Model save
Browse files- README.md +90 -0
- all_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +2577 -0
README.md
ADDED
|
@@ -0,0 +1,90 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model: mistralai/Mistral-7B-v0.1
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
library_name: peft
|
| 9 |
+
model-index:
|
| 10 |
+
- name: zephyr-7b-nca_pair-qlora-lr5e6-beta0.1
|
| 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 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/1014579852qq-tsinghua-university/huggingface/runs/p6zmvlxg)
|
| 18 |
+
# zephyr-7b-nca_pair-qlora-lr5e6-beta0.1
|
| 19 |
+
|
| 20 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on an unknown dataset.
|
| 21 |
+
It achieves the following results on the evaluation set:
|
| 22 |
+
- Loss: 1.3409
|
| 23 |
+
- Rewards/chosen: 0.1433
|
| 24 |
+
- Rewards/rejected: -0.3268
|
| 25 |
+
- Rewards/accuracies: 0.7475
|
| 26 |
+
- Rewards/margins: 0.4701
|
| 27 |
+
- Logps/rejected: -257.3707
|
| 28 |
+
- Logps/chosen: -270.7436
|
| 29 |
+
- Logits/rejected: -2.2339
|
| 30 |
+
- Logits/chosen: -2.3137
|
| 31 |
+
|
| 32 |
+
## Model description
|
| 33 |
+
|
| 34 |
+
More information needed
|
| 35 |
+
|
| 36 |
+
## Intended uses & limitations
|
| 37 |
+
|
| 38 |
+
More information needed
|
| 39 |
+
|
| 40 |
+
## Training and evaluation data
|
| 41 |
+
|
| 42 |
+
More information needed
|
| 43 |
+
|
| 44 |
+
## Training procedure
|
| 45 |
+
|
| 46 |
+
### Training hyperparameters
|
| 47 |
+
|
| 48 |
+
The following hyperparameters were used during training:
|
| 49 |
+
- learning_rate: 5e-06
|
| 50 |
+
- train_batch_size: 2
|
| 51 |
+
- eval_batch_size: 4
|
| 52 |
+
- seed: 42
|
| 53 |
+
- distributed_type: multi-GPU
|
| 54 |
+
- num_devices: 5
|
| 55 |
+
- gradient_accumulation_steps: 4
|
| 56 |
+
- total_train_batch_size: 40
|
| 57 |
+
- total_eval_batch_size: 20
|
| 58 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 59 |
+
- lr_scheduler_type: cosine
|
| 60 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 61 |
+
- num_epochs: 1
|
| 62 |
+
|
| 63 |
+
### Training results
|
| 64 |
+
|
| 65 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 66 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 67 |
+
| 1.3652 | 0.0654 | 100 | 1.3585 | 0.0179 | -0.2204 | 0.7200 | 0.2383 | -256.3067 | -271.9973 | -2.2040 | -2.2872 |
|
| 68 |
+
| 1.3457 | 0.1308 | 200 | 1.3529 | 0.1799 | -0.2015 | 0.7425 | 0.3814 | -256.1176 | -270.3774 | -2.2080 | -2.2912 |
|
| 69 |
+
| 1.3328 | 0.1963 | 300 | 1.3500 | 0.1269 | -0.2919 | 0.7150 | 0.4188 | -257.0218 | -270.9071 | -2.2303 | -2.3106 |
|
| 70 |
+
| 1.3452 | 0.2617 | 400 | 1.3536 | 0.1854 | -0.2395 | 0.7200 | 0.4249 | -256.4976 | -270.3225 | -2.2250 | -2.3062 |
|
| 71 |
+
| 1.3446 | 0.3271 | 500 | 1.3501 | 0.0859 | -0.3936 | 0.7275 | 0.4795 | -258.0389 | -271.3175 | -2.1984 | -2.2818 |
|
| 72 |
+
| 1.333 | 0.3925 | 600 | 1.3496 | 0.0493 | -0.3851 | 0.7450 | 0.4344 | -257.9544 | -271.6837 | -2.2107 | -2.2937 |
|
| 73 |
+
| 1.3577 | 0.4580 | 700 | 1.3457 | 0.1306 | -0.2688 | 0.7175 | 0.3994 | -256.7908 | -270.8706 | -2.2100 | -2.2934 |
|
| 74 |
+
| 1.343 | 0.5234 | 800 | 1.3449 | 0.0814 | -0.3810 | 0.7150 | 0.4623 | -257.9127 | -271.3629 | -2.2312 | -2.3121 |
|
| 75 |
+
| 1.3439 | 0.5888 | 900 | 1.3459 | 0.0385 | -0.4054 | 0.7250 | 0.4439 | -258.1573 | -271.7917 | -2.2327 | -2.3137 |
|
| 76 |
+
| 1.3388 | 0.6542 | 1000 | 1.3442 | 0.2150 | -0.2625 | 0.7325 | 0.4775 | -256.7277 | -270.0262 | -2.2387 | -2.3183 |
|
| 77 |
+
| 1.3186 | 0.7197 | 1100 | 1.3423 | 0.1242 | -0.3587 | 0.7325 | 0.4829 | -257.6895 | -270.9345 | -2.2306 | -2.3107 |
|
| 78 |
+
| 1.3299 | 0.7851 | 1200 | 1.3417 | 0.1468 | -0.3270 | 0.7425 | 0.4737 | -257.3728 | -270.7089 | -2.2275 | -2.3078 |
|
| 79 |
+
| 1.3248 | 0.8505 | 1300 | 1.3413 | 0.1555 | -0.3132 | 0.7525 | 0.4687 | -257.2347 | -270.6216 | -2.2306 | -2.3105 |
|
| 80 |
+
| 1.3398 | 0.9159 | 1400 | 1.3414 | 0.1409 | -0.3251 | 0.7475 | 0.4660 | -257.3535 | -270.7675 | -2.2317 | -2.3117 |
|
| 81 |
+
| 1.325 | 0.9814 | 1500 | 1.3409 | 0.1433 | -0.3268 | 0.7475 | 0.4701 | -257.3707 | -270.7436 | -2.2339 | -2.3137 |
|
| 82 |
+
|
| 83 |
+
|
| 84 |
+
### Framework versions
|
| 85 |
+
|
| 86 |
+
- PEFT 0.10.0
|
| 87 |
+
- Transformers 4.43.1
|
| 88 |
+
- Pytorch 2.1.2+cu121
|
| 89 |
+
- Datasets 2.18.0
|
| 90 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9996728819103696,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 1.343712306771603,
|
| 5 |
+
"train_runtime": 13958.9948,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 4.38,
|
| 8 |
+
"train_steps_per_second": 0.109
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9996728819103696,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 1.343712306771603,
|
| 5 |
+
"train_runtime": 13958.9948,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 4.38,
|
| 8 |
+
"train_steps_per_second": 0.109
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2577 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9996728819103696,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 1528,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0006542361792607131,
|
| 13 |
+
"grad_norm": 7.53125,
|
| 14 |
+
"learning_rate": 3.267973856209151e-08,
|
| 15 |
+
"logits/chosen": -2.690979480743408,
|
| 16 |
+
"logits/rejected": -2.4915528297424316,
|
| 17 |
+
"logps/chosen": -306.9772644042969,
|
| 18 |
+
"logps/rejected": -274.77850341796875,
|
| 19 |
+
"loss": 1.3863,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.006542361792607131,
|
| 28 |
+
"grad_norm": 6.75,
|
| 29 |
+
"learning_rate": 3.267973856209151e-07,
|
| 30 |
+
"logits/chosen": -2.5390172004699707,
|
| 31 |
+
"logits/rejected": -2.4508614540100098,
|
| 32 |
+
"logps/chosen": -286.64813232421875,
|
| 33 |
+
"logps/rejected": -271.6336669921875,
|
| 34 |
+
"loss": 1.3872,
|
| 35 |
+
"rewards/accuracies": 0.4305555522441864,
|
| 36 |
+
"rewards/chosen": 0.002395547926425934,
|
| 37 |
+
"rewards/margins": 0.0021986099891364574,
|
| 38 |
+
"rewards/rejected": 0.0001969372679013759,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.013084723585214262,
|
| 43 |
+
"grad_norm": 7.4375,
|
| 44 |
+
"learning_rate": 6.535947712418302e-07,
|
| 45 |
+
"logits/chosen": -2.479840040206909,
|
| 46 |
+
"logits/rejected": -2.3548035621643066,
|
| 47 |
+
"logps/chosen": -268.228759765625,
|
| 48 |
+
"logps/rejected": -197.14610290527344,
|
| 49 |
+
"loss": 1.3853,
|
| 50 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 51 |
+
"rewards/chosen": 0.003464195877313614,
|
| 52 |
+
"rewards/margins": -0.0019345780601724982,
|
| 53 |
+
"rewards/rejected": 0.005398774053901434,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.019627085377821395,
|
| 58 |
+
"grad_norm": 8.4375,
|
| 59 |
+
"learning_rate": 9.80392156862745e-07,
|
| 60 |
+
"logits/chosen": -2.4206929206848145,
|
| 61 |
+
"logits/rejected": -2.365497350692749,
|
| 62 |
+
"logps/chosen": -258.31414794921875,
|
| 63 |
+
"logps/rejected": -255.288818359375,
|
| 64 |
+
"loss": 1.3826,
|
| 65 |
+
"rewards/accuracies": 0.637499988079071,
|
| 66 |
+
"rewards/chosen": 0.025201931595802307,
|
| 67 |
+
"rewards/margins": 0.022868353873491287,
|
| 68 |
+
"rewards/rejected": 0.0023335753940045834,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.026169447170428524,
|
| 73 |
+
"grad_norm": 7.09375,
|
| 74 |
+
"learning_rate": 1.3071895424836604e-06,
|
| 75 |
+
"logits/chosen": -2.370481252670288,
|
| 76 |
+
"logits/rejected": -2.3122944831848145,
|
| 77 |
+
"logps/chosen": -273.23687744140625,
|
| 78 |
+
"logps/rejected": -237.98342895507812,
|
| 79 |
+
"loss": 1.3774,
|
| 80 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 81 |
+
"rewards/chosen": 0.017533209174871445,
|
| 82 |
+
"rewards/margins": 0.031263187527656555,
|
| 83 |
+
"rewards/rejected": -0.01372998021543026,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.03271180896303565,
|
| 88 |
+
"grad_norm": 7.25,
|
| 89 |
+
"learning_rate": 1.6339869281045753e-06,
|
| 90 |
+
"logits/chosen": -2.4228413105010986,
|
| 91 |
+
"logits/rejected": -2.365399122238159,
|
| 92 |
+
"logps/chosen": -237.292724609375,
|
| 93 |
+
"logps/rejected": -250.16238403320312,
|
| 94 |
+
"loss": 1.3682,
|
| 95 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 96 |
+
"rewards/chosen": 0.0359283983707428,
|
| 97 |
+
"rewards/margins": 0.08779667317867279,
|
| 98 |
+
"rewards/rejected": -0.0518682599067688,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.03925417075564279,
|
| 103 |
+
"grad_norm": 6.84375,
|
| 104 |
+
"learning_rate": 1.96078431372549e-06,
|
| 105 |
+
"logits/chosen": -2.5553746223449707,
|
| 106 |
+
"logits/rejected": -2.249553680419922,
|
| 107 |
+
"logps/chosen": -291.7508239746094,
|
| 108 |
+
"logps/rejected": -209.22256469726562,
|
| 109 |
+
"loss": 1.3681,
|
| 110 |
+
"rewards/accuracies": 0.6875,
|
| 111 |
+
"rewards/chosen": 0.006251047365367413,
|
| 112 |
+
"rewards/margins": 0.1306430548429489,
|
| 113 |
+
"rewards/rejected": -0.12439201027154922,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.04579653254824992,
|
| 118 |
+
"grad_norm": 7.15625,
|
| 119 |
+
"learning_rate": 2.2875816993464053e-06,
|
| 120 |
+
"logits/chosen": -2.4841151237487793,
|
| 121 |
+
"logits/rejected": -2.4571692943573,
|
| 122 |
+
"logps/chosen": -233.838623046875,
|
| 123 |
+
"logps/rejected": -244.6779022216797,
|
| 124 |
+
"loss": 1.3704,
|
| 125 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 126 |
+
"rewards/chosen": 0.05333448573946953,
|
| 127 |
+
"rewards/margins": 0.15633264183998108,
|
| 128 |
+
"rewards/rejected": -0.10299815982580185,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.05233889434085705,
|
| 133 |
+
"grad_norm": 7.40625,
|
| 134 |
+
"learning_rate": 2.6143790849673208e-06,
|
| 135 |
+
"logits/chosen": -2.369415760040283,
|
| 136 |
+
"logits/rejected": -2.2743725776672363,
|
| 137 |
+
"logps/chosen": -234.2888946533203,
|
| 138 |
+
"logps/rejected": -199.08114624023438,
|
| 139 |
+
"loss": 1.3623,
|
| 140 |
+
"rewards/accuracies": 0.637499988079071,
|
| 141 |
+
"rewards/chosen": 0.011479836888611317,
|
| 142 |
+
"rewards/margins": 0.14200374484062195,
|
| 143 |
+
"rewards/rejected": -0.1305239051580429,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.058881256133464184,
|
| 148 |
+
"grad_norm": 8.25,
|
| 149 |
+
"learning_rate": 2.9411764705882355e-06,
|
| 150 |
+
"logits/chosen": -2.3975484371185303,
|
| 151 |
+
"logits/rejected": -2.3238232135772705,
|
| 152 |
+
"logps/chosen": -267.9956359863281,
|
| 153 |
+
"logps/rejected": -265.9745178222656,
|
| 154 |
+
"loss": 1.3549,
|
| 155 |
+
"rewards/accuracies": 0.762499988079071,
|
| 156 |
+
"rewards/chosen": 0.1009618490934372,
|
| 157 |
+
"rewards/margins": 0.2633386254310608,
|
| 158 |
+
"rewards/rejected": -0.16237673163414001,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.0654236179260713,
|
| 163 |
+
"grad_norm": 6.96875,
|
| 164 |
+
"learning_rate": 3.2679738562091506e-06,
|
| 165 |
+
"logits/chosen": -2.496324062347412,
|
| 166 |
+
"logits/rejected": -2.4217727184295654,
|
| 167 |
+
"logps/chosen": -258.4588317871094,
|
| 168 |
+
"logps/rejected": -248.0790557861328,
|
| 169 |
+
"loss": 1.3652,
|
| 170 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 171 |
+
"rewards/chosen": 0.04371471330523491,
|
| 172 |
+
"rewards/margins": 0.1967853307723999,
|
| 173 |
+
"rewards/rejected": -0.1530705988407135,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.0654236179260713,
|
| 178 |
+
"eval_logits/chosen": -2.2871594429016113,
|
| 179 |
+
"eval_logits/rejected": -2.2039709091186523,
|
| 180 |
+
"eval_logps/chosen": -271.9973449707031,
|
| 181 |
+
"eval_logps/rejected": -256.3067321777344,
|
| 182 |
+
"eval_loss": 1.3585007190704346,
|
| 183 |
+
"eval_rewards/accuracies": 0.7200000286102295,
|
| 184 |
+
"eval_rewards/chosen": 0.01791691593825817,
|
| 185 |
+
"eval_rewards/margins": 0.23829618096351624,
|
| 186 |
+
"eval_rewards/rejected": -0.2203792780637741,
|
| 187 |
+
"eval_runtime": 193.0612,
|
| 188 |
+
"eval_samples_per_second": 10.359,
|
| 189 |
+
"eval_steps_per_second": 0.518,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.07196597971867845,
|
| 194 |
+
"grad_norm": 7.78125,
|
| 195 |
+
"learning_rate": 3.5947712418300657e-06,
|
| 196 |
+
"logits/chosen": -2.355022668838501,
|
| 197 |
+
"logits/rejected": -2.3289151191711426,
|
| 198 |
+
"logps/chosen": -282.1571960449219,
|
| 199 |
+
"logps/rejected": -228.05319213867188,
|
| 200 |
+
"loss": 1.3483,
|
| 201 |
+
"rewards/accuracies": 0.699999988079071,
|
| 202 |
+
"rewards/chosen": 0.0913485437631607,
|
| 203 |
+
"rewards/margins": 0.3209895193576813,
|
| 204 |
+
"rewards/rejected": -0.229640930891037,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.07850834151128558,
|
| 209 |
+
"grad_norm": 8.4375,
|
| 210 |
+
"learning_rate": 3.92156862745098e-06,
|
| 211 |
+
"logits/chosen": -2.3910622596740723,
|
| 212 |
+
"logits/rejected": -2.2716870307922363,
|
| 213 |
+
"logps/chosen": -274.3619079589844,
|
| 214 |
+
"logps/rejected": -280.36614990234375,
|
| 215 |
+
"loss": 1.3505,
|
| 216 |
+
"rewards/accuracies": 0.800000011920929,
|
| 217 |
+
"rewards/chosen": 0.11332492530345917,
|
| 218 |
+
"rewards/margins": 0.39289209246635437,
|
| 219 |
+
"rewards/rejected": -0.279567152261734,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.08505070330389271,
|
| 224 |
+
"grad_norm": 6.65625,
|
| 225 |
+
"learning_rate": 4.2483660130718954e-06,
|
| 226 |
+
"logits/chosen": -2.4673023223876953,
|
| 227 |
+
"logits/rejected": -2.3496053218841553,
|
| 228 |
+
"logps/chosen": -233.6593017578125,
|
| 229 |
+
"logps/rejected": -225.4088134765625,
|
| 230 |
+
"loss": 1.3577,
|
| 231 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 232 |
+
"rewards/chosen": 0.04373621195554733,
|
| 233 |
+
"rewards/margins": 0.27300602197647095,
|
| 234 |
+
"rewards/rejected": -0.22926978766918182,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.09159306509649984,
|
| 239 |
+
"grad_norm": 8.8125,
|
| 240 |
+
"learning_rate": 4.5751633986928105e-06,
