Commit
•
a74ed63
1
Parent(s):
0cbce05
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 |
-
| 0.6726 | 0.26 | 100 |
|
| 67 |
-
| 0.6419 | 0.52 | 200 |
|
| 68 |
-
| 0.5902 | 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: 0.6768
|
| 22 |
+
- Rewards/chosen: -0.2588
|
| 23 |
+
- Rewards/rejected: -0.3705
|
| 24 |
+
- Rewards/accuracies: 0.5801
|
| 25 |
+
- Rewards/margins: 0.1118
|
| 26 |
+
- Logps/rejected: -38.4429
|
| 27 |
+
- Logps/chosen: -34.6815
|
| 28 |
+
- Logits/rejected: -2.0631
|
| 29 |
+
- Logits/chosen: -2.0678
|
| 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 |
+
| 0.6726 | 0.26 | 100 | -2.2334 | -2.2286 | -34.0442 | -37.5524 | 0.6903 | 0.5482 | -0.0039 | 0.0105 | -0.0143 |
|
| 65 |
+
| 0.6419 | 0.52 | 200 | -2.2310 | -2.2261 | -34.0660 | -37.5657 | 0.6932 | 0.5042 | -0.0126 | 0.0071 | -0.0196 |
|
| 66 |
+
| 0.5902 | 0.78 | 300 | -2.2279 | -2.2231 | -34.0966 | -37.5987 | 0.6926 | 0.4954 | -0.0248 | 0.0080 | -0.0328 |
|
| 67 |
+
| 0.5071 | 1.04 | 400 | 0.6867 | -0.0833 | -0.1140 | 0.5511 | 0.0307 | -37.8015 | -34.2428 | -2.1874 | -2.1923 |
|
| 68 |
+
| 0.4361 | 1.3 | 500 | 0.6902 | -0.1634 | -0.2014 | 0.5365 | 0.0381 | -38.0202 | -34.4429 | -2.1687 | -2.1734 |
|
| 69 |
+
| 0.4356 | 1.56 | 600 | 0.6750 | -0.1197 | -0.1930 | 0.5835 | 0.0733 | -37.9990 | -34.3338 | -2.1423 | -2.1471 |
|
| 70 |
+
| 0.4036 | 1.82 | 700 | 0.6719 | -0.1493 | -0.2346 | 0.5860 | 0.0853 | -38.1032 | -34.4078 | -2.1185 | -2.1232 |
|
| 71 |
+
| 0.2952 | 2.08 | 800 | 0.6758 | -0.2049 | -0.3007 | 0.5777 | 0.0958 | -38.2684 | -34.5467 | -2.0980 | -2.1026 |
|
| 72 |
+
| 0.304 | 2.34 | 900 | 0.6779 | -0.2333 | -0.3312 | 0.5951 | 0.0979 | -38.3446 | -34.6178 | -2.0820 | -2.0867 |
|
| 73 |
+
| 0.333 | 2.6 | 1000 | 0.6770 | -0.2554 | -0.3656 | 0.5860 | 0.1102 | -38.4306 | -34.6731 | -2.0677 | -2.0723 |
|
| 74 |
+
| 0.2846 | 2.86 | 1100 | 0.6772 | -0.2533 | -0.3613 | 0.6009 | 0.1080 | -38.4198 | -34.6678 | -2.0646 | -2.0692 |
|
| 75 |
+
| 0.3301 | 3.12 | 1200 | 0.6771 | -0.2546 | -0.3652 | 0.5777 | 0.1106 | -38.4296 | -34.6710 | -2.0635 | -2.0681 |
|
| 76 |
+
| 0.2648 | 3.38 | 1300 | 0.6774 | -0.2592 | -0.3700 | 0.5835 | 0.1108 | -38.4415 | -34.6825 | -2.0632 | -2.0678 |
|
| 77 |
+
| 0.2661 | 3.64 | 1400 | 0.6737 | -0.2593 | -0.3748 | 0.5889 | 0.1155 | -38.4536 | -34.6829 | -2.0635 | -2.0681 |
|
| 78 |
+
| 0.3194 | 3.9 | 1500 | 0.6768 | -0.2588 | -0.3705 | 0.5801 | 0.1118 | -38.4429 | -34.6815 | -2.0631 | -2.0678 |
|
| 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:de477564a644348ac731a86a9eaaf488b841e462afc4204f80ca67301f2297ef
|
| 3 |
size 176183216
|
all_results.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
"eval_logits/chosen": -2.2275288105010986,
|
| 4 |
"eval_logits/rejected": -2.2226884365081787,
|
| 5 |
"eval_logps/chosen": -34.09955978393555,
|
|
@@ -13,9 +13,9 @@
|
|
| 13 |
"eval_samples": 343,
|
| 14 |
"eval_samples_per_second": 2.357,
|
| 15 |
"eval_steps_per_second": 0.295,
|
| 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": -2.2275288105010986,
|
| 4 |
"eval_logits/rejected": -2.2226884365081787,
|
| 5 |
"eval_logps/chosen": -34.09955978393555,
|
|
|
|
| 13 |
"eval_samples": 343,
|
| 14 |
"eval_samples_per_second": 2.357,
|
| 15 |
"eval_steps_per_second": 0.295,
|
| 16 |
+
"train_loss": 0.2942203808140445,
|
| 17 |
+
"train_runtime": 10768.7061,
|
| 18 |
"train_samples": 3079,
|
| 19 |
+
"train_samples_per_second": 1.144,
|
| 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": 0.2942203808140445,
|
| 4 |
+
"train_runtime": 10768.7061,
|
| 5 |
"train_samples": 3079,
|
| 6 |
+
"train_samples_per_second": 1.144,
|
| 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": 7.8125,
|
| 496 |
+
"learning_rate": 4.84533120650964e-06,
|
| 497 |
+
"logits/chosen": -2.0616166591644287,
|
| 498 |
+
"logits/rejected": -2.048832416534424,
|
| 499 |
+
"logps/chosen": -32.10347366333008,
|
| 500 |
+
"logps/rejected": -32.91118621826172,
|
| 501 |
+
"loss": 0.5726,
|
| 502 |
+
"rewards/accuracies": 0.887499988079071,
|
| 503 |
+
"rewards/chosen": 0.16996803879737854,
|
| 504 |
+
"rewards/margins": 0.2780183255672455,
|
| 505 |
+
"rewards/rejected": -0.10805028676986694,
|
| 506 |
"step": 310
|
| 507 |
},
|
| 508 |
{
|
| 509 |
"epoch": 0.83,
|
| 510 |
+
"grad_norm": 7.25,
|
| 511 |
+
"learning_rate": 4.825108134172131e-06,
|
| 512 |
+
"logits/chosen": -1.9727706909179688,
|
| 513 |
+
"logits/rejected": -1.9641412496566772,
|
| 514 |
+
"logps/chosen": -31.7288818359375,
|
| 515 |
+
"logps/rejected": -30.45233154296875,
|
| 516 |
+
"loss": 0.5546,
|
| 517 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 518 |
+
"rewards/chosen": 0.25044065713882446,
|
| 519 |
+
"rewards/margins": 0.33481746912002563,
|
| 520 |
+
"rewards/rejected": -0.08437685668468475,
|
| 521 |
"step": 320
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 0.86,
|
| 525 |
+
"grad_norm": 8.1875,
|
| 526 |
+
"learning_rate": 4.80369052967602e-06,
|
| 527 |
+
"logits/chosen": -1.9075199365615845,
|
| 528 |
+
"logits/rejected": -1.9195775985717773,
|
| 529 |
+
"logps/chosen": -29.810604095458984,
|
| 530 |
+
"logps/rejected": -33.72663497924805,
|
| 531 |
+
"loss": 0.5272,
|
| 532 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 533 |
+
"rewards/chosen": 0.2674657702445984,
|
| 534 |
+
"rewards/margins": 0.3972177803516388,
|
| 535 |
+
"rewards/rejected": -0.1297520250082016,
|
| 536 |
"step": 330
|
| 537 |
},
|
| 538 |
{
|
| 539 |
"epoch": 0.88,
|
| 540 |
+
"grad_norm": 10.0,
|
| 541 |
+
"learning_rate": 4.781089396387968e-06,
|
| 542 |
+
"logits/chosen": -1.8698316812515259,
|
| 543 |
+
"logits/rejected": -1.8606189489364624,
|
| 544 |
+
"logps/chosen": -33.96432876586914,
|
| 545 |
+
"logps/rejected": -36.236751556396484,
|
| 546 |
+
"loss": 0.5218,
|
| 547 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 548 |
+
"rewards/chosen": 0.29383981227874756,
|
| 549 |
+
"rewards/margins": 0.44584956765174866,
|
| 550 |
+
"rewards/rejected": -0.15200971066951752,
|
| 551 |
"step": 340
|
| 552 |
},
|
| 553 |
{
|
| 554 |
"epoch": 0.91,
|
| 555 |
+
"grad_norm": 7.09375,
|
| 556 |
+
"learning_rate": 4.757316345716554e-06,
|
| 557 |
+
"logits/chosen": -1.9213838577270508,
|
| 558 |
+
"logits/rejected": -1.9220517873764038,
|
| 559 |
+
"logps/chosen": -33.60356140136719,
|
| 560 |
+
"logps/rejected": -34.14032745361328,
|
| 561 |
+
"loss": 0.527,
|
| 562 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 563 |
+
"rewards/chosen": 0.3206879496574402,
|
| 564 |
+
"rewards/margins": 0.43722790479660034,
|
| 565 |
+
"rewards/rejected": -0.11653995513916016,
|
| 566 |
"step": 350
|
| 567 |
},
|
| 568 |
{
|
| 569 |
"epoch": 0.94,
|
| 570 |
+
"grad_norm": 7.4375,
|
| 571 |
+
"learning_rate": 4.73238359114687e-06,
|
| 572 |
+
"logits/chosen": -2.0467612743377686,
|
| 573 |
+
"logits/rejected": -2.0528926849365234,
|
| 574 |
+
"logps/chosen": -31.081350326538086,
|
| 575 |
+
"logps/rejected": -33.01131057739258,
|
| 576 |
+
"loss": 0.5605,
|
| 577 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 578 |
+
"rewards/chosen": 0.1967591494321823,
|
| 579 |
+
"rewards/margins": 0.3339698314666748,
|
| 580 |
+
"rewards/rejected": -0.1372106969356537,
|
| 581 |
"step": 360
|
| 582 |
},
|
| 583 |
{
|
| 584 |
"epoch": 0.96,
|
| 585 |
+
"grad_norm": 11.25,
|
| 586 |
+
"learning_rate": 4.706303941965804e-06,
|
| 587 |
+
"logits/chosen": -1.9743931293487549,
|
| 588 |
+
"logits/rejected": -1.9739511013031006,
|
| 589 |
+
"logps/chosen": -32.78725814819336,
|
| 590 |
+
"logps/rejected": -36.41063690185547,
|
| 591 |
+
"loss": 0.5304,
|
| 592 |
+
"rewards/accuracies": 0.862500011920929,
|
| 593 |
+
"rewards/chosen": 0.26307255029678345,
|
| 594 |
+
"rewards/margins": 0.4074175953865051,
|
| 595 |
+
"rewards/rejected": -0.14434504508972168,
|
| 596 |
"step": 370
|
| 597 |
},
|
| 598 |
{
|
| 599 |
"epoch": 0.99,
|
| 600 |
+
"grad_norm": 6.78125,
|
| 601 |
+
"learning_rate": 4.679090796681225e-06,
|
| 602 |
+
"logits/chosen": -2.004617214202881,
|
| 603 |
+
"logits/rejected": -2.000023365020752,
|
| 604 |
+
"logps/chosen": -30.0998592376709,
|
| 605 |
+
"logps/rejected": -29.62681007385254,
|
| 606 |
+
"loss": 0.531,
|
| 607 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 608 |
+
"rewards/chosen": 0.2442973405122757,
|
| 609 |
+
"rewards/margins": 0.4094654619693756,
|
| 610 |
+
"rewards/rejected": -0.1651681363582611,
|
| 611 |
"step": 380
|
| 612 |
},
|
| 613 |
{
|
| 614 |
+
"epoch": 1.01,
|
| 615 |
+
"grad_norm": 8.75,
|
| 616 |
+
"learning_rate": 4.650758136138454e-06,
|
| 617 |
+
"logits/chosen": -1.7745968103408813,
|
| 618 |
+
"logits/rejected": -1.7809730768203735,
|
| 619 |
+
"logps/chosen": -31.600543975830078,
|
| 620 |
+
"logps/rejected": -36.76545333862305,
|
| 621 |
+
"loss": 0.4808,
|
| 622 |
+
"rewards/accuracies": 0.887499988079071,
|
| 623 |
+
"rewards/chosen": 0.31406909227371216,
|
| 624 |
+
"rewards/margins": 0.5729340314865112,
|
| 625 |
+
"rewards/rejected": -0.2588648796081543,
|
| 626 |
+
"step": 390
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.04,
|
| 630 |
+
"grad_norm": 7.21875,
|
| 631 |
+
"learning_rate": 4.621320516337559e-06,
|
| 632 |
+
"logits/chosen": -1.9251466989517212,
|
| 633 |
+
"logits/rejected": -1.9188610315322876,
|
| 634 |
+
"logps/chosen": -32.989654541015625,
|
| 635 |
+
"logps/rejected": -32.73323059082031,
|
| 636 |
+
"loss": 0.5071,
|
| 637 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 638 |
+
"rewards/chosen": 0.33721691370010376,
|
| 639 |
+
"rewards/margins": 0.5069588422775269,
|
| 640 |
+
"rewards/rejected": -0.16974198818206787,
|
| 641 |
+
"step": 400
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 1.04,
|
| 645 |
+
"eval_logits/chosen": -2.1922616958618164,
|
| 646 |
+
"eval_logits/rejected": -2.1874282360076904,
|
| 647 |
+
"eval_logps/chosen": -34.242794036865234,
|
| 648 |
+
"eval_logps/rejected": -37.80149841308594,
|
| 649 |
+
"eval_loss": 0.6867499947547913,
|
| 650 |
+
"eval_rewards/accuracies": 0.5510797500610352,
|
| 651 |
+
"eval_rewards/chosen": -0.08329664915800095,
|
| 652 |
+
"eval_rewards/margins": 0.030653679743409157,
|
| 653 |
+
"eval_rewards/rejected": -0.11395032703876495,
|
| 654 |
+
"eval_runtime": 145.5198,
|
| 655 |
+
"eval_samples_per_second": 2.357,
|
| 656 |
+
"eval_steps_per_second": 0.295,
|
| 657 |
+
"step": 400
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.06,
|
| 661 |
+
"grad_norm": 6.8125,
|
| 662 |
+
"learning_rate": 4.590793060955158e-06,
|
| 663 |
+
"logits/chosen": -1.922146201133728,
|
| 664 |
+
"logits/rejected": -1.9294363260269165,
|
| 665 |
+
"logps/chosen": -28.543231964111328,
|
| 666 |
+
"logps/rejected": -29.782611846923828,
|
| 667 |
+
"loss": 0.4969,
|
| 668 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 669 |
+
"rewards/chosen": 0.23085805773735046,
|
| 670 |
+
"rewards/margins": 0.5046137571334839,
|
| 671 |
+
"rewards/rejected": -0.2737556993961334,
|
| 672 |
+
"step": 410
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.09,
|
| 676 |
+
"grad_norm": 7.375,
|
| 677 |
+
"learning_rate": 4.559191453574582e-06,
