Hugging Face's logo

AmberYifan
/
Llama-2-7b-sft-dpo-10k

Text Generation
Transformers
Safetensors
llama
Generated from Trainer
trl
dpo
conversational
text-generation-inference
Inference Endpoints
Model card Files Community
AmberYifan commited on
Commit
2b0ba6e
·
verified ·
1 Parent(s): b221608

Training in progress, epoch 2, checkpoint

Browse files
Files changed (18) hide show
  1. last-checkpoint/global_step313/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
  2. last-checkpoint/global_step313/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
  3. last-checkpoint/global_step313/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
  4. last-checkpoint/global_step313/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
  5. last-checkpoint/global_step313/zero_pp_rank_0_mp_rank_00_model_states.pt +3 -0
  6. last-checkpoint/global_step313/zero_pp_rank_1_mp_rank_00_model_states.pt +3 -0
  7. last-checkpoint/global_step313/zero_pp_rank_2_mp_rank_00_model_states.pt +3 -0
  8. last-checkpoint/global_step313/zero_pp_rank_3_mp_rank_00_model_states.pt +3 -0
  9. last-checkpoint/latest +1 -1
  10. last-checkpoint/model-00001-of-00003.safetensors +1 -1
  11. last-checkpoint/model-00002-of-00003.safetensors +1 -1
  12. last-checkpoint/model-00003-of-00003.safetensors +1 -1
  13. last-checkpoint/rng_state_0.pth +1 -1
  14. last-checkpoint/rng_state_1.pth +1 -1
  15. last-checkpoint/rng_state_2.pth +1 -1
  16. last-checkpoint/rng_state_3.pth +1 -1
  17. last-checkpoint/scheduler.pt +1 -1
  18. last-checkpoint/trainer_state.json +258 -2
last-checkpoint/global_step313/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:121475720cf2bbbe48861499254612f066f3b0872293cd5d8156c27eb3505041
3
+ size 13476835648
last-checkpoint/global_step313/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6a7ca3d7c7752e1a2c58d7e1ce8c5539506733e22351fce55b0794448ca33e02
3
+ size 13476835648
last-checkpoint/global_step313/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:02ba558106df3bf28e2faffc37c045fafaecf252cf954a40879b9e9ac0b1503b
3
+ size 13476835648
last-checkpoint/global_step313/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b2e62b939842439c944dcb24aece333ee0d8beded32c31f4cf8adeef092e42c5
3
+ size 13476835648
last-checkpoint/global_step313/zero_pp_rank_0_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:87b448f691cfa4a46495dc05e49e01cff1797eee23ebb09e55632c64c52dc224
3
+ size 150693
last-checkpoint/global_step313/zero_pp_rank_1_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:211e1ef93c6886703f99a126064c19878398b27bfd7d6fe55c0993b88fd20b06
3
+ size 150693
last-checkpoint/global_step313/zero_pp_rank_2_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f5a18f8c5c0e35b7a56b01112f6b3a949d7c33eb71375ee3d414c563c689d607
3
+ size 150693
last-checkpoint/global_step313/zero_pp_rank_3_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:42b9a66ad55de6e97e1ca4bcd8022a87634ef75bde4ced7652fa266ac694f25f
3
+ size 150693
last-checkpoint/latest CHANGED
@@ -1 +1 @@
1
- global_step156
 
1
+ global_step313
last-checkpoint/model-00001-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:cbca58b53cd47541d268f82399ffad5cbf319009372de99641a382b4b0bc5da1
3
  size 4938985352
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c5607479e21e60cbfd6ba75dd86857fbf07c406a070c0661031268bd8e07512f
3
  size 4938985352
last-checkpoint/model-00002-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:576e598617877202a81086b8a32ca9d9a8aee6a5b0b21768edf1cfc6ffb95009
3
