Model save
Browse files- README.md +84 -0
- adapter_model.safetensors +1 -1
- all_results.json +21 -0
- eval_results.json +16 -0
- train_results.json +8 -0
- trainer_state.json +1972 -0
README.md
ADDED
|
@@ -0,0 +1,84 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
base_model: mistralai/Mistral-7B-Instruct-v0.2
|
| 9 |
+
model-index:
|
| 10 |
+
- name: zephyr-7b-dpo-lora-pairrm
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# zephyr-7b-dpo-lora-pairrm
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.6747
|
| 22 |
+
- Rewards/chosen: -1.3181
|
| 23 |
+
- Rewards/rejected: -1.4367
|
| 24 |
+
- Rewards/accuracies: 0.5727
|
| 25 |
+
- Rewards/margins: 0.1186
|
| 26 |
+
- Logps/rejected: -357.3805
|
| 27 |
+
- Logps/chosen: -340.2056
|
| 28 |
+
- Logits/rejected: -4.5482
|
| 29 |
+
- Logits/chosen: -4.5594
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 5e-06
|
| 49 |
+
- train_batch_size: 4
|
| 50 |
+
- eval_batch_size: 8
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- gradient_accumulation_steps: 4
|
| 54 |
+
- total_train_batch_size: 16
|
| 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: 1
|
| 59 |
+
|
| 60 |
+
### Training results
|
| 61 |
+
|
| 62 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 63 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 64 |
+
| 0.6917 | 0.08 | 100 | 0.6924 | -0.0160 | -0.0177 | 0.5287 | 0.0016 | -215.4761 | -210.0007 | -2.5273 | -2.5303 |
|
| 65 |
+
| 0.6854 | 0.16 | 200 | 0.6875 | -0.0702 | -0.0835 | 0.5563 | 0.0133 | -222.0610 | -215.4225 | -2.5736 | -2.5764 |
|
| 66 |
+
| 0.682 | 0.24 | 300 | 0.6841 | -0.2388 | -0.2651 | 0.5450 | 0.0263 | -240.2197 | -232.2801 | -2.9180 | -2.9209 |
|
| 67 |
+
| 0.6634 | 0.32 | 400 | 0.6812 | -0.4832 | -0.5288 | 0.5487 | 0.0455 | -266.5857 | -256.7237 | -3.4549 | -3.4603 |
|
| 68 |
+
| 0.6296 | 0.4 | 500 | 0.6782 | -0.6896 | -0.7564 | 0.5600 | 0.0668 | -289.3543 | -277.3629 | -4.1668 | -4.1749 |
|
| 69 |
+
| 0.6503 | 0.48 | 600 | 0.6770 | -0.9588 | -1.0440 | 0.5533 | 0.0852 | -318.1134 | -304.2834 | -4.4345 | -4.4433 |
|
| 70 |
+
| 0.5974 | 0.56 | 700 | 0.6778 | -1.1455 | -1.2432 | 0.5653 | 0.0977 | -338.0312 | -322.9485 | -4.4370 | -4.4480 |
|
| 71 |
+
| 0.6508 | 0.64 | 800 | 0.6748 | -1.1002 | -1.2023 | 0.5650 | 0.1022 | -333.9435 | -318.4168 | -4.2618 | -4.2711 |
|
| 72 |
+
| 0.6746 | 0.72 | 900 | 0.6757 | -1.3289 | -1.4445 | 0.5687 | 0.1155 | -358.1558 | -341.2940 | -4.5662 | -4.5772 |
|
| 73 |
+
| 0.6151 | 0.8 | 1000 | 0.6755 | -1.3559 | -1.4746 | 0.5690 | 0.1187 | -361.1742 | -343.9893 | -4.6070 | -4.6184 |
|
| 74 |
+
| 0.6837 | 0.88 | 1100 | 0.6748 | -1.3246 | -1.4437 | 0.5710 | 0.1192 | -358.0839 | -340.8576 | -4.5607 | -4.5717 |
|
| 75 |
+
| 0.6539 | 0.96 | 1200 | 0.6746 | -1.3182 | -1.4369 | 0.5730 | 0.1187 | -357.4036 | -340.2231 | -4.5483 | -4.5595 |
|
| 76 |
+
|
| 77 |
+
|
| 78 |
+
### Framework versions
|
| 79 |
+
|
| 80 |
+
- PEFT 0.7.1
|
| 81 |
+
- Transformers 4.36.2
|
| 82 |
+
- Pytorch 2.1.2
|
| 83 |
+
- Datasets 2.14.6
|
| 84 |
+
- Tokenizers 0.15.0
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 671149616
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e69e737e1c3cc31220e7a34f1ac3283e4715c26d80bbca276c7e216632134c7a
|
| 3 |
size 671149616
|
all_results.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": -4.5593791007995605,
|
| 4 |
+
"eval_logits/rejected": -4.548151016235352,
|
| 5 |
+
"eval_logps/chosen": -340.20562744140625,
|
| 6 |
+
"eval_logps/rejected": -357.3804626464844,
|
| 7 |
+
"eval_loss": 0.6746748089790344,
|
| 8 |
+
"eval_rewards/accuracies": 0.5726666450500488,
|
| 9 |
+
"eval_rewards/chosen": -1.3180568218231201,
|
| 10 |
+
"eval_rewards/margins": 0.11864501982927322,
|
| 11 |
+
"eval_rewards/rejected": -1.436701774597168,
|
| 12 |
+
"eval_runtime": 1558.97,
|
| 13 |
+
"eval_samples": 2994,
|
| 14 |
+
"eval_samples_per_second": 1.92,
|
| 15 |
+
"eval_steps_per_second": 0.241,
|
| 16 |
+
"train_loss": 0.6508825262419409,
|
| 17 |
+
"train_runtime": 40390.1729,
|
| 18 |
+
"train_samples": 19996,
|
| 19 |
+
"train_samples_per_second": 0.495,
|
| 20 |
+
"train_steps_per_second": 0.031
|
| 21 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": -4.5593791007995605,
|
| 4 |
+
"eval_logits/rejected": -4.548151016235352,
|
| 5 |
+
"eval_logps/chosen": -340.20562744140625,
|
| 6 |
+
"eval_logps/rejected": -357.3804626464844,
|
| 7 |
+
"eval_loss": 0.6746748089790344,
|
| 8 |
+
"eval_rewards/accuracies": 0.5726666450500488,
|
| 9 |
+
"eval_rewards/chosen": -1.3180568218231201,
|
| 10 |
+
"eval_rewards/margins": 0.11864501982927322,
|
| 11 |
+
"eval_rewards/rejected": -1.436701774597168,
|
| 12 |
+
"eval_runtime": 1558.97,
|
| 13 |
+
"eval_samples": 2994,
|
| 14 |
+
"eval_samples_per_second": 1.92,
|
| 15 |
+
"eval_steps_per_second": 0.241
|
| 16 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.6508825262419409,
|
| 4 |
+
"train_runtime": 40390.1729,
|
| 5 |
+
"train_samples": 19996,
|
| 6 |
+
"train_samples_per_second": 0.495,
|
| 7 |
+
"train_steps_per_second": 0.031
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1972 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9993998799759952,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 1249,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 4e-08,
|
| 14 |
+
"logits/chosen": -2.6092891693115234,
|
| 15 |
+
"logits/rejected": -2.629662275314331,
|
| 16 |
+
"logps/chosen": -260.39178466796875,
|
| 17 |
+
"logps/rejected": -256.2431945800781,
|
| 18 |
+
"loss": 0.6931,
|
| 19 |
+
"rewards/accuracies": 0.0,
|
| 20 |
+
"rewards/chosen": 0.0,
|
| 21 |
+
"rewards/margins": 0.0,
|
| 22 |
+
"rewards/rejected": 0.0,
|
| 23 |
+
"step": 1
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.01,
|
| 27 |
+
"learning_rate": 4.0000000000000003e-07,
|
| 28 |
+
"logits/chosen": -2.542802095413208,
|
| 29 |
+
"logits/rejected": -2.5701138973236084,
|
| 30 |
+
"logps/chosen": -287.18267822265625,
|
| 31 |
+
"logps/rejected": -308.441162109375,
|
| 32 |
+
"loss": 0.6934,
|
| 33 |
+
"rewards/accuracies": 0.4166666567325592,
|
| 34 |
+
"rewards/chosen": 0.0005982329603284597,
|
| 35 |
+
"rewards/margins": -0.0005221270839683712,
|
| 36 |
+
"rewards/rejected": 0.0011203601025044918,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.02,
|
| 41 |
+
"learning_rate": 8.000000000000001e-07,
|
| 42 |
+
"logits/chosen": -2.520120859146118,
|
| 43 |
+
"logits/rejected": -2.5234670639038086,
|
| 44 |
+
"logps/chosen": -255.07748413085938,
|
| 45 |
+
"logps/rejected": -273.0997009277344,
|
| 46 |
+
"loss": 0.693,
|
| 47 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 48 |
+
"rewards/chosen": -0.00023935179342515767,
|
| 49 |
+
"rewards/margins": 0.00027643097564578056,
|
| 50 |
+
"rewards/rejected": -0.0005157828563824296,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.02,
|
| 55 |
+
"learning_rate": 1.2000000000000002e-06,
|
| 56 |
+
"logits/chosen": -2.5063071250915527,
|
| 57 |
+
"logits/rejected": -2.5213770866394043,
|
| 58 |
+
"logps/chosen": -276.6528625488281,
|
| 59 |
+
"logps/rejected": -294.9043884277344,
|
| 60 |
+
"loss": 0.693,
|
| 61 |
+
"rewards/accuracies": 0.5625,
|
| 62 |
+
"rewards/chosen": -0.0008763422374613583,
|
| 63 |
+
"rewards/margins": 0.0003111693658865988,
|
| 64 |
+
"rewards/rejected": -0.001187511719763279,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.03,
|
| 69 |
+
"learning_rate": 1.6000000000000001e-06,
|
| 70 |
+
"logits/chosen": -2.542746067047119,
|
| 71 |
+
"logits/rejected": -2.549612522125244,
|
| 72 |
+
"logps/chosen": -251.32003784179688,
|
| 73 |
+
"logps/rejected": -259.71685791015625,
|
| 74 |
+
"loss": 0.6928,
|
| 75 |
+
"rewards/accuracies": 0.581250011920929,
|
| 76 |
+
"rewards/chosen": -0.004003011621534824,
|
| 77 |
+
"rewards/margins": 0.0007247755420394242,
|
| 78 |
+
"rewards/rejected": -0.004727787338197231,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.04,
|
| 83 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 84 |
+
"logits/chosen": -2.4885387420654297,
|
| 85 |
+
"logits/rejected": -2.4705405235290527,
|
| 86 |
+
"logps/chosen": -251.05563354492188,
|
| 87 |
+
"logps/rejected": -255.7452392578125,
|
| 88 |
+
"loss": 0.6933,
|
| 89 |
+
"rewards/accuracies": 0.512499988079071,
|
| 90 |
+
"rewards/chosen": -0.006660168059170246,
|
| 91 |
+
"rewards/margins": -0.00038892580778338015,
|
| 92 |
+
"rewards/rejected": -0.006271242164075375,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.05,
|
| 97 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 98 |
+
"logits/chosen": -2.5540900230407715,
|
| 99 |
+
"logits/rejected": -2.5607895851135254,
|
| 100 |
+
"logps/chosen": -262.1423034667969,
|
| 101 |
+
"logps/rejected": -280.08209228515625,
|
| 102 |
+
"loss": 0.6928,
|
| 103 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 104 |
+
"rewards/chosen": -0.009707225486636162,
|
| 105 |
+
"rewards/margins": 0.000684750615619123,
|
| 106 |
+
"rewards/rejected": -0.010391974821686745,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.06,
|
| 111 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 112 |
+
"logits/chosen": -2.5549094676971436,
|
| 113 |
+
"logits/rejected": -2.5427558422088623,
|
| 114 |
+
"logps/chosen": -266.07891845703125,
|
| 115 |
+
"logps/rejected": -271.27740478515625,
|
| 116 |
+
"loss": 0.6926,
|
| 117 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 118 |
+
"rewards/chosen": -0.014160098508000374,
|
| 119 |
+
"rewards/margins": 0.0011990683851763606,
|
| 120 |
+
"rewards/rejected": -0.015359166078269482,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.06,
|
| 125 |
+
"learning_rate": 3.2000000000000003e-06,
|
| 126 |
+
"logits/chosen": -2.5750041007995605,
|
| 127 |
+
"logits/rejected": -2.5791172981262207,
|
| 128 |
+
"logps/chosen": -262.36407470703125,
|
| 129 |
+
"logps/rejected": -282.44720458984375,
|
| 130 |
+
"loss": 0.6928,
|
| 131 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 132 |
+
"rewards/chosen": -0.015319536440074444,
|
| 133 |
+
"rewards/margins": 0.0008405897533521056,
|
| 134 |
+
"rewards/rejected": -0.01616012677550316,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.07,
|
| 139 |
+
"learning_rate": 3.6000000000000003e-06,
|
| 140 |
+
"logits/chosen": -2.5781426429748535,
|
| 141 |
+
"logits/rejected": -2.5637879371643066,
|
| 142 |
+
"logps/chosen": -295.14898681640625,
|
| 143 |
+
"logps/rejected": -300.16131591796875,
|
| 144 |
+
"loss": 0.6906,
|
| 145 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 146 |
+
"rewards/chosen": -0.020102154463529587,
|
| 147 |
+
"rewards/margins": 0.005331115797162056,
|
| 148 |
+
"rewards/rejected": -0.025433272123336792,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.08,
|
| 153 |
+
"learning_rate": 4.000000000000001e-06,
|
| 154 |
+
"logits/chosen": -2.5696635246276855,
