Model save
Browse files- README.md +82 -0
- all_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +1294 -0
README.md
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: BramVanroy/GEITje-7B-ultra
|
| 3 |
+
library_name: peft
|
| 4 |
+
license: cc-by-nc-4.0
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- dpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: FinGEITje-7B-dpo
|
| 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/snoels/huggingface/runs/yng7mdb0)
|
| 18 |
+
# FinGEITje-7B-dpo
|
| 19 |
+
|
| 20 |
+
This model is a fine-tuned version of [BramVanroy/GEITje-7B-ultra](https://huggingface.co/BramVanroy/GEITje-7B-ultra) on an unknown dataset.
|
| 21 |
+
It achieves the following results on the evaluation set:
|
| 22 |
+
- Loss: 0.0279
|
| 23 |
+
- Rewards/chosen: -3.8974
|
| 24 |
+
- Rewards/rejected: -15.9642
|
| 25 |
+
- Rewards/accuracies: 0.9828
|
| 26 |
+
- Rewards/margins: 12.0668
|
| 27 |
+
- Logps/rejected: -1951.9310
|
| 28 |
+
- Logps/chosen: -788.9780
|
| 29 |
+
- Logits/rejected: -1.7371
|
| 30 |
+
- Logits/chosen: -1.8937
|
| 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: 1
|
| 51 |
+
- eval_batch_size: 1
|
| 52 |
+
- seed: 42
|
| 53 |
+
- distributed_type: multi-GPU
|
| 54 |
+
- num_devices: 4
|
| 55 |
+
- gradient_accumulation_steps: 16
|
| 56 |
+
- total_train_batch_size: 64
|
| 57 |
+
- total_eval_batch_size: 4
|
| 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 |
+
| 0.1029 | 0.1327 | 100 | 0.1099 | -1.8067 | -5.3683 | 0.9679 | 3.5616 | -892.3373 | -579.9115 | -2.4775 | -2.3705 |
|
| 68 |
+
| 0.042 | 0.2654 | 200 | 0.0430 | -3.5129 | -10.6778 | 0.9828 | 7.1649 | -1423.2883 | -750.5289 | -1.9744 | -1.9895 |
|
| 69 |
+
| 0.0278 | 0.3981 | 300 | 0.0344 | -3.7335 | -13.5153 | 0.9828 | 9.7818 | -1707.0360 | -772.5893 | -1.7454 | -1.8191 |
|
| 70 |
+
| 0.0223 | 0.5308 | 400 | 0.0308 | -3.6554 | -13.7712 | 0.9858 | 10.1158 | -1732.6289 | -764.7831 | -1.8020 | -1.9184 |
|
| 71 |
+
| 0.0378 | 0.6635 | 500 | 0.0297 | -4.0018 | -16.3285 | 0.9851 | 12.3266 | -1988.3542 | -799.4221 | -1.6924 | -1.8650 |
|
| 72 |
+
| 0.0352 | 0.7962 | 600 | 0.0278 | -3.8104 | -15.6430 | 0.9836 | 11.8327 | -1919.8119 | -780.2752 | -1.7437 | -1.8978 |
|
| 73 |
+
| 0.0238 | 0.9289 | 700 | 0.0279 | -3.8974 | -15.9642 | 0.9828 | 12.0668 | -1951.9310 | -788.9780 | -1.7371 | -1.8937 |
|
| 74 |
+
|
| 75 |
+
|
| 76 |
+
### Framework versions
|
| 77 |
+
|
| 78 |
+
- PEFT 0.11.1
|
| 79 |
+
- Transformers 4.42.4
|
| 80 |
+
- Pytorch 2.3.1
|
| 81 |
+
- Datasets 2.20.0
|
| 82 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.999253545658124,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.0882907345950366,
|
| 5 |
+
"train_runtime": 20220.3954,
|
| 6 |
+
"train_samples": 48227,
|
| 7 |
+
"train_samples_per_second": 2.385,
|
| 8 |
+
"train_steps_per_second": 0.037
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.999253545658124,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.0882907345950366,
|
| 5 |
+
"train_runtime": 20220.3954,
|
| 6 |
+
"train_samples": 48227,
|
| 7 |
+
"train_samples_per_second": 2.385,
|
| 8 |
+
"train_steps_per_second": 0.037
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1294 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.999253545658124,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 753,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0013270299411130464,
|
| 13 |
+
"grad_norm": 1.7421875,
|
| 14 |
+
"learning_rate": 6.578947368421053e-08,
|
| 15 |
+
"logits/chosen": -3.0980052947998047,
|
| 16 |
+
"logits/rejected": -3.127007007598877,
|
| 17 |
+
"logps/chosen": -425.274169921875,
|
| 18 |
+
"logps/rejected": -373.2780456542969,
|
| 19 |
+
"loss": 0.6931,
|
| 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": 0.013270299411130464,
|
| 28 |
+
"grad_norm": 1.765625,
|
| 29 |
+
"learning_rate": 6.578947368421053e-07,
|
| 30 |
+
"logits/chosen": -3.072850227355957,
|
| 31 |
+
"logits/rejected": -3.0915706157684326,
|
| 32 |
+
"logps/chosen": -421.3843994140625,
|
| 33 |
+
"logps/rejected": -350.215087890625,
|
| 34 |
+
"loss": 0.6919,
|
| 35 |
+
"rewards/accuracies": 0.6041666865348816,
|
| 36 |
+
"rewards/chosen": 0.0024290236178785563,
|
| 37 |
+
"rewards/margins": 0.0028885826468467712,
|
| 38 |
+
"rewards/rejected": -0.00045955937821418047,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.026540598822260928,
|
| 43 |
+
"grad_norm": 1.6875,
|
| 44 |
+
"learning_rate": 1.3157894736842106e-06,
|
| 45 |
+
"logits/chosen": -3.0872464179992676,
|
| 46 |
+
"logits/rejected": -3.0978829860687256,
|
| 47 |
+
"logps/chosen": -352.447021484375,
|
| 48 |
+
"logps/rejected": -364.5257873535156,
|
| 49 |
+
"loss": 0.6816,
|
| 50 |
+
"rewards/accuracies": 0.956250011920929,
|
| 51 |
+
"rewards/chosen": 0.015185330994427204,
|
| 52 |
+
"rewards/margins": 0.021979082375764847,
|
| 53 |
+
"rewards/rejected": -0.006793751381337643,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.039810898233391394,
|
| 58 |
+
"grad_norm": 1.53125,
|
| 59 |
+
"learning_rate": 1.973684210526316e-06,
|
| 60 |
+
"logits/chosen": -3.055328130722046,
|
| 61 |
+
"logits/rejected": -3.0872576236724854,
|
| 62 |
+
"logps/chosen": -374.4645080566406,
|
| 63 |
+
"logps/rejected": -364.3621520996094,
|
| 64 |
+
"loss": 0.6547,
|
| 65 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 66 |
+
"rewards/chosen": 0.05001773685216904,
|
| 67 |
+
"rewards/margins": 0.07953666895627975,
|
| 68 |
+
"rewards/rejected": -0.029518935829401016,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.053081197644521856,
|
| 73 |
+
"grad_norm": 1.6640625,
|
| 74 |
+
"learning_rate": 2.631578947368421e-06,
|
| 75 |
+
"logits/chosen": -3.054077625274658,
|
| 76 |
+
"logits/rejected": -3.090534210205078,
|
| 77 |
+
"logps/chosen": -394.09552001953125,
|
| 78 |
+
"logps/rejected": -394.32818603515625,
|
| 79 |
+
"loss": 0.611,
|
| 80 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 81 |
+
"rewards/chosen": 0.0969473272562027,
|
| 82 |
+
"rewards/margins": 0.17953996360301971,
|
| 83 |
+
"rewards/rejected": -0.08259265124797821,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.06635149705565231,
|
| 88 |
+
"grad_norm": 1.796875,
|
| 89 |
+
"learning_rate": 3.289473684210527e-06,
|
| 90 |
+
"logits/chosen": -3.0039563179016113,
|
| 91 |
+
"logits/rejected": -3.0520451068878174,
|
| 92 |
+
"logps/chosen": -376.13897705078125,
|
| 93 |
+
"logps/rejected": -405.67303466796875,
|
| 94 |
+
"loss": 0.5412,
|
| 95 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 96 |
+
"rewards/chosen": 0.12179754674434662,
|
| 97 |
+
"rewards/margins": 0.3453168272972107,
|
| 98 |
+
"rewards/rejected": -0.22351928055286407,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.07962179646678279,
|
| 103 |
+
"grad_norm": 1.7890625,
|
| 104 |
+
"learning_rate": 3.947368421052632e-06,
|
| 105 |
+
"logits/chosen": -2.974642515182495,
|
| 106 |
+
"logits/rejected": -3.0201048851013184,
|
| 107 |
+