|
| 678 |
+
"logits/chosen": -1.939373254776001,
|
| 679 |
+
"logits/rejected": -1.938410758972168,
|
| 680 |
+
"logps/chosen": -33.535621643066406,
|
| 681 |
+
"logps/rejected": -31.325557708740234,
|
| 682 |
+
"loss": 0.5345,
|
| 683 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 684 |
+
"rewards/chosen": 0.28603512048721313,
|
| 685 |
+
"rewards/margins": 0.4390401244163513,
|
| 686 |
+
"rewards/rejected": -0.15300500392913818,
|
| 687 |
+
"step": 420
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 1.12,
|
| 691 |
+
"grad_norm": 8.0625,
|
| 692 |
+
"learning_rate": 4.52653192962838e-06,
|
| 693 |
+
"logits/chosen": -1.9316809177398682,
|
| 694 |
+
"logits/rejected": -1.9144115447998047,
|
| 695 |
+
"logps/chosen": -30.42580223083496,
|
| 696 |
+
"logps/rejected": -33.712886810302734,
|
| 697 |
+
"loss": 0.4939,
|
| 698 |
+
"rewards/accuracies": 0.875,
|
| 699 |
+
"rewards/chosen": 0.2205621749162674,
|
| 700 |
+
"rewards/margins": 0.5352030992507935,
|
| 701 |
+
"rewards/rejected": -0.31464093923568726,
|
| 702 |
+
"step": 430
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 1.14,
|
| 706 |
+
"grad_norm": 8.125,
|
| 707 |
+
"learning_rate": 4.492831268057307e-06,
|
| 708 |
+
"logits/chosen": -1.9621837139129639,
|
| 709 |
+
"logits/rejected": -1.9643888473510742,
|
| 710 |
+
"logps/chosen": -35.697532653808594,
|
| 711 |
+
"logps/rejected": -35.595943450927734,
|
| 712 |
+
"loss": 0.4474,
|
| 713 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 714 |
+
"rewards/chosen": 0.36593085527420044,
|
| 715 |
+
"rewards/margins": 0.6603227853775024,
|
| 716 |
+
"rewards/rejected": -0.2943919003009796,
|
| 717 |
+
"step": 440
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 1.17,
|
| 721 |
+
"grad_norm": 7.125,
|
| 722 |
+
"learning_rate": 4.458106782690094e-06,
|
| 723 |
+
"logits/chosen": -2.035297155380249,
|
| 724 |
+
"logits/rejected": -2.035062074661255,
|
| 725 |
+
"logps/chosen": -31.8659725189209,
|
| 726 |
+
"logps/rejected": -34.10185241699219,
|
| 727 |
+
"loss": 0.5013,
|
| 728 |
+
"rewards/accuracies": 0.862500011920929,
|
| 729 |
+
"rewards/chosen": 0.306951105594635,
|
| 730 |
+
"rewards/margins": 0.523645281791687,
|
| 731 |
+
"rewards/rejected": -0.2166941612958908,
|
| 732 |
+
"step": 450
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.19,
|
| 736 |
+
"grad_norm": 7.78125,
|
| 737 |
+
"learning_rate": 4.422376313348405e-06,
|
| 738 |
+
"logits/chosen": -1.9762487411499023,
|
| 739 |
+
"logits/rejected": -1.968775987625122,
|
| 740 |
+
"logps/chosen": -31.517736434936523,
|
| 741 |
+
"logps/rejected": -36.918033599853516,
|
| 742 |
+
"loss": 0.4394,
|
| 743 |
+
"rewards/accuracies": 0.875,
|
| 744 |
+
"rewards/chosen": 0.3552888035774231,
|
| 745 |
+
"rewards/margins": 0.7008857131004333,
|
| 746 |
+
"rewards/rejected": -0.3455968499183655,
|
| 747 |
+
"step": 460
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.22,
|
| 751 |
+
"grad_norm": 8.375,
|
| 752 |
+
"learning_rate": 4.3856582166815696e-06,
|
| 753 |
+
"logits/chosen": -1.877435326576233,
|
| 754 |
+
"logits/rejected": -1.8740183115005493,
|
| 755 |
+
"logps/chosen": -33.043678283691406,
|
| 756 |
+
"logps/rejected": -33.61932373046875,
|
| 757 |
+
"loss": 0.4679,
|
| 758 |
+
"rewards/accuracies": 0.862500011920929,
|
| 759 |
+
"rewards/chosen": 0.3554477095603943,
|
| 760 |
+
"rewards/margins": 0.6336251497268677,
|
| 761 |
+
"rewards/rejected": -0.2781774401664734,
|
| 762 |
+
"step": 470
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.25,
|
| 766 |
+
"grad_norm": 8.5,
|
| 767 |
+
"learning_rate": 4.347971356735789e-06,
|
| 768 |
+
"logits/chosen": -2.00216007232666,
|
| 769 |
+
"logits/rejected": -1.9952083826065063,
|
| 770 |
+
"logps/chosen": -30.32107925415039,
|
| 771 |
+
"logps/rejected": -32.764427185058594,
|
| 772 |
+
"loss": 0.5165,
|
| 773 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 774 |
+
"rewards/chosen": 0.20298178493976593,
|
| 775 |
+
"rewards/margins": 0.5113586187362671,
|
| 776 |
+
"rewards/rejected": -0.30837681889533997,
|
| 777 |
+
"step": 480
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 1.27,
|
| 781 |
+
"grad_norm": 6.78125,
|
| 782 |
+
"learning_rate": 4.309335095262675e-06,
|
| 783 |
+
"logits/chosen": -1.943036437034607,
|
| 784 |
+
"logits/rejected": -1.9446159601211548,
|
| 785 |
+
"logps/chosen": -34.45885467529297,
|
| 786 |
+
"logps/rejected": -34.452476501464844,
|
| 787 |
+
"loss": 0.4477,
|
| 788 |
+
"rewards/accuracies": 0.862500011920929,
|
| 789 |
+
"rewards/chosen": 0.35773298144340515,
|
| 790 |
+
"rewards/margins": 0.6821189522743225,
|
| 791 |
+
"rewards/rejected": -0.32438600063323975,
|
| 792 |
+
"step": 490
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 1.3,
|
| 796 |
+
"grad_norm": 9.1875,
|
| 797 |
+
"learning_rate": 4.269769281772082e-06,
|
| 798 |
+
"logits/chosen": -1.8326492309570312,
|
| 799 |
+
"logits/rejected": -1.830275297164917,
|
| 800 |
+
"logps/chosen": -32.28834915161133,
|
| 801 |
+
"logps/rejected": -37.49669647216797,
|
| 802 |
+
"loss": 0.4361,
|
| 803 |
+
"rewards/accuracies": 0.862500011920929,
|
| 804 |
+
"rewards/chosen": 0.3535749316215515,
|
| 805 |
+
"rewards/margins": 0.766524076461792,
|
| 806 |
+
"rewards/rejected": -0.4129491448402405,
|
| 807 |
+
"step": 500
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 1.3,
|
| 811 |
+
"eval_logits/chosen": -2.173431634902954,
|
| 812 |
+
"eval_logits/rejected": -2.1686556339263916,
|
| 813 |
+
"eval_logps/chosen": -34.44293975830078,
|
| 814 |
+
"eval_logps/rejected": -38.02019500732422,
|
| 815 |
+
"eval_loss": 0.6902198195457458,
|
| 816 |
+
"eval_rewards/accuracies": 0.5365448594093323,
|
| 817 |
+
"eval_rewards/chosen": -0.16335560381412506,
|
| 818 |
+
"eval_rewards/margins": 0.03807440027594566,
|
| 819 |
+
"eval_rewards/rejected": -0.20143000781536102,
|
| 820 |
+
"eval_runtime": 145.4318,
|
| 821 |
+
"eval_samples_per_second": 2.358,
|
| 822 |
+
"eval_steps_per_second": 0.296,
|
| 823 |
+
"step": 500
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.32,
|
| 827 |
+
"grad_norm": 6.96875,
|
| 828 |
+
"learning_rate": 4.22929424333435e-06,
|
| 829 |
+
"logits/chosen": -1.9319264888763428,
|
| 830 |
+
"logits/rejected": -1.936668038368225,
|
| 831 |
+
"logps/chosen": -32.5339241027832,
|
| 832 |
+
"logps/rejected": -32.27501678466797,
|
| 833 |
+
"loss": 0.471,
|
| 834 |
+
"rewards/accuracies": 0.824999988079071,
|
| 835 |
+
"rewards/chosen": 0.3693718910217285,
|
| 836 |
+
"rewards/margins": 0.6209360361099243,
|
| 837 |
+
"rewards/rejected": -0.2515642046928406,
|
| 838 |
+
"step": 510
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.35,
|
| 842 |
+
"grad_norm": 7.53125,
|
| 843 |
+
"learning_rate": 4.1879307741372085e-06,
|
| 844 |
+
"logits/chosen": -1.9589513540267944,
|
| 845 |
+
"logits/rejected": -1.969935655593872,
|
| 846 |
+
"logps/chosen": -30.51776123046875,
|
| 847 |
+
"logps/rejected": -32.34286117553711,
|
| 848 |
+
"loss": 0.4588,
|
| 849 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 850 |
+
"rewards/chosen": 0.3490603566169739,
|
| 851 |
+
"rewards/margins": 0.6831297874450684,
|
| 852 |
+
"rewards/rejected": -0.3340694308280945,
|
| 853 |
+
"step": 520
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.38,
|
| 857 |
+
"grad_norm": 4.96875,
|
| 858 |
+
"learning_rate": 4.145700124802693e-06,
|
| 859 |
+
"logits/chosen": -1.8863223791122437,
|
| 860 |
+
"logits/rejected": -1.8831069469451904,
|
| 861 |
+
"logps/chosen": -31.701770782470703,
|
| 862 |
+
"logps/rejected": -33.229095458984375,
|
| 863 |
+
"loss": 0.4686,
|
| 864 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 865 |
+
"rewards/chosen": 0.2874954342842102,
|
| 866 |
+
"rewards/margins": 0.6340224742889404,
|
| 867 |
+
"rewards/rejected": -0.3465271592140198,
|
| 868 |
+
"step": 530
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 1.4,
|
| 872 |
+
"grad_norm": 6.9375,
|
| 873 |
+
"learning_rate": 4.102623991469562e-06,
|
| 874 |
+
"logits/chosen": -1.7517893314361572,
|
| 875 |
+
"logits/rejected": -1.7610219717025757,
|
| 876 |
+
"logps/chosen": -31.630868911743164,
|
| 877 |
+
"logps/rejected": -32.625404357910156,
|
| 878 |
+
"loss": 0.4697,
|
| 879 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 880 |
+
"rewards/chosen": 0.3754081726074219,
|
| 881 |
+
"rewards/margins": 0.6831758618354797,
|
| 882 |
+
"rewards/rejected": -0.3077676296234131,
|
| 883 |
+
"step": 540
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 1.43,
|
| 887 |
+
"grad_norm": 7.15625,
|
| 888 |
+
"learning_rate": 4.058724504646834e-06,
|
| 889 |
+
"logits/chosen": -1.8443384170532227,
|
| 890 |
+
"logits/rejected": -1.8382275104522705,
|
| 891 |
+
"logps/chosen": -32.76479721069336,
|
| 892 |
+
"logps/rejected": -31.679241180419922,
|
| 893 |
+
"loss": 0.4737,
|
| 894 |
+
"rewards/accuracies": 0.800000011920929,
|
| 895 |
+
"rewards/chosen": 0.3774031698703766,
|
| 896 |
+
"rewards/margins": 0.6525603532791138,
|
| 897 |
+
"rewards/rejected": -0.2751571536064148,
|
| 898 |
+
"step": 550
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 1.45,
|
| 902 |
+
"grad_norm": 6.75,
|
| 903 |
+
"learning_rate": 4.014024217844167e-06,
|
| 904 |
+
"logits/chosen": -1.932416319847107,
|
| 905 |
+
"logits/rejected": -1.9304752349853516,
|
| 906 |
+
"logps/chosen": -33.61432647705078,
|
| 907 |
+
"logps/rejected": -32.20708084106445,
|
| 908 |
+
"loss": 0.4728,
|
| 909 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 910 |
+
"rewards/chosen": 0.3827466368675232,
|
| 911 |
+
"rewards/margins": 0.6666241884231567,
|
| 912 |
+
"rewards/rejected": -0.2838776111602783,
|
| 913 |
+
"step": 560
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 1.48,
|
| 917 |
+
"grad_norm": 7.28125,
|
| 918 |
+
"learning_rate": 3.968546095984911e-06,
|
| 919 |
+
"logits/chosen": -1.7655744552612305,
|
| 920 |
+
"logits/rejected": -1.7635431289672852,
|
| 921 |
+
"logps/chosen": -31.894184112548828,
|
| 922 |
+
"logps/rejected": -31.694026947021484,
|
| 923 |
+
"loss": 0.4873,
|
| 924 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 925 |
+
"rewards/chosen": 0.3608909249305725,
|
| 926 |
+
"rewards/margins": 0.6375805735588074,
|
| 927 |
+
"rewards/rejected": -0.27668967843055725,
|
| 928 |
+
"step": 570
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 1.51,
|
| 932 |
+
"grad_norm": 10.4375,
|
| 933 |
+
"learning_rate": 3.922313503607806e-06,
|
| 934 |
+
"logits/chosen": -1.9005470275878906,
|
| 935 |
+
"logits/rejected": -1.8971643447875977,
|
| 936 |
+
"logps/chosen": -30.140361785888672,
|
| 937 |
+
"logps/rejected": -35.64023971557617,
|
| 938 |
+
"loss": 0.4311,
|
| 939 |
+
"rewards/accuracies": 0.887499988079071,
|
| 940 |
+
"rewards/chosen": 0.38571348786354065,
|
| 941 |
+
"rewards/margins": 0.7913459539413452,
|
| 942 |
+
"rewards/rejected": -0.40563249588012695,
|
| 943 |
+
"step": 580
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 1.53,
|
| 947 |
+
"grad_norm": 6.0,
|
| 948 |
+
"learning_rate": 3.875350192863368e-06,
|
| 949 |
+
"logits/chosen": -1.835296392440796,
|
| 950 |