|
| 241 |
+
"logits/chosen": -2.488830804824829,
|
| 242 |
+
"logits/rejected": -2.363779067993164,
|
| 243 |
+
"logps/chosen": -295.23980712890625,
|
| 244 |
+
"logps/rejected": -280.0679016113281,
|
| 245 |
+
"loss": 1.3551,
|
| 246 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 247 |
+
"rewards/chosen": 0.03703129664063454,
|
| 248 |
+
"rewards/margins": 0.1885422170162201,
|
| 249 |
+
"rewards/rejected": -0.15151092410087585,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.09813542688910697,
|
| 254 |
+
"grad_norm": 7.21875,
|
| 255 |
+
"learning_rate": 4.901960784313726e-06,
|
| 256 |
+
"logits/chosen": -2.4635255336761475,
|
| 257 |
+
"logits/rejected": -2.3997857570648193,
|
| 258 |
+
"logps/chosen": -271.8360900878906,
|
| 259 |
+
"logps/rejected": -239.2571563720703,
|
| 260 |
+
"loss": 1.3443,
|
| 261 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 262 |
+
"rewards/chosen": 0.13082675635814667,
|
| 263 |
+
"rewards/margins": 0.38594740629196167,
|
| 264 |
+
"rewards/rejected": -0.2551206648349762,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.1046777886817141,
|
| 269 |
+
"grad_norm": 8.1875,
|
| 270 |
+
"learning_rate": 4.999680264259825e-06,
|
| 271 |
+
"logits/chosen": -2.5628502368927,
|
| 272 |
+
"logits/rejected": -2.309502601623535,
|
| 273 |
+
"logps/chosen": -289.554931640625,
|
| 274 |
+
"logps/rejected": -228.1037139892578,
|
| 275 |
+
"loss": 1.3507,
|
| 276 |
+
"rewards/accuracies": 0.6875,
|
| 277 |
+
"rewards/chosen": 0.056355126202106476,
|
| 278 |
+
"rewards/margins": 0.30590537190437317,
|
| 279 |
+
"rewards/rejected": -0.2495502233505249,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.11122015047432122,
|
| 284 |
+
"grad_norm": 24.125,
|
| 285 |
+
"learning_rate": 4.998114408534616e-06,
|
| 286 |
+
"logits/chosen": -2.459836483001709,
|
| 287 |
+
"logits/rejected": -2.241123676300049,
|
| 288 |
+
"logps/chosen": -269.2346496582031,
|
| 289 |
+
"logps/rejected": -245.09005737304688,
|
| 290 |
+
"loss": 1.3566,
|
| 291 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 292 |
+
"rewards/chosen": 0.11544966697692871,
|
| 293 |
+
"rewards/margins": 0.3507261276245117,
|
| 294 |
+
"rewards/rejected": -0.2352764904499054,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.11776251226692837,
|
| 299 |
+
"grad_norm": 7.28125,
|
| 300 |
+
"learning_rate": 4.995244522215781e-06,
|
| 301 |
+
"logits/chosen": -2.479318857192993,
|
| 302 |
+
"logits/rejected": -2.452930450439453,
|
| 303 |
+
"logps/chosen": -253.64907836914062,
|
| 304 |
+
"logps/rejected": -255.054443359375,
|
| 305 |
+
"loss": 1.3515,
|
| 306 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 307 |
+
"rewards/chosen": 0.0358736589550972,
|
| 308 |
+
"rewards/margins": 0.2104395180940628,
|
| 309 |
+
"rewards/rejected": -0.174565851688385,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.1243048740595355,
|
| 314 |
+
"grad_norm": 6.78125,
|
| 315 |
+
"learning_rate": 4.9910721034010655e-06,
|
| 316 |
+
"logits/chosen": -2.4955029487609863,
|
| 317 |
+
"logits/rejected": -2.3607935905456543,
|
| 318 |
+
"logps/chosen": -259.48529052734375,
|
| 319 |
+
"logps/rejected": -232.8231964111328,
|
| 320 |
+
"loss": 1.3516,
|
| 321 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 322 |
+
"rewards/chosen": 0.06281731277704239,
|
| 323 |
+
"rewards/margins": 0.3066301941871643,
|
| 324 |
+
"rewards/rejected": -0.24381284415721893,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.1308472358521426,
|
| 329 |
+
"grad_norm": 7.90625,
|
| 330 |
+
"learning_rate": 4.985599330117931e-06,
|
| 331 |
+
"logits/chosen": -2.4151864051818848,
|
| 332 |
+
"logits/rejected": -2.3472375869750977,
|
| 333 |
+
"logps/chosen": -251.25210571289062,
|
| 334 |
+
"logps/rejected": -235.2193603515625,
|
| 335 |
+
"loss": 1.3457,
|
| 336 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 337 |
+
"rewards/chosen": 0.0700979083776474,
|
| 338 |
+
"rewards/margins": 0.3274100422859192,
|
| 339 |
+
"rewards/rejected": -0.25731217861175537,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.1308472358521426,
|
| 344 |
+
"eval_logits/chosen": -2.2911951541900635,
|
| 345 |
+
"eval_logits/rejected": -2.2080154418945312,
|
| 346 |
+
"eval_logps/chosen": -270.37738037109375,
|
| 347 |
+
"eval_logps/rejected": -256.1176452636719,
|
| 348 |
+
"eval_loss": 1.3529404401779175,
|
| 349 |
+
"eval_rewards/accuracies": 0.7425000071525574,
|
| 350 |
+
"eval_rewards/chosen": 0.17991353571414948,
|
| 351 |
+
"eval_rewards/margins": 0.3813881278038025,
|
| 352 |
+
"eval_rewards/rejected": -0.20147459208965302,
|
| 353 |
+
"eval_runtime": 192.5507,
|
| 354 |
+
"eval_samples_per_second": 10.387,
|
| 355 |
+
"eval_steps_per_second": 0.519,
|
| 356 |
+
"step": 200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.13738959764474976,
|
| 360 |
+
"grad_norm": 6.9375,
|
| 361 |
+
"learning_rate": 4.978829059186611e-06,
|
| 362 |
+
"logits/chosen": -2.512892246246338,
|
| 363 |
+
"logits/rejected": -2.4710421562194824,
|
| 364 |
+
"logps/chosen": -278.5118713378906,
|
| 365 |
+
"logps/rejected": -281.03924560546875,
|
| 366 |
+
"loss": 1.3592,
|
| 367 |
+
"rewards/accuracies": 0.737500011920929,
|
| 368 |
+
"rewards/chosen": 0.01397608406841755,
|
| 369 |
+
"rewards/margins": 0.33375468850135803,
|
| 370 |
+
"rewards/rejected": -0.31977859139442444,
|
| 371 |
+
"step": 210
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.1439319594373569,
|
| 375 |
+
"grad_norm": 7.8125,
|
| 376 |
+
"learning_rate": 4.97076482472884e-06,
|
| 377 |
+
"logits/chosen": -2.3786368370056152,
|
| 378 |
+
"logits/rejected": -2.4024887084960938,
|
| 379 |
+
"logps/chosen": -307.94610595703125,
|
| 380 |
+
"logps/rejected": -310.7808532714844,
|
| 381 |
+
"loss": 1.3364,
|
| 382 |
+
"rewards/accuracies": 0.762499988079071,
|
| 383 |
+
"rewards/chosen": 0.07198480516672134,
|
| 384 |
+
"rewards/margins": 0.43232816457748413,
|
| 385 |
+
"rewards/rejected": -0.3603433668613434,
|
| 386 |
+
"step": 220
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.15047432122996401,
|
| 390 |
+
"grad_norm": 7.96875,
|
| 391 |
+
"learning_rate": 4.961410836323014e-06,
|
| 392 |
+
"logits/chosen": -2.5079922676086426,
|
| 393 |
+
"logits/rejected": -2.3904035091400146,
|
| 394 |
+
"logps/chosen": -292.0177917480469,
|
| 395 |
+
"logps/rejected": -243.524658203125,
|
| 396 |
+
"loss": 1.3466,
|
| 397 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 398 |
+
"rewards/chosen": 0.0814410150051117,
|
| 399 |
+
"rewards/margins": 0.4043430685997009,
|
| 400 |
+
"rewards/rejected": -0.322902113199234,
|
| 401 |
+
"step": 230
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.15701668302257116,
|
| 405 |
+
"grad_norm": 6.75,
|
| 406 |
+
"learning_rate": 4.950771976806769e-06,
|
| 407 |
+
"logits/chosen": -2.643059730529785,
|
| 408 |
+
"logits/rejected": -2.447086811065674,
|
| 409 |
+
"logps/chosen": -279.7219543457031,
|
| 410 |
+
"logps/rejected": -228.05068969726562,
|
| 411 |
+
"loss": 1.3531,
|
| 412 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 413 |
+
"rewards/chosen": 0.1486392319202423,
|
| 414 |
+
"rewards/margins": 0.4861406683921814,
|
| 415 |
+
"rewards/rejected": -0.3375014662742615,
|
| 416 |
+
"step": 240
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.16355904481517827,
|
| 420 |
+
"grad_norm": 7.25,
|
| 421 |
+
"learning_rate": 4.938853799728112e-06,
|
| 422 |
+
"logits/chosen": -2.533738613128662,
|
| 423 |
+
"logits/rejected": -2.347072124481201,
|
| 424 |
+
"logps/chosen": -291.59527587890625,
|
| 425 |
+
"logps/rejected": -224.52340698242188,
|
| 426 |
+
"loss": 1.3444,
|
| 427 |
+
"rewards/accuracies": 0.737500011920929,
|
| 428 |
+
"rewards/chosen": 0.14931416511535645,
|
| 429 |
+
"rewards/margins": 0.5157926082611084,
|
| 430 |
+
"rewards/rejected": -0.36647850275039673,
|
| 431 |
+
"step": 250
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.17010140660778542,
|
| 435 |
+
"grad_norm": 7.96875,
|
| 436 |
+
"learning_rate": 4.925662526446431e-06,
|
| 437 |
+
"logits/chosen": -2.4311368465423584,
|
| 438 |
+
"logits/rejected": -2.3289899826049805,
|
| 439 |
+
"logps/chosen": -253.0860137939453,
|
| 440 |
+
"logps/rejected": -219.7169952392578,
|
| 441 |
+
"loss": 1.3418,
|
| 442 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 443 |
+
"rewards/chosen": 0.03077099286019802,
|
| 444 |
+
"rewards/margins": 0.30891746282577515,
|
| 445 |
+
"rewards/rejected": -0.27814650535583496,
|
| 446 |
+
"step": 260
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.17664376840039253,
|
| 450 |
+
"grad_norm": 6.90625,
|
| 451 |
+
"learning_rate": 4.911205042884912e-06,
|
| 452 |
+
"logits/chosen": -2.497413158416748,
|
| 453 |
+
"logits/rejected": -2.3586812019348145,
|
| 454 |
+
"logps/chosen": -294.7901916503906,
|
| 455 |
+
"logps/rejected": -242.64291381835938,
|
| 456 |
+
"loss": 1.3551,
|
| 457 |
+
"rewards/accuracies": 0.625,
|
| 458 |
+
"rewards/chosen": 0.0127085717394948,
|
| 459 |
+
"rewards/margins": 0.21187984943389893,
|
| 460 |
+
"rewards/rejected": -0.1991712599992752,
|
| 461 |
+
"step": 270
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.18318613019299967,
|
| 465 |
+
"grad_norm": 7.15625,
|
| 466 |
+
"learning_rate": 4.895488895936047e-06,
|
| 467 |
+
"logits/chosen": -2.4523603916168213,
|
| 468 |
+
"logits/rejected": -2.4156365394592285,
|
| 469 |
+
"logps/chosen": -240.6490020751953,
|
| 470 |
+
"logps/rejected": -219.9276580810547,
|
| 471 |
+
"loss": 1.3329,
|
| 472 |
+
"rewards/accuracies": 0.762499988079071,
|
| 473 |
+
"rewards/chosen": 0.1077442541718483,
|
| 474 |
+
"rewards/margins": 0.4334333539009094,
|
| 475 |
+
"rewards/rejected": -0.32568907737731934,
|
| 476 |
+
"step": 280
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.18972849198560682,
|
| 480 |
+
"grad_norm": 10.0,
|
| 481 |
+
"learning_rate": 4.8785222895221075e-06,
|
| 482 |
+
"logits/chosen": -2.671842336654663,
|
| 483 |
+
"logits/rejected": -2.547651767730713,
|
| 484 |
+
"logps/chosen": -322.6619567871094,
|
| 485 |
+
"logps/rejected": -319.7884216308594,
|
| 486 |
+
"loss": 1.3476,
|
| 487 |
+
"rewards/accuracies": 0.6875,
|
| 488 |
+
"rewards/chosen": 0.07628868520259857,
|
| 489 |
+
"rewards/margins": 0.34415873885154724,
|
| 490 |
+
"rewards/rejected": -0.26787006855010986,
|
| 491 |
+
"step": 290
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.19627085377821393,
|
| 495 |
+
"grad_norm": 7.8125,
|
| 496 |
+
"learning_rate": 4.860314080312651e-06,
|
| 497 |
+
"logits/chosen": -2.5084714889526367,
|
| 498 |
+
"logits/rejected": -2.4003777503967285,
|
| 499 |
+
"logps/chosen": -274.4216003417969,
|
| 500 |
+
"logps/rejected": -254.49710083007812,
|
| 501 |
+
"loss": 1.3328,
|
| 502 |
+
"rewards/accuracies": 0.737500011920929,
|
| 503 |
+
"rewards/chosen": 0.07708136737346649,
|
| 504 |
+
"rewards/margins": 0.42221707105636597,
|
| 505 |
+
"rewards/rejected": -0.3451356589794159,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.19627085377821393,
|
| 510 |
+
"eval_logits/chosen": -2.3106164932250977,
|
| 511 |
+
"eval_logits/rejected": -2.230290412902832,
|
| 512 |
+
"eval_logps/chosen": -270.9070739746094,
|
| 513 |
+
"eval_logps/rejected": -257.02178955078125,
|
| 514 |
+
"eval_loss": 1.3500434160232544,
|
| 515 |
+
"eval_rewards/accuracies": 0.7149999737739563,
|
| 516 |
+
"eval_rewards/chosen": 0.12694190442562103,
|
| 517 |
+
"eval_rewards/margins": 0.41882869601249695,
|
| 518 |
+
"eval_rewards/rejected": -0.2918868362903595,
|
| 519 |
+
"eval_runtime": 192.4649,
|
| 520 |
+
"eval_samples_per_second": 10.392,
|
| 521 |
+
"eval_steps_per_second": 0.52,
|
| 522 |
+
"step": 300
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.20281321557082108,
|
| 526 |
+
"grad_norm": 8.125,
|
| 527 |
+
"learning_rate": 4.840873773101287e-06,
|
| 528 |
+
"logits/chosen": -2.4888222217559814,
|
| 529 |
+
"logits/rejected": -2.4138476848602295,
|
| 530 |
+
"logps/chosen": -264.6638488769531,
|
| 531 |
+
"logps/rejected": -239.2753448486328,
|
| 532 |
+
"loss": 1.3448,
|
| 533 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 534 |
+
"rewards/chosen": 0.10051695257425308,
|
| 535 |
+
"rewards/margins": 0.5004209280014038,
|
| 536 |
+
"rewards/rejected": -0.39990395307540894,
|
| 537 |
+
"step": 310
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.2093555773634282,
|
| 541 |
+
"grad_norm": 7.6875,
|
| 542 |
+
"learning_rate": 4.820211515844116e-06,
|
| 543 |
+
"logits/chosen": -2.500939130783081,
|
| 544 |
+
"logits/rejected": -2.4757304191589355,
|
| 545 |
+
"logps/chosen": -203.4287109375,
|
| 546 |
+
"logps/rejected": -218.0804901123047,
|
| 547 |
+
"loss": 1.3531,
|
| 548 |
+
"rewards/accuracies": 0.675000011920929,
|
| 549 |
+
"rewards/chosen": 0.10015901178121567,
|
| 550 |
+
"rewards/margins": 0.3148805797100067,
|
| 551 |
+
"rewards/rejected": -0.21472156047821045,
|
| 552 |
+
"step": 320
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.21589793915603533,
|
| 556 |
+
"grad_norm": 7.96875,
|
| 557 |
+
"learning_rate": 4.798338094362439e-06,
|
| 558 |
+
"logits/chosen": -2.477666139602661,
|
| 559 |
+
"logits/rejected": -2.5014617443084717,
|
| 560 |
+
"logps/chosen": -288.17950439453125,
|
| 561 |
+
"logps/rejected": -281.34295654296875,
|
| 562 |
+
"loss": 1.3666,
|
| 563 |
+
"rewards/accuracies": 0.675000011920929,
|
| 564 |
+
"rewards/chosen": 0.07362223416566849,
|
| 565 |
+
"rewards/margins": 0.274558961391449,
|
| 566 |
+
"rewards/rejected": -0.2009367197751999,
|
| 567 |
+
"step": 330
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.22244030094864245,
|
| 571 |
+
"grad_norm": 6.75,
|
| 572 |
+
"learning_rate": 4.775264926712489e-06,
|
| 573 |
+
"logits/chosen": -2.4466490745544434,
|
| 574 |
+
"logits/rejected": -2.4048807621002197,
|
| 575 |
+
"logps/chosen": -246.18392944335938,
|
| 576 |
+
"logps/rejected": -259.27392578125,
|
| 577 |
+
"loss": 1.3409,
|
| 578 |
+
"rewards/accuracies": 0.762499988079071,
|
| 579 |
+
"rewards/chosen": 0.11778600513935089,
|
| 580 |
+
"rewards/margins": 0.4433900713920593,
|
| 581 |
+