"logps/chosen": -418.08135986328125,
|
| 108 |
+
"logps/rejected": -455.23846435546875,
|
| 109 |
+
"loss": 0.4179,
|
| 110 |
+
"rewards/accuracies": 0.987500011920929,
|
| 111 |
+
"rewards/chosen": 0.048082564026117325,
|
| 112 |
+
"rewards/margins": 0.6676187515258789,
|
| 113 |
+
"rewards/rejected": -0.6195362210273743,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.09289209587791325,
|
| 118 |
+
"grad_norm": 1.34375,
|
| 119 |
+
"learning_rate": 4.605263157894737e-06,
|
| 120 |
+
"logits/chosen": -2.86934757232666,
|
| 121 |
+
"logits/rejected": -2.9434354305267334,
|
| 122 |
+
"logps/chosen": -400.07568359375,
|
| 123 |
+
"logps/rejected": -482.0399475097656,
|
| 124 |
+
"loss": 0.3225,
|
| 125 |
+
"rewards/accuracies": 0.981249988079071,
|
| 126 |
+
"rewards/chosen": -0.1442013680934906,
|
| 127 |
+
"rewards/margins": 1.0631463527679443,
|
| 128 |
+
"rewards/rejected": -1.2073477506637573,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.10616239528904371,
|
| 133 |
+
"grad_norm": 1.359375,
|
| 134 |
+
"learning_rate": 4.9995693346469565e-06,
|
| 135 |
+
"logits/chosen": -2.6817970275878906,
|
| 136 |
+
"logits/rejected": -2.778237819671631,
|
| 137 |
+
"logps/chosen": -460.60394287109375,
|
| 138 |
+
"logps/rejected": -621.35498046875,
|
| 139 |
+
"loss": 0.2331,
|
| 140 |
+
"rewards/accuracies": 0.987500011920929,
|
| 141 |
+
"rewards/chosen": -0.620966374874115,
|
| 142 |
+
"rewards/margins": 1.6919386386871338,
|
| 143 |
+
"rewards/rejected": -2.3129050731658936,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.11943269470017417,
|
| 148 |
+
"grad_norm": 1.8671875,
|
| 149 |
+
"learning_rate": 4.994726053293703e-06,
|
| 150 |
+
"logits/chosen": -2.5432372093200684,
|
| 151 |
+
"logits/rejected": -2.636983633041382,
|
| 152 |
+
"logps/chosen": -504.8564453125,
|
| 153 |
+
"logps/rejected": -734.1746826171875,
|
| 154 |
+
"loss": 0.1809,
|
| 155 |
+
"rewards/accuracies": 0.981249988079071,
|
| 156 |
+
"rewards/chosen": -1.099843978881836,
|
| 157 |
+
"rewards/margins": 2.6429855823516846,
|
| 158 |
+
"rewards/rejected": -3.7428295612335205,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.13270299411130462,
|
| 163 |
+
"grad_norm": 1.140625,
|
| 164 |
+
"learning_rate": 4.984511621268103e-06,
|
| 165 |
+
"logits/chosen": -2.420698642730713,
|
| 166 |
+
"logits/rejected": -2.541872501373291,
|
| 167 |
+
"logps/chosen": -544.0470581054688,
|
| 168 |
+
"logps/rejected": -866.2991943359375,
|
| 169 |
+
"loss": 0.1029,
|
| 170 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 171 |
+
"rewards/chosen": -1.5227975845336914,
|
| 172 |
+
"rewards/margins": 3.385746479034424,
|
| 173 |
+
"rewards/rejected": -4.908544063568115,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.13270299411130462,
|
| 178 |
+
"eval_logits/chosen": -2.3704655170440674,
|
| 179 |
+
"eval_logits/rejected": -2.4775025844573975,
|
| 180 |
+
"eval_logps/chosen": -579.9114990234375,
|
| 181 |
+
"eval_logps/rejected": -892.3373413085938,
|
| 182 |
+
"eval_loss": 0.10992002487182617,
|
| 183 |
+
"eval_rewards/accuracies": 0.9679104685783386,
|
| 184 |
+
"eval_rewards/chosen": -1.8067275285720825,
|
| 185 |
+
"eval_rewards/margins": 3.5615758895874023,
|
| 186 |
+
"eval_rewards/rejected": -5.368303298950195,
|
| 187 |
+
"eval_runtime": 828.869,
|
| 188 |
+
"eval_samples_per_second": 6.465,
|
| 189 |
+
"eval_steps_per_second": 1.617,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.1459732935224351,
|
| 194 |
+
"grad_norm": 1.2421875,
|
| 195 |
+
"learning_rate": 4.968948030264743e-06,
|
| 196 |
+
"logits/chosen": -2.349208354949951,
|
| 197 |
+
"logits/rejected": -2.439763069152832,
|
| 198 |
+
"logps/chosen": -572.8814697265625,
|
| 199 |
+
"logps/rejected": -938.94091796875,
|
| 200 |
+
"loss": 0.085,
|
| 201 |
+
"rewards/accuracies": 0.987500011920929,
|
| 202 |
+
"rewards/chosen": -1.6822025775909424,
|
| 203 |
+
"rewards/margins": 4.082047939300537,
|
| 204 |
+
"rewards/rejected": -5.764250755310059,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.15924359293356558,
|
| 209 |
+
"grad_norm": 2.046875,
|
| 210 |
+
"learning_rate": 4.948068788729238e-06,
|
| 211 |
+
"logits/chosen": -2.156416177749634,
|
| 212 |
+
"logits/rejected": -2.215620756149292,
|
| 213 |
+
"logps/chosen": -626.0386962890625,
|
| 214 |
+
"logps/rejected": -1036.8035888671875,
|
| 215 |
+
"loss": 0.0874,
|
| 216 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 217 |
+
"rewards/chosen": -2.1521973609924316,
|
| 218 |
+
"rewards/margins": 4.6609344482421875,
|
| 219 |
+
"rewards/rejected": -6.813131809234619,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.17251389234469602,
|
| 224 |
+
"grad_norm": 1.8515625,
|
| 225 |
+
"learning_rate": 4.921918849714475e-06,
|
| 226 |
+
"logits/chosen": -2.1903834342956543,
|
| 227 |
+
"logits/rejected": -2.2045347690582275,
|
| 228 |
+
"logps/chosen": -673.7794189453125,
|
| 229 |
+
"logps/rejected": -1226.3961181640625,
|
| 230 |
+
"loss": 0.0559,
|
| 231 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 232 |
+
"rewards/chosen": -2.5693655014038086,
|
| 233 |
+
"rewards/margins": 5.772281646728516,
|
| 234 |
+
"rewards/rejected": -8.341647148132324,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.1857841917558265,
|
| 239 |
+
"grad_norm": 1.0078125,
|
| 240 |
+
"learning_rate": 4.890554514096592e-06,
|
| 241 |
+
"logits/chosen": -2.1850454807281494,
|
| 242 |
+
"logits/rejected": -2.19503116607666,
|
| 243 |
+
"logps/chosen": -688.4675903320312,
|
| 244 |
+
"logps/rejected": -1202.0238037109375,
|
| 245 |
+
"loss": 0.0699,
|
| 246 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 247 |
+
"rewards/chosen": -2.7659270763397217,
|
| 248 |
+
"rewards/margins": 5.731719970703125,
|
| 249 |
+
"rewards/rejected": -8.497647285461426,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.19905449116695695,
|
| 254 |
+
"grad_norm": 4.96875,
|
| 255 |
+
"learning_rate": 4.854043309359063e-06,
|
| 256 |
+
"logits/chosen": -2.045557737350464,
|
| 257 |
+
"logits/rejected": -2.022369146347046,
|
| 258 |
+
"logps/chosen": -797.4768676757812,
|
| 259 |
+
"logps/rejected": -1449.0230712890625,
|
| 260 |
+
"loss": 0.0569,
|
| 261 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 262 |
+
"rewards/chosen": -3.9922683238983154,
|
| 263 |
+
"rewards/margins": 6.881680488586426,
|
| 264 |
+
"rewards/rejected": -10.87394905090332,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.21232479057808742,
|
| 269 |
+
"grad_norm": 2.5,
|
| 270 |
+
"learning_rate": 4.8124638442058856e-06,
|
| 271 |
+
"logits/chosen": -2.0948424339294434,
|
| 272 |
+
"logits/rejected": -2.136207103729248,
|
| 273 |
+
"logps/chosen": -731.327392578125,
|
| 274 |
+
"logps/rejected": -1373.2447509765625,
|
| 275 |
+
"loss": 0.05,
|
| 276 |
+
"rewards/accuracies": 0.987500011920929,
|
| 277 |
+
"rewards/chosen": -3.1236228942871094,
|
| 278 |
+
"rewards/margins": 6.655406951904297,
|
| 279 |
+
