894736842105e-05,
|
| 920 |
+
"loss": 0.7504,
|
| 921 |
+
"step": 1280
|
| 922 |
},
|
| 923 |
{
|
| 924 |
+
"epoch": 0.03,
|
| 925 |
+
"grad_norm": 9.053960800170898,
|
| 926 |
+
"learning_rate": 1.8336842105263158e-05,
|
| 927 |
+
"loss": 0.9074,
|
| 928 |
+
"step": 1290
|
| 929 |
},
|
| 930 |
{
|
| 931 |
+
"epoch": 0.03,
|
| 932 |
+
"grad_norm": 2.2877705097198486,
|
| 933 |
+
"learning_rate": 1.831578947368421e-05,
|
| 934 |
+
"loss": 0.772,
|
| 935 |
+
"step": 1300
|
| 936 |
},
|
| 937 |
{
|
| 938 |
+
"epoch": 0.03,
|
| 939 |
+
"grad_norm": 3.4482290744781494,
|
| 940 |
+
"learning_rate": 1.8294736842105267e-05,
|
| 941 |
+
"loss": 0.8658,
|
| 942 |
+
"step": 1310
|
| 943 |
},
|
| 944 |
{
|
| 945 |
+
"epoch": 0.03,
|
| 946 |
+
"grad_norm": 6.684794902801514,
|
| 947 |
+
"learning_rate": 1.8273684210526316e-05,
|
| 948 |
+
"loss": 0.7848,
|
| 949 |
+
"step": 1320
|
| 950 |
},
|
| 951 |
{
|
| 952 |
+
"epoch": 0.03,
|
| 953 |
+
"grad_norm": 3.553828716278076,
|
| 954 |
+
"learning_rate": 1.8252631578947372e-05,
|
| 955 |
+
"loss": 0.8219,
|
| 956 |
+
"step": 1330
|
| 957 |
},
|
| 958 |
{
|
| 959 |
+
"epoch": 0.03,
|
| 960 |
+
"grad_norm": 2.5203397274017334,
|
| 961 |
+
"learning_rate": 1.823157894736842e-05,
|
| 962 |
+
"loss": 0.9071,
|
| 963 |
+
"step": 1340
|
| 964 |
},
|
| 965 |
{
|
| 966 |
+
"epoch": 0.03,
|
| 967 |
+
"grad_norm": 4.961795806884766,
|
| 968 |
+
"learning_rate": 1.8210526315789477e-05,
|
| 969 |
+
"loss": 0.6542,
|
| 970 |
+
"step": 1350
|
| 971 |
},
|
| 972 |
{
|
| 973 |
+
"epoch": 0.03,
|
| 974 |
+
"grad_norm": 3.663081645965576,
|
| 975 |
+
"learning_rate": 1.8189473684210527e-05,
|
| 976 |
+
"loss": 0.7402,
|
| 977 |
+
"step": 1360
|
| 978 |
},
|
| 979 |
{
|
| 980 |
+
"epoch": 0.03,
|
| 981 |
+
"grad_norm": 8.785040855407715,
|
| 982 |
+
"learning_rate": 1.816842105263158e-05,
|
| 983 |
+
"loss": 0.7462,
|
| 984 |
+
"step": 1370
|
| 985 |
},
|
| 986 |
{
|
| 987 |
+
"epoch": 0.03,
|
| 988 |
+
"grad_norm": 4.659074783325195,
|
| 989 |
+
"learning_rate": 1.8147368421052632e-05,
|
| 990 |
+
"loss": 0.6951,
|
| 991 |
+
"step": 1380
|
| 992 |
},
|
| 993 |
{
|
| 994 |
+
"epoch": 0.03,
|
| 995 |
+
"grad_norm": 3.5885703563690186,
|
| 996 |
+
"learning_rate": 1.8126315789473685e-05,
|
| 997 |
+
"loss": 0.7008,
|
| 998 |
+
"step": 1390
|
| 999 |
},
|
| 1000 |
{
|
| 1001 |
+
"epoch": 0.04,
|
| 1002 |
+
"grad_norm": 3.1295347213745117,
|
| 1003 |
+
"learning_rate": 1.810526315789474e-05,
|
| 1004 |
+
"loss": 0.9103,
|
| 1005 |
+
"step": 1400
|
| 1006 |
},
|
| 1007 |
{
|
| 1008 |
+
"epoch": 0.04,
|
| 1009 |
+
"grad_norm": 2.4699888229370117,
|
| 1010 |
+
"learning_rate": 1.808421052631579e-05,
|
| 1011 |
+
"loss": 0.841,
|
| 1012 |
+
"step": 1410
|
| 1013 |
},
|
| 1014 |
{
|
| 1015 |
+
"epoch": 0.04,
|
| 1016 |
+
"grad_norm": 5.3273444175720215,
|
| 1017 |
+
"learning_rate": 1.8063157894736846e-05,
|
| 1018 |
+
"loss": 0.9041,
|
| 1019 |
+
"step": 1420
|
| 1020 |
},
|
| 1021 |
{
|
| 1022 |
+
"epoch": 0.04,
|
| 1023 |
+
"grad_norm": 5.149638652801514,
|
| 1024 |
+
"learning_rate": 1.8042105263157895e-05,
|
| 1025 |
+
"loss": 0.7784,
|
| 1026 |
+
"step": 1430
|
| 1027 |
},
|
| 1028 |
{
|
| 1029 |
+
"epoch": 0.04,
|
| 1030 |
+
"grad_norm": 3.4124910831451416,
|
| 1031 |
+
"learning_rate": 1.8021052631578948e-05,
|
| 1032 |
+
"loss": 0.8208,
|
| 1033 |
+
"step": 1440
