Commit
•
059aa39
1
Parent(s):
1238de9
Model save
Browse files- README.md +32 -22
- adapter_model.safetensors +1 -1
- all_results.json +5 -5
- train_results.json +5 -5
- trainer_state.json +2030 -90
README.md
CHANGED
|
@@ -1,13 +1,11 @@
|
|
| 1 |
---
|
| 2 |
library_name: peft
|
| 3 |
tags:
|
| 4 |
-
- alignment-handbook
|
| 5 |
- trl
|
| 6 |
- dpo
|
|
|
|
| 7 |
- generated_from_trainer
|
| 8 |
base_model: NbAiLab/nb-gpt-j-6B-v2
|
| 9 |
-
datasets:
|
| 10 |
-
- hugodk-sch/aftonposten_title_prefs
|
| 11 |
model-index:
|
| 12 |
- name: aftonposten-6b-align-scan
|
| 13 |
results: []
|
|
@@ -18,17 +16,17 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 18 |
|
| 19 |
# aftonposten-6b-align-scan
|
| 20 |
|
| 21 |
-
This model is a fine-tuned version of [
|
| 22 |
It achieves the following results on the evaluation set:
|
| 23 |
-
- Loss: 0.
|
| 24 |
-
- Rewards/chosen: -0.
|
| 25 |
-
- Rewards/rejected: -0.
|
| 26 |
-
- Rewards/accuracies: 0.
|
| 27 |
-
- Rewards/margins: 0.
|
| 28 |
-
- Logps/rejected: -
|
| 29 |
-
- Logps/chosen: -34.
|
| 30 |
-
- Logits/rejected: -2.
|
| 31 |
-
- Logits/chosen: -2.
|
| 32 |
|
| 33 |
## Model description
|
| 34 |
|
|
@@ -57,21 +55,33 @@ The following hyperparameters were used during training:
|
|
| 57 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 58 |
- lr_scheduler_type: cosine
|
| 59 |
- lr_scheduler_warmup_ratio: 0.1
|
| 60 |
-
- num_epochs:
|
| 61 |
|
| 62 |
### Training results
|
| 63 |
|
| 64 |
-
| Training Loss | Epoch | Step |
|
| 65 |
-
|
| 66 |
-
| 0.9521 | 0.26 | 100 |
|
| 67 |
-
| 0.8515 | 0.52 | 200 |
|
| 68 |
-
| 0.7473 | 0.78 | 300 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 69 |
|
| 70 |
|
| 71 |
### Framework versions
|
| 72 |
|
| 73 |
-
- PEFT 0.
|
| 74 |
-
- Transformers 4.
|
| 75 |
- Pytorch 2.1.2+cu121
|
| 76 |
-
- Datasets 2.
|
| 77 |
- Tokenizers 0.15.1
|
|
|
|
| 1 |
---
|
| 2 |
library_name: peft
|
| 3 |
tags:
|
|
|
|
| 4 |
- trl
|
| 5 |
- dpo
|
| 6 |
+
- alignment-handbook
|
| 7 |
- generated_from_trainer
|
| 8 |
base_model: NbAiLab/nb-gpt-j-6B-v2
|
|
|
|
|
|
|
| 9 |
model-index:
|
| 10 |
- name: aftonposten-6b-align-scan
|
| 11 |
results: []
|
|
|
|
| 16 |
|
| 17 |
# aftonposten-6b-align-scan
|
| 18 |
|
| 19 |
+
This model is a fine-tuned version of [NbAiLab/nb-gpt-j-6B-v2](https://huggingface.co/NbAiLab/nb-gpt-j-6B-v2) on the None dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.9280
|
| 22 |
+
- Rewards/chosen: -0.1384
|
| 23 |
+
- Rewards/rejected: -0.2143
|
| 24 |
+
- Rewards/accuracies: 0.5806
|
| 25 |
+
- Rewards/margins: 0.0758
|
| 26 |
+
- Logps/rejected: -38.0523
|
| 27 |
+
- Logps/chosen: -34.3807
|
| 28 |
+
- Logits/rejected: -2.1358
|
| 29 |
+
- Logits/chosen: -2.1405
|
| 30 |
|
| 31 |
## Model description
|
| 32 |
|
|
|
|
| 55 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 56 |
- lr_scheduler_type: cosine
|
| 57 |
- lr_scheduler_warmup_ratio: 0.1
|
| 58 |
+
- num_epochs: 4
|
| 59 |
|
| 60 |
### Training results
|
| 61 |
|
| 62 |
+
| Training Loss | Epoch | Step | Logits/chosen | Logits/rejected | Logps/chosen | Logps/rejected | Validation Loss | Rewards/accuracies | Rewards/chosen | Rewards/margins | Rewards/rejected |
|
| 63 |
+
|:-------------:|:-----:|:----:|:-------------:|:---------------:|:------------:|:--------------:|:---------------:|:------------------:|:--------------:|:---------------:|:----------------:|
|
| 64 |
+
| 0.9521 | 0.26 | 100 | -2.2327 | -2.2279 | -34.0381 | -37.5401 | 0.9920 | 0.5361 | -0.0014 | 0.0080 | -0.0094 |
|
| 65 |
+
| 0.8515 | 0.52 | 200 | -2.2274 | -2.2226 | -34.0451 | -37.5730 | 0.9817 | 0.5594 | -0.0042 | 0.0184 | -0.0226 |
|
| 66 |
+
| 0.7473 | 0.78 | 300 | -2.2234 | -2.2186 | -34.0922 | -37.6165 | 0.9830 | 0.5390 | -0.0231 | 0.0169 | -0.0399 |
|
| 67 |
+
| 0.5492 | 1.04 | 400 | 0.9600 | -0.0653 | -0.1052 | 0.5565 | 0.0399 | -37.7797 | -34.1978 | -2.1852 | -2.1901 |
|
| 68 |
+
| 0.4504 | 1.3 | 500 | 0.9656 | -0.1403 | -0.1747 | 0.5685 | 0.0344 | -37.9533 | -34.3853 | -2.1770 | -2.1818 |
|
| 69 |
+
| 0.4511 | 1.56 | 600 | 0.9338 | -0.0835 | -0.1505 | 0.5951 | 0.0670 | -37.8930 | -34.2433 | -2.1565 | -2.1613 |
|
| 70 |
+
| 0.3805 | 1.82 | 700 | 0.9385 | -0.1061 | -0.1691 | 0.5511 | 0.0631 | -37.9394 | -34.2997 | -2.1442 | -2.1490 |
|
| 71 |
+
| 0.2038 | 2.08 | 800 | 0.9334 | -0.1274 | -0.1969 | 0.5689 | 0.0695 | -38.0088 | -34.3529 | -2.1367 | -2.1415 |
|
| 72 |
+
| 0.2332 | 2.34 | 900 | 0.9320 | -0.1353 | -0.2057 | 0.5718 | 0.0704 | -38.0309 | -34.3727 | -2.1352 | -2.1400 |
|
| 73 |
+
| 0.28 | 2.6 | 1000 | 0.9271 | -0.1454 | -0.2223 | 0.5714 | 0.0769 | -38.0723 | -34.3980 | -2.1345 | -2.1393 |
|
| 74 |
+
| 0.1953 | 2.86 | 1100 | 0.9399 | -0.1551 | -0.2188 | 0.5631 | 0.0637 | -38.0636 | -34.4223 | -2.1342 | -2.1390 |
|
| 75 |
+
| 0.2936 | 3.12 | 1200 | 0.9311 | -0.1412 | -0.2138 | 0.5864 | 0.0725 | -38.0510 | -34.3876 | -2.1359 | -2.1407 |
|
| 76 |
+
| 0.1526 | 3.38 | 1300 | 0.9307 | -0.1413 | -0.2150 | 0.5864 | 0.0737 | -38.0541 | -34.3877 | -2.1350 | -2.1398 |
|
| 77 |
+
| 0.1121 | 3.64 | 1400 | 0.9250 | -0.1377 | -0.2153 | 0.6038 | 0.0776 | -38.0548 | -34.3788 | -2.1356 | -2.1404 |
|
| 78 |
+
| 0.215 | 3.9 | 1500 | 0.9280 | -0.1384 | -0.2143 | 0.5806 | 0.0758 | -38.0523 | -34.3807 | -2.1358 | -2.1405 |
|
| 79 |
|
| 80 |
|
| 81 |
### Framework versions
|
| 82 |
|
| 83 |
+
- PEFT 0.10.0
|
| 84 |
+
- Transformers 4.39.0.dev0
|
| 85 |
- Pytorch 2.1.2+cu121
|
| 86 |
+
- Datasets 2.14.6
|
| 87 |
- Tokenizers 0.15.1
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 176183216
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1d0fa9fe38bf0a36e47145a69312452d9da6b70819ba3ecfd9754733dff76761
|
| 3 |
size 176183216
|
all_results.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
"eval_logits/chosen": -2.223268747329712,
|
| 4 |
"eval_logits/rejected": -2.218452215194702,
|
| 5 |
"eval_logps/chosen": -34.1085205078125,
|
|
@@ -13,9 +13,9 @@
|
|
| 13 |
"eval_samples": 343,
|
| 14 |
"eval_samples_per_second": 2.359,
|
| 15 |
"eval_steps_per_second": 0.296,
|
| 16 |
-
"train_loss": 0.
|
| 17 |
-
"train_runtime":
|
| 18 |
"train_samples": 3079,
|
| 19 |
-
"train_samples_per_second":
|
| 20 |
-
"train_steps_per_second": 0.
|
| 21 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
"eval_logits/chosen": -2.223268747329712,
|
| 4 |
"eval_logits/rejected": -2.218452215194702,
|
| 5 |
"eval_logps/chosen": -34.1085205078125,
|
|
|
|
| 13 |
"eval_samples": 343,
|
| 14 |
"eval_samples_per_second": 2.359,
|
| 15 |
"eval_steps_per_second": 0.296,
|
| 16 |
+
"train_loss": 0.2625940967689861,
|
| 17 |
+
"train_runtime": 10748.6274,
|
| 18 |
"train_samples": 3079,
|
| 19 |
+
"train_samples_per_second": 1.146,
|
| 20 |
+
"train_steps_per_second": 0.143
|
| 21 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 3079,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
+
"train_loss": 0.2625940967689861,
|
| 4 |
+
"train_runtime": 10748.6274,
|
| 5 |
"train_samples": 3079,
|
| 6 |
+
"train_samples_per_second": 1.146,
|
| 7 |
+
"train_steps_per_second": 0.143
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 100,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -492,130 +492,2070 @@
|
|
| 492 |
},
|
| 493 |
{
|
| 494 |
"epoch": 0.81,
|
| 495 |
-
"
|
| 496 |
-
"
|
| 497 |
-
"logits/
|
| 498 |
-
"
|
| 499 |
-
"logps/
|
| 500 |
-
"
|
| 501 |
-
"
|
| 502 |
-
"rewards/
|
| 503 |
-
"rewards/
|
| 504 |
-
"rewards/
|
|
|
|
| 505 |
"step": 310
|
| 506 |
},
|
| 507 |
{
|
| 508 |
"epoch": 0.83,
|
| 509 |
-
"
|
| 510 |
-
"
|
| 511 |
-
"logits/
|
| 512 |
-
"
|
| 513 |
-
"logps/
|
| 514 |
-
"
|
| 515 |
-
"
|
| 516 |
-
"rewards/
|
| 517 |
-
"rewards/
|
| 518 |
-
"rewards/
|
|
|
|
| 519 |
"step": 320
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 0.86,
|
| 523 |
-
"
|
| 524 |
-
"
|
| 525 |
-
"logits/
|
| 526 |
-
"
|
| 527 |
-
"logps/
|
| 528 |
-
"
|
| 529 |
-
"
|
| 530 |
-
"rewards/
|
| 531 |
-
"rewards/
|
| 532 |
-
"rewards/
|
|
|
|
| 533 |
"step": 330
|
| 534 |
},
|
| 535 |
{
|
| 536 |
"epoch": 0.88,
|
| 537 |
-
"
|
| 538 |
-
"
|
| 539 |
-
"logits/
|
| 540 |
-
"
|
| 541 |
-
"logps/
|
| 542 |
-
"
|
| 543 |
-
"
|
| 544 |
-
"rewards/
|
| 545 |
-
"rewards/
|
| 546 |
-
"rewards/
|
|
|
|
| 547 |
"step": 340
|
| 548 |
},
|
| 549 |
{
|
| 550 |
"epoch": 0.91,
|
| 551 |
-
"
|
| 552 |
-
"
|
| 553 |
-
"logits/
|
| 554 |
-
"
|
| 555 |
-
"logps/
|
| 556 |
-
"
|
| 557 |
-
"
|
| 558 |
-
"rewards/
|
| 559 |
-
"rewards/
|
| 560 |
-
"rewards/
|
|
|
|
| 561 |
"step": 350
|
| 562 |
},
|
| 563 |
{
|
| 564 |
"epoch": 0.94,
|
| 565 |
-
"
|
| 566 |
-
"
|
| 567 |
-
"logits/
|
| 568 |
-
"
|
| 569 |
-
"logps/
|
| 570 |
-
"
|
| 571 |
-
"
|
| 572 |
-
"rewards/
|
| 573 |
-
"rewards/
|
| 574 |
-
"rewards/
|
|
|
|
| 575 |
"step": 360
|
| 576 |
},
|
| 577 |
{
|
| 578 |
"epoch": 0.96,
|
| 579 |
-
"
|
| 580 |
-
"
|
| 581 |
-
"logits/
|
| 582 |
-
"
|
| 583 |
-
"logps/
|
| 584 |
-
"
|
| 585 |
-
"
|
| 586 |
-
"rewards/
|
| 587 |
-
"rewards/
|
| 588 |
-
"rewards/
|
|
|
|
| 589 |
"step": 370
|
| 590 |
},
|
| 591 |
{
|
| 592 |
"epoch": 0.99,
|
| 593 |
-
"
|
| 594 |
-
"
|
| 595 |
-
"logits/
|
| 596 |
-
"
|
| 597 |
-
"logps/
|
| 598 |
-
"
|
| 599 |
-
"
|
| 600 |
-
"rewards/
|
| 601 |
-
"rewards/
|
| 602 |
-
"rewards/
|
|
|
|
| 603 |
"step": 380
|
| 604 |
},
|
| 605 |
{
|
| 606 |
-
"epoch": 1.
|
| 607 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 608 |
"total_flos": 0.0,
|
| 609 |
-
"train_loss": 0.
|
| 610 |
-
"train_runtime":
|
| 611 |
-
"train_samples_per_second":
|
| 612 |
-
"train_steps_per_second": 0.