"rewards/rejected": -0.32560408115386963,
|
| 582 |
+
"step": 340
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.2289826627412496,
|
| 586 |
+
"grad_norm": 6.5,
|
| 587 |
+
"learning_rate": 4.751004057225147e-06,
|
| 588 |
+
"logits/chosen": -2.4351608753204346,
|
| 589 |
+
"logits/rejected": -2.3736894130706787,
|
| 590 |
+
"logps/chosen": -275.3951721191406,
|
| 591 |
+
"logps/rejected": -281.2572021484375,
|
| 592 |
+
"loss": 1.344,
|
| 593 |
+
"rewards/accuracies": 0.762499988079071,
|
| 594 |
+
"rewards/chosen": 0.12299034744501114,
|
| 595 |
+
"rewards/margins": 0.42316460609436035,
|
| 596 |
+
"rewards/rejected": -0.3001742362976074,
|
| 597 |
+
"step": 350
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.23552502453385674,
|
| 601 |
+
"grad_norm": 7.875,
|
| 602 |
+
"learning_rate": 4.725568150218719e-06,
|
| 603 |
+
"logits/chosen": -2.353898763656616,
|
| 604 |
+
"logits/rejected": -2.3495678901672363,
|
| 605 |
+
"logps/chosen": -247.7540283203125,
|
| 606 |
+
"logps/rejected": -262.0528564453125,
|
| 607 |
+
"loss": 1.339,
|
| 608 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 609 |
+
"rewards/chosen": 0.10569906234741211,
|
| 610 |
+
"rewards/margins": 0.41347068548202515,
|
| 611 |
+
"rewards/rejected": -0.3077716529369354,
|
| 612 |
+
"step": 360
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.24206738632646385,
|
| 616 |
+
"grad_norm": 8.6875,
|
| 617 |
+
"learning_rate": 4.6989704833880936e-06,
|
| 618 |
+
"logits/chosen": -2.455857515335083,
|
| 619 |
+
"logits/rejected": -2.2684264183044434,
|
| 620 |
+
"logps/chosen": -278.91278076171875,
|
| 621 |
+
"logps/rejected": -248.13729858398438,
|
| 622 |
+
"loss": 1.3587,
|
| 623 |
+
"rewards/accuracies": 0.637499988079071,
|
| 624 |
+
"rewards/chosen": 0.05706097558140755,
|
| 625 |
+
"rewards/margins": 0.2867721915245056,
|
| 626 |
+
"rewards/rejected": -0.22971120476722717,
|
| 627 |
+
"step": 370
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.248609748119071,
|
| 631 |
+
"grad_norm": 7.5,
|
| 632 |
+
"learning_rate": 4.671224940873704e-06,
|
| 633 |
+
"logits/chosen": -2.5477356910705566,
|
| 634 |
+
"logits/rejected": -2.4833083152770996,
|
| 635 |
+
"logps/chosen": -284.73394775390625,
|
| 636 |
+
"logps/rejected": -268.12493896484375,
|
| 637 |
+
"loss": 1.3407,
|
| 638 |
+
"rewards/accuracies": 0.762499988079071,
|
| 639 |
+
"rewards/chosen": 0.14602813124656677,
|
| 640 |
+
"rewards/margins": 0.4975086748600006,
|
| 641 |
+
"rewards/rejected": -0.35148054361343384,
|
| 642 |
+
"step": 380
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.25515210991167814,
|
| 646 |
+
"grad_norm": 8.0625,
|
| 647 |
+
"learning_rate": 4.642346006013925e-06,
|
| 648 |
+
"logits/chosen": -2.563493251800537,
|
| 649 |
+
"logits/rejected": -2.495145320892334,
|
| 650 |
+
"logps/chosen": -258.86444091796875,
|
| 651 |
+
"logps/rejected": -258.3479309082031,
|
| 652 |
+
"loss": 1.3374,
|
| 653 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 654 |
+
"rewards/chosen": 0.12897971272468567,
|
| 655 |
+
"rewards/margins": 0.4424809515476227,
|
| 656 |
+
"rewards/rejected": -0.3135012090206146,
|
| 657 |
+
"step": 390
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.2616944717042852,
|
| 661 |
+
"grad_norm": 7.6875,
|
| 662 |
+
"learning_rate": 4.612348753784682e-06,
|
| 663 |
+
"logits/chosen": -2.5260744094848633,
|
| 664 |
+
"logits/rejected": -2.3809266090393066,
|
| 665 |
+
"logps/chosen": -296.6476745605469,
|
| 666 |
+
"logps/rejected": -242.137451171875,
|
| 667 |
+
"loss": 1.3452,
|
| 668 |
+
"rewards/accuracies": 0.699999988079071,
|
| 669 |
+
"rewards/chosen": 0.08820326626300812,
|
| 670 |
+
"rewards/margins": 0.4365547299385071,
|
| 671 |
+
"rewards/rejected": -0.34835144877433777,
|
| 672 |
+
"step": 400
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.2616944717042852,
|
| 676 |
+
"eval_logits/chosen": -2.3061885833740234,
|
| 677 |
+
"eval_logits/rejected": -2.224980115890503,
|
| 678 |
+
"eval_logps/chosen": -270.3225402832031,
|
| 679 |
+
"eval_logps/rejected": -256.4975891113281,
|
| 680 |
+
"eval_loss": 1.3535617589950562,
|
| 681 |
+
"eval_rewards/accuracies": 0.7200000286102295,
|
| 682 |
+
"eval_rewards/chosen": 0.185396209359169,
|
| 683 |
+
"eval_rewards/margins": 0.42486390471458435,
|
| 684 |
+
"eval_rewards/rejected": -0.23946763575077057,
|
| 685 |
+
"eval_runtime": 192.5082,
|
| 686 |
+
"eval_samples_per_second": 10.389,
|
| 687 |
+
"eval_steps_per_second": 0.519,
|
| 688 |
+
"step": 400
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.26823683349689237,
|
| 692 |
+
"grad_norm": 7.71875,
|
| 693 |
+
"learning_rate": 4.5812488429302245e-06,
|
| 694 |
+
"logits/chosen": -2.4703800678253174,
|
| 695 |
+
"logits/rejected": -2.384880542755127,
|
| 696 |
+
"logps/chosen": -242.77249145507812,
|
| 697 |
+
"logps/rejected": -216.26919555664062,
|
| 698 |
+
"loss": 1.3521,
|
| 699 |
+
"rewards/accuracies": 0.675000011920929,
|
| 700 |
+
"rewards/chosen": 0.08403290808200836,
|
| 701 |
+
"rewards/margins": 0.23067112267017365,
|
| 702 |
+
"rewards/rejected": -0.14663821458816528,
|
| 703 |
+
"step": 410
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.2747791952894995,
|
| 707 |
+
"grad_norm": 7.625,
|
| 708 |
+
"learning_rate": 4.54906250778917e-06,
|
| 709 |
+
"logits/chosen": -2.494356632232666,
|
| 710 |
+
"logits/rejected": -2.4741995334625244,
|
| 711 |
+
"logps/chosen": -307.1471252441406,
|
| 712 |
+
"logps/rejected": -280.8088073730469,
|
| 713 |
+
"loss": 1.3253,
|
| 714 |
+
"rewards/accuracies": 0.675000011920929,
|
| 715 |
+
"rewards/chosen": 0.11699722707271576,
|
| 716 |
+
"rewards/margins": 0.491929829120636,
|
| 717 |
+
"rewards/rejected": -0.37493258714675903,
|
| 718 |
+
"step": 420
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.28132155708210665,
|
| 722 |
+
"grad_norm": 6.53125,
|
| 723 |
+
"learning_rate": 4.515806549820084e-06,
|
| 724 |
+
"logits/chosen": -2.4651083946228027,
|
| 725 |
+
"logits/rejected": -2.3330154418945312,
|
| 726 |
+
"logps/chosen": -281.57110595703125,
|
| 727 |
+
"logps/rejected": -254.07498168945312,
|
| 728 |
+
"loss": 1.3406,
|
| 729 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 730 |
+
"rewards/chosen": 0.16895495355129242,
|
| 731 |
+
"rewards/margins": 0.4621726870536804,
|
| 732 |
+
"rewards/rejected": -0.2932177186012268,
|
| 733 |
+
"step": 430
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.2878639188747138,
|
| 737 |
+
"grad_norm": 8.875,
|
| 738 |
+
"learning_rate": 4.48149832883101e-06,
|
| 739 |
+
"logits/chosen": -2.4727940559387207,
|
| 740 |
+
"logits/rejected": -2.314194917678833,
|
| 741 |
+
"logps/chosen": -266.56134033203125,
|
| 742 |
+
"logps/rejected": -236.39089965820312,
|
| 743 |
+
"loss": 1.3585,
|
| 744 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 745 |
+
"rewards/chosen": -0.08974994719028473,
|
| 746 |
+
"rewards/margins": 0.26218581199645996,
|
| 747 |
+
"rewards/rejected": -0.3519357144832611,
|
| 748 |
+
"step": 440
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.2944062806673209,
|
| 752 |
+
"grad_norm": 9.125,
|
| 753 |
+
"learning_rate": 4.446155753917559e-06,
|
| 754 |
+
"logits/chosen": -2.4373867511749268,
|
| 755 |
+
"logits/rejected": -2.3501367568969727,
|
| 756 |
+
"logps/chosen": -249.449462890625,
|
| 757 |
+
"logps/rejected": -280.8139343261719,
|
| 758 |
+
"loss": 1.355,
|
| 759 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 760 |
+
"rewards/chosen": 0.09939811378717422,
|
| 761 |
+
"rewards/margins": 0.3732451796531677,
|
| 762 |
+
"rewards/rejected": -0.2738470435142517,
|
| 763 |
+
"step": 450
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.30094864245992803,
|
| 767 |
+
"grad_norm": 7.375,
|
| 768 |
+
"learning_rate": 4.409797274114245e-06,
|
| 769 |
+
"logits/chosen": -2.5222246646881104,
|
| 770 |
+
"logits/rejected": -2.3621227741241455,
|
| 771 |
+
"logps/chosen": -293.97747802734375,
|
| 772 |
+
"logps/rejected": -264.7182312011719,
|
| 773 |
+
"loss": 1.3389,
|
| 774 |
+
"rewards/accuracies": 0.75,
|
| 775 |
+
"rewards/chosen": 0.11550019681453705,
|
| 776 |
+
"rewards/margins": 0.4369579255580902,
|
| 777 |
+
"rewards/rejected": -0.32145771384239197,
|
| 778 |
+
"step": 460
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.30749100425253517,
|
| 782 |
+
"grad_norm": 8.0,
|
| 783 |
+
"learning_rate": 4.372441868763981e-06,
|
| 784 |
+
"logits/chosen": -2.4279232025146484,
|
| 785 |
+
"logits/rejected": -2.3788094520568848,
|
| 786 |
+
"logps/chosen": -269.70782470703125,
|
| 787 |
+
"logps/rejected": -271.08380126953125,
|
| 788 |
+
"loss": 1.3499,
|
| 789 |
+
"rewards/accuracies": 0.737500011920929,
|
| 790 |
+
"rewards/chosen": 0.1701391041278839,
|
| 791 |
+
"rewards/margins": 0.48491889238357544,
|
| 792 |
+
"rewards/rejected": -0.31477978825569153,
|
| 793 |
+
"step": 470
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.3140333660451423,
|
| 797 |
+
"grad_norm": 6.8125,
|
| 798 |
+
"learning_rate": 4.334109037610757e-06,
|
| 799 |
+
"logits/chosen": -2.47053599357605,
|
| 800 |
+
"logits/rejected": -2.4681105613708496,
|
| 801 |
+
"logps/chosen": -248.77597045898438,
|
| 802 |
+
"logps/rejected": -264.0259704589844,
|
| 803 |
+
"loss": 1.3588,
|
| 804 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 805 |
+
"rewards/chosen": 0.08740128576755524,
|
| 806 |
+
"rewards/margins": 0.4287898540496826,
|
| 807 |
+
"rewards/rejected": -0.3413885235786438,
|
| 808 |
+
"step": 480
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.3205757278377494,
|
| 812 |
+
"grad_norm": 9.0625,
|
| 813 |
+
"learning_rate": 4.294818790620644e-06,
|
| 814 |
+
"logits/chosen": -2.388354539871216,
|
| 815 |
+
"logits/rejected": -2.2699804306030273,
|
| 816 |
+
"logps/chosen": -230.1956329345703,
|
| 817 |
+
"logps/rejected": -222.83560180664062,
|
| 818 |
+
"loss": 1.3372,
|
| 819 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 820 |
+
"rewards/chosen": 0.11192689836025238,
|
| 821 |
+
"rewards/margins": 0.4855028986930847,
|
| 822 |
+
"rewards/rejected": -0.37357598543167114,
|
| 823 |
+
"step": 490
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 0.32711808963035655,
|
| 827 |
+
"grad_norm": 6.46875,
|
| 828 |
+
"learning_rate": 4.2545916375364835e-06,
|
| 829 |
+
"logits/chosen": -2.4573752880096436,
|
| 830 |
+
"logits/rejected": -2.384742259979248,
|
| 831 |
+
"logps/chosen": -253.0285186767578,
|
| 832 |
+
"logps/rejected": -235.6699676513672,
|
| 833 |
+
"loss": 1.3446,
|
| 834 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 835 |
+
"rewards/chosen": 0.12988470494747162,
|
| 836 |
+
"rewards/margins": 0.4586367607116699,
|
| 837 |
+
"rewards/rejected": -0.3287521004676819,
|
| 838 |
+
"step": 500
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.32711808963035655,
|
| 842 |
+
"eval_logits/chosen": -2.2817647457122803,
|
| 843 |
+
"eval_logits/rejected": -2.1983530521392822,
|
| 844 |
+
"eval_logps/chosen": -271.3174743652344,
|
| 845 |
+
"eval_logps/rejected": -258.03887939453125,
|
| 846 |
+
"eval_loss": 1.3500897884368896,
|
| 847 |
+
"eval_rewards/accuracies": 0.7275000214576721,
|
| 848 |
+
"eval_rewards/chosen": 0.08590395003557205,
|
| 849 |
+
"eval_rewards/margins": 0.47949859499931335,
|
| 850 |
+
"eval_rewards/rejected": -0.3935946226119995,
|
| 851 |
+
"eval_runtime": 192.5361,
|
| 852 |
+
"eval_samples_per_second": 10.388,
|
| 853 |
+
"eval_steps_per_second": 0.519,
|
| 854 |
+
"step": 500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.3336604514229637,
|
| 858 |
+
"grad_norm": 6.875,
|
| 859 |
+
"learning_rate": 4.213448577171676e-06,
|
| 860 |
+
"logits/chosen": -2.4477853775024414,
|
| 861 |
+
"logits/rejected": -2.2752411365509033,
|
| 862 |
+
"logps/chosen": -286.5871887207031,
|
| 863 |
+
"logps/rejected": -277.1395568847656,
|
| 864 |
+
"loss": 1.3551,
|
| 865 |
+
"rewards/accuracies": 0.699999988079071,
|
| 866 |
+
"rewards/chosen": 0.09675732254981995,
|
| 867 |
+
"rewards/margins": 0.36776870489120483,
|
| 868 |
+
"rewards/rejected": -0.2710114121437073,
|
| 869 |
+
"step": 510
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 0.34020281321557083,
|
| 873 |
+
"grad_norm": 6.71875,
|
| 874 |
+
"learning_rate": 4.171411086448674e-06,
|
| 875 |
+
"logits/chosen": -2.4915995597839355,
|
| 876 |
+
"logits/rejected": -2.40187668800354,
|
| 877 |
+
"logps/chosen": -264.9870910644531,
|
| 878 |
+
"logps/rejected": -248.1427001953125,
|
| 879 |
+
"loss": 1.3515,
|
| 880 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 881 |
+
"rewards/chosen": 0.009617133066058159,
|
| 882 |
+
"rewards/margins": 0.3053894639015198,
|
| 883 |
+
"rewards/rejected": -0.29577234387397766,
|
| 884 |
+
"step": 520
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.346745175008178,
|
| 888 |
+
"grad_norm": 6.84375,
|
| 889 |
+
"learning_rate": 4.128501109187903e-06,
|
| 890 |
+
"logits/chosen": -2.5452780723571777,
|
| 891 |
+
"logits/rejected": -2.388065814971924,
|
| 892 |
+
"logps/chosen": -256.1615905761719,
|
| 893 |
+
"logps/rejected": -243.15542602539062,
|
| 894 |
+
"loss": 1.3448,
|
| 895 |
+
"rewards/accuracies": 0.737500011920929,
|
| 896 |
+
"rewards/chosen": 0.09919309616088867,
|
| 897 |
+
"rewards/margins": 0.45780619978904724,
|
| 898 |
+
"rewards/rejected": -0.35861313343048096,
|
| 899 |
+
"step": 530
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.35328753680078506,
|
| 903 |
+
"grad_norm": 8.0,
|
| 904 |
+
"learning_rate": 4.084741044652956e-06,
|
| 905 |
+
"logits/chosen": -2.5606534481048584,
|
| 906 |
+
"logits/rejected": -2.3131465911865234,
|
| 907 |
+
"logps/chosen": -287.9478759765625,
|
| 908 |
+
"logps/rejected": -224.75527954101562,
|
| 909 |
+
"loss": 1.3668,
|
| 910 |
+
"rewards/accuracies": 0.737500011920929,
|
| 911 |
+
"rewards/chosen": 0.17254912853240967,
|
| 912 |
+
"rewards/margins": 0.42019981145858765,
|
| 913 |
+
"rewards/rejected": -0.24765071272850037,
|
| 914 |
+
"step": 540
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.3598298985933922,
|
| 918 |
+
"grad_norm": 6.90625,
|
| 919 |
+