+
"logits/rejected": -1.8387609720230103,
|
| 951 |
+
"logps/chosen": -28.92897605895996,
|
| 952 |
+
"logps/rejected": -31.326541900634766,
|
| 953 |
+
"loss": 0.4892,
|
| 954 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 955 |
+
"rewards/chosen": 0.2744215726852417,
|
| 956 |
+
"rewards/margins": 0.5414907336235046,
|
| 957 |
+
"rewards/rejected": -0.2670692205429077,
|
| 958 |
+
"step": 590
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 1.56,
|
| 962 |
+
"grad_norm": 7.34375,
|
| 963 |
+
"learning_rate": 3.8276802913111436e-06,
|
| 964 |
+
"logits/chosen": -1.870452880859375,
|
| 965 |
+
"logits/rejected": -1.8704124689102173,
|
| 966 |
+
"logps/chosen": -31.048049926757812,
|
| 967 |
+
"logps/rejected": -31.96176528930664,
|
| 968 |
+
"loss": 0.4356,
|
| 969 |
+
"rewards/accuracies": 0.887499988079071,
|
| 970 |
+
"rewards/chosen": 0.4693926274776459,
|
| 971 |
+
"rewards/margins": 0.776892364025116,
|
| 972 |
+
"rewards/rejected": -0.3074997365474701,
|
| 973 |
+
"step": 600
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 1.56,
|
| 977 |
+
"eval_logits/chosen": -2.1470563411712646,
|
| 978 |
+
"eval_logits/rejected": -2.1422781944274902,
|
| 979 |
+
"eval_logps/chosen": -34.333770751953125,
|
| 980 |
+
"eval_logps/rejected": -37.99900436401367,
|
| 981 |
+
"eval_loss": 0.6749633550643921,
|
| 982 |
+
"eval_rewards/accuracies": 0.5834717750549316,
|
| 983 |
+
"eval_rewards/chosen": -0.11968887597322464,
|
| 984 |
+
"eval_rewards/margins": 0.07326464354991913,
|
| 985 |
+
"eval_rewards/rejected": -0.19295351207256317,
|
| 986 |
+
"eval_runtime": 145.1843,
|
| 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": 7.53125,
|
| 994 |
+
"learning_rate": 3.7793282895240927e-06,
|
| 995 |
+
"logits/chosen": -1.9326798915863037,
|
| 996 |
+
"logits/rejected": -1.9334518909454346,
|
| 997 |
+
"logps/chosen": -33.740760803222656,
|
| 998 |
+
"logps/rejected": -34.0605354309082,
|
| 999 |
+
"loss": 0.4085,
|
| 1000 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1001 |
+
"rewards/chosen": 0.5157531499862671,
|
| 1002 |
+
"rewards/margins": 0.9062908887863159,
|
| 1003 |
+
"rewards/rejected": -0.3905377686023712,
|
| 1004 |
+
"step": 610
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 1.61,
|
| 1008 |
+
"grad_norm": 6.25,
|
| 1009 |
+
"learning_rate": 3.730319028506478e-06,
|
| 1010 |
+
"logits/chosen": -1.894683837890625,
|
| 1011 |
+
"logits/rejected": -1.8922332525253296,
|
| 1012 |
+
"logps/chosen": -31.973896026611328,
|
| 1013 |
+
"logps/rejected": -32.93384552001953,
|
| 1014 |
+
"loss": 0.4327,
|
| 1015 |
+
"rewards/accuracies": 0.875,
|
| 1016 |
+
"rewards/chosen": 0.460601270198822,
|
| 1017 |
+
"rewards/margins": 0.8061720132827759,
|
| 1018 |
+
"rewards/rejected": -0.34557071328163147,
|
| 1019 |
+
"step": 620
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 1.64,
|
| 1023 |
+
"grad_norm": 13.625,
|
| 1024 |
+
"learning_rate": 3.6806776869317074e-06,
|
| 1025 |
+
"logits/chosen": -1.9046722650527954,
|
| 1026 |
+
"logits/rejected": -1.8957364559173584,
|
| 1027 |
+
"logps/chosen": -31.66204833984375,
|
| 1028 |
+
"logps/rejected": -31.843969345092773,
|
| 1029 |
+
"loss": 0.4473,
|
| 1030 |
+
"rewards/accuracies": 0.8125,
|
| 1031 |
+
"rewards/chosen": 0.44562751054763794,
|
| 1032 |
+
"rewards/margins": 0.8090687990188599,
|
| 1033 |
+
"rewards/rejected": -0.36344125866889954,
|
| 1034 |
+
"step": 630
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 1.66,
|
| 1038 |
+
"grad_norm": 9.25,
|
| 1039 |
+
"learning_rate": 3.6304297682067146e-06,
|
| 1040 |
+
"logits/chosen": -1.905128836631775,
|
| 1041 |
+
"logits/rejected": -1.901825189590454,
|
| 1042 |
+
"logps/chosen": -31.2297306060791,
|
| 1043 |
+
"logps/rejected": -33.26097106933594,
|
| 1044 |
+
"loss": 0.4435,
|
| 1045 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1046 |
+
"rewards/chosen": 0.38747596740722656,
|
| 1047 |
+
"rewards/margins": 0.7240976095199585,
|
| 1048 |
+
"rewards/rejected": -0.3366217017173767,
|
| 1049 |
+
"step": 640
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 1.69,
|
| 1053 |
+
"grad_norm": 7.21875,
|
| 1054 |
+
"learning_rate": 3.579601087369492e-06,
|
| 1055 |
+
"logits/chosen": -1.9152917861938477,
|
| 1056 |
+
"logits/rejected": -1.9175786972045898,
|
| 1057 |
+
"logps/chosen": -32.58631134033203,
|
| 1058 |
+
"logps/rejected": -34.52180099487305,
|
| 1059 |
+
"loss": 0.421,
|
| 1060 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1061 |
+
"rewards/chosen": 0.426957368850708,
|
| 1062 |
+
"rewards/margins": 0.7623184323310852,
|
| 1063 |
+
"rewards/rejected": -0.3353610634803772,
|
| 1064 |
+
"step": 650
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 1.71,
|
| 1068 |
+
"grad_norm": 10.8125,
|
| 1069 |
+
"learning_rate": 3.5282177578265295e-06,
|
| 1070 |
+
"logits/chosen": -1.8232342004776,
|
| 1071 |
+
"logits/rejected": -1.8236335515975952,
|
| 1072 |
+
"logps/chosen": -32.828582763671875,
|
| 1073 |
+
"logps/rejected": -32.33021545410156,
|
| 1074 |
+
"loss": 0.4383,
|
| 1075 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1076 |
+
"rewards/chosen": 0.447300523519516,
|
| 1077 |
+
"rewards/margins": 0.736484169960022,
|
| 1078 |
+
"rewards/rejected": -0.28918370604515076,
|
| 1079 |
+
"step": 660
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 1.74,
|
| 1083 |
+
"grad_norm": 9.9375,
|
| 1084 |
+
"learning_rate": 3.476306177936961e-06,
|
| 1085 |
+
"logits/chosen": -1.8623815774917603,
|
| 1086 |
+
"logits/rejected": -1.852773666381836,
|
| 1087 |
+
"logps/chosen": -32.85457992553711,
|
| 1088 |
+
"logps/rejected": -33.26173782348633,
|
| 1089 |
+
"loss": 0.4009,
|
| 1090 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1091 |
+
"rewards/chosen": 0.4586804509162903,
|
| 1092 |
+
"rewards/margins": 0.8402576446533203,
|
| 1093 |
+
"rewards/rejected": -0.38157716393470764,
|
| 1094 |
+
"step": 670
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 1.77,
|
| 1098 |
+
"grad_norm": 6.28125,
|
| 1099 |
+
"learning_rate": 3.423893017450324e-06,
|
| 1100 |
+
"logits/chosen": -1.7598998546600342,
|
| 1101 |
+
"logits/rejected": -1.7566627264022827,
|
| 1102 |
+
"logps/chosen": -30.202777862548828,
|
| 1103 |
+
"logps/rejected": -35.252830505371094,
|
| 1104 |
+
"loss": 0.3982,
|
| 1105 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1106 |
+
"rewards/chosen": 0.5032743215560913,
|
| 1107 |
+
"rewards/margins": 0.8769756555557251,
|
| 1108 |
+
"rewards/rejected": -0.37370121479034424,
|
| 1109 |
+
"step": 680
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 1.79,
|
| 1113 |
+
"grad_norm": 5.84375,
|
| 1114 |
+
"learning_rate": 3.3710052038048794e-06,
|
| 1115 |
+
"logits/chosen": -1.8207229375839233,
|
| 1116 |
+
"logits/rejected": -1.8200151920318604,
|
| 1117 |
+
"logps/chosen": -33.87450408935547,
|
| 1118 |
+
"logps/rejected": -36.40219497680664,
|
| 1119 |
+
"loss": 0.3646,
|
| 1120 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1121 |
+
"rewards/chosen": 0.5583134293556213,
|
| 1122 |
+
"rewards/margins": 0.9776619076728821,
|
| 1123 |
+
"rewards/rejected": -0.41934847831726074,
|
| 1124 |
+
"step": 690
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 1.82,
|
| 1128 |
+
"grad_norm": 5.9375,
|
| 1129 |
+
"learning_rate": 3.3176699082935546e-06,
|
| 1130 |
+
"logits/chosen": -1.7926433086395264,
|
| 1131 |
+
"logits/rejected": -1.7952711582183838,
|
| 1132 |
+
"logps/chosen": -31.316104888916016,
|
| 1133 |
+
"logps/rejected": -36.66993713378906,
|
| 1134 |
+
"loss": 0.4036,
|
| 1135 |
+
"rewards/accuracies": 0.875,
|
| 1136 |
+
"rewards/chosen": 0.51469886302948,
|
| 1137 |
+
"rewards/margins": 0.9260055422782898,
|
| 1138 |
+
"rewards/rejected": -0.4113067090511322,
|
| 1139 |
+
"step": 700
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 1.82,
|
| 1143 |
+
"eval_logits/chosen": -2.1232047080993652,
|
| 1144 |
+
"eval_logits/rejected": -2.11846923828125,
|
| 1145 |
+
"eval_logps/chosen": -34.40776824951172,
|
| 1146 |
+
"eval_logps/rejected": -38.10317611694336,
|
| 1147 |
+
"eval_loss": 0.6718631386756897,
|
| 1148 |
+
"eval_rewards/accuracies": 0.5859634280204773,
|
| 1149 |
+
"eval_rewards/chosen": -0.14928743243217468,
|
| 1150 |
+
"eval_rewards/margins": 0.08533468097448349,
|
| 1151 |
+
"eval_rewards/rejected": -0.23462210595607758,
|
| 1152 |
+
"eval_runtime": 145.3862,
|
| 1153 |
+
"eval_samples_per_second": 2.359,
|
| 1154 |
+
"eval_steps_per_second": 0.296,
|
| 1155 |
+
"step": 700
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 1.84,
|
| 1159 |
+
"grad_norm": 8.4375,
|
| 1160 |
+
"learning_rate": 3.2639145321045933e-06,
|
| 1161 |
+
"logits/chosen": -1.8965635299682617,
|
| 1162 |
+
"logits/rejected": -1.8991920948028564,
|
| 1163 |
+
"logps/chosen": -33.687984466552734,
|
| 1164 |
+
"logps/rejected": -35.19561004638672,
|
| 1165 |
+
"loss": 0.4379,
|
| 1166 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1167 |
+
"rewards/chosen": 0.483041912317276,
|
| 1168 |
+
"rewards/margins": 0.8532260060310364,
|
| 1169 |
+
"rewards/rejected": -0.3701840341091156,
|
| 1170 |
+
"step": 710
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 1.87,
|
| 1174 |
+
"grad_norm": 8.5625,
|
| 1175 |
+
"learning_rate": 3.2097666922441107e-06,
|
| 1176 |
+
"logits/chosen": -1.7496124505996704,
|
| 1177 |
+
"logits/rejected": -1.7437019348144531,
|
| 1178 |
+
"logps/chosen": -33.61560821533203,
|
| 1179 |
+
"logps/rejected": -33.33930587768555,
|
| 1180 |
+
"loss": 0.4064,
|
| 1181 |
+
"rewards/accuracies": 0.875,
|
| 1182 |
+
"rewards/chosen": 0.4870489239692688,
|
| 1183 |
+
"rewards/margins": 0.882034182548523,
|
| 1184 |
+
"rewards/rejected": -0.3949853479862213,
|
| 1185 |
+
"step": 720
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.9,
|
| 1189 |
+
"grad_norm": 5.875,
|
| 1190 |
+
"learning_rate": 3.1552542073477554e-06,
|
| 1191 |
+
"logits/chosen": -1.923055648803711,
|
| 1192 |
+
"logits/rejected": -1.919946312904358,
|
| 1193 |
+
"logps/chosen": -29.562047958374023,
|
| 1194 |
+
"logps/rejected": -32.68622970581055,
|
| 1195 |
+
"loss": 0.3979,
|
| 1196 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1197 |
+
"rewards/chosen": 0.4363761842250824,
|
| 1198 |
+
"rewards/margins": 0.9315892457962036,
|
| 1199 |
+
"rewards/rejected": -0.49521297216415405,
|
| 1200 |
+
"step": 730
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 1.92,
|
| 1204 |
+
"grad_norm": 6.71875,
|
| 1205 |
+
"learning_rate": 3.100405083388799e-06,
|
| 1206 |
+
"logits/chosen": -1.7599384784698486,
|
| 1207 |
+
"logits/rejected": -1.7601182460784912,
|
| 1208 |
+
"logps/chosen": -32.07990264892578,
|
| 1209 |
+
"logps/rejected": -38.58934783935547,
|
| 1210 |
+
"loss": 0.3994,
|
| 1211 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1212 |
+
"rewards/chosen": 0.5533084869384766,
|
| 1213 |
+
"rewards/margins": 0.9379235506057739,
|
| 1214 |
+
"rewards/rejected": -0.38461512327194214,
|
| 1215 |
+
"step": 740
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 1.95,
|
| 1219 |
+
"grad_norm": 5.59375,
|
| 1220 |
+
