Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +353 -3
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -6307,6 +6307,356 @@
|
|
| 6307 |
"learning_rate": 2.7091379149682685e-06,
|
| 6308 |
"loss": 0.0062,
|
| 6309 |
"step": 9000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6310 |
}
|
| 6311 |
],
|
| 6312 |
"logging_steps": 10,
|
|
@@ -6326,7 +6676,7 @@
|
|
| 6326 |
"attributes": {}
|
| 6327 |
}
|
| 6328 |
},
|
| 6329 |
-
"total_flos": 3.
|
| 6330 |
"train_batch_size": 16,
|
| 6331 |
"trial_name": null,
|
| 6332 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 19.66873706004141,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 9500,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 6307 |
"learning_rate": 2.7091379149682685e-06,
|
| 6308 |
"loss": 0.0062,
|
| 6309 |
"step": 9000
|
| 6310 |
+
},
|
| 6311 |
+
{
|
| 6312 |
+
"epoch": 18.65424430641822,
|
| 6313 |
+
"grad_norm": 0.1678360253572464,
|
| 6314 |
+
"learning_rate": 2.6557085182532582e-06,
|
| 6315 |
+
"loss": 0.0092,
|
| 6316 |
+
"step": 9010
|
| 6317 |
+
},
|
| 6318 |
+
{
|
| 6319 |
+
"epoch": 18.67494824016563,
|
| 6320 |
+
"grad_norm": 0.15553714334964752,
|
| 6321 |
+
"learning_rate": 2.602796871124663e-06,
|
| 6322 |
+
"loss": 0.0116,
|
| 6323 |
+
"step": 9020
|
| 6324 |
+
},
|
| 6325 |
+
{
|
| 6326 |
+
"epoch": 18.695652173913043,
|
| 6327 |
+
"grad_norm": 0.12781013548374176,
|
| 6328 |
+
"learning_rate": 2.5504035522157854e-06,
|
| 6329 |
+
"loss": 0.0069,
|
| 6330 |
+
"step": 9030
|
| 6331 |
+
},
|
| 6332 |
+
{
|
| 6333 |
+
"epoch": 18.716356107660456,
|
| 6334 |
+
"grad_norm": 0.10278218239545822,
|
| 6335 |
+
"learning_rate": 2.4985291344915674e-06,
|
| 6336 |
+
"loss": 0.0068,
|
| 6337 |
+
"step": 9040
|
| 6338 |
+
},
|
| 6339 |
+
{
|
| 6340 |
+
"epoch": 18.73706004140787,
|
| 6341 |
+
"grad_norm": 0.31690600514411926,
|
| 6342 |
+
"learning_rate": 2.4471741852423237e-06,
|
| 6343 |
+
"loss": 0.0093,
|
| 6344 |
+
"step": 9050
|
| 6345 |
+
},
|
| 6346 |
+
{
|
| 6347 |
+
"epoch": 18.757763975155278,
|
| 6348 |
+
"grad_norm": 0.27544263005256653,
|
| 6349 |
+
"learning_rate": 2.3963392660775575e-06,
|
| 6350 |
+
"loss": 0.0075,
|
| 6351 |
+
"step": 9060
|
| 6352 |
+
},
|
| 6353 |
+
{
|
| 6354 |
+
"epoch": 18.77846790890269,
|
| 6355 |
+
"grad_norm": 0.26249682903289795,
|
| 6356 |
+
"learning_rate": 2.3460249329197824e-06,
|
| 6357 |
+
"loss": 0.0076,
|
| 6358 |
+
"step": 9070
|
| 6359 |
+
},
|
| 6360 |
+
{
|
| 6361 |
+
"epoch": 18.799171842650104,
|