|
| 155 |
+
"logits/rejected": -2.569899797439575,
|
| 156 |
+
"logps/chosen": -263.42047119140625,
|
| 157 |
+
"logps/rejected": -276.1015319824219,
|
| 158 |
+
"loss": 0.6917,
|
| 159 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 160 |
+
"rewards/chosen": -0.024685552343726158,
|
| 161 |
+
"rewards/margins": 0.0029262492898851633,
|
| 162 |
+
"rewards/rejected": -0.027611803263425827,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.08,
|
| 167 |
+
"eval_logits/chosen": -2.530292272567749,
|
| 168 |
+
"eval_logits/rejected": -2.5272903442382812,
|
| 169 |
+
"eval_logps/chosen": -210.00074768066406,
|
| 170 |
+
"eval_logps/rejected": -215.47608947753906,
|
| 171 |
+
"eval_loss": 0.6923908591270447,
|
| 172 |
+
"eval_rewards/accuracies": 0.5286666750907898,
|
| 173 |
+
"eval_rewards/chosen": -0.016008317470550537,
|
| 174 |
+
"eval_rewards/margins": 0.001649469486437738,
|
| 175 |
+
"eval_rewards/rejected": -0.01765778660774231,
|
| 176 |
+
"eval_runtime": 1564.6229,
|
| 177 |
+
"eval_samples_per_second": 1.914,
|
| 178 |
+
"eval_steps_per_second": 0.24,
|
| 179 |
+
"step": 100
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.09,
|
| 183 |
+
"learning_rate": 4.4e-06,
|
| 184 |
+
"logits/chosen": -2.573535680770874,
|
| 185 |
+
"logits/rejected": -2.5764663219451904,
|
| 186 |
+
"logps/chosen": -283.31744384765625,
|
| 187 |
+
"logps/rejected": -293.5372314453125,
|
| 188 |
+
"loss": 0.6918,
|
| 189 |
+
"rewards/accuracies": 0.518750011920929,
|
| 190 |
+
"rewards/chosen": -0.027460157871246338,
|
| 191 |
+
"rewards/margins": 0.0029141721315681934,
|
| 192 |
+
"rewards/rejected": -0.030374327674508095,
|
| 193 |
+
"step": 110
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.1,
|
| 197 |
+
"learning_rate": 4.800000000000001e-06,
|
| 198 |
+
"logits/chosen": -2.593564033508301,
|
| 199 |
+
"logits/rejected": -2.6072158813476562,
|
| 200 |
+
"logps/chosen": -265.5439758300781,
|
| 201 |
+
"logps/rejected": -282.11444091796875,
|
| 202 |
+
"loss": 0.6911,
|
| 203 |
+
"rewards/accuracies": 0.518750011920929,
|
| 204 |
+
"rewards/chosen": -0.03183265030384064,
|
| 205 |
+
"rewards/margins": 0.004463006276637316,
|
| 206 |
+
"rewards/rejected": -0.036295659840106964,
|
| 207 |
+
"step": 120
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.1,
|
| 211 |
+
"learning_rate": 4.999755876225375e-06,
|
| 212 |
+
"logits/chosen": -2.559654712677002,
|
| 213 |
+
"logits/rejected": -2.543823480606079,
|
| 214 |
+
"logps/chosen": -279.4471435546875,
|
| 215 |
+
"logps/rejected": -297.8502502441406,
|
| 216 |
+
"loss": 0.6892,
|
| 217 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 218 |
+
"rewards/chosen": -0.0379730649292469,
|
| 219 |
+
"rewards/margins": 0.008286955766379833,
|
| 220 |
+
"rewards/rejected": -0.04626002162694931,
|
| 221 |
+
"step": 130
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.11,
|
| 225 |
+
"learning_rate": 4.997803172081864e-06,
|
| 226 |
+
"logits/chosen": -2.5811429023742676,
|
| 227 |
+
"logits/rejected": -2.581789255142212,
|
| 228 |
+
"logps/chosen": -274.745849609375,
|
| 229 |
+
"logps/rejected": -287.67303466796875,
|
| 230 |
+
"loss": 0.6865,
|
| 231 |
+
"rewards/accuracies": 0.59375,
|
| 232 |
+
"rewards/chosen": -0.04451751708984375,
|
| 233 |
+
"rewards/margins": 0.014026105403900146,
|
| 234 |
+
"rewards/rejected": -0.058543622493743896,
|
| 235 |
+
"step": 140
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.12,
|
| 239 |
+
"learning_rate": 4.9938992891651825e-06,
|
| 240 |
+
"logits/chosen": -2.545161724090576,
|
| 241 |
+
"logits/rejected": -2.5361251831054688,
|
| 242 |
+
"logps/chosen": -262.94903564453125,
|
| 243 |
+
"logps/rejected": -284.0696716308594,
|
| 244 |
+
"loss": 0.6833,
|
| 245 |
+
"rewards/accuracies": 0.606249988079071,
|
| 246 |
+
"rewards/chosen": -0.0502193346619606,
|
| 247 |
+
"rewards/margins": 0.020780760794878006,
|
| 248 |
+
"rewards/rejected": -0.07100008428096771,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.13,
|
| 253 |
+
"learning_rate": 4.988047277024456e-06,
|
| 254 |
+
"logits/chosen": -2.5855746269226074,
|
| 255 |
+
"logits/rejected": -2.5990371704101562,
|
| 256 |
+
"logps/chosen": -274.5702819824219,
|
| 257 |
+
"logps/rejected": -288.3578796386719,
|
| 258 |
+
"loss": 0.6819,
|
| 259 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 260 |
+
"rewards/chosen": -0.07741515338420868,
|
| 261 |
+
"rewards/margins": 0.023703744634985924,
|
| 262 |
+
"rewards/rejected": -0.10111888498067856,
|
| 263 |
+
"step": 160
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.14,
|
| 267 |
+
"learning_rate": 4.980251707005417e-06,
|
| 268 |
+
"logits/chosen": -2.5723748207092285,
|
| 269 |
+
"logits/rejected": -2.547863006591797,
|
| 270 |
+
"logps/chosen": -291.48736572265625,
|
| 271 |
+
"logps/rejected": -298.7716979980469,
|
| 272 |
+
"loss": 0.6899,
|
| 273 |
+
"rewards/accuracies": 0.543749988079071,
|
| 274 |
+
"rewards/chosen": -0.10616284608840942,
|
| 275 |
+
"rewards/margins": 0.007837113924324512,
|
| 276 |
+
"rewards/rejected": -0.11399996280670166,
|
| 277 |
+
"step": 170
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.14,
|
| 281 |
+
"learning_rate": 4.970518668679459e-06,
|
| 282 |
+
"logits/chosen": -2.5809338092803955,
|
| 283 |
+
"logits/rejected": -2.566509246826172,
|
| 284 |
+
"logps/chosen": -289.52642822265625,
|
| 285 |
+
"logps/rejected": -295.8729248046875,
|
| 286 |
+
"loss": 0.6849,
|
| 287 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 288 |
+
"rewards/chosen": -0.09603477269411087,
|
| 289 |
+
"rewards/margins": 0.01864427700638771,
|
| 290 |
+
"rewards/rejected": -0.11467906087636948,
|
| 291 |
+
"step": 180
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.15,
|
| 295 |
+
"learning_rate": 4.958855765086722e-06,
|
| 296 |
+
"logits/chosen": -2.608503818511963,
|
| 297 |
+
"logits/rejected": -2.6062302589416504,
|
| 298 |
+
"logps/chosen": -269.84283447265625,
|
| 299 |
+
"logps/rejected": -280.5111083984375,
|
| 300 |
+
"loss": 0.6851,
|
| 301 |
+
"rewards/accuracies": 0.625,
|
| 302 |
+
"rewards/chosen": -0.05638309568166733,
|
| 303 |
+
"rewards/margins": 0.017940795049071312,
|
| 304 |
+
"rewards/rejected": -0.07432388514280319,
|
| 305 |
+
"step": 190
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.16,
|
| 309 |
+
"learning_rate": 4.945272106796919e-06,
|
| 310 |
+
"logits/chosen": -2.605029344558716,
|
| 311 |
+
"logits/rejected": -2.6143765449523926,
|
| 312 |
+
"logps/chosen": -270.00830078125,
|
| 313 |
+
"logps/rejected": -285.1636657714844,
|
| 314 |
+
"loss": 0.6854,
|
| 315 |
+
"rewards/accuracies": 0.59375,
|
| 316 |
+
"rewards/chosen": -0.06747416406869888,
|
| 317 |
+
"rewards/margins": 0.018329909071326256,
|
| 318 |
+
"rewards/rejected": -0.08580406755208969,
|
| 319 |
+
"step": 200
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.16,
|
| 323 |
+
"eval_logits/chosen": -2.57635235786438,
|
| 324 |
+
"eval_logits/rejected": -2.573589563369751,
|
| 325 |
+
"eval_logps/chosen": -215.42254638671875,
|
| 326 |
+
"eval_logps/rejected": -222.06101989746094,
|
| 327 |
+
"eval_loss": 0.6875421404838562,
|
| 328 |
+
"eval_rewards/accuracies": 0.5563333630561829,
|
| 329 |
+
"eval_rewards/chosen": -0.07022587954998016,
|
| 330 |
+
"eval_rewards/margins": 0.013281372375786304,
|
| 331 |
+
"eval_rewards/rejected": -0.08350725471973419,
|
| 332 |
+
"eval_runtime": 1564.649,
|
| 333 |
+
"eval_samples_per_second": 1.914,
|
| 334 |
+
"eval_steps_per_second": 0.24,
|
| 335 |
+
"step": 200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.17,
|
| 339 |
+
"learning_rate": 4.929778304792537e-06,
|
| 340 |
+
"logits/chosen": -2.594109058380127,
|
| 341 |
+
"logits/rejected": -2.5997750759124756,
|
| 342 |
+
"logps/chosen": -294.9931640625,
|
| 343 |
+
"logps/rejected": -299.47515869140625,
|
| 344 |
+
"loss": 0.6783,
|
| 345 |
+
"rewards/accuracies": 0.574999988079071,
|
| 346 |
+
"rewards/chosen": -0.08973054587841034,
|
| 347 |
+
"rewards/margins": 0.03516639396548271,
|
| 348 |
+
"rewards/rejected": -0.12489694356918335,
|
| 349 |
+
"step": 210
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.18,
|
| 353 |
+
"learning_rate": 4.912386462179987e-06,
|
| 354 |
+
"logits/chosen": -2.6223044395446777,
|
| 355 |
+
"logits/rejected": -2.6194934844970703,
|
| 356 |
+
"logps/chosen": -282.9168701171875,
|
| 357 |
+
"logps/rejected": -308.5869140625,
|
| 358 |
+
"loss": 0.6757,
|
| 359 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 360 |
+
"rewards/chosen": -0.08990589529275894,
|
| 361 |
+
"rewards/margins": 0.039798181504011154,
|
| 362 |
+
"rewards/rejected": -0.1297040730714798,
|
| 363 |
+
"step": 220
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.18,
|
| 367 |
+
"learning_rate": 4.893110164735167e-06,
|
| 368 |
+
"logits/chosen": -2.686192512512207,
|
| 369 |
+
"logits/rejected": -2.6861042976379395,
|
| 370 |
+
"logps/chosen": -288.02984619140625,
|
| 371 |
+
"logps/rejected": -301.3773193359375,
|
| 372 |
+
"loss": 0.6819,
|
| 373 |
+
"rewards/accuracies": 0.59375,
|
| 374 |
+
"rewards/chosen": -0.11503495275974274,
|
| 375 |
+
"rewards/margins": 0.02641122415661812,
|
| 376 |
+
"rewards/rejected": -0.14144618809223175,
|
| 377 |
+
"step": 230
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.19,
|
| 381 |
+
"learning_rate": 4.871964470290823e-06,
|
| 382 |
+
"logits/chosen": -2.660379409790039,
|
| 383 |
+
"logits/rejected": -2.6843981742858887,
|
| 384 |
+
"logps/chosen": -289.6697998046875,
|
| 385 |
+
"logps/rejected": -310.08245849609375,
|
| 386 |
+
"loss": 0.6712,
|
| 387 |
+
"rewards/accuracies": 0.675000011920929,
|
| 388 |
+
"rewards/chosen": -0.12107042968273163,
|
| 389 |
+
"rewards/margins": 0.05008827522397041,
|
| 390 |
+
"rewards/rejected": -0.17115873098373413,
|
| 391 |
+
"step": 240
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.2,
|
| 395 |
+
"learning_rate": 4.848965896974006e-06,
|
| 396 |
+
"logits/chosen": -2.6900529861450195,
|
| 397 |
+
"logits/rejected": -2.691603183746338,
|
| 398 |
+
"logps/chosen": -281.68792724609375,
|
| 399 |
+
"logps/rejected": -301.64129638671875,
|
| 400 |
+
"loss": 0.6784,
|
| 401 |
+
"rewards/accuracies": 0.606249988079071,
|
| 402 |
+
"rewards/chosen": -0.1229969710111618,
|
| 403 |
+
"rewards/margins": 0.03473496437072754,
|
| 404 |
+
"rewards/rejected": -0.15773193538188934,
|
| 405 |
+
"step": 250
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.21,
|
| 409 |
+
"learning_rate": 4.8241324103028055e-06,
|
| 410 |
+
"logits/chosen": -2.794003963470459,
|
| 411 |
+
"logits/rejected": -2.756747007369995,
|
| 412 |
+
"logps/chosen": -288.9064025878906,
|
| 413 |
+
"logps/rejected": -300.8970947265625,
|
| 414 |
+
"loss": 0.6737,
|
| 415 |
+
"rewards/accuracies": 0.612500011920929,
|
| 416 |
+