  size 4947390880
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:87a8a69aad0f1f95dc1e092863ef0fd7033ce518b3f7f469966d2657fc2dcce0
3
  size 4947390880
last-checkpoint/model-00003-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:dafc82f3f9a5385f386a865069c04310d4a433477684ceb4da429097f35328f2
3
  size 3590488816
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f7f0dbf4f652fd9181d26c5c8295c27ef559faabc793e597c12982c4491b1fd
3
  size 3590488816
last-checkpoint/rng_state_0.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a4a1dc8b9f546123ce9b44c28828ae71d9d0ecd6f59c6b7b8337c73cb2e24c49
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:50ca27603f7c4a4d9a980223e1a102922ac31561acfa7da4f28042e4e8b3528b
3
  size 15024
last-checkpoint/rng_state_1.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0a8b225b0323772b66570a16b5c7e6446dc3c9ee076597906a2c90e3c0a383bf
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c57f658c3d9b6de611da50021823095fc4067036fb9a4b34655cc2e34fbf5c2b
3
  size 15024
last-checkpoint/rng_state_2.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c036a8d528905f7018266618c8bebe4add10c2778ef432fd0116369fa622d815
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:109da2c363a1e09432fa2ea2b7dceb6f67d1f91975069fc669e71dd536d654d8
3
  size 15024
last-checkpoint/rng_state_3.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:b93d88429d97654e76d10b1ba4c36f29a47e210c65a03efdc136f04af8282797
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bda65e490958bfe3fe01157ed131eb2354f8fc19536842d27c998ecf29d48991
3
  size 15024
last-checkpoint/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:83d5ef0cb3d7d5c4ee48b323753e4be796504f2feec3727a9d0af4a98df5f901
3
  size 1064
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:afa7143d1f3515855bc2dc683c033423673c7ec610b3bb70a98e9e265d312b47
3
  size 1064
last-checkpoint/trainer_state.json CHANGED
@@ -1,9 +1,9 @@
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
- "epoch": 0.9968051118210862,
5
  "eval_steps": 500,
6
- "global_step": 156,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
@@ -263,6 +263,262 @@
263
  "eval_samples_per_second": 17.459,
264
  "eval_steps_per_second": 0.611,
265
  "step": 156
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
266
  }
267
  ],
268
  "logging_steps": 10,
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 2.0,
5
  "eval_steps": 500,
6
+ "global_step": 313,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
 
263
  "eval_samples_per_second": 17.459,
264
  "eval_steps_per_second": 0.611,
265
  "step": 156
266
+ },
267
+ {
268
+ "epoch": 1.0223642172523961,
269
+ "grad_norm": 9.22711379768867,
270
+ "learning_rate": 3.6579572446555817e-07,
271
+ "logits/chosen": -1.4921875,
272
+ "logits/rejected": -1.46875,
273
+ "logps/chosen": -141.0,
274
+ "logps/rejected": -139.0,
275
+ "loss": 0.5811,
276
+ "rewards/accuracies": 0.675000011920929,
277
+ "rewards/chosen": -0.8828125,
278
+ "rewards/margins": 0.34765625,
279
+ "rewards/rejected": -1.2265625,
280
+ "step": 160
281
+ },
282
+ {
283
+ "epoch": 1.0862619808306708,
284
+ "grad_norm": 9.508478185274804,
285
+ "learning_rate": 3.5391923990498813e-07,
286
+ "logits/chosen": -1.5078125,
287
+ "logits/rejected": -1.5,
288
+ "logps/chosen": -145.0,
289
+ "logps/rejected": -134.0,
290
+ "loss": 0.5337,
291
+ "rewards/accuracies": 0.78125,
292
+ "rewards/chosen": -0.71875,
293
+ "rewards/margins": 0.5546875,
294
+ "rewards/rejected": -1.2734375,
295
+ "step": 170
296
+ },
297