"rewards/rejected": -9.77902889251709,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.22559508998921787,
|
| 284 |
+
"grad_norm": 1.265625,
|
| 285 |
+
"learning_rate": 4.765905639316861e-06,
|
| 286 |
+
"logits/chosen": -1.9219977855682373,
|
| 287 |
+
"logits/rejected": -1.9046119451522827,
|
| 288 |
+
"logps/chosen": -720.5792236328125,
|
| 289 |
+
"logps/rejected": -1352.9737548828125,
|
| 290 |
+
"loss": 0.0472,
|
| 291 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 292 |
+
"rewards/chosen": -3.072010040283203,
|
| 293 |
+
"rewards/margins": 6.776447296142578,
|
| 294 |
+
"rewards/rejected": -9.848457336425781,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.23886538940034835,
|
| 299 |
+
"grad_norm": 2.234375,
|
| 300 |
+
"learning_rate": 4.7144689346093814e-06,
|
| 301 |
+
"logits/chosen": -1.9168952703475952,
|
| 302 |
+
"logits/rejected": -1.8547157049179077,
|
| 303 |
+
"logps/chosen": -825.1336059570312,
|
| 304 |
+
"logps/rejected": -1616.734375,
|
| 305 |
+
"loss": 0.0578,
|
| 306 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 307 |
+
"rewards/chosen": -4.402952671051025,
|
| 308 |
+
"rewards/margins": 8.200037956237793,
|
| 309 |
+
"rewards/rejected": -12.602991104125977,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.2521356888114788,
|
| 314 |
+
"grad_norm": 2.296875,
|
| 315 |
+
"learning_rate": 4.65826447342166e-06,
|
| 316 |
+
"logits/chosen": -1.9839226007461548,
|
| 317 |
+
"logits/rejected": -1.9680023193359375,
|
| 318 |
+
"logps/chosen": -803.8746948242188,
|
| 319 |
+
"logps/rejected": -1493.1898193359375,
|
| 320 |
+
"loss": 0.0497,
|
| 321 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 322 |
+
"rewards/chosen": -3.822723865509033,
|
| 323 |
+
"rewards/margins": 7.312263488769531,
|
| 324 |
+
"rewards/rejected": -11.134986877441406,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.26540598822260925,
|
| 329 |
+
"grad_norm": 8.8125,
|
| 330 |
+
"learning_rate": 4.597413264082086e-06,
|
| 331 |
+
"logits/chosen": -1.9621204137802124,
|
| 332 |
+
"logits/rejected": -1.9449526071548462,
|
| 333 |
+
"logps/chosen": -757.6912841796875,
|
| 334 |
+
"logps/rejected": -1455.258544921875,
|
| 335 |
+
"loss": 0.042,
|
| 336 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 337 |
+
"rewards/chosen": -3.6233069896698,
|
| 338 |
+
"rewards/margins": 7.341341495513916,
|
| 339 |
+
"rewards/rejected": -10.964648246765137,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.26540598822260925,
|
| 344 |
+
"eval_logits/chosen": -1.9894516468048096,
|
| 345 |
+
"eval_logits/rejected": -1.9743856191635132,
|
| 346 |
+
"eval_logps/chosen": -750.5289306640625,
|
| 347 |
+
"eval_logps/rejected": -1423.288330078125,
|
| 348 |
+
"eval_loss": 0.04295578598976135,
|
| 349 |
+
"eval_rewards/accuracies": 0.9828358292579651,
|
| 350 |
+
"eval_rewards/chosen": -3.5129029750823975,
|
| 351 |
+
"eval_rewards/margins": 7.164910793304443,
|
| 352 |
+
"eval_rewards/rejected": -10.677813529968262,
|
| 353 |
+
"eval_runtime": 830.8607,
|
| 354 |
+
"eval_samples_per_second": 6.45,
|
| 355 |
+
"eval_steps_per_second": 1.613,
|
| 356 |
+
"step": 200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.27867628763373975,
|
| 360 |
+
"grad_norm": 9.6875,
|
| 361 |
+
"learning_rate": 4.5320463193780265e-06,
|
| 362 |
+
"logits/chosen": -1.9735777378082275,
|
| 363 |
+
"logits/rejected": -1.9742670059204102,
|
| 364 |
+
"logps/chosen": -789.7581176757812,
|
| 365 |
+
"logps/rejected": -1441.5927734375,
|
| 366 |
+
"loss": 0.0416,
|
| 367 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 368 |
+
"rewards/chosen": -3.7721946239471436,
|
| 369 |
+
"rewards/margins": 6.881577968597412,
|
| 370 |
+
"rewards/rejected": -10.653772354125977,
|
| 371 |
+
"step": 210
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.2919465870448702,
|
| 375 |
+
"grad_norm": 2.71875,
|
| 376 |
+
"learning_rate": 4.462304374485005e-06,
|
| 377 |
+
"logits/chosen": -1.910851240158081,
|
| 378 |
+
"logits/rejected": -1.8877220153808594,
|
| 379 |
+
"logps/chosen": -795.4382934570312,
|
| 380 |
+
"logps/rejected": -1552.471923828125,
|
| 381 |
+
"loss": 0.0435,
|
| 382 |
+
"rewards/accuracies": 0.987500011920929,
|
| 383 |
+
"rewards/chosen": -4.075817584991455,
|
| 384 |
+
"rewards/margins": 7.9385833740234375,
|
| 385 |
+
"rewards/rejected": -12.014400482177734,
|
| 386 |
+
"step": 220
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.30521688645600065,
|
| 390 |
+
"grad_norm": 0.9296875,
|
| 391 |
+
"learning_rate": 4.388337583963563e-06,
|
| 392 |
+
"logits/chosen": -1.8420207500457764,
|
| 393 |
+
"logits/rejected": -1.7979652881622314,
|
| 394 |
+
"logps/chosen": -831.4200439453125,
|
| 395 |
+
"logps/rejected": -1719.750244140625,
|
| 396 |
+
"loss": 0.0253,
|
| 397 |
+
"rewards/accuracies": 1.0,
|
| 398 |
+
"rewards/chosen": -4.229403018951416,
|
| 399 |
+
"rewards/margins": 9.30135440826416,
|
| 400 |
+
"rewards/rejected": -13.530756950378418,
|
| 401 |
+
"step": 230
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.31848718586713115,
|
| 405 |
+
"grad_norm": 1.15625,
|
| 406 |
+
"learning_rate": 4.310305198476161e-06,
|
| 407 |
+
"logits/chosen": -1.813542366027832,
|
| 408 |
+
"logits/rejected": -1.7659708261489868,
|
| 409 |
+
"logps/chosen": -770.3541259765625,
|
| 410 |
+
"logps/rejected": -1698.886474609375,
|
| 411 |
+
"loss": 0.0336,
|
| 412 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 413 |
+
"rewards/chosen": -3.893308162689209,
|
| 414 |
+
"rewards/margins": 9.639238357543945,
|
| 415 |
+
"rewards/rejected": -13.532546997070312,
|
| 416 |
+
"step": 240
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.3317574852782616,
|
| 420 |
+
"grad_norm": 2.03125,
|
| 421 |
+
"learning_rate": 4.228375221920147e-06,
|
| 422 |
+
"logits/chosen": -1.8265107870101929,
|
| 423 |
+
"logits/rejected": -1.765747308731079,
|
| 424 |
+
"logps/chosen": -721.0618896484375,
|
| 425 |
+
"logps/rejected": -1561.09033203125,
|
| 426 |
+
"loss": 0.0269,
|
| 427 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 428 |
+
"rewards/chosen": -3.446256160736084,
|
| 429 |
+
"rewards/margins": 8.8670015335083,
|
| 430 |
+
"rewards/rejected": -12.313258171081543,
|
| 431 |
+
"step": 250
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.34502778468939205,
|
| 435 |
+
"grad_norm": 3.875,
|
| 436 |
+
"learning_rate": 4.142724049715005e-06,
|
| 437 |
+
"logits/chosen": -1.7385963201522827,
|
| 438 |
+
"logits/rejected": -1.5829213857650757,
|
| 439 |
+
"logps/chosen": -922.6043701171875,
|
| 440 |
+
"logps/rejected": -2136.38525390625,
|
| 441 |
+
"loss": 0.0346,
|
| 442 |
+
"rewards/accuracies": 1.0,
|
| 443 |
+
"rewards/chosen": -4.981525897979736,
|
| 444 |
+
"rewards/margins": 11.962621688842773,
|
| 445 |
+
"rewards/rejected": -16.94414710998535,
|
| 446 |
+
"step": 260
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.3582980841005225,
|
| 450 |
+
"grad_norm": 5.5,
|
| 451 |
+
"learning_rate": 4.053536089022624e-06,
|
| 452 |
+
"logits/chosen": -1.841082215309143,
