ZeroUniqueness commited on
Commit
ba9b273
β€’
1 Parent(s): 392dc7b

Training in progress, step 6700

Browse files
Files changed (25) hide show
  1. adapter_model.bin +1 -1
  2. {checkpoint-6300 β†’ checkpoint-6600/adapter_model}/README.md +0 -0
  3. {checkpoint-6300 β†’ checkpoint-6600/adapter_model}/adapter_config.json +0 -0
  4. {checkpoint-6300 β†’ checkpoint-6600/adapter_model}/adapter_model.bin +1 -1
  5. {checkpoint-6300/adapter_model β†’ checkpoint-6700}/README.md +0 -0
  6. {checkpoint-6300/adapter_model β†’ checkpoint-6700}/adapter_config.json +0 -0
  7. {checkpoint-6300/adapter_model β†’ checkpoint-6700}/adapter_model.bin +1 -1
  8. {checkpoint-6300 β†’ checkpoint-6700}/optimizer.pt +1 -1
  9. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_0.pth +1 -1
  10. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_1.pth +1 -1
  11. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_10.pth +1 -1
  12. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_11.pth +1 -1
  13. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_12.pth +1 -1
  14. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_13.pth +1 -1
  15. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_2.pth +1 -1
  16. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_3.pth +1 -1
  17. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_4.pth +1 -1
  18. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_5.pth +1 -1
  19. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_6.pth +1 -1
  20. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_7.pth +1 -1
  21. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_8.pth +1 -1
  22. {checkpoint-6300 β†’ checkpoint-6700}/rng_state_9.pth +1 -1
  23. {checkpoint-6300 β†’ checkpoint-6700}/scheduler.pt +1 -1
  24. {checkpoint-6300 β†’ checkpoint-6700}/trainer_state.json +99 -3
  25. {checkpoint-6300 β†’ checkpoint-6700}/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:d81f9de68995df3ae7fe51e14a4d5a0482b3ff50173f115d320ee99247cb5e3d
3
  size 500897101
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bcb96297919ffccc23a8f48a581a3185b03d83d73a72799941d2f235cad389d3
3
  size 500897101
{checkpoint-6300 β†’ checkpoint-6600/adapter_model}/README.md RENAMED
File without changes
{checkpoint-6300 β†’ checkpoint-6600/adapter_model}/adapter_config.json RENAMED
File without changes
{checkpoint-6300 β†’ checkpoint-6600/adapter_model}/adapter_model.bin RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7a03009057bc17a55fb03559fceee9416e58df44f2b8cad404853c77ef5e25ff
3
  size 500897101
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d81f9de68995df3ae7fe51e14a4d5a0482b3ff50173f115d320ee99247cb5e3d
3
  size 500897101
{checkpoint-6300/adapter_model β†’ checkpoint-6700}/README.md RENAMED
File without changes
{checkpoint-6300/adapter_model β†’ checkpoint-6700}/adapter_config.json RENAMED
File without changes
{checkpoint-6300/adapter_model β†’ checkpoint-6700}/adapter_model.bin RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7a03009057bc17a55fb03559fceee9416e58df44f2b8cad404853c77ef5e25ff
3
  size 500897101
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bcb96297919ffccc23a8f48a581a3185b03d83d73a72799941d2f235cad389d3
3
  size 500897101
{checkpoint-6300 β†’ checkpoint-6700}/optimizer.pt RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:95b7209d5ad4284033cb4767ae444e7a78eba53804e8f6dfd2ae1d008b403ecb
3
  size 1001752701
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f62cd6479e017b9a26c6526efc05511bbf9df9083364daa29ee80aa0a5dfd359
3
  size 1001752701
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_0.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4b0892c501b6dd52efec58730585273a276e5cf9f6937d96e9fa38ae5af9abd6
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b849a067e4f8461499fff8af86ebd00e3d3bf92cc824ac056b45ce1c6f1c291d