"rewards/chosen": -0.1542636603116989,
|
| 417 |
+
"rewards/margins": 0.044557444751262665,
|
| 418 |
+
"rewards/rejected": -0.19882109761238098,
|
| 419 |
+
"step": 260
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.22,
|
| 423 |
+
"learning_rate": 4.797483409152438e-06,
|
| 424 |
+
"logits/chosen": -2.8427796363830566,
|
| 425 |
+
"logits/rejected": -2.8422060012817383,
|
| 426 |
+
"logps/chosen": -280.02099609375,
|
| 427 |
+
"logps/rejected": -300.798583984375,
|
| 428 |
+
"loss": 0.667,
|
| 429 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 430 |
+
"rewards/chosen": -0.17782548069953918,
|
| 431 |
+
"rewards/margins": 0.0627431869506836,
|
| 432 |
+
"rewards/rejected": -0.24056866765022278,
|
| 433 |
+
"step": 270
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.22,
|
| 437 |
+
"learning_rate": 4.769039710601669e-06,
|
| 438 |
+
"logits/chosen": -2.926100015640259,
|
| 439 |
+
"logits/rejected": -2.9281368255615234,
|
| 440 |
+
"logps/chosen": -280.8794250488281,
|
| 441 |
+
"logps/rejected": -298.9492492675781,
|
| 442 |
+
"loss": 0.6732,
|
| 443 |
+
"rewards/accuracies": 0.612500011920929,
|
| 444 |
+
"rewards/chosen": -0.2086031138896942,
|
| 445 |
+
"rewards/margins": 0.047397904098033905,
|
| 446 |
+
"rewards/rejected": -0.25600099563598633,
|
| 447 |
+
"step": 280
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.23,
|
| 451 |
+
"learning_rate": 4.738823533671383e-06,
|
| 452 |
+
"logits/chosen": -2.972533702850342,
|
| 453 |
+
"logits/rejected": -2.9721083641052246,
|
| 454 |
+
"logps/chosen": -306.55865478515625,
|
| 455 |
+
"logps/rejected": -322.1519470214844,
|
| 456 |
+
"loss": 0.6769,
|
| 457 |
+
"rewards/accuracies": 0.606249988079071,
|
| 458 |
+
"rewards/chosen": -0.24026934802532196,
|
| 459 |
+
"rewards/margins": 0.041927777230739594,
|
| 460 |
+
"rewards/rejected": -0.28219714760780334,
|
| 461 |
+
"step": 290
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.24,
|
| 465 |
+
"learning_rate": 4.706858481968017e-06,
|
| 466 |
+
"logits/chosen": -2.922761917114258,
|
| 467 |
+
"logits/rejected": -2.9264841079711914,
|
| 468 |
+
"logps/chosen": -295.0879821777344,
|
| 469 |
+
"logps/rejected": -304.8949279785156,
|
| 470 |
+
"loss": 0.682,
|
| 471 |
+
"rewards/accuracies": 0.5625,
|
| 472 |
+
"rewards/chosen": -0.2886089086532593,
|
| 473 |
+
"rewards/margins": 0.03215286135673523,
|
| 474 |
+
"rewards/rejected": -0.3207617700099945,
|
| 475 |
+
"step": 300
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.24,
|
| 479 |
+
"eval_logits/chosen": -2.920912981033325,
|
| 480 |
+
"eval_logits/rejected": -2.9179892539978027,
|
| 481 |
+
"eval_logps/chosen": -232.2801055908203,
|
| 482 |
+
"eval_logps/rejected": -240.21971130371094,
|
| 483 |
+
"eval_loss": 0.684112012386322,
|
| 484 |
+
"eval_rewards/accuracies": 0.5450000166893005,
|
| 485 |
+
"eval_rewards/chosen": -0.23880183696746826,
|
| 486 |
+
"eval_rewards/margins": 0.026292217895388603,
|
| 487 |
+
"eval_rewards/rejected": -0.2650940418243408,
|
| 488 |
+
"eval_runtime": 1563.5865,
|
| 489 |
+
"eval_samples_per_second": 1.915,
|
| 490 |
+
"eval_steps_per_second": 0.24,
|
| 491 |
+
"step": 300
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.25,
|
| 495 |
+
"learning_rate": 4.673169525245416e-06,
|
| 496 |
+
"logits/chosen": -2.912137269973755,
|
| 497 |
+
"logits/rejected": -2.890232563018799,
|
| 498 |
+
"logps/chosen": -291.8182678222656,
|
| 499 |
+
"logps/rejected": -320.7854309082031,
|
| 500 |
+
"loss": 0.6652,
|
| 501 |
+
"rewards/accuracies": 0.643750011920929,
|
| 502 |
+
"rewards/chosen": -0.257565438747406,
|
| 503 |
+
"rewards/margins": 0.06874646991491318,
|
| 504 |
+
"rewards/rejected": -0.3263119161128998,
|
| 505 |
+
"step": 310
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.26,
|
| 509 |
+
"learning_rate": 4.63778297989952e-06,
|
| 510 |
+
"logits/chosen": -3.007861375808716,
|
| 511 |
+
"logits/rejected": -2.9914684295654297,
|
| 512 |
+
"logps/chosen": -297.81695556640625,
|
| 513 |
+
"logps/rejected": -314.2324523925781,
|
| 514 |
+
"loss": 0.6678,
|
| 515 |
+
"rewards/accuracies": 0.643750011920929,
|
| 516 |
+
"rewards/chosen": -0.28555119037628174,
|
| 517 |
+
"rewards/margins": 0.06271664053201675,
|
| 518 |
+
"rewards/rejected": -0.3482677936553955,
|
| 519 |
+
"step": 320
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.26,
|
| 523 |
+
"learning_rate": 4.60072648841109e-06,
|
| 524 |
+
"logits/chosen": -3.0622572898864746,
|
| 525 |
+
"logits/rejected": -3.0570547580718994,
|
| 526 |
+
"logps/chosen": -311.02105712890625,
|
| 527 |
+
"logps/rejected": -333.46630859375,
|
| 528 |
+
"loss": 0.6444,
|
| 529 |
+
"rewards/accuracies": 0.6875,
|
| 530 |
+
"rewards/chosen": -0.3358006477355957,
|
| 531 |
+
"rewards/margins": 0.11871840804815292,
|
| 532 |
+
"rewards/rejected": -0.454519122838974,
|
| 533 |
+
"step": 330
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.27,
|
| 537 |
+
"learning_rate": 4.562028997752574e-06,
|
| 538 |
+
"logits/chosen": -3.153225898742676,
|
| 539 |
+
"logits/rejected": -3.1641342639923096,
|
| 540 |
+
"logps/chosen": -311.9708557128906,
|
| 541 |
+
"logps/rejected": -338.3099670410156,
|
| 542 |
+
"loss": 0.6682,
|
| 543 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 544 |
+
"rewards/chosen": -0.3542831838130951,
|
| 545 |
+
"rewards/margins": 0.068769671022892,
|
| 546 |
+
"rewards/rejected": -0.4230528473854065,
|
| 547 |
+
"step": 340
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 0.28,
|
| 551 |
+
"learning_rate": 4.521720736775947e-06,
|
| 552 |
+
"logits/chosen": -3.2916831970214844,
|
| 553 |
+
"logits/rejected": -3.3004555702209473,
|
| 554 |
+
"logps/chosen": -328.3070373535156,
|
| 555 |
+
"logps/rejected": -338.16192626953125,
|
| 556 |
+
"loss": 0.6564,
|
| 557 |
+
"rewards/accuracies": 0.637499988079071,
|
| 558 |
+
"rewards/chosen": -0.4777124524116516,
|
| 559 |
+
"rewards/margins": 0.09214212745428085,
|
| 560 |
+
"rewards/rejected": -0.5698545575141907,
|
| 561 |
+
"step": 350
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.29,
|
| 565 |
+
"learning_rate": 4.479833192599198e-06,
|
| 566 |
+
"logits/chosen": -3.4253318309783936,
|
| 567 |
+
"logits/rejected": -3.4066455364227295,
|
| 568 |
+
"logps/chosen": -329.5577697753906,
|
| 569 |
+
"logps/rejected": -350.79742431640625,
|
| 570 |
+
"loss": 0.6555,
|
| 571 |
+
"rewards/accuracies": 0.612500011920929,
|
| 572 |
+
"rewards/chosen": -0.46095147728919983,
|
| 573 |
+
"rewards/margins": 0.09699266403913498,
|
| 574 |
+
"rewards/rejected": -0.5579441785812378,
|
| 575 |
+
"step": 360
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.3,
|
| 579 |
+
"learning_rate": 4.436399086009928e-06,
|
| 580 |
+
"logits/chosen": -3.4043822288513184,
|
| 581 |
+
"logits/rejected": -3.3838889598846436,
|
| 582 |
+
"logps/chosen": -317.8493957519531,
|
| 583 |
+
"logps/rejected": -334.6202087402344,
|
| 584 |
+
"loss": 0.6576,
|
| 585 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 586 |
+
"rewards/chosen": -0.4469628930091858,
|
| 587 |
+
"rewards/margins": 0.08755813539028168,
|
| 588 |
+
"rewards/rejected": -0.5345209836959839,
|
| 589 |
+
"step": 370
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 0.3,
|
| 593 |
+
"learning_rate": 4.391452345905239e-06,
|
| 594 |
+
"logits/chosen": -3.347275972366333,
|
| 595 |
+
"logits/rejected": -3.35723614692688,
|
| 596 |
+
"logps/chosen": -325.1240234375,
|
| 597 |
+
"logps/rejected": -339.3655090332031,
|
| 598 |
+
"loss": 0.6618,
|
| 599 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 600 |
+
"rewards/chosen": -0.4934774935245514,
|
| 601 |
+
"rewards/margins": 0.09260674566030502,
|
| 602 |
+
"rewards/rejected": -0.586084246635437,
|
| 603 |
+
"step": 380
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.31,
|
| 607 |
+
"learning_rate": 4.3450280827879125e-06,
|
| 608 |
+
"logits/chosen": -3.4295692443847656,
|
| 609 |
+
"logits/rejected": -3.4535210132598877,
|
| 610 |
+
"logps/chosen": -323.19415283203125,
|
| 611 |
+
"logps/rejected": -340.07989501953125,
|
| 612 |
+
"loss": 0.6639,
|
| 613 |
+
"rewards/accuracies": 0.612500011920929,
|
| 614 |
+
"rewards/chosen": -0.4877980351448059,
|
| 615 |
+
"rewards/margins": 0.08489425480365753,
|
| 616 |
+
"rewards/rejected": -0.5726922154426575,
|
| 617 |
+
"step": 390
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.32,
|
| 621 |
+
"learning_rate": 4.297162561339554e-06,
|
| 622 |
+
"logits/chosen": -3.366345167160034,
|
| 623 |
+
"logits/rejected": -3.3470451831817627,
|
| 624 |
+
"logps/chosen": -330.50311279296875,
|
| 625 |
+
"logps/rejected": -352.4704895019531,
|
| 626 |
+
"loss": 0.6634,
|
| 627 |
+
"rewards/accuracies": 0.612500011920929,
|
| 628 |
+
"rewards/chosen": -0.5246509313583374,
|
| 629 |
+
"rewards/margins": 0.09757552295923233,
|
| 630 |
+
"rewards/rejected": -0.6222264766693115,
|
| 631 |
+
"step": 400
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 0.32,
|
| 635 |
+
"eval_logits/chosen": -3.460310935974121,
|
| 636 |
+
"eval_logits/rejected": -3.4548885822296143,
|
| 637 |
+
"eval_logps/chosen": -256.72369384765625,
|
| 638 |
+
"eval_logps/rejected": -266.585693359375,
|
| 639 |
+
"eval_loss": 0.681212306022644,
|
| 640 |
+
"eval_rewards/accuracies": 0.5486666560173035,
|
| 641 |
+
"eval_rewards/chosen": -0.4832378029823303,
|
| 642 |
+
"eval_rewards/margins": 0.04551612213253975,
|
| 643 |
+
"eval_rewards/rejected": -0.528753936290741,
|
| 644 |
+
"eval_runtime": 1563.8204,
|
| 645 |
+
"eval_samples_per_second": 1.915,
|
| 646 |
+
"eval_steps_per_second": 0.24,
|
| 647 |
+
"step": 400
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.33,
|
| 651 |
+
"learning_rate": 4.247893172092157e-06,
|
| 652 |
+
"logits/chosen": -3.3930275440216064,
|
| 653 |
+
"logits/rejected": -3.3995296955108643,
|
| 654 |
+
"logps/chosen": -320.64154052734375,
|
| 655 |
+
"logps/rejected": -351.81414794921875,
|
| 656 |
+
"loss": 0.6576,
|
| 657 |
+
"rewards/accuracies": 0.612500011920929,
|
| 658 |
+
"rewards/chosen": -0.5477547645568848,
|
| 659 |
+
"rewards/margins": 0.10663916915655136,
|
| 660 |
+
"rewards/rejected": -0.6543939709663391,
|
| 661 |
+
"step": 410
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.34,
|
| 665 |
+
"learning_rate": 4.197258402220187e-06,
|
| 666 |
+
"logits/chosen": -3.4664406776428223,
|
| 667 |
+
"logits/rejected": -3.4772841930389404,
|
| 668 |
+
"logps/chosen": -328.4302673339844,
|
| 669 |
+
"logps/rejected": -347.1799011230469,
|
| 670 |
+
"loss": 0.6516,
|
| 671 |
+
"rewards/accuracies": 0.59375,
|
| 672 |
+
"rewards/chosen": -0.5463472604751587,
|
| 673 |
+
"rewards/margins": 0.14043554663658142,
|
| 674 |
+
"rewards/rejected": -0.686782717704773,
|
| 675 |
+
"step": 420
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.34,
|
| 679 |
+
"learning_rate": 4.145297805476023e-06,