"learning_rate": 3.0452474992899645e-06,
|
| 1221 |
+
"logits/chosen": -1.6459096670150757,
|
| 1222 |
+
"logits/rejected": -1.651227355003357,
|
| 1223 |
+
"logps/chosen": -35.65028762817383,
|
| 1224 |
+
"logps/rejected": -35.04454803466797,
|
| 1225 |
+
"loss": 0.4218,
|
| 1226 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1227 |
+
"rewards/chosen": 0.5128704905509949,
|
| 1228 |
+
"rewards/margins": 0.9417527914047241,
|
| 1229 |
+
"rewards/rejected": -0.428882360458374,
|
| 1230 |
+
"step": 750
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 1.97,
|
| 1234 |
+
"grad_norm": 8.125,
|
| 1235 |
+
"learning_rate": 2.989809792446417e-06,
|
| 1236 |
+
"logits/chosen": -1.8414586782455444,
|
| 1237 |
+
"logits/rejected": -1.8431018590927124,
|
| 1238 |
+
"logps/chosen": -31.405654907226562,
|
| 1239 |
+
"logps/rejected": -33.640296936035156,
|
| 1240 |
+
"loss": 0.4105,
|
| 1241 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1242 |
+
"rewards/chosen": 0.48001009225845337,
|
| 1243 |
+
"rewards/margins": 0.8244431614875793,
|
| 1244 |
+
"rewards/rejected": -0.34443309903144836,
|
| 1245 |
+
"step": 760
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 2.0,
|
| 1249 |
+
"grad_norm": 8.1875,
|
| 1250 |
+
"learning_rate": 2.9341204441673267e-06,
|
| 1251 |
+
"logits/chosen": -1.8179328441619873,
|
| 1252 |
+
"logits/rejected": -1.8168070316314697,
|
| 1253 |
+
"logps/chosen": -30.935861587524414,
|
| 1254 |
+
"logps/rejected": -35.76841354370117,
|
| 1255 |
+
"loss": 0.433,
|
| 1256 |
+
"rewards/accuracies": 0.85833340883255,
|
| 1257 |
+
"rewards/chosen": 0.4958992004394531,
|
| 1258 |
+
"rewards/margins": 0.7920029759407043,
|
| 1259 |
+
"rewards/rejected": -0.2961038649082184,
|
| 1260 |
+
"step": 770
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 2.03,
|
| 1264 |
+
"grad_norm": 5.3125,
|
| 1265 |
+
"learning_rate": 2.878208065043501e-06,
|
| 1266 |
+
"logits/chosen": -1.8167203664779663,
|
| 1267 |
+
"logits/rejected": -1.8160884380340576,
|
| 1268 |
+
"logps/chosen": -33.24877166748047,
|
| 1269 |
+
"logps/rejected": -32.76862716674805,
|
| 1270 |
+
"loss": 0.3406,
|
| 1271 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1272 |
+
"rewards/chosen": 0.5598764419555664,
|
| 1273 |
+
"rewards/margins": 1.0995185375213623,
|
| 1274 |
+
"rewards/rejected": -0.5396420359611511,
|
| 1275 |
+
"step": 780
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 2.05,
|
| 1279 |
+
"grad_norm": 6.75,
|
| 1280 |
+
"learning_rate": 2.8221013802485974e-06,
|
| 1281 |
+
"logits/chosen": -1.8443057537078857,
|
| 1282 |
+
"logits/rejected": -1.842907190322876,
|
| 1283 |
+
"logps/chosen": -28.221912384033203,
|
| 1284 |
+
"logps/rejected": -34.14670944213867,
|
| 1285 |
+
"loss": 0.3187,
|
| 1286 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1287 |
+
"rewards/chosen": 0.536507248878479,
|
| 1288 |
+
"rewards/margins": 1.140665054321289,
|
| 1289 |
+
"rewards/rejected": -0.6041578054428101,
|
| 1290 |
+
"step": 790
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 2.08,
|
| 1294 |
+
"grad_norm": 5.0625,
|
| 1295 |
+
"learning_rate": 2.76582921478147e-06,
|
| 1296 |
+
"logits/chosen": -1.8954391479492188,
|
| 1297 |
+
"logits/rejected": -1.892087697982788,
|
| 1298 |
+
"logps/chosen": -30.82273292541504,
|
| 1299 |
+
"logps/rejected": -35.755470275878906,
|
| 1300 |
+
"loss": 0.2952,
|
| 1301 |
+
"rewards/accuracies": 0.9375,
|
| 1302 |
+
"rewards/chosen": 0.6417907476425171,
|
| 1303 |
+
"rewards/margins": 1.2643665075302124,
|
| 1304 |
+
"rewards/rejected": -0.6225756406784058,
|
| 1305 |
+
"step": 800
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 2.08,
|
| 1309 |
+
"eval_logits/chosen": -2.1026289463043213,
|
| 1310 |
+
"eval_logits/rejected": -2.097952127456665,
|
| 1311 |
+
"eval_logps/chosen": -34.54673767089844,
|
| 1312 |
+
"eval_logps/rejected": -38.26842498779297,
|
| 1313 |
+
"eval_loss": 0.6757632493972778,
|
| 1314 |
+
"eval_rewards/accuracies": 0.5776578187942505,
|
| 1315 |
+
"eval_rewards/chosen": -0.20487497746944427,
|
| 1316 |
+
"eval_rewards/margins": 0.0958474650979042,
|
| 1317 |
+
"eval_rewards/rejected": -0.3007224202156067,
|
| 1318 |
+
"eval_runtime": 145.3563,
|
| 1319 |
+
"eval_samples_per_second": 2.36,
|
| 1320 |
+
"eval_steps_per_second": 0.296,
|
| 1321 |
+
"step": 800
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 2.1,
|
| 1325 |
+
"grad_norm": 5.71875,
|
| 1326 |
+
"learning_rate": 2.7094204786572254e-06,
|
| 1327 |
+
"logits/chosen": -1.721960425376892,
|
| 1328 |
+
"logits/rejected": -1.7143983840942383,
|
| 1329 |
+
"logps/chosen": -32.55122375488281,
|
| 1330 |
+
"logps/rejected": -36.20327377319336,
|
| 1331 |
+
"loss": 0.2827,
|
| 1332 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1333 |
+
"rewards/chosen": 0.7620068192481995,
|
| 1334 |
+
"rewards/margins": 1.4192689657211304,
|
| 1335 |
+
"rewards/rejected": -0.6572622060775757,
|
| 1336 |
+
"step": 810
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 2.13,
|
| 1340 |
+
"grad_norm": 6.25,
|
| 1341 |
+
"learning_rate": 2.6529041520546072e-06,
|
| 1342 |
+
"logits/chosen": -1.797978401184082,
|
| 1343 |
+
"logits/rejected": -1.8082430362701416,
|
| 1344 |
+
"logps/chosen": -33.965431213378906,
|
| 1345 |
+
"logps/rejected": -33.7777214050293,
|
| 1346 |
+
"loss": 0.3245,
|
| 1347 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1348 |
+
"rewards/chosen": 0.6921030282974243,
|
| 1349 |
+
"rewards/margins": 1.2306172847747803,
|
| 1350 |
+
"rewards/rejected": -0.5385143160820007,
|
| 1351 |
+
"step": 820
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 2.16,
|
| 1355 |
+
"grad_norm": 7.71875,
|
| 1356 |
+
"learning_rate": 2.5963092704273302e-06,
|
| 1357 |
+
"logits/chosen": -1.847389817237854,
|
| 1358 |
+
"logits/rejected": -1.8522322177886963,
|
| 1359 |
+
"logps/chosen": -33.48334503173828,
|
| 1360 |
+
"logps/rejected": -30.594324111938477,
|
| 1361 |
+
"loss": 0.3479,
|
| 1362 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1363 |
+
"rewards/chosen": 0.5677754282951355,
|
| 1364 |
+
"rewards/margins": 1.0982062816619873,
|
| 1365 |
+
"rewards/rejected": -0.5304308533668518,
|
| 1366 |
+
"step": 830
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 2.18,
|
| 1370 |
+
"grad_norm": 5.40625,
|
| 1371 |
+
"learning_rate": 2.53966490958702e-06,
|
| 1372 |
+
"logits/chosen": -1.846261978149414,
|
| 1373 |
+
"logits/rejected": -1.8545395135879517,
|
| 1374 |
+
"logps/chosen": -33.14823913574219,
|
| 1375 |
+
"logps/rejected": -31.363750457763672,
|
| 1376 |
+
"loss": 0.3104,
|
| 1377 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1378 |
+
"rewards/chosen": 0.6559160947799683,
|
| 1379 |
+
"rewards/margins": 1.2623378038406372,
|
| 1380 |
+
"rewards/rejected": -0.6064217686653137,
|
| 1381 |
+
"step": 840
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 2.21,
|
| 1385 |
+
"grad_norm": 6.96875,
|
| 1386 |
+
"learning_rate": 2.4830001707654135e-06,
|
| 1387 |
+
"logits/chosen": -1.7743536233901978,
|
| 1388 |
+
"logits/rejected": -1.7648853063583374,
|
| 1389 |
+
"logps/chosen": -30.302188873291016,
|
| 1390 |
+
"logps/rejected": -33.442298889160156,
|
| 1391 |
+
"loss": 0.329,
|
| 1392 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1393 |
+
"rewards/chosen": 0.5964034795761108,
|
| 1394 |
+
"rewards/margins": 1.1678011417388916,
|
| 1395 |
+
"rewards/rejected": -0.5713975429534912,
|
| 1396 |
+
"step": 850
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 2.23,
|
| 1400 |
+
"grad_norm": 4.78125,
|
| 1401 |
+
"learning_rate": 2.4263441656635054e-06,
|
| 1402 |
+
"logits/chosen": -1.9121280908584595,
|
| 1403 |
+
"logits/rejected": -1.9022724628448486,
|
| 1404 |
+
"logps/chosen": -24.69386100769043,
|
| 1405 |
+
"logps/rejected": -31.476959228515625,
|
| 1406 |
+
"loss": 0.3155,
|
| 1407 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1408 |
+
"rewards/chosen": 0.6120836138725281,
|
| 1409 |
+
"rewards/margins": 1.2539312839508057,
|
| 1410 |
+
"rewards/rejected": -0.6418476104736328,
|
| 1411 |
+
"step": 860
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 2.26,
|
| 1415 |
+
"grad_norm": 5.78125,
|
| 1416 |
+
"learning_rate": 2.3697260014953107e-06,
|
| 1417 |
+
"logits/chosen": -1.7656368017196655,
|
| 1418 |
+
"logits/rejected": -1.7670552730560303,
|
| 1419 |
+
"logps/chosen": -32.631500244140625,
|
| 1420 |
+
"logps/rejected": -31.436681747436523,
|
| 1421 |
+
"loss": 0.3053,
|
| 1422 |
+
"rewards/accuracies": 0.9375,
|
| 1423 |
+
"rewards/chosen": 0.6806014180183411,
|
| 1424 |
+
"rewards/margins": 1.2622734308242798,
|
| 1425 |
+
"rewards/rejected": -0.5816720724105835,
|
| 1426 |
+
"step": 870
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 2.29,
|
| 1430 |
+
"grad_norm": 6.125,
|
| 1431 |
+
"learning_rate": 2.3131747660339396e-06,
|
| 1432 |
+
"logits/chosen": -1.7689151763916016,
|
| 1433 |
+
"logits/rejected": -1.7699298858642578,
|
| 1434 |
+
"logps/chosen": -31.100326538085938,
|
| 1435 |
+
"logps/rejected": -34.68801498413086,
|
| 1436 |
+
"loss": 0.3503,
|
| 1437 |
+
"rewards/accuracies": 0.875,
|
| 1438 |
+
"rewards/chosen": 0.515569806098938,
|
| 1439 |
+
"rewards/margins": 1.1704087257385254,
|
| 1440 |
+
"rewards/rejected": -0.6548389196395874,
|
| 1441 |
+
"step": 880
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 2.31,
|
| 1445 |
+
"grad_norm": 5.5625,
|
| 1446 |
+
"learning_rate": 2.256719512667651e-06,
|
| 1447 |
+
"logits/chosen": -1.6725788116455078,
|
| 1448 |
+
"logits/rejected": -1.671099066734314,
|
| 1449 |
+
"logps/chosen": -34.11555099487305,
|
| 1450 |
+
"logps/rejected": -37.912750244140625,
|
| 1451 |
+
"loss": 0.2954,
|
| 1452 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1453 |
+
"rewards/chosen": 0.6197448968887329,
|
| 1454 |
+
"rewards/margins": 1.4339497089385986,
|
| 1455 |
+
"rewards/rejected": -0.8142046928405762,
|
| 1456 |
+
"step": 890
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 2.34,
|
| 1460 |
+
"grad_norm": 6.25,
|
| 1461 |
+
"learning_rate": 2.2003892454735786e-06,
|
| 1462 |
+
"logits/chosen": -1.8256248235702515,
|
| 1463 |
+
"logits/rejected": -1.8188546895980835,
|
| 1464 |
+
"logps/chosen": -30.36061668395996,
|
| 1465 |
+
"logps/rejected": -34.45724105834961,
|
| 1466 |
+
"loss": 0.304,
|
| 1467 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1468 |
+
"rewards/chosen": 0.6568080186843872,
|
| 1469 |
+
"rewards/margins": 1.3364773988723755,
|
| 1470 |
+
"rewards/rejected": -0.6796693801879883,
|
| 1471 |
+
"step": 900
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 2.34,
|
| 1475 |
+
"eval_logits/chosen": -2.0867063999176025,
|
| 1476 |
+
"eval_logits/rejected": -2.082029342651367,
|
| 1477 |
+
"eval_logps/chosen": -34.61775207519531,
|
| 1478 |
+
"eval_logps/rejected": -38.34459686279297,
|
| 1479 |
+
"eval_loss": 0.6778721809387207,
|
| 1480 |
+
"eval_rewards/accuracies": 0.595099687576294,
|
| 1481 |
+
"eval_rewards/chosen": -0.23328028619289398,
|
| 1482 |
+
"eval_rewards/margins": 0.09790942072868347,
|
| 1483 |
+
"eval_rewards/rejected": -0.33118972182273865,
|
| 1484 |