3
  size 27772
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_1.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ed1c7ad44d4c44043a9a1040ab7a8d9bf8ccb6a798963c6beec8678ba53a99d5
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a4f7d4c58f6339b6807226b0366cd9b4808e6973c1c8a248799f75b54a950a71
3
  size 27772
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_10.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:51696bdc2a942c4b6659d0c7c032d47971cc912461202edc9cbf84be5b00b9d4
3
  size 27789
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:adfb08b6b4d0f86655bf2077164aac4e21f12609b7e13eaf1b85093e2bd5b575
3
  size 27789
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_11.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:cf2dc121d536b338f7f7ab71844d47b47fa50424082757312612a1ffb43ce6d6
3
  size 27789
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ee65fa4fe3df5dd0784bb0ffd9f8c43572441235f5f1584cd7e48ca6d391b3b
3
  size 27789
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_12.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:600cbb989fd1bd997e07afb75b51b0e9bb6bba91f6f55fd115d8b32524b2c2a2
3
  size 27789
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f81c59c719504cb60451e827226d5c96dd0d7b2c2029c662831e8cefda2c54c4
3
  size 27789
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_13.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4f2f55cc04908fd235b9c2e1fae05ac4e683f052b4ea8b4e70cb8d917af90168
3
  size 27789
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f42032bd3037f0ed187573de33bf981a8759d22c266510058950f3fe7fedd84
3
  size 27789
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_2.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:daf71be757b88dd90b278fe0d6abf41fdfd54d283da14639c698e92c6ee6204c
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6568983fb50a48df14df5e018a2b846aa35270eca396c9cde88d8e926a2b7e9e
3
  size 27772
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_3.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d8d3a91b3606a86dfbe9591df14b8680debec4ee7ccd893c21fc87ce694c0de6
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b63f3fa6b0e3996d213ddad5511a17eca0c4465010e640447f4b694b1b0964e2
3
  size 27772
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_4.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d6af24a322399a9faa75e21470cca01ffb561d5fc8d90ce309043c8aede63c59
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8bda1b1d8e9e2ec530f144ce7f12db517bc8a9fc24f2dada873ec9a52da5a86e
3
  size 27772
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_5.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6e40c161f1898bfa1d6decd05f837beb821a821ae27f1e34a964a8ec433edd39
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2376f635219a3647dc4c0e776c0a3111a0130b027a88d2ccb50344af206de2d4
3
  size 27772
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_6.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d5a9e186bb4ebc2493f0fa088e88be12fd79ba92df4b0d1984d04b88fb44615e
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dc6a7fccb0aa5e4d85e4648b63853053ae8aaa55e6c9b6a833460f775d1fe07e
3
  size 27772
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_7.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5389b58d9b2d1b81fdd14d9dfa214e64e33c18b1e015479871952bbf5fbc9a9f
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:39cfd7753ba250ce6a1634c239e3256daf18c248cc833b55bb6b9e6ce312d51f
3
  size 27772
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_8.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:964ebee03695528c749c601db479754170f0284c59a3215d8dfcd92e48f3ae00
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b2c755eb589e46f504ec1fa833022df48b211aa54b7f5ff4f0f1a446258e9f93
3
  size 27772
