ZeroUniqueness commited on
Commit
deec9a7
·
1 Parent(s): 6f15860

Training in progress, step 2000

Browse files
Files changed (25) hide show
  1. adapter_model.bin +1 -1
  2. {checkpoint-1600 → checkpoint-1900/adapter_model}/README.md +0 -0
  3. {checkpoint-1600 → checkpoint-1900/adapter_model}/adapter_config.json +0 -0
  4. {checkpoint-1600 → checkpoint-1900/adapter_model}/adapter_model.bin +1 -1
  5. {checkpoint-1600/adapter_model → checkpoint-2000}/README.md +0 -0
  6. {checkpoint-1600/adapter_model → checkpoint-2000}/adapter_config.json +0 -0
  7. {checkpoint-1600/adapter_model → checkpoint-2000}/adapter_model.bin +1 -1
  8. {checkpoint-1600 → checkpoint-2000}/optimizer.pt +1 -1
  9. {checkpoint-1600 → checkpoint-2000}/rng_state_0.pth +1 -1
  10. {checkpoint-1600 → checkpoint-2000}/rng_state_1.pth +1 -1
  11. {checkpoint-1600 → checkpoint-2000}/rng_state_10.pth +1 -1
  12. {checkpoint-1600 → checkpoint-2000}/rng_state_11.pth +1 -1
  13. {checkpoint-1600 → checkpoint-2000}/rng_state_12.pth +1 -1
  14. {checkpoint-1600 → checkpoint-2000}/rng_state_13.pth +1 -1
  15. {checkpoint-1600 → checkpoint-2000}/rng_state_2.pth +1 -1
  16. {checkpoint-1600 → checkpoint-2000}/rng_state_3.pth +1 -1
  17. {checkpoint-1600 → checkpoint-2000}/rng_state_4.pth +1 -1
  18. {checkpoint-1600 → checkpoint-2000}/rng_state_5.pth +1 -1
  19. {checkpoint-1600 → checkpoint-2000}/rng_state_6.pth +1 -1
  20. {checkpoint-1600 → checkpoint-2000}/rng_state_7.pth +1 -1
  21. {checkpoint-1600 → checkpoint-2000}/rng_state_8.pth +1 -1
  22. {checkpoint-1600 → checkpoint-2000}/rng_state_9.pth +1 -1
  23. {checkpoint-1600 → checkpoint-2000}/scheduler.pt +1 -1
  24. {checkpoint-1600 → checkpoint-2000}/trainer_state.json +107 -3
  25. {checkpoint-1600 → checkpoint-2000}/training_args.bin +0 -0
adapter_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6757a53516f65f459191f0812f3ff6e5c08faa4d4174d508234e365b3d269f49
3
  size 500897101
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e255c5f9085e062010ebf826aaa4734bdc7eae29433a656fa79e015e9e0dcf93
3
  size 500897101
{checkpoint-1600 → checkpoint-1900/adapter_model}/README.md RENAMED
File without changes
{checkpoint-1600 → checkpoint-1900/adapter_model}/adapter_config.json RENAMED
File without changes
{checkpoint-1600 → checkpoint-1900/adapter_model}/adapter_model.bin RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bf431539f287ae1dba4b87e2d4849cf4cece39ed2b439a95d6b40beac2eb5247
3
  size 500897101
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6757a53516f65f459191f0812f3ff6e5c08faa4d4174d508234e365b3d269f49
3
  size 500897101
{checkpoint-1600/adapter_model → checkpoint-2000}/README.md RENAMED
File without changes
{checkpoint-1600/adapter_model → checkpoint-2000}/adapter_config.json RENAMED
File without changes
{checkpoint-1600/adapter_model → checkpoint-2000}/adapter_model.bin RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bf431539f287ae1dba4b87e2d4849cf4cece39ed2b439a95d6b40beac2eb5247
3
  size 500897101
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e255c5f9085e062010ebf826aaa4734bdc7eae29433a656fa79e015e9e0dcf93
3
  size 500897101
{checkpoint-1600 → checkpoint-2000}/optimizer.pt RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:80d3d4692c287de8e2ecff9b21b7da82823721ef5d905ab99d17e14915b4eb4d
3
  size 1001752701
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28bef127f06c65c0ade137496ae38f90c86c6325d6bb1cae7e0a6af69632772c
3
  size 1001752701