|
| 453 |
+
"logits/rejected": -1.7555363178253174,
|
| 454 |
+
"logps/chosen": -756.6123657226562,
|
| 455 |
+
"logps/rejected": -1619.0562744140625,
|
| 456 |
+
"loss": 0.0434,
|
| 457 |
+
"rewards/accuracies": 0.987500011920929,
|
| 458 |
+
"rewards/chosen": -3.6720402240753174,
|
| 459 |
+
"rewards/margins": 8.794347763061523,
|
| 460 |
+
"rewards/rejected": -12.466386795043945,
|
| 461 |
+
"step": 270
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.371568383511653,
|
| 465 |
+
"grad_norm": 3.1875,
|
| 466 |
+
"learning_rate": 3.961003361718272e-06,
|
| 467 |
+
"logits/chosen": -1.8635780811309814,
|
| 468 |
+
"logits/rejected": -1.7543712854385376,
|
| 469 |
+
"logps/chosen": -745.083251953125,
|
| 470 |
+
"logps/rejected": -1585.915283203125,
|
| 471 |
+
"loss": 0.0333,
|
| 472 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 473 |
+
"rewards/chosen": -3.244401216506958,
|
| 474 |
+
"rewards/margins": 8.643343925476074,
|
| 475 |
+
"rewards/rejected": -11.887744903564453,
|
| 476 |
+
"step": 280
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.38483868292278345,
|
| 480 |
+
"grad_norm": 16.5,
|
| 481 |
+
"learning_rate": 3.8653250909670815e-06,
|
| 482 |
+
"logits/chosen": -1.6911243200302124,
|
| 483 |
+
"logits/rejected": -1.550756573677063,
|
| 484 |
+
"logps/chosen": -840.0848388671875,
|
| 485 |
+
"logps/rejected": -1973.161376953125,
|
| 486 |
+
"loss": 0.0493,
|
| 487 |
+
"rewards/accuracies": 0.987500011920929,
|
| 488 |
+
"rewards/chosen": -4.447511196136475,
|
| 489 |
+
"rewards/margins": 11.780040740966797,
|
| 490 |
+
"rewards/rejected": -16.227550506591797,
|
| 491 |
+
"step": 290
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.3981089823339139,
|
| 495 |
+
"grad_norm": 1.5859375,
|
| 496 |
+
"learning_rate": 3.7667072722961363e-06,
|
| 497 |
+
"logits/chosen": -1.8192800283432007,
|
| 498 |
+
"logits/rejected": -1.7625439167022705,
|
| 499 |
+
"logps/chosen": -772.19189453125,
|
| 500 |
+
"logps/rejected": -1698.5767822265625,
|
| 501 |
+
"loss": 0.0278,
|
| 502 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 503 |
+
"rewards/chosen": -3.4313864707946777,
|
| 504 |
+
"rewards/margins": 9.797091484069824,
|
| 505 |
+
"rewards/rejected": -13.228475570678711,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.3981089823339139,
|
| 510 |
+
"eval_logits/chosen": -1.8191460371017456,
|
| 511 |
+
"eval_logits/rejected": -1.7454465627670288,
|
| 512 |
+
"eval_logps/chosen": -772.5892944335938,
|
| 513 |
+
"eval_logps/rejected": -1707.0360107421875,
|
| 514 |
+
"eval_loss": 0.03441401198506355,
|
| 515 |
+
"eval_rewards/accuracies": 0.9828358292579651,
|
| 516 |
+
"eval_rewards/chosen": -3.733506202697754,
|
| 517 |
+
"eval_rewards/margins": 9.781785011291504,
|
| 518 |
+
"eval_rewards/rejected": -13.51529312133789,
|
| 519 |
+
"eval_runtime": 829.9568,
|
| 520 |
+
"eval_samples_per_second": 6.457,
|
| 521 |
+
"eval_steps_per_second": 1.615,
|
| 522 |
+
"step": 300
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.41137928174504435,
|
| 526 |
+
"grad_norm": 6.5625,
|
| 527 |
+
"learning_rate": 3.665362230085646e-06,
|
| 528 |
+
"logits/chosen": -1.755239725112915,
|
| 529 |
+
"logits/rejected": -1.6191673278808594,
|
| 530 |
+
"logps/chosen": -831.6585693359375,
|
| 531 |
+
"logps/rejected": -1947.4847412109375,
|
| 532 |
+
"loss": 0.0316,
|
| 533 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 534 |
+
"rewards/chosen": -4.389147758483887,
|
| 535 |
+
"rewards/margins": 11.429269790649414,
|
| 536 |
+
"rewards/rejected": -15.8184175491333,
|
| 537 |
+
"step": 310
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.42464958115617485,
|
| 541 |
+
"grad_norm": 2.9375,
|
| 542 |
+
"learning_rate": 3.5615081604340905e-06,
|
| 543 |
+
"logits/chosen": -1.812110185623169,
|
| 544 |
+
"logits/rejected": -1.6317228078842163,
|
| 545 |
+
"logps/chosen": -835.9284057617188,
|
| 546 |
+
"logps/rejected": -2101.62646484375,
|
| 547 |
+
"loss": 0.04,
|
| 548 |
+
"rewards/accuracies": 0.987500011920929,
|
| 549 |
+
"rewards/chosen": -4.403895378112793,
|
| 550 |
+
"rewards/margins": 12.521451950073242,
|
| 551 |
+
"rewards/rejected": -16.92534828186035,
|
| 552 |
+
"step": 320
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.4379198805673053,
|
| 556 |
+
"grad_norm": 1.3125,
|
| 557 |
+
"learning_rate": 3.4553686613815436e-06,
|
| 558 |
+
"logits/chosen": -1.9260807037353516,
|
| 559 |
+
"logits/rejected": -1.830836534500122,
|
| 560 |
+
"logps/chosen": -688.6129760742188,
|
| 561 |
+
"logps/rejected": -1593.781982421875,
|
| 562 |
+
"loss": 0.0215,
|
| 563 |
+
"rewards/accuracies": 0.987500011920929,
|
| 564 |
+
"rewards/chosen": -3.1708381175994873,
|
| 565 |
+
"rewards/margins": 8.699010848999023,
|
| 566 |
+
"rewards/rejected": -11.869850158691406,
|
| 567 |
+
"step": 330
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.45119017997843575,
|
| 571 |
+
"grad_norm": 1.8515625,
|
| 572 |
+
"learning_rate": 3.3471722515025986e-06,
|
| 573 |
+
"logits/chosen": -1.8425785303115845,
|
| 574 |
+
"logits/rejected": -1.7366511821746826,
|
| 575 |
+
"logps/chosen": -812.2515258789062,
|
| 576 |
+
"logps/rejected": -1760.351806640625,
|
| 577 |
+
"loss": 0.0311,
|
| 578 |
+
"rewards/accuracies": 0.981249988079071,
|
| 579 |
+
"rewards/chosen": -4.064314842224121,
|
| 580 |
+
"rewards/margins": 10.192426681518555,
|
| 581 |
+
"rewards/rejected": -14.256741523742676,
|
| 582 |
+
"step": 340
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.46446047938956625,
|
| 586 |
+
"grad_norm": 1.25,
|
| 587 |
+
"learning_rate": 3.2371518779053744e-06,
|
| 588 |
+
"logits/chosen": -1.9980113506317139,
|
| 589 |
+
"logits/rejected": -1.9302339553833008,
|
| 590 |
+
"logps/chosen": -760.2137451171875,
|
| 591 |
+
"logps/rejected": -1788.198486328125,
|
| 592 |
+
"loss": 0.0314,
|
| 593 |
+
"rewards/accuracies": 1.0,
|
| 594 |
+
"rewards/chosen": -3.4233691692352295,
|
| 595 |
+
"rewards/margins": 10.155590057373047,
|
| 596 |
+
"rewards/rejected": -13.578959465026855,
|
| 597 |
+
"step": 350
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.4777307788006967,
|
| 601 |
+
"grad_norm": 2.125,
|
| 602 |
+
"learning_rate": 3.1255444146958845e-06,
|
| 603 |
+
"logits/chosen": -1.8775854110717773,
|
| 604 |
+
"logits/rejected": -1.7533352375030518,
|
| 605 |
+
"logps/chosen": -726.8655395507812,
|
| 606 |
+
"logps/rejected": -1770.5924072265625,
|
| 607 |
+
"loss": 0.0329,
|
| 608 |
+
"rewards/accuracies": 0.981249988079071,
|
| 609 |
+
"rewards/chosen": -3.306746244430542,
|
| 610 |
+
"rewards/margins": 10.716817855834961,
|
| 611 |
+
"rewards/rejected": -14.023564338684082,
|
| 612 |
+
"step": 360
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.49100107821182715,
|
| 616 |
+
"grad_norm": 2.578125,
|
| 617 |
+
"learning_rate": 3.0125901529875612e-06,
|
| 618 |
+
"logits/chosen": -1.854692816734314,
|
| 619 |
+
"logits/rejected": -1.7414394617080688,
|
| 620 |
+
"logps/chosen": -820.2403564453125,
