ZeroUniqueness commited on
Commit
7f3dee6
·
1 Parent(s): 07a6394

Training in progress, step 2300

Browse files
Files changed (25) hide show
  1. adapter_model.bin +1 -1
  2. {checkpoint-1900 → checkpoint-2200/adapter_model}/README.md +0 -0
  3. {checkpoint-1900 → checkpoint-2200/adapter_model}/adapter_config.json +0 -0
  4. {checkpoint-1900 → checkpoint-2200/adapter_model}/adapter_model.bin +1 -1
  5. {checkpoint-1900/adapter_model → checkpoint-2300}/README.md +0 -0
  6. {checkpoint-1900/adapter_model → checkpoint-2300}/adapter_config.json +0 -0
  7. {checkpoint-1900/adapter_model → checkpoint-2300}/adapter_model.bin +1 -1
  8. {checkpoint-1900 → checkpoint-2300}/optimizer.pt +1 -1
  9. {checkpoint-1900 → checkpoint-2300}/rng_state_0.pth +1 -1
  10. {checkpoint-1900 → checkpoint-2300}/rng_state_1.pth +1 -1
  11. {checkpoint-1900 → checkpoint-2300}/rng_state_10.pth +1 -1
  12. {checkpoint-1900 → checkpoint-2300}/rng_state_11.pth +1 -1
  13. {checkpoint-1900 → checkpoint-2300}/rng_state_12.pth +1 -1
  14. {checkpoint-1900 → checkpoint-2300}/rng_state_13.pth +1 -1
  15. {checkpoint-1900 → checkpoint-2300}/rng_state_2.pth +1 -1
  16. {checkpoint-1900 → checkpoint-2300}/rng_state_3.pth +1 -1
  17. {checkpoint-1900 → checkpoint-2300}/rng_state_4.pth +1 -1
  18. {checkpoint-1900 → checkpoint-2300}/rng_state_5.pth +1 -1
  19. {checkpoint-1900 → checkpoint-2300}/rng_state_6.pth +1 -1
  20. {checkpoint-1900 → checkpoint-2300}/rng_state_7.pth +1 -1
  21. {checkpoint-1900 → checkpoint-2300}/rng_state_8.pth +1 -1
  22. {checkpoint-1900 → checkpoint-2300}/rng_state_9.pth +1 -1
  23. {checkpoint-1900 → checkpoint-2300}/scheduler.pt +1 -1
  24. {checkpoint-1900 → checkpoint-2300}/trainer_state.json +107 -3
  25. {checkpoint-1900 → checkpoint-2300}/training_args.bin +0 -0
adapter_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:44cbe159e15b8e9dbf98d3400e7358d725711929cef1f7d3da2170a07388e332
3
  size 500897101
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ccafc4bd14ec4e43b1042a0590cd94ede59a0a46b3af2268942333e14f009fe
3
  size 500897101
{checkpoint-1900 → checkpoint-2200/adapter_model}/README.md RENAMED
File without changes
{checkpoint-1900 → checkpoint-2200/adapter_model}/adapter_config.json RENAMED
File without changes
{checkpoint-1900 → checkpoint-2200/adapter_model}/adapter_model.bin RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6757a53516f65f459191f0812f3ff6e5c08faa4d4174d508234e365b3d269f49
3
  size 500897101
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:44cbe159e15b8e9dbf98d3400e7358d725711929cef1f7d3da2170a07388e332
3
  size 500897101
{checkpoint-1900/adapter_model → checkpoint-2300}/README.md RENAMED
File without changes
{checkpoint-1900/adapter_model → checkpoint-2300}/adapter_config.json RENAMED
File without changes
{checkpoint-1900/adapter_model → checkpoint-2300}/adapter_model.bin RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6757a53516f65f459191f0812f3ff6e5c08faa4d4174d508234e365b3d269f49
3
  size 500897101
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ccafc4bd14ec4e43b1042a0590cd94ede59a0a46b3af2268942333e14f009fe
3
  size 500897101
{checkpoint-1900 → checkpoint-2300}/optimizer.pt RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ae91b7ebe0e655065ee1464d56e5112920de0241421b74a0206d6529e22ff3dc
3
  size 1001752701
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:09875d8710c0b256d70a45b6165795297947cc066fffefa4eadde4a7e74447b5
3
  size 1001752701
{checkpoint-1900 → checkpoint-2300}/rng_state_0.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6b3a802bd9c256fbf9d98afcb9674e50a25f23655c922113781706a3f5bb8463
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d32bef7d638b8934d561d4bfad1a900fc4e2c41552b753e806e0029fc74fe58
3
  size 27772