|
| 1034 |
},
|
| 1035 |
{
|
| 1036 |
+
"epoch": 0.04,
|
| 1037 |
+
"grad_norm": 2.9231085777282715,
|
| 1038 |
+
"learning_rate": 1.8e-05,
|
| 1039 |
+
"loss": 0.7173,
|
| 1040 |
+
"step": 1450
|
| 1041 |
},
|
| 1042 |
{
|
| 1043 |
+
"epoch": 0.04,
|
| 1044 |
+
"grad_norm": 4.008113384246826,
|
| 1045 |
+
"learning_rate": 1.7978947368421053e-05,
|
| 1046 |
+
"loss": 0.7383,
|
| 1047 |
+
"step": 1460
|
| 1048 |
},
|
| 1049 |
{
|
| 1050 |
+
"epoch": 0.04,
|
| 1051 |
+
"grad_norm": 5.1748046875,
|
| 1052 |
+
"learning_rate": 1.795789473684211e-05,
|
| 1053 |
+
"loss": 0.8399,
|
| 1054 |
+
"step": 1470
|
|
|
|
| 1055 |
},
|
| 1056 |
{
|
| 1057 |
+
"epoch": 0.04,
|
| 1058 |
+
"grad_norm": 3.4990293979644775,
|
| 1059 |
+
"learning_rate": 1.793684210526316e-05,
|
| 1060 |
+
"loss": 0.6721,
|
| 1061 |
+
"step": 1480
|
| 1062 |
},
|
| 1063 |
{
|
| 1064 |
+
"epoch": 0.04,
|
| 1065 |
+
"grad_norm": 3.1186299324035645,
|
| 1066 |
+
"learning_rate": 1.7915789473684214e-05,
|
| 1067 |
+
"loss": 0.782,
|
| 1068 |
+
"step": 1490
|
| 1069 |
},
|
| 1070 |
{
|
| 1071 |
+
"epoch": 0.04,
|
| 1072 |
+
"grad_norm": 5.12732458114624,
|
| 1073 |
+
"learning_rate": 1.7894736842105264e-05,
|
| 1074 |
+
"loss": 0.7211,
|
| 1075 |
+
"step": 1500
|
| 1076 |
},
|
| 1077 |
{
|
| 1078 |
+
"epoch": 0.04,
|
| 1079 |
+
"eval_loss": 0.811568021774292,
|
| 1080 |
+
"eval_runtime": 67.7961,
|
| 1081 |
+
"eval_samples_per_second": 14.75,
|
| 1082 |
+
"eval_steps_per_second": 14.75,
|
| 1083 |
+
"step": 1500
|
| 1084 |
},
|
| 1085 |
{
|
| 1086 |
+
"epoch": 0.04,
|
| 1087 |
+
"grad_norm": 3.631096124649048,
|
| 1088 |
+
"learning_rate": 1.7873684210526316e-05,
|
| 1089 |
+
"loss": 0.7557,
|
| 1090 |
+
"step": 1510
|
| 1091 |
},
|
| 1092 |
{
|
| 1093 |
+
"epoch": 0.04,
|
| 1094 |
+
"grad_norm": 8.850045204162598,
|
| 1095 |
+
"learning_rate": 1.785263157894737e-05,
|
| 1096 |
+
"loss": 0.8757,
|
| 1097 |
+
"step": 1520
|
| 1098 |
},
|
| 1099 |
{
|
| 1100 |
+
"epoch": 0.04,
|
| 1101 |
+
"grad_norm": 3.1114978790283203,
|
| 1102 |
+
"learning_rate": 1.7831578947368422e-05,
|
| 1103 |
+
"loss": 0.7613,
|
| 1104 |
+
"step": 1530
|
| 1105 |
},
|
| 1106 |
{
|
| 1107 |
+
"epoch": 0.04,
|
| 1108 |
+
"grad_norm": 4.5038743019104,
|
| 1109 |
+
"learning_rate": 1.7810526315789474e-05,
|
| 1110 |
+
"loss": 0.8049,
|
| 1111 |
+
"step": 1540
|
| 1112 |
},
|
| 1113 |
{
|
| 1114 |
+
"epoch": 0.04,
|
| 1115 |
+
"grad_norm": 4.2331156730651855,
|
| 1116 |
+
"learning_rate": 1.7789473684210527e-05,
|
| 1117 |
+
"loss": 0.8277,
|
| 1118 |
+
"step": 1550
|
| 1119 |
},
|
| 1120 |
{
|
| 1121 |
+
"epoch": 0.04,
|
| 1122 |
+
"grad_norm": 5.05696964263916,
|
| 1123 |
+
"learning_rate": 1.7768421052631583e-05,
|
| 1124 |
+
"loss": 0.7973,
|
| 1125 |
+
"step": 1560
|
| 1126 |
},
|
| 1127 |
{
|
| 1128 |
+
"epoch": 0.04,
|
| 1129 |
+
"grad_norm": 2.1331920623779297,
|
| 1130 |
+
"learning_rate": 1.7747368421052632e-05,
|
| 1131 |
+
"loss": 0.7688,
|
| 1132 |
+
"step": 1570
|
| 1133 |
},
|
| 1134 |
{
|
| 1135 |
+
"epoch": 0.04,
|
| 1136 |
+
"grad_norm": 4.984541416168213,
|
| 1137 |
+
"learning_rate": 1.7726315789473685e-05,
|
| 1138 |
+
"loss": 0.7865,
|
| 1139 |
+
"step": 1580
|
| 1140 |
},
|
| 1141 |
{
|
| 1142 |
+
"epoch": 0.04,
|
| 1143 |
+
"grad_norm": 7.149406433105469,
|
| 1144 |
+
"learning_rate": 1.7705263157894738e-05,
|
| 1145 |
+