"learning_rate": 4.040153735858041e-06,
|
| 920 |
+
"logits/chosen": -2.4491143226623535,
|
| 921 |
+
"logits/rejected": -2.3129725456237793,
|
| 922 |
+
"logps/chosen": -274.43658447265625,
|
| 923 |
+
"logps/rejected": -268.6435241699219,
|
| 924 |
+
"loss": 1.3476,
|
| 925 |
+
"rewards/accuracies": 0.75,
|
| 926 |
+
"rewards/chosen": 0.14739489555358887,
|
| 927 |
+
"rewards/margins": 0.41326436400413513,
|
| 928 |
+
"rewards/rejected": -0.26586946845054626,
|
| 929 |
+
"step": 550
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.36637226038599935,
|
| 933 |
+
"grad_norm": 6.84375,
|
| 934 |
+
"learning_rate": 3.9947624576437975e-06,
|
| 935 |
+
"logits/chosen": -2.47575306892395,
|
| 936 |
+
"logits/rejected": -2.3158023357391357,
|
| 937 |
+
"logps/chosen": -230.30819702148438,
|
| 938 |
+
"logps/rejected": -211.89633178710938,
|
| 939 |
+
"loss": 1.3509,
|
| 940 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 941 |
+
"rewards/chosen": 0.07401107251644135,
|
| 942 |
+
"rewards/margins": 0.36348801851272583,
|
| 943 |
+
"rewards/rejected": -0.2894769012928009,
|
| 944 |
+
"step": 560
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.3729146221786065,
|
| 948 |
+
"grad_norm": 6.90625,
|
| 949 |
+
"learning_rate": 3.948590904527689e-06,
|
| 950 |
+
"logits/chosen": -2.395169496536255,
|
| 951 |
+
"logits/rejected": -2.330249309539795,
|
| 952 |
+
"logps/chosen": -243.09487915039062,
|
| 953 |
+
"logps/rejected": -263.39227294921875,
|
| 954 |
+
"loss": 1.3349,
|
| 955 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 956 |
+
"rewards/chosen": -0.01647457666695118,
|
| 957 |
+
"rewards/margins": 0.5050023794174194,
|
| 958 |
+
"rewards/rejected": -0.5214769244194031,
|
| 959 |
+
"step": 570
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.37945698397121363,
|
| 963 |
+
"grad_norm": 6.46875,
|
| 964 |
+
"learning_rate": 3.901663178335318e-06,
|
| 965 |
+
"logits/chosen": -2.5840940475463867,
|
| 966 |
+
"logits/rejected": -2.4302070140838623,
|
| 967 |
+
"logps/chosen": -292.9772644042969,
|
| 968 |
+
"logps/rejected": -280.25726318359375,
|
| 969 |
+
"loss": 1.3405,
|
| 970 |
+
"rewards/accuracies": 0.675000011920929,
|
| 971 |
+
"rewards/chosen": 0.11058641970157623,
|
| 972 |
+
"rewards/margins": 0.506061851978302,
|
| 973 |
+
"rewards/rejected": -0.39547544717788696,
|
| 974 |
+
"step": 580
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.3859993457638207,
|
| 978 |
+
"grad_norm": 7.59375,
|
| 979 |
+
"learning_rate": 3.854003775619142e-06,
|
| 980 |
+
"logits/chosen": -2.3781380653381348,
|
| 981 |
+
"logits/rejected": -2.372018337249756,
|
| 982 |
+
"logps/chosen": -263.7474365234375,
|
| 983 |
+
"logps/rejected": -244.7537384033203,
|
| 984 |
+
"loss": 1.3416,
|
| 985 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 986 |
+
"rewards/chosen": 0.06603378802537918,
|
| 987 |
+
"rewards/margins": 0.4166053831577301,
|
| 988 |
+
"rewards/rejected": -0.3505716323852539,
|
| 989 |
+
"step": 590
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.39254170755642787,
|
| 993 |
+
"grad_norm": 8.0625,
|
| 994 |
+
"learning_rate": 3.805637574871115e-06,
|
| 995 |
+
"logits/chosen": -2.455042600631714,
|
| 996 |
+
"logits/rejected": -2.3687422275543213,
|
| 997 |
+
"logps/chosen": -266.13043212890625,
|
| 998 |
+
"logps/rejected": -225.39614868164062,
|
| 999 |
+
"loss": 1.333,
|
| 1000 |
+
"rewards/accuracies": 0.75,
|
| 1001 |
+
"rewards/chosen": 0.10043742507696152,
|
| 1002 |
+
"rewards/margins": 0.5269637703895569,
|
| 1003 |
+
"rewards/rejected": -0.4265263080596924,
|
| 1004 |
+
"step": 600
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.39254170755642787,
|
| 1008 |
+
"eval_logits/chosen": -2.293658494949341,
|
| 1009 |
+
"eval_logits/rejected": -2.2107114791870117,
|
| 1010 |
+
"eval_logps/chosen": -271.6837463378906,
|
| 1011 |
+
"eval_logps/rejected": -257.95440673828125,
|
| 1012 |
+
"eval_loss": 1.3495758771896362,
|
| 1013 |
+
"eval_rewards/accuracies": 0.7450000047683716,
|
| 1014 |
+
"eval_rewards/chosen": 0.04927777126431465,
|
| 1015 |
+
"eval_rewards/margins": 0.43442490696907043,
|
| 1016 |
+
"eval_rewards/rejected": -0.38514718413352966,
|
| 1017 |
+
"eval_runtime": 192.5013,
|
| 1018 |
+
"eval_samples_per_second": 10.39,
|
| 1019 |
+
"eval_steps_per_second": 0.519,
|
| 1020 |
+
"step": 600
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 0.399084069349035,
|
| 1024 |
+
"grad_norm": 6.71875,
|
| 1025 |
+
"learning_rate": 3.7565898235359717e-06,
|
| 1026 |
+
"logits/chosen": -2.531266689300537,
|
| 1027 |
+
"logits/rejected": -2.4104650020599365,
|
| 1028 |
+
"logps/chosen": -268.50482177734375,
|
| 1029 |
+
"logps/rejected": -252.98202514648438,
|
| 1030 |
+
"loss": 1.3522,
|
| 1031 |
+
"rewards/accuracies": 0.75,
|
| 1032 |
+
"rewards/chosen": 0.11455075442790985,
|
| 1033 |
+
"rewards/margins": 0.33470815420150757,
|
| 1034 |
+
"rewards/rejected": -0.2201574295759201,
|
| 1035 |
+
"step": 610
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.40562643114164215,
|
| 1039 |
+
"grad_norm": 7.0,
|
| 1040 |
+
"learning_rate": 3.7068861248319127e-06,
|
| 1041 |
+
"logits/chosen": -2.420405626296997,
|
| 1042 |
+
"logits/rejected": -2.36627197265625,
|
| 1043 |
+
"logps/chosen": -252.94091796875,
|
| 1044 |
+
"logps/rejected": -260.6941833496094,
|
| 1045 |
+
"loss": 1.3502,
|
| 1046 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1047 |
+
"rewards/chosen": 0.004491160623729229,
|
| 1048 |
+
"rewards/margins": 0.4471139907836914,
|
| 1049 |
+
"rewards/rejected": -0.4426228106021881,
|
| 1050 |
+
"step": 620
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.41216879293424924,
|
| 1054 |
+
"grad_norm": 8.125,
|
| 1055 |
+
"learning_rate": 3.6565524243855695e-06,
|
| 1056 |
+
"logits/chosen": -2.482597589492798,
|
| 1057 |
+
"logits/rejected": -2.381808042526245,
|
| 1058 |
+
"logps/chosen": -270.79833984375,
|
| 1059 |
+
"logps/rejected": -254.92868041992188,
|
| 1060 |
+
"loss": 1.3455,
|
| 1061 |
+
"rewards/accuracies": 0.75,
|
| 1062 |
+
"rewards/chosen": 0.18158364295959473,
|
| 1063 |
+
"rewards/margins": 0.5938194394111633,
|
| 1064 |
+
"rewards/rejected": -0.41223573684692383,
|
| 1065 |
+
"step": 630
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.4187111547268564,
|
| 1069 |
+
"grad_norm": 8.125,
|
| 1070 |
+
"learning_rate": 3.6056149966882325e-06,
|
| 1071 |
+
"logits/chosen": -2.4741575717926025,
|
| 1072 |
+
"logits/rejected": -2.4443914890289307,
|
| 1073 |
+
"logps/chosen": -257.9027099609375,
|
| 1074 |
+
"logps/rejected": -263.8404235839844,
|
| 1075 |
+
"loss": 1.3501,
|
| 1076 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1077 |
+
"rewards/chosen": 0.12312284857034683,
|
| 1078 |
+
"rewards/margins": 0.4808879792690277,
|
| 1079 |
+
"rewards/rejected": -0.35776516795158386,
|
| 1080 |
+
"step": 640
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.4252535165194635,
|
| 1084 |
+
"grad_norm": 7.84375,
|
| 1085 |
+
"learning_rate": 3.554100431380414e-06,
|
| 1086 |
+
"logits/chosen": -2.493617534637451,
|
| 1087 |
+
"logits/rejected": -2.3935816287994385,
|
| 1088 |
+
"logps/chosen": -249.92294311523438,
|
| 1089 |
+
"logps/rejected": -238.17404174804688,
|
| 1090 |
+
"loss": 1.3504,
|
| 1091 |
+
"rewards/accuracies": 0.75,
|
| 1092 |
+
"rewards/chosen": 0.04006017744541168,
|
| 1093 |
+
"rewards/margins": 0.48104342818260193,
|
| 1094 |
+
"rewards/rejected": -0.44098323583602905,
|
| 1095 |
+
"step": 650
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.43179587831207067,
|
| 1099 |
+
"grad_norm": 9.0625,
|
| 1100 |
+
"learning_rate": 3.5020356193718934e-06,
|
| 1101 |
+
"logits/chosen": -2.4255988597869873,
|
| 1102 |
+
"logits/rejected": -2.3747265338897705,
|
| 1103 |
+
"logps/chosen": -217.7563934326172,
|
| 1104 |
+
"logps/rejected": -220.0769500732422,
|
| 1105 |
+
"loss": 1.3667,
|
| 1106 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1107 |
+
"rewards/chosen": 0.10521771758794785,
|
| 1108 |
+
"rewards/margins": 0.3761609196662903,
|
| 1109 |
+
"rewards/rejected": -0.27094319462776184,
|
| 1110 |
+
"step": 660
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 0.4383382401046778,
|
| 1114 |
+
"grad_norm": 7.96875,
|
| 1115 |
+
"learning_rate": 3.4494477388045035e-06,
|
| 1116 |
+
"logits/chosen": -2.491364002227783,
|
| 1117 |
+
"logits/rejected": -2.235619068145752,
|
| 1118 |
+
"logps/chosen": -301.46600341796875,
|
| 1119 |
+
"logps/rejected": -238.8917999267578,
|
| 1120 |
+
"loss": 1.3419,
|
| 1121 |
+
"rewards/accuracies": 0.75,
|
| 1122 |
+
"rewards/chosen": 0.07640339434146881,
|
| 1123 |
+
"rewards/margins": 0.46867623925209045,
|
| 1124 |
+
"rewards/rejected": -0.39227285981178284,
|
| 1125 |
+
"step": 670
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.4448806018972849,
|
| 1129 |
+
"grad_norm": 7.15625,
|
| 1130 |
+
"learning_rate": 3.3963642408649783e-06,
|
| 1131 |
+
"logits/chosen": -2.4565513134002686,
|
| 1132 |
+
"logits/rejected": -2.3826098442077637,
|
| 1133 |
+
"logps/chosen": -282.40228271484375,
|
| 1134 |
+
"logps/rejected": -249.9719696044922,
|
| 1135 |
+
"loss": 1.3423,
|
| 1136 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1137 |
+
"rewards/chosen": 0.1387428194284439,
|
| 1138 |
+
"rewards/margins": 0.49391037225723267,
|
| 1139 |
+
"rewards/rejected": -0.35516756772994995,
|
| 1140 |
+
"step": 680
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 0.45142296368989204,
|
| 1144 |
+
"grad_norm": 8.5,
|
| 1145 |
+
"learning_rate": 3.3428128354552727e-06,
|
| 1146 |
+
"logits/chosen": -2.3810625076293945,
|
| 1147 |
+
"logits/rejected": -2.34759521484375,
|
| 1148 |
+
"logps/chosen": -252.3175048828125,
|
| 1149 |
+
"logps/rejected": -237.6392364501953,
|
| 1150 |
+
"loss": 1.3457,
|
| 1151 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1152 |
+
"rewards/chosen": 0.10267229378223419,
|
| 1153 |
+
"rewards/margins": 0.4029284119606018,
|
| 1154 |
+
"rewards/rejected": -0.30025607347488403,
|
| 1155 |
+
"step": 690
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 0.4579653254824992,
|
| 1159 |
+
"grad_norm": 6.375,
|
| 1160 |
+
"learning_rate": 3.2888214767278246e-06,
|
| 1161 |
+
"logits/chosen": -2.5027060508728027,
|
| 1162 |
+
"logits/rejected": -2.3413376808166504,
|
| 1163 |
+
"logps/chosen": -307.34686279296875,
|
| 1164 |
+
"logps/rejected": -240.319091796875,
|
| 1165 |
+
"loss": 1.3577,
|
| 1166 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1167 |
+
"rewards/chosen": 0.04387027025222778,
|
| 1168 |
+
"rewards/margins": 0.33918899297714233,
|
| 1169 |
+
"rewards/rejected": -0.29531875252723694,
|
| 1170 |
+
"step": 700
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 0.4579653254824992,
|
| 1174 |
+
"eval_logits/chosen": -2.2934322357177734,
|
| 1175 |
+
"eval_logits/rejected": -2.2099828720092773,
|
| 1176 |
+
"eval_logps/chosen": -270.8705749511719,
|
| 1177 |
+
"eval_logps/rejected": -256.79083251953125,
|
| 1178 |
+
"eval_loss": 1.345709204673767,
|
| 1179 |
+
"eval_rewards/accuracies": 0.7174999713897705,
|
| 1180 |
+
"eval_rewards/chosen": 0.13059137761592865,
|
| 1181 |
+
"eval_rewards/margins": 0.39938145875930786,
|
| 1182 |
+
"eval_rewards/rejected": -0.268790066242218,
|
| 1183 |
+
"eval_runtime": 192.4798,
|
| 1184 |
+
"eval_samples_per_second": 10.391,
|
| 1185 |
+
"eval_steps_per_second": 0.52,
|
| 1186 |
+
"step": 700
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.46450768727510633,
|
| 1190 |
+
"grad_norm": 7.21875,
|
| 1191 |
+
"learning_rate": 3.2344183484933247e-06,
|
| 1192 |
+
"logits/chosen": -2.443279266357422,
|
| 1193 |
+
"logits/rejected": -2.3943705558776855,
|
| 1194 |
+
"logps/chosen": -256.82928466796875,
|
| 1195 |
+
"logps/rejected": -235.6039276123047,
|
| 1196 |
+
"loss": 1.3379,
|
| 1197 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1198 |
+
"rewards/chosen": 0.02693643607199192,
|
| 1199 |
+
"rewards/margins": 0.40261945128440857,
|
| 1200 |
+
"rewards/rejected": -0.3756829798221588,
|
| 1201 |
+
"step": 710
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 0.47105004906771347,
|
| 1205 |
+
"grad_norm": 9.625,
|
| 1206 |
+
"learning_rate": 3.179631849508597e-06,
|
| 1207 |
+
"logits/chosen": -2.445950984954834,
|
| 1208 |
+
"logits/rejected": -2.4440605640411377,
|
| 1209 |
+
"logps/chosen": -249.5127410888672,
|
| 1210 |
+
"logps/rejected": -268.5820007324219,
|
| 1211 |
+
"loss": 1.3417,
|
| 1212 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1213 |
+
"rewards/chosen": 0.09297636151313782,
|
| 1214 |
+
"rewards/margins": 0.33975750207901,
|
| 1215 |
+
"rewards/rejected": -0.2467811554670334,
|
| 1216 |
+
"step": 720
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 0.47759241086032056,
|
| 1220 |
+
"grad_norm": 7.59375,
|
| 1221 |
+
"learning_rate": 3.1244905786522796e-06,
|
| 1222 |
+
"logits/chosen": -2.4762067794799805,
|
| 1223 |
+
"logits/rejected": -2.4062716960906982,
|
| 1224 |
+
"logps/chosen": -264.5029602050781,
|
| 1225 |
+
"logps/rejected": -255.11361694335938,
|
| 1226 |
+
"loss": 1.3587,
|
| 1227 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1228 |
+
"rewards/chosen": 0.11461961269378662,
|
| 1229 |
+
"rewards/margins": 0.3867848515510559,
|
| 1230 |
+
"rewards/rejected": -0.2721652388572693,
|
| 1231 |
+
"step": 730
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 0.4841347726529277,
|
| 1235 |
+
"grad_norm": 7.4375,
|
| 1236 |
+
"learning_rate": 3.0690233199960393e-06,
|
| 1237 |
+
"logits/chosen": -2.3173129558563232,
|
| 1238 |
+
"logits/rejected": -2.2710585594177246,
|
| 1239 |
+
"logps/chosen": -227.1941375732422,
|
| 1240 |
+
"logps/rejected": -248.1180877685547,
|
| 1241 |
+
"loss": 1.3402,
|
| 1242 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1243 |
+
"rewards/chosen": 0.1001671776175499,
|
| 1244 |
+
"rewards/margins": 0.45164409279823303,
|
| 1245 |
+
"rewards/rejected": -0.35147690773010254,
|
| 1246 |
+
"step": 740
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 0.49067713444553485,
|
| 1250 |
+
"grad_norm": 7.4375,
|
| 1251 |
+