|
| 621 |
+
"logps/rejected": -1938.7720947265625,
|
| 622 |
+
"loss": 0.0401,
|
| 623 |
+
"rewards/accuracies": 0.987500011920929,
|
| 624 |
+
"rewards/chosen": -3.7310116291046143,
|
| 625 |
+
"rewards/margins": 12.33505630493164,
|
| 626 |
+
"rewards/rejected": -16.06606674194336,
|
| 627 |
+
"step": 370
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.5042713776229576,
|
| 631 |
+
"grad_norm": 1.4765625,
|
| 632 |
+
"learning_rate": 2.898532283553963e-06,
|
| 633 |
+
"logits/chosen": -1.8446468114852905,
|
| 634 |
+
"logits/rejected": -1.6768661737442017,
|
| 635 |
+
"logps/chosen": -750.1494750976562,
|
| 636 |
+
"logps/rejected": -1983.754638671875,
|
| 637 |
+
"loss": 0.0151,
|
| 638 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 639 |
+
"rewards/chosen": -3.6851119995117188,
|
| 640 |
+
"rewards/margins": 12.596293449401855,
|
| 641 |
+
"rewards/rejected": -16.281402587890625,
|
| 642 |
+
"step": 380
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.517541677034088,
|
| 646 |
+
"grad_norm": 3.78125,
|
| 647 |
+
"learning_rate": 2.783616373238507e-06,
|
| 648 |
+
"logits/chosen": -1.808468222618103,
|
| 649 |
+
"logits/rejected": -1.6193923950195312,
|
| 650 |
+
"logps/chosen": -817.4727783203125,
|
| 651 |
+
"logps/rejected": -1965.7611083984375,
|
| 652 |
+
"loss": 0.0232,
|
| 653 |
+
"rewards/accuracies": 1.0,
|
| 654 |
+
"rewards/chosen": -4.224337577819824,
|
| 655 |
+
"rewards/margins": 11.751169204711914,
|
| 656 |
+
"rewards/rejected": -15.975506782531738,
|
| 657 |
+
"step": 390
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.5308119764452185,
|
| 661 |
+
"grad_norm": 1.3359375,
|
| 662 |
+
"learning_rate": 2.6680898362485126e-06,
|
| 663 |
+
"logits/chosen": -1.9052226543426514,
|
| 664 |
+
"logits/rejected": -1.790858507156372,
|
| 665 |
+
"logps/chosen": -760.7432861328125,
|
| 666 |
+
"logps/rejected": -1788.7734375,
|
| 667 |
+
"loss": 0.0223,
|
| 668 |
+
"rewards/accuracies": 0.987500011920929,
|
| 669 |
+
"rewards/chosen": -3.564972400665283,
|
| 670 |
+
"rewards/margins": 10.65664291381836,
|
| 671 |
+
"rewards/rejected": -14.2216157913208,
|
| 672 |
+
"step": 400
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.5308119764452185,
|
| 676 |
+
"eval_logits/chosen": -1.918375015258789,
|
| 677 |
+
"eval_logits/rejected": -1.8019860982894897,
|
| 678 |
+
"eval_logps/chosen": -764.7831420898438,
|
| 679 |
+
"eval_logps/rejected": -1732.62890625,
|
| 680 |
+
"eval_loss": 0.030813412740826607,
|
| 681 |
+
"eval_rewards/accuracies": 0.9858208894729614,
|
| 682 |
+
"eval_rewards/chosen": -3.655445098876953,
|
| 683 |
+
"eval_rewards/margins": 10.11577320098877,
|
| 684 |
+
"eval_rewards/rejected": -13.771217346191406,
|
| 685 |
+
"eval_runtime": 830.1888,
|
| 686 |
+
"eval_samples_per_second": 6.455,
|
| 687 |
+
"eval_steps_per_second": 1.614,
|
| 688 |
+
"step": 400
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.544082275856349,
|
| 692 |
+
"grad_norm": 0.76953125,
|
| 693 |
+
"learning_rate": 2.55220140147187e-06,
|
| 694 |
+
"logits/chosen": -1.919586420059204,
|
| 695 |
+
"logits/rejected": -1.729554533958435,
|
| 696 |
+
"logps/chosen": -761.3343505859375,
|
| 697 |
+
"logps/rejected": -1788.455322265625,
|
| 698 |
+
"loss": 0.0345,
|
| 699 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 700 |
+
"rewards/chosen": -3.7783074378967285,
|
| 701 |
+
"rewards/margins": 10.44767951965332,
|
| 702 |
+
"rewards/rejected": -14.225985527038574,
|
| 703 |
+
"step": 410
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.5573525752674795,
|
| 707 |
+
"grad_norm": 0.7890625,
|
| 708 |
+
"learning_rate": 2.4362005769631985e-06,
|
| 709 |
+
"logits/chosen": -1.8321539163589478,
|
| 710 |
+
"logits/rejected": -1.664629340171814,
|
| 711 |
+
"logps/chosen": -864.9075317382812,
|
| 712 |
+
"logps/rejected": -2111.5986328125,
|
| 713 |
+
"loss": 0.0269,
|
| 714 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 715 |
+
"rewards/chosen": -4.227338790893555,
|
| 716 |
+
"rewards/margins": 13.148414611816406,
|
| 717 |
+
"rewards/rejected": -17.375751495361328,
|
| 718 |
+
"step": 420
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.57062287467861,
|
| 722 |
+
"grad_norm": 0.6796875,
|
| 723 |
+
"learning_rate": 2.320337112752459e-06,
|
| 724 |
+
"logits/chosen": -1.767311692237854,
|
| 725 |
+
"logits/rejected": -1.6259944438934326,
|
| 726 |
+
"logps/chosen": -846.6857299804688,
|
| 727 |
+
"logps/rejected": -1985.771240234375,
|
| 728 |
+
"loss": 0.0251,
|
| 729 |
+
"rewards/accuracies": 1.0,
|
| 730 |
+
"rewards/chosen": -4.594606399536133,
|
| 731 |
+
"rewards/margins": 11.99472427368164,
|
| 732 |
+
"rewards/rejected": -16.58932876586914,
|
| 733 |
+
"step": 430
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.5838931740897404,
|
| 737 |
+
"grad_norm": 2.21875,
|
| 738 |
+
"learning_rate": 2.2048604631325896e-06,
|
| 739 |
+
"logits/chosen": -1.9071296453475952,
|
| 740 |
+
"logits/rejected": -1.7615699768066406,
|
| 741 |
+
"logps/chosen": -740.1785888671875,
|
| 742 |
+
"logps/rejected": -1816.4349365234375,
|
| 743 |
+
"loss": 0.0218,
|
| 744 |
+
"rewards/accuracies": 1.0,
|
| 745 |
+
"rewards/chosen": -3.5860729217529297,
|
| 746 |
+
"rewards/margins": 11.016485214233398,
|
| 747 |
+
"rewards/rejected": -14.602559089660645,
|
| 748 |
+
"step": 440
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.5971634735008708,
|
| 752 |
+
"grad_norm": 1.3359375,
|
| 753 |
+
"learning_rate": 2.0900192495838617e-06,
|
| 754 |
+
"logits/chosen": -1.8986167907714844,
|
| 755 |
+
"logits/rejected": -1.749355673789978,
|
| 756 |
+
"logps/chosen": -749.77294921875,
|
| 757 |
+
"logps/rejected": -1782.9183349609375,
|
| 758 |
+
"loss": 0.0303,
|
| 759 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 760 |
+
"rewards/chosen": -3.5921592712402344,
|
| 761 |
+
"rewards/margins": 10.537958145141602,
|
| 762 |
+
"rewards/rejected": -14.130119323730469,
|
| 763 |
+
"step": 450
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.6104337729120013,
|
| 767 |
+
"grad_norm": 5.6875,
|
| 768 |
+
"learning_rate": 1.976060725491293e-06,
|
| 769 |
+
"logits/chosen": -1.8696537017822266,
|
| 770 |
+
"logits/rejected": -1.6489883661270142,
|
| 771 |
+
"logps/chosen": -817.318359375,
|
| 772 |
+
"logps/rejected": -2079.50732421875,
|
| 773 |
+
"loss": 0.0169,
|
| 774 |
+
"rewards/accuracies": 1.0,
|
| 775 |
+
"rewards/chosen": -4.171055316925049,
|
| 776 |
+
"rewards/margins": 12.784004211425781,
|
| 777 |
+
"rewards/rejected": -16.955059051513672,
|
| 778 |
+
"step": 460
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.6237040723231317,
|
| 782 |
+
"grad_norm": 0.466796875,
|
| 783 |
+
"learning_rate": 1.8632302438075618e-06,
|
| 784 |
+
"logits/chosen": -1.7880550622940063,
|
| 785 |
+
"logits/rejected": -1.5018467903137207,
|
| 786 |
+
"logps/chosen": -904.4384765625,
|
| 787 |
+
"logps/rejected": -2307.56103515625,
|
| 788 |
+
"loss": 0.0164,
|