+
"eval_runtime": 145.1775,
|
| 1485 |
+
"eval_samples_per_second": 2.363,
|
| 1486 |
+
"eval_steps_per_second": 0.296,
|
| 1487 |
+
"step": 900
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 2.36,
|
| 1491 |
+
"grad_norm": 6.46875,
|
| 1492 |
+
"learning_rate": 2.1442129043167877e-06,
|
| 1493 |
+
"logits/chosen": -1.858393907546997,
|
| 1494 |
+
"logits/rejected": -1.8537986278533936,
|
| 1495 |
+
"logps/chosen": -31.981470108032227,
|
| 1496 |
+
"logps/rejected": -36.285892486572266,
|
| 1497 |
+
"loss": 0.2951,
|
| 1498 |
+
"rewards/accuracies": 0.9375,
|
| 1499 |
+
"rewards/chosen": 0.6304842233657837,
|
| 1500 |
+
"rewards/margins": 1.3333914279937744,
|
| 1501 |
+
"rewards/rejected": -0.7029072642326355,
|
| 1502 |
+
"step": 910
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 2.39,
|
| 1506 |
+
"grad_norm": 5.8125,
|
| 1507 |
+
"learning_rate": 2.088219349982323e-06,
|
| 1508 |
+
"logits/chosen": -1.777714729309082,
|
| 1509 |
+
"logits/rejected": -1.7827917337417603,
|
| 1510 |
+
"logps/chosen": -33.658485412597656,
|
| 1511 |
+
"logps/rejected": -34.12433624267578,
|
| 1512 |
+
"loss": 0.2995,
|
| 1513 |
+
"rewards/accuracies": 0.9375,
|
| 1514 |
+
"rewards/chosen": 0.6760393977165222,
|
| 1515 |
+
"rewards/margins": 1.3258110284805298,
|
| 1516 |
+
"rewards/rejected": -0.649771511554718,
|
| 1517 |
+
"step": 920
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 2.42,
|
| 1521 |
+
"grad_norm": 5.0,
|
| 1522 |
+
"learning_rate": 2.0324373493478803e-06,
|
| 1523 |
+
"logits/chosen": -1.8777525424957275,
|
| 1524 |
+
"logits/rejected": -1.8685945272445679,
|
| 1525 |
+
"logps/chosen": -30.748672485351562,
|
| 1526 |
+
"logps/rejected": -36.116886138916016,
|
| 1527 |
+
"loss": 0.294,
|
| 1528 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1529 |
+
"rewards/chosen": 0.614503026008606,
|
| 1530 |
+
"rewards/margins": 1.2841379642486572,
|
| 1531 |
+
"rewards/rejected": -0.6696349382400513,
|
| 1532 |
+
"step": 930
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 2.44,
|
| 1536 |
+
"grad_norm": 7.125,
|
| 1537 |
+
"learning_rate": 1.976895560604729e-06,
|
| 1538 |
+
"logits/chosen": -1.8024845123291016,
|
| 1539 |
+
"logits/rejected": -1.7992064952850342,
|
| 1540 |
+
"logps/chosen": -30.062992095947266,
|
| 1541 |
+
"logps/rejected": -33.89337921142578,
|
| 1542 |
+
"loss": 0.356,
|
| 1543 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1544 |
+
"rewards/chosen": 0.49018430709838867,
|
| 1545 |
+
"rewards/margins": 1.0613863468170166,
|
| 1546 |
+
"rewards/rejected": -0.5712020993232727,
|
| 1547 |
+
"step": 940
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 2.47,
|
| 1551 |
+
"grad_norm": 4.4375,
|
| 1552 |
+
"learning_rate": 1.921622518534466e-06,
|
| 1553 |
+
"logits/chosen": -1.73916757106781,
|
| 1554 |
+
"logits/rejected": -1.7464996576309204,
|
| 1555 |
+
"logps/chosen": -31.560144424438477,
|
| 1556 |
+
"logps/rejected": -37.22822189331055,
|
| 1557 |
+
"loss": 0.3426,
|
| 1558 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1559 |
+
"rewards/chosen": 0.5243637561798096,
|
| 1560 |
+
"rewards/margins": 1.1956799030303955,
|
| 1561 |
+
"rewards/rejected": -0.6713162660598755,
|
| 1562 |
+
"step": 950
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 2.49,
|
| 1566 |
+
"grad_norm": 4.5,
|
| 1567 |
+
"learning_rate": 1.8666466198491794e-06,
|
| 1568 |
+
"logits/chosen": -1.7508230209350586,
|
| 1569 |
+
"logits/rejected": -1.7439250946044922,
|
| 1570 |
+
"logps/chosen": -32.305728912353516,
|
| 1571 |
+
"logps/rejected": -38.560791015625,
|
| 1572 |
+
"loss": 0.2871,
|
| 1573 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1574 |
+
"rewards/chosen": 0.6091153621673584,
|
| 1575 |
+
"rewards/margins": 1.3939114809036255,
|
| 1576 |
+
"rewards/rejected": -0.784795880317688,
|
| 1577 |
+
"step": 960
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 2.52,
|
| 1581 |
+
"grad_norm": 5.75,
|
| 1582 |
+
"learning_rate": 1.8119961086025376e-06,
|
| 1583 |
+
"logits/chosen": -1.752611517906189,
|
| 1584 |
+
"logits/rejected": -1.7530120611190796,
|
| 1585 |
+
"logps/chosen": -29.592453002929688,
|
| 1586 |
+
"logps/rejected": -34.51519775390625,
|
| 1587 |
+
"loss": 0.3156,
|
| 1588 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1589 |
+
"rewards/chosen": 0.6575800180435181,
|
| 1590 |
+
"rewards/margins": 1.2711557149887085,
|
| 1591 |
+
"rewards/rejected": -0.6135755777359009,
|
| 1592 |
+
"step": 970
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 2.55,
|
| 1596 |
+
"grad_norm": 6.53125,
|
| 1597 |
+
"learning_rate": 1.7576990616793139e-06,
|
| 1598 |
+
"logits/chosen": -1.7550010681152344,
|
| 1599 |
+
"logits/rejected": -1.7661269903182983,
|
| 1600 |
+
"logps/chosen": -31.351938247680664,
|
| 1601 |
+
"logps/rejected": -35.77216339111328,
|
| 1602 |
+
"loss": 0.3031,
|
| 1603 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1604 |
+
"rewards/chosen": 0.68990159034729,
|
| 1605 |
+
"rewards/margins": 1.3462820053100586,
|
| 1606 |
+
"rewards/rejected": -0.6563804149627686,
|
| 1607 |
+
"step": 980
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 2.57,
|
| 1611 |
+
"grad_norm": 4.59375,
|
| 1612 |
+
"learning_rate": 1.7037833743707892e-06,
|
| 1613 |
+
"logits/chosen": -1.8242708444595337,
|
| 1614 |
+
"logits/rejected": -1.821933388710022,
|
| 1615 |
+
"logps/chosen": -34.63307189941406,
|
| 1616 |
+
"logps/rejected": -33.89019012451172,
|
| 1617 |
+
"loss": 0.3595,
|
| 1618 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1619 |
+
"rewards/chosen": 0.5951918363571167,
|
| 1620 |
+
"rewards/margins": 1.1915771961212158,
|
| 1621 |
+
"rewards/rejected": -0.5963853597640991,
|
| 1622 |
+
"step": 990
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 2.6,
|
| 1626 |
+
"grad_norm": 6.53125,
|
| 1627 |
+
"learning_rate": 1.6502767460434588e-06,
|
| 1628 |
+
"logits/chosen": -1.7967815399169922,
|
| 1629 |
+
"logits/rejected": -1.8010671138763428,
|
| 1630 |
+
"logps/chosen": -32.9234619140625,
|
| 1631 |
+
"logps/rejected": -35.8967399597168,
|
| 1632 |
+
"loss": 0.333,
|
| 1633 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1634 |
+
"rewards/chosen": 0.6159319281578064,
|
| 1635 |
+
"rewards/margins": 1.1802124977111816,
|
| 1636 |
+
"rewards/rejected": -0.5642803907394409,
|
| 1637 |
+
"step": 1000
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 2.6,
|
| 1641 |
+
"eval_logits/chosen": -2.072319984436035,
|
| 1642 |
+
"eval_logits/rejected": -2.0676779747009277,
|
| 1643 |
+
"eval_logps/chosen": -34.67308044433594,
|
| 1644 |
+
"eval_logps/rejected": -38.430641174316406,
|
| 1645 |
+
"eval_loss": 0.676977813243866,
|
| 1646 |
+
"eval_rewards/accuracies": 0.5859634280204773,
|
| 1647 |
+
"eval_rewards/chosen": -0.2554103434085846,
|
| 1648 |
+
"eval_rewards/margins": 0.11019979417324066,
|
| 1649 |
+
"eval_rewards/rejected": -0.36561012268066406,
|
| 1650 |
+
"eval_runtime": 145.2025,
|
| 1651 |
+
"eval_samples_per_second": 2.362,
|
| 1652 |
+
"eval_steps_per_second": 0.296,
|
| 1653 |
+
"step": 1000
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 2.62,
|
| 1657 |
+
"grad_norm": 9.9375,
|
| 1658 |
+
"learning_rate": 1.5972066659083796e-06,
|
| 1659 |
+
"logits/chosen": -1.8064178228378296,
|
| 1660 |
+
"logits/rejected": -1.811234712600708,
|
| 1661 |
+
"logps/chosen": -31.44000244140625,
|
| 1662 |
+
"logps/rejected": -33.90888977050781,
|
| 1663 |
+
"loss": 0.3712,
|
| 1664 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1665 |
+
"rewards/chosen": 0.5441684126853943,
|
| 1666 |
+
"rewards/margins": 1.0992209911346436,
|
| 1667 |
+
"rewards/rejected": -0.5550524592399597,
|
| 1668 |
+
"step": 1010
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 2.65,
|
| 1672 |
+
"grad_norm": 7.46875,
|
| 1673 |
+
"learning_rate": 1.5446003988985041e-06,
|
| 1674 |
+
"logits/chosen": -1.7619049549102783,
|
| 1675 |
+
"logits/rejected": -1.7642395496368408,
|
| 1676 |
+
"logps/chosen": -28.927509307861328,
|
| 1677 |
+
"logps/rejected": -33.50086212158203,
|
| 1678 |
+
"loss": 0.2941,
|
| 1679 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1680 |
+
"rewards/chosen": 0.6072515845298767,
|
| 1681 |
+
"rewards/margins": 1.3279320001602173,
|
| 1682 |
+
"rewards/rejected": -0.7206803560256958,
|
| 1683 |
+
"step": 1020
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 2.68,
|
| 1687 |
+
"grad_norm": 4.5,
|
| 1688 |
+
"learning_rate": 1.4924849716612211e-06,
|
| 1689 |
+
"logits/chosen": -1.8155397176742554,
|
| 1690 |
+
"logits/rejected": -1.8104407787322998,
|
| 1691 |
+
"logps/chosen": -32.632835388183594,
|
| 1692 |
+
"logps/rejected": -35.2443962097168,
|
| 1693 |
+
"loss": 0.3301,
|
| 1694 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1695 |
+
"rewards/chosen": 0.5425110459327698,
|
| 1696 |
+
"rewards/margins": 1.233070969581604,
|
| 1697 |
+
"rewards/rejected": -0.6905598640441895,
|
| 1698 |
+
"step": 1030
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 2.7,
|
| 1702 |
+
"grad_norm": 9.5,
|
| 1703 |
+
"learning_rate": 1.440887158673332e-06,
|
| 1704 |
+
"logits/chosen": -1.7551313638687134,
|
| 1705 |
+
"logits/rejected": -1.7592474222183228,
|
| 1706 |
+
"logps/chosen": -34.41304397583008,
|
| 1707 |
+
"logps/rejected": -35.99690246582031,
|
| 1708 |
+
"loss": 0.3295,
|
| 1709 |
+
"rewards/accuracies": 0.9375,
|
| 1710 |
+
"rewards/chosen": 0.5362745523452759,
|
| 1711 |
+
"rewards/margins": 1.167203664779663,
|
| 1712 |
+
"rewards/rejected": -0.6309291124343872,
|
| 1713 |
+
"step": 1040
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 2.73,
|
| 1717 |
+
"grad_norm": 5.75,
|
| 1718 |
+
"learning_rate": 1.3898334684855647e-06,
|
| 1719 |
+
"logits/chosen": -1.751050591468811,
|
| 1720 |
+
"logits/rejected": -1.7637557983398438,
|
| 1721 |
+
"logps/chosen": -30.462039947509766,
|
| 1722 |
+
"logps/rejected": -34.61774444580078,
|
| 1723 |
+
"loss": 0.3249,
|
| 1724 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1725 |
+
"rewards/chosen": 0.5603402853012085,
|
| 1726 |
+
"rewards/margins": 1.1994216442108154,
|
| 1727 |
+
"rewards/rejected": -0.6390813589096069,
|
| 1728 |
+
"step": 1050
|
| 1729 |
+
},
|
| 1730 |
+
{
|
| 1731 |
+
"epoch": 2.75,
|
| 1732 |
+
"grad_norm": 9.4375,
|
| 1733 |
+
"learning_rate": 1.3393501301037245e-06,
|
| 1734 |
+
"logits/chosen": -1.7376810312271118,
|
| 1735 |
+
"logits/rejected": -1.7309871912002563,
|
| 1736 |
+
"logps/chosen": -30.55047035217285,
|
| 1737 |
+
"logps/rejected": -34.8571662902832,
|
| 1738 |
+
"loss": 0.3058,
|
| 1739 |
+
"rewards/accuracies": 0.9375,
|
| 1740 |
+
"rewards/chosen": 0.6435432434082031,
|
| 1741 |
+
"rewards/margins": 1.259224534034729,
|
| 1742 |
+
"rewards/rejected": -0.6156812310218811,
|
| 1743 |
+
"step": 1060
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 2.78,
|
| 1747 |
+
"grad_norm": 6.1875,
|
| 1748 |
+
"learning_rate": 1.2894630795134454e-06,
|
| 1749 |
+
"logits/chosen": -1.8616443872451782,
|
| 1750 |
+
"logits/rejected": -1.8623626232147217,
|
| 1751 |
+
"logps/chosen": -32.02442169189453,
|
| 1752 |
+
