metadata
license: mit
base_model: HuggingFaceH4/mistral-7b-sft-beta
tags:
- trl
- dpo
- generated_from_trainer
model-index:
- name: zephyr-7b-dpo-full
results: []
zephyr-7b-dpo-full
This model is a fine-tuned version of HuggingFaceH4/mistral-7b-sft-beta on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0069
- Rewards/chosen: -2.6575
- Rewards/rejected: -5.6311
- Rewards/accuracies: 0.6914
- Rewards/margins: 2.9735
- Logps/rejected: -820.4592
- Logps/chosen: -522.7909
- Logits/rejected: -2.4988
- Logits/chosen: -2.5238
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: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 2
- total_train_batch_size: 128
- total_eval_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 1
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.0071 | 0.21 | 100 | 0.0072 | -0.8922 | -1.6083 | 0.6523 | 0.7161 | -418.1843 | -346.2603 | -2.6342 | -2.6489 |
| 0.0068 | 0.42 | 200 | 0.0070 | -1.7394 | -3.1970 | 0.6680 | 1.4577 | -577.0542 | -430.9749 | -2.6058 | -2.6219 |
| 0.0069 | 0.63 | 300 | 0.0069 | -2.2358 | -4.6352 | 0.6992 | 2.3994 | -720.8748 | -480.6185 | -2.5115 | -2.5351 |
| 0.0067 | 0.84 | 400 | 0.0069 | -2.6575 | -5.6311 | 0.6914 | 2.9735 | -820.4592 | -522.7909 | -2.4988 | -2.5238 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.14.1