"loss": 0.7728,
|
| 1146 |
+
"step": 1590
|
| 1147 |
},
|
| 1148 |
{
|
| 1149 |
+
"epoch": 0.04,
|
| 1150 |
+
"grad_norm": 8.092243194580078,
|
| 1151 |
+
"learning_rate": 1.768421052631579e-05,
|
| 1152 |
+
"loss": 0.935,
|
| 1153 |
+
"step": 1600
|
| 1154 |
},
|
| 1155 |
{
|
| 1156 |
+
"epoch": 0.04,
|
| 1157 |
+
"grad_norm": 13.16551399230957,
|
| 1158 |
+
"learning_rate": 1.7663157894736843e-05,
|
| 1159 |
+
"loss": 0.8286,
|
| 1160 |
+
"step": 1610
|
| 1161 |
},
|
| 1162 |
{
|
| 1163 |
+
"epoch": 0.04,
|
| 1164 |
+
"grad_norm": 2.131350517272949,
|
| 1165 |
+
"learning_rate": 1.7642105263157896e-05,
|
| 1166 |
+
"loss": 0.7864,
|
| 1167 |
+
"step": 1620
|
| 1168 |
},
|
| 1169 |
{
|
| 1170 |
+
"epoch": 0.04,
|
| 1171 |
+
"grad_norm": 7.870023727416992,
|
| 1172 |
+
"learning_rate": 1.7621052631578948e-05,
|
| 1173 |
+
"loss": 0.8645,
|
| 1174 |
+
"step": 1630
|
| 1175 |
},
|
| 1176 |
{
|
| 1177 |
+
"epoch": 0.04,
|
| 1178 |
+
"grad_norm": 10.631692886352539,
|
| 1179 |
+
"learning_rate": 1.76e-05,
|
| 1180 |
+
"loss": 0.8473,
|
| 1181 |
+
"step": 1640
|
| 1182 |
},
|
| 1183 |
{
|
| 1184 |
+
"epoch": 0.04,
|
| 1185 |
+
"grad_norm": 6.421032905578613,
|
| 1186 |
+
"learning_rate": 1.7578947368421054e-05,
|
| 1187 |
+
"loss": 0.7868,
|
| 1188 |
+
"step": 1650
|
| 1189 |
},
|
| 1190 |
{
|
| 1191 |
+
"epoch": 0.04,
|
| 1192 |
+
"grad_norm": 4.57529878616333,
|
| 1193 |
+
"learning_rate": 1.7557894736842106e-05,
|
| 1194 |
+
"loss": 0.7882,
|
| 1195 |
+
"step": 1660
|
| 1196 |
},
|
| 1197 |
{
|
| 1198 |
+
"epoch": 0.04,
|
| 1199 |
+
"grad_norm": 3.8785624504089355,
|
| 1200 |
+
"learning_rate": 1.753684210526316e-05,
|
| 1201 |
+
"loss": 0.7543,
|
| 1202 |
+
"step": 1670
|
| 1203 |
},
|
| 1204 |
{
|
| 1205 |
+
"epoch": 0.04,
|
| 1206 |
+
"grad_norm": 5.722006320953369,
|
| 1207 |
+
"learning_rate": 1.751578947368421e-05,
|
| 1208 |
+
"loss": 0.9626,
|
| 1209 |
+
"step": 1680
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.04,
|
| 1213 |
+
"grad_norm": 2.466771364212036,
|
| 1214 |
+
"learning_rate": 1.7494736842105264e-05,
|
| 1215 |
+
"loss": 0.783,
|
| 1216 |
+
"step": 1690
|
| 1217 |
},
|
| 1218 |
{
|
| 1219 |
+
"epoch": 0.04,
|
| 1220 |
+
"grad_norm": 3.072049856185913,
|
| 1221 |
+
"learning_rate": 1.7473684210526317e-05,
|
| 1222 |
+
"loss": 0.7503,
|
| 1223 |
+
"step": 1700
|
| 1224 |
},
|
| 1225 |
{
|
| 1226 |
+
"epoch": 0.04,
|
| 1227 |
+
"grad_norm": 5.768575668334961,
|
| 1228 |
+
"learning_rate": 1.745263157894737e-05,
|
| 1229 |
+
"loss": 0.8193,
|
| 1230 |
+
"step": 1710
|
| 1231 |
},
|
| 1232 |
{
|
| 1233 |
+
"epoch": 0.04,
|
| 1234 |
+
"grad_norm": 2.585022211074829,
|
| 1235 |
+
"learning_rate": 1.7431578947368422e-05,
|
| 1236 |
+
"loss": 0.8808,
|
| 1237 |
+
"step": 1720
|
| 1238 |
},
|
| 1239 |
{
|
| 1240 |
+
"epoch": 0.04,
|
| 1241 |
+
"grad_norm": 3.0711567401885986,
|
| 1242 |
+
"learning_rate": 1.7410526315789475e-05,
|
| 1243 |
+
"loss": 0.8098,
|
| 1244 |
+
"step": 1730
|
|
|
|
| 1245 |
},
|
| 1246 |
{
|
| 1247 |
+
"epoch": 0.04,
|
| 1248 |
+
"grad_norm": 3.3020272254943848,
|
| 1249 |
+
"learning_rate": 1.7389473684210527e-05,
|
| 1250 |
+
"loss": 0.7196,
|
| 1251 |
+
"step": 1740
|
| 1252 |
},
|
| 1253 |
{
|
| 1254 |
+
"epoch": 0.04,
|
| 1255 |
+
"grad_norm": 3.645238161087036,
|
| 1256 |
+
"learning_rate": 1.736842105263158e-05,
|
| 1257 |
+
