|
--- |
|
library_name: peft |
|
tags: |
|
- trl |
|
- dpo |
|
- generated_from_trainer |
|
base_model: Weni/WeniGPT-Agents-Mistral-1.0.6-SFT-merged |
|
model-index: |
|
- name: WeniGPT-Agents-Mistral-1.0.6-SFT-1.0.5-DPO |
|
results: [] |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# WeniGPT-Agents-Mistral-1.0.6-SFT-1.0.5-DPO |
|
|
|
This model is a fine-tuned version of [Weni/WeniGPT-Agents-Mistral-1.0.6-SFT-merged](https://huggingface.co/Weni/WeniGPT-Agents-Mistral-1.0.6-SFT-merged) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.4260 |
|
- Rewards/chosen: 0.9172 |
|
- Rewards/rejected: -0.6078 |
|
- Rewards/accuracies: 0.4643 |
|
- Rewards/margins: 1.5251 |
|
- Logps/rejected: -103.4404 |
|
- Logps/chosen: -46.9008 |
|
- Logits/rejected: -1.8652 |
|
- Logits/chosen: -1.8327 |
|
|
|
## Model description |
|
|
|
More information needed |
|
|
|
## Intended uses & limitations |
|
|
|
More information needed |
|
|
|
## Training and evaluation data |
|
|
|
More information needed |
|
|
|
## Training procedure |
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 5e-06 |
|
- train_batch_size: 2 |
|
- eval_batch_size: 2 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 2 |
|
- total_train_batch_size: 4 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_ratio: 0.03 |
|
- training_steps: 366 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen | |
|
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:| |
|
| 0.6635 | 0.49 | 30 | 0.6524 | 0.0904 | 0.0036 | 0.4643 | 0.0867 | -97.3259 | -55.1696 | -1.8044 | -1.7832 | |
|
| 0.6026 | 0.98 | 60 | 0.5891 | 0.2506 | 0.0024 | 0.4643 | 0.2482 | -97.3380 | -53.5672 | -1.8099 | -1.7878 | |
|
| 0.5387 | 1.46 | 90 | 0.5295 | 0.4396 | -0.0275 | 0.4643 | 0.4671 | -97.6369 | -51.6775 | -1.8181 | -1.7943 | |
|
| 0.6033 | 1.95 | 120 | 0.4960 | 0.5751 | -0.0659 | 0.4643 | 0.6410 | -98.0210 | -50.3219 | -1.8261 | -1.8009 | |
|
| 0.5042 | 2.44 | 150 | 0.4709 | 0.6967 | -0.1479 | 0.4643 | 0.8446 | -98.8407 | -49.1060 | -1.8331 | -1.8059 | |
|
| 0.5087 | 2.93 | 180 | 0.4542 | 0.7878 | -0.2428 | 0.4643 | 1.0306 | -99.7900 | -48.1955 | -1.8425 | -1.8136 | |
|
| 0.4874 | 3.41 | 210 | 0.4428 | 0.8442 | -0.3560 | 0.4643 | 1.2002 | -100.9220 | -47.6315 | -1.8520 | -1.8219 | |
|
| 0.4229 | 3.9 | 240 | 0.4358 | 0.8750 | -0.4390 | 0.4643 | 1.3140 | -101.7521 | -47.3229 | -1.8575 | -1.8266 | |
|
| 0.5295 | 4.39 | 270 | 0.4313 | 0.9026 | -0.4960 | 0.4643 | 1.3986 | -102.3219 | -47.0471 | -1.8607 | -1.8289 | |
|
| 0.5466 | 4.88 | 300 | 0.4291 | 0.9119 | -0.5384 | 0.4643 | 1.4503 | -102.7461 | -46.9544 | -1.8629 | -1.8309 | |
|
| 0.4339 | 5.37 | 330 | 0.4268 | 0.9152 | -0.5900 | 0.4643 | 1.5052 | -103.2623 | -46.9216 | -1.8644 | -1.8320 | |
|
| 0.5438 | 5.85 | 360 | 0.4260 | 0.9172 | -0.6078 | 0.4643 | 1.5251 | -103.4404 | -46.9008 | -1.8652 | -1.8327 | |
|
|
|
|
|
### Framework versions |
|
|
|
- PEFT 0.10.0 |
|
- Transformers 4.38.2 |
|
- Pytorch 2.1.0+cu118 |
|
- Datasets 2.18.0 |
|
- Tokenizers 0.15.2 |
