Model save
Browse files- README.md +2 -6
- adapter_model.safetensors +1 -1
- all_results.json +5 -5
- runs/May15_12-44-33_gpu4-119-5/events.out.tfevents.1715741180.gpu4-119-5.3031836.0 +2 -2
- train_results.json +5 -5
- trainer_state.json +725 -67
README.md
CHANGED
|
@@ -2,14 +2,10 @@
|
|
| 2 |
license: mit
|
| 3 |
library_name: peft
|
| 4 |
tags:
|
| 5 |
-
- alignment-handbook
|
| 6 |
-
- generated_from_trainer
|
| 7 |
- trl
|
| 8 |
- dpo
|
| 9 |
- generated_from_trainer
|
| 10 |
base_model: DUAL-GPO/phi-2-gpo-new-i0
|
| 11 |
-
datasets:
|
| 12 |
-
- HuggingFaceH4/ultrafeedback_binarized
|
| 13 |
model-index:
|
| 14 |
- name: phi-2-gpo-v25-i1
|
| 15 |
results: []
|
|
@@ -20,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 20 |
|
| 21 |
# phi-2-gpo-v25-i1
|
| 22 |
|
| 23 |
-
This model is a fine-tuned version of [DUAL-GPO/phi-2-gpo-new-i0](https://huggingface.co/DUAL-GPO/phi-2-gpo-new-i0) on the
|
| 24 |
|
| 25 |
## Model description
|
| 26 |
|
|
@@ -51,7 +47,7 @@ The following hyperparameters were used during training:
|
|
| 51 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 52 |
- lr_scheduler_type: cosine
|
| 53 |
- lr_scheduler_warmup_ratio: 0.1
|
| 54 |
-
- num_epochs:
|
| 55 |
|
| 56 |
### Training results
|
| 57 |
|
|
|
|
| 2 |
license: mit
|
| 3 |
library_name: peft
|
| 4 |
tags:
|
|
|
|
|
|
|
| 5 |
- trl
|
| 6 |
- dpo
|
| 7 |
- generated_from_trainer
|
| 8 |
base_model: DUAL-GPO/phi-2-gpo-new-i0
|
|
|
|
|
|
|
| 9 |
model-index:
|
| 10 |
- name: phi-2-gpo-v25-i1
|
| 11 |
results: []
|
|
|
|
| 16 |
|
| 17 |
# phi-2-gpo-v25-i1
|
| 18 |
|
| 19 |
+
This model is a fine-tuned version of [DUAL-GPO/phi-2-gpo-new-i0](https://huggingface.co/DUAL-GPO/phi-2-gpo-new-i0) on the None dataset.
|
| 20 |
|
| 21 |
## Model description
|
| 22 |
|
|
|
|
| 47 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 48 |
- lr_scheduler_type: cosine
|
| 49 |
- lr_scheduler_warmup_ratio: 0.1
|
| 50 |
+
- num_epochs: 2
|
| 51 |
|
| 52 |
### Training results
|
| 53 |
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 167807296
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2b5d797078e8b67d31c2e54c050acbdd4fa7f6d219b5406d2ede8fd8b729d61e
|
| 3 |
size 167807296
|
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 15000,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"train_loss": 0.16023516272887206,
|
| 4 |
+
"train_runtime": 6530.1152,
|
| 5 |
"train_samples": 15000,
|
| 6 |
+
"train_samples_per_second": 4.594,
|
| 7 |
+
"train_steps_per_second": 0.143
|
| 8 |
}
|
runs/May15_12-44-33_gpu4-119-5/events.out.tfevents.1715741180.gpu4-119-5.3031836.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:61b284565f9fa8ba8bba6150923478851a8e66f9c342c3e786548af895f6e2d8
|
| 3 |
+
size 38783
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 15000,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"train_loss": 0.16023516272887206,
|
| 4 |
+
"train_runtime": 6530.1152,
|
| 5 |
"train_samples": 15000,
|
| 6 |
+
"train_samples_per_second": 4.594,
|
| 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": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -584,102 +584,760 @@
|
|
| 584 |
},
|
| 585 |
{
|
| 586 |
"epoch": 0.87,
|
| 587 |
-
"learning_rate":
|
| 588 |
-
"logits/chosen": 0.
|
| 589 |
-
"logits/rejected": 0.
|
| 590 |
-
"logps/chosen": -
|
| 591 |
-
"logps/rejected": -
|
| 592 |
-
"loss": 0.
|
| 593 |
-
"rewards/accuracies": 0.
|
| 594 |
-
"rewards/chosen": -0.
|
| 595 |
-
"rewards/margins": 0.
|
| 596 |
-
"rewards/rejected": -0.
|
| 597 |
"step": 410
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 0.9,
|
| 601 |
-
"learning_rate":
|
| 602 |
-
"logits/chosen": 0.
|
| 603 |
-
"logits/rejected": 0.
|
| 604 |
-
"logps/chosen": -
|
| 605 |
-
"logps/rejected": -
|
| 606 |
-
"loss": 0.
|
| 607 |
"rewards/accuracies": 0.706250011920929,
|
| 608 |
-
"rewards/chosen": -0.
|
| 609 |
-
"rewards/margins": 0.
|
| 610 |
-
"rewards/rejected": -0.
|
| 611 |
"step": 420
|
| 612 |
},
|
| 613 |
{
|
| 614 |
"epoch": 0.92,
|
| 615 |
-
"learning_rate":
|
| 616 |
-
"logits/chosen": 0.
|
| 617 |
-
"logits/rejected": 0.
|
| 618 |
-
"logps/chosen": -
|
| 619 |
-
"logps/rejected": -
|
| 620 |
-
"loss": 0.
|
| 621 |
-
"rewards/accuracies": 0.
|
| 622 |
-
"rewards/chosen": -0.
|
| 623 |
-
"rewards/margins": 0.
|
| 624 |
-
"rewards/rejected": -0.
|
| 625 |
"step": 430
|
| 626 |
},
|
| 627 |
{
|
| 628 |
"epoch": 0.94,
|
| 629 |
-
"learning_rate":
|
| 630 |
-
"logits/chosen": 0.
