zephyr-gemma-dpo-faithful
This model is a fine-tuned version of HuggingFaceH4/zephyr-7b-gemma-sft-v0.1 on the argilla/dpo-mix-7k dataset. It achieves the following results on the evaluation set:
- Loss: 0.4498
- Rewards/chosen: -1.7969
- Rewards/rejected: -3.5156
- Rewards/accuracies: 0.7394
- Rewards/margins: 1.7188
- Logps/rejected: -476.0
- Logps/chosen: -388.0
- Logits/rejected: 89.5
- Logits/chosen: 87.0
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-07
- train_batch_size: 2
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 8
- total_train_batch_size: 64
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- lr_scheduler_warmup_steps: 100
- num_epochs: 2
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.5155 | 0.95 | 100 | 0.5003 | -1.2422 | -2.5156 | 0.6862 | 1.2656 | -456.0 | -378.0 | 97.5 | 95.0 |
| 0.1902 | 1.9 | 200 | 0.4498 | -1.7969 | -3.5156 | 0.7394 | 1.7188 | -476.0 | -388.0 | 89.5 | 87.0 |
Framework versions
- Transformers 4.38.0
- Pytorch 2.1.1+cu121
- Datasets 2.16.1
- Tokenizers 0.15.2
- Downloads last month
- 6