Commit
•
28b0977
1
Parent(s):
317b0fc
Model save
Browse files- README.md +29 -19
- adapter_model.safetensors +1 -1
- all_results.json +5 -5
- train_results.json +5 -5
- trainer_state.json +1905 -89
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: -
|
| 30 |
-
- Logits/rejected: -
|
| 31 |
-
- Logits/chosen: -
|
| 32 |
|
| 33 |
## Model description
|
| 34 |
|
|
@@ -57,15 +55,27 @@ 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 |
-
| 0.9987 | 0.26 | 100 |
|
| 67 |
-
| 0.9965 | 0.52 | 200 |
|
| 68 |
-
| 0.9925 | 0.78 | 300 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 69 |
|
| 70 |
|
| 71 |
### Framework versions
|
|
|
|
| 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 an unknown dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.9913
|
| 22 |
+
- Rewards/chosen: -0.0567
|
| 23 |
+
- Rewards/rejected: -0.0655
|
| 24 |
+
- Rewards/accuracies: 0.5424
|
| 25 |
+
- Rewards/margins: 0.0088
|
| 26 |
+
- Logps/rejected: -44.0648
|
| 27 |
+
- Logps/chosen: -39.7075
|
| 28 |
+
- Logits/rejected: -1.5832
|
| 29 |
+
- Logits/chosen: -1.5871
|
| 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 |
+
| 0.9987 | 0.26 | 100 | -2.2313 | -2.2264 | -34.0499 | -37.5528 | 0.9998 | 0.5336 | -0.0002 | 0.0002 | -0.0004 |
|
| 65 |
+
| 0.9965 | 0.52 | 200 | -2.2252 | -2.2204 | -34.0618 | -37.5790 | 0.9996 | 0.5071 | -0.0003 | 0.0004 | -0.0006 |
|
| 66 |
+
| 0.9925 | 0.78 | 300 | -2.2214 | -2.2166 | -34.0836 | -37.6063 | 0.9996 | 0.5594 | -0.0005 | 0.0004 | -0.0009 |
|
| 67 |
+
| 0.986 | 1.04 | 400 | 0.9994 | -0.0014 | -0.0019 | 0.5212 | 0.0006 | -37.7105 | -34.1717 | -2.1877 | -2.1926 |
|
| 68 |
+
| 0.9781 | 1.3 | 500 | 0.9987 | -0.0031 | -0.0044 | 0.5855 | 0.0013 | -37.9551 | -34.3418 | -2.1137 | -2.1185 |
|
| 69 |
+
| 0.9774 | 1.56 | 600 | 0.9973 | -0.0073 | -0.0101 | 0.5743 | 0.0027 | -38.5228 | -34.7671 | -2.0162 | -2.0208 |
|
| 70 |
+
| 0.9688 | 1.82 | 700 | 0.9969 | -0.0143 | -0.0174 | 0.5482 | 0.0031 | -39.2598 | -35.4681 | -1.9235 | -1.9280 |
|
| 71 |
+
| 0.957 | 2.08 | 800 | 0.9954 | -0.0214 | -0.0260 | 0.5540 | 0.0046 | -40.1194 | -36.1733 | -1.8363 | -1.8407 |
|
| 72 |
+
| 0.9358 | 2.34 | 900 | 0.9939 | -0.0362 | -0.0423 | 0.5365 | 0.0061 | -41.7483 | -37.6532 | -1.6988 | -1.7029 |
|
| 73 |
+
| 0.9535 | 2.6 | 1000 | 0.9921 | -0.0511 | -0.0591 | 0.5453 | 0.0079 | -43.4237 | -39.1479 | -1.6143 | -1.6183 |
|
| 74 |
+
| 0.9616 | 2.86 | 1100 | 0.9916 | -0.0562 | -0.0646 | 0.5453 | 0.0084 | -43.9754 | -39.6505 | -1.5880 | -1.5920 |
|
| 75 |
+
| 0.9167 | 3.12 | 1200 | 0.9912 | -0.0563 | -0.0651 | 0.5482 | 0.0088 | -44.0289 | -39.6666 | -1.5851 | -1.5890 |
|
| 76 |
+
| 0.9033 | 3.38 | 1300 | 0.9913 | -0.0570 | -0.0657 | 0.5453 | 0.0087 | -44.0868 | -39.7316 | -1.5817 | -1.5856 |
|
| 77 |
+
| 0.9285 | 3.64 | 1400 | 0.9912 | -0.0569 | -0.0657 | 0.5395 | 0.0088 | -44.0852 | -39.7216 | -1.5825 | -1.5864 |
|
| 78 |
+
| 0.9196 | 3.9 | 1500 | 0.9913 | -0.0567 | -0.0655 | 0.5424 | 0.0088 | -44.0648 | -39.7075 | -1.5832 | -1.5871 |
|
| 79 |
|
| 80 |
|
| 81 |
### Framework versions
|
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:f9968a4bb556b78754d04833393473ea61e6cc9ca517cac1ac531d1dbcbe29e5
|
| 3 |
size 176183216
|
all_results.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
"eval_logits/chosen": -2.221278667449951,
|
| 4 |
"eval_logits/rejected": -2.2164483070373535,
|
| 5 |
"eval_logps/chosen": -34.091766357421875,
|
|
@@ -13,9 +13,9 @@
|
|
| 13 |
"eval_samples": 343,
|
| 14 |
"eval_samples_per_second": 2.356,
|
| 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": -2.221278667449951,
|
| 4 |
"eval_logits/rejected": -2.2164483070373535,
|
| 5 |
"eval_logps/chosen": -34.091766357421875,
|
|
|
|
| 13 |
"eval_samples": 343,
|
| 14 |
"eval_samples_per_second": 2.356,
|
| 15 |
"eval_steps_per_second": 0.295,
|
| 16 |
+
"train_loss": 0.7654441864459546,
|
| 17 |
+
"train_runtime": 10797.8934,
|
| 18 |
"train_samples": 3079,
|
| 19 |
+
"train_samples_per_second": 1.141,
|
| 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": 0.7654441864459546,
|
| 4 |
+
"train_runtime": 10797.8934,
|
| 5 |
"train_samples": 3079,
|
| 6 |
+
"train_samples_per_second": 1.141,
|
| 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 +492,1946 @@
|
|
| 492 |
},
|
| 493 |
{
|
| 494 |
"epoch": 0.81,
|
| 495 |
-
"learning_rate":
|
| 496 |
-
"logits/chosen": -1.
|
| 497 |
-
"logits/rejected": -1.
|
| 498 |
-
"logps/chosen": -31.
|
| 499 |
-
"logps/rejected": -33.
|
| 500 |
-
"loss": 0.
|
| 501 |
-
"rewards/accuracies": 0.
|
| 502 |
-
"rewards/chosen": 0.
|
| 503 |
-
"rewards/margins": 0.
|
| 504 |
-
"rewards/rejected": -0.
|
| 505 |
"step": 310
|
| 506 |
},
|
| 507 |
{
|
| 508 |
"epoch": 0.83,
|
| 509 |
-
"learning_rate": 4.
|
| 510 |
-
"logits/chosen": -1.
|
| 511 |
-
"logits/rejected": -1.
|
| 512 |
-
"logps/chosen": -34.
|
| 513 |
-
"logps/rejected": -33.
|
| 514 |
-
"loss": 0.
|
| 515 |
-
"rewards/accuracies": 0.
|
| 516 |
-
"rewards/chosen": 0.
|
| 517 |
-
"rewards/margins": 0.
|
| 518 |
-
"rewards/rejected": -0.
|
| 519 |
"step": 320
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 0.86,
|
| 523 |
-
"learning_rate":
|
| 524 |
-
"logits/chosen": -1.
|
| 525 |
-
"logits/rejected": -1.
|
| 526 |
-
"logps/chosen": -33.
|
| 527 |
-
"logps/rejected": -32.
|
| 528 |
-
"loss": 0.
|
| 529 |
"rewards/accuracies": 0.737500011920929,
|
| 530 |
-
"rewards/chosen": 0.
|
| 531 |
-
"rewards/margins": 0.
|
| 532 |
-
"rewards/rejected": -0.
|
| 533 |
"step": 330
|
| 534 |
},
|
| 535 |
{
|
| 536 |
"epoch": 0.88,
|
| 537 |
-
"learning_rate":
|
| 538 |
-
"logits/chosen": -2.
|
| 539 |
-
"logits/rejected": -2.
|
| 540 |
-
"logps/chosen": -33.
|
| 541 |
-
"logps/rejected": -33.
|
| 542 |
-
"loss": 0.
|
| 543 |
-
"rewards/accuracies": 0.
|
| 544 |
-
"rewards/chosen": 0.
|
| 545 |
-
"rewards/margins": 0.
|
| 546 |
-
"rewards/rejected":
|
| 547 |
"step": 340
|
| 548 |
},
|
| 549 |
{
|
| 550 |
"epoch": 0.91,
|
| 551 |
-
"learning_rate":
|
| 552 |
-
"logits/chosen": -1.
|
| 553 |
-
"logits/rejected": -1.
|
| 554 |
-
"logps/chosen": -32.
|
| 555 |
-
"logps/rejected": -32.
|
| 556 |
-
"loss": 0.
|
| 557 |
-
"rewards/accuracies": 0.
|
| 558 |
-
"rewards/chosen": 0.
|
| 559 |
-
"rewards/margins": 0.
|
| 560 |
-
"rewards/rejected": -0.
|
| 561 |
"step": 350
|
| 562 |
},
|
| 563 |
{
|
| 564 |
"epoch": 0.94,
|
| 565 |
-
"learning_rate":
|
| 566 |
-
"logits/chosen": -1.
|
| 567 |
-
"logits/rejected": -1.
|
| 568 |
-
"logps/chosen": -31.
|
| 569 |
-
"logps/rejected": -35.
|
| 570 |
-
"loss": 0.
|
| 571 |
-
"rewards/accuracies": 0.
|
| 572 |
-
"rewards/chosen": 0.
|
| 573 |
-
"rewards/margins": 0.
|
| 574 |
-
"rewards/rejected": -0.
|
| 575 |
"step": 360
|
| 576 |
},
|
| 577 |
{
|
| 578 |
"epoch": 0.96,
|
| 579 |
-
"learning_rate":
|
| 580 |
-
"logits/chosen": -2.
|
| 581 |
-
"logits/rejected": -2.
|
| 582 |
-
"logps/chosen": -33.
|
| 583 |
-
"logps/rejected": -29.
|
| 584 |
-
"loss": 0.
|
| 585 |
-
"rewards/accuracies": 0.
|
| 586 |
-
"rewards/chosen": 0.
|
| 587 |
-
"rewards/margins": 0.
|
| 588 |
-
"rewards/rejected": -0.
|
| 589 |
"step": 370
|
| 590 |
},
|
| 591 |
{
|
| 592 |
"epoch": 0.99,
|
| 593 |
-
"learning_rate":
|
| 594 |
-
"logits/chosen": -1.
|
| 595 |
-
"logits/rejected": -1.
|
| 596 |
-
"logps/chosen": -33.
|
| 597 |
-
"logps/rejected": -
|
| 598 |
-
"loss": 0.
|
| 599 |
-
"rewards/accuracies": 0.
|
| 600 |
-
"rewards/chosen": 0.
|
| 601 |
-
"rewards/margins": 0.
|
| 602 |
-
"rewards/rejected": -0.