{checkpoint-1600 → checkpoint-2000}/rng_state_0.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:48a5d505b1c01b2b69f6517cf50d2aaa5ebf16e07731c5642a5f65d49262d951
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:27579ec3d9b91002c52255be7b19d3672a1f1ec3d429da98c8c1ef2633d0f251
3
  size 27772
{checkpoint-1600 → checkpoint-2000}/rng_state_1.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:095a268a99c08b7e3d610ed61ae84f6a27f6844746846573127b3a33df4de242
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:36aa526f7071e27d49bc000e919db98302c51d02b19d500fc264f3e0104c3cea
3
  size 27772
{checkpoint-1600 → checkpoint-2000}/rng_state_10.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:520e6e5f7ced8af633c38bd45e6f6df79479350618291ccf9a241ffe53cd3210
3
  size 27789
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4477c8c81e33334e074cce529d4d2e1aca0ebc913cbff90da614b6721c62e750
3
  size 27789
{checkpoint-1600 → checkpoint-2000}/rng_state_11.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:eafb507adba67e6a1d08266d4a4fc6fa6d253de9a1cf9a8bc679d67841101d68
3
  size 27789
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4f09945398b8c4def355346fb2902311eee35a47828fe712e4e12da2cf1088e2
3
  size 27789
{checkpoint-1600 → checkpoint-2000}/rng_state_12.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fa1dff2167bd8bf87dcd2c18f7c0a0657160fc353b6c2cdc903a705286e69225
3
  size 27789
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7e5231ab3fa56f538d8bf8dffc51620ca0b1c2b4be79895e7033a2d8961db87b
3
  size 27789
{checkpoint-1600 → checkpoint-2000}/rng_state_13.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ac85c882576b1e5483b6c2d0288e23b15c8ac0f2d8e6d7c82f1b2ad5799191d3
3
  size 27789
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:209766a80558d37b531dd5a9ece109bba80871085d75fba9d12ee04914d9d026
3
  size 27789
{checkpoint-1600 → checkpoint-2000}/rng_state_2.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:1131f5c97d79f2d2e02fedab30ef318979d5eaf0576c2156d6bd3f4b2dd23eb7
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1f7b057c3857c8cf42ac255dc1f7ac76a7f0dd8fdcc135cfae39ecc1eabf9f23
3
  size 27772
{checkpoint-1600 → checkpoint-2000}/rng_state_3.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fac0d37e645f8dd70bdc29707f5fdcc1c5fd30db1bcc146793709d8030b945a5
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:941e6b1a442dfc2d1b2b7c75c57e3bdd075babe8cce2f08a91138c93791d72eb
3
  size 27772
{checkpoint-1600 → checkpoint-2000}/rng_state_4.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f766a08bf0befd1cc26c67dce6fac9420a473620d0d5c215093331e7c575c408
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a1cbca6fe49890e1ccfca7586f80985a6851928705fbf9966c443ffc513fe81b
3
  size 27772
{checkpoint-1600 → checkpoint-2000}/rng_state_5.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:02df72ae6be7fdf053ebec2bdbebd30eba8cde3dd592fa1e7de8b965dacdaced
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:004d6dfb04b5fd343b24f7be1634cb7a20bcf37ac38d640d3a1a99d43d3d07ed
3
  size 27772
{checkpoint-1600 → checkpoint-2000}/rng_state_6.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:b041cefa32e536e9d9830f60fd1bafae40a6618c29f992f5a1e013b7d4c2de2b
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c239f45c218b397b8ca21537396eab08b3b14c3555ba4f49a1e9b08508c48b84
3
  size 27772
{checkpoint-1600 → checkpoint-2000}/rng_state_7.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a96bfee948fdbcf4a703afb0f1286672bd6667e9d481b3381c8adfaba450828c