|
| 631 |
-
"logits/rejected": 0.
|
| 632 |
-
"logps/chosen": -
|
| 633 |
-
"logps/rejected": -
|
| 634 |
-
"loss": 0.
|
| 635 |
"rewards/accuracies": 0.6812499761581421,
|
| 636 |
-
"rewards/chosen": -0.
|
| 637 |
-
"rewards/margins": 0.
|
| 638 |
-
"rewards/rejected": -0.
|
| 639 |
"step": 440
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 0.96,
|
| 643 |
-
"learning_rate":
|
| 644 |
-
"logits/chosen": 0.
|
| 645 |
-
"logits/rejected": 0.
|
| 646 |
-
"logps/chosen": -
|
| 647 |
-
"logps/rejected": -
|
| 648 |
-
"loss": 0.
|
| 649 |
-
"rewards/accuracies": 0.
|
| 650 |
-
"rewards/chosen": -0.
|
| 651 |
-
"rewards/margins": 0.
|
| 652 |
-
"rewards/rejected": -0.
|
| 653 |
"step": 450
|
| 654 |
},
|
| 655 |
{
|
| 656 |
"epoch": 0.98,
|
| 657 |
-
"learning_rate":
|
| 658 |
-
"logits/chosen": 0.
|
| 659 |
-
"logits/rejected": 0.
|
| 660 |
-
"logps/chosen": -
|
| 661 |
-
"logps/rejected": -
|
| 662 |
-
"loss": 0.
|
| 663 |
-
"rewards/accuracies": 0.
|
| 664 |
-
"rewards/chosen": -0.
|
| 665 |
-
"rewards/margins": 0.
|
| 666 |
-
"rewards/rejected": -0.
|
| 667 |
"step": 460
|
| 668 |
},
|
| 669 |
{
|
| 670 |
"epoch": 1.0,
|
| 671 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 672 |
"total_flos": 0.0,
|
| 673 |
-
"train_loss": 0.
|
| 674 |
-
"train_runtime":
|
| 675 |
-
"train_samples_per_second":
|
| 676 |
-
"train_steps_per_second": 0.
|
| 677 |
}
|
| 678 |
],
|
| 679 |
"logging_steps": 10,
|
| 680 |
-
"max_steps":
|
| 681 |
"num_input_tokens_seen": 0,
|
| 682 |
-
"num_train_epochs":
|
| 683 |
"save_steps": 100,
|
| 684 |
"total_flos": 0.0,
|
| 685 |
"train_batch_size": 4,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.9968,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 936,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 584 |
},
|
| 585 |
{
|
| 586 |
"epoch": 0.87,
|
| 587 |
+
"learning_rate": 3.4545537312690565e-06,
|
| 588 |
+
"logits/chosen": 0.1286882907152176,
|
| 589 |
+
"logits/rejected": 0.13096146285533905,
|
| 590 |
+
"logps/chosen": -515.2964477539062,
|
| 591 |
+
"logps/rejected": -708.9202270507812,
|
| 592 |
+
"loss": 0.2656,
|
| 593 |
+
"rewards/accuracies": 0.737500011920929,
|
| 594 |
+
"rewards/chosen": -0.19799669086933136,
|
| 595 |
+
"rewards/margins": 0.1586463451385498,
|
| 596 |
+
"rewards/rejected": -0.35664302110671997,
|
| 597 |
"step": 410
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 0.9,
|
| 601 |
+
"learning_rate": 3.3676987756445894e-06,
|
| 602 |
+
"logits/chosen": 0.13132785260677338,
|
| 603 |
+
"logits/rejected": 0.13156192004680634,
|
| 604 |
+
"logps/chosen": -523.5271606445312,
|
| 605 |
+
"logps/rejected": -717.6295166015625,
|
| 606 |
+
"loss": 0.3083,
|
| 607 |
"rewards/accuracies": 0.706250011920929,
|
| 608 |
+
"rewards/chosen": -0.2122844010591507,
|
| 609 |
+
"rewards/margins": 0.15919235348701477,
|
| 610 |
+
"rewards/rejected": -0.3714767396450043,
|
| 611 |
"step": 420
|
| 612 |
},
|
| 613 |
{
|
| 614 |
"epoch": 0.92,
|
| 615 |
+
"learning_rate": 3.2796360224151587e-06,
|
| 616 |
+
"logits/chosen": 0.1404297947883606,
|
| 617 |
+
"logits/rejected": 0.13226640224456787,
|
| 618 |
+
"logps/chosen": -557.2911987304688,
|
| 619 |
+
"logps/rejected": -633.6729736328125,
|
| 620 |
+
"loss": 0.2935,
|
| 621 |
+
"rewards/accuracies": 0.612500011920929,
|
| 622 |
+
"rewards/chosen": -0.22657549381256104,
|
| 623 |
+
"rewards/margins": 0.11841969192028046,
|
| 624 |
+
"rewards/rejected": -0.3449952006340027,
|
| 625 |
"step": 430
|
| 626 |
},
|
| 627 |
{
|
| 628 |
"epoch": 0.94,
|
| 629 |
+
"learning_rate": 3.1904880509659397e-06,
|
| 630 |
+
"logits/chosen": 0.12834230065345764,
|
| 631 |
+
"logits/rejected": 0.22366562485694885,
|
| 632 |
+
"logps/chosen": -551.0172119140625,
|
| 633 |
+
"logps/rejected": -710.9666748046875,
|
| 634 |
+
"loss": 0.2803,
|
| 635 |
"rewards/accuracies": 0.6812499761581421,
|
| 636 |
+
"rewards/chosen": -0.22843453288078308,
|
| 637 |
+
"rewards/margins": 0.14828652143478394,
|
| 638 |
+
"rewards/rejected": -0.376721054315567,
|
| 639 |
"step": 440
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 0.96,
|
| 643 |
+
"learning_rate": 3.100378951256981e-06,
|
| 644 |
+
"logits/chosen": 0.11333411931991577,
|
| 645 |
+
"logits/rejected": 0.18246471881866455,
|
| 646 |
+
"logps/chosen": -568.0606689453125,
|
| 647 |
+
"logps/rejected": -714.0242309570312,
|
| 648 |
+
"loss": 0.2998,
|
| 649 |
+
"rewards/accuracies": 0.675000011920929,
|
| 650 |
+
"rewards/chosen": -0.24061532318592072,
|
| 651 |
+
"rewards/margins": 0.12436308711767197,
|
| 652 |
+
"rewards/rejected": -0.3649784326553345,
|
| 653 |
"step": 450
|
| 654 |
},
|
| 655 |
{
|
| 656 |
"epoch": 0.98,
|
| 657 |
+
"learning_rate": 3.0094341510955697e-06,