|
| 680 |
+
"logits/chosen": -3.5236573219299316,
|
| 681 |
+
"logits/rejected": -3.5294857025146484,
|
| 682 |
+
"logps/chosen": -326.1194763183594,
|
| 683 |
+
"logps/rejected": -350.4049987792969,
|
| 684 |
+
"loss": 0.6477,
|
| 685 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 686 |
+
"rewards/chosen": -0.5520918965339661,
|
| 687 |
+
"rewards/margins": 0.13327525556087494,
|
| 688 |
+
"rewards/rejected": -0.685367226600647,
|
| 689 |
+
"step": 430
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.35,
|
| 693 |
+
"learning_rate": 4.092051971292228e-06,
|
| 694 |
+
"logits/chosen": -3.597280502319336,
|
| 695 |
+
"logits/rejected": -3.597381591796875,
|
| 696 |
+
"logps/chosen": -324.05584716796875,
|
| 697 |
+
"logps/rejected": -346.5114440917969,
|
| 698 |
+
"loss": 0.654,
|
| 699 |
+
"rewards/accuracies": 0.643750011920929,
|
| 700 |
+
"rewards/chosen": -0.5706614255905151,
|
| 701 |
+
"rewards/margins": 0.11499401181936264,
|
| 702 |
+
"rewards/rejected": -0.6856554746627808,
|
| 703 |
+
"step": 440
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.36,
|
| 707 |
+
"learning_rate": 4.037562493074792e-06,
|
| 708 |
+
"logits/chosen": -3.742100238800049,
|
| 709 |
+
"logits/rejected": -3.7566616535186768,
|
| 710 |
+
"logps/chosen": -349.8996887207031,
|
| 711 |
+
"logps/rejected": -364.50506591796875,
|
| 712 |
+
"loss": 0.6477,
|
| 713 |
+
"rewards/accuracies": 0.606249988079071,
|
| 714 |
+
"rewards/chosen": -0.6562848091125488,
|
| 715 |
+
"rewards/margins": 0.1411934792995453,
|
| 716 |
+
"rewards/rejected": -0.7974783182144165,
|
| 717 |
+
"step": 450
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.37,
|
| 721 |
+
"learning_rate": 3.981871935712112e-06,
|
| 722 |
+
"logits/chosen": -3.968726396560669,
|
| 723 |
+
"logits/rejected": -3.941209077835083,
|
| 724 |
+
"logps/chosen": -332.73443603515625,
|
| 725 |
+
"logps/rejected": -360.47625732421875,
|
| 726 |
+
"loss": 0.6493,
|
| 727 |
+
"rewards/accuracies": 0.581250011920929,
|
| 728 |
+
"rewards/chosen": -0.6563684344291687,
|
| 729 |
+
"rewards/margins": 0.13294601440429688,
|
| 730 |
+
"rewards/rejected": -0.7893144488334656,
|
| 731 |
+
"step": 460
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.38,
|
| 735 |
+
"learning_rate": 3.925023802325094e-06,
|
| 736 |
+
"logits/chosen": -4.149798393249512,
|
| 737 |
+
"logits/rejected": -4.129707336425781,
|
| 738 |
+
"logps/chosen": -354.22802734375,
|
| 739 |
+
"logps/rejected": -380.85040283203125,
|
| 740 |
+
"loss": 0.6505,
|
| 741 |
+
"rewards/accuracies": 0.59375,
|
| 742 |
+
"rewards/chosen": -0.7849351763725281,
|
| 743 |
+
"rewards/margins": 0.16557727754116058,
|
| 744 |
+
"rewards/rejected": -0.9505125284194946,
|
| 745 |
+
"step": 470
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.38,
|
| 749 |
+
"learning_rate": 3.867062500284342e-06,
|
| 750 |
+
"logits/chosen": -4.189931392669678,
|
| 751 |
+
"logits/rejected": -4.1646270751953125,
|
| 752 |
+
"logps/chosen": -344.6732482910156,
|
| 753 |
+
"logps/rejected": -376.7868957519531,
|
| 754 |
+
"loss": 0.647,
|
| 755 |
+
"rewards/accuracies": 0.612500011920929,
|
| 756 |
+
"rewards/chosen": -0.8410800099372864,
|
| 757 |
+
"rewards/margins": 0.15282076597213745,
|
| 758 |
+
"rewards/rejected": -0.9939007759094238,
|
| 759 |
+
"step": 480
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.39,
|
| 763 |
+
"learning_rate": 3.8080333065209885e-06,
|
| 764 |
+
"logits/chosen": -4.1885223388671875,
|
| 765 |
+
"logits/rejected": -4.1956305503845215,
|
| 766 |
+
"logps/chosen": -351.62237548828125,
|
| 767 |
+
"logps/rejected": -351.5168151855469,
|
| 768 |
+
"loss": 0.6998,
|
| 769 |
+
"rewards/accuracies": 0.543749988079071,
|
| 770 |
+
"rewards/chosen": -0.8627797961235046,
|
| 771 |
+
"rewards/margins": 0.05857674032449722,
|
| 772 |
+
"rewards/rejected": -0.9213564991950989,
|
| 773 |
+
"step": 490
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.4,
|
| 777 |
+
"learning_rate": 3.7479823321582624e-06,
|
| 778 |
+
"logits/chosen": -4.04768705368042,
|
| 779 |
+
"logits/rejected": -4.022109031677246,
|
| 780 |
+
"logps/chosen": -345.1391296386719,
|
| 781 |
+
"logps/rejected": -380.88482666015625,
|
| 782 |
+
"loss": 0.6296,
|
| 783 |
+
"rewards/accuracies": 0.637499988079071,
|
| 784 |
+
"rewards/chosen": -0.7372187972068787,
|
| 785 |
+
"rewards/margins": 0.17785954475402832,
|
| 786 |
+
"rewards/rejected": -0.915078341960907,
|
| 787 |
+
"step": 500
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.4,
|
| 791 |
+
"eval_logits/chosen": -4.174871921539307,
|
| 792 |
+
"eval_logits/rejected": -4.1668291091918945,
|
| 793 |
+
"eval_logps/chosen": -277.3629150390625,
|
| 794 |
+
"eval_logps/rejected": -289.35430908203125,
|
| 795 |
+
"eval_loss": 0.678155779838562,
|
| 796 |
+
"eval_rewards/accuracies": 0.5600000023841858,
|
| 797 |
+
"eval_rewards/chosen": -0.6896299123764038,
|
| 798 |
+
"eval_rewards/margins": 0.06681026518344879,
|
| 799 |
+
"eval_rewards/rejected": -0.7564401626586914,
|
| 800 |
+
"eval_runtime": 1565.1126,
|
| 801 |
+
"eval_samples_per_second": 1.913,
|
| 802 |
+
"eval_steps_per_second": 0.24,
|
| 803 |
+
"step": 500
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 0.41,
|
| 807 |
+
"learning_rate": 3.686956486491419e-06,
|
| 808 |
+
"logits/chosen": -4.035499572753906,
|
| 809 |
+
"logits/rejected": -4.026083469390869,
|
| 810 |
+
"logps/chosen": -354.2256774902344,
|
| 811 |
+
"logps/rejected": -383.2754821777344,
|
| 812 |
+
"loss": 0.642,
|
| 813 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 814 |
+
"rewards/chosen": -0.7423269748687744,
|
| 815 |
+
"rewards/margins": 0.21402092278003693,
|
| 816 |
+
"rewards/rejected": -0.9563478231430054,
|
| 817 |
+
"step": 510
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 0.42,
|
| 821 |
+
"learning_rate": 3.625003440344166e-06,
|
| 822 |
+
"logits/chosen": -4.110401630401611,
|
| 823 |
+
"logits/rejected": -4.130579948425293,
|
| 824 |
+
"logps/chosen": -333.2506103515625,
|
| 825 |
+
"logps/rejected": -345.42529296875,
|
| 826 |
+
"loss": 0.6662,
|
| 827 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 828 |
+
"rewards/chosen": -0.6982457637786865,
|
| 829 |
+
"rewards/margins": 0.08196356892585754,
|
| 830 |
+
"rewards/rejected": -0.7802093625068665,
|
| 831 |
+
"step": 520
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.42,
|
| 835 |
+
"learning_rate": 3.562171588830231e-06,
|
| 836 |
+
"logits/chosen": -4.092467308044434,
|
| 837 |
+
"logits/rejected": -4.074677467346191,
|
| 838 |
+
"logps/chosen": -342.26513671875,
|
| 839 |
+
"logps/rejected": -370.6748352050781,
|
| 840 |
+
"loss": 0.6627,
|
| 841 |
+
"rewards/accuracies": 0.637499988079071,
|
| 842 |
+
"rewards/chosen": -0.6801224946975708,
|
| 843 |
+
"rewards/margins": 0.11174317449331284,
|
| 844 |
+
"rewards/rejected": -0.7918656468391418,
|
| 845 |
+
"step": 530
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 0.43,
|
| 849 |
+
"learning_rate": 3.4985100135491245e-06,
|
| 850 |
+
"logits/chosen": -4.104280471801758,
|
| 851 |
+
"logits/rejected": -4.07126522064209,
|
| 852 |
+
"logps/chosen": -353.3872375488281,
|
| 853 |
+
"logps/rejected": -391.28289794921875,
|
| 854 |
+
"loss": 0.6383,
|
| 855 |
+
"rewards/accuracies": 0.59375,
|
| 856 |
+
"rewards/chosen": -0.7574983239173889,
|
| 857 |
+
"rewards/margins": 0.18485240638256073,
|
| 858 |
+
"rewards/rejected": -0.9423506855964661,
|
| 859 |
+
"step": 540
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 0.44,
|
| 863 |
+
"learning_rate": 3.4340684442456673e-06,
|
| 864 |
+
"logits/chosen": -4.108157634735107,
|
| 865 |
+
"logits/rejected": -4.100740432739258,
|
| 866 |
+
"logps/chosen": -353.9938049316406,
|
| 867 |
+
"logps/rejected": -375.70562744140625,
|
| 868 |
+
"loss": 0.6498,
|
| 869 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 870 |
+
"rewards/chosen": -0.8029891848564148,
|
| 871 |
+
"rewards/margins": 0.12438172101974487,
|
| 872 |
+
"rewards/rejected": -0.9273708462715149,
|
| 873 |
+
"step": 550
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.45,
|
| 877 |
+
"learning_rate": 3.3688972199631974e-06,
|
| 878 |
+
"logits/chosen": -4.0825605392456055,
|
| 879 |
+
"logits/rejected": -4.094653129577637,
|
| 880 |
+
"logps/chosen": -352.5472412109375,
|
| 881 |
+
"logps/rejected": -372.22125244140625,
|
| 882 |
+
"loss": 0.646,
|
| 883 |
+
"rewards/accuracies": 0.668749988079071,
|
| 884 |
+
"rewards/chosen": -0.7980318069458008,
|
| 885 |
+
"rewards/margins": 0.1836821585893631,
|
| 886 |
+
"rewards/rejected": -0.9817139506340027,
|
| 887 |
+
"step": 560
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 0.46,
|
| 891 |
+
"learning_rate": 3.3030472497208354e-06,
|
| 892 |
+
"logits/chosen": -4.089273452758789,
|
| 893 |
+
"logits/rejected": -4.051713943481445,
|
| 894 |
+
"logps/chosen": -349.65447998046875,
|
| 895 |
+
"logps/rejected": -411.3336486816406,
|
| 896 |
+
"loss": 0.6145,
|
| 897 |
+
"rewards/accuracies": 0.637499988079071,
|
| 898 |
+
"rewards/chosen": -0.8661645650863647,
|
| 899 |
+
"rewards/margins": 0.2521691918373108,
|
| 900 |
+
"rewards/rejected": -1.1183335781097412,
|
| 901 |
+
"step": 570
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 0.46,
|
| 905 |
+
"learning_rate": 3.236569972745492e-06,
|
| 906 |
+
"logits/chosen": -4.18710994720459,
|
| 907 |
+
"logits/rejected": -4.17570161819458,
|
| 908 |
+
"logps/chosen": -347.78521728515625,
|
| 909 |
+
"logps/rejected": -368.5164489746094,
|
| 910 |
+
"loss": 0.6671,
|
| 911 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 912 |
+
"rewards/chosen": -0.878218948841095,
|
| 913 |
+
"rewards/margins": 0.11807211488485336,
|
| 914 |
+
"rewards/rejected": -0.9962911605834961,
|
| 915 |
+
"step": 580
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 0.47,
|
| 919 |
+
"learning_rate": 3.1695173182897126e-06,
|
| 920 |
+
"logits/chosen": -4.180907249450684,
|
| 921 |
+
"logits/rejected": -4.169510841369629,
|
| 922 |
+
"logps/chosen": -363.77081298828125,
|
| 923 |
+
"logps/rejected": -398.118408203125,
|
| 924 |
+
"loss": 0.6542,
|
| 925 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 926 |
+
"rewards/chosen": -0.982130229473114,
|
| 927 |
+
"rewards/margins": 0.1635296642780304,
|
| 928 |
+
"rewards/rejected": -1.1456598043441772,
|
| 929 |
+
"step": 590
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.48,
|
| 933 |
+
"learning_rate": 3.10194166506673e-06,
|
| 934 |
+
"logits/chosen": -4.227340221405029,
|
| 935 |
+
"logits/rejected": -4.191034317016602,
|
| 936 |
+
"logps/chosen": -362.04327392578125,
|
| 937 |
+
"logps/rejected": -401.2042541503906,
|
| 938 |
+
"loss": 0.6503,
|
| 939 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 940 |
+
"rewards/chosen": -0.9192866086959839,
|
| 941 |
+