|
| 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 |
+
"learning_rate": 4.84533120650964e-06,
|
| 496 |
+
"logits/chosen": -1.9058300256729126,
|
| 497 |
+
"logits/rejected": -1.902580976486206,
|
| 498 |
+
"logps/chosen": -31.3160400390625,
|
| 499 |
+
"logps/rejected": -33.86355972290039,
|
| 500 |
+
"loss": 0.9959,
|
| 501 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 502 |
+
"rewards/chosen": 0.0028731029015034437,
|
| 503 |
+
"rewards/margins": 0.004087135195732117,
|
| 504 |
+
"rewards/rejected": -0.0012140319449827075,
|
| 505 |
"step": 310
|
| 506 |
},
|
| 507 |
{
|
| 508 |
"epoch": 0.83,
|
| 509 |
+
"learning_rate": 4.825108134172131e-06,
|
| 510 |
+
"logits/chosen": -1.9523051977157593,
|
| 511 |
+
"logits/rejected": -1.9401578903198242,
|
| 512 |
+
"logps/chosen": -34.279632568359375,
|
| 513 |
+
"logps/rejected": -33.72241973876953,
|
| 514 |
+
"loss": 0.995,
|
| 515 |
+
"rewards/accuracies": 0.675000011920929,
|
| 516 |
+
"rewards/chosen": 0.003164437832310796,
|
| 517 |
+
"rewards/margins": 0.005000208970159292,
|
| 518 |
+
"rewards/rejected": -0.0018357712542638183,
|
| 519 |
"step": 320
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 0.86,
|
| 523 |
+
"learning_rate": 4.80369052967602e-06,
|
| 524 |
+
"logits/chosen": -1.985467553138733,
|
| 525 |
+
"logits/rejected": -1.9840377569198608,
|
| 526 |
+
"logps/chosen": -33.06474685668945,
|
| 527 |
+
"logps/rejected": -32.610504150390625,
|
| 528 |
+
"loss": 0.9945,
|
| 529 |
"rewards/accuracies": 0.737500011920929,
|
| 530 |
+
"rewards/chosen": 0.004210514482110739,
|
| 531 |
+
"rewards/margins": 0.005519717000424862,
|
| 532 |
+
"rewards/rejected": -0.0013092018198221922,
|
| 533 |
"step": 330
|
| 534 |
},
|
| 535 |
{
|
| 536 |
"epoch": 0.88,
|
| 537 |
+
"learning_rate": 4.781089396387968e-06,
|
| 538 |
+
"logits/chosen": -2.0711922645568848,
|
| 539 |
+
"logits/rejected": -2.0555691719055176,
|
| 540 |
+
"logps/chosen": -33.74380874633789,
|
| 541 |
+
"logps/rejected": -33.10470199584961,
|
| 542 |
+
"loss": 0.9958,
|
| 543 |
+
"rewards/accuracies": 0.637499988079071,
|
| 544 |
+
"rewards/chosen": 0.0043558296747505665,
|
| 545 |
+
"rewards/margins": 0.004222516901791096,
|
| 546 |
+
"rewards/rejected": 0.00013331293303053826,
|
| 547 |
"step": 340
|
| 548 |
},
|
| 549 |
{
|
| 550 |
"epoch": 0.91,
|
| 551 |
+
"learning_rate": 4.757316345716554e-06,
|
| 552 |
+
"logits/chosen": -1.94492506980896,
|
| 553 |
+
"logits/rejected": -1.9441289901733398,
|
| 554 |
+
"logps/chosen": -32.790000915527344,
|
| 555 |
+
"logps/rejected": -32.51346969604492,
|
| 556 |
+
"loss": 0.9946,
|
| 557 |
+
"rewards/accuracies": 0.675000011920929,
|
| 558 |
+
"rewards/chosen": 0.004920940846204758,
|
| 559 |
+
"rewards/margins": 0.005404182709753513,
|
| 560 |
+
"rewards/rejected": -0.0004832421545870602,
|
| 561 |
"step": 350
|
| 562 |
},
|
| 563 |
{
|
| 564 |
"epoch": 0.94,
|
| 565 |
+
"learning_rate": 4.73238359114687e-06,
|
| 566 |
+
"logits/chosen": -1.8960033655166626,
|
| 567 |
+
"logits/rejected": -1.9061963558197021,
|
| 568 |
+
"logps/chosen": -31.744653701782227,
|
| 569 |
+
"logps/rejected": -35.42363739013672,
|
| 570 |
+
"loss": 0.9941,
|
| 571 |
+
"rewards/accuracies": 0.699999988079071,
|
| 572 |
+
"rewards/chosen": 0.004600368440151215,
|
| 573 |
+
"rewards/margins": 0.0059491791762411594,
|
| 574 |
+
"rewards/rejected": -0.0013488102704286575,
|
| 575 |
"step": 360
|
| 576 |
},
|
| 577 |
{
|
| 578 |
"epoch": 0.96,
|
| 579 |
+
"learning_rate": 4.706303941965804e-06,
|
| 580 |
+
"logits/chosen": -2.029266834259033,
|
| 581 |
+
"logits/rejected": -2.0229573249816895,
|
| 582 |
+
"logps/chosen": -33.29435729980469,
|
| 583 |
+
"logps/rejected": -29.330352783203125,
|
| 584 |
+
"loss": 0.9954,
|
| 585 |
+
"rewards/accuracies": 0.675000011920929,
|
| 586 |
+
"rewards/chosen": 0.003576122224330902,
|
| 587 |
+
"rewards/margins": 0.004615402314811945,
|
| 588 |
+
"rewards/rejected": -0.0010392797412350774,
|
| 589 |
"step": 370
|
| 590 |
},
|
| 591 |
{
|
| 592 |
"epoch": 0.99,
|
| 593 |
+
"learning_rate": 4.679090796681225e-06,
|
| 594 |
+
"logits/chosen": -1.8853864669799805,
|
| 595 |
+
"logits/rejected": -1.8876174688339233,
|
| 596 |
+
"logps/chosen": -33.62360382080078,
|
| 597 |
+
"logps/rejected": -31.0107479095459,
|
| 598 |
+
"loss": 0.9925,
|
| 599 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 600 |
+
"rewards/chosen": 0.00615445151925087,
|
| 601 |
+
"rewards/margins": 0.007535641081631184,
|
| 602 |
+
"rewards/rejected": -0.0013811895623803139,
|
| 603 |
"step": 380
|
| 604 |
},
|
| 605 |
{
|
| 606 |
+
"epoch": 1.01,
|
| 607 |
+
"learning_rate": 4.650758136138454e-06,
|
| 608 |
+
"logits/chosen": -1.9119482040405273,
|
| 609 |
+
"logits/rejected": -1.910658597946167,
|
| 610 |
+
"logps/chosen": -33.74114227294922,
|
| 611 |
+
"logps/rejected": -36.10308074951172,
|
| 612 |
+
"loss": 0.9891,
|
| 613 |
+
"rewards/accuracies": 0.7333333492279053,
|
| 614 |
+
"rewards/chosen": 0.006436466239392757,
|
| 615 |
+
"rewards/margins": 0.010907178744673729,
|
| 616 |
+
"rewards/rejected": -0.004470714367926121,
|
| 617 |
+
"step": 390
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 1.04,
|
| 621 |
+
"learning_rate": 4.621320516337559e-06,
|
| 622 |
+
"logits/chosen": -1.845229148864746,
|
| 623 |
+
"logits/rejected": -1.8368085622787476,
|
| 624 |
+
"logps/chosen": -30.943897247314453,
|
| 625 |
+
"logps/rejected": -36.51512908935547,
|
| 626 |
+
"loss": 0.986,
|
| 627 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 628 |
+
"rewards/chosen": 0.00830076914280653,
|
| 629 |
+
"rewards/margins": 0.014046055264770985,
|
| 630 |
+
"rewards/rejected": -0.005745284259319305,
|
| 631 |
+
"step": 400
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 1.04,
|
| 635 |
+
"eval_logits/chosen": -2.1925649642944336,
|
| 636 |
+
"eval_logits/rejected": -2.1877267360687256,
|
| 637 |
+
"eval_logps/chosen": -34.17170333862305,
|
| 638 |
+
"eval_logps/rejected": -37.71051788330078,
|
| 639 |
+
"eval_loss": 0.9994248747825623,
|
| 640 |
+
"eval_rewards/accuracies": 0.5211793780326843,
|
| 641 |
+
"eval_rewards/chosen": -0.0013715263921767473,
|
| 642 |
+
"eval_rewards/margins": 0.0005674380226992071,
|
| 643 |
+
"eval_rewards/rejected": -0.0019389643566682935,
|
| 644 |
+
"eval_runtime": 146.3059,
|
| 645 |
+
"eval_samples_per_second": 2.344,
|
| 646 |
+
"eval_steps_per_second": 0.294,
|
| 647 |
+
"step": 400
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.06,
|
| 651 |
+
"learning_rate": 4.590793060955158e-06,
|
| 652 |
+
"logits/chosen": -2.012714385986328,
|
| 653 |
+
"logits/rejected": -2.015584945678711,
|
| 654 |
+
"logps/chosen": -32.184898376464844,
|
| 655 |
+
"logps/rejected": -35.35422897338867,
|
| 656 |
+
"loss": 0.9857,
|
| 657 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 658 |
+
"rewards/chosen": 0.00869043916463852,
|
| 659 |
+
"rewards/margins": 0.014304302632808685,
|
| 660 |
+
"rewards/rejected": -0.005613864399492741,
|
| 661 |
+
"step": 410
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 1.09,
|
| 665 |
+
"learning_rate": 4.559191453574582e-06,
|
| 666 |
+
"logits/chosen": -1.8472423553466797,
|
| 667 |
+
"logits/rejected": -1.8458467721939087,
|
| 668 |
+
"logps/chosen": -28.332477569580078,
|
| 669 |
+
"logps/rejected": -32.81929397583008,
|
| 670 |
+
"loss": 0.9876,
|
| 671 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 672 |
+
"rewards/chosen": 0.0073507861234247684,
|
| 673 |
+
"rewards/margins": 0.012367108836770058,
|
| 674 |
+
"rewards/rejected": -0.005016324110329151,
|
| 675 |
+
"step": 420
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.12,
|
| 679 |
+
"learning_rate": 4.52653192962838e-06,
|
| 680 |
+
"logits/chosen": -1.8031494617462158,
|
| 681 |
+
"logits/rejected": -1.796241044998169,
|
| 682 |
+
"logps/chosen": -33.12567901611328,
|
| 683 |
+
"logps/rejected": -34.58681106567383,
|
| 684 |
+
"loss": 0.9871,
|
| 685 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 686 |
+
"rewards/chosen": 0.0094840619713068,
|
| 687 |
+
"rewards/margins": 0.012899428606033325,
|
| 688 |
+
"rewards/rejected": -0.003415366169065237,
|
| 689 |
+
"step": 430
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 1.14,
|
| 693 |
+
"learning_rate": 4.492831268057307e-06,
|
| 694 |
+
"logits/chosen": -1.9694416522979736,
|
| 695 |
+
"logits/rejected": -1.9643234014511108,
|
| 696 |
+
"logps/chosen": -30.745025634765625,
|
| 697 |
+
"logps/rejected": -32.631690979003906,
|
| 698 |
+
"loss": 0.983,
|
| 699 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 700 |
+
"rewards/chosen": 0.010382669046521187,
|
| 701 |
+
"rewards/margins": 0.017023462802171707,
|
| 702 |
+
"rewards/rejected": -0.0066407956182956696,
|
| 703 |
+
"step": 440
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.17,
|
| 707 |
+
"learning_rate": 4.458106782690094e-06,
|
| 708 |
+
"logits/chosen": -1.8479034900665283,
|
| 709 |
+
"logits/rejected": -1.8521947860717773,
|
| 710 |
+
"logps/chosen": -33.47832107543945,
|
| 711 |
+
"logps/rejected": -33.297340393066406,
|
| 712 |
+
"loss": 0.9825,
|
| 713 |
+
"rewards/accuracies": 0.925000011920929,
|
| 714 |
+
"rewards/chosen": 0.010375277139246464,
|
| 715 |
+
"rewards/margins": 0.017462292686104774,
|
| 716 |
+
"rewards/rejected": -0.007087015546858311,
|
| 717 |
+
"step": 450
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 1.19,
|
| 721 |
+
"learning_rate": 4.422376313348405e-06,
|
| 722 |
+
"logits/chosen": -1.848722219467163,
|
| 723 |
+
"logits/rejected": -1.8431631326675415,
|
| 724 |
+
"logps/chosen": -34.28997039794922,
|
| 725 |
+
"logps/rejected": -35.908599853515625,
|
| 726 |
+
"loss": 0.9791,
|
| 727 |
+
"rewards/accuracies": 0.925000011920929,
|
| 728 |
+
"rewards/chosen": 0.011295564472675323,
|
| 729 |
+
"rewards/margins": 0.0208906102925539,
|
| 730 |
+
"rewards/rejected": -0.009595044888556004,
|
| 731 |
+
"step": 460
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 1.22,
|
| 735 |
+
"learning_rate": 4.3856582166815696e-06,
|
| 736 |
+
"logits/chosen": -1.8663924932479858,
|
| 737 |
+
"logits/rejected": -1.8662691116333008,
|
| 738 |
+
"logps/chosen": -33.07716369628906,
|
| 739 |
+
"logps/rejected": -34.7927131652832,
|
| 740 |
+
"loss": 0.983,
|
| 741 |
+
"rewards/accuracies": 0.875,
|
| 742 |
+
"rewards/chosen": 0.010911665856838226,
|
| 743 |
+
"rewards/margins": 0.016983961686491966,
|
| 744 |
+
"rewards/rejected": -0.006072297692298889,
|
| 745 |
+
"step": 470
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 1.25,
|
| 749 |
+
"learning_rate": 4.347971356735789e-06,
|
| 750 |
+
"logits/chosen": -1.9087026119232178,
|
| 751 |
+
"logits/rejected": -1.8900855779647827,
|
| 752 |
+
"logps/chosen": -32.9819450378418,
|
| 753 |
+
"logps/rejected": -33.98347854614258,
|
| 754 |
+
"loss": 0.9791,
|
| 755 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 756 |
+
"rewards/chosen": 0.011788198724389076,
|
| 757 |
+
"rewards/margins": 0.02092314139008522,
|
| 758 |
+
"rewards/rejected": -0.009134944528341293,
|
| 759 |
+
"step": 480
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 1.27,
|
| 763 |
+
"learning_rate": 4.309335095262675e-06,
|
| 764 |
+
"logits/chosen": -1.8713194131851196,
|
| 765 |
+
"logits/rejected": -1.870599389076233,
|
| 766 |
+
"logps/chosen": -30.50040054321289,
|
| 767 |
+
"logps/rejected": -31.814233779907227,
|
| 768 |
+
"loss": 0.9832,
|
| 769 |
+
"rewards/accuracies": 0.875,
|
| 770 |
+
"rewards/chosen": 0.011109036393463612,
|
| 771 |
+
"rewards/margins": 0.016790907829999924,
|
| 772 |
+
"rewards/rejected": -0.005681873299181461,
|
| 773 |
+
"step": 490
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.3,