{checkpoint-6300 β†’ checkpoint-6700}/rng_state_9.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:20f2b548b149282f75edcff3a569214186d9d16e83d775ade7543af31b397f99
3
  size 27772
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2635059e555f5db3a1c952c755fa2e418c4d0df4b1ef65c28677fc28d92db24d
3
  size 27772
{checkpoint-6300 β†’ checkpoint-6700}/scheduler.pt RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fc486d593be71181a7b350459598105a51de879e181677c7b4a0bcff932974dc
3
  size 627
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e004103cf4786af8b81629072cde0f9b18441c01ffdcd8cae42df8cd38f3d412
3
  size 627
{checkpoint-6300 β†’ checkpoint-6700}/trainer_state.json RENAMED
@@ -1,8 +1,8 @@
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
- "epoch": 2.44280728964715,
5
- "global_step": 6300,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
@@ -1542,11 +1542,107 @@
1542
  "learning_rate": 1.6587177861727454e-05,
1543
  "loss": 0.7703,
1544
  "step": 6300
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1545
  }
1546
  ],
1547
  "max_steps": 7737,
1548
  "num_train_epochs": 3,
1549
- "total_flos": 2.7132330413479952e+19,
1550
  "trial_name": null,
1551
  "trial_params": null
1552
  }
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 2.5979061651803024,
5
+ "global_step": 6700,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
 
1542
  "learning_rate": 1.6587177861727454e-05,
1543
  "loss": 0.7703,
1544
  "step": 6300
1545
+ },
1546
+ {
1547
+ "epoch": 2.45,
1548
+ "learning_rate": 1.6030861503198204e-05,
1549
+ "loss": 0.7658,
1550
+ "step": 6325
1551
+ },
1552
+ {
1553
+ "epoch": 2.46,
1554
+ "learning_rate": 1.548322023282235e-05,
1555
+ "loss": 0.7677,
1556
+ "step": 6350
1557
+ },
1558
+ {
1559
+ "epoch": 2.47,
1560
+ "learning_rate": 1.4944310628961666e-05,
1561
+ "loss": 0.7694,
1562
+ "step": 6375
1563
+ },
1564
+ {
1565
+ "epoch": 2.48,
1566
+ "learning_rate": 1.4414188367884907e-05,
1567
+ "loss": 0.7668,
1568
+ "step": 6400
1569
+ },
1570
+ {
1571
+ "epoch": 2.49,
1572
+ "learning_rate": 1.3892908218015654e-05,
1573
+ "loss": 0.7662,
1574
+ "step": 6425
1575
+ },
1576
+ {
1577
+ "epoch": 2.5,
1578
+ "learning_rate": 1.3380524034274122e-05,
1579
+ "loss": 0.7689,
1580
+ "step": 6450
1581
+ },
1582
+ {
1583
+ "epoch": 2.51,
1584
+ "learning_rate": 1.2877088752513189e-05,
1585
+ "loss": 0.7694,
1586
+ "step": 6475
1587
+ },
1588
+ {
1589
+ "epoch": 2.52,
1590
+ "learning_rate": 1.2382654384049475e-05,
1591
+ "loss": 0.7714,
1592
+ "step": 6500
1593
+ },
1594
+ {
1595
+ "epoch": 2.53,
1596
+ "learning_rate": 1.1897272010289884e-05,
1597
+ "loss": 0.7701,
1598
+ "step": 6525
1599
+ },
1600
+ {
1601
+ "epoch": 2.54,
1602
+ "learning_rate": 1.1420991777454315e-05,
1603
+ "loss": 0.7628,
1604
+ "step": 6550
1605
+ },
1606
+ {
1607
+ "epoch": 2.55,
1608
+ "learning_rate": 1.0953862891394795e-05,
1609
+ "loss": 0.7661,
1610
+ "step": 6575
1611
+ },
1612
+ {
1613
+ "epoch": 2.56,
1614
+ "learning_rate": 1.0495933612511976e-05,
1615
+ "loss": 0.7729,
1616
+ "step": 6600
1617
+ },
1618
+ {
1619
+ "epoch": 2.57,
1620
+ "learning_rate": 1.0047251250769175e-05,
1621
+ "loss": 0.772,
1622
+ "step": 6625
1623
+ },
1624
+ {
1625
+ "epoch": 2.58,
1626
+ "learning_rate": 9.60786216080466e-06,
1627
+ "loss": 0.7702,
1628
+ "step": 6650
1629
+ },
1630
+ {
1631
+ "epoch": 2.59,
1632
+ "learning_rate": 9.177811737142627e-06,
1633
+ "loss": 0.7711,
1634
+ "step": 6675
1635
+ },
1636
+ {
1637
+ "epoch": 2.6,
1638
+ "learning_rate": 8.757144409503359e-06,
1639
+ "loss": 0.765,
1640
+ "step": 6700
1641
  }
1642
  ],
1643
  "max_steps": 7737,
1644
  "num_train_epochs": 3,
1645
+ "total_flos": 2.885490295679274e+19,
1646
  "trial_name": null,
1647
  "trial_params": null
1648
  }
{checkpoint-6300 β†’ checkpoint-6700}/training_args.bin RENAMED
File without changes