"learning_rate": 3.0132590277791163e-06,
|
| 1252 |
+
"logits/chosen": -2.524819850921631,
|
| 1253 |
+
"logits/rejected": -2.406615734100342,
|
| 1254 |
+
"logps/chosen": -263.93133544921875,
|
| 1255 |
+
"logps/rejected": -243.890625,
|
| 1256 |
+
"loss": 1.3328,
|
| 1257 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1258 |
+
"rewards/chosen": 0.14402124285697937,
|
| 1259 |
+
"rewards/margins": 0.5698949098587036,
|
| 1260 |
+
"rewards/rejected": -0.42587366700172424,
|
| 1261 |
+
"step": 750
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 0.497219496238142,
|
| 1265 |
+
"grad_norm": 7.96875,
|
| 1266 |
+
"learning_rate": 2.9572268112940354e-06,
|
| 1267 |
+
"logits/chosen": -2.495116710662842,
|
| 1268 |
+
"logits/rejected": -2.4868061542510986,
|
| 1269 |
+
"logps/chosen": -254.06405639648438,
|
| 1270 |
+
"logps/rejected": -259.20379638671875,
|
| 1271 |
+
"loss": 1.3341,
|
| 1272 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1273 |
+
"rewards/chosen": 0.02839145064353943,
|
| 1274 |
+
"rewards/margins": 0.419182687997818,
|
| 1275 |
+
"rewards/rejected": -0.39079123735427856,
|
| 1276 |
+
"step": 760
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.5037618580307491,
|
| 1280 |
+
"grad_norm": 8.1875,
|
| 1281 |
+
"learning_rate": 2.9009559196913882e-06,
|
| 1282 |
+
"logits/chosen": -2.5047004222869873,
|
| 1283 |
+
"logits/rejected": -2.386676788330078,
|
| 1284 |
+
"logps/chosen": -291.96478271484375,
|
| 1285 |
+
"logps/rejected": -252.67971801757812,
|
| 1286 |
+
"loss": 1.3422,
|
| 1287 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1288 |
+
"rewards/chosen": 0.1507759988307953,
|
| 1289 |
+
"rewards/margins": 0.5127144455909729,
|
| 1290 |
+
"rewards/rejected": -0.3619384467601776,
|
| 1291 |
+
"step": 770
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.5103042198233563,
|
| 1295 |
+
"grad_norm": 7.84375,
|
| 1296 |
+
"learning_rate": 2.844475726711595e-06,
|
| 1297 |
+
"logits/chosen": -2.4655020236968994,
|
| 1298 |
+
"logits/rejected": -2.294151782989502,
|
| 1299 |
+
"logps/chosen": -249.9043731689453,
|
| 1300 |
+
"logps/rejected": -220.50216674804688,
|
| 1301 |
+
"loss": 1.3446,
|
| 1302 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1303 |
+
"rewards/chosen": 0.06338523328304291,
|
| 1304 |
+
"rewards/margins": 0.47769251465797424,
|
| 1305 |
+
"rewards/rejected": -0.41430729627609253,
|
| 1306 |
+
"step": 780
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.5168465816159633,
|
| 1310 |
+
"grad_norm": 6.0625,
|
| 1311 |
+
"learning_rate": 2.7878157153516446e-06,
|
| 1312 |
+
"logits/chosen": -2.4090731143951416,
|
| 1313 |
+
"logits/rejected": -2.3185999393463135,
|
| 1314 |
+
"logps/chosen": -270.53857421875,
|
| 1315 |
+
"logps/rejected": -269.4087829589844,
|
| 1316 |
+
"loss": 1.33,
|
| 1317 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1318 |
+
"rewards/chosen": 0.11772044748067856,
|
| 1319 |
+
"rewards/margins": 0.4702891707420349,
|
| 1320 |
+
"rewards/rejected": -0.35256871581077576,
|
| 1321 |
+
"step": 790
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 0.5233889434085705,
|
| 1325 |
+
"grad_norm": 7.75,
|
| 1326 |
+
"learning_rate": 2.731005462474787e-06,
|
| 1327 |
+
"logits/chosen": -2.5132055282592773,
|
| 1328 |
+
"logits/rejected": -2.4322190284729004,
|
| 1329 |
+
"logps/chosen": -271.2453308105469,
|
| 1330 |
+
"logps/rejected": -266.37701416015625,
|
| 1331 |
+
"loss": 1.343,
|
| 1332 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1333 |
+
"rewards/chosen": 0.21996447443962097,
|
| 1334 |
+
"rewards/margins": 0.497943639755249,
|
| 1335 |
+
"rewards/rejected": -0.27797916531562805,
|
| 1336 |
+
"step": 800
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 0.5233889434085705,
|
| 1340 |
+
"eval_logits/chosen": -2.3120627403259277,
|
| 1341 |
+
"eval_logits/rejected": -2.2312381267547607,
|
| 1342 |
+
"eval_logps/chosen": -271.3629150390625,
|
| 1343 |
+
"eval_logps/rejected": -257.9126892089844,
|
| 1344 |
+
"eval_loss": 1.3448688983917236,
|
| 1345 |
+
"eval_rewards/accuracies": 0.7149999737739563,
|
| 1346 |
+
"eval_rewards/chosen": 0.08135941624641418,
|
| 1347 |
+
"eval_rewards/margins": 0.4623354375362396,
|
| 1348 |
+
"eval_rewards/rejected": -0.38097602128982544,
|
| 1349 |
+
"eval_runtime": 192.5442,
|
| 1350 |
+
"eval_samples_per_second": 10.387,
|
| 1351 |
+
"eval_steps_per_second": 0.519,
|
| 1352 |
+
"step": 800
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 0.5299313052011776,
|
| 1356 |
+
"grad_norm": 8.25,
|
| 1357 |
+
"learning_rate": 2.67407462337124e-06,
|
| 1358 |
+
"logits/chosen": -2.516996383666992,
|
| 1359 |
+
"logits/rejected": -2.399326801300049,
|
| 1360 |
+
"logps/chosen": -245.3385772705078,
|
| 1361 |
+
"logps/rejected": -241.2891082763672,
|
| 1362 |
+
"loss": 1.334,
|
| 1363 |
+
"rewards/accuracies": 0.75,
|
| 1364 |
+
"rewards/chosen": 0.0016076326137408614,
|
| 1365 |
+
"rewards/margins": 0.4354288578033447,
|
| 1366 |
+
"rewards/rejected": -0.43382126092910767,
|
| 1367 |
+
"step": 810
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.5364736669937847,
|
| 1371 |
+
"grad_norm": 7.0,
|
| 1372 |
+
"learning_rate": 2.617052916277952e-06,
|
| 1373 |
+
"logits/chosen": -2.5313477516174316,
|
| 1374 |
+
"logits/rejected": -2.338625431060791,
|
| 1375 |
+
"logps/chosen": -253.85226440429688,
|
| 1376 |
+
"logps/rejected": -227.67898559570312,
|
| 1377 |
+
"loss": 1.3429,
|
| 1378 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1379 |
+
"rewards/chosen": 0.07979673147201538,
|
| 1380 |
+
"rewards/margins": 0.5423867106437683,
|
| 1381 |
+
"rewards/rejected": -0.4625900387763977,
|
| 1382 |
+
"step": 820
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.5430160287863919,
|
| 1386 |
+
"grad_norm": 7.1875,
|
| 1387 |
+
"learning_rate": 2.5599701068654985e-06,
|
| 1388 |
+
"logits/chosen": -2.5015134811401367,
|
| 1389 |
+
"logits/rejected": -2.4792141914367676,
|
| 1390 |
+
"logps/chosen": -250.22543334960938,
|
| 1391 |
+
"logps/rejected": -245.0906219482422,
|
| 1392 |
+
"loss": 1.3437,
|
| 1393 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1394 |
+
"rewards/chosen": 0.13883474469184875,
|
| 1395 |
+
"rewards/margins": 0.5313934087753296,
|
| 1396 |
+
"rewards/rejected": -0.39255863428115845,
|
| 1397 |
+
"step": 830
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 0.549558390578999,
|
| 1401 |
+
"grad_norm": 7.15625,
|
| 1402 |
+
"learning_rate": 2.5028559927002326e-06,
|
| 1403 |
+
"logits/chosen": -2.562939405441284,
|
| 1404 |
+
"logits/rejected": -2.4629690647125244,
|
| 1405 |
+
"logps/chosen": -295.15142822265625,
|
| 1406 |
+
"logps/rejected": -249.9164276123047,
|
| 1407 |
+
"loss": 1.3382,
|
| 1408 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1409 |
+
"rewards/chosen": 0.11802862584590912,
|
| 1410 |
+
"rewards/margins": 0.46983498334884644,
|
| 1411 |
+
"rewards/rejected": -0.3518063426017761,
|
| 1412 |
+
"step": 840
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 0.5561007523716062,
|
| 1416 |
+
"grad_norm": 10.4375,
|
| 1417 |
+
"learning_rate": 2.4457403876897756e-06,
|
| 1418 |
+
"logits/chosen": -2.3556885719299316,
|
| 1419 |
+
"logits/rejected": -2.300297737121582,
|
| 1420 |
+
"logps/chosen": -241.96975708007812,
|
| 1421 |
+
"logps/rejected": -196.66481018066406,
|
| 1422 |
+
"loss": 1.3402,
|
| 1423 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1424 |
+
"rewards/chosen": 0.06055239588022232,
|
| 1425 |
+
"rewards/margins": 0.29153305292129517,
|
| 1426 |
+
"rewards/rejected": -0.23098066449165344,
|
| 1427 |
+
"step": 850
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 0.5626431141642133,
|
| 1431 |
+
"grad_norm": 7.84375,
|
| 1432 |
+
"learning_rate": 2.388653106519975e-06,
|
| 1433 |
+
"logits/chosen": -2.504599094390869,
|
| 1434 |
+
"logits/rejected": -2.3005897998809814,
|
| 1435 |
+
"logps/chosen": -257.0250549316406,
|
| 1436 |
+
"logps/rejected": -192.78662109375,
|
| 1437 |
+
"loss": 1.3466,
|
| 1438 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1439 |
+
"rewards/chosen": 0.04565539211034775,
|
| 1440 |
+
"rewards/margins": 0.4032720923423767,
|
| 1441 |
+
"rewards/rejected": -0.35761672258377075,
|
| 1442 |
+
"step": 860
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 0.5691854759568205,
|
| 1446 |
+
"grad_norm": 7.21875,
|
| 1447 |
+
"learning_rate": 2.331623949091467e-06,
|
| 1448 |
+
"logits/chosen": -2.5008959770202637,
|
| 1449 |
+
"logits/rejected": -2.472198009490967,
|
| 1450 |
+
"logps/chosen": -277.51580810546875,
|
| 1451 |
+
"logps/rejected": -266.95245361328125,
|
| 1452 |
+
"loss": 1.3361,
|
| 1453 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1454 |
+
"rewards/chosen": 0.13331273198127747,
|
| 1455 |
+
"rewards/margins": 0.45952072739601135,
|
| 1456 |
+
"rewards/rejected": -0.3262080252170563,
|
| 1457 |
+
"step": 870
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 0.5757278377494276,
|
| 1461 |
+
"grad_norm": 7.75,
|
| 1462 |
+
"learning_rate": 2.2746826849639513e-06,
|
| 1463 |
+
"logits/chosen": -2.505166530609131,
|
| 1464 |
+
"logits/rejected": -2.3743433952331543,
|
| 1465 |
+
"logps/chosen": -281.8514709472656,
|
| 1466 |
+
"logps/rejected": -239.01443481445312,
|
| 1467 |
+
"loss": 1.3391,
|
| 1468 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1469 |
+
"rewards/chosen": 0.11163483560085297,
|
| 1470 |
+
"rewards/margins": 0.4709078371524811,
|
| 1471 |
+
"rewards/rejected": -0.35927295684814453,
|
| 1472 |
+
"step": 880
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.5822701995420346,
|
| 1476 |
+
"grad_norm": 7.1875,
|
| 1477 |
+
"learning_rate": 2.2178590378162957e-06,
|
| 1478 |
+
"logits/chosen": -2.5283544063568115,
|
| 1479 |
+
"logits/rejected": -2.402369260787964,
|
| 1480 |
+
"logps/chosen": -319.1461486816406,
|
| 1481 |
+
"logps/rejected": -279.74462890625,
|
| 1482 |
+
"loss": 1.3521,
|
| 1483 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1484 |
+
"rewards/chosen": 0.02043941244482994,
|
| 1485 |
+
"rewards/margins": 0.45712727308273315,
|
| 1486 |
+
"rewards/rejected": -0.4366879463195801,
|
| 1487 |
+
"step": 890
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.5888125613346418,
|
| 1491 |
+
"grad_norm": 7.15625,
|
| 1492 |
+
"learning_rate": 2.1611826699306104e-06,
|
| 1493 |
+
"logits/chosen": -2.4243669509887695,
|
| 1494 |
+
"logits/rejected": -2.32483172416687,
|
| 1495 |
+
"logps/chosen": -271.6533203125,
|
| 1496 |
+
"logps/rejected": -232.5511474609375,
|
| 1497 |
+
"loss": 1.3439,
|
| 1498 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1499 |
+
"rewards/chosen": 0.13815268874168396,
|
| 1500 |
+
"rewards/margins": 0.49229878187179565,
|
| 1501 |
+
"rewards/rejected": -0.35414618253707886,
|
| 1502 |
+
"step": 900
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 0.5888125613346418,
|
| 1506 |
+
"eval_logits/chosen": -2.313720941543579,
|
| 1507 |
+
"eval_logits/rejected": -2.2327077388763428,
|
| 1508 |
+
"eval_logps/chosen": -271.791748046875,
|
| 1509 |
+
"eval_logps/rejected": -258.1572570800781,
|
| 1510 |
+
"eval_loss": 1.3458954095840454,
|
| 1511 |
+
"eval_rewards/accuracies": 0.7250000238418579,
|
| 1512 |
+
"eval_rewards/chosen": 0.038476575165987015,
|
| 1513 |
+
"eval_rewards/margins": 0.443908154964447,
|
| 1514 |
+
"eval_rewards/rejected": -0.4054316282272339,
|
| 1515 |
+
"eval_runtime": 192.5142,
|
| 1516 |
+
"eval_samples_per_second": 10.389,
|
| 1517 |
+
"eval_steps_per_second": 0.519,
|
| 1518 |
+
"step": 900
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 0.5953549231272489,
|
| 1522 |
+
"grad_norm": 8.0625,
|
| 1523 |
+
"learning_rate": 2.1046831667083483e-06,
|
| 1524 |
+
"logits/chosen": -2.4782989025115967,
|
| 1525 |
+
"logits/rejected": -2.3582215309143066,
|
| 1526 |
+
"logps/chosen": -341.8036193847656,
|
| 1527 |
+
"logps/rejected": -267.46795654296875,
|
| 1528 |
+
"loss": 1.3463,
|
| 1529 |
+
"rewards/accuracies": 0.8125,
|
| 1530 |
+
"rewards/chosen": 0.08111786842346191,
|
| 1531 |
+
"rewards/margins": 0.4356920123100281,
|
| 1532 |
+
"rewards/rejected": -0.35457414388656616,
|
| 1533 |
+
"step": 910
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 0.6018972849198561,
|
| 1537 |
+
"grad_norm": 7.09375,
|
| 1538 |
+
"learning_rate": 2.048390021226559e-06,
|
| 1539 |
+
"logits/chosen": -2.445460557937622,
|
| 1540 |
+
"logits/rejected": -2.3675286769866943,
|
| 1541 |
+
"logps/chosen": -262.48907470703125,
|
| 1542 |
+
"logps/rejected": -266.7914733886719,
|
| 1543 |
+
"loss": 1.3489,
|
| 1544 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1545 |
+
"rewards/chosen": 0.1052175760269165,
|
| 1546 |
+
"rewards/margins": 0.4468202590942383,
|
| 1547 |
+
"rewards/rejected": -0.341602623462677,
|
| 1548 |
+
"step": 920
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 0.6084396467124632,
|
| 1552 |
+
"grad_norm": 7.25,
|
| 1553 |
+
"learning_rate": 1.9923326188423212e-06,
|
| 1554 |
+
"logits/chosen": -2.5053839683532715,
|
| 1555 |
+
"logits/rejected": -2.4001247882843018,
|
| 1556 |
+
"logps/chosen": -236.89382934570312,
|
| 1557 |
+
"logps/rejected": -205.64291381835938,
|
| 1558 |
+
"loss": 1.3523,
|
| 1559 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1560 |
+
"rewards/chosen": 0.08720345795154572,
|
| 1561 |
+
"rewards/margins": 0.3279283344745636,
|
| 1562 |
+
"rewards/rejected": -0.24072487652301788,
|
| 1563 |
+
"step": 930
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.6149820085050703,
|
| 1567 |
+
"grad_norm": 7.4375,
|
| 1568 |
+
"learning_rate": 1.936540221853415e-06,
|
| 1569 |
+
"logits/chosen": -2.4638781547546387,
|
| 1570 |
+
"logits/rejected": -2.3390696048736572,
|
| 1571 |
+
"logps/chosen": -283.2751770019531,
|
| 1572 |
+
"logps/rejected": -238.634765625,
|
| 1573 |
+
"loss": 1.3297,
|
| 1574 |
+
"rewards/accuracies": 0.75,
|
| 1575 |
+
"rewards/chosen": 0.058457307517528534,
|
| 1576 |
+
"rewards/margins": 0.6001859307289124,
|
| 1577 |
+
"rewards/rejected": -0.5417286157608032,
|
| 1578 |
+
"step": 940
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.6215243702976775,