|
| 777 |
+
"learning_rate": 4.269769281772082e-06,
|
| 778 |
+
"logits/chosen": -1.8269096612930298,
|
| 779 |
+
"logits/rejected": -1.820021629333496,
|
| 780 |
+
"logps/chosen": -31.442739486694336,
|
| 781 |
+
"logps/rejected": -35.589012145996094,
|
| 782 |
+
"loss": 0.9781,
|
| 783 |
+
"rewards/accuracies": 0.862500011920929,
|
| 784 |
+
"rewards/chosen": 0.012273869477212429,
|
| 785 |
+
"rewards/margins": 0.02189657650887966,
|
| 786 |
+
"rewards/rejected": -0.009622708894312382,
|
| 787 |
+
"step": 500
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.3,
|
| 791 |
+
"eval_logits/chosen": -2.118459939956665,
|
| 792 |
+
"eval_logits/rejected": -2.113690137863159,
|
| 793 |
+
"eval_logps/chosen": -34.34184646606445,
|
| 794 |
+
"eval_logps/rejected": -37.955074310302734,
|
| 795 |
+
"eval_loss": 0.9986757040023804,
|
| 796 |
+
"eval_rewards/accuracies": 0.5855481624603271,
|
| 797 |
+
"eval_rewards/chosen": -0.0030729183927178383,
|
| 798 |
+
"eval_rewards/margins": 0.0013115763431414962,
|
| 799 |
+
"eval_rewards/rejected": -0.004384494852274656,
|
| 800 |
+
"eval_runtime": 145.8957,
|
| 801 |
+
"eval_samples_per_second": 2.351,
|
| 802 |
+
"eval_steps_per_second": 0.295,
|
| 803 |
+
"step": 500
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 1.32,
|
| 807 |
+
"learning_rate": 4.22929424333435e-06,
|
| 808 |
+
"logits/chosen": -1.818110466003418,
|
| 809 |
+
"logits/rejected": -1.821658730506897,
|
| 810 |
+
"logps/chosen": -28.320688247680664,
|
| 811 |
+
"logps/rejected": -33.88779830932617,
|
| 812 |
+
"loss": 0.9812,
|
| 813 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 814 |
+
"rewards/chosen": 0.008541365154087543,
|
| 815 |
+
"rewards/margins": 0.01882646046578884,
|
| 816 |
+
"rewards/rejected": -0.010285094380378723,
|
| 817 |
+
"step": 510
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 1.35,
|
| 821 |
+
"learning_rate": 4.1879307741372085e-06,
|
| 822 |
+
"logits/chosen": -1.8071190118789673,
|
| 823 |
+
"logits/rejected": -1.817929983139038,
|
| 824 |
+
"logps/chosen": -32.183380126953125,
|
| 825 |
+
"logps/rejected": -31.759241104125977,
|
| 826 |
+
"loss": 0.9781,
|
| 827 |
+
"rewards/accuracies": 0.862500011920929,
|
| 828 |
+
"rewards/chosen": 0.010789523832499981,
|
| 829 |
+
"rewards/margins": 0.021865110844373703,
|
| 830 |
+
"rewards/rejected": -0.011075586080551147,
|
| 831 |
+
"step": 520
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 1.38,
|
| 835 |
+
"learning_rate": 4.145700124802693e-06,
|
| 836 |
+
"logits/chosen": -1.7446613311767578,
|
| 837 |
+
"logits/rejected": -1.74239182472229,
|
| 838 |
+
"logps/chosen": -30.616100311279297,
|
| 839 |
+
"logps/rejected": -31.322246551513672,
|
| 840 |
+
"loss": 0.978,
|
| 841 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 842 |
+
"rewards/chosen": 0.010509822517633438,
|
| 843 |
+
"rewards/margins": 0.02196466363966465,
|
| 844 |
+
"rewards/rejected": -0.011454842053353786,
|
| 845 |
+
"step": 530
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.4,
|
| 849 |
+
"learning_rate": 4.102623991469562e-06,
|
| 850 |
+
"logits/chosen": -1.8112165927886963,
|
| 851 |
+
"logits/rejected": -1.8044904470443726,
|
| 852 |
+
"logps/chosen": -33.239593505859375,
|
| 853 |
+
"logps/rejected": -34.19357681274414,
|
| 854 |
+
"loss": 0.9783,
|
| 855 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 856 |
+
"rewards/chosen": 0.009932359680533409,
|
| 857 |
+
"rewards/margins": 0.021693259477615356,
|
| 858 |
+
"rewards/rejected": -0.011760897003114223,
|
| 859 |
+
"step": 540
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.43,
|
| 863 |
+
"learning_rate": 4.058724504646834e-06,
|
| 864 |
+
"logits/chosen": -1.77358078956604,
|
| 865 |
+
"logits/rejected": -1.7800006866455078,
|
| 866 |
+
"logps/chosen": -30.977636337280273,
|
| 867 |
+
"logps/rejected": -33.71123123168945,
|
| 868 |
+
"loss": 0.9827,
|
| 869 |
+
"rewards/accuracies": 0.762499988079071,
|
| 870 |
+
"rewards/chosen": 0.007755486760288477,
|
| 871 |
+
"rewards/margins": 0.01728045754134655,
|
| 872 |
+
"rewards/rejected": -0.00952497310936451,
|
| 873 |
+
"step": 550
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 1.45,
|
| 877 |
+
"learning_rate": 4.014024217844167e-06,
|
| 878 |
+
"logits/chosen": -1.8399150371551514,
|
| 879 |
+
"logits/rejected": -1.8170970678329468,
|
| 880 |
+
"logps/chosen": -30.506546020507812,
|
| 881 |
+
"logps/rejected": -33.8461799621582,
|
| 882 |
+
"loss": 0.9822,
|
| 883 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 884 |
+
"rewards/chosen": 0.009310225024819374,
|
| 885 |
+
"rewards/margins": 0.01781143620610237,
|
| 886 |
+
"rewards/rejected": -0.008501212112605572,
|
| 887 |
+
"step": 560
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 1.48,
|
| 891 |
+
"learning_rate": 3.968546095984911e-06,
|
| 892 |
+
"logits/chosen": -1.768343210220337,
|
| 893 |
+
"logits/rejected": -1.7633851766586304,
|
| 894 |
+
"logps/chosen": -31.467700958251953,
|
| 895 |
+
"logps/rejected": -33.090362548828125,
|
| 896 |
+
"loss": 0.9808,
|
| 897 |
+
"rewards/accuracies": 0.8125,
|
| 898 |
+
"rewards/chosen": 0.009490304626524448,
|
| 899 |
+
"rewards/margins": 0.019184768199920654,
|
| 900 |
+
"rewards/rejected": -0.009694463573396206,
|
| 901 |
+
"step": 570
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 1.51,
|
| 905 |
+
"learning_rate": 3.922313503607806e-06,
|
| 906 |
+
"logits/chosen": -1.7968637943267822,
|
| 907 |
+
"logits/rejected": -1.798661231994629,
|
| 908 |
+
"logps/chosen": -33.596519470214844,
|
| 909 |
+
"logps/rejected": -36.36685562133789,
|
| 910 |
+
"loss": 0.9765,
|
| 911 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 912 |
+
"rewards/chosen": 0.007792733609676361,
|
| 913 |
+
"rewards/margins": 0.02345210127532482,
|
| 914 |
+
"rewards/rejected": -0.01565936766564846,
|
| 915 |
+
"step": 580
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 1.53,
|
| 919 |
+
"learning_rate": 3.875350192863368e-06,
|
| 920 |
+
"logits/chosen": -1.7735790014266968,
|
| 921 |
+
"logits/rejected": -1.773097038269043,
|
| 922 |
+
"logps/chosen": -29.556873321533203,
|
| 923 |
+
"logps/rejected": -32.799781799316406,
|
| 924 |
+
"loss": 0.9763,
|
| 925 |
+
"rewards/accuracies": 0.862500011920929,
|
| 926 |
+
"rewards/chosen": 0.01035183947533369,
|
| 927 |
+
"rewards/margins": 0.02373330667614937,
|
| 928 |
+
"rewards/rejected": -0.013381466269493103,
|
| 929 |
+
"step": 590
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 1.56,
|
| 933 |
+
"learning_rate": 3.8276802913111436e-06,
|
| 934 |
+
"logits/chosen": -1.77323317527771,
|
| 935 |
+
"logits/rejected": -1.770957589149475,
|
| 936 |
+
"logps/chosen": -32.05171585083008,
|
| 937 |
+
"logps/rejected": -33.61422348022461,
|
| 938 |
+
"loss": 0.9774,
|
| 939 |
+
"rewards/accuracies": 0.862500011920929,
|
| 940 |
+
"rewards/chosen": 0.009465712122619152,
|
| 941 |
+
"rewards/margins": 0.02260836958885193,
|
| 942 |
+
"rewards/rejected": -0.013142657466232777,
|
| 943 |
+
"step": 600
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 1.56,
|
| 947 |
+
"eval_logits/chosen": -2.0208382606506348,
|
| 948 |
+
"eval_logits/rejected": -2.0161898136138916,
|
| 949 |
+
"eval_logps/chosen": -34.76713562011719,
|
| 950 |
+
"eval_logps/rejected": -38.52279281616211,
|
| 951 |
+
"eval_loss": 0.997253954410553,
|
| 952 |
+
"eval_rewards/accuracies": 0.574335515499115,
|
| 953 |
+
"eval_rewards/chosen": -0.007325790822505951,
|
| 954 |
+
"eval_rewards/margins": 0.002735937014222145,
|
| 955 |
+
"eval_rewards/rejected": -0.01006172876805067,
|
| 956 |
+
"eval_runtime": 145.8309,
|
| 957 |
+
"eval_samples_per_second": 2.352,
|
| 958 |
+
"eval_steps_per_second": 0.295,
|
| 959 |
+
"step": 600
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 1.58,
|
| 963 |
+
"learning_rate": 3.7793282895240927e-06,
|
| 964 |
+
"logits/chosen": -1.7979720830917358,
|
| 965 |
+
"logits/rejected": -1.8042287826538086,
|
| 966 |
+
"logps/chosen": -31.683053970336914,
|
| 967 |
+
"logps/rejected": -33.631996154785156,
|
| 968 |
+
"loss": 0.9788,
|
| 969 |
+
"rewards/accuracies": 0.862500011920929,
|
| 970 |
+
"rewards/chosen": 0.005543296225368977,
|
| 971 |
+
"rewards/margins": 0.02123137377202511,
|
| 972 |
+
"rewards/rejected": -0.015688076615333557,
|
| 973 |
+
"step": 610
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 1.61,
|
| 977 |
+
"learning_rate": 3.730319028506478e-06,
|
| 978 |
+
"logits/chosen": -1.741254210472107,
|
| 979 |
+
"logits/rejected": -1.7390741109848022,
|
| 980 |
+
"logps/chosen": -33.84577178955078,
|
| 981 |
+
"logps/rejected": -32.38778305053711,
|
| 982 |
+
"loss": 0.9769,
|
| 983 |
+
"rewards/accuracies": 0.800000011920929,
|
| 984 |
+
"rewards/chosen": 0.009101121686398983,
|
| 985 |
+
"rewards/margins": 0.023123882710933685,
|
| 986 |
+
"rewards/rejected": -0.014022761955857277,
|
| 987 |
+
"step": 620
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 1.64,
|
| 991 |
+
"learning_rate": 3.6806776869317074e-06,
|
| 992 |
+
"logits/chosen": -1.681199312210083,
|
| 993 |
+
"logits/rejected": -1.6748110055923462,
|
| 994 |
+
"logps/chosen": -34.46784210205078,
|
| 995 |
+
"logps/rejected": -34.00514221191406,
|
| 996 |
+
"loss": 0.9736,
|
| 997 |
+
"rewards/accuracies": 0.800000011920929,
|
| 998 |
+
"rewards/chosen": 0.00996384583413601,
|
| 999 |
+
"rewards/margins": 0.02644318714737892,
|
| 1000 |
+
"rewards/rejected": -0.016479339450597763,
|
| 1001 |
+
"step": 630
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 1.66,
|
| 1005 |
+
"learning_rate": 3.6304297682067146e-06,
|
| 1006 |
+
"logits/chosen": -1.6963602304458618,
|
| 1007 |
+
"logits/rejected": -1.7026523351669312,
|
| 1008 |
+
"logps/chosen": -33.33148193359375,
|
| 1009 |
+
"logps/rejected": -34.69775390625,
|
| 1010 |
+
"loss": 0.9789,
|
| 1011 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1012 |
+
"rewards/chosen": 0.00647822767496109,
|
| 1013 |
+
"rewards/margins": 0.021053150296211243,
|
| 1014 |
+
"rewards/rejected": -0.014574920758605003,
|
| 1015 |
+
"step": 640
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 1.69,
|
| 1019 |
+
"learning_rate": 3.579601087369492e-06,
|
| 1020 |
+
"logits/chosen": -1.767584204673767,
|
| 1021 |
+
"logits/rejected": -1.7815086841583252,
|
| 1022 |
+
"logps/chosen": -31.229949951171875,
|
| 1023 |
+
"logps/rejected": -33.55590057373047,
|
| 1024 |
+
"loss": 0.9782,
|
| 1025 |
+
"rewards/accuracies": 0.8125,
|
| 1026 |
+
"rewards/chosen": 0.006225164048373699,
|
| 1027 |
+
"rewards/margins": 0.02180730551481247,
|
| 1028 |
+
"rewards/rejected": -0.015582139603793621,
|
| 1029 |
+
"step": 650
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 1.71,
|
| 1033 |
+
"learning_rate": 3.5282177578265295e-06,
|
| 1034 |
+
"logits/chosen": -1.6283632516860962,
|
| 1035 |
+
"logits/rejected": -1.6250991821289062,
|
| 1036 |
+
"logps/chosen": -32.852622985839844,
|
| 1037 |
+
"logps/rejected": -36.96779251098633,
|
| 1038 |
+
"loss": 0.9666,
|
| 1039 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1040 |
+
"rewards/chosen": 0.01070336066186428,
|
| 1041 |
+
"rewards/margins": 0.03337875381112099,
|
| 1042 |
+
"rewards/rejected": -0.022675391286611557,
|
| 1043 |
+
"step": 660
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 1.74,
|
| 1047 |
+
"learning_rate": 3.476306177936961e-06,
|
| 1048 |
+
"logits/chosen": -1.7106907367706299,
|
| 1049 |
+
"logits/rejected": -1.710681676864624,
|
| 1050 |
+
"logps/chosen": -30.867549896240234,
|
| 1051 |
+
"logps/rejected": -36.25127410888672,
|
| 1052 |
+
"loss": 0.9716,
|
| 1053 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1054 |
+
"rewards/chosen": 0.00418041180819273,
