---
base_model: mistralai/Mistral-7B-v0.1
library_name: peft
license: apache-2.0
tags:
- trl
- dpo
- generated_from_trainer
model-index:
- name: zephyr-7b-dpo-qlora
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# zephyr-7b-dpo-qlora

This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4952
- Rewards/chosen: -2.8107
- Rewards/rejected: -3.8708
- Rewards/accuracies: 0.7718
- Rewards/margins: 1.0601
- Logps/rejected: -631.7385
- Logps/chosen: -545.9743
- Logits/rejected: -1.0385
- Logits/chosen: -1.1509

## 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: 4
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- 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: 1

### Training results

| Training Loss | Epoch  | Step | Logits/chosen | Logits/rejected | Logps/chosen | Logps/rejected | Validation Loss | Rewards/accuracies | Rewards/chosen | Rewards/margins | Rewards/rejected |
|:-------------:|:------:|:----:|:-------------:|:---------------:|:------------:|:--------------:|:---------------:|:------------------:|:--------------:|:---------------:|:----------------:|
| 0.6163        | 0.1047 | 100  | -2.1006       | -2.0162         | -303.8351    | -310.3097      | 0.6178          | 0.6806             | -0.3893        | 0.2672          | -0.6565          |
| 0.5679        | 0.2094 | 200  | -1.8227       | -1.7394         | -352.2879    | -389.6575      | 0.5567          | 0.7401             | -0.8739        | 0.5761          | -1.4500          |
| 0.5412        | 0.3141 | 300  | -1.3111       | -1.2181         | -421.3257    | -483.0423      | 0.5305          | 0.7460             | -1.5642        | 0.8196          | -2.3838          |
| 0.5364        | 0.4187 | 400  | -1.2334       | -1.1332         | -416.6979    | -476.3458      | 0.5143          | 0.7579             | -1.5180        | 0.7989          | -2.3169          |
| 0.5046        | 0.5234 | 500  | -1.1373       | -1.0302         | -529.9542    | -605.2977      | 0.5062          | 0.7579             | -2.6505        | 0.9559          | -3.6064          |
| 0.4736        | 0.6281 | 600  | 0.5059        | -2.7244         | -3.7650      | 0.7639         | 1.0406          | -621.1549          | -537.3406      | -1.0135         | -1.1253          |
| 0.4619        | 0.7328 | 700  | 0.4994        | -2.9240         | -3.9991      | 0.7619         | 1.0750          | -644.5651          | -557.3041      | -1.0064         | -1.1194          |
| 0.4926        | 0.8375 | 800  | 0.4962        | -2.7247         | -3.7455      | 0.7659         | 1.0207          | -619.2051          | -537.3770      | -1.0516         | -1.1641          |
| 0.4856        | 0.9422 | 900  | 0.4952        | -2.8107         | -3.8708      | 0.7718         | 1.0601          | -631.7385          | -545.9743      | -1.0385         | -1.1509          |


### Framework versions

- PEFT 0.12.0
- Transformers 4.44.2
- Pytorch 2.4.0
- Datasets 2.21.0
- Tokenizers 0.19.1