|
| 1582 |
+
"grad_norm": 6.71875,
|
| 1583 |
+
"learning_rate": 1.8810419542232245e-06,
|
| 1584 |
+
"logits/chosen": -2.5065360069274902,
|
| 1585 |
+
"logits/rejected": -2.3907275199890137,
|
| 1586 |
+
"logps/chosen": -273.1107177734375,
|
| 1587 |
+
"logps/rejected": -269.38140869140625,
|
| 1588 |
+
"loss": 1.3301,
|
| 1589 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1590 |
+
"rewards/chosen": 0.07303529977798462,
|
| 1591 |
+
"rewards/margins": 0.37555113434791565,
|
| 1592 |
+
"rewards/rejected": -0.30251583456993103,
|
| 1593 |
+
"step": 950
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.6280667320902846,
|
| 1597 |
+
"grad_norm": 6.71875,
|
| 1598 |
+
"learning_rate": 1.8258667863778573e-06,
|
| 1599 |
+
"logits/chosen": -2.415342092514038,
|
| 1600 |
+
"logits/rejected": -2.3436694145202637,
|
| 1601 |
+
"logps/chosen": -305.41937255859375,
|
| 1602 |
+
"logps/rejected": -254.4953155517578,
|
| 1603 |
+
"loss": 1.3252,
|
| 1604 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1605 |
+
"rewards/chosen": 0.03756101056933403,
|
| 1606 |
+
"rewards/margins": 0.33726876974105835,
|
| 1607 |
+
"rewards/rejected": -0.2997077405452728,
|
| 1608 |
+
"step": 960
|
| 1609 |
+
},
|
| 1610 |
+
{
|
| 1611 |
+
"epoch": 0.6346090938828918,
|
| 1612 |
+
"grad_norm": 7.09375,
|
| 1613 |
+
"learning_rate": 1.7710435200834126e-06,
|
| 1614 |
+
"logits/chosen": -2.424187660217285,
|
| 1615 |
+
"logits/rejected": -2.4013800621032715,
|
| 1616 |
+
"logps/chosen": -253.3881378173828,
|
| 1617 |
+
"logps/rejected": -242.24853515625,
|
| 1618 |
+
"loss": 1.3294,
|
| 1619 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1620 |
+
"rewards/chosen": 0.09904266893863678,
|
| 1621 |
+
"rewards/margins": 0.44573745131492615,
|
| 1622 |
+
"rewards/rejected": -0.3466947376728058,
|
| 1623 |
+
"step": 970
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 0.6411514556754988,
|
| 1627 |
+
"grad_norm": 10.1875,
|
| 1628 |
+
"learning_rate": 1.7166007734112808e-06,
|
| 1629 |
+
"logits/chosen": -2.512207269668579,
|
| 1630 |
+
"logits/rejected": -2.3953731060028076,
|
| 1631 |
+
"logps/chosen": -282.1387939453125,
|
| 1632 |
+
"logps/rejected": -238.2812042236328,
|
| 1633 |
+
"loss": 1.3447,
|
| 1634 |
+
"rewards/accuracies": 0.75,
|
| 1635 |
+
"rewards/chosen": 0.10133460909128189,
|
| 1636 |
+
"rewards/margins": 0.4784146845340729,
|
| 1637 |
+
"rewards/rejected": -0.3770800232887268,
|
| 1638 |
+
"step": 980
|
| 1639 |
+
},
|
| 1640 |
+
{
|
| 1641 |
+
"epoch": 0.647693817468106,
|
| 1642 |
+
"grad_norm": 6.84375,
|
| 1643 |
+
"learning_rate": 1.6625669657993483e-06,
|
| 1644 |
+
"logits/chosen": -2.588611125946045,
|
| 1645 |
+
"logits/rejected": -2.5047097206115723,
|
| 1646 |
+
"logps/chosen": -316.20037841796875,
|
| 1647 |
+
"logps/rejected": -254.8091278076172,
|
| 1648 |
+
"loss": 1.3237,
|
| 1649 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1650 |
+
"rewards/chosen": 0.18245932459831238,
|
| 1651 |
+
"rewards/margins": 0.5891641974449158,
|
| 1652 |
+
"rewards/rejected": -0.40670496225357056,
|
| 1653 |
+
"step": 990
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 0.6542361792607131,
|
| 1657 |
+
"grad_norm": 6.90625,
|
| 1658 |
+
"learning_rate": 1.6089703032168736e-06,
|
| 1659 |
+
"logits/chosen": -2.417908191680908,
|
| 1660 |
+
"logits/rejected": -2.41746187210083,
|
| 1661 |
+
"logps/chosen": -254.3357391357422,
|
| 1662 |
+
"logps/rejected": -241.4535675048828,
|
| 1663 |
+
"loss": 1.3388,
|
| 1664 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1665 |
+
"rewards/chosen": 0.06749463826417923,
|
| 1666 |
+
"rewards/margins": 0.35123783349990845,
|
| 1667 |
+
"rewards/rejected": -0.2837432026863098,
|
| 1668 |
+
"step": 1000
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 0.6542361792607131,
|
| 1672 |
+
"eval_logits/chosen": -2.318289041519165,
|
| 1673 |
+
"eval_logits/rejected": -2.238693952560425,
|
| 1674 |
+
"eval_logps/chosen": -270.0261535644531,
|
| 1675 |
+
"eval_logps/rejected": -256.7276611328125,
|
| 1676 |
+
"eval_loss": 1.344198226928711,
|
| 1677 |
+
"eval_rewards/accuracies": 0.7325000166893005,
|
| 1678 |
+
"eval_rewards/chosen": 0.215036541223526,
|
| 1679 |
+
"eval_rewards/margins": 0.4775146245956421,
|
| 1680 |
+
"eval_rewards/rejected": -0.2624781131744385,
|
| 1681 |
+
"eval_runtime": 192.5272,
|
| 1682 |
+
"eval_samples_per_second": 10.388,
|
| 1683 |
+
"eval_steps_per_second": 0.519,
|
| 1684 |
+
"step": 1000
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 0.6607785410533202,
|
| 1688 |
+
"grad_norm": 7.8125,
|
| 1689 |
+
"learning_rate": 1.55583876344081e-06,
|
| 1690 |
+
"logits/chosen": -2.4103312492370605,
|
| 1691 |
+
"logits/rejected": -2.1900105476379395,
|
| 1692 |
+
"logps/chosen": -260.7649230957031,
|
| 1693 |
+
"logps/rejected": -209.84048461914062,
|
| 1694 |
+
"loss": 1.3325,
|
| 1695 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1696 |
+
"rewards/chosen": 0.10994074493646622,
|
| 1697 |
+
"rewards/margins": 0.5784383416175842,
|
| 1698 |
+
"rewards/rejected": -0.46849751472473145,
|
| 1699 |
+
"step": 1010
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 0.6673209028459274,
|
| 1703 |
+
"grad_norm": 7.3125,
|
| 1704 |
+
"learning_rate": 1.5032000814512372e-06,
|
| 1705 |
+
"logits/chosen": -2.4817819595336914,
|
| 1706 |
+
"logits/rejected": -2.4498400688171387,
|
| 1707 |
+
"logps/chosen": -240.53134155273438,
|
| 1708 |
+
"logps/rejected": -238.8572998046875,
|
| 1709 |
+
"loss": 1.3307,
|
| 1710 |
+
"rewards/accuracies": 0.6875,
|
| 1711 |
+
"rewards/chosen": -0.0427989736199379,
|
| 1712 |
+
"rewards/margins": 0.3493194580078125,
|
| 1713 |
+
"rewards/rejected": -0.3921184539794922,
|
| 1714 |
+
"step": 1020
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 0.6738632646385345,
|
| 1718 |
+
"grad_norm": 5.59375,
|
| 1719 |
+
"learning_rate": 1.4510817349535323e-06,
|
| 1720 |
+
"logits/chosen": -2.453907012939453,
|
| 1721 |
+
"logits/rejected": -2.399127244949341,
|
| 1722 |
+
"logps/chosen": -262.92559814453125,
|
| 1723 |
+
"logps/rejected": -249.8006134033203,
|
| 1724 |
+
"loss": 1.3411,
|
| 1725 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1726 |
+
"rewards/chosen": 0.13903044164180756,
|
| 1727 |
+
"rewards/margins": 0.5323747396469116,
|
| 1728 |
+
"rewards/rejected": -0.39334431290626526,
|
| 1729 |
+
"step": 1030
|
| 1730 |
+
},
|
| 1731 |
+
{
|
| 1732 |
+
"epoch": 0.6804056264311417,
|
| 1733 |
+
"grad_norm": 8.25,
|
| 1734 |
+
"learning_rate": 1.3995109300348537e-06,
|
| 1735 |
+
"logits/chosen": -2.5314221382141113,
|
| 1736 |
+
"logits/rejected": -2.424898624420166,
|
| 1737 |
+
"logps/chosen": -296.252197265625,
|
| 1738 |
+
"logps/rejected": -279.55096435546875,
|
| 1739 |
+
"loss": 1.3372,
|
| 1740 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1741 |
+
"rewards/chosen": -0.001908986596390605,
|
| 1742 |
+
"rewards/margins": 0.4961877763271332,
|
| 1743 |
+
"rewards/rejected": -0.49809688329696655,
|
| 1744 |
+
"step": 1040
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 0.6869479882237488,
|
| 1748 |
+
"grad_norm": 7.03125,
|
| 1749 |
+
"learning_rate": 1.348514586962389e-06,
|
| 1750 |
+
"logits/chosen": -2.543003559112549,
|
| 1751 |
+
"logits/rejected": -2.466203212738037,
|
| 1752 |
+
"logps/chosen": -244.98037719726562,
|
| 1753 |
+
"logps/rejected": -231.51449584960938,
|
| 1754 |
+
"loss": 1.3373,
|
| 1755 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1756 |
+
"rewards/chosen": 0.1332973688840866,
|
| 1757 |
+
"rewards/margins": 0.5175554752349854,
|
| 1758 |
+
"rewards/rejected": -0.38425812125205994,
|
| 1759 |
+
"step": 1050
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 0.693490350016356,
|
| 1763 |
+
"grad_norm": 6.625,
|
| 1764 |
+
"learning_rate": 1.2981193261308284e-06,
|
| 1765 |
+
"logits/chosen": -2.485957622528076,
|
| 1766 |
+
"logits/rejected": -2.440640687942505,
|
| 1767 |
+
"logps/chosen": -243.56118774414062,
|
| 1768 |
+
"logps/rejected": -264.2271728515625,
|
| 1769 |
+
"loss": 1.3184,
|
| 1770 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1771 |
+
"rewards/chosen": 0.058649301528930664,
|
| 1772 |
+
"rewards/margins": 0.4494194984436035,
|
| 1773 |
+
"rewards/rejected": -0.39077019691467285,
|
| 1774 |
+
"step": 1060
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.700032711808963,
|
| 1778 |
+
"grad_norm": 7.625,
|
| 1779 |
+
"learning_rate": 1.2483514541663501e-06,
|
| 1780 |
+
"logits/chosen": -2.567582607269287,
|
| 1781 |
+
"logits/rejected": -2.398489475250244,
|
| 1782 |
+
"logps/chosen": -268.79248046875,
|
| 1783 |
+
"logps/rejected": -236.4168701171875,
|
| 1784 |
+
"loss": 1.3346,
|
| 1785 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1786 |
+
"rewards/chosen": 0.1122722402215004,
|
| 1787 |
+
"rewards/margins": 0.54461270570755,
|
| 1788 |
+
"rewards/rejected": -0.43234047293663025,
|
| 1789 |
+
"step": 1070
|
| 1790 |
+
},
|
| 1791 |
+
{
|
| 1792 |
+
"epoch": 0.7065750736015701,
|
| 1793 |
+
"grad_norm": 9.875,
|
| 1794 |
+
"learning_rate": 1.1992369501944096e-06,
|
| 1795 |
+
"logits/chosen": -2.6160035133361816,
|
| 1796 |
+
"logits/rejected": -2.485849380493164,
|
| 1797 |
+
"logps/chosen": -301.0981140136719,
|
| 1798 |
+
"logps/rejected": -261.76226806640625,
|
| 1799 |
+
"loss": 1.3344,
|
| 1800 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1801 |
+
"rewards/chosen": 0.12587206065654755,
|
| 1802 |
+
"rewards/margins": 0.49982064962387085,
|
| 1803 |
+
"rewards/rejected": -0.3739486336708069,
|
| 1804 |
+
"step": 1080
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 0.7131174353941773,
|
| 1808 |
+
"grad_norm": 5.15625,
|
| 1809 |
+
"learning_rate": 1.1508014522784803e-06,
|
| 1810 |
+
"logits/chosen": -2.5031826496124268,
|
| 1811 |
+
"logits/rejected": -2.380375862121582,
|
| 1812 |
+
"logps/chosen": -259.3973693847656,
|
| 1813 |
+
"logps/rejected": -242.9427490234375,
|
| 1814 |
+
"loss": 1.3325,
|
| 1815 |
+
"rewards/accuracies": 0.75,
|
| 1816 |
+
"rewards/chosen": 0.098679319024086,
|
| 1817 |
+
"rewards/margins": 0.5007960796356201,
|
| 1818 |
+
"rewards/rejected": -0.40211671590805054,
|
| 1819 |
+
"step": 1090
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 0.7196597971867844,
|
| 1823 |
+
"grad_norm": 6.3125,
|
| 1824 |
+
"learning_rate": 1.1030702440368319e-06,
|
| 1825 |
+
"logits/chosen": -2.4038383960723877,
|
| 1826 |
+
"logits/rejected": -2.402168035507202,
|
| 1827 |
+
"logps/chosen": -282.268798828125,
|
| 1828 |
+
"logps/rejected": -266.29132080078125,
|
| 1829 |
+
"loss": 1.3186,
|
| 1830 |
+
"rewards/accuracies": 0.8125,
|
| 1831 |
+
"rewards/chosen": 0.16258536279201508,
|
| 1832 |
+
"rewards/margins": 0.7569273114204407,
|
| 1833 |
+
"rewards/rejected": -0.5943418741226196,
|
| 1834 |
+
"step": 1100
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 0.7196597971867844,
|
| 1838 |
+
"eval_logits/chosen": -2.3106985092163086,
|
| 1839 |
+
"eval_logits/rejected": -2.2305588722229004,
|
| 1840 |
+
"eval_logps/chosen": -270.9344787597656,
|
| 1841 |
+
"eval_logps/rejected": -257.6895446777344,
|
| 1842 |
+
"eval_loss": 1.3422751426696777,
|
| 1843 |
+
"eval_rewards/accuracies": 0.7325000166893005,
|
| 1844 |
+
"eval_rewards/chosen": 0.12420222908258438,
|
| 1845 |
+
"eval_rewards/margins": 0.48286232352256775,
|
| 1846 |
+
"eval_rewards/rejected": -0.3586600422859192,
|
| 1847 |
+
"eval_runtime": 192.5343,
|
| 1848 |
+
"eval_samples_per_second": 10.388,
|
| 1849 |
+
"eval_steps_per_second": 0.519,
|
| 1850 |
+
"step": 1100
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.7262021589793916,
|
| 1854 |
+
"grad_norm": 8.0625,
|
| 1855 |
+
"learning_rate": 1.0560682414443315e-06,
|
| 1856 |
+
"logits/chosen": -2.492471694946289,
|
| 1857 |
+
"logits/rejected": -2.3619418144226074,
|
| 1858 |
+
"logps/chosen": -296.5441589355469,
|
| 1859 |
+
"logps/rejected": -249.51174926757812,
|
| 1860 |
+
"loss": 1.333,
|
| 1861 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1862 |
+
"rewards/chosen": 0.08190511167049408,
|
| 1863 |
+
"rewards/margins": 0.522915244102478,
|
| 1864 |
+
"rewards/rejected": -0.44101008772850037,
|
| 1865 |
+
"step": 1110
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 0.7327445207719987,
|
| 1869 |
+
"grad_norm": 7.03125,
|
| 1870 |
+
"learning_rate": 1.009819979826156e-06,
|
| 1871 |
+
"logits/chosen": -2.49381947517395,
|
| 1872 |
+
"logits/rejected": -2.4158735275268555,
|
| 1873 |
+
"logps/chosen": -277.04742431640625,
|
| 1874 |
+
"logps/rejected": -277.2239074707031,
|
| 1875 |
+
"loss": 1.3209,
|
| 1876 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1877 |
+
"rewards/chosen": 0.1789507120847702,
|
| 1878 |
+
"rewards/margins": 0.5798290967941284,
|
| 1879 |
+
"rewards/rejected": -0.4008784294128418,
|
| 1880 |
+
"step": 1120
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 0.7392868825646058,
|
| 1884 |
+
"grad_norm": 9.875,
|
| 1885 |
+
"learning_rate": 9.643496010502054e-07,
|
| 1886 |
+
"logits/chosen": -2.383232831954956,
|
| 1887 |
+
"logits/rejected": -2.3699889183044434,
|
| 1888 |
+
"logps/chosen": -282.37872314453125,
|
| 1889 |
+
"logps/rejected": -288.27764892578125,
|
| 1890 |
+
"loss": 1.3569,
|
| 1891 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1892 |
+
"rewards/chosen": 0.049896519631147385,
|
| 1893 |
+
"rewards/margins": 0.3302707076072693,
|
| 1894 |
+
"rewards/rejected": -0.280374139547348,
|
| 1895 |
+
"step": 1130
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 0.745829244357213,
|
| 1899 |
+
"grad_norm": 7.75,
|
| 1900 |
+
"learning_rate": 9.196808409249086e-07,
|
| 1901 |
+
"logits/chosen": -2.4595417976379395,
|
| 1902 |
+
"logits/rejected": -2.346752882003784,
|
| 1903 |
+
"logps/chosen": -263.95172119140625,
|
| 1904 |
+
"logps/rejected": -211.8850555419922,
|
| 1905 |
+
"loss": 1.3457,
|
| 1906 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1907 |
+
"rewards/chosen": 0.07734758406877518,
|
| 1908 |
+
"rewards/margins": 0.34833234548568726,
|
| 1909 |
+