|
| 658 |
+
"logits/chosen": 0.14087212085723877,
|
| 659 |
+
"logits/rejected": 0.2087050974369049,
|
| 660 |
+
"logps/chosen": -552.45947265625,
|
| 661 |
+
"logps/rejected": -642.1560668945312,
|
| 662 |
+
"loss": 0.2889,
|
| 663 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 664 |
+
"rewards/chosen": -0.20147624611854553,
|
| 665 |
+
"rewards/margins": 0.1283041536808014,
|
| 666 |
+
"rewards/rejected": -0.32978036999702454,
|
| 667 |
"step": 460
|
| 668 |
},
|
| 669 |
{
|
| 670 |
"epoch": 1.0,
|
| 671 |
+
"learning_rate": 2.9177802415463714e-06,
|
| 672 |
+
"logits/chosen": 0.16158871352672577,
|
| 673 |
+
"logits/rejected": 0.23096832633018494,
|
| 674 |
+
"logps/chosen": -573.5584716796875,
|
| 675 |
+
"logps/rejected": -712.2202758789062,
|
| 676 |
+
"loss": 0.2774,
|
| 677 |
+
"rewards/accuracies": 0.706250011920929,
|
| 678 |
+
"rewards/chosen": -0.22124040126800537,
|
| 679 |
+
"rewards/margins": 0.14029137790203094,
|
| 680 |
+
"rewards/rejected": -0.3615317940711975,
|
| 681 |
+
"step": 470
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 1.02,
|
| 685 |
+
"learning_rate": 2.825544800722376e-06,
|
| 686 |
+
"logits/chosen": 0.1098761335015297,
|
| 687 |
+
"logits/rejected": 0.2546694278717041,
|
| 688 |
+
"logps/chosen": -615.3218994140625,
|
| 689 |
+
"logps/rejected": -713.9712524414062,
|
| 690 |
+
"loss": 0.2699,
|
| 691 |
+
"rewards/accuracies": 0.643750011920929,
|
| 692 |
+
"rewards/chosen": -0.23255982995033264,
|
| 693 |
+
"rewards/margins": 0.13802029192447662,
|
| 694 |
+
"rewards/rejected": -0.3705800771713257,
|
| 695 |
+
"step": 480
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 1.05,
|
| 699 |
+
"learning_rate": 2.732856216201906e-06,
|
| 700 |
+
"logits/chosen": 0.08082452416419983,
|
| 701 |
+
"logits/rejected": 0.1814456284046173,
|
| 702 |
+
"logps/chosen": -614.1076049804688,
|
| 703 |
+
"logps/rejected": -726.0872192382812,
|
| 704 |
+
"loss": 0.2972,
|
| 705 |
+
"rewards/accuracies": 0.699999988079071,
|
| 706 |
+
"rewards/chosen": -0.24807974696159363,
|
| 707 |
+
"rewards/margins": 0.13322189450263977,
|
| 708 |
+
"rewards/rejected": -0.3813017010688782,
|
| 709 |
+
"step": 490
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 1.07,
|
| 713 |
+
"learning_rate": 2.639843506318899e-06,
|
| 714 |
+
"logits/chosen": 0.1431538164615631,
|
| 715 |
+
"logits/rejected": 0.16376420855522156,
|
| 716 |
+
"logps/chosen": -603.7203369140625,
|
| 717 |
+
"logps/rejected": -724.11669921875,
|
| 718 |
+
"loss": 0.2644,
|
| 719 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 720 |
+
"rewards/chosen": -0.24805834889411926,
|
| 721 |
+
"rewards/margins": 0.13241049647331238,
|
| 722 |
+
"rewards/rejected": -0.38046884536743164,
|
| 723 |
+
"step": 500
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 1.09,
|
| 727 |
+
"learning_rate": 2.5466361405751914e-06,
|
| 728 |
+
"logits/chosen": 0.09880776703357697,
|
| 729 |
+
"logits/rejected": 0.15729175508022308,
|
| 730 |
+
"logps/chosen": -563.4605712890625,
|
| 731 |
+
"logps/rejected": -697.0057373046875,
|
| 732 |
+
"loss": 0.3066,
|
| 733 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 734 |
+
"rewards/chosen": -0.22475488483905792,
|
| 735 |
+
"rewards/margins": 0.13303039968013763,
|
| 736 |
+
"rewards/rejected": -0.35778528451919556,
|
| 737 |
+
"step": 510
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.11,
|
| 741 |
+
"learning_rate": 2.4533638594248094e-06,
|
| 742 |
+
"logits/chosen": 0.08443008363246918,
|
| 743 |
+
"logits/rejected": 0.15599215030670166,
|
| 744 |
+
"logps/chosen": -553.6532592773438,
|
| 745 |
+
"logps/rejected": -658.5653076171875,
|
| 746 |
+
"loss": 0.2935,
|
| 747 |
+
"rewards/accuracies": 0.643750011920929,
|
| 748 |
+
"rewards/chosen": -0.22460079193115234,
|
| 749 |
+
"rewards/margins": 0.11079039424657822,
|
| 750 |
+
"rewards/rejected": -0.3353911340236664,
|
| 751 |
+
"step": 520
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.13,
|
| 755 |
+
"learning_rate": 2.360156493681102e-06,
|
| 756 |
+
"logits/chosen": 0.10354779660701752,
|
| 757 |
+
"logits/rejected": 0.14787225425243378,
|
| 758 |
+
"logps/chosen": -592.5408325195312,
|
| 759 |
+
"logps/rejected": -705.7700805664062,
|
| 760 |
+
"loss": 0.2888,
|
| 761 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 762 |
+
"rewards/chosen": -0.25297555327415466,
|
| 763 |
+
"rewards/margins": 0.1141052097082138,
|
| 764 |
+
"rewards/rejected": -0.36708077788352966,
|
| 765 |
+
"step": 530
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 1.15,
|
| 769 |
+
"learning_rate": 2.2671437837980943e-06,
|
| 770 |
+
"logits/chosen": 0.08793269097805023,
|
| 771 |
+
"logits/rejected": 0.17164576053619385,
|
| 772 |