"rewards/margins": 0.18541966378688812,
|
| 942 |
+
"rewards/rejected": -1.1047064065933228,
|
| 943 |
+
"step": 600
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 0.48,
|
| 947 |
+
"eval_logits/chosen": -4.443284511566162,
|
| 948 |
+
"eval_logits/rejected": -4.434456825256348,
|
| 949 |
+
"eval_logps/chosen": -304.2834167480469,
|
| 950 |
+
"eval_logps/rejected": -318.1134033203125,
|
| 951 |
+
"eval_loss": 0.6769664287567139,
|
| 952 |
+
"eval_rewards/accuracies": 0.5533333420753479,
|
| 953 |
+
"eval_rewards/chosen": -0.9588348269462585,
|
| 954 |
+
"eval_rewards/margins": 0.0851958841085434,
|
| 955 |
+
"eval_rewards/rejected": -1.0440306663513184,
|
| 956 |
+
"eval_runtime": 1565.9018,
|
| 957 |
+
"eval_samples_per_second": 1.912,
|
| 958 |
+
"eval_steps_per_second": 0.239,
|
| 959 |
+
"step": 600
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.49,
|
| 963 |
+
"learning_rate": 3.0338958003344115e-06,
|
| 964 |
+
"logits/chosen": -4.3410210609436035,
|
| 965 |
+
"logits/rejected": -4.294773101806641,
|
| 966 |
+
"logps/chosen": -360.5263977050781,
|
| 967 |
+
"logps/rejected": -391.7284240722656,
|
| 968 |
+
"loss": 0.6595,
|
| 969 |
+
"rewards/accuracies": 0.668749988079071,
|
| 970 |
+
"rewards/chosen": -0.9942121505737305,
|
| 971 |
+
"rewards/margins": 0.21516022086143494,
|
| 972 |
+
"rewards/rejected": -1.2093722820281982,
|
| 973 |
+
"step": 610
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 0.5,
|
| 977 |
+
"learning_rate": 2.9654328786600823e-06,
|
| 978 |
+
"logits/chosen": -4.326909065246582,
|
| 979 |
+
"logits/rejected": -4.2857208251953125,
|
| 980 |
+
"logps/chosen": -359.14569091796875,
|
| 981 |
+
"logps/rejected": -399.8367004394531,
|
| 982 |
+
"loss": 0.627,
|
| 983 |
+
"rewards/accuracies": 0.65625,
|
| 984 |
+
"rewards/chosen": -0.9758028984069824,
|
| 985 |
+
"rewards/margins": 0.18271294236183167,
|
| 986 |
+
"rewards/rejected": -1.1585158109664917,
|
| 987 |
+
"step": 620
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.5,
|
| 991 |
+
"learning_rate": 2.896606380398402e-06,
|
| 992 |
+
"logits/chosen": -4.396684169769287,
|
| 993 |
+
"logits/rejected": -4.430912971496582,
|
| 994 |
+
"logps/chosen": -377.33551025390625,
|
| 995 |
+
"logps/rejected": -398.4137878417969,
|
| 996 |
+
"loss": 0.6688,
|
| 997 |
+
"rewards/accuracies": 0.612500011920929,
|
| 998 |
+
"rewards/chosen": -1.0871975421905518,
|
| 999 |
+
"rewards/margins": 0.13647660613059998,
|
| 1000 |
+
"rewards/rejected": -1.2236744165420532,
|
| 1001 |
+
"step": 630
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.51,
|
| 1005 |
+
"learning_rate": 2.827470069914772e-06,
|
| 1006 |
+
"logits/chosen": -4.29737663269043,
|
| 1007 |
+
"logits/rejected": -4.263535499572754,
|
| 1008 |
+
"logps/chosen": -376.9189453125,
|
| 1009 |
+
"logps/rejected": -404.60247802734375,
|
| 1010 |
+
"loss": 0.6551,
|
| 1011 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1012 |
+
"rewards/chosen": -1.0662126541137695,
|
| 1013 |
+
"rewards/margins": 0.14261041581630707,
|
| 1014 |
+
"rewards/rejected": -1.2088232040405273,
|
| 1015 |
+
"step": 640
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 0.52,
|
| 1019 |
+
"learning_rate": 2.7580779535868675e-06,
|
| 1020 |
+
"logits/chosen": -4.269147872924805,
|
| 1021 |
+
"logits/rejected": -4.276608467102051,
|
| 1022 |
+
"logps/chosen": -366.36492919921875,
|
| 1023 |
+
"logps/rejected": -390.6036376953125,
|
| 1024 |
+
"loss": 0.6583,
|
| 1025 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1026 |
+
"rewards/chosen": -0.9485187530517578,
|
| 1027 |
+
"rewards/margins": 0.15207722783088684,
|
| 1028 |
+
"rewards/rejected": -1.1005959510803223,
|
| 1029 |
+
"step": 650
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.53,
|
| 1033 |
+
"learning_rate": 2.688484237617129e-06,
|
| 1034 |
+
"logits/chosen": -4.208004474639893,
|
| 1035 |
+
"logits/rejected": -4.182279586791992,
|
| 1036 |
+
"logps/chosen": -362.94586181640625,
|
| 1037 |
+
"logps/rejected": -393.9739990234375,
|
| 1038 |
+
"loss": 0.6322,
|
| 1039 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1040 |
+
"rewards/chosen": -0.8923367261886597,
|
| 1041 |
+
"rewards/margins": 0.20138171315193176,
|
| 1042 |
+
"rewards/rejected": -1.0937182903289795,
|
| 1043 |
+
"step": 660
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 0.54,
|
| 1047 |
+
"learning_rate": 2.6187432856891585e-06,
|
| 1048 |
+
"logits/chosen": -4.1537322998046875,
|
| 1049 |
+
"logits/rejected": -4.151703357696533,
|
| 1050 |
+
"logps/chosen": -372.31201171875,
|
| 1051 |
+
"logps/rejected": -405.5547180175781,
|
| 1052 |
+
"loss": 0.6489,
|
| 1053 |
+
"rewards/accuracies": 0.59375,
|
| 1054 |
+
"rewards/chosen": -0.953785240650177,
|
| 1055 |
+
"rewards/margins": 0.15488819777965546,
|
| 1056 |
+
"rewards/rejected": -1.108673334121704,
|
| 1057 |
+
"step": 670
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 0.54,
|
| 1061 |
+
"learning_rate": 2.548909576501096e-06,
|
| 1062 |
+
"logits/chosen": -4.182751655578613,
|
| 1063 |
+
"logits/rejected": -4.181895732879639,
|
| 1064 |
+
"logps/chosen": -372.65435791015625,
|
| 1065 |
+
"logps/rejected": -398.80902099609375,
|
| 1066 |
+
"loss": 0.6442,
|
| 1067 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1068 |
+
"rewards/chosen": -0.9745942950248718,
|
| 1069 |
+
"rewards/margins": 0.160622701048851,
|
| 1070 |
+
"rewards/rejected": -1.1352169513702393,
|
| 1071 |
+
"step": 680
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 0.55,
|
| 1075 |
+
"learning_rate": 2.4790376612091503e-06,
|
| 1076 |
+
"logits/chosen": -4.2613630294799805,
|
| 1077 |
+
"logits/rejected": -4.227023124694824,
|
| 1078 |
+
"logps/chosen": -391.2955017089844,
|
| 1079 |
+
"logps/rejected": -419.54656982421875,
|
| 1080 |
+
"loss": 0.6233,
|
| 1081 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1082 |
+
"rewards/chosen": -1.0887335538864136,
|
| 1083 |
+
"rewards/margins": 0.2128380537033081,
|
| 1084 |
+
"rewards/rejected": -1.3015714883804321,
|
| 1085 |
+
"step": 690
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 0.56,
|
| 1089 |
+
"learning_rate": 2.40918212081453e-06,
|
| 1090 |
+
"logits/chosen": -4.308588981628418,
|
| 1091 |
+
"logits/rejected": -4.264666557312012,
|
| 1092 |
+
"logps/chosen": -360.5702819824219,
|
| 1093 |
+
"logps/rejected": -409.14422607421875,
|
| 1094 |
+
"loss": 0.5974,
|
| 1095 |
+
"rewards/accuracies": 0.6875,
|
| 1096 |
+
"rewards/chosen": -1.0383028984069824,
|
| 1097 |
+
"rewards/margins": 0.29866236448287964,
|
| 1098 |
+
"rewards/rejected": -1.3369653224945068,
|
| 1099 |
+
"step": 700
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 0.56,
|
| 1103 |
+
"eval_logits/chosen": -4.447990417480469,
|
| 1104 |
+
"eval_logits/rejected": -4.436967849731445,
|
| 1105 |
+
"eval_logps/chosen": -322.94854736328125,
|
| 1106 |
+
"eval_logps/rejected": -338.0312194824219,
|
| 1107 |
+
"eval_loss": 0.6777762174606323,
|
| 1108 |
+
"eval_rewards/accuracies": 0.5653333067893982,
|
| 1109 |
+
"eval_rewards/chosen": -1.1454858779907227,
|
| 1110 |
+
"eval_rewards/margins": 0.09772318601608276,
|
| 1111 |
+
"eval_rewards/rejected": -1.2432090044021606,
|
| 1112 |
+
"eval_runtime": 1566.0394,
|
| 1113 |
+
"eval_samples_per_second": 1.912,
|
| 1114 |
+
"eval_steps_per_second": 0.239,
|
| 1115 |
+
"step": 700
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.57,
|
| 1119 |
+
"learning_rate": 2.3393975235270654e-06,
|
| 1120 |
+
"logits/chosen": -4.278247833251953,
|
| 1121 |
+
"logits/rejected": -4.263747692108154,
|
| 1122 |
+
"logps/chosen": -394.7868347167969,
|
| 1123 |
+
"logps/rejected": -433.32373046875,
|
| 1124 |
+
"loss": 0.6411,
|
| 1125 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1126 |
+
"rewards/chosen": -1.1327521800994873,
|
| 1127 |
+
"rewards/margins": 0.22177498042583466,
|
| 1128 |
+
"rewards/rejected": -1.354527235031128,
|
| 1129 |
+
"step": 710
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.58,
|
| 1133 |
+
"learning_rate": 2.2697383821388153e-06,
|
| 1134 |
+
"logits/chosen": -4.283724308013916,
|
| 1135 |
+
"logits/rejected": -4.2910637855529785,
|
| 1136 |
+
"logps/chosen": -383.4498596191406,
|
| 1137 |
+
"logps/rejected": -406.142578125,
|
| 1138 |
+
"loss": 0.6479,
|
| 1139 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1140 |
+
"rewards/chosen": -1.1030470132827759,
|
| 1141 |
+
"rewards/margins": 0.19455528259277344,
|
| 1142 |
+
"rewards/rejected": -1.2976024150848389,
|
| 1143 |
+
"step": 720
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.58,
|
| 1147 |
+
"learning_rate": 2.2002591114409657e-06,
|
| 1148 |
+
"logits/chosen": -4.243215560913086,
|
| 1149 |
+
"logits/rejected": -4.230997562408447,
|
| 1150 |
+
"logps/chosen": -392.0690612792969,
|
| 1151 |
+
"logps/rejected": -419.01092529296875,
|
| 1152 |
+
"loss": 0.6517,
|
| 1153 |
+
"rewards/accuracies": 0.581250011920929,
|
| 1154 |
+
"rewards/chosen": -1.1523319482803345,
|
| 1155 |
+
"rewards/margins": 0.17104671895503998,
|
| 1156 |
+
"rewards/rejected": -1.323378562927246,
|
| 1157 |
+
"step": 730
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.59,
|
| 1161 |
+
"learning_rate": 2.131013985717285e-06,
|
| 1162 |
+
"logits/chosen": -4.2963361740112305,
|
| 1163 |
+
"logits/rejected": -4.248631477355957,
|
| 1164 |
+
"logps/chosen": -402.02130126953125,
|
| 1165 |
+
"logps/rejected": -451.04400634765625,
|
| 1166 |
+
"loss": 0.6211,
|
| 1167 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1168 |
+
"rewards/chosen": -1.197863221168518,
|
| 1169 |
+
"rewards/margins": 0.26271212100982666,
|
| 1170 |
+
"rewards/rejected": -1.4605753421783447,
|
| 1171 |
+
"step": 740
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.6,
|
| 1175 |
+
"learning_rate": 2.062057096347338e-06,
|
| 1176 |
+
"logits/chosen": -4.2642927169799805,
|
| 1177 |
+
"logits/rejected": -4.245224952697754,
|
| 1178 |
+
"logps/chosen": -386.35394287109375,
|
| 1179 |
+
"logps/rejected": -399.6516418457031,
|
| 1180 |
+
"loss": 0.6611,
|
| 1181 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1182 |
+
"rewards/chosen": -1.1031478643417358,
|
| 1183 |
+
"rewards/margins": 0.12747173011302948,
|
| 1184 |
+
"rewards/rejected": -1.2306196689605713,
|
| 1185 |
+
"step": 750
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.61,
|
| 1189 |
+
"learning_rate": 1.9934423095525733e-06,
|
| 1190 |
+
"logits/chosen": -4.144771575927734,
|
| 1191 |
+
"logits/rejected": -4.146181583404541,
|
| 1192 |
+
"logps/chosen": -381.7077941894531,
|
| 1193 |
+
"logps/rejected": -401.769287109375,
|
| 1194 |
+
"loss": 0.6555,
|
| 1195 |
+
"rewards/accuracies": 0.606249988079071,
|
| 1196 |
+
"rewards/chosen": -0.9921267628669739,
|
| 1197 |
+
"rewards/margins": 0.19879314303398132,
|
| 1198 |
+
"rewards/rejected": -1.190920114517212,