|
| 613 |
}
|
| 614 |
],
|
| 615 |
"logging_steps": 10,
|
| 616 |
-
"max_steps":
|
| 617 |
"num_input_tokens_seen": 0,
|
| 618 |
-
"num_train_epochs":
|
| 619 |
"save_steps": 100,
|
| 620 |
"total_flos": 0.0,
|
| 621 |
"train_batch_size": 4,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 4.0,
|
| 5 |
"eval_steps": 100,
|
| 6 |
+
"global_step": 1540,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 492 |
},
|
| 493 |
{
|
| 494 |
"epoch": 0.81,
|
| 495 |
+
"grad_norm": 19.375,
|
| 496 |
+
"learning_rate": 4.84533120650964e-06,
|
| 497 |
+
"logits/chosen": -2.057250499725342,
|
| 498 |
+
"logits/rejected": -2.0444836616516113,
|
| 499 |
+
"logps/chosen": -32.08330535888672,
|
| 500 |
+
"logps/rejected": -32.93002700805664,
|
| 501 |
+
"loss": 0.7174,
|
| 502 |
+
"rewards/accuracies": 0.875,
|
| 503 |
+
"rewards/chosen": 0.17803600430488586,
|
| 504 |
+
"rewards/margins": 0.29362356662750244,
|
| 505 |
+
"rewards/rejected": -0.11558754742145538,
|
| 506 |
"step": 310
|
| 507 |
},
|
| 508 |
{
|
| 509 |
"epoch": 0.83,
|
| 510 |
+
"grad_norm": 18.375,
|
| 511 |
+
"learning_rate": 4.825108134172131e-06,
|
| 512 |
+
"logits/chosen": -1.9667476415634155,
|
| 513 |
+
"logits/rejected": -1.9581434726715088,
|
| 514 |
+
"logps/chosen": -31.69952964782715,
|
| 515 |
+
"logps/rejected": -30.477031707763672,
|
| 516 |
+
"loss": 0.6654,
|
| 517 |
+
"rewards/accuracies": 0.824999988079071,
|
| 518 |
+
"rewards/chosen": 0.26218193769454956,
|
| 519 |
+
"rewards/margins": 0.35643792152404785,
|
| 520 |
+
"rewards/rejected": -0.09425600618124008,
|
| 521 |
"step": 320
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 0.86,
|
| 525 |
+
"grad_norm": 13.5,
|
| 526 |
+
"learning_rate": 4.80369052967602e-06,
|
| 527 |
+
"logits/chosen": -1.902342438697815,
|
| 528 |
+
"logits/rejected": -1.914393424987793,
|
| 529 |
+
"logps/chosen": -29.752674102783203,
|
| 530 |
+
"logps/rejected": -33.74560546875,
|
| 531 |
+
"loss": 0.5789,
|
| 532 |
+
"rewards/accuracies": 0.824999988079071,
|
| 533 |
+
"rewards/chosen": 0.29063910245895386,
|
| 534 |
+
"rewards/margins": 0.427977979183197,
|
| 535 |
+
"rewards/rejected": -0.13733889162540436,
|
| 536 |
"step": 330
|
| 537 |
},
|
| 538 |
{
|
| 539 |
"epoch": 0.88,
|
| 540 |
+
"grad_norm": 23.375,
|
| 541 |
+
"learning_rate": 4.781089396387968e-06,
|
| 542 |
+
"logits/chosen": -1.8653125762939453,
|
| 543 |
+
"logits/rejected": -1.8561639785766602,
|
| 544 |
+
"logps/chosen": -33.982444763183594,
|
| 545 |
+
"logps/rejected": -36.24190902709961,
|
| 546 |
+
"loss": 0.6225,
|
| 547 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 548 |
+
"rewards/chosen": 0.2865905463695526,
|
| 549 |
+
"rewards/margins": 0.44066372513771057,
|
| 550 |
+
"rewards/rejected": -0.15407314896583557,
|
| 551 |
"step": 340
|
| 552 |
},
|
| 553 |
{
|
| 554 |
"epoch": 0.91,
|
| 555 |
+
"grad_norm": 16.875,
|
| 556 |
+
"learning_rate": 4.757316345716554e-06,
|
| 557 |
+
"logits/chosen": -1.9161920547485352,
|
| 558 |
+
"logits/rejected": -1.9168342351913452,
|
| 559 |
+
"logps/chosen": -33.598670959472656,
|
| 560 |
+
"logps/rejected": -34.16885757446289,
|
| 561 |
+
"loss": 0.6015,
|
| 562 |
+
"rewards/accuracies": 0.8125,
|
| 563 |
+
"rewards/chosen": 0.3226447105407715,
|
| 564 |
+
"rewards/margins": 0.4505980908870697,
|
| 565 |
+
"rewards/rejected": -0.1279533952474594,
|
| 566 |
"step": 350
|
| 567 |
},
|
| 568 |
{
|
| 569 |
"epoch": 0.94,
|
| 570 |
+
"grad_norm": 17.75,
|
| 571 |
+
"learning_rate": 4.73238359114687e-06,
|
| 572 |
+
"logits/chosen": -2.0425257682800293,
|
| 573 |
+
"logits/rejected": -2.0486459732055664,
|
| 574 |
+
"logps/chosen": -31.08384132385254,
|
| 575 |
+
"logps/rejected": -33.032264709472656,
|
| 576 |
+
"loss": 0.6861,
|
| 577 |
+
"rewards/accuracies": 0.800000011920929,
|
| 578 |
+
"rewards/chosen": 0.19576172530651093,
|
| 579 |
+
"rewards/margins": 0.3413549065589905,
|
| 580 |
+
"rewards/rejected": -0.14559319615364075,
|
| 581 |
"step": 360
|
| 582 |
},
|
| 583 |
{
|
| 584 |
"epoch": 0.96,
|
| 585 |
+
"grad_norm": 20.5,
|
| 586 |
+
"learning_rate": 4.706303941965804e-06,
|
| 587 |
+
"logits/chosen": -1.9708837270736694,
|
| 588 |
+
"logits/rejected": -1.970446228981018,
|
| 589 |
+
"logps/chosen": -32.829166412353516,
|
| 590 |
+
"logps/rejected": -36.46767044067383,
|
| 591 |
+
"loss": 0.6072,
|
| 592 |
+
"rewards/accuracies": 0.862500011920929,
|
| 593 |
+
"rewards/chosen": 0.24631083011627197,
|
| 594 |
+
"rewards/margins": 0.41346845030784607,
|
| 595 |
+
"rewards/rejected": -0.16715766489505768,
|
| 596 |
"step": 370
|
| 597 |
},
|
| 598 |
{
|
| 599 |
"epoch": 0.99,
|
| 600 |
+
"grad_norm": 19.0,
|
| 601 |
+
"learning_rate": 4.679090796681225e-06,
|
| 602 |
+
"logits/chosen": -2.002481698989868,
|
| 603 |
+
"logits/rejected": -1.9979221820831299,
|
| 604 |
+
"logps/chosen": -30.1049747467041,
|
| 605 |
+
"logps/rejected": -29.63289451599121,
|
| 606 |
+
"loss": 0.6387,
|
| 607 |
+
"rewards/accuracies": 0.8125,
|
| 608 |
+
"rewards/chosen": 0.2422492504119873,
|
| 609 |
+
"rewards/margins": 0.40985146164894104,
|
| 610 |
+
"rewards/rejected": -0.16760216653347015,
|
| 611 |
"step": 380
|
| 612 |
},
|
| 613 |
{
|
| 614 |
+
"epoch": 1.01,
|
| 615 |
+
"grad_norm": 22.625,
|
| 616 |
+
"learning_rate": 4.650758136138454e-06,
|
| 617 |
+
"logits/chosen": -1.773651123046875,
|
| 618 |
+
"logits/rejected": -1.7800319194793701,
|
| 619 |
+
"logps/chosen": -31.610965728759766,
|
| 620 |
+
"logps/rejected": -36.748817443847656,
|
| 621 |
+
"loss": 0.523,
|
| 622 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 623 |
+
"rewards/chosen": 0.30989953875541687,
|
| 624 |
+
"rewards/margins": 0.5621098279953003,
|
| 625 |
+
"rewards/rejected": -0.25221019983291626,
|
| 626 |
+
"step": 390
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.04,
|
| 630 |
+
"grad_norm": 17.75,
|
| 631 |
+
"learning_rate": 4.621320516337559e-06,
|
| 632 |
+
"logits/chosen": -1.9248863458633423,
|
| 633 |
+
"logits/rejected": -1.918617844581604,
|
| 634 |
+
"logps/chosen": -32.99136734008789,
|
| 635 |
+
"logps/rejected": -32.704185485839844,
|
| 636 |
+
"loss": 0.5492,
|
| 637 |
+
"rewards/accuracies": 0.800000011920929,
|
| 638 |
+
"rewards/chosen": 0.33653250336647034,
|
| 639 |
+
"rewards/margins": 0.49465543031692505,
|
| 640 |
+
"rewards/rejected": -0.15812288224697113,
|
| 641 |
+
"step": 400
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 1.04,
|
| 645 |
+
"eval_logits/chosen": -2.1900622844696045,
|
| 646 |
+
"eval_logits/rejected": -2.185243606567383,
|
| 647 |
+
"eval_logps/chosen": -34.19782638549805,
|
| 648 |
+
"eval_logps/rejected": -37.779701232910156,
|
| 649 |
+
"eval_loss": 0.9599968194961548,
|
| 650 |
+
"eval_rewards/accuracies": 0.5564784407615662,
|
| 651 |
+
"eval_rewards/chosen": -0.06531065702438354,
|
| 652 |
+
"eval_rewards/margins": 0.03992270678281784,
|
| 653 |
+
"eval_rewards/rejected": -0.10523335635662079,
|
| 654 |
+
"eval_runtime": 145.4889,
|
| 655 |
+
"eval_samples_per_second": 2.358,
|
| 656 |
+
"eval_steps_per_second": 0.296,
|
| 657 |
+
"step": 400
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.06,
|
| 661 |
+
"grad_norm": 14.9375,
|
| 662 |
+
"learning_rate": 4.590793060955158e-06,
|
| 663 |
+
"logits/chosen": -1.9203557968139648,
|
| 664 |
+
"logits/rejected": -1.9276371002197266,
|
| 665 |
+
"logps/chosen": -28.529979705810547,
|
| 666 |
+
"logps/rejected": -29.745304107666016,
|
| 667 |
+
"loss": 0.5725,
|
| 668 |
+
"rewards/accuracies": 0.925000011920929,
|
| 669 |
+
"rewards/chosen": 0.23615682125091553,
|
| 670 |
+
"rewards/margins": 0.4949887692928314,
|
| 671 |
+
"rewards/rejected": -0.2588319778442383,
|
| 672 |
+
"step": 410
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.09,
|
| 676 |
+
"grad_norm": 17.0,
|
| 677 |
+
"learning_rate": 4.559191453574582e-06,
|
| 678 |
+
"logits/chosen": -1.9383413791656494,
|
| 679 |
+
"logits/rejected": -1.9373687505722046,
|
| 680 |
+
"logps/chosen": -33.60287094116211,
|
| 681 |
+
"logps/rejected": -31.25448989868164,
|
| 682 |
+
"loss": 0.6687,
|
| 683 |
+
"rewards/accuracies": 0.762499988079071,
|
| 684 |
+
"rewards/chosen": 0.2591347098350525,
|
| 685 |
+
"rewards/margins": 0.3837122917175293,
|
| 686 |
+
"rewards/rejected": -0.12457761913537979,
|
| 687 |
+
"step": 420
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 1.12,
|
| 691 |
+
"grad_norm": 16.875,
|
| 692 |
+
"learning_rate": 4.52653192962838e-06,
|
| 693 |
+
"logits/chosen": -1.9323232173919678,
|
| 694 |
+
"logits/rejected": -1.9150965213775635,
|
| 695 |
+
"logps/chosen": -30.429235458374023,
|
| 696 |
+
"logps/rejected": -33.68767547607422,
|
| 697 |
+
"loss": 0.5576,
|
| 698 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 699 |
+
"rewards/chosen": 0.21918778121471405,
|
| 700 |
+
"rewards/margins": 0.5237432718276978,
|
| 701 |
+
"rewards/rejected": -0.3045555651187897,
|
| 702 |
+
"step": 430
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 1.14,
|
| 706 |
+
"grad_norm": 17.5,
|
| 707 |
+
"learning_rate": 4.492831268057307e-06,
|
| 708 |
+
"logits/chosen": -1.9633067846298218,
|
| 709 |
+
"logits/rejected": -1.9654438495635986,
|
| 710 |
+
"logps/chosen": -35.656089782714844,
|
| 711 |
+
"logps/rejected": -35.58705520629883,
|
| 712 |
+
"loss": 0.4182,
|
| 713 |
+
"rewards/accuracies": 0.925000011920929,
|
| 714 |
+
"rewards/chosen": 0.3825095295906067,
|
| 715 |
+
"rewards/margins": 0.6733442544937134,
|
| 716 |
+
"rewards/rejected": -0.2908347249031067,
|
| 717 |
+
"step": 440
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 1.17,
|
| 721 |
+
"grad_norm": 16.75,
|
| 722 |
+
"learning_rate": 4.458106782690094e-06,
|
| 723 |
+
"logits/chosen": -2.039092779159546,
|
| 724 |
+
"logits/rejected": -2.03885817527771,
|
| 725 |
+
"logps/chosen": -31.8747501373291,
|
| 726 |
+
"logps/rejected": -34.06696319580078,
|
| 727 |
+
"loss": 0.5685,
|
| 728 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 729 |
+
"rewards/chosen": 0.3034403622150421,
|
| 730 |
+
"rewards/margins": 0.5061808228492737,
|
| 731 |
+
"rewards/rejected": -0.2027404010295868,
|
| 732 |
+
"step": 450
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.19,
|
| 736 |
+
"grad_norm": 18.0,
|
| 737 |
+
"learning_rate": 4.422376313348405e-06,
|
| 738 |
+
"logits/chosen": -1.9820168018341064,
|
| 739 |
+
"logits/rejected": -1.974509835243225,
|
| 740 |
+
"logps/chosen": -31.496234893798828,
|
| 741 |
+
"logps/rejected": -36.78831100463867,
|
| 742 |
+
"loss": 0.435,
|
| 743 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 744 |
+
"rewards/chosen": 0.3638893961906433,
|
| 745 |
+
"rewards/margins": 0.6575988531112671,
|
| 746 |
+
"rewards/rejected": -0.2937094271183014,
|
| 747 |
+
"step": 460
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.22,
|
| 751 |
+
"grad_norm": 14.75,
|
| 752 |
+
"learning_rate": 4.3856582166815696e-06,
|
| 753 |
+
"logits/chosen": -1.8861879110336304,
|
| 754 |
+
"logits/rejected": -1.8827335834503174,
|
| 755 |
+
"logps/chosen": -33.01527404785156,
|
| 756 |
+
"logps/rejected": -33.59056854248047,
|
| 757 |
+
"loss": 0.4878,
|
| 758 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 759 |
+
"rewards/chosen": 0.3668093681335449,
|
| 760 |
+
"rewards/margins": 0.6334856748580933,
|
| 761 |
+
"rewards/rejected": -0.26667624711990356,
|
| 762 |
+
"step": 470
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.25,
|
| 766 |
+
"grad_norm": 14.625,
|
| 767 |
+
"learning_rate": 4.347971356735789e-06,
|
| 768 |
+
"logits/chosen": -2.0107829570770264,
|
| 769 |
+
"logits/rejected": -2.003845691680908,
|
| 770 |
+
"logps/chosen": -30.23215103149414,
|
| 771 |
+
"logps/rejected": -32.65163803100586,
|
| 772 |
+
"loss": 0.5881,
|
| 773 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 774 |
+
"rewards/chosen": 0.2385544776916504,
|
| 775 |
+
"rewards/margins": 0.5018148422241211,
|
| 776 |
+
"rewards/rejected": -0.2632603049278259,
|
| 777 |
+
"step": 480
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 1.27,
|
| 781 |
+
"grad_norm": 11.625,
|
| 782 |
+
"learning_rate": 4.309335095262675e-06,
|
| 783 |
+
"logits/chosen": -1.9544703960418701,
|
| 784 |
+
"logits/rejected": -1.9560562372207642,
|
| 785 |
+
"logps/chosen": -34.39679718017578,
|
| 786 |
+
"logps/rejected": -34.34023666381836,
|