+
"logps/chosen": -603.9788208007812,
|
| 773 |
+
"logps/rejected": -723.2301025390625,
|
| 774 |
+
"loss": 0.284,
|
| 775 |
+
"rewards/accuracies": 0.643750011920929,
|
| 776 |
+
"rewards/chosen": -0.2590700089931488,
|
| 777 |
+
"rewards/margins": 0.13130523264408112,
|
| 778 |
+
"rewards/rejected": -0.3903752863407135,
|
| 779 |
+
"step": 540
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 1.17,
|
| 783 |
+
"learning_rate": 2.1744551992776247e-06,
|
| 784 |
+
"logits/chosen": 0.09502136707305908,
|
| 785 |
+
"logits/rejected": 0.12443023920059204,
|
| 786 |
+
"logps/chosen": -556.7479858398438,
|
| 787 |
+
"logps/rejected": -695.36669921875,
|
| 788 |
+
"loss": 0.2874,
|
| 789 |
+
"rewards/accuracies": 0.668749988079071,
|
| 790 |
+
"rewards/chosen": -0.24774689972400665,
|
| 791 |
+
"rewards/margins": 0.13886813819408417,
|
| 792 |
+
"rewards/rejected": -0.3866150975227356,
|
| 793 |
+
"step": 550
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 1.19,
|
| 797 |
+
"learning_rate": 2.082219758453629e-06,
|
| 798 |
+
"logits/chosen": 0.03692219406366348,
|
| 799 |
+
"logits/rejected": 0.1840008795261383,
|
| 800 |
+
"logps/chosen": -606.75,
|
| 801 |
+
"logps/rejected": -705.9107666015625,
|
| 802 |
+
"loss": 0.2798,
|
| 803 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 804 |
+
"rewards/chosen": -0.27475666999816895,
|
| 805 |
+
"rewards/margins": 0.11195492744445801,
|
| 806 |
+
"rewards/rejected": -0.38671156764030457,
|
| 807 |
+
"step": 560
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 1.22,
|
| 811 |
+
"learning_rate": 1.990565848904431e-06,
|
| 812 |
+
"logits/chosen": 0.07638466358184814,
|
| 813 |
+
"logits/rejected": 0.12104539573192596,
|
| 814 |
+
"logps/chosen": -611.66650390625,
|
| 815 |
+
"logps/rejected": -736.2400512695312,
|
| 816 |
+
"loss": 0.3029,
|
| 817 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 818 |
+
"rewards/chosen": -0.2728550434112549,
|
| 819 |
+
"rewards/margins": 0.14484602212905884,
|
| 820 |
+
"rewards/rejected": -0.4177010655403137,
|
| 821 |
+
"step": 570
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 1.24,
|
| 825 |
+
"learning_rate": 1.899621048743019e-06,
|
| 826 |
+
"logits/chosen": 0.12088838964700699,
|
| 827 |
+
"logits/rejected": 0.07981579750776291,
|
| 828 |
+
"logps/chosen": -601.095947265625,
|
| 829 |
+
"logps/rejected": -716.2054443359375,
|
| 830 |
+
"loss": 0.2954,
|
| 831 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 832 |
+
"rewards/chosen": -0.2841927409172058,
|
| 833 |
+
"rewards/margins": 0.11602810770273209,
|
| 834 |
+
"rewards/rejected": -0.4002207815647125,
|
| 835 |
+
"step": 580
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 1.26,
|
| 839 |
+
"learning_rate": 1.8095119490340618e-06,
|
| 840 |
+
"logits/chosen": 0.1057354211807251,
|
| 841 |
+
"logits/rejected": 0.20554998517036438,
|
| 842 |
+
"logps/chosen": -656.5835571289062,
|
| 843 |
+
"logps/rejected": -745.124267578125,
|
| 844 |
+
"loss": 0.2951,
|
| 845 |
+
"rewards/accuracies": 0.706250011920929,
|
| 846 |
+
"rewards/chosen": -0.28658175468444824,
|
| 847 |
+
"rewards/margins": 0.13431188464164734,
|
| 848 |
+
"rewards/rejected": -0.42089366912841797,
|
| 849 |
+
"step": 590
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 1.28,
|
| 853 |
+
"learning_rate": 1.7203639775848423e-06,
|
| 854 |
+
"logits/chosen": 0.08960368484258652,
|
| 855 |
+
"logits/rejected": 0.1500837355852127,
|
| 856 |
+
"logps/chosen": -605.3450317382812,
|
| 857 |
+
"logps/rejected": -729.7435913085938,
|
| 858 |
+
"loss": 0.2824,
|
| 859 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 860 |
+
"rewards/chosen": -0.26503902673721313,
|
| 861 |
+
"rewards/margins": 0.13953007757663727,
|
| 862 |
+
"rewards/rejected": -0.4045690894126892,
|
| 863 |
+
"step": 600
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 1.3,
|
| 867 |
+
"learning_rate": 1.632301224355411e-06,
|
| 868 |
+
"logits/chosen": 0.0958481878042221,
|
| 869 |
+
"logits/rejected": 0.14852683246135712,
|
| 870 |
+
"logps/chosen": -632.3499145507812,
|
| 871 |
+
"logps/rejected": -763.3027954101562,
|
| 872 |
+
"loss": 0.2912,
|
| 873 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 874 |
+
"rewards/chosen": -0.26938048005104065,
|
| 875 |
+
"rewards/margins": 0.15196532011032104,
|
| 876 |
+
"rewards/rejected": -0.4213457703590393,
|
| 877 |
+
"step": 610
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 1.32,
|
| 881 |
+
"learning_rate": 1.5454462687309445e-06,
|
| 882 |
+
"logits/chosen": 0.07703549414873123,
|
| 883 |
+
"logits/rejected": 0.14966604113578796,
|
| 884 |
+
"logps/chosen": -574.7545166015625,
|
| 885 |
+
"logps/rejected": -733.3427734375,
|
| 886 |
+
"loss": 0.2753,