| 6362 |
+
"grad_norm": 0.09675180912017822,
|
| 6363 |
+
"learning_rate": 2.296231735998511e-06,
|
| 6364 |
+
"loss": 0.0059,
|
| 6365 |
+
"step": 9080
|
| 6366 |
+
},
|
| 6367 |
+
{
|
| 6368 |
+
"epoch": 18.819875776397517,
|
| 6369 |
+
"grad_norm": 0.08136120438575745,
|
| 6370 |
+
"learning_rate": 2.2469602198441573e-06,
|
| 6371 |
+
"loss": 0.0049,
|
| 6372 |
+
"step": 9090
|
| 6373 |
+
},
|
| 6374 |
+
{
|
| 6375 |
+
"epoch": 18.840579710144926,
|
| 6376 |
+
"grad_norm": 0.17611472308635712,
|
| 6377 |
+
"learning_rate": 2.1982109232821178e-06,
|
| 6378 |
+
"loss": 0.0063,
|
| 6379 |
+
"step": 9100
|
| 6380 |
+
},
|
| 6381 |
+
{
|
| 6382 |
+
"epoch": 18.86128364389234,
|
| 6383 |
+
"grad_norm": 0.09379903227090836,
|
| 6384 |
+
"learning_rate": 2.149984379426906e-06,
|
| 6385 |
+
"loss": 0.0064,
|
| 6386 |
+
"step": 9110
|
| 6387 |
+
},
|
| 6388 |
+
{
|
| 6389 |
+
"epoch": 18.88198757763975,
|
| 6390 |
+
"grad_norm": 0.05829261615872383,
|
| 6391 |
+
"learning_rate": 2.102281115676258e-06,
|
| 6392 |
+
"loss": 0.0042,
|
| 6393 |
+
"step": 9120
|
| 6394 |
+
},
|
| 6395 |
+
{
|
| 6396 |
+
"epoch": 18.902691511387165,
|
| 6397 |
+
"grad_norm": 0.05887362360954285,
|
| 6398 |
+
"learning_rate": 2.0551016537054493e-06,
|
| 6399 |
+
"loss": 0.0091,
|
| 6400 |
+
"step": 9130
|
| 6401 |
+
},
|
| 6402 |
+
{
|
| 6403 |
+
"epoch": 18.923395445134574,
|
| 6404 |
+
"grad_norm": 0.08341336995363235,
|
| 6405 |
+
"learning_rate": 2.008446509461498e-06,
|
| 6406 |
+
"loss": 0.0075,
|
| 6407 |
+
"step": 9140
|
| 6408 |
+
},
|
| 6409 |
+
{
|
| 6410 |
+
"epoch": 18.944099378881987,
|
| 6411 |
+
"grad_norm": 0.13443836569786072,
|
| 6412 |
+
"learning_rate": 1.962316193157593e-06,
|
| 6413 |
+
"loss": 0.0051,
|
| 6414 |
+
"step": 9150
|
| 6415 |
+
},
|
| 6416 |
+
{
|
| 6417 |
+
"epoch": 18.9648033126294,
|
| 6418 |
+
"grad_norm": 0.16308577358722687,
|
| 6419 |
+
"learning_rate": 1.91671120926748e-06,
|
| 6420 |
+
"loss": 0.0084,
|
| 6421 |
+
"step": 9160
|
| 6422 |
+
},
|
| 6423 |
+
{
|
| 6424 |
+
"epoch": 18.985507246376812,
|
| 6425 |
+
"grad_norm": 0.080412857234478,
|
| 6426 |
+
"learning_rate": 1.8716320565199618e-06,
|
| 6427 |
+
"loss": 0.0094,
|
| 6428 |
+
"step": 9170
|
| 6429 |
+
},
|
| 6430 |
+
{
|
| 6431 |
+
"epoch": 19.006211180124225,
|
| 6432 |
+
"grad_norm": 0.29631340503692627,
|
| 6433 |
+
"learning_rate": 1.8270792278934302e-06,
|
| 6434 |
+
"loss": 0.0051,
|
| 6435 |
+
"step": 9180
|
| 6436 |
+
},
|
| 6437 |
+
{
|
| 6438 |
+
"epoch": 19.026915113871635,