| 787 |
+
"loss": 0.4545,
|
| 788 |
+
"rewards/accuracies": 0.862500011920929,
|
| 789 |
+
"rewards/chosen": 0.38255801796913147,
|
| 790 |
+
"rewards/margins": 0.6620479822158813,
|
| 791 |
+
"rewards/rejected": -0.2794899046421051,
|
| 792 |
+
"step": 490
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 1.3,
|
| 796 |
+
"grad_norm": 16.75,
|
| 797 |
+
"learning_rate": 4.269769281772082e-06,
|
| 798 |
+
"logits/chosen": -1.8426096439361572,
|
| 799 |
+
"logits/rejected": -1.840172529220581,
|
| 800 |
+
"logps/chosen": -32.309085845947266,
|
| 801 |
+
"logps/rejected": -37.37702178955078,
|
| 802 |
+
"loss": 0.4504,
|
| 803 |
+
"rewards/accuracies": 0.862500011920929,
|
| 804 |
+
"rewards/chosen": 0.345283180475235,
|
| 805 |
+
"rewards/margins": 0.7103637456893921,
|
| 806 |
+
"rewards/rejected": -0.3650805950164795,
|
| 807 |
+
"step": 500
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 1.3,
|
| 811 |
+
"eval_logits/chosen": -2.1818087100982666,
|
| 812 |
+
"eval_logits/rejected": -2.1769938468933105,
|
| 813 |
+
"eval_logps/chosen": -34.38530349731445,
|
| 814 |
+
"eval_logps/rejected": -37.95327377319336,
|
| 815 |
+
"eval_loss": 0.9655585289001465,
|
| 816 |
+
"eval_rewards/accuracies": 0.5685215592384338,
|
| 817 |
+
"eval_rewards/chosen": -0.14030161499977112,
|
| 818 |
+
"eval_rewards/margins": 0.03436039760708809,
|
| 819 |
+
"eval_rewards/rejected": -0.1746620088815689,
|
| 820 |
+
"eval_runtime": 145.206,
|
| 821 |
+
"eval_samples_per_second": 2.362,
|
| 822 |
+
"eval_steps_per_second": 0.296,
|
| 823 |
+
"step": 500
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.32,
|
| 827 |
+
"grad_norm": 13.4375,
|
| 828 |
+
"learning_rate": 4.22929424333435e-06,
|
| 829 |
+
"logits/chosen": -1.9420890808105469,
|
| 830 |
+
"logits/rejected": -1.9468368291854858,
|
| 831 |
+
"logps/chosen": -32.48661804199219,
|
| 832 |
+
"logps/rejected": -32.22886276245117,
|
| 833 |
+
"loss": 0.4787,
|
| 834 |
+
"rewards/accuracies": 0.824999988079071,
|
| 835 |
+
"rewards/chosen": 0.38829392194747925,
|
| 836 |
+
"rewards/margins": 0.6213949918746948,
|
| 837 |
+
"rewards/rejected": -0.23310112953186035,
|
| 838 |
+
"step": 510
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.35,
|
| 842 |
+
"grad_norm": 14.8125,
|
| 843 |
+
"learning_rate": 4.1879307741372085e-06,
|
| 844 |
+
"logits/chosen": -1.9701076745986938,
|
| 845 |
+
"logits/rejected": -1.981092095375061,
|
| 846 |
+
"logps/chosen": -30.474559783935547,
|
| 847 |
+
"logps/rejected": -32.24431610107422,
|
| 848 |
+
"loss": 0.4762,
|
| 849 |
+
"rewards/accuracies": 0.887499988079071,
|
| 850 |
+
"rewards/chosen": 0.36634185910224915,
|
| 851 |
+
"rewards/margins": 0.6609948873519897,
|
| 852 |
+
"rewards/rejected": -0.2946530282497406,
|
| 853 |
+
"step": 520
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.38,
|
| 857 |
+
"grad_norm": 10.4375,
|
| 858 |
+
"learning_rate": 4.145700124802693e-06,
|
| 859 |
+
"logits/chosen": -1.897570252418518,
|
| 860 |
+
"logits/rejected": -1.8943248987197876,
|
| 861 |
+
"logps/chosen": -31.66860580444336,
|
| 862 |
+
"logps/rejected": -33.14496994018555,
|
| 863 |
+
"loss": 0.4876,
|
| 864 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 865 |
+
"rewards/chosen": 0.30076172947883606,
|
| 866 |
+
"rewards/margins": 0.6136397123336792,
|
| 867 |
+
"rewards/rejected": -0.3128780126571655,
|
| 868 |
+
"step": 530
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 1.4,
|
| 872 |
+
"grad_norm": 14.0,
|
| 873 |
+
"learning_rate": 4.102623991469562e-06,
|
| 874 |
+
"logits/chosen": -1.7630268335342407,
|
| 875 |
+
"logits/rejected": -1.7722015380859375,
|
| 876 |
+
"logps/chosen": -31.767126083374023,
|
| 877 |
+
"logps/rejected": -32.539676666259766,
|
| 878 |
+
"loss": 0.5367,
|
| 879 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 880 |
+
"rewards/chosen": 0.3209052085876465,
|
| 881 |
+
"rewards/margins": 0.5943812727928162,
|
| 882 |
+
"rewards/rejected": -0.27347609400749207,
|
| 883 |
+
"step": 540
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 1.43,
|
| 887 |
+
"grad_norm": 13.9375,
|
| 888 |
+
"learning_rate": 4.058724504646834e-06,
|
| 889 |
+
"logits/chosen": -1.856566071510315,
|
| 890 |
+
"logits/rejected": -1.8504016399383545,
|
| 891 |
+
"logps/chosen": -32.787315368652344,
|
| 892 |
+
"logps/rejected": -31.569156646728516,
|
| 893 |
+
"loss": 0.5202,
|
| 894 |
+
"rewards/accuracies": 0.800000011920929,
|
| 895 |
+
"rewards/chosen": 0.36839646100997925,
|
| 896 |
+
"rewards/margins": 0.5995180606842041,
|
| 897 |
+
"rewards/rejected": -0.23112154006958008,
|
| 898 |
+
"step": 550
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 1.45,
|
| 902 |
+
"grad_norm": 13.5625,
|
| 903 |
+
"learning_rate": 4.014024217844167e-06,
|
| 904 |
+
"logits/chosen": -1.943418264389038,
|
| 905 |
+
"logits/rejected": -1.941474199295044,
|
| 906 |
+
"logps/chosen": -33.62010192871094,
|
| 907 |
+
"logps/rejected": -32.09187316894531,
|
| 908 |
+
"loss": 0.5174,
|
| 909 |
+
"rewards/accuracies": 0.824999988079071,
|
| 910 |
+
"rewards/chosen": 0.3804393410682678,
|
| 911 |
+
"rewards/margins": 0.6182351112365723,
|
| 912 |
+
"rewards/rejected": -0.23779578506946564,
|
| 913 |
+
"step": 560
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 1.48,
|
| 917 |
+
"grad_norm": 13.9375,
|
| 918 |
+
"learning_rate": 3.968546095984911e-06,
|
| 919 |
+
"logits/chosen": -1.778063416481018,
|
| 920 |
+
"logits/rejected": -1.7760202884674072,
|
| 921 |
+
"logps/chosen": -31.914138793945312,
|
| 922 |
+
"logps/rejected": -31.619396209716797,
|
| 923 |
+
"loss": 0.5162,
|
| 924 |
+
"rewards/accuracies": 0.824999988079071,
|
| 925 |
+
"rewards/chosen": 0.352909654378891,
|
| 926 |
+
"rewards/margins": 0.5997473001480103,
|
| 927 |
+
"rewards/rejected": -0.24683766067028046,
|
| 928 |
+
"step": 570
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 1.51,
|
| 932 |
+
"grad_norm": 18.25,
|
| 933 |
+
"learning_rate": 3.922313503607806e-06,
|
| 934 |
+
"logits/chosen": -1.9141117334365845,
|
| 935 |
+
"logits/rejected": -1.910735845565796,
|
| 936 |
+
"logps/chosen": -30.1756591796875,
|
| 937 |
+
"logps/rejected": -35.52814865112305,
|
| 938 |
+
"loss": 0.441,
|
| 939 |
+
"rewards/accuracies": 0.887499988079071,
|
| 940 |
+
"rewards/chosen": 0.37159496545791626,
|
| 941 |
+
"rewards/margins": 0.7323898077011108,
|
| 942 |
+
"rewards/rejected": -0.36079490184783936,
|
| 943 |
+
"step": 580
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 1.53,
|
| 947 |
+
"grad_norm": 12.5,
|
| 948 |
+
"learning_rate": 3.875350192863368e-06,
|
| 949 |
+
"logits/chosen": -1.8503096103668213,
|
| 950 |
+
"logits/rejected": -1.8538156747817993,
|
| 951 |
+
"logps/chosen": -28.902246475219727,
|
| 952 |
+
"logps/rejected": -31.208913803100586,
|
| 953 |
+
"loss": 0.5458,
|
| 954 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 955 |
+
"rewards/chosen": 0.2851153612136841,
|
| 956 |
+
"rewards/margins": 0.5051326751708984,
|
| 957 |
+
"rewards/rejected": -0.22001728415489197,
|
| 958 |
+
"step": 590
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 1.56,
|
| 962 |
+
"grad_norm": 14.6875,
|
| 963 |
+
"learning_rate": 3.8276802913111436e-06,
|
| 964 |
+
"logits/chosen": -1.8863645792007446,
|
| 965 |
+
"logits/rejected": -1.8863203525543213,
|
| 966 |
+
"logps/chosen": -31.125957489013672,
|
| 967 |
+
"logps/rejected": -31.79779624938965,
|
| 968 |
+
"loss": 0.4511,
|
| 969 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 970 |
+
"rewards/chosen": 0.4382299780845642,
|
| 971 |
+
"rewards/margins": 0.6801418662071228,
|
| 972 |
+
"rewards/rejected": -0.2419118881225586,
|
| 973 |
+
"step": 600
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 1.56,
|
| 977 |
+
"eval_logits/chosen": -2.161299228668213,
|
| 978 |
+
"eval_logits/rejected": -2.1565115451812744,
|
| 979 |
+
"eval_logps/chosen": -34.24331283569336,
|
| 980 |
+
"eval_logps/rejected": -37.89296340942383,
|
| 981 |
+
"eval_loss": 0.9337846636772156,
|
| 982 |
+
"eval_rewards/accuracies": 0.595099687576294,
|
| 983 |
+
"eval_rewards/chosen": -0.08350436389446259,
|
| 984 |
+
"eval_rewards/margins": 0.06703372299671173,
|
| 985 |
+
"eval_rewards/rejected": -0.1505381017923355,
|
| 986 |
+
"eval_runtime": 145.1637,
|
| 987 |
+
"eval_samples_per_second": 2.363,
|
| 988 |
+
"eval_steps_per_second": 0.296,
|
| 989 |
+
"step": 600
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 1.58,
|
| 993 |
+
"grad_norm": 18.625,
|
| 994 |
+
"learning_rate": 3.7793282895240927e-06,
|
| 995 |
+
"logits/chosen": -1.9492238759994507,
|
| 996 |
+
"logits/rejected": -1.950005292892456,
|
| 997 |
+
"logps/chosen": -33.86152648925781,
|
| 998 |
+
"logps/rejected": -33.876121520996094,
|
| 999 |
+
"loss": 0.4213,
|
| 1000 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1001 |
+
"rewards/chosen": 0.4674473702907562,
|
| 1002 |
+
"rewards/margins": 0.7842208743095398,
|
| 1003 |
+
"rewards/rejected": -0.3167734146118164,
|
| 1004 |
+
"step": 610
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 1.61,
|
| 1008 |
+
"grad_norm": 16.25,
|
| 1009 |
+
"learning_rate": 3.730319028506478e-06,
|
| 1010 |
+
"logits/chosen": -1.9146831035614014,
|
| 1011 |
+
"logits/rejected": -1.912245750427246,
|
| 1012 |
+
"logps/chosen": -32.04179000854492,
|
| 1013 |
+
"logps/rejected": -32.73381042480469,
|
| 1014 |
+
"loss": 0.4604,
|
| 1015 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1016 |
+
"rewards/chosen": 0.4334433972835541,
|
| 1017 |
+
"rewards/margins": 0.6990013122558594,
|
| 1018 |
+
"rewards/rejected": -0.2655579149723053,
|
| 1019 |
+
"step": 620
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 1.64,
|
| 1023 |
+
"grad_norm": 21.25,
|
| 1024 |
+
"learning_rate": 3.6806776869317074e-06,
|
| 1025 |
+
"logits/chosen": -1.9271072149276733,
|
| 1026 |
+
"logits/rejected": -1.9181263446807861,
|
| 1027 |
+
"logps/chosen": -31.6562557220459,
|
| 1028 |
+
"logps/rejected": -31.68838119506836,
|
| 1029 |
+
"loss": 0.4662,
|
| 1030 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1031 |
+
"rewards/chosen": 0.44794517755508423,
|
| 1032 |
+
"rewards/margins": 0.7491516470909119,
|
| 1033 |
+
"rewards/rejected": -0.30120649933815,
|
| 1034 |
+
"step": 630
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 1.66,
|
| 1038 |
+
"grad_norm": 19.25,
|
| 1039 |
+
"learning_rate": 3.6304297682067146e-06,
|
| 1040 |
+
"logits/chosen": -1.928035020828247,
|
| 1041 |
+
"logits/rejected": -1.9247719049453735,
|
| 1042 |
+
"logps/chosen": -31.21347427368164,
|
| 1043 |
+
"logps/rejected": -33.109100341796875,
|
| 1044 |
+
"loss": 0.4472,
|
| 1045 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1046 |
+
"rewards/chosen": 0.3939800262451172,
|
| 1047 |
+
"rewards/margins": 0.669854462146759,
|
| 1048 |
+
"rewards/rejected": -0.27587443590164185,
|
| 1049 |
+
"step": 640
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 1.69,
|
| 1053 |
+
"grad_norm": 17.125,
|
| 1054 |
+
"learning_rate": 3.579601087369492e-06,
|
| 1055 |
+
"logits/chosen": -1.936981201171875,
|
| 1056 |
+
"logits/rejected": -1.9392879009246826,
|
| 1057 |
+
"logps/chosen": -32.55559158325195,
|
| 1058 |
+
"logps/rejected": -34.394744873046875,
|
| 1059 |
+
"loss": 0.3874,
|
| 1060 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1061 |
+
"rewards/chosen": 0.4392463266849518,
|
| 1062 |
+
"rewards/margins": 0.7237855792045593,
|
| 1063 |
+
"rewards/rejected": -0.28453919291496277,
|
| 1064 |
+
"step": 650
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 1.71,
|
| 1068 |
+
"grad_norm": 22.75,
|
| 1069 |
+
"learning_rate": 3.5282177578265295e-06,
|
| 1070 |
+
"logits/chosen": -1.848418951034546,
|
| 1071 |
+
"logits/rejected": -1.8488578796386719,
|
| 1072 |
+
"logps/chosen": -32.838523864746094,
|
| 1073 |
+
"logps/rejected": -32.17537307739258,
|
| 1074 |
+
"loss": 0.4407,
|
| 1075 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1076 |
+
"rewards/chosen": 0.44332486391067505,
|
| 1077 |
+
"rewards/margins": 0.6705711483955383,
|
| 1078 |
+
"rewards/rejected": -0.22724632918834686,
|
| 1079 |
+
"step": 660
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 1.74,
|
| 1083 |
+
"grad_norm": 21.0,
|
| 1084 |
+
"learning_rate": 3.476306177936961e-06,
|
| 1085 |
+
"logits/chosen": -1.8874480724334717,