{checkpoint-1900 → checkpoint-2300}/rng_state_1.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3e7f2239d042a55cda85c2e7758244e8792f7beb8e5ba981233ffdb2aa732b79
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:91d7e27372a96cbe26696b4d82834096fe258ad20f726deb5cafafff231c413a
3
  size 27772
{checkpoint-1900 → checkpoint-2300}/rng_state_10.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8330d83d575223fac963ca6a4707ecc9e4b1ef58ef555104235ea89a727c5d4b
3
  size 27789
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d9806ad8996ad73655276c702e9bf582d2f26f57c2055304271731c76fc61448
3
  size 27789
{checkpoint-1900 → checkpoint-2300}/rng_state_11.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:903b008d79f769b82405ea8d8b45fb950081d6aa74f3428b09f1fb82ea232fe1
3
  size 27789
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2304ae934a1ca1b566d82fc3b91a900f5a0e57004ba7c3747d3680c9bf7994a2
3
  size 27789
{checkpoint-1900 → checkpoint-2300}/rng_state_12.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3d1edee8b7340ce996421a9fd149e4e6a8942787541d062323516ba9d30ecf8f
3
  size 27789
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:095ef2621d7ad17b6d8b81bd18ce2b6fe2c178312e1e8ba1a86de571c45e2de5
3
  size 27789
{checkpoint-1900 → checkpoint-2300}/rng_state_13.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6283f9a9370c88d6837c7316107dce92668515e916d1179ef4f1efd41c913dd9
3
  size 27789
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dc9e9be373c3ebb63188ca728493103bb092268d7b6fcf92e0a89b09894cd715
3
  size 27789
{checkpoint-1900 → checkpoint-2300}/rng_state_2.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:98bf87f7dcfe1d127cc320729b4080ec2e82e2f88d4f7de32bc8a715c87972fe
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4c239354dbdd86d563374469538b584871ae2a28594f5ff63b8abde299fdbf2b
3
  size 27772
{checkpoint-1900 → checkpoint-2300}/rng_state_3.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:45b2dcd2f6316e5db48f888bdae509222dc7d5273099fc663bd5e3c165c9fa02
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:214c1bd964accc3329ceb0082946ac028de6fff1d26af23610d5e72192f25e06
3
  size 27772
{checkpoint-1900 → checkpoint-2300}/rng_state_4.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ee8e8118850847e086e829686ae4cf257b7a2898bdae2a6ba26534f9c85bcb7c
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a51ae82338efebd3f9959ec89be63f51f4fd7447d7c81b0d5ced4b15d1471971
3
  size 27772
{checkpoint-1900 → checkpoint-2300}/rng_state_5.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:059ec6310b63d0c409817144f7b2d080119ab00b3aaf86cbdfa4f7ce3c8c741a
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4e425296388d7987a837bfc17bf8521aaffafc95bfdaa4ed71a252a8ee6e4785
3
  size 27772
{checkpoint-1900 → checkpoint-2300}/rng_state_6.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a8b9d06c4f388f3ecd8c2980dfdd9a7453ef1a70f0475fe02cd5f65f86b165b7
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a4c0480d1aa56b841109174fd836fb9e139546505439291aa5228bd8284cd42d
3
  size 27772
{checkpoint-1900 → checkpoint-2300}/rng_state_7.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5a8a03ee273ba6f2f6c2df0641e330c27bb2b7bc063f6543e9217febb5a3e320
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f9cf82ee205afbfaca239c8f3606beac9926d4f89934cfba78fc70c7965c04c
3
  size 27772
{checkpoint-1900 → checkpoint-2300}/rng_state_8.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fe3cec27c7c7d26e278c4c909f291cfc8e746c0ef796674e371d24e67bfdcdf8
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4f54cc251e64fc43ea7906bcda268979b7ccf5730ecf572e5f7e35696eaef1fc
3
  size 27772
{checkpoint-1900 → checkpoint-2300}/rng_state_9.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2902c9d6da9c90e9824b8f085cb5711e2525da6171a58ba51b1db3542ae98b58