"loss": 0.8904,
|
| 1258 |
+
"step": 1750
|
| 1259 |
},
|
| 1260 |
{
|
| 1261 |
+
"epoch": 0.04,
|
| 1262 |
+
"grad_norm": 6.018638610839844,
|
| 1263 |
+
"learning_rate": 1.7347368421052633e-05,
|
| 1264 |
+
"loss": 0.7937,
|
| 1265 |
+
"step": 1760
|
| 1266 |
},
|
| 1267 |
{
|
| 1268 |
+
"epoch": 0.04,
|
| 1269 |
+
"grad_norm": 3.629096746444702,
|
| 1270 |
+
"learning_rate": 1.7326315789473685e-05,
|
| 1271 |
+
"loss": 0.9171,
|
| 1272 |
+
"step": 1770
|
| 1273 |
},
|
| 1274 |
{
|
| 1275 |
+
"epoch": 0.04,
|
| 1276 |
+
"grad_norm": 2.5619189739227295,
|
| 1277 |
+
"learning_rate": 1.7305263157894738e-05,
|
| 1278 |
+
"loss": 0.9488,
|
| 1279 |
+
"step": 1780
|
| 1280 |
},
|
| 1281 |
{
|
| 1282 |
+
"epoch": 0.04,
|
| 1283 |
+
"grad_norm": 9.464752197265625,
|
| 1284 |
+
"learning_rate": 1.728421052631579e-05,
|
| 1285 |
+
"loss": 0.8459,
|
| 1286 |
+
"step": 1790
|
| 1287 |
},
|
| 1288 |
{
|
| 1289 |
+
"epoch": 0.04,
|
| 1290 |
+
"grad_norm": 3.9856364727020264,
|
| 1291 |
+
"learning_rate": 1.7263157894736843e-05,
|
| 1292 |
+
"loss": 0.8378,
|
| 1293 |
+
"step": 1800
|
| 1294 |
},
|
| 1295 |
{
|
| 1296 |
+
"epoch": 0.05,
|
| 1297 |
+
"grad_norm": 3.753553867340088,
|
| 1298 |
+
"learning_rate": 1.7242105263157896e-05,
|
| 1299 |
+
"loss": 0.8093,
|
| 1300 |
+
"step": 1810
|
| 1301 |
},
|
| 1302 |
{
|
| 1303 |
+
"epoch": 0.05,
|
| 1304 |
+
"grad_norm": 3.4593358039855957,
|
| 1305 |
+
"learning_rate": 1.722105263157895e-05,
|
| 1306 |
+
"loss": 0.7896,
|
| 1307 |
+
"step": 1820
|
| 1308 |
},
|
| 1309 |
{
|
| 1310 |
+
"epoch": 0.05,
|
| 1311 |
+
"grad_norm": 2.7163546085357666,
|
| 1312 |
+
"learning_rate": 1.72e-05,
|
| 1313 |
+
"loss": 0.7188,
|
| 1314 |
+
"step": 1830
|
| 1315 |
},
|
| 1316 |
{
|
| 1317 |
+
"epoch": 0.05,
|
| 1318 |
+
"grad_norm": 3.105628728866577,
|
| 1319 |
+
"learning_rate": 1.7178947368421054e-05,
|
| 1320 |
+
"loss": 0.7643,
|
| 1321 |
+
"step": 1840
|
| 1322 |
},
|
| 1323 |
{
|
| 1324 |
+
"epoch": 0.05,
|
| 1325 |
+
"grad_norm": 2.387368679046631,
|
| 1326 |
+
"learning_rate": 1.7157894736842107e-05,
|
| 1327 |
+
"loss": 0.8465,
|
| 1328 |
+
"step": 1850
|
| 1329 |
},
|
| 1330 |
{
|
| 1331 |
+
"epoch": 0.05,
|
| 1332 |
+
"grad_norm": 6.020385265350342,
|
| 1333 |
+
"learning_rate": 1.713684210526316e-05,
|
| 1334 |
+
"loss": 0.7798,
|
| 1335 |
+
"step": 1860
|
| 1336 |
},
|
| 1337 |
{
|
| 1338 |
+
"epoch": 0.05,
|
| 1339 |
+
"grad_norm": 4.560520172119141,
|
| 1340 |
+
"learning_rate": 1.7115789473684212e-05,
|
| 1341 |
+
"loss": 0.7704,
|
| 1342 |
+
"step": 1870
|
| 1343 |
},
|
| 1344 |
{
|
| 1345 |
+
"epoch": 0.05,
|
| 1346 |
+
"grad_norm": 15.739727973937988,
|
| 1347 |
+
"learning_rate": 1.7094736842105265e-05,
|
| 1348 |
+
"loss": 0.7148,
|
| 1349 |
+
"step": 1880
|
| 1350 |
},
|
| 1351 |
{
|
| 1352 |
+
"epoch": 0.05,
|
| 1353 |
+
"grad_norm": 5.79690408706665,
|
| 1354 |
+
"learning_rate": 1.7073684210526317e-05,
|
| 1355 |
+
"loss": 0.798,
|
| 1356 |
+
"step": 1890
|
| 1357 |
},
|
| 1358 |
{
|
| 1359 |
+
"epoch": 0.05,
|
| 1360 |
+
"grad_norm": 2.6939146518707275,
|
| 1361 |
+
"learning_rate": 1.705263157894737e-05,
|
| 1362 |
+
"loss": 0.7641,
|
| 1363 |
+
"step": 1900
|
| 1364 |
},
|
| 1365 |
{
|
| 1366 |
+
"epoch": 0.05,
|
| 1367 |
+
"grad_norm": 5.193384170532227,
|
| 1368 |
+
"learning_rate": 1.7031578947368423e-05,
|
| 1369 |
+
"loss": 0.7866,