|
| 1199 |
+
"step": 760
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.62,
|
| 1203 |
+
"learning_rate": 1.9252232243182986e-06,
|
| 1204 |
+
"logits/chosen": -4.229475498199463,
|
| 1205 |
+
"logits/rejected": -4.177742958068848,
|
| 1206 |
+
"logps/chosen": -345.7065734863281,
|
| 1207 |
+
"logps/rejected": -388.74462890625,
|
| 1208 |
+
"loss": 0.6144,
|
| 1209 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1210 |
+
"rewards/chosen": -0.8882685899734497,
|
| 1211 |
+
"rewards/margins": 0.26380977034568787,
|
| 1212 |
+
"rewards/rejected": -1.15207839012146,
|
| 1213 |
+
"step": 770
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.62,
|
| 1217 |
+
"learning_rate": 1.8574531305244043e-06,
|
| 1218 |
+
"logits/chosen": -4.0229105949401855,
|
| 1219 |
+
"logits/rejected": -4.011281490325928,
|
| 1220 |
+
"logps/chosen": -381.92205810546875,
|
| 1221 |
+
"logps/rejected": -418.5514221191406,
|
| 1222 |
+
"loss": 0.6197,
|
| 1223 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1224 |
+
"rewards/chosen": -1.013808250427246,
|
| 1225 |
+
"rewards/margins": 0.23891642689704895,
|
| 1226 |
+
"rewards/rejected": -1.2527246475219727,
|
| 1227 |
+
"step": 780
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.63,
|
| 1231 |
+
"learning_rate": 1.7901849673175559e-06,
|
| 1232 |
+
"logits/chosen": -4.090008735656738,
|
| 1233 |
+
"logits/rejected": -4.056280612945557,
|
| 1234 |
+
"logps/chosen": -375.6288146972656,
|
| 1235 |
+
"logps/rejected": -409.71490478515625,
|
| 1236 |
+
"loss": 0.634,
|
| 1237 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1238 |
+
"rewards/chosen": -0.9446262121200562,
|
| 1239 |
+
"rewards/margins": 0.2019636631011963,
|
| 1240 |
+
"rewards/rejected": -1.1465898752212524,
|
| 1241 |
+
"step": 790
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.64,
|
| 1245 |
+
"learning_rate": 1.7234712817573555e-06,
|
| 1246 |
+
"logits/chosen": -4.081742763519287,
|
| 1247 |
+
"logits/rejected": -4.069535255432129,
|
| 1248 |
+
"logps/chosen": -408.01580810546875,
|
| 1249 |
+
"logps/rejected": -428.19830322265625,
|
| 1250 |
+
"loss": 0.6508,
|
| 1251 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1252 |
+
"rewards/chosen": -1.1142655611038208,
|
| 1253 |
+
"rewards/margins": 0.18682199716567993,
|
| 1254 |
+
"rewards/rejected": -1.301087498664856,
|
| 1255 |
+
"step": 800
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.64,
|
| 1259 |
+
"eval_logits/chosen": -4.271145343780518,
|
| 1260 |
+
"eval_logits/rejected": -4.261835098266602,
|
| 1261 |
+
"eval_logps/chosen": -318.4167785644531,
|
| 1262 |
+
"eval_logps/rejected": -333.9435119628906,
|
| 1263 |
+
"eval_loss": 0.6748104691505432,
|
| 1264 |
+
"eval_rewards/accuracies": 0.5649999976158142,
|
| 1265 |
+
"eval_rewards/chosen": -1.1001687049865723,
|
| 1266 |
+
"eval_rewards/margins": 0.10216327011585236,
|
| 1267 |
+
"eval_rewards/rejected": -1.2023320198059082,
|
| 1268 |
+
"eval_runtime": 1566.0255,
|
| 1269 |
+
"eval_samples_per_second": 1.912,
|
| 1270 |
+
"eval_steps_per_second": 0.239,
|
| 1271 |
+
"step": 800
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 0.65,
|
| 1275 |
+
"learning_rate": 1.6573641877687936e-06,
|
| 1276 |
+
"logits/chosen": -4.110816478729248,
|
| 1277 |
+
"logits/rejected": -4.093571662902832,
|
| 1278 |
+
"logps/chosen": -381.05877685546875,
|
| 1279 |
+
"logps/rejected": -423.17364501953125,
|
| 1280 |
+
"loss": 0.6202,
|
| 1281 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1282 |
+
"rewards/chosen": -1.0917795896530151,
|
| 1283 |
+
"rewards/margins": 0.21704904735088348,
|
| 1284 |
+
"rewards/rejected": -1.308828592300415,
|
| 1285 |
+
"step": 810
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 0.66,
|
| 1289 |
+
"learning_rate": 1.591915325433034e-06,
|
| 1290 |
+
"logits/chosen": -4.177929878234863,
|
| 1291 |
+
"logits/rejected": -4.188526630401611,
|
| 1292 |
+
"logps/chosen": -370.15655517578125,
|
| 1293 |
+
"logps/rejected": -401.153564453125,
|
| 1294 |
+
"loss": 0.6176,
|
| 1295 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1296 |
+
"rewards/chosen": -1.020612359046936,
|
| 1297 |
+
"rewards/margins": 0.26860466599464417,
|
| 1298 |
+
"rewards/rejected": -1.2892169952392578,
|
| 1299 |
+
"step": 820
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 0.66,
|
| 1303 |
+
"learning_rate": 1.5271758206483664e-06,
|
| 1304 |
+
"logits/chosen": -4.208674430847168,
|
| 1305 |
+
"logits/rejected": -4.197338104248047,
|
| 1306 |
+
"logps/chosen": -399.302490234375,
|
| 1307 |
+
"logps/rejected": -428.2151794433594,
|
| 1308 |
+
"loss": 0.6565,
|
| 1309 |
+
"rewards/accuracies": 0.71875,
|
| 1310 |
+
"rewards/chosen": -1.2150713205337524,
|
| 1311 |
+
"rewards/margins": 0.18397213518619537,
|
| 1312 |
+
"rewards/rejected": -1.3990434408187866,
|
| 1313 |
+
"step": 830
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.67,
|
| 1317 |
+
"learning_rate": 1.4631962451927966e-06,
|
| 1318 |
+
"logits/chosen": -4.116250514984131,
|
| 1319 |
+
"logits/rejected": -4.103257656097412,
|
| 1320 |
+
"logps/chosen": -396.60467529296875,
|
| 1321 |
+
"logps/rejected": -433.5016174316406,
|
| 1322 |
+
"loss": 0.6155,
|
| 1323 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1324 |
+
"rewards/chosen": -1.1897716522216797,
|
| 1325 |
+
"rewards/margins": 0.23897738754749298,
|
| 1326 |
+
"rewards/rejected": -1.4287490844726562,
|
| 1327 |
+
"step": 840
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 0.68,
|
| 1331 |
+
"learning_rate": 1.4000265772195032e-06,
|
| 1332 |
+
"logits/chosen": -4.289558410644531,
|
| 1333 |
+
"logits/rejected": -4.233942985534668,
|
| 1334 |
+
"logps/chosen": -391.7837219238281,
|
| 1335 |
+
"logps/rejected": -433.24261474609375,
|
| 1336 |
+
"loss": 0.6144,
|
| 1337 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1338 |
+
"rewards/chosen": -1.1339555978775024,
|
| 1339 |
+
"rewards/margins": 0.25826430320739746,
|
| 1340 |
+
"rewards/rejected": -1.3922197818756104,
|
| 1341 |
+
"step": 850
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 0.69,
|
| 1345 |
+
"learning_rate": 1.3377161622160077e-06,
|
| 1346 |
+
"logits/chosen": -4.226014137268066,
|
| 1347 |
+
"logits/rejected": -4.22386360168457,
|
| 1348 |
+
"logps/chosen": -392.7188415527344,
|
| 1349 |
+
"logps/rejected": -428.2996520996094,
|
| 1350 |
+
"loss": 0.6116,
|
| 1351 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1352 |
+
"rewards/chosen": -1.1736845970153809,
|
| 1353 |
+
"rewards/margins": 0.23641009628772736,
|
| 1354 |
+
"rewards/rejected": -1.4100948572158813,
|
| 1355 |
+
"step": 860
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 0.7,
|
| 1359 |
+
"learning_rate": 1.276313674457553e-06,
|
| 1360 |
+
"logits/chosen": -4.321101188659668,
|
| 1361 |
+
"logits/rejected": -4.316857814788818,
|
| 1362 |
+
"logps/chosen": -376.7432556152344,
|
| 1363 |
+
"logps/rejected": -420.355224609375,
|
| 1364 |
+
"loss": 0.6062,
|
| 1365 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1366 |
+
"rewards/chosen": -1.1783860921859741,
|
| 1367 |
+
"rewards/margins": 0.2604502737522125,
|
| 1368 |
+
"rewards/rejected": -1.4388363361358643,
|
| 1369 |
+
"step": 870
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 0.7,
|
| 1373 |
+
"learning_rate": 1.2158670789848095e-06,
|
| 1374 |
+
"logits/chosen": -4.357269287109375,
|
| 1375 |
+
"logits/rejected": -4.3427653312683105,
|
| 1376 |
+
"logps/chosen": -412.49603271484375,
|
| 1377 |
+
"logps/rejected": -446.0122985839844,
|
| 1378 |
+
"loss": 0.6207,
|
| 1379 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1380 |
+
"rewards/chosen": -1.3438339233398438,
|
| 1381 |
+
"rewards/margins": 0.2622073292732239,
|
| 1382 |
+
"rewards/rejected": -1.6060411930084229,
|
| 1383 |
+
"step": 880
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 0.71,
|
| 1387 |
+
"learning_rate": 1.1564235941356016e-06,
|
| 1388 |
+
"logits/chosen": -4.453230381011963,
|
| 1389 |
+
"logits/rejected": -4.401095390319824,
|
| 1390 |
+
"logps/chosen": -398.6358337402344,
|
| 1391 |
+
"logps/rejected": -450.500244140625,
|
| 1392 |
+
"loss": 0.6262,
|
| 1393 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1394 |
+
"rewards/chosen": -1.3652451038360596,
|
| 1395 |
+
"rewards/margins": 0.3177439570426941,
|
| 1396 |
+
"rewards/rejected": -1.6829888820648193,
|
| 1397 |
+
"step": 890
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 0.72,
|
| 1401 |
+
"learning_rate": 1.0980296546599254e-06,
|
| 1402 |
+
"logits/chosen": -4.331755638122559,
|
| 1403 |
+
"logits/rejected": -4.3203840255737305,
|
| 1404 |
+
"logps/chosen": -422.264404296875,
|
| 1405 |
+
"logps/rejected": -445.89739990234375,
|
| 1406 |
+
"loss": 0.6746,
|
| 1407 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 1408 |
+
"rewards/chosen": -1.3867082595825195,
|
| 1409 |
+
"rewards/margins": 0.28304582834243774,
|
| 1410 |
+
"rewards/rejected": -1.6697540283203125,
|
| 1411 |
+
"step": 900
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 0.72,
|
| 1415 |
+
"eval_logits/chosen": -4.577242374420166,
|
| 1416 |
+
"eval_logits/rejected": -4.566162586212158,
|
| 1417 |
+
"eval_logps/chosen": -341.29400634765625,
|
| 1418 |
+
"eval_logps/rejected": -358.15576171875,
|
| 1419 |
+
"eval_loss": 0.6757350564002991,
|
| 1420 |
+
"eval_rewards/accuracies": 0.5686666369438171,
|
| 1421 |
+
"eval_rewards/chosen": -1.3289406299591064,
|
| 1422 |
+
"eval_rewards/margins": 0.11551385372877121,
|
| 1423 |
+
"eval_rewards/rejected": -1.4444544315338135,
|
| 1424 |
+
"eval_runtime": 1566.2371,
|
| 1425 |
+
"eval_samples_per_second": 1.912,
|
| 1426 |
+
"eval_steps_per_second": 0.239,
|
| 1427 |
+
"step": 900
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 0.73,
|
| 1431 |
+
"learning_rate": 1.040730875447083e-06,
|
| 1432 |
+
"logits/chosen": -4.31623649597168,
|
| 1433 |
+
"logits/rejected": -4.329934597015381,
|
| 1434 |
+
"logps/chosen": -407.89544677734375,
|
| 1435 |
+
"logps/rejected": -431.8106384277344,
|
| 1436 |
+
"loss": 0.6508,
|
| 1437 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1438 |
+
"rewards/chosen": -1.282503604888916,
|
| 1439 |
+
"rewards/margins": 0.16822777688503265,
|
| 1440 |
+
"rewards/rejected": -1.4507312774658203,
|
| 1441 |
+
"step": 910
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 0.74,
|
| 1445 |
+
"learning_rate": 9.845720158932414e-07,
|
| 1446 |
+
"logits/chosen": -4.317078590393066,
|
| 1447 |
+
"logits/rejected": -4.328789710998535,
|
| 1448 |
+
"logps/chosen": -384.6829833984375,
|
| 1449 |
+
"logps/rejected": -404.79107666015625,
|
| 1450 |
+
"loss": 0.6351,
|
| 1451 |
+
"rewards/accuracies": 0.625,
|
| 1452 |
+
"rewards/chosen": -1.2274483442306519,
|
| 1453 |
+
"rewards/margins": 0.19847750663757324,
|
| 1454 |
+
"rewards/rejected": -1.4259259700775146,