|
| 1086 |
+
"logits/rejected": -1.8778259754180908,
|
| 1087 |
+
"logps/chosen": -32.84857940673828,
|
| 1088 |
+
"logps/rejected": -33.05742645263672,
|
| 1089 |
+
"loss": 0.382,
|
| 1090 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1091 |
+
"rewards/chosen": 0.4610818326473236,
|
| 1092 |
+
"rewards/margins": 0.7609356641769409,
|
| 1093 |
+
"rewards/rejected": -0.29985374212265015,
|
| 1094 |
+
"step": 670
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 1.77,
|
| 1098 |
+
"grad_norm": 19.0,
|
| 1099 |
+
"learning_rate": 3.423893017450324e-06,
|
| 1100 |
+
"logits/chosen": -1.7845916748046875,
|
| 1101 |
+
"logits/rejected": -1.7813985347747803,
|
| 1102 |
+
"logps/chosen": -30.31686782836914,
|
| 1103 |
+
"logps/rejected": -35.026180267333984,
|
| 1104 |
+
"loss": 0.3954,
|
| 1105 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1106 |
+
"rewards/chosen": 0.45763739943504333,
|
| 1107 |
+
"rewards/margins": 0.7406758069992065,
|
| 1108 |
+
"rewards/rejected": -0.2830384373664856,
|
| 1109 |
+
"step": 680
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 1.79,
|
| 1113 |
+
"grad_norm": 14.75,
|
| 1114 |
+
"learning_rate": 3.3710052038048794e-06,
|
| 1115 |
+
"logits/chosen": -1.8450958728790283,
|
| 1116 |
+
"logits/rejected": -1.8443653583526611,
|
| 1117 |
+
"logps/chosen": -33.978050231933594,
|
| 1118 |
+
"logps/rejected": -36.23003387451172,
|
| 1119 |
+
"loss": 0.3186,
|
| 1120 |
+
"rewards/accuracies": 0.875,
|
| 1121 |
+
"rewards/chosen": 0.5168943405151367,
|
| 1122 |
+
"rewards/margins": 0.8673771619796753,
|
| 1123 |
+
"rewards/rejected": -0.35048285126686096,
|
| 1124 |
+
"step": 690
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 1.82,
|
| 1128 |
+
"grad_norm": 11.9375,
|
| 1129 |
+
"learning_rate": 3.3176699082935546e-06,
|
| 1130 |
+
"logits/chosen": -1.8192574977874756,
|
| 1131 |
+
"logits/rejected": -1.822011947631836,
|
| 1132 |
+
"logps/chosen": -31.353174209594727,
|
| 1133 |
+
"logps/rejected": -36.575157165527344,
|
| 1134 |
+
"loss": 0.3805,
|
| 1135 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1136 |
+
"rewards/chosen": 0.4998700022697449,
|
| 1137 |
+
"rewards/margins": 0.873264491558075,
|
| 1138 |
+
"rewards/rejected": -0.37339454889297485,
|
| 1139 |
+
"step": 700
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 1.82,
|
| 1143 |
+
"eval_logits/chosen": -2.1489837169647217,
|
| 1144 |
+
"eval_logits/rejected": -2.1442174911499023,
|
| 1145 |
+
"eval_logps/chosen": -34.29972457885742,
|
| 1146 |
+
"eval_logps/rejected": -37.9394416809082,
|
| 1147 |
+
"eval_loss": 0.9385247826576233,
|
| 1148 |
+
"eval_rewards/accuracies": 0.5510797500610352,
|
| 1149 |
+
"eval_rewards/chosen": -0.10606933385133743,
|
| 1150 |
+
"eval_rewards/margins": 0.06305938214063644,
|
| 1151 |
+
"eval_rewards/rejected": -0.16912870109081268,
|
| 1152 |
+
"eval_runtime": 145.1991,
|
| 1153 |
+
"eval_samples_per_second": 2.362,
|
| 1154 |
+
"eval_steps_per_second": 0.296,
|
| 1155 |
+
"step": 700
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 1.84,
|
| 1159 |
+
"grad_norm": 16.0,
|
| 1160 |
+
"learning_rate": 3.2639145321045933e-06,
|
| 1161 |
+
"logits/chosen": -1.9265644550323486,
|
| 1162 |
+
"logits/rejected": -1.9291988611221313,
|
| 1163 |
+
"logps/chosen": -33.78008270263672,
|
| 1164 |
+
"logps/rejected": -34.95719909667969,
|
| 1165 |
+
"loss": 0.4644,
|
| 1166 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1167 |
+
"rewards/chosen": 0.44620281457901,
|
| 1168 |
+
"rewards/margins": 0.7210227847099304,
|
| 1169 |
+
"rewards/rejected": -0.2748199999332428,
|
| 1170 |
+
"step": 710
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 1.87,
|
| 1174 |
+
"grad_norm": 20.875,
|
| 1175 |
+
"learning_rate": 3.2097666922441107e-06,
|
| 1176 |
+
"logits/chosen": -1.7806246280670166,
|
| 1177 |
+
"logits/rejected": -1.7747358083724976,
|
| 1178 |
+
"logps/chosen": -33.649070739746094,
|
| 1179 |
+
"logps/rejected": -33.15039825439453,
|
| 1180 |
+
"loss": 0.4038,
|
| 1181 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1182 |
+
"rewards/chosen": 0.4736657738685608,
|
| 1183 |
+
"rewards/margins": 0.7930890917778015,
|
| 1184 |
+
"rewards/rejected": -0.3194233477115631,
|
| 1185 |
+
"step": 720
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.9,
|
| 1189 |
+
"grad_norm": 12.9375,
|
| 1190 |
+
"learning_rate": 3.1552542073477554e-06,
|
| 1191 |
+
"logits/chosen": -1.9554284811019897,
|
| 1192 |
+
"logits/rejected": -1.9523767232894897,
|
| 1193 |
+
"logps/chosen": -29.604736328125,
|
| 1194 |
+
"logps/rejected": -32.440574645996094,
|
| 1195 |
+
"loss": 0.3749,
|
| 1196 |
+
"rewards/accuracies": 0.9375,
|
| 1197 |
+
"rewards/chosen": 0.4192999005317688,
|
| 1198 |
+
"rewards/margins": 0.8162514567375183,
|
| 1199 |
+
"rewards/rejected": -0.39695149660110474,
|
| 1200 |
+
"step": 730
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 1.92,
|
| 1204 |
+
"grad_norm": 16.125,
|
| 1205 |
+
"learning_rate": 3.100405083388799e-06,
|
| 1206 |
+
"logits/chosen": -1.7934890985488892,
|
| 1207 |
+
"logits/rejected": -1.7935874462127686,
|
| 1208 |
+
"logps/chosen": -32.12221145629883,
|
| 1209 |
+
"logps/rejected": -38.396541595458984,
|
| 1210 |
+
"loss": 0.3706,
|
| 1211 |
+
"rewards/accuracies": 0.875,
|
| 1212 |
+
"rewards/chosen": 0.5363865494728088,
|
| 1213 |
+
"rewards/margins": 0.8438798785209656,
|
| 1214 |
+
"rewards/rejected": -0.3074934482574463,
|
| 1215 |
+
"step": 740
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 1.95,
|
| 1219 |
+
"grad_norm": 9.125,
|
| 1220 |
+
"learning_rate": 3.0452474992899645e-06,
|
| 1221 |
+
"logits/chosen": -1.6800241470336914,
|
| 1222 |
+
"logits/rejected": -1.685310959815979,
|
| 1223 |
+
"logps/chosen": -35.7508659362793,
|
| 1224 |
+
"logps/rejected": -34.849327087402344,
|
| 1225 |
+
"loss": 0.4332,
|
| 1226 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1227 |
+
"rewards/chosen": 0.47263655066490173,
|
| 1228 |
+
"rewards/margins": 0.8234331011772156,
|
| 1229 |
+
"rewards/rejected": -0.35079658031463623,
|
| 1230 |
+
"step": 750
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 1.97,
|
| 1234 |
+
"grad_norm": 11.0,
|
| 1235 |
+
"learning_rate": 2.989809792446417e-06,
|
| 1236 |
+
"logits/chosen": -1.8796085119247437,
|
| 1237 |
+
"logits/rejected": -1.8812284469604492,
|
| 1238 |
+
"logps/chosen": -31.44708251953125,
|
| 1239 |
+
"logps/rejected": -33.469627380371094,
|
| 1240 |
+
"loss": 0.3873,
|
| 1241 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1242 |
+
"rewards/chosen": 0.46344003081321716,
|
| 1243 |
+
"rewards/margins": 0.7396036982536316,
|
| 1244 |
+
"rewards/rejected": -0.27616366744041443,
|
| 1245 |
+
"step": 760
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 2.0,
|
| 1249 |
+
"grad_norm": 20.25,
|
| 1250 |
+
"learning_rate": 2.9341204441673267e-06,
|
| 1251 |
+
"logits/chosen": -1.8549083471298218,
|
| 1252 |
+
"logits/rejected": -1.8539108037948608,
|
| 1253 |
+
"logps/chosen": -30.935928344726562,
|
| 1254 |
+
"logps/rejected": -35.55149841308594,
|
| 1255 |
+
"loss": 0.4385,
|
| 1256 |
+
"rewards/accuracies": 0.8708333969116211,
|
| 1257 |
+
"rewards/chosen": 0.4958726763725281,
|
| 1258 |
+
"rewards/margins": 0.7052101492881775,
|
| 1259 |
+
"rewards/rejected": -0.20933744311332703,
|
| 1260 |
+
"step": 770
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 2.03,
|
| 1264 |
+
"grad_norm": 12.8125,
|
| 1265 |
+
"learning_rate": 2.878208065043501e-06,
|
| 1266 |
+
"logits/chosen": -1.8549137115478516,
|
| 1267 |
+
"logits/rejected": -1.8542522192001343,
|
| 1268 |
+
"logps/chosen": -33.281410217285156,
|
| 1269 |
+
"logps/rejected": -32.49152755737305,
|
| 1270 |
+
"loss": 0.2642,
|
| 1271 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1272 |
+
"rewards/chosen": 0.546822190284729,
|
| 1273 |
+
"rewards/margins": 0.9756248593330383,
|
| 1274 |
+
"rewards/rejected": -0.4288027286529541,
|
| 1275 |
+
"step": 780
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 2.05,
|
| 1279 |
+
"grad_norm": 13.0,
|
| 1280 |
+
"learning_rate": 2.8221013802485974e-06,
|
| 1281 |
+
"logits/chosen": -1.8834869861602783,
|
| 1282 |
+
"logits/rejected": -1.8820085525512695,
|
| 1283 |
+
"logps/chosen": -28.202373504638672,
|
| 1284 |
+
"logps/rejected": -33.88237380981445,
|
| 1285 |
+
"loss": 0.1975,
|
| 1286 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1287 |
+
"rewards/chosen": 0.5443224906921387,
|
| 1288 |
+
"rewards/margins": 1.0427463054656982,
|
| 1289 |
+
"rewards/rejected": -0.4984237253665924,
|
| 1290 |
+
"step": 790
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 2.08,
|
| 1294 |
+
"grad_norm": 0.006591796875,
|
| 1295 |
+
"learning_rate": 2.76582921478147e-06,
|
| 1296 |
+
"logits/chosen": -1.9375505447387695,
|
| 1297 |
+
"logits/rejected": -1.9341075420379639,
|
| 1298 |
+
"logps/chosen": -30.97635269165039,
|
| 1299 |
+
"logps/rejected": -35.407142639160156,
|
| 1300 |
+
"loss": 0.2038,
|
| 1301 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1302 |
+
"rewards/chosen": 0.5803438425064087,
|
| 1303 |
+
"rewards/margins": 1.0635874271392822,
|
| 1304 |
+
"rewards/rejected": -0.4832436144351959,
|
| 1305 |
+
"step": 800
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 2.08,
|
| 1309 |
+
"eval_logits/chosen": -2.1414618492126465,
|
| 1310 |
+
"eval_logits/rejected": -2.1367087364196777,
|
| 1311 |
+
"eval_logps/chosen": -34.35294723510742,
|
| 1312 |
+
"eval_logps/rejected": -38.00876235961914,
|
| 1313 |
+
"eval_loss": 0.9334202408790588,
|
| 1314 |
+
"eval_rewards/accuracies": 0.5689368844032288,
|
| 1315 |
+
"eval_rewards/chosen": -0.12735822796821594,
|
| 1316 |
+
"eval_rewards/margins": 0.06950045377016068,
|
| 1317 |
+
"eval_rewards/rejected": -0.19685867428779602,
|
| 1318 |
+
"eval_runtime": 145.1666,
|
| 1319 |
+
"eval_samples_per_second": 2.363,
|
| 1320 |
+
"eval_steps_per_second": 0.296,
|
| 1321 |
+
"step": 800
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 2.1,
|
| 1325 |
+
"grad_norm": 11.0,
|
| 1326 |
+
"learning_rate": 2.7094204786572254e-06,
|
| 1327 |
+
"logits/chosen": -1.7686500549316406,
|
| 1328 |
+
"logits/rejected": -1.7609798908233643,
|
| 1329 |
+
"logps/chosen": -32.71022033691406,
|
| 1330 |
+
"logps/rejected": -35.735992431640625,
|
| 1331 |
+
"loss": 0.1833,
|
| 1332 |
+
"rewards/accuracies": 0.9375,
|
| 1333 |
+
"rewards/chosen": 0.6984082460403442,
|
| 1334 |
+
"rewards/margins": 1.1687581539154053,
|
| 1335 |
+
"rewards/rejected": -0.4703500270843506,
|
| 1336 |
+
"step": 810
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 2.13,
|
| 1340 |
+
"grad_norm": 14.125,
|
| 1341 |
+
"learning_rate": 2.6529041520546072e-06,
|
| 1342 |
+
"logits/chosen": -1.8466291427612305,
|
| 1343 |
+
"logits/rejected": -1.8570162057876587,
|
| 1344 |
+
"logps/chosen": -34.17173385620117,
|
| 1345 |
+
"logps/rejected": -33.32606506347656,
|
| 1346 |
+
"loss": 0.28,
|
| 1347 |
+
"rewards/accuracies": 0.9375,
|
| 1348 |
+
"rewards/chosen": 0.6095821261405945,
|
| 1349 |
+
"rewards/margins": 0.9674333333969116,
|
| 1350 |
+
"rewards/rejected": -0.35785120725631714,
|
| 1351 |
+
"step": 820
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 2.16,
|
| 1355 |
+
"grad_norm": 16.5,
|
| 1356 |
+
"learning_rate": 2.5963092704273302e-06,
|
| 1357 |
+
"logits/chosen": -1.8971784114837646,
|
| 1358 |
+
"logits/rejected": -1.9018369913101196,
|
| 1359 |
+
"logps/chosen": -33.51736831665039,
|
| 1360 |
+
"logps/rejected": -30.209964752197266,
|
| 1361 |
+
"loss": 0.2911,
|
| 1362 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1363 |
+
"rewards/chosen": 0.5541660189628601,
|
| 1364 |
+
"rewards/margins": 0.9308546781539917,
|
| 1365 |
+
"rewards/rejected": -0.3766886591911316,
|
| 1366 |
+
"step": 830
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 2.18,
|
| 1370 |
+
"grad_norm": 12.9375,
|
| 1371 |
+
"learning_rate": 2.53966490958702e-06,
|
| 1372 |
+
"logits/chosen": -1.8965809345245361,
|
| 1373 |
+
"logits/rejected": -1.9047677516937256,
|
| 1374 |
+
"logps/chosen": -33.255287170410156,
|
| 1375 |
+
"logps/rejected": -30.980331420898438,
|
| 1376 |
+
"loss": 0.2265,
|
| 1377 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1378 |
+
"rewards/chosen": 0.6130979061126709,
|
| 1379 |
+