|
| 1055 |
+
"rewards/margins": 0.028429334983229637,
|
| 1056 |
+
"rewards/rejected": -0.024248924106359482,
|
| 1057 |
+
"step": 670
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 1.77,
|
| 1061 |
+
"learning_rate": 3.423893017450324e-06,
|
| 1062 |
+
"logits/chosen": -1.6477956771850586,
|
| 1063 |
+
"logits/rejected": -1.6444308757781982,
|
| 1064 |
+
"logps/chosen": -30.36174964904785,
|
| 1065 |
+
"logps/rejected": -34.880332946777344,
|
| 1066 |
+
"loss": 0.9743,
|
| 1067 |
+
"rewards/accuracies": 0.8125,
|
| 1068 |
+
"rewards/chosen": 0.005105969496071339,
|
| 1069 |
+
"rewards/margins": 0.025731250643730164,
|
| 1070 |
+
"rewards/rejected": -0.0206252820789814,
|
| 1071 |
+
"step": 680
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 1.79,
|
| 1075 |
+
"learning_rate": 3.3710052038048794e-06,
|
| 1076 |
+
"logits/chosen": -1.6614688634872437,
|
| 1077 |
+
"logits/rejected": -1.6616519689559937,
|
| 1078 |
+
"logps/chosen": -29.40378189086914,
|
| 1079 |
+
"logps/rejected": -32.798709869384766,
|
| 1080 |
+
"loss": 0.9711,
|
| 1081 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1082 |
+
"rewards/chosen": 0.007593109272420406,
|
| 1083 |
+
"rewards/margins": 0.028875216841697693,
|
| 1084 |
+
"rewards/rejected": -0.021282104775309563,
|
| 1085 |
+
"step": 690
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 1.82,
|
| 1089 |
+
"learning_rate": 3.3176699082935546e-06,
|
| 1090 |
+
"logits/chosen": -1.579671025276184,
|
| 1091 |
+
"logits/rejected": -1.583088994026184,
|
| 1092 |
+
"logps/chosen": -33.53199005126953,
|
| 1093 |
+
"logps/rejected": -33.63783264160156,
|
| 1094 |
+
"loss": 0.9688,
|
| 1095 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1096 |
+
"rewards/chosen": 0.009464827366173267,
|
| 1097 |
+
"rewards/margins": 0.031196285039186478,
|
| 1098 |
+
"rewards/rejected": -0.021731454879045486,
|
| 1099 |
+
"step": 700
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 1.82,
|
| 1103 |
+
"eval_logits/chosen": -1.9279817342758179,
|
| 1104 |
+
"eval_logits/rejected": -1.923504114151001,
|
| 1105 |
+
"eval_logps/chosen": -35.46812438964844,
|
| 1106 |
+
"eval_logps/rejected": -39.259788513183594,
|
| 1107 |
+
"eval_loss": 0.9969056844711304,
|
| 1108 |
+
"eval_rewards/accuracies": 0.5481727719306946,
|
| 1109 |
+
"eval_rewards/chosen": -0.014335726387798786,
|
| 1110 |
+
"eval_rewards/margins": 0.0030959637369960546,
|
| 1111 |
+
"eval_rewards/rejected": -0.01743169128894806,
|
| 1112 |
+
"eval_runtime": 145.8801,
|
| 1113 |
+
"eval_samples_per_second": 2.351,
|
| 1114 |
+
"eval_steps_per_second": 0.295,
|
| 1115 |
+
"step": 700
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 1.84,
|
| 1119 |
+
"learning_rate": 3.2639145321045933e-06,
|
| 1120 |
+
"logits/chosen": -1.6492347717285156,
|
| 1121 |
+
"logits/rejected": -1.6412155628204346,
|
| 1122 |
+
"logps/chosen": -36.10542297363281,
|
| 1123 |
+
"logps/rejected": -34.03607940673828,
|
| 1124 |
+
"loss": 0.9746,
|
| 1125 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1126 |
+
"rewards/chosen": 0.004065921995788813,
|
| 1127 |
+
"rewards/margins": 0.025368575006723404,
|
| 1128 |
+
"rewards/rejected": -0.02130264975130558,
|
| 1129 |
+
"step": 710
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.87,
|
| 1133 |
+
"learning_rate": 3.2097666922441107e-06,
|
| 1134 |
+
"logits/chosen": -1.649939775466919,
|
| 1135 |
+
"logits/rejected": -1.651545524597168,
|
| 1136 |
+
"logps/chosen": -36.18703842163086,
|
| 1137 |
+
"logps/rejected": -35.8452262878418,
|
| 1138 |
+
"loss": 0.9698,
|
| 1139 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1140 |
+
"rewards/chosen": 0.002483480144292116,
|
| 1141 |
+
"rewards/margins": 0.03017548657953739,
|
| 1142 |
+
"rewards/rejected": -0.027692005038261414,
|
| 1143 |
+
"step": 720
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.9,
|
| 1147 |
+
"learning_rate": 3.1552542073477554e-06,
|
| 1148 |
+
"logits/chosen": -1.6672731637954712,
|
| 1149 |
+
"logits/rejected": -1.6649030447006226,
|
| 1150 |
+
"logps/chosen": -31.656982421875,
|
| 1151 |
+
"logps/rejected": -35.12821578979492,
|
| 1152 |
+
"loss": 0.9696,
|
| 1153 |
+
"rewards/accuracies": 0.8125,
|
| 1154 |
+
"rewards/chosen": 0.009063473902642727,
|
| 1155 |
+
"rewards/margins": 0.030376678332686424,
|
| 1156 |
+
"rewards/rejected": -0.02131320908665657,
|
| 1157 |
+
"step": 730
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 1.92,
|
| 1161 |
+
"learning_rate": 3.100405083388799e-06,
|
| 1162 |
+
"logits/chosen": -1.6315555572509766,
|
| 1163 |
+
"logits/rejected": -1.6366676092147827,
|
| 1164 |
+
"logps/chosen": -30.99042320251465,
|
| 1165 |
+
"logps/rejected": -35.70110321044922,
|
| 1166 |
+
"loss": 0.9666,
|
| 1167 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1168 |
+
"rewards/chosen": 0.007953016087412834,
|
| 1169 |
+
"rewards/margins": 0.03341587260365486,
|
| 1170 |
+
"rewards/rejected": -0.025462854653596878,
|
| 1171 |
+
"step": 740
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 1.95,
|
| 1175 |
+
"learning_rate": 3.0452474992899645e-06,
|
| 1176 |
+
"logits/chosen": -1.5864739418029785,
|
| 1177 |
+
"logits/rejected": -1.5849554538726807,
|
| 1178 |
+
"logps/chosen": -32.77734375,
|
| 1179 |
+
"logps/rejected": -37.71215057373047,
|
| 1180 |
+
"loss": 0.9663,
|
| 1181 |
+
"rewards/accuracies": 0.75,
|
| 1182 |
+
"rewards/chosen": 0.0034763626754283905,
|
| 1183 |
+
"rewards/margins": 0.03373824805021286,
|
| 1184 |
+
"rewards/rejected": -0.03026188537478447,
|
| 1185 |
+
"step": 750
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.97,
|
| 1189 |
+
"learning_rate": 2.989809792446417e-06,
|
| 1190 |
+
"logits/chosen": -1.453434705734253,
|
| 1191 |
+
"logits/rejected": -1.448988676071167,
|
| 1192 |
+
"logps/chosen": -35.56426239013672,
|
| 1193 |
+
"logps/rejected": -38.61144256591797,
|
| 1194 |
+
"loss": 0.9616,
|
| 1195 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1196 |
+
"rewards/chosen": 0.004341802094131708,
|
| 1197 |
+
"rewards/margins": 0.03837207704782486,
|
| 1198 |
+
"rewards/rejected": -0.03403027355670929,
|
| 1199 |
+
"step": 760
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 2.0,
|
| 1203 |
+
"learning_rate": 2.9341204441673267e-06,
|
| 1204 |
+
"logits/chosen": -1.5692594051361084,
|
| 1205 |
+
"logits/rejected": -1.5738866329193115,
|
| 1206 |
+
"logps/chosen": -35.17799758911133,
|
| 1207 |
+
"logps/rejected": -36.24814224243164,
|
| 1208 |
+
"loss": 0.9731,
|
| 1209 |
+
"rewards/accuracies": 0.7208333611488342,
|
| 1210 |
+
"rewards/chosen": 0.0003566344385035336,
|
| 1211 |
+
"rewards/margins": 0.02688964083790779,
|
| 1212 |
+
"rewards/rejected": -0.026533011347055435,
|
| 1213 |
+
"step": 770
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 2.03,
|
| 1217 |
+
"learning_rate": 2.878208065043501e-06,
|
| 1218 |
+
"logits/chosen": -1.5177220106124878,
|
| 1219 |
+
"logits/rejected": -1.5161765813827515,
|
| 1220 |
+
"logps/chosen": -32.8176155090332,
|
| 1221 |
+
"logps/rejected": -39.03325653076172,
|
| 1222 |
+
"loss": 0.9429,
|
| 1223 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1224 |
+
"rewards/chosen": 0.010839789174497128,
|
| 1225 |
+
"rewards/margins": 0.0571257583796978,
|
| 1226 |
+
"rewards/rejected": -0.0462859682738781,
|
| 1227 |
+
"step": 780
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 2.05,
|
| 1231 |
+
"learning_rate": 2.8221013802485974e-06,
|
| 1232 |
+
"logits/chosen": -1.5499645471572876,
|
| 1233 |
+
"logits/rejected": -1.5473253726959229,
|
| 1234 |
+
"logps/chosen": -32.2547607421875,
|
| 1235 |
+
"logps/rejected": -36.87131881713867,
|
| 1236 |
+
"loss": 0.9509,
|
| 1237 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1238 |
+
"rewards/chosen": 0.011209758929908276,
|
| 1239 |
+
"rewards/margins": 0.04905728995800018,
|
| 1240 |
+
"rewards/rejected": -0.037847526371479034,
|
| 1241 |
+
"step": 790
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 2.08,
|
| 1245 |
+
"learning_rate": 2.76582921478147e-06,
|
| 1246 |
+
"logits/chosen": -1.4648706912994385,
|
| 1247 |
+
"logits/rejected": -1.459632396697998,
|
| 1248 |
+
"logps/chosen": -33.919517517089844,
|
| 1249 |
+
"logps/rejected": -35.301055908203125,
|
| 1250 |
+
"loss": 0.957,
|
| 1251 |
+
"rewards/accuracies": 0.875,
|
| 1252 |
+
"rewards/chosen": 0.004635663237422705,
|
| 1253 |
+
"rewards/margins": 0.04300360754132271,
|
| 1254 |
+
"rewards/rejected": -0.038367945700883865,
|
| 1255 |
+
"step": 800
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 2.08,
|
| 1259 |
+
"eval_logits/chosen": -1.840654969215393,
|
| 1260 |
+
"eval_logits/rejected": -1.8362884521484375,
|
| 1261 |
+
"eval_logps/chosen": -36.17332458496094,
|
| 1262 |
+
"eval_logps/rejected": -40.119380950927734,
|
| 1263 |
+
"eval_loss": 0.9953566193580627,
|
| 1264 |
+
"eval_rewards/accuracies": 0.5539867281913757,
|
| 1265 |
+
"eval_rewards/chosen": -0.021387748420238495,
|
| 1266 |
+
"eval_rewards/margins": 0.004639865830540657,
|
| 1267 |
+
"eval_rewards/rejected": -0.0260276161134243,
|
| 1268 |
+
"eval_runtime": 145.904,
|
| 1269 |
+
"eval_samples_per_second": 2.351,
|
| 1270 |
+
"eval_steps_per_second": 0.295,
|
| 1271 |
+
"step": 800
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 2.1,
|
| 1275 |
+
"learning_rate": 2.7094204786572254e-06,
|
| 1276 |
+
"logits/chosen": -1.547839641571045,
|
| 1277 |
+
"logits/rejected": -1.5548707246780396,
|
| 1278 |
+
"logps/chosen": -31.346927642822266,
|
| 1279 |
+
"logps/rejected": -39.03870391845703,
|
| 1280 |
+
"loss": 0.9444,
|
| 1281 |
+
"rewards/accuracies": 0.875,
|
| 1282 |
+
"rewards/chosen": 0.007813268341124058,
|
| 1283 |
+
"rewards/margins": 0.05564187094569206,
|
| 1284 |
+
"rewards/rejected": -0.04782859981060028,
|
| 1285 |
+
"step": 810
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 2.13,
|
| 1289 |
+
"learning_rate": 2.6529041520546072e-06,
|
| 1290 |
+
"logits/chosen": -1.514336347579956,
|
| 1291 |
+
"logits/rejected": -1.5161477327346802,
|
| 1292 |
+
"logps/chosen": -32.147666931152344,
|
| 1293 |
+
"logps/rejected": -36.56987762451172,
|
| 1294 |
+
"loss": 0.9624,
|
| 1295 |
+
"rewards/accuracies": 0.8125,
|
| 1296 |
+
"rewards/chosen": 0.004234147723764181,
|
| 1297 |
+
"rewards/margins": 0.03763490542769432,
|
| 1298 |
+
"rewards/rejected": -0.0334007553756237,
|
| 1299 |
+
"step": 820
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 2.16,
|
| 1303 |
+
"learning_rate": 2.5963092704273302e-06,
|
| 1304 |
+
"logits/chosen": -1.3985604047775269,
|
| 1305 |
+
"logits/rejected": -1.4026976823806763,
|
| 1306 |
+
"logps/chosen": -32.4110221862793,
|
| 1307 |
+
"logps/rejected": -40.09288787841797,
|
| 1308 |
+
"loss": 0.9464,
|
| 1309 |
+
"rewards/accuracies": 0.875,
|
| 1310 |
+
"rewards/chosen": 0.0018043376039713621,
|
| 1311 |
+
"rewards/margins": 0.053631413727998734,
|
| 1312 |
+
"rewards/rejected": -0.051827073097229004,
|
| 1313 |
+
"step": 830
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 2.18,
|
| 1317 |
+
"learning_rate": 2.53966490958702e-06,
|
| 1318 |
+
"logits/chosen": -1.4642579555511475,
|
| 1319 |
+
"logits/rejected": -1.4605650901794434,
|
| 1320 |
+
"logps/chosen": -32.868621826171875,
|
| 1321 |
+
"logps/rejected": -37.319786071777344,
|
| 1322 |
+
"loss": 0.9557,
|
| 1323 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1324 |
+
"rewards/chosen": 0.0010002745548263192,
|
| 1325 |
+
"rewards/margins": 0.04425168037414551,
|
| 1326 |
+
"rewards/rejected": -0.0432514064013958,
|
| 1327 |
+
"step": 840
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 2.21,
|
| 1331 |
+
"learning_rate": 2.4830001707654135e-06,
|
| 1332 |
+