|
| 887 |
+
"rewards/accuracies": 0.71875,
|
| 888 |
+
"rewards/chosen": -0.2730166018009186,
|
| 889 |
+
"rewards/margins": 0.15502889454364777,
|
| 890 |
+
"rewards/rejected": -0.42804545164108276,
|
| 891 |
+
"step": 620
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1.34,
|
| 895 |
+
"learning_rate": 1.4599200088967652e-06,
|
| 896 |
+
"logits/chosen": 0.06834821403026581,
|
| 897 |
+
"logits/rejected": 0.09840573370456696,
|
| 898 |
+
"logps/chosen": -588.8328857421875,
|
| 899 |
+
"logps/rejected": -734.0791625976562,
|
| 900 |
+
"loss": 0.2827,
|
| 901 |
+
"rewards/accuracies": 0.637499988079071,
|
| 902 |
+
"rewards/chosen": -0.29259082674980164,
|
| 903 |
+
"rewards/margins": 0.11803393065929413,
|
| 904 |
+
"rewards/rejected": -0.41062480211257935,
|
| 905 |
+
"step": 630
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 1.37,
|
| 909 |
+
"learning_rate": 1.3758414935535147e-06,
|
| 910 |
+
"logits/chosen": 0.05101291462779045,
|
| 911 |
+
"logits/rejected": 0.19463148713111877,
|
| 912 |
+
"logps/chosen": -635.7034912109375,
|
| 913 |
+
"logps/rejected": -769.3609619140625,
|
| 914 |
+
"loss": 0.287,
|
| 915 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 916 |
+
"rewards/chosen": -0.27811720967292786,
|
| 917 |
+
"rewards/margins": 0.15194621682167053,
|
| 918 |
+
"rewards/rejected": -0.4300634264945984,
|
| 919 |
+
"step": 640
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 1.39,
|
| 923 |
+
"learning_rate": 1.293327756206729e-06,
|
| 924 |
+
"logits/chosen": 0.06910277158021927,
|
| 925 |
+
"logits/rejected": 0.08267343044281006,
|
| 926 |
+
"logps/chosen": -630.0482177734375,
|
| 927 |
+
"logps/rejected": -760.6334228515625,
|
| 928 |
+
"loss": 0.2618,
|
| 929 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 930 |
+
"rewards/chosen": -0.28737881779670715,
|
| 931 |
+
"rewards/margins": 0.14742901921272278,
|
| 932 |
+
"rewards/rejected": -0.4348078668117523,
|
| 933 |
+
"step": 650
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 1.41,
|
| 937 |
+
"learning_rate": 1.2124936522614622e-06,
|
| 938 |
+
"logits/chosen": 0.08306999504566193,
|
| 939 |
+
"logits/rejected": 0.18402306735515594,
|
| 940 |
+
"logps/chosen": -614.7750244140625,
|
| 941 |
+
"logps/rejected": -746.0162353515625,
|
| 942 |
+
"loss": 0.243,
|
| 943 |
+
"rewards/accuracies": 0.71875,
|
| 944 |
+
"rewards/chosen": -0.2686184048652649,
|
| 945 |
+
"rewards/margins": 0.14612407982349396,
|
| 946 |
+
"rewards/rejected": -0.41474246978759766,
|
| 947 |
+
"step": 660
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 1.43,
|
| 951 |
+
"learning_rate": 1.1334516991487473e-06,
|
| 952 |
+
"logits/chosen": 0.10118447244167328,
|
| 953 |
+
"logits/rejected": 0.17648069560527802,
|
| 954 |
+
"logps/chosen": -568.6295166015625,
|
| 955 |
+
"logps/rejected": -708.2867431640625,
|
| 956 |
+
"loss": 0.29,
|
| 957 |
+
"rewards/accuracies": 0.6875,
|
| 958 |
+
"rewards/chosen": -0.2447134256362915,
|
| 959 |
+
"rewards/margins": 0.14246779680252075,
|
| 960 |
+
"rewards/rejected": -0.38718122243881226,
|
| 961 |
+
"step": 670
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 1.45,
|
| 965 |
+
"learning_rate": 1.0563119197063934e-06,
|
| 966 |
+
"logits/chosen": 0.09860675781965256,
|
| 967 |
+
"logits/rejected": 0.16828308999538422,
|
| 968 |
+
"logps/chosen": -644.8552856445312,
|
| 969 |
+
"logps/rejected": -772.2024536132812,
|
| 970 |
+
"loss": 0.2985,
|
| 971 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 972 |
+
"rewards/chosen": -0.2842712104320526,
|
| 973 |
+
"rewards/margins": 0.14040768146514893,
|
| 974 |
+
"rewards/rejected": -0.42467889189720154,
|
| 975 |
+
"step": 680
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 1.47,
|
| 979 |
+
"learning_rate": 9.81181689032158e-07,
|
| 980 |
+
"logits/chosen": 0.05354565382003784,
|
| 981 |
+
"logits/rejected": 0.14584000408649445,
|
| 982 |
+
"logps/chosen": -637.1915893554688,
|
| 983 |
+
"logps/rejected": -777.9390869140625,
|
| 984 |
+
"loss": 0.2788,
|
| 985 |
+
"rewards/accuracies": 0.75,
|
| 986 |
+
"rewards/chosen": -0.28641483187675476,
|
| 987 |
+
"rewards/margins": 0.14343242347240448,
|
| 988 |
+
"rewards/rejected": -0.4298473000526428,
|
| 989 |
+
"step": 690
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 1.49,
|
| 993 |
+
"learning_rate": 9.081655850224449e-07,
|
| 994 |
+
"logits/chosen": 0.05313466861844063,
|
| 995 |
+
"logits/rejected": 0.1454547941684723,
|
| 996 |
+
"logps/chosen": -599.1509399414062,
|
| 997 |
+
"logps/rejected": -701.4476318359375,
|
| 998 |
+
"loss": 0.2915,
|
| 999 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1000 |
+
"rewards/chosen": -0.2644391655921936,
|
| 1001 |
+