"rewards/margins": 1.0661498308181763,
|
| 1380 |
+
"rewards/rejected": -0.4530518651008606,
|
| 1381 |
+
"step": 840
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 2.21,
|
| 1385 |
+
"grad_norm": 13.375,
|
| 1386 |
+
"learning_rate": 2.4830001707654135e-06,
|
| 1387 |
+
"logits/chosen": -1.8259570598602295,
|
| 1388 |
+
"logits/rejected": -1.8164736032485962,
|
| 1389 |
+
"logps/chosen": -30.39957618713379,
|
| 1390 |
+
"logps/rejected": -33.05730056762695,
|
| 1391 |
+
"loss": 0.2609,
|
| 1392 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1393 |
+
"rewards/chosen": 0.5574477314949036,
|
| 1394 |
+
"rewards/margins": 0.9748461842536926,
|
| 1395 |
+
"rewards/rejected": -0.4173983633518219,
|
| 1396 |
+
"step": 850
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 2.23,
|
| 1400 |
+
"grad_norm": 11.875,
|
| 1401 |
+
"learning_rate": 2.4263441656635054e-06,
|
| 1402 |
+
"logits/chosen": -1.9653680324554443,
|
| 1403 |
+
"logits/rejected": -1.9555402994155884,
|
| 1404 |
+
"logps/chosen": -24.850261688232422,
|
| 1405 |
+
"logps/rejected": -31.08827781677246,
|
| 1406 |
+
"loss": 0.2561,
|
| 1407 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1408 |
+
"rewards/chosen": 0.5495240688323975,
|
| 1409 |
+
"rewards/margins": 1.035900354385376,
|
| 1410 |
+
"rewards/rejected": -0.48637622594833374,
|
| 1411 |
+
"step": 860
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 2.26,
|
| 1415 |
+
"grad_norm": 13.1875,
|
| 1416 |
+
"learning_rate": 2.3697260014953107e-06,
|
| 1417 |
+
"logits/chosen": -1.8186180591583252,
|
| 1418 |
+
"logits/rejected": -1.819954514503479,
|
| 1419 |
+
"logps/chosen": -32.77277374267578,
|
| 1420 |
+
"logps/rejected": -31.007610321044922,
|
| 1421 |
+
"loss": 0.218,
|
| 1422 |
+
"rewards/accuracies": 0.9375,
|
| 1423 |
+
"rewards/chosen": 0.6240935325622559,
|
| 1424 |
+
"rewards/margins": 1.0341370105743408,
|
| 1425 |
+
"rewards/rejected": -0.41004353761672974,
|
| 1426 |
+
"step": 870
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 2.29,
|
| 1430 |
+
"grad_norm": 12.25,
|
| 1431 |
+
"learning_rate": 2.3131747660339396e-06,
|
| 1432 |
+
"logits/chosen": -1.8255703449249268,
|
| 1433 |
+
"logits/rejected": -1.826511025428772,
|
| 1434 |
+
"logps/chosen": -31.154870986938477,
|
| 1435 |
+
"logps/rejected": -34.21940994262695,
|
| 1436 |
+
"loss": 0.3146,
|
| 1437 |
+
"rewards/accuracies": 0.875,
|
| 1438 |
+
"rewards/chosen": 0.4937514662742615,
|
| 1439 |
+
"rewards/margins": 0.9611461758613586,
|
| 1440 |
+
"rewards/rejected": -0.4673948287963867,
|
| 1441 |
+
"step": 880
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 2.31,
|
| 1445 |
+
"grad_norm": 11.6875,
|
| 1446 |
+
"learning_rate": 2.256719512667651e-06,
|
| 1447 |
+
"logits/chosen": -1.7308677434921265,
|
| 1448 |
+
"logits/rejected": -1.7293297052383423,
|
| 1449 |
+
"logps/chosen": -34.149600982666016,
|
| 1450 |
+
"logps/rejected": -37.341552734375,
|
| 1451 |
+
"loss": 0.2179,
|
| 1452 |
+
"rewards/accuracies": 0.9375,
|
| 1453 |
+
"rewards/chosen": 0.6061262488365173,
|
| 1454 |
+
"rewards/margins": 1.1918519735336304,
|
| 1455 |
+
"rewards/rejected": -0.5857258439064026,
|
| 1456 |
+
"step": 890
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 2.34,
|
| 1460 |
+
"grad_norm": 8.375,
|
| 1461 |
+
"learning_rate": 2.2003892454735786e-06,
|
| 1462 |
+
"logits/chosen": -1.8834432363510132,
|
| 1463 |
+
"logits/rejected": -1.8766663074493408,
|
| 1464 |
+
"logps/chosen": -30.528228759765625,
|
| 1465 |
+
"logps/rejected": -33.975677490234375,
|
| 1466 |
+
"loss": 0.2332,
|
| 1467 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1468 |
+
"rewards/chosen": 0.5897632837295532,
|
| 1469 |
+
"rewards/margins": 1.076806664466858,
|
| 1470 |
+
"rewards/rejected": -0.4870433807373047,
|
| 1471 |
+
"step": 900
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 2.34,
|
| 1475 |
+
"eval_logits/chosen": -2.139965057373047,
|
| 1476 |
+
"eval_logits/rejected": -2.135190963745117,
|
| 1477 |
+
"eval_logps/chosen": -34.37273025512695,
|
| 1478 |
+
"eval_logps/rejected": -38.030879974365234,
|
| 1479 |
+
"eval_loss": 0.9319750070571899,
|
| 1480 |
+
"eval_rewards/accuracies": 0.5718438625335693,
|
| 1481 |
+
"eval_rewards/chosen": -0.13527235388755798,
|
| 1482 |
+
"eval_rewards/margins": 0.0704314336180687,
|
| 1483 |
+
"eval_rewards/rejected": -0.20570378005504608,
|
| 1484 |
+
"eval_runtime": 145.1066,
|
| 1485 |
+
"eval_samples_per_second": 2.364,
|
| 1486 |
+
"eval_steps_per_second": 0.296,
|
| 1487 |
+
"step": 900
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 2.36,
|
| 1491 |
+
"grad_norm": 13.6875,
|
| 1492 |
+
"learning_rate": 2.1442129043167877e-06,
|
| 1493 |
+
"logits/chosen": -1.922201156616211,
|
| 1494 |
+
"logits/rejected": -1.9175071716308594,
|
| 1495 |
+
"logps/chosen": -32.03234100341797,
|
| 1496 |
+
"logps/rejected": -35.83629608154297,
|
| 1497 |
+
"loss": 0.213,
|
| 1498 |
+
"rewards/accuracies": 0.9375,
|
| 1499 |
+
"rewards/chosen": 0.6101375818252563,
|
| 1500 |
+
"rewards/margins": 1.1332075595855713,
|
| 1501 |
+
"rewards/rejected": -0.5230700373649597,
|
| 1502 |
+
"step": 910
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 2.39,
|
| 1506 |
+
"grad_norm": 10.5625,
|
| 1507 |
+
"learning_rate": 2.088219349982323e-06,
|
| 1508 |
+
"logits/chosen": -1.8422584533691406,
|
| 1509 |
+
"logits/rejected": -1.8473327159881592,
|
| 1510 |
+
"logps/chosen": -33.719722747802734,
|
| 1511 |
+
"logps/rejected": -33.628021240234375,
|
| 1512 |
+
"loss": 0.2188,
|
| 1513 |
+
"rewards/accuracies": 0.9375,
|
| 1514 |
+
"rewards/chosen": 0.6515445709228516,
|
| 1515 |
+
"rewards/margins": 1.1027911901474,
|
| 1516 |
+
"rewards/rejected": -0.4512465000152588,
|
| 1517 |
+
"step": 920
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 2.42,
|
| 1521 |
+
"grad_norm": 9.4375,
|
| 1522 |
+
"learning_rate": 2.0324373493478803e-06,
|
| 1523 |
+
"logits/chosen": -1.9423255920410156,
|
| 1524 |
+
"logits/rejected": -1.9330682754516602,
|
| 1525 |
+
"logps/chosen": -30.859844207763672,
|
| 1526 |
+
"logps/rejected": -35.690673828125,
|
| 1527 |
+
"loss": 0.2137,
|
| 1528 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1529 |
+
"rewards/chosen": 0.5700357556343079,
|
| 1530 |
+
"rewards/margins": 1.0691864490509033,
|
| 1531 |
+
"rewards/rejected": -0.49915066361427307,
|
| 1532 |
+
"step": 930
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 2.44,
|
| 1536 |
+
"grad_norm": 15.125,
|
| 1537 |
+
"learning_rate": 1.976895560604729e-06,
|
| 1538 |
+
"logits/chosen": -1.8686740398406982,
|
| 1539 |
+
"logits/rejected": -1.8654989004135132,
|
| 1540 |
+
"logps/chosen": -30.049386978149414,
|
| 1541 |
+
"logps/rejected": -33.48214340209961,
|
| 1542 |
+
"loss": 0.321,
|
| 1543 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1544 |
+
"rewards/chosen": 0.49562782049179077,
|
| 1545 |
+
"rewards/margins": 0.9023343920707703,
|
| 1546 |
+
"rewards/rejected": -0.40670666098594666,
|
| 1547 |
+
"step": 940
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 2.47,
|
| 1551 |
+
"grad_norm": 4.75,
|
| 1552 |
+
"learning_rate": 1.921622518534466e-06,
|
| 1553 |
+
"logits/chosen": -1.8089139461517334,
|
| 1554 |
+
"logits/rejected": -1.8163315057754517,
|
| 1555 |
+
"logps/chosen": -31.528472900390625,
|
| 1556 |
+
"logps/rejected": -36.56946563720703,
|
| 1557 |
+
"loss": 0.3027,
|
| 1558 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1559 |
+
"rewards/chosen": 0.5370327234268188,
|
| 1560 |
+
"rewards/margins": 0.9448447227478027,
|
| 1561 |
+
"rewards/rejected": -0.4078119397163391,
|
| 1562 |
+
"step": 950
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 2.49,
|
| 1566 |
+
"grad_norm": 10.375,
|
| 1567 |
+
"learning_rate": 1.8666466198491794e-06,
|
| 1568 |
+
"logits/chosen": -1.8195326328277588,
|
| 1569 |
+
"logits/rejected": -1.8125743865966797,
|
| 1570 |
+
"logps/chosen": -32.26793670654297,
|
| 1571 |
+
"logps/rejected": -37.91437530517578,
|
| 1572 |
+
"loss": 0.2086,
|
| 1573 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1574 |
+
"rewards/chosen": 0.6242316961288452,
|
| 1575 |
+
"rewards/margins": 1.1504604816436768,
|
| 1576 |
+
"rewards/rejected": -0.5262287259101868,
|
| 1577 |
+
"step": 960
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 2.52,
|
| 1581 |
+
"grad_norm": 12.875,
|
| 1582 |
+
"learning_rate": 1.8119961086025376e-06,
|
| 1583 |
+
"logits/chosen": -1.8216333389282227,
|
| 1584 |
+
"logits/rejected": -1.8221938610076904,
|
| 1585 |
+
"logps/chosen": -29.730484008789062,
|
| 1586 |
+
"logps/rejected": -34.02881622314453,
|
| 1587 |
+
"loss": 0.2371,
|
| 1588 |
+
"rewards/accuracies": 0.9375,
|
| 1589 |
+
"rewards/chosen": 0.602368175983429,
|
| 1590 |
+
"rewards/margins": 1.021390438079834,
|
| 1591 |
+
"rewards/rejected": -0.41902226209640503,
|
| 1592 |
+
"step": 970
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 2.55,
|
| 1596 |
+
"grad_norm": 14.0625,
|
| 1597 |
+
"learning_rate": 1.7576990616793139e-06,
|
| 1598 |
+
"logits/chosen": -1.8246610164642334,
|
| 1599 |
+
"logits/rejected": -1.8358711004257202,
|
| 1600 |
+
"logps/chosen": -31.346548080444336,
|
| 1601 |
+
"logps/rejected": -35.217620849609375,
|
| 1602 |
+
"loss": 0.2028,
|
| 1603 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1604 |
+
"rewards/chosen": 0.6920571327209473,
|
| 1605 |
+
"rewards/margins": 1.1266204118728638,
|
| 1606 |
+
"rewards/rejected": -0.43456321954727173,
|
| 1607 |
+
"step": 980
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 2.57,
|
| 1611 |
+
"grad_norm": 5.0,
|
| 1612 |
+
"learning_rate": 1.7037833743707892e-06,
|
| 1613 |
+
"logits/chosen": -1.8958324193954468,
|
| 1614 |
+
"logits/rejected": -1.893398642539978,
|
| 1615 |
+
"logps/chosen": -34.60765838623047,
|
| 1616 |
+
"logps/rejected": -33.333709716796875,
|
| 1617 |
+
"loss": 0.3265,
|
| 1618 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1619 |
+
"rewards/chosen": 0.605356752872467,
|
| 1620 |
+
"rewards/margins": 0.9791496396064758,
|
| 1621 |
+
"rewards/rejected": -0.3737927973270416,
|
| 1622 |
+
"step": 990
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 2.6,
|
| 1626 |
+
"grad_norm": 12.6875,
|
| 1627 |
+
"learning_rate": 1.6502767460434588e-06,
|
| 1628 |
+
"logits/chosen": -1.8685047626495361,
|
| 1629 |
+
"logits/rejected": -1.8726377487182617,
|
| 1630 |
+
"logps/chosen": -33.05529022216797,
|
| 1631 |
+
"logps/rejected": -35.44232177734375,
|
| 1632 |
+
"loss": 0.28,
|
| 1633 |
+
"rewards/accuracies": 0.9375,
|
| 1634 |
+
"rewards/chosen": 0.5631993412971497,
|
| 1635 |
+
"rewards/margins": 0.9457098245620728,
|
| 1636 |
+
"rewards/rejected": -0.3825104832649231,
|
| 1637 |
+
"step": 1000
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 2.6,
|
| 1641 |
+
"eval_logits/chosen": -2.139305830001831,
|
| 1642 |
+
"eval_logits/rejected": -2.134549379348755,
|
| 1643 |
+
"eval_logps/chosen": -34.397972106933594,
|
| 1644 |
+
"eval_logps/rejected": -38.07229995727539,
|
| 1645 |
+
"eval_loss": 0.9270963668823242,
|
| 1646 |
+
"eval_rewards/accuracies": 0.5714285373687744,
|
| 1647 |
+
"eval_rewards/chosen": -0.14536714553833008,
|
| 1648 |
+
"eval_rewards/margins": 0.07690288871526718,
|
| 1649 |
+
"eval_rewards/rejected": -0.22227002680301666,
|
| 1650 |
+
"eval_runtime": 145.3536,
|
| 1651 |
+
"eval_samples_per_second": 2.36,
|
| 1652 |
+
"eval_steps_per_second": 0.296,
|
| 1653 |
+
"step": 1000
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 2.62,
|
| 1657 |
+
"grad_norm": 17.5,
|
| 1658 |
+
"learning_rate": 1.5972066659083796e-06,
|
| 1659 |
+
"logits/chosen": -1.8788774013519287,
|
| 1660 |
+
"logits/rejected": -1.8837945461273193,
|
| 1661 |
+
"logps/chosen": -31.51852035522461,
|
| 1662 |
+
"logps/rejected": -33.45092010498047,
|
| 1663 |
+
"loss": 0.3491,
|
| 1664 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1665 |
+
"rewards/chosen": 0.5127608180046082,
|
| 1666 |
+
"rewards/margins": 0.8846257328987122,
|
| 1667 |
+
"rewards/rejected": -0.3718649446964264,
|
| 1668 |
+
"step": 1010
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 2.65,
|
| 1672 |
+
"grad_norm": 13.5,
|
| 1673 |
+
"learning_rate": 1.5446003988985041e-06,
|
| 1674 |
+
"logits/chosen": -1.8375968933105469,
|
| 1675 |
+
"logits/rejected": -1.8401010036468506,