|
| 1370 |
+
"step": 1910
|
| 1371 |
},
|
| 1372 |
{
|
| 1373 |
+
"epoch": 0.05,
|
| 1374 |
+
"grad_norm": 4.940731525421143,
|
| 1375 |
+
"learning_rate": 1.7010526315789475e-05,
|
| 1376 |
+
"loss": 0.8261,
|
| 1377 |
+
"step": 1920
|
| 1378 |
},
|
| 1379 |
{
|
| 1380 |
+
"epoch": 0.05,
|
| 1381 |
+
"grad_norm": 2.1812446117401123,
|
| 1382 |
+
"learning_rate": 1.6989473684210528e-05,
|
| 1383 |
+
"loss": 0.7973,
|
| 1384 |
+
"step": 1930
|
| 1385 |
},
|
| 1386 |
{
|
| 1387 |
+
"epoch": 0.05,
|
| 1388 |
+
"grad_norm": 3.7413289546966553,
|
| 1389 |
+
"learning_rate": 1.696842105263158e-05,
|
| 1390 |
+
"loss": 0.7818,
|
| 1391 |
+
"step": 1940
|
| 1392 |
},
|
| 1393 |
{
|
| 1394 |
+
"epoch": 0.05,
|
| 1395 |
+
"grad_norm": 4.024014472961426,
|
| 1396 |
+
"learning_rate": 1.6947368421052633e-05,
|
| 1397 |
+
"loss": 0.7237,
|
| 1398 |
+
"step": 1950
|
| 1399 |
},
|
| 1400 |
{
|
| 1401 |
+
"epoch": 0.05,
|
| 1402 |
+
"grad_norm": 3.0871291160583496,
|
| 1403 |
+
"learning_rate": 1.6926315789473686e-05,
|
| 1404 |
+
"loss": 0.772,
|
| 1405 |
+
"step": 1960
|
| 1406 |
},
|
| 1407 |
{
|
| 1408 |
+
"epoch": 0.05,
|
| 1409 |
+
"grad_norm": 3.28814435005188,
|
| 1410 |
+
"learning_rate": 1.690526315789474e-05,
|
| 1411 |
+
"loss": 0.7067,
|
| 1412 |
+
"step": 1970
|
| 1413 |
},
|
| 1414 |
{
|
| 1415 |
+
"epoch": 0.05,
|
| 1416 |
+
"grad_norm": 2.8241286277770996,
|
| 1417 |
+
"learning_rate": 1.688421052631579e-05,
|
| 1418 |
+
"loss": 0.8175,
|
| 1419 |
+
"step": 1980
|
| 1420 |
},
|
| 1421 |
{
|
| 1422 |
+
"epoch": 0.05,
|
| 1423 |
+
"grad_norm": 2.5942068099975586,
|
| 1424 |
+
"learning_rate": 1.6863157894736844e-05,
|
| 1425 |
+
"loss": 0.9265,
|
| 1426 |
+
"step": 1990
|
|
|
|
| 1427 |
},
|
| 1428 |
{
|
| 1429 |
+
"epoch": 0.05,
|
| 1430 |
+
"grad_norm": 6.6822662353515625,
|
| 1431 |
+
"learning_rate": 1.6842105263157896e-05,
|
| 1432 |
+
"loss": 0.8593,
|
| 1433 |
+
"step": 2000
|
| 1434 |
},
|
| 1435 |
{
|
| 1436 |
+
"epoch": 0.05,
|
| 1437 |
+
"eval_loss": 0.8064771890640259,
|
| 1438 |
+
"eval_runtime": 67.7887,
|
| 1439 |
+
"eval_samples_per_second": 14.752,
|
| 1440 |
+
"eval_steps_per_second": 14.752,
|
| 1441 |
+
"step": 2000
|
| 1442 |
},
|
| 1443 |
{
|
| 1444 |
+
"epoch": 0.05,
|
| 1445 |
+
"grad_norm": 7.032164573669434,
|
| 1446 |
+
"learning_rate": 1.682105263157895e-05,
|
| 1447 |
+
"loss": 0.8819,
|
| 1448 |
+
"step": 2010
|
| 1449 |
},
|
| 1450 |
{
|
| 1451 |
+
"epoch": 0.05,
|
| 1452 |
+
"grad_norm": 4.874982833862305,
|
| 1453 |
+
"learning_rate": 1.6800000000000002e-05,
|
| 1454 |
+
"loss": 0.8021,
|
| 1455 |
+
"step": 2020
|
| 1456 |
},
|
| 1457 |
{
|
| 1458 |
+
"epoch": 0.05,
|
| 1459 |
+
"grad_norm": 2.6172547340393066,
|
| 1460 |
+
"learning_rate": 1.6778947368421054e-05,
|
| 1461 |
+
"loss": 0.8017,
|
| 1462 |
+
"step": 2030
|
| 1463 |
},
|
| 1464 |
{
|
| 1465 |
+
"epoch": 0.05,
|
| 1466 |
+
"grad_norm": 10.659741401672363,
|
| 1467 |
+
"learning_rate": 1.6757894736842107e-05,
|
| 1468 |
+
"loss": 0.8896,
|
| 1469 |
+
"step": 2040
|
| 1470 |
},
|
| 1471 |
{
|
| 1472 |
+
"epoch": 0.05,
|
| 1473 |
+
"grad_norm": 6.189141750335693,
|
| 1474 |
+
"learning_rate": 1.673684210526316e-05,
|
| 1475 |
+
"loss": 0.7997,
|
| 1476 |
+
"step": 2050
|
| 1477 |
},
|
| 1478 |
{
|
| 1479 |
+
"epoch": 0.05,
|
| 1480 |
+
"grad_norm": 4.523468971252441,
|
| 1481 |
+