| 789 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 790 |
+
"rewards/chosen": -4.95499324798584,
|
| 791 |
+
"rewards/margins": 14.273486137390137,
|
| 792 |
+
"rewards/rejected": -19.22848129272461,
|
| 793 |
+
"step": 470
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.6369743717342623,
|
| 797 |
+
"grad_norm": 0.8046875,
|
| 798 |
+
"learning_rate": 1.7517707288075617e-06,
|
| 799 |
+
"logits/chosen": -1.7752134799957275,
|
| 800 |
+
"logits/rejected": -1.5753581523895264,
|
| 801 |
+
"logps/chosen": -829.3978271484375,
|
| 802 |
+
"logps/rejected": -2190.319091796875,
|
| 803 |
+
"loss": 0.0281,
|
| 804 |
+
"rewards/accuracies": 0.987500011920929,
|
| 805 |
+
"rewards/chosen": -4.3248748779296875,
|
| 806 |
+
"rewards/margins": 13.842320442199707,
|
| 807 |
+
"rewards/rejected": -18.16719627380371,
|
| 808 |
+
"step": 480
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.6502446711453927,
|
| 812 |
+
"grad_norm": 5.03125,
|
| 813 |
+
"learning_rate": 1.6419221530719062e-06,
|
| 814 |
+
"logits/chosen": -1.8694307804107666,
|
| 815 |
+
"logits/rejected": -1.6351697444915771,
|
| 816 |
+
"logps/chosen": -807.4894409179688,
|
| 817 |
+
"logps/rejected": -2210.259765625,
|
| 818 |
+
"loss": 0.0224,
|
| 819 |
+
"rewards/accuracies": 0.987500011920929,
|
| 820 |
+
"rewards/chosen": -4.219082832336426,
|
| 821 |
+
"rewards/margins": 13.85973834991455,
|
| 822 |
+
"rewards/rejected": -18.078821182250977,
|
| 823 |
+
"step": 490
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 0.6635149705565232,
|
| 827 |
+
"grad_norm": 5.25,
|
| 828 |
+
"learning_rate": 1.5339210208254345e-06,
|
| 829 |
+
"logits/chosen": -1.796460509300232,
|
| 830 |
+
"logits/rejected": -1.5865790843963623,
|
| 831 |
+
"logps/chosen": -825.5130615234375,
|
| 832 |
+
"logps/rejected": -2155.32177734375,
|
| 833 |
+
"loss": 0.0378,
|
| 834 |
+
"rewards/accuracies": 0.987500011920929,
|
| 835 |
+
"rewards/chosen": -4.26907205581665,
|
| 836 |
+
"rewards/margins": 13.452142715454102,
|
| 837 |
+
"rewards/rejected": -17.721214294433594,
|
| 838 |
+
"step": 500
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.6635149705565232,
|
| 842 |
+
"eval_logits/chosen": -1.8649603128433228,
|
| 843 |
+
"eval_logits/rejected": -1.6923589706420898,
|
| 844 |
+
"eval_logps/chosen": -799.422119140625,
|
| 845 |
+
"eval_logps/rejected": -1988.354248046875,
|
| 846 |
+
"eval_loss": 0.029703186824917793,
|
| 847 |
+
"eval_rewards/accuracies": 0.9850746393203735,
|
| 848 |
+
"eval_rewards/chosen": -4.001834392547607,
|
| 849 |
+
"eval_rewards/margins": 12.326638221740723,
|
| 850 |
+
"eval_rewards/rejected": -16.328474044799805,
|
| 851 |
+
"eval_runtime": 830.0926,
|
| 852 |
+
"eval_samples_per_second": 6.456,
|
| 853 |
+
"eval_steps_per_second": 1.614,
|
| 854 |
+
"step": 500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.6767852699676536,
|
| 858 |
+
"grad_norm": 2.53125,
|
| 859 |
+
"learning_rate": 1.4279998587430944e-06,
|
| 860 |
+
"logits/chosen": -1.9209420680999756,
|
| 861 |
+
"logits/rejected": -1.7078396081924438,
|
| 862 |
+
"logps/chosen": -800.6217041015625,
|
| 863 |
+
"logps/rejected": -2088.8896484375,
|
| 864 |
+
"loss": 0.0344,
|
| 865 |
+
"rewards/accuracies": 1.0,
|
| 866 |
+
"rewards/chosen": -3.650468349456787,
|
| 867 |
+
"rewards/margins": 13.276763916015625,
|
| 868 |
+
"rewards/rejected": -16.927234649658203,
|
| 869 |
+
"step": 510
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 0.6900555693787841,
|
| 873 |
+
"grad_norm": 0.44921875,
|
| 874 |
+
"learning_rate": 1.3243867153195033e-06,
|
| 875 |
+
"logits/chosen": -1.9058411121368408,
|
| 876 |
+
"logits/rejected": -1.7081434726715088,
|
| 877 |
+
"logps/chosen": -756.5093994140625,
|
| 878 |
+
"logps/rejected": -1919.1995849609375,
|
| 879 |
+
"loss": 0.0323,
|
| 880 |
+
"rewards/accuracies": 1.0,
|
| 881 |
+
"rewards/chosen": -3.5202860832214355,
|
| 882 |
+
"rewards/margins": 11.93966007232666,
|
| 883 |
+
"rewards/rejected": -15.459945678710938,
|
| 884 |
+
"step": 520
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.7033258687899145,
|
| 888 |
+
"grad_norm": 10.4375,
|
| 889 |
+
"learning_rate": 1.2233046698800343e-06,
|
| 890 |
+
"logits/chosen": -1.9104827642440796,
|
| 891 |
+
"logits/rejected": -1.6314716339111328,
|
| 892 |
+
"logps/chosen": -794.9130859375,
|
| 893 |
+
"logps/rejected": -2090.015380859375,
|
| 894 |
+
"loss": 0.0287,
|
| 895 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 896 |
+
"rewards/chosen": -3.8667774200439453,
|
| 897 |
+
"rewards/margins": 13.207371711730957,
|
| 898 |
+
"rewards/rejected": -17.074146270751953,
|
| 899 |
+
"step": 530
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.716596168201045,
|
| 903 |
+
"grad_norm": 1.3125,
|
| 904 |
+
"learning_rate": 1.124971352290545e-06,
|
| 905 |
+
"logits/chosen": -1.8662798404693604,
|
| 906 |
+
"logits/rejected": -1.6738961935043335,
|
| 907 |
+
"logps/chosen": -800.027587890625,
|
| 908 |
+
"logps/rejected": -2160.337158203125,
|
| 909 |
+
"loss": 0.0261,
|
| 910 |
+
"rewards/accuracies": 0.981249988079071,
|
| 911 |
+
"rewards/chosen": -4.005196571350098,
|
| 912 |
+
"rewards/margins": 13.762578964233398,
|
| 913 |
+
"rewards/rejected": -17.76777458190918,
|
| 914 |
+
"step": 540
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.7298664676121756,
|
| 918 |
+
"grad_norm": 1.3046875,
|
| 919 |
+
"learning_rate": 1.0295984743997911e-06,
|
| 920 |
+
"logits/chosen": -1.8652830123901367,
|
| 921 |
+
"logits/rejected": -1.6665403842926025,
|
| 922 |
+
"logps/chosen": -794.8709106445312,
|
| 923 |
+
"logps/rejected": -1995.255859375,
|
| 924 |
+
"loss": 0.03,
|
| 925 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 926 |
+
"rewards/chosen": -3.9191575050354004,
|
| 927 |
+
"rewards/margins": 12.423823356628418,
|
| 928 |
+
"rewards/rejected": -16.34298324584961,
|
| 929 |
+
"step": 550
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.743136767023306,
|
| 933 |
+
"grad_norm": 0.470703125,
|
| 934 |
+
"learning_rate": 9.37391374223355e-07,
|
| 935 |
+
"logits/chosen": -1.8799976110458374,
|
| 936 |
+
"logits/rejected": -1.6955276727676392,
|
| 937 |
+
"logps/chosen": -788.1275634765625,
|
| 938 |
+
"logps/rejected": -1998.5726318359375,
|
| 939 |
+
"loss": 0.0196,
|
| 940 |
+
"rewards/accuracies": 1.0,
|
| 941 |
+
"rewards/chosen": -3.821469783782959,
|
| 942 |
+
"rewards/margins": 12.33929443359375,
|
| 943 |
+
"rewards/rejected": -16.160762786865234,
|
| 944 |
+
"step": 560
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.7564070664344364,
|
| 948 |
+
"grad_norm": 1.6171875,
|
| 949 |
+
"learning_rate": 8.48548573850449e-07,
|
| 950 |
+
"logits/chosen": -1.8499343395233154,
|
| 951 |
+
"logits/rejected": -1.6717488765716553,
|
| 952 |
+
"logps/chosen": -778.8970947265625,
|
| 953 |
+
"logps/rejected": -1945.9202880859375,
|
| 954 |
+
"loss": 0.0362,
|
| 955 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 956 |
+
"rewards/chosen": -3.6539416313171387,
|
| 957 |
+