+ {
298
+ "epoch": 1.1501597444089458,
299
+ "grad_norm": 9.316918076373987,
300
+ "learning_rate": 3.42042755344418e-07,
301
+ "logits/chosen": -1.5625,
302
+ "logits/rejected": -1.53125,
303
+ "logps/chosen": -157.0,
304
+ "logps/rejected": -146.0,
305
+ "loss": 0.544,
306
+ "rewards/accuracies": 0.7875000238418579,
307
+ "rewards/chosen": -0.66015625,
308
+ "rewards/margins": 0.56640625,
309
+ "rewards/rejected": -1.2265625,
310
+ "step": 180
311
+ },
312
+ {
313
+ "epoch": 1.2140575079872205,
314
+ "grad_norm": 8.759483351227418,
315
+ "learning_rate": 3.3016627078384796e-07,
316
+ "logits/chosen": -1.5703125,
317
+ "logits/rejected": -1.5234375,
318
+ "logps/chosen": -152.0,
319
+ "logps/rejected": -146.0,
320
+ "loss": 0.5362,
321
+ "rewards/accuracies": 0.7250000238418579,
322
+ "rewards/chosen": -0.74609375,
323
+ "rewards/margins": 0.5,
324
+ "rewards/rejected": -1.25,
325
+ "step": 190
326
+ },
327
+ {
328
+ "epoch": 1.2779552715654952,
329
+ "grad_norm": 9.97195116624129,
330
+ "learning_rate": 3.182897862232779e-07,
331
+ "logits/chosen": -1.515625,
332
+ "logits/rejected": -1.515625,
333
+ "logps/chosen": -151.0,
334
+ "logps/rejected": -141.0,
335
+ "loss": 0.5373,
336
+ "rewards/accuracies": 0.768750011920929,
337
+ "rewards/chosen": -0.7890625,
338
+ "rewards/margins": 0.55859375,
339
+ "rewards/rejected": -1.34375,
340
+ "step": 200
341
+ },
342
+ {
343
+ "epoch": 1.34185303514377,
344
+ "grad_norm": 8.576220811968858,
345
+ "learning_rate": 3.0641330166270784e-07,
346
+ "logits/chosen": -1.578125,
347
+ "logits/rejected": -1.5234375,
348
+ "logps/chosen": -151.0,
349
+ "logps/rejected": -140.0,
350
+ "loss": 0.5339,
351
+ "rewards/accuracies": 0.6875,
352
+ "rewards/chosen": -0.83984375,
353
+ "rewards/margins": 0.462890625,
354
+ "rewards/rejected": -1.3046875,
355
+ "step": 210
356
+ },
357
+ {
358
+ "epoch": 1.4057507987220448,
359
+ "grad_norm": 8.688671845017401,
360
+ "learning_rate": 2.9453681710213776e-07,
361
+ "logits/chosen": -1.53125,
362
+ "logits/rejected": -1.53125,
363
+ "logps/chosen": -148.0,
364
+ "logps/rejected": -143.0,
365
+ "loss": 0.5366,
366
+ "rewards/accuracies": 0.731249988079071,
367
+ "rewards/chosen": -0.8671875,
368
+ "rewards/margins": 0.5078125,
369
+ "rewards/rejected": -1.375,
370
+ "step": 220
371
+ },
372
+ {
373
+ "epoch": 1.4696485623003195,
374
+ "grad_norm": 9.030983047385451,
375
+ "learning_rate": 2.8266033254156767e-07,
376
+ "logits/chosen": -1.546875,
377
+ "logits/rejected": -1.515625,
378
+ "logps/chosen": -145.0,
379
+ "logps/rejected": -142.0,
380
+ "loss": 0.5346,
381
+ "rewards/accuracies": 0.737500011920929,
382
+ "rewards/chosen": -0.8046875,
383
+ "rewards/margins": 0.53515625,
384
+ "rewards/rejected": -1.34375,
385
+ "step": 230
386
+ },
387
+ {
388
+ "epoch": 1.5335463258785942,
389
+ "grad_norm": 11.622095941624893,
390
+ "learning_rate": 2.7078384798099764e-07,
391
+ "logits/chosen": -1.484375,
392
+ "logits/rejected": -1.4609375,
393
+ "logps/chosen": -145.0,
394
+ "logps/rejected": -153.0,
395
+ "loss": 0.553,
396
+ "rewards/accuracies": 0.75,
397
+ "rewards/chosen": -0.80859375,
398
+ "rewards/margins": 0.51171875,
399
+ "rewards/rejected": -1.3203125,
400
+ "step": 240
401
+ },
402
+ {
403
+ "epoch": 1.5974440894568689,
404
+ "grad_norm": 9.38942767820535,
405
+ "learning_rate": 2.589073634204275e-07,
406
+ "logits/chosen": -1.5625,
407
+ "logits/rejected": -1.4765625,
408
+ "logps/chosen": -146.0,
409
+ "logps/rejected": -132.0,
410
+ "loss": 0.54,
411