"learning_rate": 1.6715789473684212e-05,
|
| 1482 |
+
"loss": 0.8498,
|
| 1483 |
+
"step": 2060
|
| 1484 |
},
|
| 1485 |
{
|
| 1486 |
+
"epoch": 0.05,
|
| 1487 |
+
"grad_norm": 8.533658981323242,
|
| 1488 |
+
"learning_rate": 1.6694736842105265e-05,
|
| 1489 |
+
"loss": 0.8857,
|
| 1490 |
+
"step": 2070
|
| 1491 |
},
|
| 1492 |
{
|
| 1493 |
+
"epoch": 0.05,
|
| 1494 |
+
"grad_norm": 3.0041606426239014,
|
| 1495 |
+
"learning_rate": 1.6673684210526318e-05,
|
| 1496 |
+
"loss": 0.8112,
|
| 1497 |
+
"step": 2080
|
| 1498 |
},
|
| 1499 |
{
|
| 1500 |
+
"epoch": 0.05,
|
| 1501 |
+
"grad_norm": 5.055651664733887,
|
| 1502 |
+
"learning_rate": 1.665263157894737e-05,
|
| 1503 |
+
"loss": 0.7872,
|
| 1504 |
+
"step": 2090
|
| 1505 |
},
|
| 1506 |
{
|
| 1507 |
+
"epoch": 0.05,
|
| 1508 |
+
"grad_norm": 5.761922836303711,
|
| 1509 |
+
"learning_rate": 1.6631578947368423e-05,
|
| 1510 |
+
"loss": 0.7727,
|
| 1511 |
+
"step": 2100
|
| 1512 |
},
|
| 1513 |
{
|
| 1514 |
+
"epoch": 0.05,
|
| 1515 |
+
"grad_norm": 2.518223524093628,
|
| 1516 |
+
"learning_rate": 1.6610526315789476e-05,
|
| 1517 |
+
"loss": 0.7997,
|
| 1518 |
+
"step": 2110
|
| 1519 |
},
|
| 1520 |
{
|
| 1521 |
+
"epoch": 0.05,
|
| 1522 |
+
"grad_norm": 4.975761890411377,
|
| 1523 |
+
"learning_rate": 1.658947368421053e-05,
|
| 1524 |
+
"loss": 0.7457,
|
| 1525 |
+
"step": 2120
|
| 1526 |
},
|
| 1527 |
{
|
| 1528 |
+
"epoch": 0.05,
|
| 1529 |
+
"grad_norm": 3.2227561473846436,
|
| 1530 |
+
"learning_rate": 1.656842105263158e-05,
|
| 1531 |
+
"loss": 0.816,
|
| 1532 |
+
"step": 2130
|
| 1533 |
},
|
| 1534 |
{
|
| 1535 |
+
"epoch": 0.05,
|
| 1536 |
+
"grad_norm": 4.705923080444336,
|
| 1537 |
+
"learning_rate": 1.6547368421052634e-05,
|
| 1538 |
+
"loss": 0.8113,
|
| 1539 |
+
"step": 2140
|
| 1540 |
},
|
| 1541 |
{
|
| 1542 |
+
"epoch": 0.05,
|
| 1543 |
+
"grad_norm": 2.655057430267334,
|
| 1544 |
+
"learning_rate": 1.6526315789473686e-05,
|
| 1545 |
+
"loss": 0.7912,
|
| 1546 |
+
"step": 2150
|
| 1547 |
},
|
| 1548 |
{
|
| 1549 |
+
"epoch": 0.05,
|
| 1550 |
+
"grad_norm": 3.0186755657196045,
|
| 1551 |
+
"learning_rate": 1.650526315789474e-05,
|
| 1552 |
+
"loss": 0.8608,
|
| 1553 |
+
"step": 2160
|
| 1554 |
},
|
| 1555 |
{
|
| 1556 |
+
"epoch": 0.05,
|
| 1557 |
+
"grad_norm": 1.232386827468872,
|
| 1558 |
+
"learning_rate": 1.648421052631579e-05,
|
| 1559 |
+
"loss": 0.8549,
|
| 1560 |
+
"step": 2170
|
| 1561 |
},
|
| 1562 |
{
|
| 1563 |
+
"epoch": 0.05,
|
| 1564 |
+
"grad_norm": 11.968620300292969,
|
| 1565 |
+
"learning_rate": 1.6463157894736844e-05,
|
| 1566 |
+
"loss": 0.868,
|
| 1567 |
+
"step": 2180
|
| 1568 |
},
|
| 1569 |
{
|
| 1570 |
+
"epoch": 0.05,
|
| 1571 |
+
"grad_norm": 3.5853216648101807,
|
| 1572 |
+
"learning_rate": 1.6442105263157897e-05,
|
| 1573 |
+
"loss": 0.8388,
|
| 1574 |
+
"step": 2190
|
| 1575 |
},
|
| 1576 |
{
|
| 1577 |
+
"epoch": 0.06,
|
| 1578 |
+
"grad_norm": 2.375610589981079,
|
| 1579 |
+
"learning_rate": 1.642105263157895e-05,
|
| 1580 |
+
"loss": 0.9111,
|
| 1581 |
+
"step": 2200
|
| 1582 |
},
|
| 1583 |
{
|
| 1584 |
+
"epoch": 0.06,
|
| 1585 |
+
"grad_norm": 1.9734487533569336,
|
| 1586 |
+
"learning_rate": 1.64e-05,
|
| 1587 |
+
"loss": 0.7288,
|
| 1588 |
+
"step": 2210
|
| 1589 |
},
|
| 1590 |
{
|
| 1591 |
+
"epoch": 0.06,
|
| 1592 |
+
"grad_norm": 10.517192840576172,
|
| 1593 |
+