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:79c9beb6328a8bf32a2ebe4ac82b5901d131a60a7dc20376e4f388428a37dce9
3
  size 27772
{checkpoint-1600 → checkpoint-2000}/rng_state_8.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7ae907eefc9027a96bf5f9853790b5600a54254215b147ad9f8868e433a2cabd
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:056fe2846efa06081e041d52ab372d86a13d0750aa4104d759eb2a50f503424c
3
  size 27772
{checkpoint-1600 → checkpoint-2000}/rng_state_9.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d7c381466400808a5ccdb730543ed011128c0465075464739a5da7c5f3003641
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d058c79c3ccf68d8b85de9450fb58c574dfe9bcb5e30663becaa9548d64ff5ce
3
  size 27772
{checkpoint-1600 → checkpoint-2000}/scheduler.pt RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a2d3d1b9becf2b722060450ced20a29a5568255b78e3e01e310ed6863548230b
3
  size 627
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5aa05ae2d8bfb7f529ed4f5f986a046fc64331a971b9e53a2aa8cdd3a0259708
3
  size 627
{checkpoint-1600 → checkpoint-2000}/trainer_state.json RENAMED
@@ -1,8 +1,8 @@
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
- "epoch": 0.6203955021326095,
5
- "global_step": 1600,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
@@ -374,11 +374,115 @@
374
  "learning_rate": 0.00017982261628553842,
375
  "loss": 0.8349,
376
  "step": 1600
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
377
  }
378
  ],
379
  "max_steps": 7737,
380
  "num_train_epochs": 3,
381
- "total_flos": 6.890881779226378e+18,
382
  "trial_name": null,
383
  "trial_params": null
384
  }
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 0.7754943776657619,
5
+ "global_step": 2000,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
 
374
  "learning_rate": 0.00017982261628553842,
375
  "loss": 0.8349,
376
  "step": 1600
377
+ },
378
+ {
379
+ "epoch": 0.63,
380
+ "learning_rate": 0.0001792062467707703,
381
+ "loss": 0.8483,
382
+ "step": 1625
383
+ },
384
+ {
385
+ "epoch": 0.64,
386
+ "learning_rate": 0.0001785816942356052,
387
+ "loss": 0.8387,
388
+ "step": 1650
389
+ },
390
+ {
391
+ "epoch": 0.65,
392
+ "learning_rate": 0.00017794902320432429,
393
+ "loss": 0.843,
394
+ "step": 1675
395
+ },
396
+ {
397
+ "epoch": 0.66,
398
+ "learning_rate": 0.00017730829903995333,
399
+ "loss": 0.8424,
400
+ "step": 1700
401
+ },
402
+ {
403
+ "epoch": 0.67,
404
+ "learning_rate": 0.00017665958793751006,
405
+ "loss": 0.8418,
406
+ "step": 1725
407
+ },
408
+ {
409
+ "epoch": 0.68,
410
+ "learning_rate": 0.00017600295691716522,
411
+ "loss": 0.8384,
412
+ "step": 1750
413
+ },
414
+ {
415
+ "epoch": 0.69,
416
+ "learning_rate": 0.00017533847381731856,
417
+ "loss": 0.8445,
418
+ "step": 1775
419
+ },
420
+ {
421
+ "epoch": 0.7,
422
+ "learning_rate": 0.00017466620728759033,
423
+ "loss": 0.8446,
424
+ "step": 1800
425
+ },
426
+ {
427
+ "epoch": 0.71,
428
+ "learning_rate": 0.00017398622678172878,
429
+ "loss": 0.838,
430
+ "step": 1825
431
+ },
432
+ {
433
+ "epoch": 0.72,
434
+ "learning_rate": 0.0001732986025504348,
435
+ "loss": 0.8415,
436
+ "step": 1850
437
+ },
438
+ {
439
+ "epoch": 0.73,
440
+ "learning_rate": 0.000172603405634104,
441
+ "loss": 0.8357,
442
+ "step": 1875
443
+ },
444
+ {
445
+ "epoch": 0.74,
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
  "max_steps": 7737,
484
  "num_train_epochs": 3,
485
+ "total_flos": 8.614008003953164e+18,
486
  "trial_name": null,
487
  "trial_params": null
488
  }
{checkpoint-1600 → checkpoint-2000}/training_args.bin RENAMED
File without changes