"rewards/margins": 12.185694694519043,
|
| 958 |
+
"rewards/rejected": -15.839635848999023,
|
| 959 |
+
"step": 570
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.7696773658455669,
|
| 963 |
+
"grad_norm": 2.46875,
|
| 964 |
+
"learning_rate": 7.632613520254159e-07,
|
| 965 |
+
"logits/chosen": -1.9424989223480225,
|
| 966 |
+
"logits/rejected": -1.8162052631378174,
|
| 967 |
+
"logps/chosen": -791.0227661132812,
|
| 968 |
+
"logps/rejected": -1805.901123046875,
|
| 969 |
+
"loss": 0.035,
|
| 970 |
+
"rewards/accuracies": 0.987500011920929,
|
| 971 |
+
"rewards/chosen": -3.653287410736084,
|
| 972 |
+
"rewards/margins": 11.017667770385742,
|
| 973 |
+
"rewards/rejected": -14.670953750610352,
|
| 974 |
+
"step": 580
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.7829476652566973,
|
| 978 |
+
"grad_norm": 0.94921875,
|
| 979 |
+
"learning_rate": 6.817133323241757e-07,
|
| 980 |
+
"logits/chosen": -1.9590984582901,
|
| 981 |
+
"logits/rejected": -1.754612922668457,
|
| 982 |
+
"logps/chosen": -768.0120239257812,
|
| 983 |
+
"logps/rejected": -2016.432373046875,
|
| 984 |
+
"loss": 0.0217,
|
| 985 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 986 |
+
"rewards/chosen": -3.4281082153320312,
|
| 987 |
+
"rewards/margins": 12.803730964660645,
|
| 988 |
+
"rewards/rejected": -16.23183822631836,
|
| 989 |
+
"step": 590
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.7962179646678278,
|
| 993 |
+
"grad_norm": 2.203125,
|
| 994 |
+
"learning_rate": 6.040800878122655e-07,
|
| 995 |
+
"logits/chosen": -1.9362680912017822,
|
| 996 |
+
"logits/rejected": -1.8174184560775757,
|
| 997 |
+
"logps/chosen": -807.0682373046875,
|
| 998 |
+
"logps/rejected": -1890.33984375,
|
| 999 |
+
"loss": 0.0352,
|
| 1000 |
+
"rewards/accuracies": 1.0,
|
| 1001 |
+
"rewards/chosen": -3.6727454662323,
|
| 1002 |
+
"rewards/margins": 11.275753021240234,
|
| 1003 |
+
"rewards/rejected": -14.948498725891113,
|
| 1004 |
+
"step": 600
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.7962179646678278,
|
| 1008 |
+
"eval_logits/chosen": -1.8977892398834229,
|
| 1009 |
+
"eval_logits/rejected": -1.7437201738357544,
|
| 1010 |
+
"eval_logps/chosen": -780.2752075195312,
|
| 1011 |
+
"eval_logps/rejected": -1919.8118896484375,
|
| 1012 |
+
"eval_loss": 0.02784702554345131,
|
| 1013 |
+
"eval_rewards/accuracies": 0.983582079410553,
|
| 1014 |
+
"eval_rewards/chosen": -3.810366153717041,
|
| 1015 |
+
"eval_rewards/margins": 11.832680702209473,
|
| 1016 |
+
"eval_rewards/rejected": -15.643047332763672,
|
| 1017 |
+
"eval_runtime": 829.9624,
|
| 1018 |
+
"eval_samples_per_second": 6.457,
|
| 1019 |
+
"eval_steps_per_second": 1.615,
|
| 1020 |
+
"step": 600
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 0.8094882640789582,
|
| 1024 |
+
"grad_norm": 1.171875,
|
| 1025 |
+
"learning_rate": 5.305287630356363e-07,
|
| 1026 |
+
"logits/chosen": -1.8633983135223389,
|
| 1027 |
+
"logits/rejected": -1.6604722738265991,
|
| 1028 |
+
"logps/chosen": -751.8145141601562,
|
| 1029 |
+
"logps/rejected": -1894.828857421875,
|
| 1030 |
+
"loss": 0.0299,
|
| 1031 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1032 |
+
"rewards/chosen": -3.7778866291046143,
|
| 1033 |
+
"rewards/margins": 11.620625495910645,
|
| 1034 |
+
"rewards/rejected": -15.39851188659668,
|
| 1035 |
+
"step": 610
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.8227585634900887,
|
| 1039 |
+
"grad_norm": 0.37109375,
|
| 1040 |
+
"learning_rate": 4.612177141580876e-07,
|
| 1041 |
+
"logits/chosen": -1.8507238626480103,
|
| 1042 |
+
"logits/rejected": -1.6502704620361328,
|
| 1043 |
+
"logps/chosen": -739.475341796875,
|
| 1044 |
+
"logps/rejected": -1800.6304931640625,
|
| 1045 |
+
"loss": 0.028,
|
| 1046 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 1047 |
+
"rewards/chosen": -3.580106735229492,
|
| 1048 |
+
"rewards/margins": 10.93178939819336,
|
| 1049 |
+
"rewards/rejected": -14.511896133422852,
|
| 1050 |
+
"step": 620
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.8360288629012192,
|
| 1054 |
+
"grad_norm": 2.375,
|
| 1055 |
+
"learning_rate": 3.962961680200927e-07,
|
| 1056 |
+
"logits/chosen": -1.9084770679473877,
|
| 1057 |
+
"logits/rejected": -1.7507511377334595,
|
| 1058 |
+
"logps/chosen": -785.728515625,
|
| 1059 |
+
"logps/rejected": -1896.435791015625,
|
| 1060 |
+
"loss": 0.0222,
|
| 1061 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 1062 |
+
"rewards/chosen": -3.7701239585876465,
|
| 1063 |
+
"rewards/margins": 11.418277740478516,
|
| 1064 |
+
"rewards/rejected": -15.18840217590332,
|
| 1065 |
+
"step": 630
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.8492991623123497,
|
| 1069 |
+
"grad_norm": 1.96875,
|
| 1070 |
+
"learning_rate": 3.3590390085308457e-07,
|
| 1071 |
+
"logits/chosen": -1.9369093179702759,
|
| 1072 |
+
"logits/rejected": -1.782634973526001,
|
| 1073 |
+
"logps/chosen": -792.2256469726562,
|
| 1074 |
+
"logps/rejected": -1962.2239990234375,
|
| 1075 |
+
"loss": 0.0246,
|
| 1076 |
+
"rewards/accuracies": 0.981249988079071,
|
| 1077 |
+
"rewards/chosen": -3.8103549480438232,
|
| 1078 |
+
"rewards/margins": 11.93663501739502,
|
| 1079 |
+
"rewards/rejected": -15.746989250183105,
|
| 1080 |
+
"step": 640
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.8625694617234801,
|
| 1084 |
+
"grad_norm": 0.373046875,
|
| 1085 |
+
"learning_rate": 2.801709373409248e-07,
|
| 1086 |
+
"logits/chosen": -1.8602193593978882,
|
| 1087 |
+
"logits/rejected": -1.7032759189605713,
|
| 1088 |
+
"logps/chosen": -804.9444580078125,
|
| 1089 |
+
"logps/rejected": -2019.756103515625,
|
| 1090 |
+
"loss": 0.0177,
|
| 1091 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 1092 |
+
"rewards/chosen": -3.675445556640625,
|
| 1093 |
+
"rewards/margins": 12.645570755004883,
|
| 1094 |
+
"rewards/rejected": -16.321016311645508,
|
| 1095 |
+
"step": 650
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.8758397611346106,
|
| 1099 |
+
"grad_norm": 1.90625,
|
| 1100 |
+
"learning_rate": 2.2921727067647032e-07,
|
| 1101 |
+
"logits/chosen": -1.9662708044052124,
|
| 1102 |
+
"logits/rejected": -1.7979133129119873,
|
| 1103 |
+
"logps/chosen": -745.5474243164062,
|
| 1104 |
+
"logps/rejected": -1877.61328125,
|
| 1105 |
+
"loss": 0.0226,
|
| 1106 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1107 |
+
"rewards/chosen": -3.5280399322509766,
|
| 1108 |
+
"rewards/margins": 11.784358024597168,
|
| 1109 |
+
"rewards/rejected": -15.312397956848145,
|
| 1110 |
+
"step": 660
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 0.889110060545741,
|
| 1114 |
+
"grad_norm": 15.375,
|
| 1115 |
+
"learning_rate": 1.8315260421596925e-07,
|
| 1116 |
+
"logits/chosen": -1.9315414428710938,
|
| 1117 |
+
"logits/rejected": -1.7226520776748657,
|
| 1118 |
+
"logps/chosen": -760.8677978515625,
|
| 1119 |
+
"logps/rejected": -1978.8160400390625,
|
| 1120 |
+
"loss": 0.0338,
|
| 1121 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1122 |
+