"logits/chosen": -1.5074342489242554,
|
| 1333 |
+
"logits/rejected": -1.5097802877426147,
|
| 1334 |
+
"logps/chosen": -32.264610290527344,
|
| 1335 |
+
"logps/rejected": -41.21857833862305,
|
| 1336 |
+
"loss": 0.9393,
|
| 1337 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1338 |
+
"rewards/chosen": 0.005107272416353226,
|
| 1339 |
+
"rewards/margins": 0.06067372113466263,
|
| 1340 |
+
"rewards/rejected": -0.0555664524435997,
|
| 1341 |
+
"step": 850
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 2.23,
|
| 1345 |
+
"learning_rate": 2.4263441656635054e-06,
|
| 1346 |
+
"logits/chosen": -1.3380223512649536,
|
| 1347 |
+
"logits/rejected": -1.3343451023101807,
|
| 1348 |
+
"logps/chosen": -37.09096908569336,
|
| 1349 |
+
"logps/rejected": -38.383792877197266,
|
| 1350 |
+
"loss": 0.9532,
|
| 1351 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1352 |
+
"rewards/chosen": -0.008092949166893959,
|
| 1353 |
+
"rewards/margins": 0.04682592675089836,
|
| 1354 |
+
"rewards/rejected": -0.05491887778043747,
|
| 1355 |
+
"step": 860
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 2.26,
|
| 1359 |
+
"learning_rate": 2.3697260014953107e-06,
|
| 1360 |
+
"logits/chosen": -1.360643744468689,
|
| 1361 |
+
"logits/rejected": -1.3605453968048096,
|
| 1362 |
+
"logps/chosen": -36.00204086303711,
|
| 1363 |
+
"logps/rejected": -40.74755096435547,
|
| 1364 |
+
"loss": 0.9426,
|
| 1365 |
+
"rewards/accuracies": 0.8125,
|
| 1366 |
+
"rewards/chosen": -0.0015923639293760061,
|
| 1367 |
+
"rewards/margins": 0.057420529425144196,
|
| 1368 |
+
"rewards/rejected": -0.059012897312641144,
|
| 1369 |
+
"step": 870
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 2.29,
|
| 1373 |
+
"learning_rate": 2.3131747660339396e-06,
|
| 1374 |
+
"logits/chosen": -1.3869389295578003,
|
| 1375 |
+
"logits/rejected": -1.3750654458999634,
|
| 1376 |
+
"logps/chosen": -34.1107292175293,
|
| 1377 |
+
"logps/rejected": -38.999202728271484,
|
| 1378 |
+
"loss": 0.9393,
|
| 1379 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1380 |
+
"rewards/chosen": -4.148450534557924e-05,
|
| 1381 |
+
"rewards/margins": 0.06073024123907089,
|
| 1382 |
+
"rewards/rejected": -0.06077173352241516,
|
| 1383 |
+
"step": 880
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 2.31,
|
| 1387 |
+
"learning_rate": 2.256719512667651e-06,
|
| 1388 |
+
"logits/chosen": -1.476839303970337,
|
| 1389 |
+
"logits/rejected": -1.4818614721298218,
|
| 1390 |
+
"logps/chosen": -34.29611587524414,
|
| 1391 |
+
"logps/rejected": -39.490264892578125,
|
| 1392 |
+
"loss": 0.9391,
|
| 1393 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1394 |
+
"rewards/chosen": -0.009661799296736717,
|
| 1395 |
+
"rewards/margins": 0.06087601184844971,
|
| 1396 |
+
"rewards/rejected": -0.07053781300783157,
|
| 1397 |
+
"step": 890
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 2.34,
|
| 1401 |
+
"learning_rate": 2.2003892454735786e-06,
|
| 1402 |
+
"logits/chosen": -1.4013174772262573,
|
| 1403 |
+
"logits/rejected": -1.3942521810531616,
|
| 1404 |
+
"logps/chosen": -35.21047592163086,
|
| 1405 |
+
"logps/rejected": -39.179725646972656,
|
| 1406 |
+
"loss": 0.9358,
|
| 1407 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1408 |
+
"rewards/chosen": -0.0035639957059174776,
|
| 1409 |
+
"rewards/margins": 0.064211905002594,
|
| 1410 |
+
"rewards/rejected": -0.0677758976817131,
|
| 1411 |
+
"step": 900
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 2.34,
|
| 1415 |
+
"eval_logits/chosen": -1.7029222249984741,
|
| 1416 |
+
"eval_logits/rejected": -1.6987833976745605,
|
| 1417 |
+
"eval_logps/chosen": -37.65321350097656,
|
| 1418 |
+
"eval_logps/rejected": -41.74834060668945,
|
| 1419 |
+
"eval_loss": 0.9938713312149048,
|
| 1420 |
+
"eval_rewards/accuracies": 0.5365448594093323,
|
| 1421 |
+
"eval_rewards/chosen": -0.03618660196661949,
|
| 1422 |
+
"eval_rewards/margins": 0.00613059988245368,
|
| 1423 |
+
"eval_rewards/rejected": -0.04231720417737961,
|
| 1424 |
+
"eval_runtime": 145.9089,
|
| 1425 |
+
"eval_samples_per_second": 2.351,
|
| 1426 |
+
"eval_steps_per_second": 0.295,
|
| 1427 |
+
"step": 900
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 2.36,
|
| 1431 |
+
"learning_rate": 2.1442129043167877e-06,
|
| 1432 |
+
"logits/chosen": -1.3909523487091064,
|
| 1433 |
+
"logits/rejected": -1.3909533023834229,
|
| 1434 |
+
"logps/chosen": -31.57651138305664,
|
| 1435 |
+
"logps/rejected": -42.50416564941406,
|
| 1436 |
+
"loss": 0.9259,
|
| 1437 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1438 |
+
"rewards/chosen": -0.004093508701771498,
|
| 1439 |
+
"rewards/margins": 0.07405127584934235,
|
| 1440 |
+
"rewards/rejected": -0.07814478874206543,
|
| 1441 |
+
"step": 910
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 2.39,
|
| 1445 |
+
"learning_rate": 2.088219349982323e-06,
|
| 1446 |
+
"logits/chosen": -1.3331390619277954,
|
| 1447 |
+
"logits/rejected": -1.3248008489608765,
|
| 1448 |
+
"logps/chosen": -33.280738830566406,
|
| 1449 |
+
"logps/rejected": -41.13130569458008,
|
| 1450 |
+
"loss": 0.9388,
|
| 1451 |
+
"rewards/accuracies": 0.8125,
|
| 1452 |
+
"rewards/chosen": -0.014417916536331177,
|
| 1453 |
+
"rewards/margins": 0.06118142604827881,
|
| 1454 |
+
"rewards/rejected": -0.07559934258460999,
|
| 1455 |
+
"step": 920
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 2.42,
|
| 1459 |
+
"learning_rate": 2.0324373493478803e-06,
|
| 1460 |
+
"logits/chosen": -1.474799394607544,
|
| 1461 |
+
"logits/rejected": -1.473085641860962,
|
| 1462 |
+
"logps/chosen": -30.739898681640625,
|
| 1463 |
+
"logps/rejected": -40.02002716064453,
|
| 1464 |
+
"loss": 0.9359,
|
| 1465 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1466 |
+
"rewards/chosen": -0.007428646087646484,
|
| 1467 |
+
"rewards/margins": 0.06412303447723389,
|
| 1468 |
+
"rewards/rejected": -0.07155168056488037,
|
| 1469 |
+
"step": 930
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 2.44,
|
| 1473 |
+
"learning_rate": 1.976895560604729e-06,
|
| 1474 |
+
"logits/chosen": -1.3616539239883423,
|
| 1475 |
+
"logits/rejected": -1.372312307357788,
|
| 1476 |
+
"logps/chosen": -36.05510330200195,
|
| 1477 |
+
"logps/rejected": -41.12104034423828,
|
| 1478 |
+
"loss": 0.935,
|
| 1479 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1480 |
+
"rewards/chosen": -0.016300354152917862,
|
| 1481 |
+
"rewards/margins": 0.06501360237598419,
|
| 1482 |
+
"rewards/rejected": -0.08131395280361176,
|
| 1483 |
+
"step": 940
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 2.47,
|
| 1487 |
+
"learning_rate": 1.921622518534466e-06,
|
| 1488 |
+
"logits/chosen": -1.3951683044433594,
|
| 1489 |
+
"logits/rejected": -1.3979324102401733,
|
| 1490 |
+
"logps/chosen": -32.248931884765625,
|
| 1491 |
+
"logps/rejected": -39.3216552734375,
|
| 1492 |
+
"loss": 0.9402,
|
| 1493 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1494 |
+
"rewards/chosen": -0.016012068837881088,
|
| 1495 |
+
"rewards/margins": 0.05984366685152054,
|
| 1496 |
+
"rewards/rejected": -0.07585573941469193,
|
| 1497 |
+
"step": 950
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 2.49,
|
| 1501 |
+
"learning_rate": 1.8666466198491794e-06,
|
| 1502 |
+
"logits/chosen": -1.3717939853668213,
|
| 1503 |
+
"logits/rejected": -1.3667659759521484,
|
| 1504 |
+
"logps/chosen": -35.57506561279297,
|
| 1505 |
+
"logps/rejected": -42.31549835205078,
|
| 1506 |
+
"loss": 0.9309,
|
| 1507 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1508 |
+
"rewards/chosen": -0.014998522587120533,
|
| 1509 |
+
"rewards/margins": 0.06909255683422089,
|
| 1510 |
+
"rewards/rejected": -0.08409108221530914,
|
| 1511 |
+
"step": 960
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 2.52,
|
| 1515 |
+
"learning_rate": 1.8119961086025376e-06,
|
| 1516 |
+
"logits/chosen": -1.294325590133667,
|
| 1517 |
+
"logits/rejected": -1.2963002920150757,
|
| 1518 |
+
"logps/chosen": -34.213260650634766,
|
| 1519 |
+
"logps/rejected": -43.49393081665039,
|
| 1520 |
+
"loss": 0.9309,
|
| 1521 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1522 |
+
"rewards/chosen": -0.014005112461745739,
|
| 1523 |
+
"rewards/margins": 0.06911824643611908,
|
| 1524 |
+
"rewards/rejected": -0.0831233561038971,
|
| 1525 |
+
"step": 970
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 2.55,
|
| 1529 |
+
"learning_rate": 1.7576990616793139e-06,
|
| 1530 |
+
"logits/chosen": -1.3165125846862793,
|
| 1531 |
+
"logits/rejected": -1.310859203338623,
|
| 1532 |
+
"logps/chosen": -38.791927337646484,
|
| 1533 |
+
"logps/rejected": -45.34046936035156,
|
| 1534 |
+
"loss": 0.9431,
|
| 1535 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1536 |
+
"rewards/chosen": -0.03386862203478813,
|
| 1537 |
+
"rewards/margins": 0.05691651254892349,
|
| 1538 |
+
"rewards/rejected": -0.09078512340784073,
|
| 1539 |
+
"step": 980
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 2.57,
|
| 1543 |
+
"learning_rate": 1.7037833743707892e-06,
|
| 1544 |
+
"logits/chosen": -1.3041751384735107,
|
| 1545 |
+
"logits/rejected": -1.2975685596466064,
|
| 1546 |
+
"logps/chosen": -32.80427551269531,
|
| 1547 |
+
"logps/rejected": -45.00722885131836,
|
| 1548 |
+
"loss": 0.9278,
|
| 1549 |
+
"rewards/accuracies": 0.8125,
|
| 1550 |
+
"rewards/chosen": -0.018013231456279755,
|
| 1551 |
+
"rewards/margins": 0.07216015458106995,
|
| 1552 |
+
"rewards/rejected": -0.0901733785867691,
|
| 1553 |
+
"step": 990
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 2.6,
|
| 1557 |
+
"learning_rate": 1.6502767460434588e-06,
|
| 1558 |
+
"logits/chosen": -1.2888100147247314,
|
| 1559 |
+
"logits/rejected": -1.2791659832000732,
|
| 1560 |
+
"logps/chosen": -34.17406463623047,
|
| 1561 |
+
"logps/rejected": -36.677513122558594,
|
| 1562 |
+
"loss": 0.9535,
|
| 1563 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1564 |
+
"rewards/chosen": -0.024253757670521736,
|
| 1565 |
+
"rewards/margins": 0.04652407765388489,
|
| 1566 |
+
"rewards/rejected": -0.07077784091234207,
|
| 1567 |
+
"step": 1000
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 2.6,
|
| 1571 |
+
"eval_logits/chosen": -1.6182585954666138,
|
| 1572 |
+
"eval_logits/rejected": -1.614295244216919,
|
| 1573 |
+
"eval_logps/chosen": -39.14792251586914,
|
| 1574 |
+
"eval_logps/rejected": -43.423744201660156,
|
| 1575 |
+
"eval_loss": 0.9920738935470581,
|
| 1576 |
+
"eval_rewards/accuracies": 0.545265793800354,
|
| 1577 |
+
"eval_rewards/chosen": -0.05113373324275017,
|
| 1578 |
+
"eval_rewards/margins": 0.007937532849609852,
|
| 1579 |
+
"eval_rewards/rejected": -0.059071265161037445,
|
| 1580 |
+
"eval_runtime": 145.6856,
|
| 1581 |
+
"eval_samples_per_second": 2.354,
|
| 1582 |
+
"eval_steps_per_second": 0.295,
|
| 1583 |
+
"step": 1000
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 2.62,
|
| 1587 |
+
"learning_rate": 1.5972066659083796e-06,
|
| 1588 |
+
"logits/chosen": -1.353121042251587,
|
| 1589 |
+
"logits/rejected": -1.35282301902771,
|
| 1590 |
+
"logps/chosen": -34.075523376464844,
|
| 1591 |
+
"logps/rejected": -38.43246078491211,
|
| 1592 |
+
"loss": 0.9408,
|
| 1593 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1594 |
+
"rewards/chosen": -0.022513313218951225,
|
| 1595 |
+
"rewards/margins": 0.059226490557193756,
|
| 1596 |
+
"rewards/rejected": -0.08173979073762894,
|
| 1597 |
+
"step": 1010
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 2.65,
|
| 1601 |
+
"learning_rate": 1.5446003988985041e-06,
|
| 1602 |
+
"logits/chosen": -1.403580904006958,
|
| 1603 |
+
"logits/rejected": -1.4037662744522095,
|
| 1604 |
+
"logps/chosen": -33.98898696899414,
|
| 1605 |
+
"logps/rejected": -39.30225372314453,