|
| 1455 |
+
"step": 920
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 0.74,
|
| 1459 |
+
"learning_rate": 9.295969449372796e-07,
|
| 1460 |
+
"logits/chosen": -4.379042625427246,
|
| 1461 |
+
"logits/rejected": -4.360438346862793,
|
| 1462 |
+
"logps/chosen": -385.20245361328125,
|
| 1463 |
+
"logps/rejected": -416.4458923339844,
|
| 1464 |
+
"loss": 0.6352,
|
| 1465 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1466 |
+
"rewards/chosen": -1.1640335321426392,
|
| 1467 |
+
"rewards/margins": 0.19592450559139252,
|
| 1468 |
+
"rewards/rejected": -1.3599579334259033,
|
| 1469 |
+
"step": 930
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 0.75,
|
| 1473 |
+
"learning_rate": 8.758486067922176e-07,
|
| 1474 |
+
"logits/chosen": -4.383969306945801,
|
| 1475 |
+
"logits/rejected": -4.344205379486084,
|
| 1476 |
+
"logps/chosen": -384.3717346191406,
|
| 1477 |
+
"logps/rejected": -427.7298889160156,
|
| 1478 |
+
"loss": 0.6135,
|
| 1479 |
+
"rewards/accuracies": 0.71875,
|
| 1480 |
+
"rewards/chosen": -1.207880973815918,
|
| 1481 |
+
"rewards/margins": 0.2520582675933838,
|
| 1482 |
+
"rewards/rejected": -1.4599391222000122,
|
| 1483 |
+
"step": 940
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 0.76,
|
| 1487 |
+
"learning_rate": 8.233689873990006e-07,
|
| 1488 |
+
"logits/chosen": -4.391127109527588,
|
| 1489 |
+
"logits/rejected": -4.357823848724365,
|
| 1490 |
+
"logps/chosen": -389.40570068359375,
|
| 1491 |
+
"logps/rejected": -443.3841247558594,
|
| 1492 |
+
"loss": 0.6165,
|
| 1493 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1494 |
+
"rewards/chosen": -1.2565414905548096,
|
| 1495 |
+
"rewards/margins": 0.3726215362548828,
|
| 1496 |
+
"rewards/rejected": -1.6291630268096924,
|
| 1497 |
+
"step": 950
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 0.77,
|
| 1501 |
+
"learning_rate": 7.721990816288555e-07,
|
| 1502 |
+
"logits/chosen": -4.3561787605285645,
|
| 1503 |
+
"logits/rejected": -4.322731971740723,
|
| 1504 |
+
"logps/chosen": -363.4510803222656,
|
| 1505 |
+
"logps/rejected": -400.2040710449219,
|
| 1506 |
+
"loss": 0.6128,
|
| 1507 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1508 |
+
"rewards/chosen": -1.1432923078536987,
|
| 1509 |
+
"rewards/margins": 0.24396805465221405,
|
| 1510 |
+
"rewards/rejected": -1.3872603178024292,
|
| 1511 |
+
"step": 960
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 0.78,
|
| 1515 |
+
"learning_rate": 7.223788612598148e-07,
|
| 1516 |
+
"logits/chosen": -4.374851703643799,
|
| 1517 |
+
"logits/rejected": -4.3536810874938965,
|
| 1518 |
+
"logps/chosen": -397.66680908203125,
|
| 1519 |
+
"logps/rejected": -428.77227783203125,
|
| 1520 |
+
"loss": 0.639,
|
| 1521 |
+
"rewards/accuracies": 0.65625,
|
| 1522 |
+
"rewards/chosen": -1.2013370990753174,
|
| 1523 |
+
"rewards/margins": 0.15685971081256866,
|
| 1524 |
+
"rewards/rejected": -1.3581968545913696,
|
| 1525 |
+
"step": 970
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 0.78,
|
| 1529 |
+
"learning_rate": 6.73947243752448e-07,
|
| 1530 |
+
"logits/chosen": -4.431571006774902,
|
| 1531 |
+
"logits/rejected": -4.440985679626465,
|
| 1532 |
+
"logps/chosen": -415.3916931152344,
|
| 1533 |
+
"logps/rejected": -431.7449645996094,
|
| 1534 |
+
"loss": 0.656,
|
| 1535 |
+
"rewards/accuracies": 0.65625,
|
| 1536 |
+
"rewards/chosen": -1.3690643310546875,
|
| 1537 |
+
"rewards/margins": 0.22862248122692108,
|
| 1538 |
+
"rewards/rejected": -1.5976868867874146,
|
| 1539 |
+
"step": 980
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 0.79,
|
| 1543 |
+
"learning_rate": 6.269420618491759e-07,
|
| 1544 |
+
"logits/chosen": -4.3968610763549805,
|
| 1545 |
+
"logits/rejected": -4.365904808044434,
|
| 1546 |
+
"logps/chosen": -384.7822265625,
|
| 1547 |
+
"logps/rejected": -417.7525939941406,
|
| 1548 |
+
"loss": 0.6459,
|
| 1549 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1550 |
+
"rewards/chosen": -1.2827244997024536,
|
| 1551 |
+
"rewards/margins": 0.17925159633159637,
|
| 1552 |
+
"rewards/rejected": -1.461976170539856,
|
| 1553 |
+
"step": 990
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 0.8,
|
| 1557 |
+
"learning_rate": 5.814000340209267e-07,
|
| 1558 |
+
"logits/chosen": -4.392641067504883,
|
| 1559 |
+
"logits/rejected": -4.334468841552734,
|
| 1560 |
+
"logps/chosen": -398.47662353515625,
|
| 1561 |
+
"logps/rejected": -445.6675720214844,
|
| 1562 |
+
"loss": 0.6151,
|
| 1563 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1564 |
+
"rewards/chosen": -1.2670553922653198,
|
| 1565 |
+
"rewards/margins": 0.2884874939918518,
|
| 1566 |
+
"rewards/rejected": -1.5555428266525269,
|
| 1567 |
+
"step": 1000
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 0.8,
|
| 1571 |
+
"eval_logits/chosen": -4.618439197540283,
|
| 1572 |
+
"eval_logits/rejected": -4.607028484344482,
|
| 1573 |
+
"eval_logps/chosen": -343.98931884765625,
|
| 1574 |
+
"eval_logps/rejected": -361.1741943359375,
|
| 1575 |
+
"eval_loss": 0.6755098700523376,
|
| 1576 |
+
"eval_rewards/accuracies": 0.5690000057220459,
|
| 1577 |
+
"eval_rewards/chosen": -1.3558939695358276,
|
| 1578 |
+
"eval_rewards/margins": 0.11874517053365707,
|
| 1579 |
+
"eval_rewards/rejected": -1.4746389389038086,
|
| 1580 |
+
"eval_runtime": 1565.397,
|
| 1581 |
+
"eval_samples_per_second": 1.913,
|
| 1582 |
+
"eval_steps_per_second": 0.24,
|
| 1583 |
+
"step": 1000
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 0.81,
|
| 1587 |
+
"learning_rate": 5.373567357842111e-07,
|
| 1588 |
+
"logits/chosen": -4.363351821899414,
|
| 1589 |
+
"logits/rejected": -4.336891174316406,
|
| 1590 |
+
"logps/chosen": -393.9075012207031,
|
| 1591 |
+
"logps/rejected": -432.1922302246094,
|
| 1592 |
+
"loss": 0.6011,
|
| 1593 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1594 |
+
"rewards/chosen": -1.2202444076538086,
|
| 1595 |
+
"rewards/margins": 0.2966518700122833,
|
| 1596 |
+
"rewards/rejected": -1.516896367073059,
|
| 1597 |
+
"step": 1010
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 0.82,
|
| 1601 |
+
"learning_rate": 4.948465719110226e-07,
|
| 1602 |
+
"logits/chosen": -4.430235862731934,
|
| 1603 |
+
"logits/rejected": -4.40949010848999,
|
| 1604 |
+
"logps/chosen": -379.41375732421875,
|
| 1605 |
+
"logps/rejected": -404.0538635253906,
|
| 1606 |
+
"loss": 0.6481,
|
| 1607 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 1608 |
+
"rewards/chosen": -1.1881725788116455,
|
| 1609 |
+
"rewards/margins": 0.17870518565177917,
|
| 1610 |
+
"rewards/rejected": -1.366877794265747,
|
| 1611 |
+
"step": 1020
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 0.82,
|
| 1615 |
+
"learning_rate": 4.539027495532766e-07,
|
| 1616 |
+
"logits/chosen": -4.403011322021484,
|
| 1617 |
+
"logits/rejected": -4.427884578704834,
|
| 1618 |
+
"logps/chosen": -381.40582275390625,
|
| 1619 |
+
"logps/rejected": -415.64044189453125,
|
| 1620 |
+
"loss": 0.6201,
|
| 1621 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1622 |
+
"rewards/chosen": -1.2115557193756104,
|
| 1623 |
+
"rewards/margins": 0.2422032654285431,
|
| 1624 |
+
"rewards/rejected": -1.4537590742111206,
|
| 1625 |
+
"step": 1030
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 0.83,
|
| 1629 |
+
"learning_rate": 4.14557252302783e-07,
|
| 1630 |
+
"logits/chosen": -4.416018962860107,
|
| 1631 |
+
"logits/rejected": -4.378045082092285,
|
| 1632 |
+
"logps/chosen": -401.25286865234375,
|
| 1633 |
+
"logps/rejected": -434.80523681640625,
|
| 1634 |
+
"loss": 0.6795,
|
| 1635 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1636 |
+
"rewards/chosen": -1.350279450416565,
|
| 1637 |
+
"rewards/margins": 0.18791106343269348,
|
| 1638 |
+
"rewards/rejected": -1.538190484046936,
|
| 1639 |
+
"step": 1040
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 0.84,
|
| 1643 |
+
"learning_rate": 3.7684081520700884e-07,
|
| 1644 |
+
"logits/chosen": -4.323966026306152,
|
| 1645 |
+
"logits/rejected": -4.319721221923828,
|
| 1646 |
+
"logps/chosen": -411.38812255859375,
|
| 1647 |
+
"logps/rejected": -426.6982421875,
|
| 1648 |
+
"loss": 0.6426,
|
| 1649 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1650 |
+
"rewards/chosen": -1.2739022970199585,
|
| 1651 |
+
"rewards/margins": 0.20139846205711365,
|
| 1652 |
+
"rewards/rejected": -1.475300908088684,
|
| 1653 |
+
"step": 1050
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 0.85,
|
| 1657 |
+
"learning_rate": 3.407829007601507e-07,
|
| 1658 |
+
"logits/chosen": -4.3614420890808105,
|
| 1659 |
+
"logits/rejected": -4.317663192749023,
|
| 1660 |
+
"logps/chosen": -395.0092468261719,
|
| 1661 |
+
"logps/rejected": -438.6324157714844,
|
| 1662 |
+
"loss": 0.6132,
|
| 1663 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1664 |
+
"rewards/chosen": -1.2276242971420288,
|
| 1665 |
+
"rewards/margins": 0.28230905532836914,
|
| 1666 |
+
"rewards/rejected": -1.5099334716796875,
|
| 1667 |
+
"step": 1060
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 0.86,
|
| 1671 |
+
"learning_rate": 3.064116758882724e-07,
|
| 1672 |
+
"logits/chosen": -4.330022811889648,
|
| 1673 |
+
"logits/rejected": -4.274947643280029,
|
| 1674 |
+
"logps/chosen": -406.9677429199219,
|
| 1675 |
+
"logps/rejected": -459.18048095703125,
|
| 1676 |
+
"loss": 0.6063,
|
| 1677 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1678 |
+
"rewards/chosen": -1.304404616355896,
|
| 1679 |
+
"rewards/margins": 0.33513855934143066,
|
| 1680 |
+
"rewards/rejected": -1.6395431756973267,
|
| 1681 |
+
"step": 1070
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 0.86,
|
| 1685 |
+
"learning_rate": 2.737539899464908e-07,
|
| 1686 |
+
"logits/chosen": -4.385619163513184,
|
| 1687 |
+
"logits/rejected": -4.378711700439453,
|
| 1688 |
+
"logps/chosen": -370.54156494140625,
|
| 1689 |
+
"logps/rejected": -411.8634338378906,
|
| 1690 |
+
"loss": 0.6197,
|
| 1691 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1692 |
+
"rewards/chosen": -1.1883795261383057,
|
| 1693 |
+
"rewards/margins": 0.2932683825492859,
|
| 1694 |
+
"rewards/rejected": -1.4816478490829468,
|
| 1695 |
+
"step": 1080
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 0.87,
|
| 1699 |
+
"learning_rate": 2.4283535374538645e-07,
|
| 1700 |
+
"logits/chosen": -4.3373942375183105,
|
| 1701 |
+
"logits/rejected": -4.339685916900635,
|
| 1702 |
+
"logps/chosen": -400.1960754394531,
|
| 1703 |
+
"logps/rejected": -430.27557373046875,
|
| 1704 |
+
"loss": 0.6378,
|
| 1705 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 1706 |
+
"rewards/chosen": -1.2183153629302979,
|
| 1707 |
+
"rewards/margins": 0.2266024649143219,
|
| 1708 |
+
"rewards/rejected": -1.444917917251587,
|
| 1709 |
+
"step": 1090
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 0.88,
|
| 1713 |
+