"learning_rate": 1.6378947368421055e-05,
|
| 1594 |
+
"loss": 0.698,
|
| 1595 |
+
"step": 2220
|
| 1596 |
},
|
| 1597 |
{
|
| 1598 |
+
"epoch": 0.06,
|
| 1599 |
+
"grad_norm": 4.183718204498291,
|
| 1600 |
+
"learning_rate": 1.6357894736842108e-05,
|
| 1601 |
+
"loss": 0.7759,
|
| 1602 |
+
"step": 2230
|
| 1603 |
},
|
| 1604 |
{
|
| 1605 |
+
"epoch": 0.06,
|
| 1606 |
+
"grad_norm": 3.9075675010681152,
|
| 1607 |
+
"learning_rate": 1.633684210526316e-05,
|
| 1608 |
+
"loss": 0.7829,
|
| 1609 |
+
"step": 2240
|
|
|
|
| 1610 |
},
|
| 1611 |
{
|
| 1612 |
+
"epoch": 0.06,
|
| 1613 |
+
"grad_norm": 5.287744998931885,
|
| 1614 |
+
"learning_rate": 1.6315789473684213e-05,
|
| 1615 |
+
"loss": 0.7057,
|
| 1616 |
+
"step": 2250
|
| 1617 |
},
|
| 1618 |
{
|
| 1619 |
+
"epoch": 0.06,
|
| 1620 |
+
"grad_norm": 4.977657318115234,
|
| 1621 |
+
"learning_rate": 1.6294736842105265e-05,
|
| 1622 |
+
"loss": 0.8346,
|
| 1623 |
+
"step": 2260
|
| 1624 |
},
|
| 1625 |
{
|
| 1626 |
+
"epoch": 0.06,
|
| 1627 |
+
"grad_norm": 7.196689128875732,
|
| 1628 |
+
"learning_rate": 1.6273684210526318e-05,
|
| 1629 |
+
"loss": 0.8508,
|
| 1630 |
+
"step": 2270
|
| 1631 |
},
|
| 1632 |
{
|
| 1633 |
+
"epoch": 0.06,
|
| 1634 |
+
"grad_norm": 2.467477798461914,
|
| 1635 |
+
"learning_rate": 1.6252631578947367e-05,
|
| 1636 |
+
"loss": 0.7179,
|
| 1637 |
+
"step": 2280
|
| 1638 |
},
|
| 1639 |
{
|
| 1640 |
+
"epoch": 0.06,
|
| 1641 |
+
"grad_norm": 7.059762954711914,
|
| 1642 |
+
"learning_rate": 1.6231578947368423e-05,
|
| 1643 |
+
"loss": 0.7549,
|
| 1644 |
+
"step": 2290
|
| 1645 |
},
|
| 1646 |
{
|
| 1647 |
+
"epoch": 0.06,
|
| 1648 |
+
"grad_norm": 3.980865955352783,
|
| 1649 |
+
"learning_rate": 1.6210526315789473e-05,
|
| 1650 |
+
"loss": 0.814,
|
| 1651 |
+
"step": 2300
|
| 1652 |
},
|
| 1653 |
{
|
| 1654 |
+
"epoch": 0.06,
|
| 1655 |
+
"grad_norm": 7.675939559936523,
|
| 1656 |
+
"learning_rate": 1.618947368421053e-05,
|
| 1657 |
+
"loss": 0.8227,
|
| 1658 |
+
"step": 2310
|
| 1659 |
},
|
| 1660 |
{
|
| 1661 |
+
"epoch": 0.06,
|
| 1662 |
+
"grad_norm": 3.530073642730713,
|
| 1663 |
+
"learning_rate": 1.616842105263158e-05,
|
| 1664 |
+
"loss": 0.8517,
|
| 1665 |
+
"step": 2320
|
| 1666 |
},
|
| 1667 |
{
|
| 1668 |
+
"epoch": 0.06,
|
| 1669 |
+
"grad_norm": 3.6851344108581543,
|
| 1670 |
+
"learning_rate": 1.6147368421052634e-05,
|
| 1671 |
+
"loss": 0.7684,
|
| 1672 |
+
"step": 2330
|
| 1673 |
},
|
| 1674 |
{
|
| 1675 |
+
"epoch": 0.06,
|
| 1676 |
+
"grad_norm": 5.206923961639404,
|
| 1677 |
+
"learning_rate": 1.6126315789473687e-05,
|
| 1678 |
+
"loss": 0.8199,
|
| 1679 |
+
"step": 2340
|
| 1680 |
},
|
| 1681 |
{
|
| 1682 |
+
"epoch": 0.06,
|
| 1683 |
+
"grad_norm": 5.220828056335449,
|
| 1684 |
+
"learning_rate": 1.6105263157894736e-05,
|
| 1685 |
+
"loss": 0.8871,
|
| 1686 |
+
"step": 2350
|
| 1687 |
},
|
| 1688 |
{
|
| 1689 |
+
"epoch": 0.06,
|
| 1690 |
+
"grad_norm": 3.5062482357025146,
|
| 1691 |
+
"learning_rate": 1.6084210526315792e-05,
|
| 1692 |
+
"loss": 0.8281,
|
| 1693 |
+
"step": 2360
|
| 1694 |
},
|
| 1695 |
{
|
| 1696 |
+
"epoch": 0.06,
|
| 1697 |
+
"grad_norm": 1.9830796718597412,
|
| 1698 |
+
"learning_rate": 1.606315789473684e-05,
|
| 1699 |
+
"loss": 0.8678,
|
| 1700 |
+
"step": 2370
|
| 1701 |
},
|
| 1702 |
{
|
| 1703 |
+
"epoch": 0.06,
|
| 1704 |
+
"grad_norm": 3.3255491256713867,
|
| 1705 |
+