"logps/rejected": -34.386295318603516,
|
| 1753 |
+
"loss": 0.3145,
|
| 1754 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1755 |
+
"rewards/chosen": 0.6090625524520874,
|
| 1756 |
+
"rewards/margins": 1.2257022857666016,
|
| 1757 |
+
"rewards/rejected": -0.6166397333145142,
|
| 1758 |
+
"step": 1070
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 2.81,
|
| 1762 |
+
"grad_norm": 6.59375,
|
| 1763 |
+
"learning_rate": 1.2401979463554984e-06,
|
| 1764 |
+
"logits/chosen": -1.7933332920074463,
|
| 1765 |
+
"logits/rejected": -1.7921574115753174,
|
| 1766 |
+
"logps/chosen": -32.67540740966797,
|
| 1767 |
+
"logps/rejected": -34.564029693603516,
|
| 1768 |
+
"loss": 0.3561,
|
| 1769 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1770 |
+
"rewards/chosen": 0.5657684206962585,
|
| 1771 |
+
"rewards/margins": 1.1146466732025146,
|
| 1772 |
+
"rewards/rejected": -0.5488781929016113,
|
| 1773 |
+
"step": 1080
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 2.83,
|
| 1777 |
+
"grad_norm": 5.84375,
|
| 1778 |
+
"learning_rate": 1.1915800407584705e-06,
|
| 1779 |
+
"logits/chosen": -1.8151838779449463,
|
| 1780 |
+
"logits/rejected": -1.8076585531234741,
|
| 1781 |
+
"logps/chosen": -32.7107048034668,
|
| 1782 |
+
"logps/rejected": -32.47800827026367,
|
| 1783 |
+
"loss": 0.3302,
|
| 1784 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1785 |
+
"rewards/chosen": 0.6060681939125061,
|
| 1786 |
+
"rewards/margins": 1.1806586980819702,
|
| 1787 |
+
"rewards/rejected": -0.5745903849601746,
|
| 1788 |
+
"step": 1090
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 2.86,
|
| 1792 |
+
"grad_norm": 4.59375,
|
| 1793 |
+
"learning_rate": 1.1436343403356019e-06,
|
| 1794 |
+
"logits/chosen": -1.7889270782470703,
|
| 1795 |
+
"logits/rejected": -1.787825584411621,
|
| 1796 |
+
"logps/chosen": -33.79990005493164,
|
| 1797 |
+
"logps/rejected": -38.28570556640625,
|
| 1798 |
+
"loss": 0.2846,
|
| 1799 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1800 |
+
"rewards/chosen": 0.6474052667617798,
|
| 1801 |
+
"rewards/margins": 1.4206043481826782,
|
| 1802 |
+
"rewards/rejected": -0.7731989622116089,
|
| 1803 |
+
"step": 1100
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 2.86,
|
| 1807 |
+
"eval_logits/chosen": -2.069221258163452,
|
| 1808 |
+
"eval_logits/rejected": -2.0645899772644043,
|
| 1809 |
+
"eval_logps/chosen": -34.66777420043945,
|
| 1810 |
+
"eval_logps/rejected": -38.41979217529297,
|
| 1811 |
+
"eval_loss": 0.6772051453590393,
|
| 1812 |
+
"eval_rewards/accuracies": 0.6009136438369751,
|
| 1813 |
+
"eval_rewards/chosen": -0.2532878816127777,
|
| 1814 |
+
"eval_rewards/margins": 0.10798129439353943,
|
| 1815 |
+
"eval_rewards/rejected": -0.36126917600631714,
|
| 1816 |
+
"eval_runtime": 145.1766,
|
| 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": 6.9375,
|
| 1824 |
+
"learning_rate": 1.0963854773524548e-06,
|
| 1825 |
+
"logits/chosen": -1.836913824081421,
|
| 1826 |
+
"logits/rejected": -1.8424360752105713,
|
| 1827 |
+
"logps/chosen": -34.19898223876953,
|
| 1828 |
+
"logps/rejected": -37.27791213989258,
|
| 1829 |
+
"loss": 0.3393,
|
| 1830 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1831 |
+
"rewards/chosen": 0.5072540044784546,
|
| 1832 |
+
"rewards/margins": 1.200984239578247,
|
| 1833 |
+
"rewards/rejected": -0.6937301158905029,
|
| 1834 |
+
"step": 1110
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 2.91,
|
| 1838 |
+
"grad_norm": 5.09375,
|
| 1839 |
+
"learning_rate": 1.049857726072005e-06,
|
| 1840 |
+
"logits/chosen": -1.8054929971694946,
|
| 1841 |
+
"logits/rejected": -1.8034013509750366,
|
| 1842 |
+
"logps/chosen": -31.228893280029297,
|
| 1843 |
+
"logps/rejected": -34.67009353637695,
|
| 1844 |
+
"loss": 0.3249,
|
| 1845 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1846 |
+
"rewards/chosen": 0.5549246668815613,
|
| 1847 |
+
"rewards/margins": 1.229860544204712,
|
| 1848 |
+
"rewards/rejected": -0.6749356985092163,
|
| 1849 |
+
"step": 1120
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 2.94,
|
| 1853 |
+
"grad_norm": 4.78125,
|
| 1854 |
+
"learning_rate": 1.0040749902836508e-06,
|
| 1855 |
+
"logits/chosen": -1.7074657678604126,
|
| 1856 |
+
"logits/rejected": -1.710054636001587,
|
| 1857 |
+
"logps/chosen": -28.319087982177734,
|
| 1858 |
+
"logps/rejected": -32.21437454223633,
|
| 1859 |
+
"loss": 0.2869,
|
| 1860 |
+
"rewards/accuracies": 0.9375,
|
| 1861 |
+
"rewards/chosen": 0.5718191266059875,
|
| 1862 |
+
"rewards/margins": 1.3595529794692993,
|
| 1863 |
+
"rewards/rejected": -0.7877337336540222,
|
| 1864 |
+
"step": 1130
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 2.96,
|
| 1868 |
+
"grad_norm": 5.8125,
|
| 1869 |
+
"learning_rate": 9.59060791022566e-07,
|
| 1870 |
+
"logits/chosen": -1.8046897649765015,
|
| 1871 |
+
"logits/rejected": -1.8014905452728271,
|
| 1872 |
+
"logps/chosen": -31.595922470092773,
|
| 1873 |
+
"logps/rejected": -33.86729431152344,
|
| 1874 |
+
"loss": 0.3331,
|
| 1875 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1876 |
+
"rewards/chosen": 0.6431301832199097,
|
| 1877 |
+
"rewards/margins": 1.1792962551116943,
|
| 1878 |
+
"rewards/rejected": -0.5361660718917847,
|
| 1879 |
+
"step": 1140
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 2.99,
|
| 1883 |
+
"grad_norm": 6.40625,
|
| 1884 |
+
"learning_rate": 9.148382544856885e-07,
|
| 1885 |
+
"logits/chosen": -1.7242072820663452,
|
| 1886 |
+
"logits/rejected": -1.7179205417633057,
|
| 1887 |
+
"logps/chosen": -26.89887046813965,
|
| 1888 |
+
"logps/rejected": -33.96310806274414,
|
| 1889 |
+
"loss": 0.3234,
|
| 1890 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1891 |
+
"rewards/chosen": 0.49564680457115173,
|
| 1892 |
+
"rewards/margins": 1.2215702533721924,
|
| 1893 |
+
"rewards/rejected": -0.7259235382080078,
|
| 1894 |
+
"step": 1150
|
| 1895 |
+
},
|
| 1896 |
+
{
|
| 1897 |
+
"epoch": 3.01,
|
| 1898 |
+
"grad_norm": 5.34375,
|
| 1899 |
+
"learning_rate": 8.714301001505568e-07,
|
| 1900 |
+
"logits/chosen": -1.833547830581665,
|
| 1901 |
+
"logits/rejected": -1.8309831619262695,
|
| 1902 |
+
"logps/chosen": -31.566417694091797,
|
| 1903 |
+
"logps/rejected": -36.94977569580078,
|
| 1904 |
+
"loss": 0.3005,
|
| 1905 |
+
"rewards/accuracies": 0.9458333849906921,
|
| 1906 |
+
"rewards/chosen": 0.611860990524292,
|
| 1907 |
+
"rewards/margins": 1.3551568984985352,
|
| 1908 |
+
"rewards/rejected": -0.7432958483695984,
|
| 1909 |
+
"step": 1160
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 3.04,
|
| 1913 |
+
"grad_norm": 4.875,
|
| 1914 |
+
"learning_rate": 8.288586291031025e-07,
|
| 1915 |
+
"logits/chosen": -1.8228060007095337,
|
| 1916 |
+
"logits/rejected": -1.8206332921981812,
|
| 1917 |
+
"logps/chosen": -30.116840362548828,
|
| 1918 |
+
"logps/rejected": -34.350860595703125,
|
| 1919 |
+
"loss": 0.2812,
|
| 1920 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1921 |
+
"rewards/chosen": 0.5990268588066101,
|
| 1922 |
+
"rewards/margins": 1.3937623500823975,
|
| 1923 |
+
"rewards/rejected": -0.7947354912757874,
|
| 1924 |
+
"step": 1170
|
| 1925 |
+
},
|
| 1926 |
+
{
|
| 1927 |
+
"epoch": 3.06,
|
| 1928 |
+
"grad_norm": 4.65625,
|
| 1929 |
+
"learning_rate": 7.871457125803897e-07,
|
| 1930 |
+
"logits/chosen": -1.8162040710449219,
|
| 1931 |
+
"logits/rejected": -1.804120659828186,
|
| 1932 |
+
"logps/chosen": -34.508934020996094,
|
| 1933 |
+
"logps/rejected": -36.088863372802734,
|
| 1934 |
+
"loss": 0.2775,
|
| 1935 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1936 |
+
"rewards/chosen": 0.6145464181900024,
|
| 1937 |
+
"rewards/margins": 1.4215012788772583,
|
| 1938 |
+
"rewards/rejected": -0.8069549798965454,
|
| 1939 |
+
"step": 1180
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 3.09,
|
| 1943 |
+
"grad_norm": 5.625,
|
| 1944 |
+
"learning_rate": 7.463127807341966e-07,
|
| 1945 |
+
"logits/chosen": -1.7128658294677734,
|
| 1946 |
+
"logits/rejected": -1.7075669765472412,
|
| 1947 |
+
"logps/chosen": -33.452003479003906,
|
| 1948 |
+
"logps/rejected": -35.91048812866211,
|
| 1949 |
+
"loss": 0.2738,
|
| 1950 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1951 |
+
"rewards/chosen": 0.7590751647949219,
|
| 1952 |
+
"rewards/margins": 1.42814040184021,
|
| 1953 |
+
"rewards/rejected": -0.6690651774406433,
|
| 1954 |
+
"step": 1190
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 3.12,
|
| 1958 |
+
"grad_norm": 5.125,
|
| 1959 |
+
"learning_rate": 7.063808116212021e-07,
|
| 1960 |
+
"logits/chosen": -1.7600603103637695,
|
| 1961 |
+
"logits/rejected": -1.7632497549057007,
|
| 1962 |
+
"logps/chosen": -31.24569320678711,
|
| 1963 |
+
"logps/rejected": -33.11017608642578,
|
| 1964 |
+
"loss": 0.3301,
|
| 1965 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1966 |
+
"rewards/chosen": 0.5477610230445862,
|
| 1967 |
+
"rewards/margins": 1.1973613500595093,
|
| 1968 |
+
"rewards/rejected": -0.6496003866195679,
|
| 1969 |
+
"step": 1200
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 3.12,
|
| 1973 |
+
"eval_logits/chosen": -2.0681374073028564,
|
| 1974 |
+
"eval_logits/rejected": -2.0634970664978027,
|
| 1975 |
+
"eval_logps/chosen": -34.671016693115234,
|
| 1976 |
+
"eval_logps/rejected": -38.42964172363281,
|
| 1977 |
+
"eval_loss": 0.6771031618118286,
|
| 1978 |
+
"eval_rewards/accuracies": 0.5776578187942505,
|
| 1979 |
+
"eval_rewards/chosen": -0.25458672642707825,
|
| 1980 |
+
"eval_rewards/margins": 0.11062110960483551,
|
| 1981 |
+
"eval_rewards/rejected": -0.36520785093307495,
|
| 1982 |
+
"eval_runtime": 145.3934,
|
| 1983 |
+
"eval_samples_per_second": 2.359,
|
| 1984 |
+
"eval_steps_per_second": 0.296,
|
| 1985 |
+
"step": 1200
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 3.14,
|
| 1989 |
+
"grad_norm": 6.3125,
|
| 1990 |
+
"learning_rate": 6.673703204254348e-07,
|
| 1991 |
+
"logits/chosen": -1.7789433002471924,
|
| 1992 |
+
"logits/rejected": -1.7744804620742798,
|
| 1993 |
+
"logps/chosen": -29.76243019104004,
|
| 1994 |
+
"logps/rejected": -32.72071075439453,
|
| 1995 |
+
"loss": 0.2864,
|
| 1996 |
+
"rewards/accuracies": 0.9375,
|
| 1997 |
+
"rewards/chosen": 0.6044963002204895,
|
| 1998 |
+
"rewards/margins": 1.3295423984527588,
|
| 1999 |
+
"rewards/rejected": -0.7250461578369141,
|
| 2000 |
+
"step": 1210
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 3.17,
|
| 2004 |
+
"grad_norm": 5.03125,
|
| 2005 |
+
"learning_rate": 6.293013489185315e-07,
|
| 2006 |
+
"logits/chosen": -1.7943885326385498,
|
| 2007 |
+
"logits/rejected": -1.7889045476913452,
|
| 2008 |
+
"logps/chosen": -33.75127410888672,
|
| 2009 |
+
"logps/rejected": -36.586204528808594,
|
| 2010 |
+
"loss": 0.2973,
|
| 2011 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2012 |
+
"rewards/chosen": 0.6090667247772217,
|
| 2013 |
+
"rewards/margins": 1.4155981540679932,
|
| 2014 |
+
"rewards/rejected": -0.8065314292907715,
|
| 2015 |
+
"step": 1220
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 3.19,
|
| 2019 |
+
"grad_norm": 4.9375,
|
| 2020 |
+
"learning_rate": 5.921934551632086e-07,
|