|
| 1606 |
+
"loss": 0.9391,
|
| 1607 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1608 |
+
"rewards/chosen": -0.020454101264476776,
|
| 1609 |
+
"rewards/margins": 0.06091039255261421,
|
| 1610 |
+
"rewards/rejected": -0.08136448264122009,
|
| 1611 |
+
"step": 1020
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 2.68,
|
| 1615 |
+
"learning_rate": 1.4924849716612211e-06,
|
| 1616 |
+
"logits/chosen": -1.3622567653656006,
|
| 1617 |
+
"logits/rejected": -1.3672521114349365,
|
| 1618 |
+
"logps/chosen": -35.09298324584961,
|
| 1619 |
+
"logps/rejected": -35.331912994384766,
|
| 1620 |
+
"loss": 0.9546,
|
| 1621 |
+
"rewards/accuracies": 0.6875,
|
| 1622 |
+
"rewards/chosen": -0.02812912128865719,
|
| 1623 |
+
"rewards/margins": 0.04535282403230667,
|
| 1624 |
+
"rewards/rejected": -0.07348194718360901,
|
| 1625 |
+
"step": 1030
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 2.7,
|
| 1629 |
+
"learning_rate": 1.440887158673332e-06,
|
| 1630 |
+
"logits/chosen": -1.3483327627182007,
|
| 1631 |
+
"logits/rejected": -1.3397239446640015,
|
| 1632 |
+
"logps/chosen": -33.612525939941406,
|
| 1633 |
+
"logps/rejected": -43.52740478515625,
|
| 1634 |
+
"loss": 0.9241,
|
| 1635 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1636 |
+
"rewards/chosen": -0.024346303194761276,
|
| 1637 |
+
"rewards/margins": 0.07592549920082092,
|
| 1638 |
+
"rewards/rejected": -0.1002717837691307,
|
| 1639 |
+
"step": 1040
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 2.73,
|
| 1643 |
+
"learning_rate": 1.3898334684855647e-06,
|
| 1644 |
+
"logits/chosen": -1.3182249069213867,
|
| 1645 |
+
"logits/rejected": -1.3285300731658936,
|
| 1646 |
+
"logps/chosen": -36.65705108642578,
|
| 1647 |
+
"logps/rejected": -41.1858024597168,
|
| 1648 |
+
"loss": 0.9416,
|
| 1649 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1650 |
+
"rewards/chosen": -0.0310394074767828,
|
| 1651 |
+
"rewards/margins": 0.05836848169565201,
|
| 1652 |
+
"rewards/rejected": -0.08940788358449936,
|
| 1653 |
+
"step": 1050
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 2.75,
|
| 1657 |
+
"learning_rate": 1.3393501301037245e-06,
|
| 1658 |
+
"logits/chosen": -1.3787955045700073,
|
| 1659 |
+
"logits/rejected": -1.3705497980117798,
|
| 1660 |
+
"logps/chosen": -35.9283561706543,
|
| 1661 |
+
"logps/rejected": -47.05866241455078,
|
| 1662 |
+
"loss": 0.917,
|
| 1663 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1664 |
+
"rewards/chosen": -0.024275777861475945,
|
| 1665 |
+
"rewards/margins": 0.08299694955348969,
|
| 1666 |
+
"rewards/rejected": -0.10727272927761078,
|
| 1667 |
+
"step": 1060
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 2.78,
|
| 1671 |
+
"learning_rate": 1.2894630795134454e-06,
|
| 1672 |
+
"logits/chosen": -1.2876031398773193,
|
| 1673 |
+
"logits/rejected": -1.2903258800506592,
|
| 1674 |
+
"logps/chosen": -38.158668518066406,
|
| 1675 |
+
"logps/rejected": -41.497459411621094,
|
| 1676 |
+
"loss": 0.9337,
|
| 1677 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1678 |
+
"rewards/chosen": -0.021693874150514603,
|
| 1679 |
+
"rewards/margins": 0.06630205363035202,
|
| 1680 |
+
"rewards/rejected": -0.08799593150615692,
|
| 1681 |
+
"step": 1070
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 2.81,
|
| 1685 |
+
"learning_rate": 1.2401979463554984e-06,
|
| 1686 |
+
"logits/chosen": -1.3835922479629517,
|
| 1687 |
+
"logits/rejected": -1.38229238986969,
|
| 1688 |
+
"logps/chosen": -35.45813751220703,
|
| 1689 |
+
"logps/rejected": -44.54060363769531,
|
| 1690 |
+
"loss": 0.9198,
|
| 1691 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1692 |
+
"rewards/chosen": -0.024004962295293808,
|
| 1693 |
+
"rewards/margins": 0.08024422824382782,
|
| 1694 |
+
"rewards/rejected": -0.10424919426441193,
|
| 1695 |
+
"step": 1080
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 2.83,
|
| 1699 |
+
"learning_rate": 1.1915800407584705e-06,
|
| 1700 |
+
"logits/chosen": -1.3808025121688843,
|
| 1701 |
+
"logits/rejected": -1.3838340044021606,
|
| 1702 |
+
"logps/chosen": -33.26921463012695,
|
| 1703 |
+
"logps/rejected": -42.89200210571289,
|
| 1704 |
+
"loss": 0.9303,
|
| 1705 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1706 |
+
"rewards/chosen": -0.023758988827466965,
|
| 1707 |
+
"rewards/margins": 0.06968275457620621,
|
| 1708 |
+
"rewards/rejected": -0.09344174712896347,
|
| 1709 |
+
"step": 1090
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 2.86,
|
| 1713 |
+
"learning_rate": 1.1436343403356019e-06,
|
| 1714 |
+
"logits/chosen": -1.3661361932754517,
|
| 1715 |
+
"logits/rejected": -1.3705623149871826,
|
| 1716 |
+
"logps/chosen": -36.603370666503906,
|
| 1717 |
+
"logps/rejected": -37.271080017089844,
|
| 1718 |
+
"loss": 0.9616,
|
| 1719 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1720 |
+
"rewards/chosen": -0.030343661084771156,
|
| 1721 |
+
"rewards/margins": 0.03835037350654602,
|
| 1722 |
+
"rewards/rejected": -0.06869403272867203,
|
| 1723 |
+
"step": 1100
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 2.86,
|
| 1727 |
+
"eval_logits/chosen": -1.5919773578643799,
|
| 1728 |
+
"eval_logits/rejected": -1.5880482196807861,
|
| 1729 |
+
"eval_logps/chosen": -39.65049743652344,
|
| 1730 |
+
"eval_logps/rejected": -43.97543716430664,
|
| 1731 |
+
"eval_loss": 0.9915903806686401,
|
| 1732 |
+
"eval_rewards/accuracies": 0.545265793800354,
|
| 1733 |
+
"eval_rewards/chosen": -0.05615944042801857,
|
| 1734 |
+
"eval_rewards/margins": 0.008428744971752167,
|
| 1735 |
+
"eval_rewards/rejected": -0.06458818912506104,
|
| 1736 |
+
"eval_runtime": 145.863,
|
| 1737 |
+
"eval_samples_per_second": 2.352,
|
| 1738 |
+
"eval_steps_per_second": 0.295,
|
| 1739 |
+
"step": 1100
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 2.88,
|
| 1743 |
+
"learning_rate": 1.0963854773524548e-06,
|
| 1744 |
+
"logits/chosen": -1.355208158493042,
|
| 1745 |
+
"logits/rejected": -1.3564389944076538,
|
| 1746 |
+
"logps/chosen": -34.99750900268555,
|
| 1747 |
+
"logps/rejected": -39.34065628051758,
|
| 1748 |
+
"loss": 0.9373,
|
| 1749 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1750 |
+
"rewards/chosen": -0.020573578774929047,
|
| 1751 |
+
"rewards/margins": 0.06267382204532623,
|
| 1752 |
+
"rewards/rejected": -0.08324739336967468,
|
| 1753 |
+
"step": 1110
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 2.91,
|
| 1757 |
+
"learning_rate": 1.049857726072005e-06,
|
| 1758 |
+
"logits/chosen": -1.227752923965454,
|
| 1759 |
+
"logits/rejected": -1.2306647300720215,
|
| 1760 |
+
"logps/chosen": -37.04521942138672,
|
| 1761 |
+
"logps/rejected": -41.808509826660156,
|
| 1762 |
+
"loss": 0.9371,
|
| 1763 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1764 |
+
"rewards/chosen": -0.02480238489806652,
|
| 1765 |
+
"rewards/margins": 0.06287000328302383,
|
| 1766 |
+
"rewards/rejected": -0.0876723974943161,
|
| 1767 |
+
"step": 1120
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 2.94,
|
| 1771 |
+
"learning_rate": 1.0040749902836508e-06,
|
| 1772 |
+
"logits/chosen": -1.250138521194458,
|
| 1773 |
+
"logits/rejected": -1.2487337589263916,
|
| 1774 |
+
"logps/chosen": -34.29961013793945,
|
| 1775 |
+
"logps/rejected": -39.423301696777344,
|
| 1776 |
+
"loss": 0.9503,
|
| 1777 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1778 |
+
"rewards/chosen": -0.032878875732421875,
|
| 1779 |
+
"rewards/margins": 0.04969673603773117,
|
| 1780 |
+
"rewards/rejected": -0.08257561177015305,
|
| 1781 |
+
"step": 1130
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 2.96,
|
| 1785 |
+
"learning_rate": 9.59060791022566e-07,
|
| 1786 |
+
"logits/chosen": -1.363937258720398,
|
| 1787 |
+
"logits/rejected": -1.360128402709961,
|
| 1788 |
+
"logps/chosen": -34.89466094970703,
|
| 1789 |
+
"logps/rejected": -41.99665832519531,
|
| 1790 |
+
"loss": 0.9281,
|
| 1791 |
+
"rewards/accuracies": 0.8125,
|
| 1792 |
+
"rewards/chosen": -0.018548697233200073,
|
| 1793 |
+
"rewards/margins": 0.07187779247760773,
|
| 1794 |
+
"rewards/rejected": -0.0904264897108078,
|
| 1795 |
+
"step": 1140
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 2.99,
|
| 1799 |
+
"learning_rate": 9.148382544856885e-07,
|
| 1800 |
+
"logits/chosen": -1.2763797044754028,
|
| 1801 |
+
"logits/rejected": -1.2683088779449463,
|
| 1802 |
+
"logps/chosen": -36.54337692260742,
|
| 1803 |
+
"logps/rejected": -40.25540542602539,
|
| 1804 |
+
"loss": 0.9413,
|
| 1805 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1806 |
+
"rewards/chosen": -0.030916398391127586,
|
| 1807 |
+
"rewards/margins": 0.05871362239122391,
|
| 1808 |
+
"rewards/rejected": -0.08963000774383545,
|
| 1809 |
+
"step": 1150
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 3.01,
|
| 1813 |
+
"learning_rate": 8.714301001505568e-07,
|
| 1814 |
+
"logits/chosen": -1.301361083984375,
|
| 1815 |
+
"logits/rejected": -1.3027942180633545,
|
| 1816 |
+
"logps/chosen": -36.6050910949707,
|
| 1817 |
+
"logps/rejected": -39.68968963623047,
|
| 1818 |
+
"loss": 0.9444,
|
| 1819 |
+
"rewards/accuracies": 0.7666667103767395,
|
| 1820 |
+
"rewards/chosen": -0.029233410954475403,
|
| 1821 |
+
"rewards/margins": 0.05564901977777481,
|
| 1822 |
+
"rewards/rejected": -0.08488242328166962,
|
| 1823 |
+
"step": 1160
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 3.04,
|
| 1827 |
+
"learning_rate": 8.288586291031025e-07,
|
| 1828 |
+
"logits/chosen": -1.3797643184661865,
|
| 1829 |
+
"logits/rejected": -1.3745005130767822,
|
| 1830 |
+
"logps/chosen": -36.120018005371094,
|
| 1831 |
+
"logps/rejected": -41.199337005615234,
|
| 1832 |
+
"loss": 0.9434,
|
| 1833 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1834 |
+
"rewards/chosen": -0.02594051882624626,
|
| 1835 |
+
"rewards/margins": 0.05664067342877388,
|
| 1836 |
+
"rewards/rejected": -0.08258119225502014,
|
| 1837 |
+
"step": 1170
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 3.06,
|
| 1841 |
+
"learning_rate": 7.871457125803897e-07,
|
| 1842 |
+
"logits/chosen": -1.2806499004364014,
|
| 1843 |
+
"logits/rejected": -1.289930820465088,
|
| 1844 |
+
"logps/chosen": -36.737998962402344,
|
| 1845 |
+
"logps/rejected": -41.14373016357422,
|
| 1846 |
+
"loss": 0.9462,
|
| 1847 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1848 |
+
"rewards/chosen": -0.03404999524354935,
|
| 1849 |
+
"rewards/margins": 0.053772974759340286,
|
| 1850 |
+
"rewards/rejected": -0.08782295882701874,
|
| 1851 |
+
"step": 1180
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 3.09,
|
| 1855 |
+
"learning_rate": 7.463127807341966e-07,
|
| 1856 |
+
"logits/chosen": -1.2992810010910034,
|
| 1857 |
+
"logits/rejected": -1.2938917875289917,
|
| 1858 |
+
"logps/chosen": -33.985870361328125,
|
| 1859 |
+
"logps/rejected": -42.187801361083984,
|
| 1860 |
+
"loss": 0.9291,
|
| 1861 |
+
"rewards/accuracies": 0.8125,
|
| 1862 |
+
"rewards/chosen": -0.018122535198926926,
|
| 1863 |
+
"rewards/margins": 0.07092103362083435,
|
| 1864 |
+
"rewards/rejected": -0.08904357254505157,
|
| 1865 |
+
"step": 1190
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 3.12,
|
| 1869 |
+
"learning_rate": 7.063808116212021e-07,
|
| 1870 |
+
"logits/chosen": -1.261099934577942,
|
| 1871 |
+
"logits/rejected": -1.262424349784851,
|
| 1872 |
+
"logps/chosen": -35.9713249206543,
|
| 1873 |
+
"logps/rejected": -43.56898880004883,
|
| 1874 |
+
"loss": 0.9167,
|
| 1875 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1876 |
+
"rewards/chosen": -0.02480882592499256,
|
| 1877 |
+
"rewards/margins": 0.08328056335449219,
|
| 1878 |
+
"rewards/rejected": -0.1080893874168396,
|
| 1879 |
+
"step": 1200
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 3.12,