"learning_rate": 1.6042105263157897e-05,
|
| 1706 |
+
"loss": 0.8337,
|
| 1707 |
+
"step": 2380
|
| 1708 |
},
|
| 1709 |
{
|
| 1710 |
+
"epoch": 0.06,
|
| 1711 |
+
"grad_norm": 5.259572505950928,
|
| 1712 |
+
"learning_rate": 1.6021052631578947e-05,
|
| 1713 |
+
"loss": 0.7954,
|
| 1714 |
+
"step": 2390
|
| 1715 |
},
|
| 1716 |
{
|
| 1717 |
+
"epoch": 0.06,
|
| 1718 |
+
"grad_norm": 3.6201376914978027,
|
| 1719 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 1720 |
+
"loss": 0.818,
|
| 1721 |
+
"step": 2400
|
| 1722 |
},
|
| 1723 |
{
|
| 1724 |
+
"epoch": 0.06,
|
| 1725 |
+
"grad_norm": 3.3598544597625732,
|
| 1726 |
+
"learning_rate": 1.5978947368421055e-05,
|
| 1727 |
+
"loss": 0.7697,
|
| 1728 |
+
"step": 2410
|
| 1729 |
},
|
| 1730 |
{
|
| 1731 |
+
"epoch": 0.06,
|
| 1732 |
+
"grad_norm": 6.34808349609375,
|
| 1733 |
+
"learning_rate": 1.5957894736842105e-05,
|
| 1734 |
+
"loss": 0.6347,
|
| 1735 |
+
"step": 2420
|
| 1736 |
},
|
| 1737 |
{
|
| 1738 |
+
"epoch": 0.06,
|
| 1739 |
+
"grad_norm": 3.967682361602783,
|
| 1740 |
+
"learning_rate": 1.593684210526316e-05,
|
| 1741 |
+
"loss": 0.7178,
|
| 1742 |
+
"step": 2430
|
| 1743 |
},
|
| 1744 |
{
|
| 1745 |
+
"epoch": 0.06,
|
| 1746 |
+
"grad_norm": 10.222978591918945,
|
| 1747 |
+
"learning_rate": 1.591578947368421e-05,
|
| 1748 |
+
"loss": 0.7642,
|
| 1749 |
+
"step": 2440
|
| 1750 |
},
|
| 1751 |
{
|
| 1752 |
+
"epoch": 0.06,
|
| 1753 |
+
"grad_norm": 3.9339826107025146,
|
| 1754 |
+
"learning_rate": 1.5894736842105266e-05,
|
| 1755 |
+
"loss": 0.8197,
|
| 1756 |
+
"step": 2450
|
| 1757 |
},
|
| 1758 |
{
|
| 1759 |
+
"epoch": 0.06,
|
| 1760 |
+
"grad_norm": 2.3337771892547607,
|
| 1761 |
+
"learning_rate": 1.5873684210526315e-05,
|
| 1762 |
+
"loss": 0.9375,
|
| 1763 |
+
"step": 2460
|
| 1764 |
},
|
| 1765 |
{
|
| 1766 |
+
"epoch": 0.06,
|
| 1767 |
+
"grad_norm": 2.8479838371276855,
|
| 1768 |
+
"learning_rate": 1.585263157894737e-05,
|
| 1769 |
+
"loss": 0.9196,
|
| 1770 |
+
"step": 2470
|
| 1771 |
},
|
| 1772 |
{
|
| 1773 |
+
"epoch": 0.06,
|
| 1774 |
+
"grad_norm": 9.294541358947754,
|
| 1775 |
+
"learning_rate": 1.5831578947368424e-05,
|
| 1776 |
+
"loss": 0.7144,
|
| 1777 |
+
"step": 2480
|
| 1778 |
},
|
| 1779 |
{
|
| 1780 |
+
"epoch": 0.06,
|
| 1781 |
+
"grad_norm": 5.325323104858398,
|
| 1782 |
+
"learning_rate": 1.5810526315789473e-05,
|
| 1783 |
+
"loss": 0.7897,
|
| 1784 |
+
"step": 2490
|
| 1785 |
},
|
| 1786 |
{
|
| 1787 |
+
"epoch": 0.06,
|
| 1788 |
+
"grad_norm": 4.377369403839111,
|
| 1789 |
+
"learning_rate": 1.578947368421053e-05,
|
| 1790 |
+
"loss": 0.9008,
|
| 1791 |
+
"step": 2500
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 0.06,
|
| 1795 |
+
"eval_loss": 0.8163847923278809,
|
| 1796 |
+
"eval_runtime": 67.7994,
|
| 1797 |
+
"eval_samples_per_second": 14.749,
|
| 1798 |
+
"eval_steps_per_second": 14.749,
|
| 1799 |
+
"step": 2500
|
| 1800 |
}
|
| 1801 |
],
|
| 1802 |
+
"logging_steps": 10,
|
| 1803 |
+
"max_steps": 10000,
|
| 1804 |
"num_input_tokens_seen": 0,
|
| 1805 |
"num_train_epochs": 1,
|
| 1806 |
"save_steps": 2500,
|
| 1807 |
+
"total_flos": 4.025531498496e+16,
|
| 1808 |
"train_batch_size": 1,
|
| 1809 |
"trial_name": null,
|
| 1810 |
"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 sha256:51ac7424107c168679594d767b2ffefa42eac9e349caa7916abcb7990d9f453e
|
| 3 |
+
size 4920
|