"learning_rate": 2.1367991962303298e-07,
|
| 1714 |
+
"logits/chosen": -4.3288164138793945,
|
| 1715 |
+
"logits/rejected": -4.31368350982666,
|
| 1716 |
+
"logps/chosen": -382.65814208984375,
|
| 1717 |
+
"logps/rejected": -405.8753967285156,
|
| 1718 |
+
"loss": 0.6837,
|
| 1719 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 1720 |
+
"rewards/chosen": -1.2913353443145752,
|
| 1721 |
+
"rewards/margins": 0.10410255193710327,
|
| 1722 |
+
"rewards/rejected": -1.3954378366470337,
|
| 1723 |
+
"step": 1100
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 0.88,
|
| 1727 |
+
"eval_logits/chosen": -4.571716785430908,
|
| 1728 |
+
"eval_logits/rejected": -4.5607380867004395,
|
| 1729 |
+
"eval_logps/chosen": -340.8576354980469,
|
| 1730 |
+
"eval_logps/rejected": -358.08392333984375,
|
| 1731 |
+
"eval_loss": 0.6748068928718567,
|
| 1732 |
+
"eval_rewards/accuracies": 0.5709999799728394,
|
| 1733 |
+
"eval_rewards/chosen": -1.3245766162872314,
|
| 1734 |
+
"eval_rewards/margins": 0.11915931850671768,
|
| 1735 |
+
"eval_rewards/rejected": -1.4437358379364014,
|
| 1736 |
+
"eval_runtime": 1563.7394,
|
| 1737 |
+
"eval_samples_per_second": 1.915,
|
| 1738 |
+
"eval_steps_per_second": 0.24,
|
| 1739 |
+
"step": 1100
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.89,
|
| 1743 |
+
"learning_rate": 1.8631046257820278e-07,
|
| 1744 |
+
"logits/chosen": -4.340999603271484,
|
| 1745 |
+
"logits/rejected": -4.353137969970703,
|
| 1746 |
+
"logps/chosen": -399.4720153808594,
|
| 1747 |
+
"logps/rejected": -427.0414123535156,
|
| 1748 |
+
"loss": 0.6277,
|
| 1749 |
+
"rewards/accuracies": 0.65625,
|
| 1750 |
+
"rewards/chosen": -1.2465407848358154,
|
| 1751 |
+
"rewards/margins": 0.23380056023597717,
|
| 1752 |
+
"rewards/rejected": -1.4803413152694702,
|
| 1753 |
+
"step": 1110
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.9,
|
| 1757 |
+
"learning_rate": 1.6074836247950143e-07,
|
| 1758 |
+
"logits/chosen": -4.328183174133301,
|
| 1759 |
+
"logits/rejected": -4.327004432678223,
|
| 1760 |
+
"logps/chosen": -406.1539306640625,
|
| 1761 |
+
"logps/rejected": -430.02044677734375,
|
| 1762 |
+
"loss": 0.6483,
|
| 1763 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1764 |
+
"rewards/chosen": -1.2271990776062012,
|
| 1765 |
+
"rewards/margins": 0.17618249356746674,
|
| 1766 |
+
"rewards/rejected": -1.4033817052841187,
|
| 1767 |
+
"step": 1120
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.9,
|
| 1771 |
+
"learning_rate": 1.370135873643097e-07,
|
| 1772 |
+
"logits/chosen": -4.33071231842041,
|
| 1773 |
+
"logits/rejected": -4.364874839782715,
|
| 1774 |
+
"logps/chosen": -403.37628173828125,
|
| 1775 |
+
"logps/rejected": -413.484375,
|
| 1776 |
+
"loss": 0.6427,
|
| 1777 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1778 |
+
"rewards/chosen": -1.2655575275421143,
|
| 1779 |
+
"rewards/margins": 0.19985264539718628,
|
| 1780 |
+
"rewards/rejected": -1.4654099941253662,
|
| 1781 |
+
"step": 1130
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.91,
|
| 1785 |
+
"learning_rate": 1.1512467784059372e-07,
|
| 1786 |
+
"logits/chosen": -4.4061384201049805,
|
| 1787 |
+
"logits/rejected": -4.369538307189941,
|
| 1788 |
+
"logps/chosen": -367.4423522949219,
|
| 1789 |
+
"logps/rejected": -402.65911865234375,
|
| 1790 |
+
"loss": 0.6186,
|
| 1791 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1792 |
+
"rewards/chosen": -1.1879181861877441,
|
| 1793 |
+
"rewards/margins": 0.27168983221054077,
|
| 1794 |
+
"rewards/rejected": -1.4596080780029297,
|
| 1795 |
+
"step": 1140
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 0.92,
|
| 1799 |
+
"learning_rate": 9.509873260376251e-08,
|
| 1800 |
+
"logits/chosen": -4.3491435050964355,
|
| 1801 |
+
"logits/rejected": -4.2928547859191895,
|
| 1802 |
+
"logps/chosen": -394.6898498535156,
|
| 1803 |
+
"logps/rejected": -461.10845947265625,
|
| 1804 |
+
"loss": 0.583,
|
| 1805 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1806 |
+
"rewards/chosen": -1.2356455326080322,
|
| 1807 |
+
"rewards/margins": 0.3976030647754669,
|
| 1808 |
+
"rewards/rejected": -1.6332485675811768,
|
| 1809 |
+
"step": 1150
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.93,
|
| 1813 |
+
"learning_rate": 7.695139507988559e-08,
|
| 1814 |
+
"logits/chosen": -4.359416961669922,
|
| 1815 |
+
"logits/rejected": -4.381959915161133,
|
| 1816 |
+
"logps/chosen": -409.65643310546875,
|
| 1817 |
+
"logps/rejected": -434.98675537109375,
|
| 1818 |
+
"loss": 0.6521,
|
| 1819 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1820 |
+
"rewards/chosen": -1.2608633041381836,
|
| 1821 |
+
"rewards/margins": 0.1726626455783844,
|
| 1822 |
+
"rewards/rejected": -1.4335259199142456,
|
| 1823 |
+
"step": 1160
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 0.94,
|
| 1827 |
+
"learning_rate": 6.069684120570684e-08,
|
| 1828 |
+
"logits/chosen": -4.381880283355713,
|
| 1829 |
+
"logits/rejected": -4.29121208190918,
|
| 1830 |
+
"logps/chosen": -395.4755554199219,
|
| 1831 |
+
"logps/rejected": -438.13427734375,
|
| 1832 |
+
"loss": 0.6418,
|
| 1833 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1834 |
+
"rewards/chosen": -1.3113107681274414,
|
| 1835 |
+
"rewards/margins": 0.23519746959209442,
|
| 1836 |
+
"rewards/rejected": -1.5465081930160522,
|
| 1837 |
+
"step": 1170
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 0.94,
|
| 1841 |
+
"learning_rate": 4.634776835499871e-08,
|
| 1842 |
+
"logits/chosen": -4.314273357391357,
|
| 1843 |
+
"logits/rejected": -4.296536445617676,
|
| 1844 |
+
"logps/chosen": -382.25958251953125,
|
| 1845 |
+
"logps/rejected": -419.88897705078125,
|
| 1846 |
+
"loss": 0.6354,
|
| 1847 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1848 |
+
"rewards/chosen": -1.2596309185028076,
|
| 1849 |
+
"rewards/margins": 0.24562938511371613,
|
| 1850 |
+
"rewards/rejected": -1.5052604675292969,
|
| 1851 |
+
"step": 1180
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 0.95,
|
| 1855 |
+
"learning_rate": 3.3915385419908964e-08,
|
| 1856 |
+
"logits/chosen": -4.292089939117432,
|
| 1857 |
+
"logits/rejected": -4.3260178565979,
|
| 1858 |
+
"logps/chosen": -391.3450927734375,
|
| 1859 |
+
"logps/rejected": -419.9061584472656,
|
| 1860 |
+
"loss": 0.6332,
|
| 1861 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1862 |
+
"rewards/chosen": -1.2074798345565796,
|
| 1863 |
+
"rewards/margins": 0.2409057915210724,
|
| 1864 |
+
"rewards/rejected": -1.4483857154846191,
|
| 1865 |
+
"step": 1190
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 0.96,
|
| 1869 |
+
"learning_rate": 2.3409404055043938e-08,
|
| 1870 |
+
"logits/chosen": -4.403561592102051,
|
| 1871 |
+
"logits/rejected": -4.3759870529174805,
|
| 1872 |
+
"logps/chosen": -400.16522216796875,
|
| 1873 |
+
"logps/rejected": -427.40606689453125,
|
| 1874 |
+
"loss": 0.6539,
|
| 1875 |
+
"rewards/accuracies": 0.606249988079071,
|
| 1876 |
+
"rewards/chosen": -1.288109302520752,
|
| 1877 |
+
"rewards/margins": 0.17616990208625793,
|
| 1878 |
+
"rewards/rejected": -1.4642791748046875,
|
| 1879 |
+
"step": 1200
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 0.96,
|
| 1883 |
+
"eval_logits/chosen": -4.559507846832275,
|
| 1884 |
+
"eval_logits/rejected": -4.548335075378418,
|
| 1885 |
+
"eval_logps/chosen": -340.22308349609375,
|
| 1886 |
+
"eval_logps/rejected": -357.4035949707031,
|
| 1887 |
+
"eval_loss": 0.6746300458908081,
|
| 1888 |
+
"eval_rewards/accuracies": 0.5730000138282776,
|
| 1889 |
+
"eval_rewards/chosen": -1.318231225013733,
|
| 1890 |
+
"eval_rewards/margins": 0.11870139837265015,
|
| 1891 |
+
"eval_rewards/rejected": -1.4369325637817383,
|
| 1892 |
+
"eval_runtime": 1563.0275,
|
| 1893 |
+
"eval_samples_per_second": 1.916,
|
| 1894 |
+
"eval_steps_per_second": 0.24,
|
| 1895 |
+
"step": 1200
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 0.97,
|
| 1899 |
+
"learning_rate": 1.4838031091134186e-08,
|
| 1900 |
+
"logits/chosen": -4.341926574707031,
|
| 1901 |
+
"logits/rejected": -4.2704925537109375,
|
| 1902 |
+
"logps/chosen": -373.17681884765625,
|
| 1903 |
+
"logps/rejected": -431.8099060058594,
|
| 1904 |
+
"loss": 0.6065,
|
| 1905 |
+
"rewards/accuracies": 0.65625,
|
| 1906 |
+
"rewards/chosen": -1.2158408164978027,
|
| 1907 |
+
"rewards/margins": 0.33925992250442505,
|
| 1908 |
+
"rewards/rejected": -1.555100679397583,
|
| 1909 |
+
"step": 1210
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 0.98,
|
| 1913 |
+
"learning_rate": 8.207962124201774e-09,
|
| 1914 |
+
"logits/chosen": -4.371233940124512,
|
| 1915 |
+
"logits/rejected": -4.343363285064697,
|
| 1916 |
+
"logps/chosen": -401.8610534667969,
|
| 1917 |
+
"logps/rejected": -437.0784606933594,
|
| 1918 |
+
"loss": 0.6005,
|
| 1919 |
+
"rewards/accuracies": 0.6875,
|
| 1920 |
+
"rewards/chosen": -1.1877615451812744,
|
| 1921 |
+
"rewards/margins": 0.29727452993392944,
|
| 1922 |
+
"rewards/rejected": -1.4850361347198486,
|
| 1923 |
+
"step": 1220
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 0.98,
|
| 1927 |
+
"learning_rate": 3.5243762852441023e-09,
|
| 1928 |
+
"logits/chosen": -4.321781158447266,
|
| 1929 |
+
"logits/rejected": -4.277449131011963,
|
| 1930 |
+
"logps/chosen": -392.8436279296875,
|
| 1931 |
+
"logps/rejected": -430.36279296875,
|
| 1932 |
+
"loss": 0.6401,
|
| 1933 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1934 |
+
"rewards/chosen": -1.2518620491027832,
|
| 1935 |
+
"rewards/margins": 0.2195155918598175,
|
| 1936 |
+
"rewards/rejected": -1.4713777303695679,
|
| 1937 |
+
"step": 1230
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 0.99,
|
| 1941 |
+
"learning_rate": 7.909321945129278e-10,
|
| 1942 |
+
"logits/chosen": -4.296388626098633,
|
| 1943 |
+
"logits/rejected": -4.257707118988037,
|
| 1944 |
+
"logps/chosen": -408.13409423828125,
|
| 1945 |
+
"logps/rejected": -451.42462158203125,
|
| 1946 |
+
"loss": 0.6106,
|
| 1947 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1948 |
+
"rewards/chosen": -1.2207772731781006,
|
| 1949 |
+
"rewards/margins": 0.29706281423568726,
|
| 1950 |
+
"rewards/rejected": -1.5178401470184326,
|
| 1951 |
+
"step": 1240
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"epoch": 1.0,
|
| 1955 |
+
"step": 1249,
|
| 1956 |
+
"total_flos": 0.0,
|
| 1957 |
+
"train_loss": 0.6508825262419409,
|
| 1958 |
+
"train_runtime": 40390.1729,
|
| 1959 |
+
"train_samples_per_second": 0.495,
|
| 1960 |
+
"train_steps_per_second": 0.031
|
| 1961 |
+
}
|
| 1962 |
+
],
|
| 1963 |
+
"logging_steps": 10,
|
| 1964 |
+
"max_steps": 1249,
|
| 1965 |
+
"num_input_tokens_seen": 0,
|
| 1966 |
+
"num_train_epochs": 1,
|
| 1967 |
+
"save_steps": 100,
|
| 1968 |
+
"total_flos": 0.0,
|
| 1969 |
+
"train_batch_size": 4,
|
| 1970 |
+
"trial_name": null,
|
| 1971 |
+
"trial_params": null
|
| 1972 |
+
}
|