|
| 6439 |
+
"grad_norm": 0.15507416427135468,
|
| 6440 |
+
"learning_rate": 1.7830532106104747e-06,
|
| 6441 |
+
"loss": 0.0062,
|
| 6442 |
+
"step": 9190
|
| 6443 |
+
},
|
| 6444 |
+
{
|
| 6445 |
+
"epoch": 19.047619047619047,
|
| 6446 |
+
"grad_norm": 0.23660752177238464,
|
| 6447 |
+
"learning_rate": 1.7395544861325718e-06,
|
| 6448 |
+
"loss": 0.0073,
|
| 6449 |
+
"step": 9200
|
| 6450 |
+
},
|
| 6451 |
+
{
|
| 6452 |
+
"epoch": 19.06832298136646,
|
| 6453 |
+
"grad_norm": 0.2019672840833664,
|
| 6454 |
+
"learning_rate": 1.696583530154794e-06,
|
| 6455 |
+
"loss": 0.0115,
|
| 6456 |
+
"step": 9210
|
| 6457 |
+
},
|
| 6458 |
+
{
|
| 6459 |
+
"epoch": 19.089026915113873,
|
| 6460 |
+
"grad_norm": 0.06961517781019211,
|
| 6461 |
+
"learning_rate": 1.6541408126006463e-06,
|
| 6462 |
+
"loss": 0.0061,
|
| 6463 |
+
"step": 9220
|
| 6464 |
+
},
|
| 6465 |
+
{
|
| 6466 |
+
"epoch": 19.109730848861282,
|
| 6467 |
+
"grad_norm": 0.2266739010810852,
|
| 6468 |
+
"learning_rate": 1.6122267976168781e-06,
|
| 6469 |
+
"loss": 0.0104,
|
| 6470 |
+
"step": 9230
|
| 6471 |
+
},
|
| 6472 |
+
{
|
| 6473 |
+
"epoch": 19.130434782608695,
|
| 6474 |
+
"grad_norm": 0.10511913150548935,
|
| 6475 |
+
"learning_rate": 1.5708419435684462e-06,
|
| 6476 |
+
"loss": 0.0094,
|
| 6477 |
+
"step": 9240
|
| 6478 |
+
},
|
| 6479 |
+
{
|
| 6480 |
+
"epoch": 19.151138716356108,
|
| 6481 |
+
"grad_norm": 0.1996021419763565,
|
| 6482 |
+
"learning_rate": 1.5299867030334814e-06,
|
| 6483 |
+
"loss": 0.0093,
|
| 6484 |
+
"step": 9250
|
| 6485 |
+
},
|
| 6486 |
+
{
|
| 6487 |
+
"epoch": 19.17184265010352,
|
| 6488 |
+
"grad_norm": 0.12162107229232788,
|
| 6489 |
+
"learning_rate": 1.4896615227983468e-06,
|
| 6490 |
+
"loss": 0.0044,
|
| 6491 |
+
"step": 9260
|
| 6492 |
+
},
|
| 6493 |
+
{
|
| 6494 |
+
"epoch": 19.19254658385093,
|
| 6495 |
+
"grad_norm": 0.09749267995357513,
|
| 6496 |
+
"learning_rate": 1.4498668438527597e-06,
|
| 6497 |
+
"loss": 0.0064,
|
| 6498 |
+
"step": 9270
|
| 6499 |
+
},
|
| 6500 |
+
{
|
| 6501 |
+
"epoch": 19.213250517598343,
|
| 6502 |
+
"grad_norm": 0.26853621006011963,
|
| 6503 |
+
"learning_rate": 1.4106031013849496e-06,
|
| 6504 |
+
"loss": 0.0141,
|
| 6505 |
+
"step": 9280
|
| 6506 |
+
},
|
| 6507 |
+
{
|
| 6508 |
+
"epoch": 19.233954451345756,
|
| 6509 |
+
"grad_norm": 0.08542772382497787,
|
| 6510 |
+
"learning_rate": 1.3718707247769135e-06,
|
| 6511 |
+
"loss": 0.0049,
|
| 6512 |
+
"step": 9290
|
| 6513 |
+
},
|
| 6514 |
+
{
|
| 6515 |
+
