---
license: apache-2.0
base_model: alignment-handbook/zephyr-7b-sft-full
tags:
- trl
- dpo
- generated_from_trainer
model-index:
- name: zephyr-7b-UFB-ref
  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-UFB-ref

This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.5067
- Rewards/chosen: -1.1023
- Rewards/rejected: -2.3762
- Rewards/accuracies: 0.7098
- Rewards/margins: 1.2739
- Logps/rejected: -120.8096
- Logps/chosen: -110.8900
- Logits/rejected: -2.1865
- Logits/chosen: -2.2284

## 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: 4
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- total_train_batch_size: 16
- total_eval_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- 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 |
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
| 0.5598        | 0.15  | 500  | 0.6348          | -0.4646        | -1.5732          | 0.7121             | 1.1087          | -112.7802      | -104.5124    | -2.2989         | -2.3398       |
| 0.6708        | 0.3   | 1000 | 0.5807          | -1.9508        | -2.8042          | 0.6830             | 0.8534          | -125.0895      | -119.3747    | -2.2155         | -2.2623       |
| 0.5984        | 0.45  | 1500 | 0.5244          | -1.4451        | -2.6765          | 0.7188             | 1.2313          | -123.8126      | -114.3180    | -2.1383         | -2.1824       |
| 0.5508        | 0.6   | 2000 | 0.5644          | -1.7905        | -2.8869          | 0.6786             | 1.0964          | -125.9164      | -117.7717    | -2.0760         | -2.1208       |
| 0.5218        | 0.74  | 2500 | 0.5183          | -1.3228        | -2.5470          | 0.7031             | 1.2242          | -122.5180      | -113.0946    | -2.2172         | -2.2616       |
| 0.4914        | 0.89  | 3000 | 0.5079          | -1.0825        | -2.3551          | 0.7121             | 1.2725          | -120.5985      | -110.6918    | -2.2149         | -2.2567       |


### Framework versions

- Transformers 4.36.1
- Pytorch 2.0.1+cu117
- Datasets 2.16.1
- Tokenizers 0.15.0