+ "rewards/accuracies": 0.78125,
412
+ "rewards/chosen": -0.50390625,
413
+ "rewards/margins": 0.578125,
414
+ "rewards/rejected": -1.078125,
415
+ "step": 250
416
+ },
417
+ {
418
+ "epoch": 1.6613418530351438,
419
+ "grad_norm": 9.87005880531835,
420
+ "learning_rate": 2.4703087885985747e-07,
421
+ "logits/chosen": -1.5546875,
422
+ "logits/rejected": -1.5234375,
423
+ "logps/chosen": -160.0,
424
+ "logps/rejected": -148.0,
425
+ "loss": 0.5312,
426
+ "rewards/accuracies": 0.737500011920929,
427
+ "rewards/chosen": -0.55859375,
428
+ "rewards/margins": 0.5546875,
429
+ "rewards/rejected": -1.1171875,
430
+ "step": 260
431
+ },
432
+ {
433
+ "epoch": 1.7252396166134185,
434
+ "grad_norm": 8.251251897175216,
435
+ "learning_rate": 2.351543942992874e-07,
436
+ "logits/chosen": -1.5234375,
437
+ "logits/rejected": -1.4921875,
438
+ "logps/chosen": -146.0,
439
+ "logps/rejected": -143.0,
440
+ "loss": 0.5354,
441
+ "rewards/accuracies": 0.7875000238418579,
442
+ "rewards/chosen": -0.609375,
443
+ "rewards/margins": 0.56640625,
444
+ "rewards/rejected": -1.171875,
445
+ "step": 270
446
+ },
447
+ {
448
+ "epoch": 1.7891373801916934,
449
+ "grad_norm": 9.817987165601615,
450
+ "learning_rate": 2.2327790973871732e-07,
451
+ "logits/chosen": -1.53125,
452
+ "logits/rejected": -1.4921875,
453
+ "logps/chosen": -148.0,
454
+ "logps/rejected": -144.0,
455
+ "loss": 0.5347,
456
+ "rewards/accuracies": 0.78125,
457
+ "rewards/chosen": -0.5546875,
458
+ "rewards/margins": 0.58984375,
459
+ "rewards/rejected": -1.1484375,
460
+ "step": 280
461
+ },
462
+ {
463
+ "epoch": 1.8530351437699681,
464
+ "grad_norm": 9.610664273576093,
465
+ "learning_rate": 2.1140142517814726e-07,
466
+ "logits/chosen": -1.5390625,
467
+ "logits/rejected": -1.4921875,
468
+ "logps/chosen": -164.0,
469
+ "logps/rejected": -148.0,
470
+ "loss": 0.5186,
471
+ "rewards/accuracies": 0.762499988079071,
472
+ "rewards/chosen": -0.57421875,
473
+ "rewards/margins": 0.59375,
474
+ "rewards/rejected": -1.1640625,
475
+ "step": 290
476
+ },
477
+ {
478
+ "epoch": 1.9169329073482428,
479
+ "grad_norm": 9.80773406850179,
480
+ "learning_rate": 1.9952494061757718e-07,
481
+ "logits/chosen": -1.53125,
482
+ "logits/rejected": -1.5078125,
483
+ "logps/chosen": -151.0,
484
+ "logps/rejected": -148.0,
485
+ "loss": 0.5476,
486
+ "rewards/accuracies": 0.731249988079071,
487
+ "rewards/chosen": -0.75,
488
+ "rewards/margins": 0.439453125,
489
+ "rewards/rejected": -1.1875,
490
+ "step": 300
491
+ },
492
+ {
493
+ "epoch": 1.9808306709265175,
494
+ "grad_norm": 9.05899810741374,
495
+ "learning_rate": 1.876484560570071e-07,
496
+ "logits/chosen": -1.578125,
497
+ "logits/rejected": -1.53125,
498
+ "logps/chosen": -156.0,
499
+ "logps/rejected": -144.0,
500
+ "loss": 0.5245,
501
+ "rewards/accuracies": 0.8062499761581421,
502
+ "rewards/chosen": -0.71875,
503
+ "rewards/margins": 0.5703125,
504
+ "rewards/rejected": -1.2890625,
505
+ "step": 310
506
+ },
507
+ {
508
+ "epoch": 2.0,
509
+ "eval_logits/chosen": -1.5,
510
+ "eval_logits/rejected": -1.4921875,
511
+ "eval_logps/chosen": -144.0,
512
+ "eval_logps/rejected": -134.0,
513
+ "eval_loss": 0.5994531512260437,
514
+ "eval_rewards/accuracies": 0.6964285969734192,
515
+ "eval_rewards/chosen": -0.546875,
516
+ "eval_rewards/margins": 0.458984375,
517
+ "eval_rewards/rejected": -1.0078125,
518
+ "eval_runtime": 11.4667,
519
+ "eval_samples_per_second": 17.442,
520
+ "eval_steps_per_second": 0.61,
521
+ "step": 313
522
  }
523
  ],
524
  "logging_steps": 10,