|
| 1676 |
+
"logps/chosen": -29.037750244140625,
|
| 1677 |
+
"logps/rejected": -32.918941497802734,
|
| 1678 |
+
"loss": 0.198,
|
| 1679 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1680 |
+
"rewards/chosen": 0.563154935836792,
|
| 1681 |
+
"rewards/margins": 1.0510679483413696,
|
| 1682 |
+
"rewards/rejected": -0.48791298270225525,
|
| 1683 |
+
"step": 1020
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 2.68,
|
| 1687 |
+
"grad_norm": 8.5625,
|
| 1688 |
+
"learning_rate": 1.4924849716612211e-06,
|
| 1689 |
+
"logits/chosen": -1.8911283016204834,
|
| 1690 |
+
"logits/rejected": -1.8856918811798096,
|
| 1691 |
+
"logps/chosen": -32.591796875,
|
| 1692 |
+
"logps/rejected": -34.6553955078125,
|
| 1693 |
+
"loss": 0.2597,
|
| 1694 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1695 |
+
"rewards/chosen": 0.5589276552200317,
|
| 1696 |
+
"rewards/margins": 1.0138864517211914,
|
| 1697 |
+
"rewards/rejected": -0.4549587368965149,
|
| 1698 |
+
"step": 1030
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 2.7,
|
| 1702 |
+
"grad_norm": 15.8125,
|
| 1703 |
+
"learning_rate": 1.440887158673332e-06,
|
| 1704 |
+
"logits/chosen": -1.8267465829849243,
|
| 1705 |
+
"logits/rejected": -1.8307063579559326,
|
| 1706 |
+
"logps/chosen": -34.41118240356445,
|
| 1707 |
+
"logps/rejected": -35.5274543762207,
|
| 1708 |
+
"loss": 0.252,
|
| 1709 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1710 |
+
"rewards/chosen": 0.5370186567306519,
|
| 1711 |
+
"rewards/margins": 0.9801691174507141,
|
| 1712 |
+
"rewards/rejected": -0.4431503713130951,
|
| 1713 |
+
"step": 1040
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 2.73,
|
| 1717 |
+
"grad_norm": 11.9375,
|
| 1718 |
+
"learning_rate": 1.3898334684855647e-06,
|
| 1719 |
+
"logits/chosen": -1.8240203857421875,
|
| 1720 |
+
"logits/rejected": -1.8369216918945312,
|
| 1721 |
+
"logps/chosen": -30.501928329467773,
|
| 1722 |
+
"logps/rejected": -34.156166076660156,
|
| 1723 |
+
"loss": 0.251,
|
| 1724 |
+
"rewards/accuracies": 0.9375,
|
| 1725 |
+
"rewards/chosen": 0.5443831086158752,
|
| 1726 |
+
"rewards/margins": 0.9988336563110352,
|
| 1727 |
+
"rewards/rejected": -0.4544506072998047,
|
| 1728 |
+
"step": 1050
|
| 1729 |
+
},
|
| 1730 |
+
{
|
| 1731 |
+
"epoch": 2.75,
|
| 1732 |
+
"grad_norm": 14.125,
|
| 1733 |
+
"learning_rate": 1.3393501301037245e-06,
|
| 1734 |
+
"logits/chosen": -1.814233422279358,
|
| 1735 |
+
"logits/rejected": -1.807539701461792,
|
| 1736 |
+
"logps/chosen": -30.65509605407715,
|
| 1737 |
+
"logps/rejected": -34.37706756591797,
|
| 1738 |
+
"loss": 0.2287,
|
| 1739 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1740 |
+
"rewards/chosen": 0.6016949415206909,
|
| 1741 |
+
"rewards/margins": 1.0253394842147827,
|
| 1742 |
+
"rewards/rejected": -0.4236446022987366,
|
| 1743 |
+
"step": 1060
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 2.78,
|
| 1747 |
+
"grad_norm": 16.375,
|
| 1748 |
+
"learning_rate": 1.2894630795134454e-06,
|
| 1749 |
+
"logits/chosen": -1.9399135112762451,
|
| 1750 |
+
"logits/rejected": -1.940683126449585,
|
| 1751 |
+
"logps/chosen": -32.10429763793945,
|
| 1752 |
+
"logps/rejected": -33.903560638427734,
|
| 1753 |
+
"loss": 0.2239,
|
| 1754 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1755 |
+
"rewards/chosen": 0.5771129131317139,
|
| 1756 |
+
"rewards/margins": 1.0006603002548218,
|
| 1757 |
+
"rewards/rejected": -0.4235473573207855,
|
| 1758 |
+
"step": 1070
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 2.81,
|
| 1762 |
+
"grad_norm": 12.3125,
|
| 1763 |
+
"learning_rate": 1.2401979463554984e-06,
|
| 1764 |
+
"logits/chosen": -1.8709838390350342,
|
| 1765 |
+
"logits/rejected": -1.8698453903198242,
|
| 1766 |
+
"logps/chosen": -32.744606018066406,
|
| 1767 |
+
"logps/rejected": -34.11382293701172,
|
| 1768 |
+
"loss": 0.3118,
|
| 1769 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1770 |
+
"rewards/chosen": 0.5380905866622925,
|
| 1771 |
+
"rewards/margins": 0.906886100769043,
|
| 1772 |
+
"rewards/rejected": -0.3687955439090729,
|
| 1773 |
+
"step": 1080
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 2.83,
|
| 1777 |
+
"grad_norm": 10.4375,
|
| 1778 |
+
"learning_rate": 1.1915800407584705e-06,
|
| 1779 |
+
"logits/chosen": -1.8952579498291016,
|
| 1780 |
+
"logits/rejected": -1.8877201080322266,
|
| 1781 |
+
"logps/chosen": -32.717063903808594,
|
| 1782 |
+
"logps/rejected": -31.973791122436523,
|
| 1783 |
+
"loss": 0.239,
|
| 1784 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1785 |
+
"rewards/chosen": 0.6035231947898865,
|
| 1786 |
+
"rewards/margins": 0.9764264822006226,
|
| 1787 |
+
"rewards/rejected": -0.3729034662246704,
|
| 1788 |
+
"step": 1090
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 2.86,
|
| 1792 |
+
"grad_norm": 7.875,
|
| 1793 |
+
"learning_rate": 1.1436343403356019e-06,
|
| 1794 |
+
"logits/chosen": -1.8662681579589844,
|
| 1795 |
+
"logits/rejected": -1.8652747869491577,
|
| 1796 |
+
"logps/chosen": -33.863346099853516,
|
| 1797 |
+
"logps/rejected": -37.67915344238281,
|
| 1798 |
+
"loss": 0.1953,
|
| 1799 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1800 |
+
"rewards/chosen": 0.622026801109314,
|
| 1801 |
+
"rewards/margins": 1.152605652809143,
|
| 1802 |
+
"rewards/rejected": -0.5305787324905396,
|
| 1803 |
+
"step": 1100
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 2.86,
|
| 1807 |
+
"eval_logits/chosen": -2.1389670372009277,
|
| 1808 |
+
"eval_logits/rejected": -2.134211301803589,
|
| 1809 |
+
"eval_logps/chosen": -34.42226791381836,
|
| 1810 |
+
"eval_logps/rejected": -38.063568115234375,
|
| 1811 |
+
"eval_loss": 0.9399018287658691,
|
| 1812 |
+
"eval_rewards/accuracies": 0.5631229281425476,
|
| 1813 |
+
"eval_rewards/chosen": -0.1550864279270172,
|
| 1814 |
+
"eval_rewards/margins": 0.06369131803512573,
|
| 1815 |
+
"eval_rewards/rejected": -0.21877776086330414,
|
| 1816 |
+
"eval_runtime": 145.1823,
|
| 1817 |
+
"eval_samples_per_second": 2.363,
|
| 1818 |
+
"eval_steps_per_second": 0.296,
|
| 1819 |
+
"step": 1100
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 2.88,
|
| 1823 |
+
"grad_norm": 7.40625,
|
| 1824 |
+
"learning_rate": 1.0963854773524548e-06,
|
| 1825 |
+
"logits/chosen": -1.9152100086212158,
|
| 1826 |
+
"logits/rejected": -1.9206510782241821,
|
| 1827 |
+
"logps/chosen": -34.14377975463867,
|
| 1828 |
+
"logps/rejected": -36.70298767089844,
|
| 1829 |
+
"loss": 0.2774,
|
| 1830 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1831 |
+
"rewards/chosen": 0.5293335914611816,
|
| 1832 |
+
"rewards/margins": 0.9930937886238098,
|
| 1833 |
+
"rewards/rejected": -0.4637601971626282,
|
| 1834 |
+
"step": 1110
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 2.91,
|
| 1838 |
+
"grad_norm": 8.5625,
|
| 1839 |
+
"learning_rate": 1.049857726072005e-06,
|
| 1840 |
+
"logits/chosen": -1.885117769241333,
|
| 1841 |
+
"logits/rejected": -1.8830986022949219,
|
| 1842 |
+
"logps/chosen": -31.27766990661621,
|
| 1843 |
+
"logps/rejected": -34.183876037597656,
|
| 1844 |
+
"loss": 0.2761,
|
| 1845 |
+
"rewards/accuracies": 0.9375,
|
| 1846 |
+
"rewards/chosen": 0.5354140996932983,
|
| 1847 |
+
"rewards/margins": 1.0158647298812866,
|
| 1848 |
+
"rewards/rejected": -0.4804508090019226,
|
| 1849 |
+
"step": 1120
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 2.94,
|
| 1853 |
+
"grad_norm": 10.4375,
|
| 1854 |
+
"learning_rate": 1.0040749902836508e-06,
|
| 1855 |
+
"logits/chosen": -1.7881187200546265,
|
| 1856 |
+
"logits/rejected": -1.790825605392456,
|
| 1857 |
+
"logps/chosen": -28.32158851623535,
|
| 1858 |
+
"logps/rejected": -31.530282974243164,
|
| 1859 |
+
"loss": 0.1986,
|
| 1860 |
+
"rewards/accuracies": 0.9375,
|
| 1861 |
+
"rewards/chosen": 0.5708181262016296,
|
| 1862 |
+
"rewards/margins": 1.0849167108535767,
|
| 1863 |
+
"rewards/rejected": -0.514098584651947,
|
| 1864 |
+
"step": 1130
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 2.96,
|
| 1868 |
+
"grad_norm": 13.9375,
|
| 1869 |
+
"learning_rate": 9.59060791022566e-07,
|
| 1870 |
+
"logits/chosen": -1.886063814163208,
|
| 1871 |
+
"logits/rejected": -1.882871389389038,
|
| 1872 |
+
"logps/chosen": -31.68621826171875,
|
| 1873 |
+
"logps/rejected": -33.42014694213867,
|
| 1874 |
+
"loss": 0.249,
|
| 1875 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1876 |
+
"rewards/chosen": 0.6070117354393005,
|
| 1877 |
+
"rewards/margins": 0.9643187522888184,
|
| 1878 |
+
"rewards/rejected": -0.35730695724487305,
|
| 1879 |
+
"step": 1140
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 2.99,
|
| 1883 |
+
"grad_norm": 14.4375,
|
| 1884 |
+
"learning_rate": 9.148382544856885e-07,
|
| 1885 |
+
"logits/chosen": -1.8054192066192627,
|
| 1886 |
+
"logits/rejected": -1.7991243600845337,
|
| 1887 |
+
"logps/chosen": -26.93368911743164,
|
| 1888 |
+
"logps/rejected": -33.427791595458984,
|
| 1889 |
+
"loss": 0.2445,
|
| 1890 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1891 |
+
"rewards/chosen": 0.4817207455635071,
|
| 1892 |
+
"rewards/margins": 0.993517279624939,
|
| 1893 |
+
"rewards/rejected": -0.5117965936660767,
|
| 1894 |
+
"step": 1150
|
| 1895 |
+
},
|
| 1896 |
+
{
|
| 1897 |
+
"epoch": 3.01,
|
| 1898 |
+
"grad_norm": 10.75,
|
| 1899 |
+
"learning_rate": 8.714301001505568e-07,
|
| 1900 |
+
"logits/chosen": -1.9188390970230103,
|
| 1901 |
+
"logits/rejected": -1.9163001775741577,
|
| 1902 |
+
"logps/chosen": -31.702442169189453,
|
| 1903 |
+
"logps/rejected": -36.311798095703125,
|
| 1904 |
+
"loss": 0.2506,
|
| 1905 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1906 |
+
"rewards/chosen": 0.557451069355011,
|
| 1907 |
+
"rewards/margins": 1.0455574989318848,
|
| 1908 |
+
"rewards/rejected": -0.4881063997745514,
|
| 1909 |
+
"step": 1160
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 3.04,
|
| 1913 |
+
"grad_norm": 6.78125,
|
| 1914 |
+
"learning_rate": 8.288586291031025e-07,
|
| 1915 |
+
"logits/chosen": -1.9051227569580078,
|
| 1916 |
+
"logits/rejected": -1.9030368328094482,
|
| 1917 |
+
"logps/chosen": -30.17287254333496,
|
| 1918 |
+
"logps/rejected": -33.75264358520508,
|
| 1919 |
+
"loss": 0.1641,
|
| 1920 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1921 |
+
"rewards/chosen": 0.5766139030456543,
|
| 1922 |
+
"rewards/margins": 1.1320630311965942,
|
| 1923 |
+
"rewards/rejected": -0.5554491877555847,
|
| 1924 |
+
"step": 1170
|
| 1925 |
+
},
|
| 1926 |
+
{
|
| 1927 |
+
"epoch": 3.06,
|
| 1928 |
+
"grad_norm": 12.4375,
|
| 1929 |
+
"learning_rate": 7.871457125803897e-07,
|
| 1930 |
+
"logits/chosen": -1.8981797695159912,
|
| 1931 |
+
"logits/rejected": -1.8859472274780273,
|
| 1932 |
+
"logps/chosen": -34.58381271362305,
|
| 1933 |
+
"logps/rejected": -35.41051483154297,
|
| 1934 |
+
"loss": 0.1915,
|
| 1935 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1936 |
+
"rewards/chosen": 0.5845947861671448,
|
| 1937 |
+
"rewards/margins": 1.1202127933502197,
|
| 1938 |
+
"rewards/rejected": -0.535618007183075,
|
| 1939 |
+
"step": 1180
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 3.09,
|
| 1943 |
+
"grad_norm": 14.5625,
|
| 1944 |
+
"learning_rate": 7.463127807341966e-07,
|
| 1945 |
+
"logits/chosen": -1.7928569316864014,
|
| 1946 |
+
"logits/rejected": -1.7875295877456665,
|
| 1947 |
+
"logps/chosen": -33.692710876464844,
|
| 1948 |
+
"logps/rejected": -35.34910583496094,
|
| 1949 |
+
"loss": 0.1739,
|
| 1950 |
+
"rewards/accuracies": 0.9375,
|
| 1951 |
+
"rewards/chosen": 0.6627923250198364,
|
| 1952 |
+
"rewards/margins": 1.1073065996170044,
|
| 1953 |
+
"rewards/rejected": -0.4445141851902008,
|
| 1954 |
+
"step": 1190
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 3.12,
|
| 1958 |
+
"grad_norm": 9.875,
|
| 1959 |
+
"learning_rate": 7.063808116212021e-07,
|
| 1960 |
+
"logits/chosen": -1.838303804397583,
|
| 1961 |
+
"logits/rejected": -1.8414958715438843,
|
| 1962 |
+
"logps/chosen": -31.297800064086914,
|
| 1963 |
+
"logps/rejected": -32.53749084472656,
|
| 1964 |
+
"loss": 0.2936,
|
| 1965 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1966 |
+
"rewards/chosen": 0.526917576789856,
|
| 1967 |
+
"rewards/margins": 0.9474434852600098,
|
| 1968 |
+
"rewards/rejected": -0.4205259382724762,
|
| 1969 |
+
"step": 1200