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:411f28f7aa334979626223b80737ac45d841073b4663a97895d7c8a8a90716d4
3
  size 27772
{checkpoint-1900 → checkpoint-2300}/scheduler.pt RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fb84c13748245de4501f13af61cd2bc02aa1649f62194e99b0d7e9c2a5675b25
3
  size 627
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9a4eb98896140d186e72cc6934fa7664ebaea1f79a33825c5ffebfaf2a088e1
3
  size 627
{checkpoint-1900 → checkpoint-2300}/trainer_state.json RENAMED
@@ -1,8 +1,8 @@
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
- "epoch": 0.7367196587824738,
5
- "global_step": 1900,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
@@ -446,11 +446,115 @@
446
  "learning_rate": 0.00017190070785548755,
447
  "loss": 0.8311,
448
  "step": 1900
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
449
  }
450
  ],
451
  "max_steps": 7737,
452
  "num_train_epochs": 3,
453
- "total_flos": 8.183794238153032e+18,
454
  "trial_name": null,
455
  "trial_params": null
456
  }
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 0.8918185343156262,
5
+ "global_step": 2300,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
 
446
  "learning_rate": 0.00017190070785548755,
447
  "loss": 0.8311,
448
  "step": 1900
449
+ },
450
+ {
451
+ "epoch": 0.75,
452
+ "learning_rate": 0.0001711905818122717,
453
+ "loss": 0.8333,
454
+ "step": 1925
455
+ },
456
+ {
457
+ "epoch": 0.76,
458
+ "learning_rate": 0.0001704731008695777,
459
+ "loss": 0.8387,
460
+ "step": 1950
461
+ },
462
+ {
463
+ "epoch": 0.77,
464
+ "learning_rate": 0.0001697483391523821,
465
+ "loss": 0.8442,
466
+ "step": 1975
467
+ },
468
+ {
469
+ "epoch": 0.78,
470
+ "learning_rate": 0.00016901637153785885,
471
+ "loss": 0.8399,
472
+ "step": 2000
473
+ },
474
+ {
475
+ "epoch": 0.78,
476
+ "eval_loss": 0.8339959383010864,
477
+ "eval_runtime": 58.5829,
478
+ "eval_samples_per_second": 12.461,
479
+ "eval_steps_per_second": 0.905,
480
+ "step": 2000
481
+ },
482
+ {
483
+ "epoch": 0.79,
484
+ "learning_rate": 0.0001682772736476434,
485
+ "loss": 0.8334,
486
+ "step": 2025
487
+ },
488
+ {
489
+ "epoch": 0.79,
490
+ "learning_rate": 0.0001675311218400201,
491
+ "loss": 0.835,
492
+ "step": 2050
493
+ },
494
+ {
495
+ "epoch": 0.8,
496
+ "learning_rate": 0.00016677799320203332,
497
+ "loss": 0.8368,
498
+ "step": 2075
499
+ },
500
+ {
501
+ "epoch": 0.81,
502
+ "learning_rate": 0.00016601796554152344,
503
+ "loss": 0.8278,
504
+ "step": 2100
505
+ },
506
+ {
507
+ "epoch": 0.82,
508
+ "learning_rate": 0.00016525111737908827,
509
+ "loss": 0.8334,
510
+ "step": 2125
511
+ },
512
+ {
513
+ "epoch": 0.83,
514
+ "learning_rate": 0.00016447752793997096,
515
+ "loss": 0.8416,
516
+ "step": 2150
517
+ },
518
+ {
519
+ "epoch": 0.84,
520
+ "learning_rate": 0.00016369727714587483,
521
+ "loss": 0.8297,
522
+ "step": 2175
523
+ },
524
+ {
525
+ "epoch": 0.85,
526
+ "learning_rate": 0.0001629104456067066,
527
+ "loss": 0.8327,
528
+ "step": 2200
529
+ },
530
+ {
531
+ "epoch": 0.86,
532
+ "learning_rate": 0.00016211711461224825,
533
+ "loss": 0.8324,
534
+ "step": 2225
535
+ },
536
+ {
537
+ "epoch": 0.87,
538
+ "learning_rate": 0.0001613173661237589,
539
+ "loss": 0.8313,
540
+ "step": 2250
541
+ },
542
+ {
543
+ "epoch": 0.88,
544
+ "learning_rate": 0.0001605112827655069,
545
+ "loss": 0.8292,
546
+ "step": 2275
547
+ },
548
+ {
549
+ "epoch": 0.89,
550
+ "learning_rate": 0.0001596989478162339,
551
+ "loss": 0.8334,
552
+ "step": 2300
553
  }
554
  ],
555
  "max_steps": 7737,
556
  "num_train_epochs": 3,
557
+ "total_flos": 9.906168731933868e+18,
558
  "trial_name": null,
559
  "trial_params": null
560
  }
{checkpoint-1900 → checkpoint-2300}/training_args.bin RENAMED
File without changes