| 2021 |
+
"logits/chosen": -1.7955095767974854,
|
| 2022 |
+
"logits/rejected": -1.782134771347046,
|
| 2023 |
+
"logps/chosen": -31.721553802490234,
|
| 2024 |
+
"logps/rejected": -36.032196044921875,
|
| 2025 |
+
"loss": 0.3156,
|
| 2026 |
+
"rewards/accuracies": 0.862500011920929,
|
| 2027 |
+
"rewards/chosen": 0.5489839315414429,
|
| 2028 |
+
"rewards/margins": 1.3139245510101318,
|
| 2029 |
+
"rewards/rejected": -0.764940619468689,
|
| 2030 |
+
"step": 1230
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 3.22,
|
| 2034 |
+
"grad_norm": 4.5625,
|
| 2035 |
+
"learning_rate": 5.560657034652405e-07,
|
| 2036 |
+
"logits/chosen": -1.8425153493881226,
|
| 2037 |
+
"logits/rejected": -1.8402408361434937,
|
| 2038 |
+
"logps/chosen": -33.340415954589844,
|
| 2039 |
+
"logps/rejected": -33.357208251953125,
|
| 2040 |
+
"loss": 0.2976,
|
| 2041 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 2042 |
+
"rewards/chosen": 0.6769941449165344,
|
| 2043 |
+
"rewards/margins": 1.2839314937591553,
|
| 2044 |
+
"rewards/rejected": -0.6069372296333313,
|
| 2045 |
+
"step": 1240
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 3.25,
|
| 2049 |
+
"grad_norm": 4.4375,
|
| 2050 |
+
"learning_rate": 5.2093665457911e-07,
|
| 2051 |
+
"logits/chosen": -1.7631075382232666,
|
| 2052 |
+
"logits/rejected": -1.7602026462554932,
|
| 2053 |
+
"logps/chosen": -32.7774658203125,
|
| 2054 |
+
"logps/rejected": -37.03419876098633,
|
| 2055 |
+
"loss": 0.2631,
|
| 2056 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2057 |
+
"rewards/chosen": 0.6474747061729431,
|
| 2058 |
+
"rewards/margins": 1.5423027276992798,
|
| 2059 |
+
"rewards/rejected": -0.8948280215263367,
|
| 2060 |
+
"step": 1250
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 3.27,
|
| 2064 |
+
"grad_norm": 5.0,
|
| 2065 |
+
"learning_rate": 4.868243561723535e-07,
|
| 2066 |
+
"logits/chosen": -1.836340308189392,
|
| 2067 |
+
"logits/rejected": -1.8310245275497437,
|
| 2068 |
+
"logps/chosen": -29.116525650024414,
|
| 2069 |
+
"logps/rejected": -34.146141052246094,
|
| 2070 |
+
"loss": 0.2836,
|
| 2071 |
+
"rewards/accuracies": 0.9375,
|
| 2072 |
+
"rewards/chosen": 0.6577095985412598,
|
| 2073 |
+
"rewards/margins": 1.3631455898284912,
|
| 2074 |
+
"rewards/rejected": -0.7054358720779419,
|
| 2075 |
+
"step": 1260
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 3.3,
|
| 2079 |
+
"grad_norm": 4.90625,
|
| 2080 |
+
"learning_rate": 4.537463335535161e-07,
|
| 2081 |
+
"logits/chosen": -1.8614356517791748,
|
| 2082 |
+
"logits/rejected": -1.8663842678070068,
|
| 2083 |
+
"logps/chosen": -30.975250244140625,
|
| 2084 |
+
"logps/rejected": -33.32194137573242,
|
| 2085 |
+
"loss": 0.2845,
|
| 2086 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2087 |
+
"rewards/chosen": 0.6942988634109497,
|
| 2088 |
+
"rewards/margins": 1.350528359413147,
|
| 2089 |
+
"rewards/rejected": -0.656229555606842,
|
| 2090 |
+
"step": 1270
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 3.32,
|
| 2094 |
+
"grad_norm": 6.34375,
|
| 2095 |
+
"learning_rate": 4.217195806684629e-07,
|
| 2096 |
+
"logits/chosen": -1.7541017532348633,
|
| 2097 |
+
"logits/rejected": -1.761639952659607,
|
| 2098 |
+
"logps/chosen": -33.46928024291992,
|
| 2099 |
+
"logps/rejected": -34.43730163574219,
|
| 2100 |
+
"loss": 0.3088,
|
| 2101 |
+
"rewards/accuracies": 0.9375,
|
| 2102 |
+
"rewards/chosen": 0.5751846432685852,
|
| 2103 |
+
"rewards/margins": 1.3107969760894775,
|
| 2104 |
+
"rewards/rejected": -0.7356122732162476,
|
| 2105 |
+
"step": 1280
|
| 2106 |
+
},
|
| 2107 |
+
{
|
| 2108 |
+
"epoch": 3.35,
|
| 2109 |
+
"grad_norm": 5.78125,
|
| 2110 |
+
"learning_rate": 3.907605513696808e-07,
|
| 2111 |
+
"logits/chosen": -1.6694676876068115,
|
| 2112 |
+
"logits/rejected": -1.671893835067749,
|
| 2113 |
+
"logps/chosen": -31.393539428710938,
|
| 2114 |
+
"logps/rejected": -38.413612365722656,
|
| 2115 |
+
"loss": 0.2938,
|
| 2116 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2117 |
+
"rewards/chosen": 0.6953436732292175,
|
| 2118 |
+
"rewards/margins": 1.4390814304351807,
|
| 2119 |
+
"rewards/rejected": -0.7437376976013184,
|
| 2120 |
+
"step": 1290
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 3.38,
|
| 2124 |
+
"grad_norm": 5.34375,
|
| 2125 |
+
"learning_rate": 3.6088515096305675e-07,
|
| 2126 |
+
"logits/chosen": -1.7161693572998047,
|
| 2127 |
+
"logits/rejected": -1.719686508178711,
|
| 2128 |
+
"logps/chosen": -31.36373519897461,
|
| 2129 |
+
"logps/rejected": -34.30319595336914,
|
| 2130 |
+
"loss": 0.2648,
|
| 2131 |
+
"rewards/accuracies": 0.987500011920929,
|
| 2132 |
+
"rewards/chosen": 0.741172194480896,
|
| 2133 |
+
"rewards/margins": 1.4767869710922241,
|
| 2134 |
+
"rewards/rejected": -0.7356146574020386,
|
| 2135 |
+
"step": 1300
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 3.38,
|
| 2139 |
+
"eval_logits/chosen": -2.067815065383911,
|
| 2140 |
+
"eval_logits/rejected": -2.0631678104400635,
|
| 2141 |
+
"eval_logps/chosen": -34.68254089355469,
|
| 2142 |
+
"eval_logps/rejected": -38.4415397644043,
|
| 2143 |
+
"eval_loss": 0.6774489283561707,
|
| 2144 |
+
"eval_rewards/accuracies": 0.5834717750549316,
|
| 2145 |
+
"eval_rewards/chosen": -0.2591961622238159,
|
| 2146 |
+
"eval_rewards/margins": 0.11077102273702621,
|
| 2147 |
+
"eval_rewards/rejected": -0.3699672222137451,
|
| 2148 |
+
"eval_runtime": 145.3523,
|
| 2149 |
+
"eval_samples_per_second": 2.36,
|
| 2150 |
+
"eval_steps_per_second": 0.296,
|
| 2151 |
+
"step": 1300
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 3.4,
|
| 2155 |
+
"grad_norm": 3.765625,
|
| 2156 |
+
"learning_rate": 3.321087280364757e-07,
|
| 2157 |
+
"logits/chosen": -1.8252456188201904,
|
| 2158 |
+
"logits/rejected": -1.8048315048217773,
|
| 2159 |
+
"logps/chosen": -29.41412925720215,
|
| 2160 |
+
"logps/rejected": -37.30854415893555,
|
| 2161 |
+
"loss": 0.2709,
|
| 2162 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2163 |
+
"rewards/chosen": 0.6797068119049072,
|
| 2164 |
+
"rewards/margins": 1.5365630388259888,
|
| 2165 |
+
"rewards/rejected": -0.8568561673164368,
|
| 2166 |
+
"step": 1310
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 3.43,
|
| 2170 |
+
"grad_norm": 4.90625,
|
| 2171 |
+
"learning_rate": 3.044460665744284e-07,
|
| 2172 |
+
"logits/chosen": -1.799944519996643,
|
| 2173 |
+
"logits/rejected": -1.8057048320770264,
|
| 2174 |
+
"logps/chosen": -31.14408302307129,
|
| 2175 |
+
"logps/rejected": -34.24466323852539,
|
| 2176 |
+
"loss": 0.2632,
|
| 2177 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2178 |
+
"rewards/chosen": 0.7558823823928833,
|
| 2179 |
+
"rewards/margins": 1.4574472904205322,
|
| 2180 |
+
"rewards/rejected": -0.7015649080276489,
|
| 2181 |
+
"step": 1320
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 3.45,
|
| 2185 |
+
"grad_norm": 4.59375,
|
| 2186 |
+
"learning_rate": 2.779113783626916e-07,
|
| 2187 |
+
"logits/chosen": -1.7793442010879517,
|
| 2188 |
+
"logits/rejected": -1.7743394374847412,
|
| 2189 |
+
"logps/chosen": -31.542285919189453,
|
| 2190 |
+
"logps/rejected": -36.73578643798828,
|
| 2191 |
+
"loss": 0.2666,
|
| 2192 |
+
"rewards/accuracies": 0.9375,
|
| 2193 |
+
"rewards/chosen": 0.5920882821083069,
|
| 2194 |
+
"rewards/margins": 1.4574869871139526,
|
| 2195 |
+
"rewards/rejected": -0.8653987646102905,
|
| 2196 |
+
"step": 1330
|
| 2197 |
+
},
|
| 2198 |
+
{
|
| 2199 |
+
"epoch": 3.48,
|
| 2200 |
+
"grad_norm": 4.3125,
|
| 2201 |
+
"learning_rate": 2.5251829568697204e-07,
|
| 2202 |
+
"logits/chosen": -1.6390937566757202,
|
| 2203 |
+
"logits/rejected": -1.648639440536499,
|
| 2204 |
+
"logps/chosen": -31.974777221679688,
|
| 2205 |
+
"logps/rejected": -32.479408264160156,
|
| 2206 |
+
"loss": 0.3034,
|
| 2207 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2208 |
+
"rewards/chosen": 0.7299244999885559,
|
| 2209 |
+
"rewards/margins": 1.3080428838729858,
|
| 2210 |
+
"rewards/rejected": -0.5781184434890747,
|
| 2211 |
+
"step": 1340
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 3.51,
|
| 2215 |
+
"grad_norm": 5.25,
|
| 2216 |
+
"learning_rate": 2.2827986432927774e-07,
|
| 2217 |
+
"logits/chosen": -1.6959750652313232,
|
| 2218 |
+
"logits/rejected": -1.7006906270980835,
|
| 2219 |
+
"logps/chosen": -32.08546447753906,
|
| 2220 |
+
"logps/rejected": -34.80610656738281,
|
| 2221 |
+
"loss": 0.2624,
|
| 2222 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 2223 |
+
"rewards/chosen": 0.7461883425712585,
|
| 2224 |
+
"rewards/margins": 1.3867065906524658,
|
| 2225 |
+
"rewards/rejected": -0.6405184268951416,
|
| 2226 |
+
"step": 1350
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 3.53,
|
| 2230 |
+
"grad_norm": 11.0625,
|
| 2231 |
+
"learning_rate": 2.0520853686560177e-07,
|
| 2232 |
+
"logits/chosen": -1.7251228094100952,
|
| 2233 |
+
"logits/rejected": -1.7304246425628662,
|
| 2234 |
+
"logps/chosen": -32.209083557128906,
|
| 2235 |
+
"logps/rejected": -36.53895950317383,
|
| 2236 |
+
"loss": 0.2908,
|
| 2237 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2238 |
+
"rewards/chosen": 0.6490303874015808,
|
| 2239 |
+
"rewards/margins": 1.4393888711929321,
|
| 2240 |
+
"rewards/rejected": -0.790358304977417,
|
| 2241 |
+
"step": 1360
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 3.56,
|
| 2245 |
+
"grad_norm": 3.953125,
|
| 2246 |
+
"learning_rate": 1.833161662683672e-07,
|
| 2247 |
+
"logits/chosen": -1.8194055557250977,
|
| 2248 |
+
"logits/rejected": -1.8149350881576538,
|
| 2249 |
+
"logps/chosen": -31.16534423828125,
|
| 2250 |
+
"logps/rejected": -32.77348709106445,
|
| 2251 |
+
"loss": 0.2875,
|
| 2252 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2253 |
+
"rewards/chosen": 0.5881365537643433,
|
| 2254 |
+
"rewards/margins": 1.3327710628509521,
|
| 2255 |
+
"rewards/rejected": -0.7446345090866089,
|
| 2256 |
+
"step": 1370
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 3.58,
|
| 2260 |
+
"grad_norm": 6.03125,
|
| 2261 |
+
"learning_rate": 1.626139998169246e-07,
|
| 2262 |
+
"logits/chosen": -1.7297048568725586,
|
| 2263 |
+
"logits/rejected": -1.7319259643554688,
|
| 2264 |
+
"logps/chosen": -28.14961814880371,
|
| 2265 |
+
"logps/rejected": -32.501651763916016,
|
| 2266 |
+
"loss": 0.299,
|
| 2267 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2268 |
+
"rewards/chosen": 0.6038740277290344,
|
| 2269 |
+
"rewards/margins": 1.306862235069275,
|
| 2270 |
+
"rewards/rejected": -0.7029882669448853,
|
| 2271 |
+
"step": 1380
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 3.61,
|
| 2275 |
+
"grad_norm": 6.15625,
|
| 2276 |
+
"learning_rate": 1.4311267331922535e-07,
|
| 2277 |
+
"logits/chosen": -1.7326571941375732,
|
| 2278 |
+
"logits/rejected": -1.7373113632202148,
|
| 2279 |
+
"logps/chosen": -30.526447296142578,
|
| 2280 |
+
"logps/rejected": -34.334877014160156,
|
| 2281 |
+
"loss": 0.322,
|
| 2282 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2283 |
+
"rewards/chosen": 0.5972187519073486,
|
| 2284 |
+
"rewards/margins": 1.1359431743621826,
|
| 2285 |
+
"rewards/rejected": -0.5387245416641235,