"rewards/rejected": -0.2709847092628479,
|
| 1910 |
+
"step": 1140
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 0.7523716061498201,
|
| 1914 |
+
"grad_norm": 8.25,
|
| 1915 |
+
"learning_rate": 8.758370168089797e-07,
|
| 1916 |
+
"logits/chosen": -2.46266770362854,
|
| 1917 |
+
"logits/rejected": -2.4414191246032715,
|
| 1918 |
+
"logps/chosen": -258.66815185546875,
|
| 1919 |
+
"logps/rejected": -272.05242919921875,
|
| 1920 |
+
"loss": 1.3453,
|
| 1921 |
+
"rewards/accuracies": 0.75,
|
| 1922 |
+
"rewards/chosen": 0.12471766769886017,
|
| 1923 |
+
"rewards/margins": 0.5029908418655396,
|
| 1924 |
+
"rewards/rejected": -0.3782731890678406,
|
| 1925 |
+
"step": 1150
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 0.7589139679424273,
|
| 1929 |
+
"grad_norm": 7.90625,
|
| 1930 |
+
"learning_rate": 8.328410154396318e-07,
|
| 1931 |
+
"logits/chosen": -2.473407030105591,
|
| 1932 |
+
"logits/rejected": -2.4644668102264404,
|
| 1933 |
+
"logps/chosen": -270.7806396484375,
|
| 1934 |
+
"logps/rejected": -228.4518280029297,
|
| 1935 |
+
"loss": 1.3378,
|
| 1936 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1937 |
+
"rewards/chosen": 0.08448322862386703,
|
| 1938 |
+
"rewards/margins": 0.4490872323513031,
|
| 1939 |
+
"rewards/rejected": -0.36460399627685547,
|
| 1940 |
+
"step": 1160
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 0.7654563297350343,
|
| 1944 |
+
"grad_norm": 6.28125,
|
| 1945 |
+
"learning_rate": 7.907152809855529e-07,
|
| 1946 |
+
"logits/chosen": -2.4258522987365723,
|
| 1947 |
+
"logits/rejected": -2.3456382751464844,
|
| 1948 |
+
"logps/chosen": -260.29730224609375,
|
| 1949 |
+
"logps/rejected": -251.9761199951172,
|
| 1950 |
+
"loss": 1.3517,
|
| 1951 |
+
"rewards/accuracies": 0.625,
|
| 1952 |
+
"rewards/chosen": 0.03364313393831253,
|
| 1953 |
+
"rewards/margins": 0.29857534170150757,
|
| 1954 |
+
"rewards/rejected": -0.26493218541145325,
|
| 1955 |
+
"step": 1170
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 0.7719986915276414,
|
| 1959 |
+
"grad_norm": 11.1875,
|
| 1960 |
+
"learning_rate": 7.494818033309207e-07,
|
| 1961 |
+
"logits/chosen": -2.4767394065856934,
|
| 1962 |
+
"logits/rejected": -2.3014492988586426,
|
| 1963 |
+
"logps/chosen": -238.3695526123047,
|
| 1964 |
+
"logps/rejected": -213.4428253173828,
|
| 1965 |
+
"loss": 1.3469,
|
| 1966 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1967 |
+
"rewards/chosen": 0.12686462700366974,
|
| 1968 |
+
"rewards/margins": 0.3980763554573059,
|
| 1969 |
+
"rewards/rejected": -0.27121174335479736,
|
| 1970 |
+
"step": 1180
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 0.7785410533202486,
|
| 1974 |
+
"grad_norm": 7.1875,
|
| 1975 |
+
"learning_rate": 7.091621065965521e-07,
|
| 1976 |
+
"logits/chosen": -2.454749584197998,
|
| 1977 |
+
"logits/rejected": -2.4374210834503174,
|
| 1978 |
+
"logps/chosen": -276.31488037109375,
|
| 1979 |
+
"logps/rejected": -310.75,
|
| 1980 |
+
"loss": 1.3416,
|
| 1981 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1982 |
+
"rewards/chosen": 0.041704945266246796,
|
| 1983 |
+
"rewards/margins": 0.38392168283462524,
|
| 1984 |
+
"rewards/rejected": -0.34221673011779785,
|
| 1985 |
+
"step": 1190
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 0.7850834151128557,
|
| 1989 |
+
"grad_norm": 7.34375,
|
| 1990 |
+
"learning_rate": 6.697772379041823e-07,
|
| 1991 |
+
"logits/chosen": -2.4566853046417236,
|
| 1992 |
+
"logits/rejected": -2.416715145111084,
|
| 1993 |
+
"logps/chosen": -245.9681396484375,
|
| 1994 |
+
"logps/rejected": -241.95858764648438,
|
| 1995 |
+
"loss": 1.3299,
|
| 1996 |
+
"rewards/accuracies": 0.6875,
|
| 1997 |
+
"rewards/chosen": 0.12603041529655457,
|
| 1998 |
+
"rewards/margins": 0.42295509576797485,
|
| 1999 |
+
"rewards/rejected": -0.2969246804714203,
|
| 2000 |
+
"step": 1200
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 0.7850834151128557,
|
| 2004 |
+
"eval_logits/chosen": -2.3078200817108154,
|
| 2005 |
+
"eval_logits/rejected": -2.2274787425994873,
|
| 2006 |
+
"eval_logps/chosen": -270.70892333984375,
|
| 2007 |
+
"eval_logps/rejected": -257.372802734375,
|
| 2008 |
+
"eval_loss": 1.3416602611541748,
|
| 2009 |
+
"eval_rewards/accuracies": 0.7425000071525574,
|
| 2010 |
+
"eval_rewards/chosen": 0.14675647020339966,
|
| 2011 |
+
"eval_rewards/margins": 0.4737465977668762,
|
| 2012 |
+
"eval_rewards/rejected": -0.3269902169704437,
|
| 2013 |
+
"eval_runtime": 192.5364,
|
| 2014 |
+
"eval_samples_per_second": 10.388,
|
| 2015 |
+
"eval_steps_per_second": 0.519,
|
| 2016 |
+
"step": 1200
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 0.7916257769054629,
|
| 2020 |
+
"grad_norm": 5.8125,
|
| 2021 |
+
"learning_rate": 6.313477563897466e-07,
|
| 2022 |
+
"logits/chosen": -2.516838550567627,
|
| 2023 |
+
"logits/rejected": -2.4841699600219727,
|
| 2024 |
+
"logps/chosen": -243.35372924804688,
|
| 2025 |
+
"logps/rejected": -246.02450561523438,
|
| 2026 |
+
"loss": 1.3459,
|
| 2027 |
+
"rewards/accuracies": 0.762499988079071,
|
| 2028 |
+
"rewards/chosen": 0.11812669038772583,
|
| 2029 |
+
"rewards/margins": 0.5516510009765625,
|
| 2030 |
+
"rewards/rejected": -0.43352431058883667,
|
| 2031 |
+
"step": 1210
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 0.79816813869807,
|
| 2035 |
+
"grad_norm": 21.5,
|
| 2036 |
+
"learning_rate": 5.9389372247138e-07,
|
| 2037 |
+
"logits/chosen": -2.493438482284546,
|
| 2038 |
+
"logits/rejected": -2.3540115356445312,
|
| 2039 |
+
"logps/chosen": -215.5032501220703,
|
| 2040 |
+
"logps/rejected": -212.7133026123047,
|
| 2041 |
+
"loss": 1.3261,
|
| 2042 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 2043 |
+
"rewards/chosen": 0.16954867541790009,
|
| 2044 |
+
"rewards/margins": 0.5659071803092957,
|
| 2045 |
+
"rewards/rejected": -0.39635851979255676,
|
| 2046 |
+
"step": 1220
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 0.8047105004906772,
|
| 2050 |
+
"grad_norm": 13.0625,
|
| 2051 |
+
"learning_rate": 5.574346873777714e-07,
|
| 2052 |
+
"logits/chosen": -2.4212934970855713,
|
| 2053 |
+
"logits/rejected": -2.4491488933563232,
|
| 2054 |
+
"logps/chosen": -249.1747589111328,
|
| 2055 |
+
"logps/rejected": -282.01385498046875,
|
| 2056 |
+
"loss": 1.3411,
|
| 2057 |
+
"rewards/accuracies": 0.75,
|
| 2058 |
+
"rewards/chosen": 0.17586642503738403,
|
| 2059 |
+
"rewards/margins": 0.467965304851532,
|
| 2060 |
+
"rewards/rejected": -0.29209887981414795,
|
| 2061 |
+
"step": 1230
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 0.8112528622832843,
|
| 2065 |
+
"grad_norm": 7.5625,
|
| 2066 |
+
"learning_rate": 5.219896829422927e-07,
|
| 2067 |
+
"logits/chosen": -2.491405487060547,
|
| 2068 |
+
"logits/rejected": -2.3750882148742676,
|
| 2069 |
+
"logps/chosen": -252.0865020751953,
|
| 2070 |
+
"logps/rejected": -231.3611602783203,
|
| 2071 |
+
"loss": 1.3306,
|
| 2072 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 2073 |
+
"rewards/chosen": 0.12007620185613632,
|
| 2074 |
+
"rewards/margins": 0.5494218468666077,
|
| 2075 |
+
"rewards/rejected": -0.42934560775756836,
|
| 2076 |
+
"step": 1240
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"epoch": 0.8177952240758914,
|
| 2080 |
+
"grad_norm": 7.5,
|
| 2081 |
+
"learning_rate": 4.875772116682817e-07,
|
| 2082 |
+
"logits/chosen": -2.359609842300415,
|
| 2083 |
+
"logits/rejected": -2.25789475440979,
|
| 2084 |
+
"logps/chosen": -228.32534790039062,
|
| 2085 |
+
"logps/rejected": -250.2344512939453,
|
| 2086 |
+
"loss": 1.3465,
|
| 2087 |
+
"rewards/accuracies": 0.699999988079071,
|
| 2088 |
+
"rewards/chosen": 0.0579691156744957,
|
| 2089 |
+
"rewards/margins": 0.4779117703437805,
|
| 2090 |
+
"rewards/rejected": -0.41994261741638184,
|
| 2091 |
+
"step": 1250
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 0.8243375858684985,
|
| 2095 |
+
"grad_norm": 8.125,
|
| 2096 |
+
"learning_rate": 4.542152370706149e-07,
|
| 2097 |
+
"logits/chosen": -2.395474672317505,
|
| 2098 |
+
"logits/rejected": -2.3001809120178223,
|
| 2099 |
+
"logps/chosen": -254.91268920898438,
|
| 2100 |
+
"logps/rejected": -258.6715087890625,
|
| 2101 |
+
"loss": 1.331,
|
| 2102 |
+
"rewards/accuracies": 0.675000011920929,
|
| 2103 |
+
"rewards/chosen": 0.06895018368959427,
|
| 2104 |
+
"rewards/margins": 0.4956684112548828,
|
| 2105 |
+
"rewards/rejected": -0.4267183244228363,
|
| 2106 |
+
"step": 1260
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 0.8308799476611056,
|
| 2110 |
+
"grad_norm": 6.25,
|
| 2111 |
+
"learning_rate": 4.2192117429865067e-07,
|
| 2112 |
+
"logits/chosen": -2.28416109085083,
|
| 2113 |
+
"logits/rejected": -2.2375693321228027,
|
| 2114 |
+
"logps/chosen": -239.7650909423828,
|
| 2115 |
+
"logps/rejected": -263.7681579589844,
|
| 2116 |
+
"loss": 1.3323,
|
| 2117 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 2118 |
+
"rewards/chosen": 0.11223635822534561,
|
| 2119 |
+
"rewards/margins": 0.42569345235824585,
|
| 2120 |
+
"rewards/rejected": -0.3134571313858032,
|
| 2121 |
+
"step": 1270
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 0.8374223094537128,
|
| 2125 |
+
"grad_norm": 6.96875,
|
| 2126 |
+
"learning_rate": 3.907118810454172e-07,
|
| 2127 |
+
"logits/chosen": -2.401803970336914,
|
| 2128 |
+
"logits/rejected": -2.3998618125915527,
|
| 2129 |
+
"logps/chosen": -274.4956359863281,
|
| 2130 |
+
"logps/rejected": -298.844970703125,
|
| 2131 |
+
"loss": 1.3443,
|
| 2132 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 2133 |
+
"rewards/chosen": 0.061115562915802,
|
| 2134 |
+
"rewards/margins": 0.4573536813259125,
|
| 2135 |
+
"rewards/rejected": -0.39623817801475525,
|
| 2136 |
+
"step": 1280
|
| 2137 |
+
},
|
| 2138 |
+
{
|
| 2139 |
+
"epoch": 0.8439646712463199,
|
| 2140 |
+
"grad_norm": 6.46875,
|
| 2141 |
+
"learning_rate": 3.6060364874779455e-07,
|
| 2142 |
+
"logits/chosen": -2.399519443511963,
|
| 2143 |
+
"logits/rejected": -2.386014461517334,
|
| 2144 |
+
"logps/chosen": -262.71044921875,
|
| 2145 |
+
"logps/rejected": -251.5861358642578,
|
| 2146 |
+
"loss": 1.351,
|
| 2147 |
+
"rewards/accuracies": 0.637499988079071,
|
| 2148 |
+
"rewards/chosen": 0.02161099575459957,
|
| 2149 |
+
"rewards/margins": 0.339490681886673,
|
| 2150 |
+
"rewards/rejected": -0.31787967681884766,
|
| 2151 |
+
"step": 1290
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 0.850507033038927,
|
| 2155 |
+
"grad_norm": 6.25,
|
| 2156 |
+
"learning_rate": 3.3161219408229026e-07,
|
| 2157 |
+
"logits/chosen": -2.5390801429748535,
|
| 2158 |
+
"logits/rejected": -2.4665403366088867,
|
| 2159 |
+
"logps/chosen": -261.7703552246094,
|
| 2160 |
+
"logps/rejected": -245.41683959960938,
|
| 2161 |
+
"loss": 1.3248,
|
| 2162 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 2163 |
+
"rewards/chosen": 0.11996345221996307,
|
| 2164 |
+
"rewards/margins": 0.4374065399169922,
|
| 2165 |
+
"rewards/rejected": -0.31744304299354553,
|
| 2166 |
+
"step": 1300
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 0.850507033038927,
|
| 2170 |
+
"eval_logits/chosen": -2.3104758262634277,
|
| 2171 |
+
"eval_logits/rejected": -2.2305808067321777,
|
| 2172 |
+
"eval_logps/chosen": -270.62164306640625,
|
| 2173 |
+
"eval_logps/rejected": -257.2347106933594,
|
| 2174 |
+
"eval_loss": 1.341292381286621,
|
| 2175 |
+
"eval_rewards/accuracies": 0.7524999976158142,
|
| 2176 |
+
"eval_rewards/chosen": 0.15548919141292572,
|
| 2177 |
+
"eval_rewards/margins": 0.4686690866947174,
|
| 2178 |
+
"eval_rewards/rejected": -0.3131798803806305,
|
| 2179 |
+
"eval_runtime": 192.5429,
|
| 2180 |
+
"eval_samples_per_second": 10.387,
|
| 2181 |
+
"eval_steps_per_second": 0.519,
|
| 2182 |
+
"step": 1300
|
| 2183 |
+
},
|
| 2184 |
+
{
|
| 2185 |
+
"epoch": 0.8570493948315342,
|
| 2186 |
+
"grad_norm": 6.34375,
|
| 2187 |
+
"learning_rate": 3.0375265076083796e-07,
|
| 2188 |
+
"logits/chosen": -2.5264501571655273,
|
| 2189 |
+
"logits/rejected": -2.514683246612549,
|
| 2190 |
+
"logps/chosen": -277.8392028808594,
|
| 2191 |
+
"logps/rejected": -238.16134643554688,
|
| 2192 |
+
"loss": 1.3384,
|
| 2193 |
+
"rewards/accuracies": 0.699999988079071,
|
| 2194 |
+
"rewards/chosen": 0.1312626153230667,
|
| 2195 |
+
"rewards/margins": 0.3692128658294678,
|
| 2196 |
+
"rewards/rejected": -0.23795023560523987,
|
| 2197 |
+
"step": 1310
|
| 2198 |
+
},
|
| 2199 |
+
{
|
| 2200 |
+
"epoch": 0.8635917566241413,
|
| 2201 |
+
"grad_norm": 7.1875,
|
| 2202 |
+
"learning_rate": 2.7703956163091153e-07,
|
| 2203 |
+
"logits/chosen": -2.4872682094573975,
|
| 2204 |
+
"logits/rejected": -2.399378538131714,
|
| 2205 |
+
"logps/chosen": -252.8599853515625,
|
| 2206 |
+
"logps/rejected": -243.26425170898438,
|
| 2207 |
+
"loss": 1.3411,
|
| 2208 |
+
"rewards/accuracies": 0.824999988079071,
|
| 2209 |
+
"rewards/chosen": 0.09871190041303635,
|
| 2210 |
+
"rewards/margins": 0.4442734718322754,
|
| 2211 |
+
"rewards/rejected": -0.3455616533756256,
|
| 2212 |
+
"step": 1320
|
| 2213 |
+
},
|
| 2214 |
+
{
|
| 2215 |
+
"epoch": 0.8701341184167485,
|
| 2216 |
+
"grad_norm": 7.71875,
|
| 2217 |
+
"learning_rate": 2.514868710840723e-07,
|
| 2218 |
+
"logits/chosen": -2.482853889465332,
|
| 2219 |
+
"logits/rejected": -2.420342445373535,
|
| 2220 |
+
"logps/chosen": -310.83123779296875,
|
| 2221 |
+
"logps/rejected": -267.9164123535156,
|
| 2222 |
+
"loss": 1.3399,
|
| 2223 |
+
"rewards/accuracies": 0.625,
|
| 2224 |
+
"rewards/chosen": 0.10147074609994888,
|
| 2225 |
+
"rewards/margins": 0.3591291904449463,
|
| 2226 |
+
"rewards/rejected": -0.2576584219932556,
|
| 2227 |
+
"step": 1330
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 0.8766764802093556,
|
| 2231 |
+
"grad_norm": 7.09375,
|
| 2232 |
+
"learning_rate": 2.271079177769117e-07,
|
| 2233 |
+
"logits/chosen": -2.4147746562957764,
|
| 2234 |
+
"logits/rejected": -2.420971632003784,
|
| 2235 |
+
"logps/chosen": -284.0067138671875,
|
| 2236 |
+
"logps/rejected": -307.20989990234375,
|
| 2237 |
+
"loss": 1.3301,
|
| 2238 |
+
"rewards/accuracies": 0.6875,
|
| 2239 |
+
"rewards/chosen": 0.12000395357608795,
|
| 2240 |
+