"epoch": 19.25465838509317,
|
| 6516 |
+
"grad_norm": 0.4365129768848419,
|
| 6517 |
+
"learning_rate": 1.333670137599713e-06,
|
| 6518 |
+
"loss": 0.01,
|
| 6519 |
+
"step": 9300
|
| 6520 |
+
},
|
| 6521 |
+
{
|
| 6522 |
+
"epoch": 19.27536231884058,
|
| 6523 |
+
"grad_norm": 0.3198767602443695,
|
| 6524 |
+
"learning_rate": 1.2960017576088446e-06,
|
| 6525 |
+
"loss": 0.0104,
|
| 6526 |
+
"step": 9310
|
| 6527 |
+
},
|
| 6528 |
+
{
|
| 6529 |
+
"epoch": 19.29606625258799,
|
| 6530 |
+
"grad_norm": 0.2250336855649948,
|
| 6531 |
+
"learning_rate": 1.2588659967397e-06,
|
| 6532 |
+
"loss": 0.0175,
|
| 6533 |
+
"step": 9320
|
| 6534 |
+
},
|
| 6535 |
+
{
|
| 6536 |
+
"epoch": 19.316770186335404,
|
| 6537 |
+
"grad_norm": 0.22645168006420135,
|
| 6538 |
+
"learning_rate": 1.222263261102985e-06,
|
| 6539 |
+
"loss": 0.0079,
|
| 6540 |
+
"step": 9330
|
| 6541 |
+
},
|
| 6542 |
+
{
|
| 6543 |
+
"epoch": 19.337474120082817,
|
| 6544 |
+
"grad_norm": 0.18786205351352692,
|
| 6545 |
+
"learning_rate": 1.1861939509803687e-06,
|
| 6546 |
+
"loss": 0.0074,
|
| 6547 |
+
"step": 9340
|
| 6548 |
+
},
|
| 6549 |
+
{
|
| 6550 |
+
"epoch": 19.358178053830226,
|
| 6551 |
+
"grad_norm": 0.1437540352344513,
|
| 6552 |
+
"learning_rate": 1.1506584608200367e-06,
|
| 6553 |
+
"loss": 0.0044,
|
| 6554 |
+
"step": 9350
|
| 6555 |
+
},
|
| 6556 |
+
{
|
| 6557 |
+
"epoch": 19.37888198757764,
|
| 6558 |
+
"grad_norm": 0.13397444784641266,
|
| 6559 |
+
"learning_rate": 1.1156571792324211e-06,
|
| 6560 |
+
"loss": 0.0055,
|
| 6561 |
+
"step": 9360
|
| 6562 |
+
},
|
| 6563 |
+
{
|
| 6564 |
+
"epoch": 19.399585921325052,
|
| 6565 |
+
"grad_norm": 0.1180369108915329,
|
| 6566 |
+
"learning_rate": 1.0811904889859336e-06,
|
| 6567 |
+
"loss": 0.0147,
|
| 6568 |
+
"step": 9370
|
| 6569 |
+
},
|
| 6570 |
+
{
|
| 6571 |
+
"epoch": 19.420289855072465,
|
| 6572 |
+
"grad_norm": 0.13278833031654358,
|
| 6573 |
+
"learning_rate": 1.0472587670027678e-06,
|
| 6574 |
+
"loss": 0.0038,
|
| 6575 |
+
"step": 9380
|
| 6576 |
+
},
|
| 6577 |
+
{
|
| 6578 |
+
"epoch": 19.440993788819874,
|
| 6579 |
+
"grad_norm": 0.2261534333229065,
|
| 6580 |
+
"learning_rate": 1.0138623843548078e-06,
|
| 6581 |
+
"loss": 0.0145,
|
| 6582 |
+
"step": 9390
|
| 6583 |
+
},
|
| 6584 |
+
{
|
| 6585 |
+
"epoch": 19.461697722567287,
|
| 6586 |
+
"grad_norm": 0.16062599420547485,
|
| 6587 |
+
"learning_rate": 9.810017062595322e-07,
|
| 6588 |
+
"loss": 0.0091,
|
| 6589 |
+
"step": 9400
|
| 6590 |
+
},
|
| 6591 |
+
{
|
| 6592 |
+