|
| 1883 |
+
"eval_logits/chosen": -1.5890045166015625,
|
| 1884 |
+
"eval_logits/rejected": -1.585091471672058,
|
| 1885 |
+
"eval_logps/chosen": -39.66655349731445,
|
| 1886 |
+
"eval_logps/rejected": -44.02885818481445,
|
| 1887 |
+
"eval_loss": 0.9912154078483582,
|
| 1888 |
+
"eval_rewards/accuracies": 0.5481727719306946,
|
| 1889 |
+
"eval_rewards/chosen": -0.056320033967494965,
|
| 1890 |
+
"eval_rewards/margins": 0.00880234595388174,
|
| 1891 |
+
"eval_rewards/rejected": -0.06512238085269928,
|
| 1892 |
+
"eval_runtime": 145.8882,
|
| 1893 |
+
"eval_samples_per_second": 2.351,
|
| 1894 |
+
"eval_steps_per_second": 0.295,
|
| 1895 |
+
"step": 1200
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 3.14,
|
| 1899 |
+
"learning_rate": 6.673703204254348e-07,
|
| 1900 |
+
"logits/chosen": -1.2175233364105225,
|
| 1901 |
+
"logits/rejected": -1.2172149419784546,
|
| 1902 |
+
"logps/chosen": -37.891151428222656,
|
| 1903 |
+
"logps/rejected": -42.90294647216797,
|
| 1904 |
+
"loss": 0.9165,
|
| 1905 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1906 |
+
"rewards/chosen": -0.01838667131960392,
|
| 1907 |
+
"rewards/margins": 0.08354915678501129,
|
| 1908 |
+
"rewards/rejected": -0.10193584114313126,
|
| 1909 |
+
"step": 1210
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 3.17,
|
| 1913 |
+
"learning_rate": 6.293013489185315e-07,
|
| 1914 |
+
"logits/chosen": -1.3444710969924927,
|
| 1915 |
+
"logits/rejected": -1.3372966051101685,
|
| 1916 |
+
"logps/chosen": -34.07860565185547,
|
| 1917 |
+
"logps/rejected": -43.33241653442383,
|
| 1918 |
+
"loss": 0.9192,
|
| 1919 |
+
"rewards/accuracies": 0.8125,
|
| 1920 |
+
"rewards/chosen": -0.02374918945133686,
|
| 1921 |
+
"rewards/margins": 0.08078955113887787,
|
| 1922 |
+
"rewards/rejected": -0.10453873872756958,
|
| 1923 |
+
"step": 1220
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 3.19,
|
| 1927 |
+
"learning_rate": 5.921934551632086e-07,
|
| 1928 |
+
"logits/chosen": -1.2269846200942993,
|
| 1929 |
+
"logits/rejected": -1.2168152332305908,
|
| 1930 |
+
"logps/chosen": -36.28278350830078,
|
| 1931 |
+
"logps/rejected": -42.86933517456055,
|
| 1932 |
+
"loss": 0.9192,
|
| 1933 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1934 |
+
"rewards/chosen": -0.01897098869085312,
|
| 1935 |
+
"rewards/margins": 0.08078411966562271,
|
| 1936 |
+
"rewards/rejected": -0.09975510835647583,
|
| 1937 |
+
"step": 1230
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 3.22,
|
| 1941 |
+
"learning_rate": 5.560657034652405e-07,
|
| 1942 |
+
"logits/chosen": -1.3140621185302734,
|
| 1943 |
+
"logits/rejected": -1.308857798576355,
|
| 1944 |
+
"logps/chosen": -33.93239212036133,
|
| 1945 |
+
"logps/rejected": -37.61920928955078,
|
| 1946 |
+
"loss": 0.9448,
|
| 1947 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1948 |
+
"rewards/chosen": -0.03232160210609436,
|
| 1949 |
+
"rewards/margins": 0.0551949217915535,
|
| 1950 |
+
"rewards/rejected": -0.08751652389764786,
|
| 1951 |
+
"step": 1240
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"epoch": 3.25,
|
| 1955 |
+
"learning_rate": 5.2093665457911e-07,
|
| 1956 |
+
"logits/chosen": -1.3194572925567627,
|
| 1957 |
+
"logits/rejected": -1.3273426294326782,
|
| 1958 |
+
"logps/chosen": -38.09012985229492,
|
| 1959 |
+
"logps/rejected": -40.47280502319336,
|
| 1960 |
+
"loss": 0.9374,
|
| 1961 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1962 |
+
"rewards/chosen": -0.026947390288114548,
|
| 1963 |
+
"rewards/margins": 0.06259752810001373,
|
| 1964 |
+
"rewards/rejected": -0.08954491466283798,
|
| 1965 |
+
"step": 1250
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 3.27,
|
| 1969 |
+
"learning_rate": 4.868243561723535e-07,
|
| 1970 |
+
"logits/chosen": -1.3223885297775269,
|
| 1971 |
+
"logits/rejected": -1.3221849203109741,
|
| 1972 |
+
"logps/chosen": -36.7206916809082,
|
| 1973 |
+
"logps/rejected": -43.311153411865234,
|
| 1974 |
+
"loss": 0.932,
|
| 1975 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1976 |
+
"rewards/chosen": -0.03692134469747543,
|
| 1977 |
+
"rewards/margins": 0.06804148107767105,
|
| 1978 |
+
"rewards/rejected": -0.10496282577514648,
|
| 1979 |
+
"step": 1260
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 3.3,
|
| 1983 |
+
"learning_rate": 4.537463335535161e-07,
|
| 1984 |
+
"logits/chosen": -1.2348535060882568,
|
| 1985 |
+
"logits/rejected": -1.2325520515441895,
|
| 1986 |
+
"logps/chosen": -34.88422393798828,
|
| 1987 |
+
"logps/rejected": -44.12788009643555,
|
| 1988 |
+
"loss": 0.9119,
|
| 1989 |
+
"rewards/accuracies": 0.8125,
|
| 1990 |
+
"rewards/chosen": -0.01793385110795498,
|
| 1991 |
+
"rewards/margins": 0.08813884109258652,
|
| 1992 |
+
"rewards/rejected": -0.10607268661260605,
|
| 1993 |
+
"step": 1270
|
| 1994 |
+
},
|
| 1995 |
+
{
|
| 1996 |
+
"epoch": 3.32,
|
| 1997 |
+
"learning_rate": 4.217195806684629e-07,
|
| 1998 |
+
"logits/chosen": -1.1531862020492554,
|
| 1999 |
+
"logits/rejected": -1.1495788097381592,
|
| 2000 |
+
"logps/chosen": -37.46773147583008,
|
| 2001 |
+
"logps/rejected": -39.666324615478516,
|
| 2002 |
+
"loss": 0.9349,
|
| 2003 |
+
"rewards/accuracies": 0.737500011920929,
|
| 2004 |
+
"rewards/chosen": -0.022302847355604172,
|
| 2005 |
+
"rewards/margins": 0.06509546935558319,
|
| 2006 |
+
"rewards/rejected": -0.08739831298589706,
|
| 2007 |
+
"step": 1280
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 3.35,
|
| 2011 |
+
"learning_rate": 3.907605513696808e-07,
|
| 2012 |
+
"logits/chosen": -1.3214530944824219,
|
| 2013 |
+
"logits/rejected": -1.3079793453216553,
|
| 2014 |
+
"logps/chosen": -37.78839111328125,
|
| 2015 |
+
"logps/rejected": -46.277854919433594,
|
| 2016 |
+
"loss": 0.9193,
|
| 2017 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 2018 |
+
"rewards/chosen": -0.03440812602639198,
|
| 2019 |
+
"rewards/margins": 0.0807172879576683,
|
| 2020 |
+
"rewards/rejected": -0.11512543261051178,
|
| 2021 |
+
"step": 1290
|
| 2022 |
+
},
|
| 2023 |
+
{
|
| 2024 |
+
"epoch": 3.38,
|
| 2025 |
+
"learning_rate": 3.6088515096305675e-07,
|
| 2026 |
+
"logits/chosen": -1.274196982383728,
|
| 2027 |
+
"logits/rejected": -1.2778257131576538,
|
| 2028 |
+
"logps/chosen": -36.133689880371094,
|
| 2029 |
+
"logps/rejected": -48.38513946533203,
|
| 2030 |
+
"loss": 0.9033,
|
| 2031 |
+
"rewards/accuracies": 0.875,
|
| 2032 |
+
"rewards/chosen": -0.026160676032304764,
|
| 2033 |
+
"rewards/margins": 0.09665848314762115,
|
| 2034 |
+
"rewards/rejected": -0.12281917035579681,
|
| 2035 |
+
"step": 1300
|
| 2036 |
+
},
|
| 2037 |
+
{
|
| 2038 |
+
"epoch": 3.38,
|
| 2039 |
+
"eval_logits/chosen": -1.5856225490570068,
|
| 2040 |
+
"eval_logits/rejected": -1.5817017555236816,
|
| 2041 |
+
"eval_logps/chosen": -39.73164367675781,
|
| 2042 |
+
"eval_logps/rejected": -44.086830139160156,
|
| 2043 |
+
"eval_loss": 0.9912906289100647,
|
| 2044 |
+
"eval_rewards/accuracies": 0.545265793800354,
|
| 2045 |
+
"eval_rewards/chosen": -0.056970901787281036,
|
| 2046 |
+
"eval_rewards/margins": 0.00873124971985817,
|
| 2047 |
+
"eval_rewards/rejected": -0.06570214778184891,
|
| 2048 |
+
"eval_runtime": 145.8872,
|
| 2049 |
+
"eval_samples_per_second": 2.351,
|
| 2050 |
+
"eval_steps_per_second": 0.295,
|
| 2051 |
+
"step": 1300
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 3.4,
|
| 2055 |
+
"learning_rate": 3.321087280364757e-07,
|
| 2056 |
+
"logits/chosen": -1.2609808444976807,
|
| 2057 |
+
"logits/rejected": -1.2618134021759033,
|
| 2058 |
+
"logps/chosen": -38.979774475097656,
|
| 2059 |
+
"logps/rejected": -48.166404724121094,
|
| 2060 |
+
"loss": 0.9165,
|
| 2061 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 2062 |
+
"rewards/chosen": -0.02735757827758789,
|
| 2063 |
+
"rewards/margins": 0.0834653228521347,
|
| 2064 |
+
"rewards/rejected": -0.11082291603088379,
|
| 2065 |
+
"step": 1310
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 3.43,
|
| 2069 |
+
"learning_rate": 3.044460665744284e-07,
|
| 2070 |
+
"logits/chosen": -1.3282499313354492,
|
| 2071 |
+
"logits/rejected": -1.3269433975219727,
|
| 2072 |
+
"logps/chosen": -34.594669342041016,
|
| 2073 |
+
"logps/rejected": -40.66201400756836,
|
| 2074 |
+
"loss": 0.9279,
|
| 2075 |
+
"rewards/accuracies": 0.75,
|
| 2076 |
+
"rewards/chosen": -0.02500814199447632,
|
| 2077 |
+
"rewards/margins": 0.07207842171192169,
|
| 2078 |
+
"rewards/rejected": -0.09708657115697861,
|
| 2079 |
+
"step": 1320
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 3.45,
|
| 2083 |
+
"learning_rate": 2.779113783626916e-07,
|
| 2084 |
+
"logits/chosen": -1.2696067094802856,
|
| 2085 |
+
"logits/rejected": -1.2711015939712524,
|
| 2086 |
+
"logps/chosen": -36.31203079223633,
|
| 2087 |
+
"logps/rejected": -43.704010009765625,
|
| 2088 |
+
"loss": 0.9188,
|
| 2089 |
+
"rewards/accuracies": 0.8125,
|
| 2090 |
+
"rewards/chosen": -0.019377777352929115,
|
| 2091 |
+
"rewards/margins": 0.08117558807134628,
|
| 2092 |
+
"rewards/rejected": -0.10055337101221085,
|
| 2093 |
+
"step": 1330
|
| 2094 |
+
},
|
| 2095 |
+
{
|
| 2096 |
+
"epoch": 3.48,
|
| 2097 |
+
"learning_rate": 2.5251829568697204e-07,
|
| 2098 |
+
"logits/chosen": -1.312336802482605,
|
| 2099 |
+
"logits/rejected": -1.3112585544586182,
|
| 2100 |
+
"logps/chosen": -33.062835693359375,
|
| 2101 |
+
"logps/rejected": -41.20447540283203,
|
| 2102 |
+
"loss": 0.9265,
|
| 2103 |
+
"rewards/accuracies": 0.800000011920929,
|
| 2104 |
+
"rewards/chosen": -0.018641695380210876,
|
| 2105 |
+
"rewards/margins": 0.07348335534334183,
|
| 2106 |
+
"rewards/rejected": -0.0921250432729721,
|
| 2107 |
+
"step": 1340
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 3.51,
|
| 2111 |
+
"learning_rate": 2.2827986432927774e-07,
|
| 2112 |
+
"logits/chosen": -1.3330471515655518,
|
| 2113 |
+
"logits/rejected": -1.318232774734497,
|
| 2114 |
+
"logps/chosen": -37.70689392089844,
|
| 2115 |
+
"logps/rejected": -48.62843322753906,
|
| 2116 |
+
"loss": 0.9103,
|
| 2117 |
+
"rewards/accuracies": 0.862500011920929,
|
| 2118 |
+
"rewards/chosen": -0.033902253955602646,
|
| 2119 |
+
"rewards/margins": 0.08973778784275055,
|
| 2120 |
+
"rewards/rejected": -0.1236400380730629,
|
| 2121 |
+
"step": 1350
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 3.53,
|
| 2125 |
+
"learning_rate": 2.0520853686560177e-07,
|
| 2126 |
+
"logits/chosen": -1.2979753017425537,
|
| 2127 |
+
"logits/rejected": -1.3076903820037842,
|
| 2128 |
+
"logps/chosen": -33.92186737060547,
|
| 2129 |
+
"logps/rejected": -42.06645965576172,
|
| 2130 |
+
"loss": 0.9203,
|
| 2131 |
+
"rewards/accuracies": 0.800000011920929,
|
| 2132 |
+
"rewards/chosen": -0.018533948808908463,
|
| 2133 |
+
"rewards/margins": 0.07973320782184601,
|
| 2134 |
+
"rewards/rejected": -0.09826715290546417,
|
| 2135 |
+
"step": 1360
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 3.56,
|
| 2139 |
+
"learning_rate": 1.833161662683672e-07,
|
| 2140 |
+
"logits/chosen": -1.4156075716018677,
|
| 2141 |
+
"logits/rejected": -1.4154436588287354,
|
| 2142 |
+
"logps/chosen": -33.43849563598633,
|
| 2143 |
+
"logps/rejected": -48.292179107666016,
|
| 2144 |
+
"loss": 0.8895,
|
| 2145 |
+
"rewards/accuracies": 0.887499988079071,
|
| 2146 |
+
"rewards/chosen": -0.014080649241805077,
|
| 2147 |
+
"rewards/margins": 0.11053232103586197,
|
| 2148 |
+
"rewards/rejected": -0.1246129646897316,
|
| 2149 |
+
"step": 1370
|
| 2150 |
+
},
|
| 2151 |
+
{
|
| 2152 |
+
"epoch": 3.58,
|
| 2153 |
+
"learning_rate": 1.626139998169246e-07,
|
| 2154 |
+
"logits/chosen": -1.2881263494491577,
|
| 2155 |
+
"logits/rejected": -1.295348882675171,
|
| 2156 |
+
"logps/chosen": -36.35157775878906,
|