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 3.12,
|
| 1973 |
+
"eval_logits/chosen": -2.1406960487365723,
|
| 1974 |
+
"eval_logits/rejected": -2.1359305381774902,
|
| 1975 |
+
"eval_logps/chosen": -34.38763427734375,
|
| 1976 |
+
"eval_logps/rejected": -38.051002502441406,
|
| 1977 |
+
"eval_loss": 0.9310786128044128,
|
| 1978 |
+
"eval_rewards/accuracies": 0.5863787531852722,
|
| 1979 |
+
"eval_rewards/chosen": -0.14123326539993286,
|
| 1980 |
+
"eval_rewards/margins": 0.07251953333616257,
|
| 1981 |
+
"eval_rewards/rejected": -0.21375279128551483,
|
| 1982 |
+
"eval_runtime": 145.1819,
|
| 1983 |
+
"eval_samples_per_second": 2.363,
|
| 1984 |
+
"eval_steps_per_second": 0.296,
|
| 1985 |
+
"step": 1200
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 3.14,
|
| 1989 |
+
"grad_norm": 13.125,
|
| 1990 |
+
"learning_rate": 6.673703204254348e-07,
|
| 1991 |
+
"logits/chosen": -1.8620998859405518,
|
| 1992 |
+
"logits/rejected": -1.857629418373108,
|
| 1993 |
+
"logps/chosen": -29.83384132385254,
|
| 1994 |
+
"logps/rejected": -32.17254638671875,
|
| 1995 |
+
"loss": 0.1952,
|
| 1996 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1997 |
+
"rewards/chosen": 0.5759321451187134,
|
| 1998 |
+
"rewards/margins": 1.0817142724990845,
|
| 1999 |
+
"rewards/rejected": -0.5057822465896606,
|
| 2000 |
+
"step": 1210
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 3.17,
|
| 2004 |
+
"grad_norm": 10.9375,
|
| 2005 |
+
"learning_rate": 6.293013489185315e-07,
|
| 2006 |
+
"logits/chosen": -1.8760480880737305,
|
| 2007 |
+
"logits/rejected": -1.870548963546753,
|
| 2008 |
+
"logps/chosen": -33.693180084228516,
|
| 2009 |
+
"logps/rejected": -35.85017013549805,
|
| 2010 |
+
"loss": 0.201,
|
| 2011 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2012 |
+
"rewards/chosen": 0.6323033571243286,
|
| 2013 |
+
"rewards/margins": 1.1444201469421387,
|
| 2014 |
+
"rewards/rejected": -0.5121166706085205,
|
| 2015 |
+
"step": 1220
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 3.19,
|
| 2019 |
+
"grad_norm": 8.75,
|
| 2020 |
+
"learning_rate": 5.921934551632086e-07,
|
| 2021 |
+
"logits/chosen": -1.8780237436294556,
|
| 2022 |
+
"logits/rejected": -1.8647384643554688,
|
| 2023 |
+
"logps/chosen": -31.742191314697266,
|
| 2024 |
+
"logps/rejected": -35.399742126464844,
|
| 2025 |
+
"loss": 0.2536,
|
| 2026 |
+
"rewards/accuracies": 0.862500011920929,
|
| 2027 |
+
"rewards/chosen": 0.5407308340072632,
|
| 2028 |
+
"rewards/margins": 1.0526906251907349,
|
| 2029 |
+
"rewards/rejected": -0.5119598507881165,
|
| 2030 |
+
"step": 1230
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 3.22,
|
| 2034 |
+
"grad_norm": 7.96875,
|
| 2035 |
+
"learning_rate": 5.560657034652405e-07,
|
| 2036 |
+
"logits/chosen": -1.923841118812561,
|
| 2037 |
+
"logits/rejected": -1.9213125705718994,
|
| 2038 |
+
"logps/chosen": -33.42062759399414,
|
| 2039 |
+
"logps/rejected": -32.888607025146484,
|
| 2040 |
+
"loss": 0.198,
|
| 2041 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 2042 |
+
"rewards/chosen": 0.6449113488197327,
|
| 2043 |
+
"rewards/margins": 1.064406156539917,
|
| 2044 |
+
"rewards/rejected": -0.41949471831321716,
|
| 2045 |
+
"step": 1240
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 3.25,
|
| 2049 |
+
"grad_norm": 7.9375,
|
| 2050 |
+
"learning_rate": 5.2093665457911e-07,
|
| 2051 |
+
"logits/chosen": -1.8493897914886475,
|
| 2052 |
+
"logits/rejected": -1.846502661705017,
|
| 2053 |
+
"logps/chosen": -32.840755462646484,
|
| 2054 |
+
"logps/rejected": -36.272361755371094,
|
| 2055 |
+
"loss": 0.1617,
|
| 2056 |
+
"rewards/accuracies": 0.9375,
|
| 2057 |
+
"rewards/chosen": 0.6221579909324646,
|
| 2058 |
+
"rewards/margins": 1.2122505903244019,
|
| 2059 |
+
"rewards/rejected": -0.5900925397872925,
|
| 2060 |
+
"step": 1250
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 3.27,
|
| 2064 |
+
"grad_norm": 9.0,
|
| 2065 |
+
"learning_rate": 4.868243561723535e-07,
|
| 2066 |
+
"logits/chosen": -1.9173166751861572,
|
| 2067 |
+
"logits/rejected": -1.912136435508728,
|
| 2068 |
+
"logps/chosen": -29.191509246826172,
|
| 2069 |
+
"logps/rejected": -33.58539581298828,
|
| 2070 |
+
"loss": 0.1642,
|
| 2071 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2072 |
+
"rewards/chosen": 0.6277153491973877,
|
| 2073 |
+
"rewards/margins": 1.1088515520095825,
|
| 2074 |
+
"rewards/rejected": -0.4811362624168396,
|
| 2075 |
+
"step": 1260
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 3.3,
|
| 2079 |
+
"grad_norm": 8.75,
|
| 2080 |
+
"learning_rate": 4.537463335535161e-07,
|
| 2081 |
+
"logits/chosen": -1.943853735923767,
|
| 2082 |
+
"logits/rejected": -1.9487136602401733,
|
| 2083 |
+
"logps/chosen": -31.14084815979004,
|
| 2084 |
+
"logps/rejected": -32.768775939941406,
|
| 2085 |
+
"loss": 0.1881,
|
| 2086 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2087 |
+
"rewards/chosen": 0.6280597448348999,
|
| 2088 |
+
"rewards/margins": 1.063023567199707,
|
| 2089 |
+
"rewards/rejected": -0.43496379256248474,
|
| 2090 |
+
"step": 1270
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 3.32,
|
| 2094 |
+
"grad_norm": 13.25,
|
| 2095 |
+
"learning_rate": 4.217195806684629e-07,
|
| 2096 |
+
"logits/chosen": -1.8363254070281982,
|
| 2097 |
+
"logits/rejected": -1.8437111377716064,
|
| 2098 |
+
"logps/chosen": -33.44670867919922,
|
| 2099 |
+
"logps/rejected": -33.85987854003906,
|
| 2100 |
+
"loss": 0.2191,
|
| 2101 |
+
"rewards/accuracies": 0.9375,
|
| 2102 |
+
"rewards/chosen": 0.5842136144638062,
|
| 2103 |
+
"rewards/margins": 1.0888583660125732,
|
| 2104 |
+
"rewards/rejected": -0.5046447515487671,
|
| 2105 |
+
"step": 1280
|
| 2106 |
+
},
|
| 2107 |
+
{
|
| 2108 |
+
"epoch": 3.35,
|
| 2109 |
+
"grad_norm": 9.3125,
|
| 2110 |
+
"learning_rate": 3.907605513696808e-07,
|
| 2111 |
+
"logits/chosen": -1.74712336063385,
|
| 2112 |
+
"logits/rejected": -1.7494428157806396,
|
| 2113 |
+
"logps/chosen": -31.577123641967773,
|
| 2114 |
+
"logps/rejected": -37.742897033691406,
|
| 2115 |
+
"loss": 0.2161,
|
| 2116 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2117 |
+
"rewards/chosen": 0.6219089031219482,
|
| 2118 |
+
"rewards/margins": 1.0973594188690186,
|
| 2119 |
+
"rewards/rejected": -0.47545066475868225,
|
| 2120 |
+
"step": 1290
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 3.38,
|
| 2124 |
+
"grad_norm": 5.6875,
|
| 2125 |
+
"learning_rate": 3.6088515096305675e-07,
|
| 2126 |
+
"logits/chosen": -1.7973334789276123,
|
| 2127 |
+
"logits/rejected": -1.8008067607879639,
|
| 2128 |
+
"logps/chosen": -31.566543579101562,
|
| 2129 |
+
"logps/rejected": -33.696510314941406,
|
| 2130 |
+
"loss": 0.1526,
|
| 2131 |
+
"rewards/accuracies": 0.987500011920929,
|
| 2132 |
+
"rewards/chosen": 0.6600514650344849,
|
| 2133 |
+
"rewards/margins": 1.1529910564422607,
|
| 2134 |
+
"rewards/rejected": -0.4929395616054535,
|
| 2135 |
+
"step": 1300
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 3.38,
|
| 2139 |
+
"eval_logits/chosen": -2.139777898788452,
|
| 2140 |
+
"eval_logits/rejected": -2.135021686553955,
|
| 2141 |
+
"eval_logps/chosen": -34.38774108886719,
|
| 2142 |
+
"eval_logps/rejected": -38.054115295410156,
|
| 2143 |
+
"eval_loss": 0.9307055473327637,
|
| 2144 |
+
"eval_rewards/accuracies": 0.5863787531852722,
|
| 2145 |
+
"eval_rewards/chosen": -0.14127586781978607,
|
| 2146 |
+
"eval_rewards/margins": 0.07372146099805832,
|
| 2147 |
+
"eval_rewards/rejected": -0.2149973064661026,
|
| 2148 |
+
"eval_runtime": 145.1112,
|
| 2149 |
+
"eval_samples_per_second": 2.364,
|
| 2150 |
+
"eval_steps_per_second": 0.296,
|
| 2151 |
+
"step": 1300
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 3.4,
|
| 2155 |
+
"grad_norm": 8.0625,
|
| 2156 |
+
"learning_rate": 3.321087280364757e-07,
|
| 2157 |
+
"logits/chosen": -1.9071658849716187,
|
| 2158 |
+
"logits/rejected": -1.886819839477539,
|
| 2159 |
+
"logps/chosen": -29.60116195678711,
|
| 2160 |
+
"logps/rejected": -36.554786682128906,
|
| 2161 |
+
"loss": 0.2053,
|
| 2162 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2163 |
+
"rewards/chosen": 0.6048943400382996,
|
| 2164 |
+
"rewards/margins": 1.1602458953857422,
|
| 2165 |
+
"rewards/rejected": -0.5553516149520874,
|
| 2166 |
+
"step": 1310
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 3.43,
|
| 2170 |
+
"grad_norm": 13.375,
|
| 2171 |
+
"learning_rate": 3.044460665744284e-07,
|
| 2172 |
+
"logits/chosen": -1.8822952508926392,
|
| 2173 |
+
"logits/rejected": -1.888043761253357,
|
| 2174 |
+
"logps/chosen": -31.35972023010254,
|
| 2175 |
+
"logps/rejected": -33.60492706298828,
|
| 2176 |
+
"loss": 0.1474,
|
| 2177 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2178 |
+
"rewards/chosen": 0.6696279048919678,
|
| 2179 |
+
"rewards/margins": 1.115299105644226,
|
| 2180 |
+
"rewards/rejected": -0.4456712603569031,
|
| 2181 |
+
"step": 1320
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 3.45,
|
| 2185 |
+
"grad_norm": 7.28125,
|
| 2186 |
+
"learning_rate": 2.779113783626916e-07,
|
| 2187 |
+
"logits/chosen": -1.8583717346191406,
|
| 2188 |
+
"logits/rejected": -1.8534027338027954,
|
| 2189 |
+
"logps/chosen": -31.66836166381836,
|
| 2190 |
+
"logps/rejected": -35.97147750854492,
|
| 2191 |
+
"loss": 0.1801,
|
| 2192 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2193 |
+
"rewards/chosen": 0.5416578650474548,
|
| 2194 |
+
"rewards/margins": 1.1013325452804565,
|
| 2195 |
+
"rewards/rejected": -0.5596746802330017,
|
| 2196 |
+
"step": 1330
|
| 2197 |
+
},
|
| 2198 |
+
{
|
| 2199 |
+
"epoch": 3.48,
|
| 2200 |
+
"grad_norm": 10.5625,
|
| 2201 |
+
"learning_rate": 2.5251829568697204e-07,
|
| 2202 |
+
"logits/chosen": -1.720171570777893,
|
| 2203 |
+
"logits/rejected": -1.729686975479126,
|
| 2204 |
+
"logps/chosen": -32.17263412475586,
|
| 2205 |
+
"logps/rejected": -31.95621681213379,
|
| 2206 |
+
"loss": 0.2302,
|
| 2207 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2208 |
+
"rewards/chosen": 0.6507819294929504,
|
| 2209 |
+
"rewards/margins": 1.019622564315796,
|
| 2210 |
+
"rewards/rejected": -0.368840754032135,
|
| 2211 |
+
"step": 1340
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 3.51,
|
| 2215 |
+
"grad_norm": 9.25,
|
| 2216 |
+
"learning_rate": 2.2827986432927774e-07,
|
| 2217 |
+
"logits/chosen": -1.7789561748504639,
|
| 2218 |
+
"logits/rejected": -1.7837011814117432,
|
| 2219 |
+
"logps/chosen": -32.18306350708008,
|
| 2220 |
+
"logps/rejected": -34.33165740966797,
|
| 2221 |
+
"loss": 0.1217,
|
| 2222 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 2223 |
+
"rewards/chosen": 0.7071508169174194,
|
| 2224 |
+
"rewards/margins": 1.157888650894165,
|
| 2225 |
+
"rewards/rejected": -0.45073771476745605,
|
| 2226 |
+
"step": 1350
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 3.53,
|
| 2230 |
+
"grad_norm": 17.25,
|
| 2231 |
+
"learning_rate": 2.0520853686560177e-07,
|
| 2232 |
+
"logits/chosen": -1.8064203262329102,
|
| 2233 |
+
"logits/rejected": -1.811870813369751,
|
| 2234 |
+
"logps/chosen": -32.34109878540039,
|
| 2235 |
+
"logps/rejected": -35.7831916809082,
|
| 2236 |
+
"loss": 0.2097,
|
| 2237 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 2238 |
+
"rewards/chosen": 0.5962255597114563,
|
| 2239 |
+
"rewards/margins": 1.0842751264572144,
|
| 2240 |
+
"rewards/rejected": -0.4880496561527252,
|
| 2241 |
+
"step": 1360
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 3.56,
|
| 2245 |
+
"grad_norm": 7.28125,
|
| 2246 |
+
"learning_rate": 1.833161662683672e-07,
|
| 2247 |
+
"logits/chosen": -1.8998634815216064,
|
| 2248 |
+
"logits/rejected": -1.8953298330307007,
|
| 2249 |
+
"logps/chosen": -31.270008087158203,
|
| 2250 |
+
"logps/rejected": -32.123897552490234,
|
| 2251 |
+
"loss": 0.1836,
|
| 2252 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 2253 |
+
"rewards/chosen": 0.546270489692688,
|
| 2254 |
+
"rewards/margins": 1.0310702323913574,
|
| 2255 |
+
"rewards/rejected": -0.4847997725009918,
|
| 2256 |
+
"step": 1370
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 3.58,
|
| 2260 |
+
"grad_norm": 14.0625,
|
| 2261 |
+
"learning_rate": 1.626139998169246e-07,
|
| 2262 |
+
"logits/chosen": -1.8143882751464844,
|
| 2263 |
+
"logits/rejected": -1.816428542137146,
|
| 2264 |
+
"logps/chosen": -28.297687530517578,