"epoch": 19.4824016563147,
|
| 6593 |
+
"grad_norm": 0.09465906769037247,
|
| 6594 |
+
"learning_rate": 9.486770920760668e-07,
|
| 6595 |
+
"loss": 0.0079,
|
| 6596 |
+
"step": 9410
|
| 6597 |
+
},
|
| 6598 |
+
{
|
| 6599 |
+
"epoch": 19.503105590062113,
|
| 6600 |
+
"grad_norm": 0.06441060453653336,
|
| 6601 |
+
"learning_rate": 9.168888953011989e-07,
|
| 6602 |
+
"loss": 0.0056,
|
| 6603 |
+
"step": 9420
|
| 6604 |
+
},
|
| 6605 |
+
{
|
| 6606 |
+
"epoch": 19.523809523809526,
|
| 6607 |
+
"grad_norm": 0.3446926176548004,
|
| 6608 |
+
"learning_rate": 8.856374635655695e-07,
|
| 6609 |
+
"loss": 0.0066,
|
| 6610 |
+
"step": 9430
|
| 6611 |
+
},
|
| 6612 |
+
{
|
| 6613 |
+
"epoch": 19.544513457556935,
|
| 6614 |
+
"grad_norm": 0.264846533536911,
|
| 6615 |
+
"learning_rate": 8.549231386298151e-07,
|
| 6616 |
+
"loss": 0.008,
|
| 6617 |
+
"step": 9440
|
| 6618 |
+
},
|
| 6619 |
+
{
|
| 6620 |
+
"epoch": 19.565217391304348,
|
| 6621 |
+
"grad_norm": 0.23693622648715973,
|
| 6622 |
+
"learning_rate": 8.247462563808817e-07,
|
| 6623 |
+
"loss": 0.0085,
|
| 6624 |
+
"step": 9450
|
| 6625 |
+
},
|
| 6626 |
+
{
|
| 6627 |
+
"epoch": 19.58592132505176,
|
| 6628 |
+
"grad_norm": 0.1094195693731308,
|
| 6629 |
+
"learning_rate": 7.951071468283167e-07,
|
| 6630 |
+
"loss": 0.0042,
|
| 6631 |
+
"step": 9460
|
| 6632 |
+
},
|
| 6633 |
+
{
|
| 6634 |
+
"epoch": 19.606625258799173,
|
| 6635 |
+
"grad_norm": 0.058433957397937775,
|
| 6636 |
+
"learning_rate": 7.66006134100672e-07,
|
| 6637 |
+
"loss": 0.0063,
|
| 6638 |
+
"step": 9470
|
| 6639 |
+
},
|
| 6640 |
+
{
|
| 6641 |
+
"epoch": 19.627329192546583,
|
| 6642 |
+
"grad_norm": 0.28139904141426086,
|
| 6643 |
+
"learning_rate": 7.374435364419674e-07,
|
| 6644 |
+
"loss": 0.0064,
|
| 6645 |
+
"step": 9480
|
| 6646 |
+
},
|
| 6647 |
+
{
|
| 6648 |
+
"epoch": 19.648033126293996,
|
| 6649 |
+
"grad_norm": 0.13973474502563477,
|
| 6650 |
+
"learning_rate": 7.094196662081831e-07,
|
| 6651 |
+
"loss": 0.0054,
|
| 6652 |
+
"step": 9490
|
| 6653 |
+
},
|
| 6654 |
+
{
|
| 6655 |
+
"epoch": 19.66873706004141,
|
| 6656 |
+
"grad_norm": 0.27205267548561096,
|
| 6657 |
+
"learning_rate": 6.819348298638839e-07,
|
| 6658 |
+
"loss": 0.0077,
|
| 6659 |
+
"step": 9500
|
| 6660 |
}
|
| 6661 |
],
|
| 6662 |
"logging_steps": 10,
|
|
|
|
| 6676 |
"attributes": {}
|
| 6677 |
}
|
| 6678 |
},
|
| 6679 |
+
"total_flos": 3.4453830668706816e+17,
|
| 6680 |
"train_batch_size": 16,
|
| 6681 |
"trial_name": null,
|
| 6682 |
"trial_params": null
|