metadata
library_name: peft
tags:
- trl
- dpo
- DPO
- WeniGPT
- generated_from_trainer
base_model: Weni/WeniGPT-Agents-Mistral-1.0.6-SFT-merged
model-index:
- name: WeniGPT-Agents-Mistral-1.0.6-SFT-1.0.8-DPO
results: []
WeniGPT-Agents-Mistral-1.0.6-SFT-1.0.8-DPO
This model is a fine-tuned version of Weni/WeniGPT-Agents-Mistral-1.0.6-SFT-merged on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.2005
- Rewards/chosen: 1.7621
- Rewards/rejected: -0.7248
- Rewards/accuracies: 1.0
- Rewards/margins: 2.4869
- Logps/rejected: -271.5133
- Logps/chosen: -102.9727
- Logits/rejected: -1.8958
- Logits/chosen: -1.8013
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: 1
- eval_batch_size: 1
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- total_eval_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: 180
- 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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.5477 | 0.97 | 30 | 0.4843 | 0.4718 | -0.1109 | 0.8571 | 0.5827 | -269.4668 | -107.2735 | -1.8863 | -1.7949 |
| 0.3542 | 1.94 | 60 | 0.3440 | 0.9609 | -0.2516 | 1.0 | 1.2125 | -269.9360 | -105.6431 | -1.8903 | -1.7979 |
| 0.2892 | 2.9 | 90 | 0.2756 | 1.3199 | -0.4182 | 1.0 | 1.7381 | -270.4914 | -104.4467 | -1.8928 | -1.7995 |
| 0.1858 | 3.87 | 120 | 0.2327 | 1.6010 | -0.5696 | 1.0 | 2.1706 | -270.9958 | -103.5094 | -1.8947 | -1.8008 |
| 0.1811 | 4.84 | 150 | 0.2076 | 1.7245 | -0.6925 | 1.0 | 2.4170 | -271.4054 | -103.0979 | -1.8954 | -1.8010 |
| 0.2065 | 5.81 | 180 | 0.2005 | 1.7621 | -0.7248 | 1.0 | 2.4869 | -271.5133 | -102.9727 | -1.8958 | -1.8013 |
Framework versions
- PEFT 0.10.0
- Transformers 4.38.2
- Pytorch 2.1.0+cu118
- Datasets 2.18.0
- Tokenizers 0.15.2