|
| 2265 |
+
"logps/rejected": -31.918853759765625,
|
| 2266 |
+
"loss": 0.2172,
|
| 2267 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2268 |
+
"rewards/chosen": 0.5446463823318481,
|
| 2269 |
+
"rewards/margins": 1.0145161151885986,
|
| 2270 |
+
"rewards/rejected": -0.4698697626590729,
|
| 2271 |
+
"step": 1380
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 3.61,
|
| 2275 |
+
"grad_norm": 13.75,
|
| 2276 |
+
"learning_rate": 1.4311267331922535e-07,
|
| 2277 |
+
"logits/chosen": -1.8110907077789307,
|
| 2278 |
+
"logits/rejected": -1.8157243728637695,
|
| 2279 |
+
"logps/chosen": -30.599262237548828,
|
| 2280 |
+
"logps/rejected": -33.90684127807617,
|
| 2281 |
+
"loss": 0.228,
|
| 2282 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2283 |
+
"rewards/chosen": 0.5680925250053406,
|
| 2284 |
+
"rewards/margins": 0.9356026649475098,
|
| 2285 |
+
"rewards/rejected": -0.3675101101398468,
|
| 2286 |
+
"step": 1390
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 3.64,
|
| 2290 |
+
"grad_norm": 5.28125,
|
| 2291 |
+
"learning_rate": 1.2482220564763669e-07,
|
| 2292 |
+
"logits/chosen": -1.8766982555389404,
|
| 2293 |
+
"logits/rejected": -1.8799679279327393,
|
| 2294 |
+
"logps/chosen": -33.26051712036133,
|
| 2295 |
+
"logps/rejected": -34.686241149902344,
|
| 2296 |
+
"loss": 0.1121,
|
| 2297 |
+
"rewards/accuracies": 0.987500011920929,
|
| 2298 |
+
"rewards/chosen": 0.6680229902267456,
|
| 2299 |
+
"rewards/margins": 1.1730940341949463,
|
| 2300 |
+
"rewards/rejected": -0.5050708651542664,
|
| 2301 |
+
"step": 1400
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 3.64,
|
| 2305 |
+
"eval_logits/chosen": -2.14040207862854,
|
| 2306 |
+
"eval_logits/rejected": -2.135646104812622,
|
| 2307 |
+
"eval_logps/chosen": -34.378780364990234,
|
| 2308 |
+
"eval_logps/rejected": -38.054847717285156,
|
| 2309 |
+
"eval_loss": 0.925031840801239,
|
| 2310 |
+
"eval_rewards/accuracies": 0.6038206219673157,
|
| 2311 |
+
"eval_rewards/chosen": -0.13769178092479706,
|
| 2312 |
+
"eval_rewards/margins": 0.07760081440210342,
|
| 2313 |
+
"eval_rewards/rejected": -0.21529260277748108,
|
| 2314 |
+
"eval_runtime": 145.1344,
|
| 2315 |
+
"eval_samples_per_second": 2.363,
|
| 2316 |
+
"eval_steps_per_second": 0.296,
|
| 2317 |
+
"step": 1400
|
| 2318 |
+
},
|
| 2319 |
+
{
|
| 2320 |
+
"epoch": 3.66,
|
| 2321 |
+
"grad_norm": 13.4375,
|
| 2322 |
+
"learning_rate": 1.0775199359171346e-07,
|
| 2323 |
+
"logits/chosen": -1.945010781288147,
|
| 2324 |
+
"logits/rejected": -1.9381306171417236,
|
| 2325 |
+
"logps/chosen": -32.14293670654297,
|
| 2326 |
+
"logps/rejected": -35.293968200683594,
|
| 2327 |
+
"loss": 0.1774,
|
| 2328 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 2329 |
+
"rewards/chosen": 0.6638438105583191,
|
| 2330 |
+
"rewards/margins": 1.0836031436920166,
|
| 2331 |
+
"rewards/rejected": -0.4197593331336975,
|
| 2332 |
+
"step": 1410
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 3.69,
|
| 2336 |
+
"grad_norm": 12.1875,
|
| 2337 |
+
"learning_rate": 9.191080703056604e-08,
|
| 2338 |
+
"logits/chosen": -1.8345167636871338,
|
| 2339 |
+
"logits/rejected": -1.8454965353012085,
|
| 2340 |
+
"logps/chosen": -32.55244827270508,
|
| 2341 |
+
"logps/rejected": -34.81972122192383,
|
| 2342 |
+
"loss": 0.1625,
|
| 2343 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2344 |
+
"rewards/chosen": 0.6387423276901245,
|
| 2345 |
+
"rewards/margins": 1.1301978826522827,
|
| 2346 |
+
"rewards/rejected": -0.49145546555519104,
|
| 2347 |
+
"step": 1420
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 3.71,
|
| 2351 |
+
"grad_norm": 13.0,
|
| 2352 |
+
"learning_rate": 7.730678442730539e-08,
|
| 2353 |
+
"logits/chosen": -1.9138675928115845,
|
| 2354 |
+
"logits/rejected": -1.9257293939590454,
|
| 2355 |
+
"logps/chosen": -33.43199920654297,
|
| 2356 |
+
"logps/rejected": -34.919532775878906,
|
| 2357 |
+
"loss": 0.1706,
|
| 2358 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2359 |
+
"rewards/chosen": 0.6515201926231384,
|
| 2360 |
+
"rewards/margins": 1.1639997959136963,
|
| 2361 |
+
"rewards/rejected": -0.5124797224998474,
|
| 2362 |
+
"step": 1430
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 3.74,
|
| 2366 |
+
"grad_norm": 11.1875,
|
| 2367 |
+
"learning_rate": 6.394742864787806e-08,
|
| 2368 |
+
"logits/chosen": -1.904793381690979,
|
| 2369 |
+
"logits/rejected": -1.9069957733154297,
|
| 2370 |
+
"logps/chosen": -31.7263126373291,
|
| 2371 |
+
"logps/rejected": -35.0744743347168,
|
| 2372 |
+
"loss": 0.2078,
|
| 2373 |
+
"rewards/accuracies": 0.9375,
|
| 2374 |
+
"rewards/chosen": 0.5928308367729187,
|
| 2375 |
+
"rewards/margins": 1.0786316394805908,
|
| 2376 |
+
"rewards/rejected": -0.4858008325099945,
|
| 2377 |
+
"step": 1440
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 3.77,
|
| 2381 |
+
"grad_norm": 9.5625,
|
| 2382 |
+
"learning_rate": 5.183960310644748e-08,
|
| 2383 |
+
"logits/chosen": -1.8884856700897217,
|
| 2384 |
+
"logits/rejected": -1.8813756704330444,
|
| 2385 |
+
"logps/chosen": -33.934364318847656,
|
| 2386 |
+
"logps/rejected": -34.85586929321289,
|
| 2387 |
+
"loss": 0.2029,
|
| 2388 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2389 |
+
"rewards/chosen": 0.6535843014717102,
|
| 2390 |
+
"rewards/margins": 1.0877697467803955,
|
| 2391 |
+
"rewards/rejected": -0.43418556451797485,
|
| 2392 |
+
"step": 1450
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"epoch": 3.79,
|
| 2396 |
+
"grad_norm": 18.875,
|
| 2397 |
+
"learning_rate": 4.098952823928693e-08,
|
| 2398 |
+
"logits/chosen": -1.8547000885009766,
|
| 2399 |
+
"logits/rejected": -1.860727310180664,
|
| 2400 |
+
"logps/chosen": -29.743526458740234,
|
| 2401 |
+
"logps/rejected": -34.97495651245117,
|
| 2402 |
+
"loss": 0.1719,
|
| 2403 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2404 |
+
"rewards/chosen": 0.5878596305847168,
|
| 2405 |
+
"rewards/margins": 1.1098417043685913,
|
| 2406 |
+
"rewards/rejected": -0.5219820737838745,
|
| 2407 |
+
"step": 1460
|
| 2408 |
+
},
|
| 2409 |
+
{
|
| 2410 |
+
"epoch": 3.82,
|
| 2411 |
+
"grad_norm": 12.875,
|
| 2412 |
+
"learning_rate": 3.1402778309014284e-08,
|
| 2413 |
+
"logits/chosen": -1.8023372888565063,
|
| 2414 |
+
"logits/rejected": -1.8003690242767334,
|
| 2415 |
+
"logps/chosen": -29.268962860107422,
|
| 2416 |
+
"logps/rejected": -31.533855438232422,
|
| 2417 |
+
"loss": 0.2267,
|
| 2418 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2419 |
+
"rewards/chosen": 0.6306362152099609,
|
| 2420 |
+
"rewards/margins": 1.006486177444458,
|
| 2421 |
+
"rewards/rejected": -0.37585002183914185,
|
| 2422 |
+
"step": 1470
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 3.84,
|
| 2426 |
+
"grad_norm": 9.5625,
|
| 2427 |
+
"learning_rate": 2.3084278540791427e-08,
|
| 2428 |
+
"logits/chosen": -1.9781367778778076,
|
| 2429 |
+
"logits/rejected": -1.9725608825683594,
|
| 2430 |
+
"logps/chosen": -33.99125671386719,
|
| 2431 |
+
"logps/rejected": -33.27691650390625,
|
| 2432 |
+
"loss": 0.1815,
|
| 2433 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 2434 |
+
"rewards/chosen": 0.5176017880439758,
|
| 2435 |
+
"rewards/margins": 1.0056849718093872,
|
| 2436 |
+
"rewards/rejected": -0.4880831241607666,
|
| 2437 |
+
"step": 1480
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 3.87,
|
| 2441 |
+
"grad_norm": 14.25,
|
| 2442 |
+
"learning_rate": 1.6038302591975807e-08,
|
| 2443 |
+
"logits/chosen": -1.8709309101104736,
|
| 2444 |
+
"logits/rejected": -1.873234510421753,
|
| 2445 |
+
"logps/chosen": -27.13213539123535,
|
| 2446 |
+
"logps/rejected": -29.021121978759766,
|
| 2447 |
+
"loss": 0.2587,
|
| 2448 |
+
"rewards/accuracies": 0.875,
|
| 2449 |
+
"rewards/chosen": 0.5376989245414734,
|
| 2450 |
+
"rewards/margins": 0.9372351765632629,
|
| 2451 |
+
"rewards/rejected": -0.39953625202178955,
|
| 2452 |
+
"step": 1490
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 3.9,
|
| 2456 |
+
"grad_norm": 16.25,
|
| 2457 |
+
"learning_rate": 1.0268470356514237e-08,
|
| 2458 |
+
"logits/chosen": -1.8781601190567017,
|
| 2459 |
+
"logits/rejected": -1.8725292682647705,
|
| 2460 |
+
"logps/chosen": -31.48541831970215,
|
| 2461 |
+
"logps/rejected": -33.22844696044922,
|
| 2462 |
+
"loss": 0.215,
|
| 2463 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2464 |
+
"rewards/chosen": 0.5261479020118713,
|
| 2465 |
+
"rewards/margins": 1.0299475193023682,
|
| 2466 |
+
"rewards/rejected": -0.5037996768951416,
|
| 2467 |
+
"step": 1500
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 3.9,
|
| 2471 |
+
"eval_logits/chosen": -2.140532970428467,
|
| 2472 |
+
"eval_logits/rejected": -2.135777473449707,
|
| 2473 |
+
"eval_logps/chosen": -34.38066101074219,
|
| 2474 |
+
"eval_logps/rejected": -38.052310943603516,
|
| 2475 |
+
"eval_loss": 0.9279635548591614,
|
| 2476 |
+
"eval_rewards/accuracies": 0.5805647969245911,
|
| 2477 |
+
"eval_rewards/chosen": -0.1384437084197998,
|
| 2478 |
+
"eval_rewards/margins": 0.07583153247833252,
|
| 2479 |
+
"eval_rewards/rejected": -0.21427522599697113,
|
| 2480 |
+
"eval_runtime": 145.1697,
|
| 2481 |
+
"eval_samples_per_second": 2.363,
|
| 2482 |
+
"eval_steps_per_second": 0.296,
|
| 2483 |
+
"step": 1500
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 3.92,
|
| 2487 |
+
"grad_norm": 16.375,
|
| 2488 |
+
"learning_rate": 5.777746105209147e-09,
|
| 2489 |
+
"logits/chosen": -1.803832769393921,
|
| 2490 |
+
"logits/rejected": -1.8080018758773804,
|
| 2491 |
+
"logps/chosen": -32.76865005493164,
|
| 2492 |
+
"logps/rejected": -35.45244216918945,
|
| 2493 |
+
"loss": 0.2319,
|
| 2494 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2495 |
+
"rewards/chosen": 0.56093829870224,
|
| 2496 |
+
"rewards/margins": 0.9777617454528809,
|
| 2497 |
+
"rewards/rejected": -0.41682347655296326,
|
| 2498 |
+
"step": 1510
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 3.95,
|
| 2502 |
+
"grad_norm": 5.53125,
|
| 2503 |
+
"learning_rate": 2.5684369628148352e-09,
|
| 2504 |
+
"logits/chosen": -1.8591749668121338,
|
| 2505 |
+
"logits/rejected": -1.8575620651245117,
|
| 2506 |
+
"logps/chosen": -29.235681533813477,
|
| 2507 |
+
"logps/rejected": -33.72986602783203,
|
| 2508 |
+
"loss": 0.2297,
|
| 2509 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2510 |
+
"rewards/chosen": 0.5993990302085876,
|
| 2511 |
+
"rewards/margins": 1.0554559230804443,
|
| 2512 |
+
"rewards/rejected": -0.4560568928718567,
|
| 2513 |
+
"step": 1520
|
| 2514 |
+
},
|
| 2515 |
+
{
|
| 2516 |
+
"epoch": 3.97,
|
| 2517 |
+
"grad_norm": 14.1875,
|
| 2518 |
+
"learning_rate": 6.421917227455999e-10,
|
| 2519 |
+
"logits/chosen": -1.9594846963882446,
|
| 2520 |
+
"logits/rejected": -1.951703667640686,
|
| 2521 |
+
"logps/chosen": -26.725296020507812,
|
| 2522 |
+
"logps/rejected": -29.879215240478516,
|
| 2523 |
+
"loss": 0.2118,
|
| 2524 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2525 |
+
"rewards/chosen": 0.5350590944290161,
|
| 2526 |
+
"rewards/margins": 1.0181024074554443,
|
| 2527 |
+
"rewards/rejected": -0.4830433428287506,
|
| 2528 |
+
"step": 1530
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 4.0,
|
| 2532 |
+
"grad_norm": 13.9375,
|
| 2533 |
+
"learning_rate": 0.0,
|
| 2534 |
+
"logits/chosen": -1.8623542785644531,
|
| 2535 |
+
"logits/rejected": -1.8520355224609375,
|
| 2536 |
+
"logps/chosen": -31.616281509399414,
|
| 2537 |
+
"logps/rejected": -36.66792297363281,
|
| 2538 |
+
"loss": 0.1728,
|
| 2539 |
+
"rewards/accuracies": 0.987500011920929,
|
| 2540 |
+
"rewards/chosen": 0.5799667835235596,
|
| 2541 |
+
"rewards/margins": 1.0826661586761475,
|
| 2542 |
+
"rewards/rejected": -0.5026994943618774,
|
| 2543 |
+
"step": 1540
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 4.0,
|
| 2547 |
+
"step": 1540,
|
| 2548 |
"total_flos": 0.0,
|
| 2549 |
+
"train_loss": 0.2625940967689861,
|
| 2550 |
+
"train_runtime": 10748.6274,
|
| 2551 |
+
"train_samples_per_second": 1.146,
|
| 2552 |
+
"train_steps_per_second": 0.143
|
| 2553 |
}
|
| 2554 |
],
|
| 2555 |
"logging_steps": 10,
|
| 2556 |
+
"max_steps": 1540,
|
| 2557 |
"num_input_tokens_seen": 0,
|
| 2558 |
+
"num_train_epochs": 4,
|
| 2559 |
"save_steps": 100,
|
| 2560 |
"total_flos": 0.0,
|
| 2561 |
"train_batch_size": 4,
|