"rewards/margins": 0.11923892796039581,
|
| 1002 |
+
"rewards/rejected": -0.3836781084537506,
|
| 1003 |
+
"step": 700
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 1.51,
|
| 1007 |
+
"learning_rate": 8.373652428045831e-07,
|
| 1008 |
+
"logits/chosen": 0.08170448243618011,
|
| 1009 |
+
"logits/rejected": 0.17704859375953674,
|
| 1010 |
+
"logps/chosen": -674.1668701171875,
|
| 1011 |
+
"logps/rejected": -791.4132080078125,
|
| 1012 |
+
"loss": 0.28,
|
| 1013 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1014 |
+
"rewards/chosen": -0.2925952076911926,
|
| 1015 |
+
"rewards/margins": 0.15293267369270325,
|
| 1016 |
+
"rewards/rejected": -0.4455278813838959,
|
| 1017 |
+
"step": 710
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 1.54,
|
| 1021 |
+
"learning_rate": 7.688792132653111e-07,
|
| 1022 |
+
"logits/chosen": 0.13877324759960175,
|
| 1023 |
+
"logits/rejected": 0.11766058206558228,
|
| 1024 |
+
"logps/chosen": -636.9558715820312,
|
| 1025 |
+
"logps/rejected": -835.0140380859375,
|
| 1026 |
+
"loss": 0.244,
|
| 1027 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1028 |
+
"rewards/chosen": -0.26449164748191833,
|
| 1029 |
+
"rewards/margins": 0.2034236639738083,
|
| 1030 |
+
"rewards/rejected": -0.46791529655456543,
|
| 1031 |
+
"step": 720
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 1.56,
|
| 1035 |
+
"learning_rate": 7.028028258723818e-07,
|
| 1036 |
+
"logits/chosen": 0.07956353574991226,
|
| 1037 |
+
"logits/rejected": 0.15025413036346436,
|
| 1038 |
+
"logps/chosen": -635.7805786132812,
|
| 1039 |
+
"logps/rejected": -744.8255615234375,
|
| 1040 |
+
"loss": 0.2902,
|
| 1041 |
+
"rewards/accuracies": 0.65625,
|
| 1042 |
+
"rewards/chosen": -0.29231467843055725,
|
| 1043 |
+
"rewards/margins": 0.12245229631662369,
|
| 1044 |
+
"rewards/rejected": -0.41476696729660034,
|
| 1045 |
+
"step": 730
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 1.58,
|
| 1049 |
+
"learning_rate": 6.392280559802341e-07,
|
| 1050 |
+
"logits/chosen": 0.0947863757610321,
|
| 1051 |
+
"logits/rejected": 0.14226123690605164,
|
| 1052 |
+
"logps/chosen": -580.0279541015625,
|
| 1053 |
+
"logps/rejected": -725.8198852539062,
|
| 1054 |
+
"loss": 0.2555,
|
| 1055 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1056 |
+
"rewards/chosen": -0.23229190707206726,
|
| 1057 |
+
"rewards/margins": 0.16553141176700592,
|
| 1058 |
+
"rewards/rejected": -0.3978233337402344,
|
| 1059 |
+
"step": 740
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.6,
|
| 1063 |
+
"learning_rate": 5.782433968044495e-07,
|
| 1064 |
+
"logits/chosen": 0.09284155815839767,
|
| 1065 |
+
"logits/rejected": 0.19596152007579803,
|
| 1066 |
+
"logps/chosen": -648.7670288085938,
|
| 1067 |
+
"logps/rejected": -761.2144775390625,
|
| 1068 |
+
"loss": 0.2766,
|
| 1069 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1070 |
+
"rewards/chosen": -0.27112412452697754,
|
| 1071 |
+
"rewards/margins": 0.1452399045228958,
|
| 1072 |
+
"rewards/rejected": -0.4163641035556793,
|
| 1073 |
+
"step": 750
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 1.62,
|
| 1077 |
+
"learning_rate": 5.199337362431792e-07,
|
| 1078 |
+
"logits/chosen": 0.08202387392520905,
|
| 1079 |
+
"logits/rejected": 0.1395001858472824,
|
| 1080 |
+
"logps/chosen": -638.7788696289062,
|
| 1081 |
+
"logps/rejected": -739.9161376953125,
|
| 1082 |
+
"loss": 0.2912,
|
| 1083 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1084 |
+
"rewards/chosen": -0.2824631333351135,
|
| 1085 |
+
"rewards/margins": 0.1440524160861969,
|
| 1086 |
+
"rewards/rejected": -0.42651551961898804,
|
| 1087 |
+
"step": 760
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 1.64,
|
| 1091 |
+
"learning_rate": 4.643802387170118e-07,
|
| 1092 |
+
"logits/chosen": 0.031006410717964172,
|
| 1093 |
+
"logits/rejected": 0.20315364003181458,
|
| 1094 |
+
"logps/chosen": -635.1582641601562,
|
| 1095 |
+
"logps/rejected": -805.4976196289062,
|
| 1096 |
+
"loss": 0.2579,
|
| 1097 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1098 |
+
"rewards/chosen": -0.28585267066955566,
|
| 1099 |
+
"rewards/margins": 0.1737949550151825,
|
| 1100 |
+
"rewards/rejected": -0.45964759588241577,
|
| 1101 |
+
"step": 770
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 1.66,
|
| 1105 |
+
"learning_rate": 4.1166023219176176e-07,
|
| 1106 |
+
"logits/chosen": 0.11464326083660126,
|
| 1107 |
+
"logits/rejected": 0.14991115033626556,
|
| 1108 |
+
"logps/chosen": -589.2257690429688,
|
| 1109 |
+
"logps/rejected": -750.9774169921875,
|
| 1110 |
+
"loss": 0.2713,
|
| 1111 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1112 |
+
"rewards/chosen": -0.2886643409729004,
|
| 1113 |
+
"rewards/margins": 0.15789894759655,