|
| 2286 |
+
"step": 1390
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 3.64,
|
| 2290 |
+
"grad_norm": 4.5,
|
| 2291 |
+
"learning_rate": 1.2482220564763669e-07,
|
| 2292 |
+
"logits/chosen": -1.7946516275405884,
|
| 2293 |
+
"logits/rejected": -1.7978696823120117,
|
| 2294 |
+
"logps/chosen": -33.16937255859375,
|
| 2295 |
+
"logps/rejected": -35.18634033203125,
|
| 2296 |
+
"loss": 0.2661,
|
| 2297 |
+
"rewards/accuracies": 0.987500011920929,
|
| 2298 |
+
"rewards/chosen": 0.7044798135757446,
|
| 2299 |
+
"rewards/margins": 1.409590244293213,
|
| 2300 |
+
"rewards/rejected": -0.7051103711128235,
|
| 2301 |
+
"step": 1400
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 3.64,
|
| 2305 |
+
"eval_logits/chosen": -2.068145990371704,
|
| 2306 |
+
"eval_logits/rejected": -2.0635111331939697,
|
| 2307 |
+
"eval_logps/chosen": -34.682861328125,
|
| 2308 |
+
"eval_logps/rejected": -38.453575134277344,
|
| 2309 |
+
"eval_loss": 0.6737242937088013,
|
| 2310 |
+
"eval_rewards/accuracies": 0.5888704061508179,
|
| 2311 |
+
"eval_rewards/chosen": -0.25932323932647705,
|
| 2312 |
+
"eval_rewards/margins": 0.11545901745557785,
|
| 2313 |
+
"eval_rewards/rejected": -0.3747822642326355,
|
| 2314 |
+
"eval_runtime": 145.2165,
|
| 2315 |
+
"eval_samples_per_second": 2.362,
|
| 2316 |
+
"eval_steps_per_second": 0.296,
|
| 2317 |
+
"step": 1400
|
| 2318 |
+
},
|
| 2319 |
+
{
|
| 2320 |
+
"epoch": 3.66,
|
| 2321 |
+
"grad_norm": 5.5625,
|
| 2322 |
+
"learning_rate": 1.0775199359171346e-07,
|
| 2323 |
+
"logits/chosen": -1.861598253250122,
|
| 2324 |
+
"logits/rejected": -1.8546861410140991,
|
| 2325 |
+
"logps/chosen": -32.03349685668945,
|
| 2326 |
+
"logps/rejected": -35.89234924316406,
|
| 2327 |
+
"loss": 0.2836,
|
| 2328 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2329 |
+
"rewards/chosen": 0.7076212763786316,
|
| 2330 |
+
"rewards/margins": 1.3667352199554443,
|
| 2331 |
+
"rewards/rejected": -0.6591139435768127,
|
| 2332 |
+
"step": 1410
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 3.69,
|
| 2336 |
+
"grad_norm": 7.25,
|
| 2337 |
+
"learning_rate": 9.191080703056604e-08,
|
| 2338 |
+
"logits/chosen": -1.754303216934204,
|
| 2339 |
+
"logits/rejected": -1.7654063701629639,
|
| 2340 |
+
"logps/chosen": -32.40322494506836,
|
| 2341 |
+
"logps/rejected": -35.30275344848633,
|
| 2342 |
+
"loss": 0.2752,
|
| 2343 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2344 |
+
"rewards/chosen": 0.6984306573867798,
|
| 2345 |
+
"rewards/margins": 1.383098840713501,
|
| 2346 |
+
"rewards/rejected": -0.6846679449081421,
|
| 2347 |
+
"step": 1420
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 3.71,
|
| 2351 |
+
"grad_norm": 6.09375,
|
| 2352 |
+
"learning_rate": 7.730678442730539e-08,
|
| 2353 |
+
"logits/chosen": -1.8308874368667603,
|
| 2354 |
+
"logits/rejected": -1.8427495956420898,
|
| 2355 |
+
"logps/chosen": -33.228111267089844,
|
| 2356 |
+
"logps/rejected": -35.49666213989258,
|
| 2357 |
+
"loss": 0.2652,
|
| 2358 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2359 |
+
"rewards/chosen": 0.7330778241157532,
|
| 2360 |
+
"rewards/margins": 1.476409912109375,
|
| 2361 |
+
"rewards/rejected": -0.743332028388977,
|
| 2362 |
+
"step": 1430
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 3.74,
|
| 2366 |
+
"grad_norm": 6.875,
|
| 2367 |
+
"learning_rate": 6.394742864787806e-08,
|
| 2368 |
+
"logits/chosen": -1.8214858770370483,
|
| 2369 |
+
"logits/rejected": -1.823809266090393,
|
| 2370 |
+
"logps/chosen": -31.59054183959961,
|
| 2371 |
+
"logps/rejected": -35.78374481201172,
|
| 2372 |
+
"loss": 0.2894,
|
| 2373 |
+
"rewards/accuracies": 0.9375,
|
| 2374 |
+
"rewards/chosen": 0.6471369862556458,
|
| 2375 |
+
"rewards/margins": 1.4166474342346191,
|
| 2376 |
+
"rewards/rejected": -0.7695104479789734,
|
| 2377 |
+
"step": 1440
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 3.77,
|
| 2381 |
+
"grad_norm": 4.71875,
|
| 2382 |
+
"learning_rate": 5.183960310644748e-08,
|
| 2383 |
+
"logits/chosen": -1.8054840564727783,
|
| 2384 |
+
"logits/rejected": -1.798434853553772,
|
| 2385 |
+
"logps/chosen": -33.86231231689453,
|
| 2386 |
+
"logps/rejected": -35.45820617675781,
|
| 2387 |
+
"loss": 0.2973,
|
| 2388 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2389 |
+
"rewards/chosen": 0.682405948638916,
|
| 2390 |
+
"rewards/margins": 1.3575280904769897,
|
| 2391 |
+
"rewards/rejected": -0.6751221418380737,
|
| 2392 |
+
"step": 1450
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"epoch": 3.79,
|
| 2396 |
+
"grad_norm": 5.15625,
|
| 2397 |
+
"learning_rate": 4.098952823928693e-08,
|
| 2398 |
+
"logits/chosen": -1.772991418838501,
|
| 2399 |
+
"logits/rejected": -1.7789325714111328,
|
| 2400 |
+
"logps/chosen": -29.6343994140625,
|
| 2401 |
+
"logps/rejected": -35.68324661254883,
|
| 2402 |
+
"loss": 0.2694,
|
| 2403 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2404 |
+
"rewards/chosen": 0.6315112709999084,
|
| 2405 |
+
"rewards/margins": 1.4368103742599487,
|
| 2406 |
+
"rewards/rejected": -0.8052991032600403,
|
| 2407 |
+
"step": 1460
|
| 2408 |
+
},
|
| 2409 |
+
{
|
| 2410 |
+
"epoch": 3.82,
|
| 2411 |
+
"grad_norm": 4.875,
|
| 2412 |
+
"learning_rate": 3.1402778309014284e-08,
|
| 2413 |
+
"logits/chosen": -1.7217044830322266,
|
| 2414 |
+
"logits/rejected": -1.719930648803711,
|
| 2415 |
+
"logps/chosen": -29.1605281829834,
|
| 2416 |
+
"logps/rejected": -32.21342086791992,
|
| 2417 |
+
"loss": 0.2994,
|
| 2418 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2419 |
+
"rewards/chosen": 0.6740103363990784,
|
| 2420 |
+
"rewards/margins": 1.3216867446899414,
|
| 2421 |
+
"rewards/rejected": -0.6476765275001526,
|
| 2422 |
+
"step": 1470
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 3.84,
|
| 2426 |
+
"grad_norm": 5.96875,
|
| 2427 |
+
"learning_rate": 2.3084278540791427e-08,
|
| 2428 |
+
"logits/chosen": -1.896645188331604,
|
| 2429 |
+
"logits/rejected": -1.8910773992538452,
|
| 2430 |
+
"logps/chosen": -33.91704177856445,
|
| 2431 |
+
"logps/rejected": -33.85138702392578,
|
| 2432 |
+
"loss": 0.3006,
|
| 2433 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2434 |
+
"rewards/chosen": 0.5472885370254517,
|
| 2435 |
+
"rewards/margins": 1.2651593685150146,
|
| 2436 |
+
"rewards/rejected": -0.7178710103034973,
|
| 2437 |
+
"step": 1480
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 3.87,
|
| 2441 |
+
"grad_norm": 5.78125,
|
| 2442 |
+
"learning_rate": 1.6038302591975807e-08,
|
| 2443 |
+
"logits/chosen": -1.790889024734497,
|
| 2444 |
+
"logits/rejected": -1.793336272239685,
|
| 2445 |
+
"logps/chosen": -27.0684757232666,
|
| 2446 |
+
"logps/rejected": -29.4942684173584,
|
| 2447 |
+
"loss": 0.338,
|
| 2448 |
+
"rewards/accuracies": 0.875,
|
| 2449 |
+
"rewards/chosen": 0.5631623864173889,
|
| 2450 |
+
"rewards/margins": 1.1519577503204346,
|
| 2451 |
+
"rewards/rejected": -0.5887953639030457,
|
| 2452 |
+
"step": 1490
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 3.9,
|
| 2456 |
+
"grad_norm": 6.21875,
|
| 2457 |
+
"learning_rate": 1.0268470356514237e-08,
|
| 2458 |
+
"logits/chosen": -1.796856164932251,
|
| 2459 |
+
"logits/rejected": -1.7912604808807373,
|
| 2460 |
+
"logps/chosen": -31.488414764404297,
|
| 2461 |
+
"logps/rejected": -33.81616973876953,
|
| 2462 |
+
"loss": 0.3194,
|
| 2463 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2464 |
+
"rewards/chosen": 0.5249518156051636,
|
| 2465 |
+
"rewards/margins": 1.2638393640518188,
|
| 2466 |
+
"rewards/rejected": -0.7388876080513,
|
| 2467 |
+
"step": 1500
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 3.9,
|
| 2471 |
+
"eval_logits/chosen": -2.067756175994873,
|
| 2472 |
+
"eval_logits/rejected": -2.0631155967712402,
|
| 2473 |
+
"eval_logps/chosen": -34.68146514892578,
|
| 2474 |
+
"eval_logps/rejected": -38.44291687011719,
|
| 2475 |
+
"eval_loss": 0.676816999912262,
|
| 2476 |
+
"eval_rewards/accuracies": 0.5801494717597961,
|
| 2477 |
+
"eval_rewards/chosen": -0.25876519083976746,
|
| 2478 |
+
"eval_rewards/margins": 0.11175353080034256,
|
| 2479 |
+
"eval_rewards/rejected": -0.3705187141895294,
|
| 2480 |
+
"eval_runtime": 145.241,
|
| 2481 |
+
"eval_samples_per_second": 2.362,
|
| 2482 |
+
"eval_steps_per_second": 0.296,
|
| 2483 |
+
"step": 1500
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 3.92,
|
| 2487 |
+
"grad_norm": 7.375,
|
| 2488 |
+
"learning_rate": 5.777746105209147e-09,
|
| 2489 |
+
"logits/chosen": -1.7225477695465088,
|
| 2490 |
+
"logits/rejected": -1.7265936136245728,
|
| 2491 |
+
"logps/chosen": -32.69451141357422,
|
| 2492 |
+
"logps/rejected": -35.99700164794922,
|
| 2493 |
+
"loss": 0.3195,
|
| 2494 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 2495 |
+
"rewards/chosen": 0.5905929803848267,
|
| 2496 |
+
"rewards/margins": 1.2252399921417236,
|
| 2497 |
+
"rewards/rejected": -0.6346471309661865,
|
| 2498 |
+
"step": 1510
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 3.95,
|
| 2502 |
+
"grad_norm": 3.671875,
|
| 2503 |
+
"learning_rate": 2.5684369628148352e-09,
|
| 2504 |
+
"logits/chosen": -1.7768518924713135,
|
| 2505 |
+
"logits/rejected": -1.7753241062164307,
|
| 2506 |
+
"logps/chosen": -29.13608741760254,
|
| 2507 |
+
"logps/rejected": -34.37911605834961,
|
| 2508 |
+
"loss": 0.3089,
|
| 2509 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2510 |
+
"rewards/chosen": 0.6392368078231812,
|
| 2511 |
+
"rewards/margins": 1.3549953699111938,
|
| 2512 |
+
"rewards/rejected": -0.7157586216926575,
|
| 2513 |
+
"step": 1520
|
| 2514 |
+
},
|
| 2515 |
+
{
|
| 2516 |
+
"epoch": 3.97,
|
| 2517 |
+
"grad_norm": 7.34375,
|
| 2518 |
+
"learning_rate": 6.421917227455999e-10,
|
| 2519 |
+
"logits/chosen": -1.8761787414550781,
|
| 2520 |
+
"logits/rejected": -1.8682851791381836,
|
| 2521 |
+
"logps/chosen": -26.693634033203125,
|
| 2522 |
+
"logps/rejected": -30.43243408203125,
|
| 2523 |
+
"loss": 0.3054,
|
| 2524 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2525 |
+
"rewards/chosen": 0.547723114490509,
|
| 2526 |
+
"rewards/margins": 1.2520538568496704,
|
| 2527 |
+
"rewards/rejected": -0.7043307423591614,
|
| 2528 |
+
"step": 1530
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 4.0,
|
| 2532 |
+
"grad_norm": 5.3125,
|
| 2533 |
+
"learning_rate": 0.0,
|
| 2534 |
+
"logits/chosen": -1.7809568643569946,
|
| 2535 |
+
"logits/rejected": -1.7705955505371094,
|
| 2536 |
+
"logps/chosen": -31.539371490478516,
|
| 2537 |
+
"logps/rejected": -37.31859588623047,
|
| 2538 |
+
"loss": 0.2762,
|
| 2539 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 2540 |
+
"rewards/chosen": 0.6107311248779297,
|
| 2541 |
+
"rewards/margins": 1.3737008571624756,
|
| 2542 |
+
"rewards/rejected": -0.7629695534706116,
|
| 2543 |
+
"step": 1540
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 4.0,
|
| 2547 |
+
"step": 1540,
|
| 2548 |
"total_flos": 0.0,
|
| 2549 |
+
"train_loss": 0.2942203808140445,
|
| 2550 |
+
"train_runtime": 10768.7061,
|
| 2551 |
+
"train_samples_per_second": 1.144,
|
| 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,
|