metadata
license: apache-2.0
base_model: mistralai/Mistral-7B-v0.1
tags:
- generated_from_trainer
model-index:
- name: zephyr-7b-dpo-lora
results: []
zephyr-7b-dpo-lora
This model is a fine-tuned version of mistralai/Mistral-7B-v0.1 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.5270
- Rewards/chosen: -0.1210
- Rewards/rejected: -0.9978
- Rewards/accuracies: 0.7812
- Rewards/margins: 0.8768
- Logps/rejected: -198.5849
- Logps/chosen: -248.6519
- Logits/rejected: -1.9190
- Logits/chosen: -2.0860
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: 32
- total_train_batch_size: 64
- total_eval_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.5491 | 1.0 | 969 | 0.5563 | -0.0962 | -0.7226 | 0.7812 | 0.6263 | -195.8333 | -248.4046 | -1.9755 | -2.1375 |
| 0.5454 | 2.0 | 1938 | 0.5312 | -0.1249 | -0.9600 | 0.7969 | 0.8351 | -198.2077 | -248.6910 | -1.9316 | -2.0971 |
| 0.5242 | 3.0 | 2907 | 0.5270 | -0.1210 | -0.9978 | 0.7812 | 0.8768 | -198.5849 | -248.6519 | -1.9190 | -2.0860 |
Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1