zephyr-7b-gemma-dpo
This model is a fine-tuned version of tanliboy/zephyr-7b-gemma-sft on the argilla/dpo-mix-7k dataset. It achieves the following results on the evaluation set:
- Loss: 0.4833
- Rewards/chosen: -0.1121
- Rewards/rejected: -1.2301
- Rewards/accuracies: 0.7396
- Rewards/margins: 1.1180
- Logps/rejected: -719.5310
- Logps/chosen: -698.5295
- Logits/rejected: 153.0066
- Logits/chosen: 153.1078
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: 8
- gradient_accumulation_steps: 8
- total_train_batch_size: 128
- total_eval_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- 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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.1424 | 1.8957 | 100 | 0.4722 | -0.0658 | -1.2673 | 0.7396 | 1.2015 | -720.2745 | -697.6023 | 152.9660 | 153.1356 |
Framework versions
- Transformers 4.40.2
- Pytorch 2.3.0+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1
- Downloads last month
- 5
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.