"rewards/margins": 0.36496061086654663,
|
| 2241 |
+
"rewards/rejected": -0.24495668709278107,
|
| 2242 |
+
"step": 1340
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 0.8832188420019627,
|
| 2246 |
+
"grad_norm": 6.53125,
|
| 2247 |
+
"learning_rate": 2.0391542766819456e-07,
|
| 2248 |
+
"logits/chosen": -2.473135471343994,
|
| 2249 |
+
"logits/rejected": -2.308255195617676,
|
| 2250 |
+
"logps/chosen": -278.90167236328125,
|
| 2251 |
+
"logps/rejected": -262.17694091796875,
|
| 2252 |
+
"loss": 1.3217,
|
| 2253 |
+
"rewards/accuracies": 0.800000011920929,
|
| 2254 |
+
"rewards/chosen": 0.08379217237234116,
|
| 2255 |
+
"rewards/margins": 0.525702953338623,
|
| 2256 |
+
"rewards/rejected": -0.4419107437133789,
|
| 2257 |
+
"step": 1350
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 0.8897612037945698,
|
| 2261 |
+
"grad_norm": 8.1875,
|
| 2262 |
+
"learning_rate": 1.8192150737583264e-07,
|
| 2263 |
+
"logits/chosen": -2.471273899078369,
|
| 2264 |
+
"logits/rejected": -2.355404853820801,
|
| 2265 |
+
"logps/chosen": -262.5338439941406,
|
| 2266 |
+
"logps/rejected": -240.73681640625,
|
| 2267 |
+
"loss": 1.3361,
|
| 2268 |
+
"rewards/accuracies": 0.675000011920929,
|
| 2269 |
+
"rewards/chosen": 0.20122098922729492,
|
| 2270 |
+
"rewards/margins": 0.45535793900489807,
|
| 2271 |
+
"rewards/rejected": -0.25413697957992554,
|
| 2272 |
+
"step": 1360
|
| 2273 |
+
},
|
| 2274 |
+
{
|
| 2275 |
+
"epoch": 0.8963035655871769,
|
| 2276 |
+
"grad_norm": 8.75,
|
| 2277 |
+
"learning_rate": 1.61137637857158e-07,
|
| 2278 |
+
"logits/chosen": -2.5680222511291504,
|
| 2279 |
+
"logits/rejected": -2.452162027359009,
|
| 2280 |
+
"logps/chosen": -331.45050048828125,
|
| 2281 |
+
"logps/rejected": -277.1683654785156,
|
| 2282 |
+
"loss": 1.3416,
|
| 2283 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 2284 |
+
"rewards/chosen": 0.10738255083560944,
|
| 2285 |
+
"rewards/margins": 0.42853283882141113,
|
| 2286 |
+
"rewards/rejected": -0.3211502432823181,
|
| 2287 |
+
"step": 1370
|
| 2288 |
+
},
|
| 2289 |
+
{
|
| 2290 |
+
"epoch": 0.9028459273797841,
|
| 2291 |
+
"grad_norm": 6.6875,
|
| 2292 |
+
"learning_rate": 1.415746684157951e-07,
|
| 2293 |
+
"logits/chosen": -2.469827890396118,
|
| 2294 |
+
"logits/rejected": -2.433368444442749,
|
| 2295 |
+
"logps/chosen": -222.18994140625,
|
| 2296 |
+
"logps/rejected": -224.70907592773438,
|
| 2297 |
+
"loss": 1.3334,
|
| 2298 |
+
"rewards/accuracies": 0.75,
|
| 2299 |
+
"rewards/chosen": 0.13235194981098175,
|
| 2300 |
+
"rewards/margins": 0.4484756886959076,
|
| 2301 |
+
"rewards/rejected": -0.3161238133907318,
|
| 2302 |
+
"step": 1380
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 0.9093882891723912,
|
| 2306 |
+
"grad_norm": 7.03125,
|
| 2307 |
+
"learning_rate": 1.232428110382586e-07,
|
| 2308 |
+
"logits/chosen": -2.569434642791748,
|
| 2309 |
+
"logits/rejected": -2.4922871589660645,
|
| 2310 |
+
"logps/chosen": -249.40328979492188,
|
| 2311 |
+
"logps/rejected": -248.40054321289062,
|
| 2312 |
+
"loss": 1.3386,
|
| 2313 |
+
"rewards/accuracies": 0.675000011920929,
|
| 2314 |
+
"rewards/chosen": -0.005796324461698532,
|
| 2315 |
+
"rewards/margins": 0.4979252815246582,
|
| 2316 |
+
"rewards/rejected": -0.5037215948104858,
|
| 2317 |
+
"step": 1390
|
| 2318 |
+
},
|
| 2319 |
+
{
|
| 2320 |
+
"epoch": 0.9159306509649984,
|
| 2321 |
+
"grad_norm": 6.9375,
|
| 2322 |
+
"learning_rate": 1.0615163506323856e-07,
|
| 2323 |
+
"logits/chosen": -2.4826645851135254,
|
| 2324 |
+
"logits/rejected": -2.3935964107513428,
|
| 2325 |
+
"logps/chosen": -256.60003662109375,
|
| 2326 |
+
"logps/rejected": -247.3723602294922,
|
| 2327 |
+
"loss": 1.3398,
|
| 2328 |
+
"rewards/accuracies": 0.6875,
|
| 2329 |
+
"rewards/chosen": 0.10640044510364532,
|
| 2330 |
+
"rewards/margins": 0.3504992723464966,
|
| 2331 |
+
"rewards/rejected": -0.24409881234169006,
|
| 2332 |
+
"step": 1400
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 0.9159306509649984,
|
| 2336 |
+
"eval_logits/chosen": -2.3116648197174072,
|
| 2337 |
+
"eval_logits/rejected": -2.2316980361938477,
|
| 2338 |
+
"eval_logps/chosen": -270.7674560546875,
|
| 2339 |
+
"eval_logps/rejected": -257.3534851074219,
|
| 2340 |
+
"eval_loss": 1.3414233922958374,
|
| 2341 |
+
"eval_rewards/accuracies": 0.7475000023841858,
|
| 2342 |
+
"eval_rewards/chosen": 0.14090880751609802,
|
| 2343 |
+
"eval_rewards/margins": 0.46596458554267883,
|
| 2344 |
+
"eval_rewards/rejected": -0.3250557780265808,
|
| 2345 |
+
"eval_runtime": 192.5164,
|
| 2346 |
+
"eval_samples_per_second": 10.389,
|
| 2347 |
+
"eval_steps_per_second": 0.519,
|
| 2348 |
+
"step": 1400
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 0.9224730127576055,
|
| 2352 |
+
"grad_norm": 6.5625,
|
| 2353 |
+
"learning_rate": 9.031006218634342e-08,
|
| 2354 |
+
"logits/chosen": -2.4582056999206543,
|
| 2355 |
+
"logits/rejected": -2.4208710193634033,
|
| 2356 |
+
"logps/chosen": -294.5101623535156,
|
| 2357 |
+
"logps/rejected": -322.93707275390625,
|
| 2358 |
+
"loss": 1.3372,
|
| 2359 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 2360 |
+
"rewards/chosen": 0.17300447821617126,
|
| 2361 |
+
"rewards/margins": 0.528673529624939,
|
| 2362 |
+
"rewards/rejected": -0.3556690812110901,
|
| 2363 |
+
"step": 1410
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 0.9290153745502127,
|
| 2367 |
+
"grad_norm": 7.46875,
|
| 2368 |
+
"learning_rate": 7.572636180292831e-08,
|
| 2369 |
+
"logits/chosen": -2.4200243949890137,
|
| 2370 |
+
"logits/rejected": -2.357060194015503,
|
| 2371 |
+
"logps/chosen": -254.03231811523438,
|
| 2372 |
+
"logps/rejected": -243.3819580078125,
|
| 2373 |
+
"loss": 1.3276,
|
| 2374 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 2375 |
+
"rewards/chosen": 0.08049002289772034,
|
| 2376 |
+
"rewards/margins": 0.46086424589157104,
|
| 2377 |
+
"rewards/rejected": -0.3803742825984955,
|
| 2378 |
+
"step": 1420
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 0.9355577363428198,
|
| 2382 |
+
"grad_norm": 7.4375,
|
| 2383 |
+
"learning_rate": 6.240814669141559e-08,
|
| 2384 |
+
"logits/chosen": -2.4541358947753906,
|
| 2385 |
+
"logits/rejected": -2.4223341941833496,
|
| 2386 |
+
"logps/chosen": -266.42156982421875,
|
| 2387 |
+
"logps/rejected": -304.84600830078125,
|
| 2388 |
+
"loss": 1.3296,
|
| 2389 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 2390 |
+
"rewards/chosen": 0.07571353018283844,
|
| 2391 |
+
"rewards/margins": 0.49027562141418457,
|
| 2392 |
+
"rewards/rejected": -0.4145621359348297,
|
| 2393 |
+
"step": 1430
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 0.9421000981354269,
|
| 2397 |
+
"grad_norm": 8.25,
|
| 2398 |
+
"learning_rate": 5.036236903938285e-08,
|
| 2399 |
+
"logits/chosen": -2.5123915672302246,
|
| 2400 |
+
"logits/rejected": -2.4232895374298096,
|
| 2401 |
+
"logps/chosen": -289.10601806640625,
|
| 2402 |
+
"logps/rejected": -232.24868774414062,
|
| 2403 |
+
"loss": 1.3312,
|
| 2404 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 2405 |
+
"rewards/chosen": 0.08486177772283554,
|
| 2406 |
+
"rewards/margins": 0.33679136633872986,
|
| 2407 |
+
"rewards/rejected": -0.2519296109676361,
|
| 2408 |
+
"step": 1440
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 0.948642459928034,
|
| 2412 |
+
"grad_norm": 6.90625,
|
| 2413 |
+
"learning_rate": 3.959531681447859e-08,
|
| 2414 |
+
"logits/chosen": -2.3544886112213135,
|
| 2415 |
+
"logits/rejected": -2.306253433227539,
|
| 2416 |
+
"logps/chosen": -259.6851501464844,
|
| 2417 |
+
"logps/rejected": -238.7523956298828,
|
| 2418 |
+
"loss": 1.3431,
|
| 2419 |
+
"rewards/accuracies": 0.699999988079071,
|
| 2420 |
+
"rewards/chosen": 0.04705673083662987,
|
| 2421 |
+
"rewards/margins": 0.3753374218940735,
|
| 2422 |
+
"rewards/rejected": -0.3282806873321533,
|
| 2423 |
+
"step": 1450
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 0.9551848217206411,
|
| 2427 |
+
"grad_norm": 7.03125,
|
| 2428 |
+
"learning_rate": 3.0112610482064544e-08,
|
| 2429 |
+
"logits/chosen": -2.586221933364868,
|
| 2430 |
+
"logits/rejected": -2.4770185947418213,
|
| 2431 |
+
"logps/chosen": -299.3462829589844,
|
| 2432 |
+
"logps/rejected": -263.6593017578125,
|
| 2433 |
+
"loss": 1.342,
|
| 2434 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 2435 |
+
"rewards/chosen": 0.11202242225408554,
|
| 2436 |
+
"rewards/margins": 0.4375079572200775,
|
| 2437 |
+
"rewards/rejected": -0.3254855275154114,
|
| 2438 |
+
"step": 1460
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 0.9617271835132483,
|
| 2442 |
+
"grad_norm": 6.96875,
|
| 2443 |
+
"learning_rate": 2.1919200071301715e-08,
|
| 2444 |
+
"logits/chosen": -2.5899202823638916,
|
| 2445 |
+
"logits/rejected": -2.45597767829895,
|
| 2446 |
+
"logps/chosen": -292.31683349609375,
|
| 2447 |
+
"logps/rejected": -233.8472900390625,
|
| 2448 |
+
"loss": 1.3513,
|
| 2449 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 2450 |
+
"rewards/chosen": 0.09856925159692764,
|
| 2451 |
+
"rewards/margins": 0.4128998816013336,
|
| 2452 |
+
"rewards/rejected": -0.31433066725730896,
|
| 2453 |
+
"step": 1470
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 0.9682695453058554,
|
| 2457 |
+
"grad_norm": 6.78125,
|
| 2458 |
+
"learning_rate": 1.50193625912029e-08,
|
| 2459 |
+
"logits/chosen": -2.3694183826446533,
|
| 2460 |
+
"logits/rejected": -2.3877651691436768,
|
| 2461 |
+
"logps/chosen": -229.5559539794922,
|
| 2462 |
+
"logps/rejected": -265.0757141113281,
|
| 2463 |
+
"loss": 1.3336,
|
| 2464 |
+
"rewards/accuracies": 0.737500011920929,
|
| 2465 |
+
"rewards/chosen": -0.014900955371558666,
|
| 2466 |
+
"rewards/margins": 0.37212496995925903,
|
| 2467 |
+
"rewards/rejected": -0.38702592253685,
|
| 2468 |
+
"step": 1480
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 0.9748119070984625,
|
| 2472 |
+
"grad_norm": 8.25,
|
| 2473 |
+
"learning_rate": 9.416699798010521e-09,
|
| 2474 |
+
"logits/chosen": -2.3672139644622803,
|
| 2475 |
+
"logits/rejected": -2.2716662883758545,
|
| 2476 |
+
"logps/chosen": -253.46322631835938,
|
| 2477 |
+
"logps/rejected": -208.71768188476562,
|
| 2478 |
+
"loss": 1.3194,
|
| 2479 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 2480 |
+
"rewards/chosen": 0.11514721810817719,
|
| 2481 |
+
"rewards/margins": 0.5433107018470764,
|
| 2482 |
+
"rewards/rejected": -0.42816343903541565,
|
| 2483 |
+
"step": 1490
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 0.9813542688910697,
|
| 2487 |
+
"grad_norm": 6.5,
|
| 2488 |
+
"learning_rate": 5.114136315058083e-09,
|
| 2489 |
+
"logits/chosen": -2.3912510871887207,
|
| 2490 |
+
"logits/rejected": -2.3454089164733887,
|
| 2491 |
+
"logps/chosen": -262.8589782714844,
|
| 2492 |
+
"logps/rejected": -278.1678161621094,
|
| 2493 |
+
"loss": 1.325,
|
| 2494 |
+
"rewards/accuracies": 0.800000011920929,
|
| 2495 |
+
"rewards/chosen": 0.12093720585107803,
|
| 2496 |
+
"rewards/margins": 0.5334186553955078,
|
| 2497 |
+
"rewards/rejected": -0.41248148679733276,
|
| 2498 |
+
"step": 1500
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 0.9813542688910697,
|
| 2502 |
+
"eval_logits/chosen": -2.313671827316284,
|
| 2503 |
+
"eval_logits/rejected": -2.23390793800354,
|
| 2504 |
+
"eval_logps/chosen": -270.74359130859375,
|
| 2505 |
+
"eval_logps/rejected": -257.3707275390625,
|
| 2506 |
+
"eval_loss": 1.3408766984939575,
|
| 2507 |
+
"eval_rewards/accuracies": 0.7475000023841858,
|
| 2508 |
+
"eval_rewards/chosen": 0.14329300820827484,
|
| 2509 |
+
"eval_rewards/margins": 0.47007519006729126,
|
| 2510 |
+
"eval_rewards/rejected": -0.3267821967601776,
|
| 2511 |
+
"eval_runtime": 192.5128,
|
| 2512 |
+
"eval_samples_per_second": 10.389,
|
| 2513 |
+
"eval_steps_per_second": 0.519,
|
| 2514 |
+
"step": 1500
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 0.9878966306836768,
|
| 2518 |
+
"grad_norm": 7.09375,
|
| 2519 |
+
"learning_rate": 2.113918106098345e-09,
|
| 2520 |
+
"logits/chosen": -2.456124782562256,
|
| 2521 |
+
"logits/rejected": -2.4603703022003174,
|
| 2522 |
+
"logps/chosen": -279.5404052734375,
|
| 2523 |
+
"logps/rejected": -286.7142333984375,
|
| 2524 |
+
"loss": 1.3363,
|
| 2525 |
+
"rewards/accuracies": 0.675000011920929,
|
| 2526 |
+
"rewards/chosen": 0.1024564728140831,
|
| 2527 |
+
"rewards/margins": 0.3759761452674866,
|
| 2528 |
+
"rewards/rejected": -0.2735196650028229,
|
| 2529 |
+
"step": 1510
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 0.994438992476284,
|
| 2533 |
+
"grad_norm": 8.6875,
|
| 2534 |
+
"learning_rate": 4.176113028983575e-10,
|
| 2535 |
+
"logits/chosen": -2.407866954803467,
|
| 2536 |
+
"logits/rejected": -2.355156421661377,
|
| 2537 |
+
"logps/chosen": -266.18231201171875,
|
| 2538 |
+
"logps/rejected": -265.04022216796875,
|
| 2539 |
+
"loss": 1.337,
|
| 2540 |
+
"rewards/accuracies": 0.675000011920929,
|
| 2541 |
+
"rewards/chosen": 0.07989536225795746,
|
| 2542 |
+
"rewards/margins": 0.4557567238807678,
|
| 2543 |
+
"rewards/rejected": -0.37586134672164917,
|
| 2544 |
+
"step": 1520
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 0.9996728819103696,
|
| 2548 |
+
"step": 1528,
|
| 2549 |
+
"total_flos": 0.0,
|
| 2550 |
+
"train_loss": 1.343712306771603,
|
| 2551 |
+
"train_runtime": 13958.9948,
|
| 2552 |
+
"train_samples_per_second": 4.38,
|
| 2553 |
+
"train_steps_per_second": 0.109
|
| 2554 |
+
}
|
| 2555 |
+
],
|
| 2556 |
+
"logging_steps": 10,
|
| 2557 |
+
"max_steps": 1528,
|
| 2558 |
+
"num_input_tokens_seen": 0,
|
| 2559 |
+
"num_train_epochs": 1,
|
| 2560 |
+
"save_steps": 100,
|
| 2561 |
+
"stateful_callbacks": {
|
| 2562 |
+
"TrainerControl": {
|
| 2563 |
+
"args": {
|
| 2564 |
+
"should_epoch_stop": false,
|
| 2565 |
+
"should_evaluate": false,
|
| 2566 |
+
"should_log": false,
|
| 2567 |
+
"should_save": true,
|
| 2568 |
+
"should_training_stop": true
|
| 2569 |
+
},
|
| 2570 |
+
"attributes": {}
|
| 2571 |
+
}
|
| 2572 |
+
},
|
| 2573 |
+
"total_flos": 0.0,
|
| 2574 |
+
"train_batch_size": 2,
|
| 2575 |
+
"trial_name": null,
|
| 2576 |
+
"trial_params": null
|
| 2577 |
+
}
|