|
| 1114 |
+
"rewards/rejected": -0.4465633034706116,
|
| 1115 |
+
"step": 780
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 1.69,
|
| 1119 |
+
"learning_rate": 3.618471005414215e-07,
|
| 1120 |
+
"logits/chosen": 0.07082124054431915,
|
| 1121 |
+
"logits/rejected": 0.11936893314123154,
|
| 1122 |
+
"logps/chosen": -596.2206420898438,
|
| 1123 |
+
"logps/rejected": -772.8016357421875,
|
| 1124 |
+
"loss": 0.2372,
|
| 1125 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1126 |
+
"rewards/chosen": -0.2700226902961731,
|
| 1127 |
+
"rewards/margins": 0.17619924247264862,
|
| 1128 |
+
"rewards/rejected": -0.4462219178676605,
|
| 1129 |
+
"step": 790
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.71,
|
| 1133 |
+
"learning_rate": 3.150101814011136e-07,
|
| 1134 |
+
"logits/chosen": 0.060673005878925323,
|
| 1135 |
+
"logits/rejected": 0.15200214087963104,
|
| 1136 |
+
"logps/chosen": -641.08056640625,
|
| 1137 |
+
"logps/rejected": -811.6611328125,
|
| 1138 |
+
"loss": 0.2697,
|
| 1139 |
+
"rewards/accuracies": 0.6875,
|
| 1140 |
+
"rewards/chosen": -0.2966248095035553,
|
| 1141 |
+
"rewards/margins": 0.16086846590042114,
|
| 1142 |
+
"rewards/rejected": -0.45749321579933167,
|
| 1143 |
+
"step": 800
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.73,
|
| 1147 |
+
"learning_rate": 2.712146696522305e-07,
|
| 1148 |
+
"logits/chosen": 0.0920618399977684,
|
| 1149 |
+
"logits/rejected": 0.11318673193454742,
|
| 1150 |
+
"logps/chosen": -665.89599609375,
|
| 1151 |
+
"logps/rejected": -780.0318603515625,
|
| 1152 |
+
"loss": 0.2911,
|
| 1153 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1154 |
+
"rewards/chosen": -0.3211720585823059,
|
| 1155 |
+
"rewards/margins": 0.14485250413417816,
|
| 1156 |
+
"rewards/rejected": -0.46602457761764526,
|
| 1157 |
+
"step": 810
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 1.75,
|
| 1161 |
+
"learning_rate": 2.3052152667409289e-07,
|
| 1162 |
+
"logits/chosen": 0.10792404413223267,
|
| 1163 |
+
"logits/rejected": 0.1413693130016327,
|
| 1164 |
+
"logps/chosen": -645.6889038085938,
|
| 1165 |
+
"logps/rejected": -751.7999877929688,
|
| 1166 |
+
"loss": 0.3049,
|
| 1167 |
+
"rewards/accuracies": 0.6875,
|
| 1168 |
+
"rewards/chosen": -0.30210328102111816,
|
| 1169 |
+
"rewards/margins": 0.12790395319461823,
|
| 1170 |
+
"rewards/rejected": -0.43000727891921997,
|
| 1171 |
+
"step": 820
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 1.77,
|
| 1175 |
+
"learning_rate": 1.9298739548845813e-07,
|
| 1176 |
+
"logits/chosen": 0.09733234345912933,
|
| 1177 |
+
"logits/rejected": 0.16931495070457458,
|
| 1178 |
+
"logps/chosen": -679.776123046875,
|
| 1179 |
+
"logps/rejected": -784.3840942382812,
|
| 1180 |
+
"loss": 0.2662,
|
| 1181 |
+
"rewards/accuracies": 0.71875,
|
| 1182 |
+
"rewards/chosen": -0.30325835943222046,
|
| 1183 |
+
"rewards/margins": 0.15936632454395294,
|
| 1184 |
+
"rewards/rejected": -0.4626246392726898,
|
| 1185 |
+
"step": 830
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.79,
|
| 1189 |
+
"learning_rate": 1.5866452191498488e-07,
|
| 1190 |
+
"logits/chosen": 0.07468713819980621,
|
| 1191 |
+
"logits/rejected": 0.15481439232826233,
|
| 1192 |
+
"logps/chosen": -627.0779418945312,
|
| 1193 |
+
"logps/rejected": -745.9903564453125,
|
| 1194 |
+
"loss": 0.2667,
|
| 1195 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1196 |
+
"rewards/chosen": -0.27285081148147583,
|
| 1197 |
+
"rewards/margins": 0.14448818564414978,
|
| 1198 |
+
"rewards/rejected": -0.41733893752098083,
|
| 1199 |
+
"step": 840
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 1.81,
|
| 1203 |
+
"learning_rate": 1.2760068184740597e-07,
|
| 1204 |
+
"logits/chosen": 0.07968030869960785,
|
| 1205 |
+
"logits/rejected": 0.14501354098320007,
|
| 1206 |
+
"logps/chosen": -628.3474731445312,
|
| 1207 |
+
"logps/rejected": -817.0506591796875,
|
| 1208 |
+
"loss": 0.2647,
|
| 1209 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1210 |
+
"rewards/chosen": -0.28495317697525024,
|
| 1211 |
+
"rewards/margins": 0.17006665468215942,
|
| 1212 |
+
"rewards/rejected": -0.4550197720527649,
|
| 1213 |
+
"step": 850
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 1.83,
|
| 1217 |
+
"learning_rate": 9.983911475163727e-08,
|
| 1218 |
+
"logits/chosen": 0.05757877230644226,
|
| 1219 |
+
"logits/rejected": 0.20654790103435516,
|
| 1220 |
+
"logps/chosen": -694.041748046875,
|
| 1221 |
+
"logps/rejected": -830.3031005859375,
|
| 1222 |
+
"loss": 0.279,
|
| 1223 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1224 |
+
"rewards/chosen": -0.3297106921672821,
|
| 1225 |
+
"rewards/margins": 0.1504661738872528,
|
| 1226 |
+
"rewards/rejected": -0.4801768362522125,