| 2157 |
+
"logps/rejected": -48.889339447021484,
|
| 2158 |
+
"loss": 0.9073,
|
| 2159 |
+
"rewards/accuracies": 0.862500011920929,
|
| 2160 |
+
"rewards/chosen": -0.021710969507694244,
|
| 2161 |
+
"rewards/margins": 0.09268596768379211,
|
| 2162 |
+
"rewards/rejected": -0.11439694464206696,
|
| 2163 |
+
"step": 1380
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 3.61,
|
| 2167 |
+
"learning_rate": 1.4311267331922535e-07,
|
| 2168 |
+
"logits/chosen": -1.2490915060043335,
|
| 2169 |
+
"logits/rejected": -1.2452831268310547,
|
| 2170 |
+
"logps/chosen": -36.50374984741211,
|
| 2171 |
+
"logps/rejected": -40.645694732666016,
|
| 2172 |
+
"loss": 0.9269,
|
| 2173 |
+
"rewards/accuracies": 0.8125,
|
| 2174 |
+
"rewards/chosen": -0.01736537739634514,
|
| 2175 |
+
"rewards/margins": 0.07311762124300003,
|
| 2176 |
+
"rewards/rejected": -0.09048298746347427,
|
| 2177 |
+
"step": 1390
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 3.64,
|
| 2181 |
+
"learning_rate": 1.2482220564763669e-07,
|
| 2182 |
+
"logits/chosen": -1.37045419216156,
|
| 2183 |
+
"logits/rejected": -1.3693218231201172,
|
| 2184 |
+
"logps/chosen": -33.22849655151367,
|
| 2185 |
+
"logps/rejected": -41.22184371948242,
|
| 2186 |
+
"loss": 0.9285,
|
| 2187 |
+
"rewards/accuracies": 0.762499988079071,
|
| 2188 |
+
"rewards/chosen": -0.01931949146091938,
|
| 2189 |
+
"rewards/margins": 0.07145430892705917,
|
| 2190 |
+
"rewards/rejected": -0.090773805975914,
|
| 2191 |
+
"step": 1400
|
| 2192 |
+
},
|
| 2193 |
+
{
|
| 2194 |
+
"epoch": 3.64,
|
| 2195 |
+
"eval_logits/chosen": -1.586363673210144,
|
| 2196 |
+
"eval_logits/rejected": -1.5824512243270874,
|
| 2197 |
+
"eval_logps/chosen": -39.72161102294922,
|
| 2198 |
+
"eval_logps/rejected": -44.085166931152344,
|
| 2199 |
+
"eval_loss": 0.9912044405937195,
|
| 2200 |
+
"eval_rewards/accuracies": 0.5394518375396729,
|
| 2201 |
+
"eval_rewards/chosen": -0.056870583444833755,
|
| 2202 |
+
"eval_rewards/margins": 0.0088148582726717,
|
| 2203 |
+
"eval_rewards/rejected": -0.06568543612957001,
|
| 2204 |
+
"eval_runtime": 145.8196,
|
| 2205 |
+
"eval_samples_per_second": 2.352,
|
| 2206 |
+
"eval_steps_per_second": 0.295,
|
| 2207 |
+
"step": 1400
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 3.66,
|
| 2211 |
+
"learning_rate": 1.0775199359171346e-07,
|
| 2212 |
+
"logits/chosen": -1.3426971435546875,
|
| 2213 |
+
"logits/rejected": -1.337646484375,
|
| 2214 |
+
"logps/chosen": -36.17827224731445,
|
| 2215 |
+
"logps/rejected": -37.915504455566406,
|
| 2216 |
+
"loss": 0.9417,
|
| 2217 |
+
"rewards/accuracies": 0.675000011920929,
|
| 2218 |
+
"rewards/chosen": -0.025357365608215332,
|
| 2219 |
+
"rewards/margins": 0.05825306102633476,
|
| 2220 |
+
"rewards/rejected": -0.0836104229092598,
|
| 2221 |
+
"step": 1410
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 3.69,
|
| 2225 |
+
"learning_rate": 9.191080703056604e-08,
|
| 2226 |
+
"logits/chosen": -1.2913800477981567,
|
| 2227 |
+
"logits/rejected": -1.2924379110336304,
|
| 2228 |
+
"logps/chosen": -35.52798843383789,
|
| 2229 |
+
"logps/rejected": -44.0432243347168,
|
| 2230 |
+
"loss": 0.9304,
|
| 2231 |
+
"rewards/accuracies": 0.8125,
|
| 2232 |
+
"rewards/chosen": -0.021631117910146713,
|
| 2233 |
+
"rewards/margins": 0.06956102699041367,
|
| 2234 |
+
"rewards/rejected": -0.09119214862585068,
|
| 2235 |
+
"step": 1420
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 3.71,
|
| 2239 |
+
"learning_rate": 7.730678442730539e-08,
|
| 2240 |
+
"logits/chosen": -1.241626501083374,
|
| 2241 |
+
"logits/rejected": -1.2360563278198242,
|
| 2242 |
+
"logps/chosen": -35.85869216918945,
|
| 2243 |
+
"logps/rejected": -48.056610107421875,
|
| 2244 |
+
"loss": 0.9035,
|
| 2245 |
+
"rewards/accuracies": 0.8125,
|
| 2246 |
+
"rewards/chosen": -0.017169218510389328,
|
| 2247 |
+
"rewards/margins": 0.09650282561779022,
|
| 2248 |
+
"rewards/rejected": -0.11367203295230865,
|
| 2249 |
+
"step": 1430
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 3.74,
|
| 2253 |
+
"learning_rate": 6.394742864787806e-08,
|
| 2254 |
+
"logits/chosen": -1.2561070919036865,
|
| 2255 |
+
"logits/rejected": -1.2504056692123413,
|
| 2256 |
+
"logps/chosen": -31.515949249267578,
|
| 2257 |
+
"logps/rejected": -41.77183151245117,
|
| 2258 |
+
"loss": 0.9216,
|
| 2259 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 2260 |
+
"rewards/chosen": -0.02202703058719635,
|
| 2261 |
+
"rewards/margins": 0.07838472723960876,
|
| 2262 |
+
"rewards/rejected": -0.10041175782680511,
|
| 2263 |
+
"step": 1440
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 3.77,
|
| 2267 |
+
"learning_rate": 5.183960310644748e-08,
|
| 2268 |
+
"logits/chosen": -1.3071388006210327,
|
| 2269 |
+
"logits/rejected": -1.2970083951950073,
|
| 2270 |
+
"logps/chosen": -35.57398986816406,
|
| 2271 |
+
"logps/rejected": -45.840171813964844,
|
| 2272 |
+
"loss": 0.9254,
|
| 2273 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 2274 |
+
"rewards/chosen": -0.030924629420042038,
|
| 2275 |
+
"rewards/margins": 0.0746234804391861,
|
| 2276 |
+
"rewards/rejected": -0.10554809868335724,
|
| 2277 |
+
"step": 1450
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 3.79,
|
| 2281 |
+
"learning_rate": 4.098952823928693e-08,
|
| 2282 |
+
"logits/chosen": -1.2653499841690063,
|
| 2283 |
+
"logits/rejected": -1.2617781162261963,
|
| 2284 |
+
"logps/chosen": -35.99557876586914,
|
| 2285 |
+
"logps/rejected": -39.989891052246094,
|
| 2286 |
+
"loss": 0.9359,
|
| 2287 |
+
"rewards/accuracies": 0.75,
|
| 2288 |
+
"rewards/chosen": -0.026857968419790268,
|
| 2289 |
+
"rewards/margins": 0.06413265317678452,
|
| 2290 |
+
"rewards/rejected": -0.09099061787128448,
|
| 2291 |
+
"step": 1460
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 3.82,
|
| 2295 |
+
"learning_rate": 3.1402778309014284e-08,
|
| 2296 |
+
"logits/chosen": -1.3204753398895264,
|
| 2297 |
+
"logits/rejected": -1.3249455690383911,
|
| 2298 |
+
"logps/chosen": -33.896812438964844,
|
| 2299 |
+
"logps/rejected": -42.878475189208984,
|
| 2300 |
+
"loss": 0.9191,
|
| 2301 |
+
"rewards/accuracies": 0.8125,
|
| 2302 |
+
"rewards/chosen": -0.018276356160640717,
|
| 2303 |
+
"rewards/margins": 0.08088420331478119,
|
| 2304 |
+
"rewards/rejected": -0.0991605594754219,
|
| 2305 |
+
"step": 1470
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 3.84,
|
| 2309 |
+
"learning_rate": 2.3084278540791427e-08,
|
| 2310 |
+
"logits/chosen": -1.3220521211624146,
|
| 2311 |
+
"logits/rejected": -1.3311010599136353,
|
| 2312 |
+
"logps/chosen": -33.49136734008789,
|
| 2313 |
+
"logps/rejected": -38.53743362426758,
|
| 2314 |
+
"loss": 0.9289,
|
| 2315 |
+
"rewards/accuracies": 0.824999988079071,
|
| 2316 |
+
"rewards/chosen": -0.017961082980036736,
|
| 2317 |
+
"rewards/margins": 0.07112761586904526,
|
| 2318 |
+
"rewards/rejected": -0.08908869326114655,
|
| 2319 |
+
"step": 1480
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 3.87,
|
| 2323 |
+
"learning_rate": 1.6038302591975807e-08,
|
| 2324 |
+
"logits/chosen": -1.258320689201355,
|
| 2325 |
+
"logits/rejected": -1.2530874013900757,
|
| 2326 |
+
"logps/chosen": -36.1824951171875,
|
| 2327 |
+
"logps/rejected": -41.13605880737305,
|
| 2328 |
+
"loss": 0.936,
|
| 2329 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 2330 |
+
"rewards/chosen": -0.02526010572910309,
|
| 2331 |
+
"rewards/margins": 0.06399567425251007,
|
| 2332 |
+
"rewards/rejected": -0.08925577253103256,
|
| 2333 |
+
"step": 1490
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 3.9,
|
| 2337 |
+
"learning_rate": 1.0268470356514237e-08,
|
| 2338 |
+
"logits/chosen": -1.3191821575164795,
|
| 2339 |
+
"logits/rejected": -1.3164719343185425,
|
| 2340 |
+
"logps/chosen": -36.40666961669922,
|
| 2341 |
+
"logps/rejected": -44.079254150390625,
|
| 2342 |
+
"loss": 0.9196,
|
| 2343 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 2344 |
+
"rewards/chosen": -0.02781098522245884,
|
| 2345 |
+
"rewards/margins": 0.08037930727005005,
|
| 2346 |
+
"rewards/rejected": -0.10819028317928314,
|
| 2347 |
+
"step": 1500
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 3.9,
|
| 2351 |
+
"eval_logits/chosen": -1.587085485458374,
|
| 2352 |
+
"eval_logits/rejected": -1.5831539630889893,
|
| 2353 |
+
"eval_logps/chosen": -39.70748519897461,
|
| 2354 |
+
"eval_logps/rejected": -44.0648307800293,
|
| 2355 |
+
"eval_loss": 0.9912670850753784,
|
| 2356 |
+
"eval_rewards/accuracies": 0.5423588156700134,
|
| 2357 |
+
"eval_rewards/chosen": -0.0567292720079422,
|
| 2358 |
+
"eval_rewards/margins": 0.008752820082008839,
|
| 2359 |
+
"eval_rewards/rejected": -0.06548209488391876,
|
| 2360 |
+
"eval_runtime": 145.8973,
|
| 2361 |
+
"eval_samples_per_second": 2.351,
|
| 2362 |
+
"eval_steps_per_second": 0.295,
|
| 2363 |
+
"step": 1500
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 3.92,
|
| 2367 |
+
"learning_rate": 5.777746105209147e-09,
|
| 2368 |
+
"logits/chosen": -1.3802618980407715,
|
| 2369 |
+
"logits/rejected": -1.3801379203796387,
|
| 2370 |
+
"logps/chosen": -31.544485092163086,
|
| 2371 |
+
"logps/rejected": -42.76384735107422,
|
| 2372 |
+
"loss": 0.917,
|
| 2373 |
+
"rewards/accuracies": 0.75,
|
| 2374 |
+
"rewards/chosen": -0.017282087355852127,
|
| 2375 |
+
"rewards/margins": 0.08295184373855591,
|
| 2376 |
+
"rewards/rejected": -0.10023393481969833,
|
| 2377 |
+
"step": 1510
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 3.95,
|
| 2381 |
+
"learning_rate": 2.5684369628148352e-09,
|
| 2382 |
+
"logits/chosen": -1.2462044954299927,
|
| 2383 |
+
"logits/rejected": -1.2448112964630127,
|
| 2384 |
+
"logps/chosen": -34.93328857421875,
|
| 2385 |
+
"logps/rejected": -43.264244079589844,
|
| 2386 |
+
"loss": 0.925,
|
| 2387 |
+
"rewards/accuracies": 0.800000011920929,
|
| 2388 |
+
"rewards/chosen": -0.01891419291496277,
|
| 2389 |
+
"rewards/margins": 0.07499094307422638,
|
| 2390 |
+
"rewards/rejected": -0.09390512108802795,
|
| 2391 |
+
"step": 1520
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 3.97,
|
| 2395 |
+
"learning_rate": 6.421917227455999e-10,
|
| 2396 |
+
"logits/chosen": -1.3795892000198364,
|
| 2397 |
+
"logits/rejected": -1.3771770000457764,
|
| 2398 |
+
"logps/chosen": -33.931644439697266,
|
| 2399 |
+
"logps/rejected": -40.94723892211914,
|
| 2400 |
+
"loss": 0.9299,
|
| 2401 |
+
"rewards/accuracies": 0.8125,
|
| 2402 |
+
"rewards/chosen": -0.024575814604759216,
|
| 2403 |
+
"rewards/margins": 0.07006003707647324,
|
| 2404 |
+
"rewards/rejected": -0.09463585913181305,
|
| 2405 |
+
"step": 1530
|
| 2406 |
+
},
|
| 2407 |
+
{
|
| 2408 |
+
"epoch": 4.0,
|
| 2409 |
+
"learning_rate": 0.0,
|
| 2410 |
+
"logits/chosen": -1.3671410083770752,
|
| 2411 |
+
"logits/rejected": -1.3676393032073975,
|
| 2412 |
+
"logps/chosen": -33.82143020629883,
|
| 2413 |
+
"logps/rejected": -37.72132873535156,
|
| 2414 |
+
"loss": 0.9527,
|
| 2415 |
+
"rewards/accuracies": 0.762499988079071,
|
| 2416 |
+
"rewards/chosen": -0.0320148840546608,
|
| 2417 |
+
"rewards/margins": 0.04729658365249634,
|
| 2418 |
+
"rewards/rejected": -0.07931147515773773,
|
| 2419 |
+
"step": 1540
|
| 2420 |
+
},
|
| 2421 |
+
{
|
| 2422 |
+
"epoch": 4.0,
|
| 2423 |
+
"step": 1540,
|
| 2424 |
"total_flos": 0.0,
|
| 2425 |
+
"train_loss": 0.7654441864459546,
|
| 2426 |
+
"train_runtime": 10797.8934,
|
| 2427 |
+
"train_samples_per_second": 1.141,
|
| 2428 |
+
"train_steps_per_second": 0.143
|
| 2429 |
}
|
| 2430 |
],
|
| 2431 |
"logging_steps": 10,
|
| 2432 |
+
"max_steps": 1540,
|
| 2433 |
"num_input_tokens_seen": 0,
|
| 2434 |
+
"num_train_epochs": 4,
|
| 2435 |
"save_steps": 100,
|
| 2436 |
"total_flos": 0.0,
|
| 2437 |
"train_batch_size": 4,
|