"rewards/chosen": -3.70927357673645,
|
| 1123 |
+
"rewards/margins": 12.64477252960205,
|
| 1124 |
+
"rewards/rejected": -16.35404396057129,
|
| 1125 |
+
"step": 670
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.9023803599568715,
|
| 1129 |
+
"grad_norm": 0.412109375,
|
| 1130 |
+
"learning_rate": 1.4207611528749e-07,
|
| 1131 |
+
"logits/chosen": -1.9155975580215454,
|
| 1132 |
+
"logits/rejected": -1.6965796947479248,
|
| 1133 |
+
"logps/chosen": -769.4285278320312,
|
| 1134 |
+
"logps/rejected": -1877.641845703125,
|
| 1135 |
+
"loss": 0.0319,
|
| 1136 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1137 |
+
"rewards/chosen": -3.767559766769409,
|
| 1138 |
+
"rewards/margins": 11.334938049316406,
|
| 1139 |
+
"rewards/rejected": -15.102499008178711,
|
| 1140 |
+
"step": 680
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 0.9156506593680019,
|
| 1144 |
+
"grad_norm": 0.56640625,
|
| 1145 |
+
"learning_rate": 1.060762416619196e-07,
|
| 1146 |
+
"logits/chosen": -1.946947693824768,
|
| 1147 |
+
"logits/rejected": -1.7013956308364868,
|
| 1148 |
+
"logps/chosen": -765.5987548828125,
|
| 1149 |
+
"logps/rejected": -2080.13134765625,
|
| 1150 |
+
"loss": 0.0215,
|
| 1151 |
+
"rewards/accuracies": 0.981249988079071,
|
| 1152 |
+
"rewards/chosen": -3.7093491554260254,
|
| 1153 |
+
"rewards/margins": 12.91864013671875,
|
| 1154 |
+
"rewards/rejected": -16.627988815307617,
|
| 1155 |
+
"step": 690
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 0.9289209587791325,
|
| 1159 |
+
"grad_norm": 0.515625,
|
| 1160 |
+
"learning_rate": 7.523049114624647e-08,
|
| 1161 |
+
"logits/chosen": -1.9617208242416382,
|
| 1162 |
+
"logits/rejected": -1.8319323062896729,
|
| 1163 |
+
"logps/chosen": -802.1334838867188,
|
| 1164 |
+
"logps/rejected": -1892.474609375,
|
| 1165 |
+
"loss": 0.0238,
|
| 1166 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1167 |
+
"rewards/chosen": -3.8938992023468018,
|
| 1168 |
+
"rewards/margins": 11.566261291503906,
|
| 1169 |
+
"rewards/rejected": -15.460162162780762,
|
| 1170 |
+
"step": 700
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 0.9289209587791325,
|
| 1174 |
+
"eval_logits/chosen": -1.8936866521835327,
|
| 1175 |
+
"eval_logits/rejected": -1.737060785293579,
|
| 1176 |
+
"eval_logps/chosen": -788.9779663085938,
|
| 1177 |
+
"eval_logps/rejected": -1951.9310302734375,
|
| 1178 |
+
"eval_loss": 0.027877720072865486,
|
| 1179 |
+
"eval_rewards/accuracies": 0.9828358292579651,
|
| 1180 |
+
"eval_rewards/chosen": -3.897392511367798,
|
| 1181 |
+
"eval_rewards/margins": 12.06684684753418,
|
| 1182 |
+
"eval_rewards/rejected": -15.964240074157715,
|
| 1183 |
+
"eval_runtime": 830.3344,
|
| 1184 |
+
"eval_samples_per_second": 6.454,
|
| 1185 |
+
"eval_steps_per_second": 1.614,
|
| 1186 |
+
"step": 700
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.942191258190263,
|
| 1190 |
+
"grad_norm": 0.92578125,
|
| 1191 |
+
"learning_rate": 4.9605274709082774e-08,
|
| 1192 |
+
"logits/chosen": -1.883724570274353,
|
| 1193 |
+
"logits/rejected": -1.7111318111419678,
|
| 1194 |
+
"logps/chosen": -796.9892578125,
|
| 1195 |
+
"logps/rejected": -2012.552001953125,
|
| 1196 |
+
"loss": 0.0267,
|
| 1197 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1198 |
+
"rewards/chosen": -3.85164213180542,
|
| 1199 |
+
"rewards/margins": 12.14155101776123,
|
| 1200 |
+
"rewards/rejected": -15.993194580078125,
|
| 1201 |
+
"step": 710
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 0.9554615576013934,
|
| 1205 |
+
"grad_norm": 6.9375,
|
| 1206 |
+
"learning_rate": 2.9255763497703373e-08,
|
| 1207 |
+
"logits/chosen": -1.8911021947860718,
|
| 1208 |
+
"logits/rejected": -1.6832084655761719,
|
| 1209 |
+
"logps/chosen": -790.8480224609375,
|
| 1210 |
+
"logps/rejected": -2087.356689453125,
|
| 1211 |
+
"loss": 0.0314,
|
| 1212 |
+
"rewards/accuracies": 1.0,
|
| 1213 |
+
"rewards/chosen": -3.734605312347412,
|
| 1214 |
+
"rewards/margins": 13.477132797241211,
|
| 1215 |
+
"rewards/rejected": -17.21173858642578,
|
| 1216 |
+
"step": 720
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 0.9687318570125238,
|
| 1220 |
+
"grad_norm": 3.46875,
|
| 1221 |
+
"learning_rate": 1.42257700544432e-08,
|
| 1222 |
+
"logits/chosen": -1.9454247951507568,
|
| 1223 |
+
"logits/rejected": -1.7627389430999756,
|
| 1224 |
+
"logps/chosen": -774.6414794921875,
|
| 1225 |
+
"logps/rejected": -1878.0687255859375,
|
| 1226 |
+
"loss": 0.0266,
|
| 1227 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 1228 |
+
"rewards/chosen": -3.7591774463653564,
|
| 1229 |
+
"rewards/margins": 11.635783195495605,
|
| 1230 |
+
"rewards/rejected": -15.3949613571167,
|
| 1231 |
+
"step": 730
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 0.9820021564236543,
|
| 1235 |
+
"grad_norm": 0.6640625,
|
| 1236 |
+
"learning_rate": 4.547653988198619e-09,
|
| 1237 |
+
"logits/chosen": -1.8585050106048584,
|
| 1238 |
+
"logits/rejected": -1.6734154224395752,
|
| 1239 |
+
"logps/chosen": -771.71044921875,
|
| 1240 |
+
"logps/rejected": -1891.1732177734375,
|
| 1241 |
+
"loss": 0.0211,
|
| 1242 |
+
"rewards/accuracies": 1.0,
|
| 1243 |
+
"rewards/chosen": -3.8070855140686035,
|
| 1244 |
+
"rewards/margins": 11.580972671508789,
|
| 1245 |
+
"rewards/rejected": -15.38805866241455,
|
| 1246 |
+
"step": 740
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 0.9952724558347847,
|
| 1250 |
+
"grad_norm": 0.46875,
|
| 1251 |
+
"learning_rate": 2.422523041178959e-10,
|
| 1252 |
+
"logits/chosen": -1.9344615936279297,
|
| 1253 |
+
"logits/rejected": -1.733974814414978,
|
| 1254 |
+
"logps/chosen": -803.4793090820312,
|
| 1255 |
+
"logps/rejected": -2031.9625244140625,
|
| 1256 |
+
"loss": 0.0295,
|
| 1257 |
+
"rewards/accuracies": 0.981249988079071,
|
| 1258 |
+
"rewards/chosen": -3.7710418701171875,
|
| 1259 |
+
"rewards/margins": 12.417600631713867,
|
| 1260 |
+
"rewards/rejected": -16.188644409179688,
|
| 1261 |
+
"step": 750
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 0.999253545658124,
|
| 1265 |
+
"step": 753,
|
| 1266 |
+
"total_flos": 0.0,
|
| 1267 |
+
"train_loss": 0.0882907345950366,
|
| 1268 |
+
"train_runtime": 20220.3954,
|
| 1269 |
+
"train_samples_per_second": 2.385,
|
| 1270 |
+
"train_steps_per_second": 0.037
|
| 1271 |
+
}
|
| 1272 |
+
],
|
| 1273 |
+
"logging_steps": 10,
|
| 1274 |
+
"max_steps": 753,
|
| 1275 |
+
"num_input_tokens_seen": 0,
|
| 1276 |
+
"num_train_epochs": 1,
|
| 1277 |
+
"save_steps": 100,
|
| 1278 |
+
"stateful_callbacks": {
|
| 1279 |
+
"TrainerControl": {
|
| 1280 |
+
"args": {
|
| 1281 |
+
"should_epoch_stop": false,
|
| 1282 |
+
"should_evaluate": false,
|
| 1283 |
+
"should_log": false,
|
| 1284 |
+
"should_save": true,
|
| 1285 |
+
"should_training_stop": true
|
| 1286 |
+
},
|
| 1287 |
+
"attributes": {}
|
| 1288 |
+
}
|
| 1289 |
+
},
|
| 1290 |
+
"total_flos": 0.0,
|
| 1291 |
+
"train_batch_size": 1,
|
| 1292 |
+
"trial_name": null,
|
| 1293 |
+
"trial_params": null
|
| 1294 |
+
}
|