|
| 1227 |
+
"step": 860
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 1.86,
|
| 1231 |
+
"learning_rate": 7.541846347838915e-08,
|
| 1232 |
+
"logits/chosen": 0.07009818404912949,
|
| 1233 |
+
"logits/rejected": 0.08682046085596085,
|
| 1234 |
+
"logps/chosen": -622.718505859375,
|
| 1235 |
+
"logps/rejected": -736.4935302734375,
|
| 1236 |
+
"loss": 0.2882,
|
| 1237 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1238 |
+
"rewards/chosen": -0.2856903374195099,
|
| 1239 |
+
"rewards/margins": 0.15370842814445496,
|
| 1240 |
+
"rewards/rejected": -0.4393988251686096,
|
| 1241 |
+
"step": 870
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.88,
|
| 1245 |
+
"learning_rate": 5.437272047405712e-08,
|
| 1246 |
+
"logits/chosen": 0.05985601618885994,
|
| 1247 |
+
"logits/rejected": 0.19025097787380219,
|
| 1248 |
+
"logps/chosen": -640.6109619140625,
|
| 1249 |
+
"logps/rejected": -787.9774169921875,
|
| 1250 |
+
"loss": 0.2732,
|
| 1251 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1252 |
+
"rewards/chosen": -0.29887470602989197,
|
| 1253 |
+
"rewards/margins": 0.16820885241031647,
|
| 1254 |
+
"rewards/rejected": -0.467083603143692,
|
| 1255 |
+
"step": 880
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.9,
|
| 1259 |
+
"learning_rate": 3.673118046477159e-08,
|
| 1260 |
+
"logits/chosen": 0.07184259593486786,
|
| 1261 |
+
"logits/rejected": 0.19540506601333618,
|
| 1262 |
+
"logps/chosen": -646.6096801757812,
|
| 1263 |
+
"logps/rejected": -782.7240600585938,
|
| 1264 |
+
"loss": 0.278,
|
| 1265 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1266 |
+
"rewards/chosen": -0.3017248213291168,
|
| 1267 |
+
"rewards/margins": 0.1628948152065277,
|
| 1268 |
+
"rewards/rejected": -0.46461963653564453,
|
| 1269 |
+
"step": 890
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.92,
|
| 1273 |
+
"learning_rate": 2.251839967945535e-08,
|
| 1274 |
+
"logits/chosen": 0.05988436937332153,
|
| 1275 |
+
"logits/rejected": 0.11050447076559067,
|
| 1276 |
+
"logps/chosen": -601.7271728515625,
|
| 1277 |
+
"logps/rejected": -758.0606689453125,
|
| 1278 |
+
"loss": 0.2704,
|
| 1279 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1280 |
+
"rewards/chosen": -0.28261813521385193,
|
| 1281 |
+
"rewards/margins": 0.15564511716365814,
|
| 1282 |
+
"rewards/rejected": -0.4382632374763489,
|
| 1283 |
+
"step": 900
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 1.94,
|
| 1287 |
+
"learning_rate": 1.1754161668660612e-08,
|
| 1288 |
+
"logits/chosen": 0.07712358981370926,
|
| 1289 |
+
"logits/rejected": 0.17667053639888763,
|
| 1290 |
+
"logps/chosen": -672.001708984375,
|
| 1291 |
+
"logps/rejected": -807.9947509765625,
|
| 1292 |
+
"loss": 0.2561,
|
| 1293 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1294 |
+
"rewards/chosen": -0.2962462306022644,
|
| 1295 |
+
"rewards/margins": 0.16602984070777893,
|
| 1296 |
+
"rewards/rejected": -0.46227607131004333,
|
| 1297 |
+
"step": 910
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 1.96,
|
| 1301 |
+
"learning_rate": 4.453449766758933e-09,
|
| 1302 |
+
"logits/chosen": 0.1141105443239212,
|
| 1303 |
+
"logits/rejected": 0.20343729853630066,
|
| 1304 |
+
"logps/chosen": -630.8961181640625,
|
| 1305 |
+
"logps/rejected": -757.9405517578125,
|
| 1306 |
+
"loss": 0.2796,
|
| 1307 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1308 |
+
"rewards/chosen": -0.2968885600566864,
|
| 1309 |
+
"rewards/margins": 0.12835559248924255,
|
| 1310 |
+
"rewards/rejected": -0.42524418234825134,
|
| 1311 |
+
"step": 920
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.98,
|
| 1315 |
+
"learning_rate": 6.264262358129936e-10,
|
| 1316 |
+
"logits/chosen": 0.07077296823263168,
|
| 1317 |
+
"logits/rejected": 0.14096608757972717,
|
| 1318 |
+
"logps/chosen": -652.6800537109375,
|
| 1319 |
+
"logps/rejected": -770.9761352539062,
|
| 1320 |
+
"loss": 0.291,
|
| 1321 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1322 |
+
"rewards/chosen": -0.30475863814353943,
|
| 1323 |
+
"rewards/margins": 0.14434154331684113,
|
| 1324 |
+
"rewards/rejected": -0.44910019636154175,
|
| 1325 |
+
"step": 930
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 2.0,
|
| 1329 |
+
"step": 936,
|
| 1330 |
"total_flos": 0.0,
|
| 1331 |
+
"train_loss": 0.16023516272887206,
|
| 1332 |
+
"train_runtime": 6530.1152,
|
| 1333 |
+
"train_samples_per_second": 4.594,
|
| 1334 |
+
"train_steps_per_second": 0.143
|
| 1335 |
}
|
| 1336 |
],
|
| 1337 |
"logging_steps": 10,
|
| 1338 |
+
"max_steps": 936,
|
| 1339 |
"num_input_tokens_seen": 0,
|
| 1340 |
+
"num_train_epochs": 2,
|
| 1341 |
"save_steps": 100,
|
| 1342 |
"total_flos": 0.0,
|
| 1343 |
"train_batch_size": 4,
|