README.md

---
model-index:
- name: llama-3-tulu-2-8b-uf-mean-rm
  results: []
datasets:
- allenai/tulu-2.5-preference-data
- allenai/tulu-v2-sft-mixture
language:
- en
base_model: allenai/llama-3-tulu-2-8b
license: apache-2.0
---
<center>
<img src="https://huggingface.co/datasets/allenai/blog-images/resolve/main/tulu-2.5/tulu_25_banner.png" alt="Tulu 2.5 banner image" width="800px"/>
</center>

# Model Card for Llama 3 Tulu V2 8B RM - UltraFeedback

Tulu is a series of language models that are trained to act as helpful assistants.
This is a 8B reward model used for PPO training trained on the UltraFeedback dataset.

For more details, read the paper:
[Unpacking DPO and PPO: Disentangling Best Practices for Learning from Preference Feedback](https://arxiv.org/abs/2406.09279).

**Built with Meta Llama 3!**
Note that Llama 3 is released under the Meta Llama 3 community license, included here under `llama_3_license.txt`.

## Performance

We evaluate the model on [RewardBench](https://github.com/allenai/reward-bench):

| Model            | Score | Chat  | Chat Hard | Safety | Reasoning | Prior Sets (0.5 weight) |
|------------------|-------|-------|-----------|--------|-----------|-------------------------|
| **[Llama 3 Tulu 2 8b UF RM](https://huggingface.co/allenai/llama-3-tulu-2-8b-uf-mean-rm) (this model)**  | 66.3  | 96.6  |    59.4   |  61.4  |    80.7   |                         |
| [Llama 3 Tulu 2 70b UF RM](https://huggingface.co/allenai/llama-3-tulu-2-70b-uf-mean-rm) |    65.3   |   89.1    |      52.6     |    64.0    |      88.3     |                         |


## Model description

- **Model type:** A reward model trained on UltraFeedback, designed to be used in RLHF training.
- **Language(s) (NLP):** English
- **License:** Apache 2.0.
- **Finetuned from model:** [allenai/llama-3-tulu-2-8b](https://huggingface.co/allenai/llama-3-tulu-2-8b)

### Model Sources

- **Repository:** https://github.com/allenai/open-instruct
- **Dataset:** Data used to train this model can be found [here](https://huggingface.co/datasets/allenai/tulu-2.5-preference-data) - specifically the `ultrafeedback_mean_aspects` split.


## Input Format

The model is trained to use the following format (note the newlines):
```
<|user|>
Your message here!
<|assistant|>
```

For best results, format all inputs in this manner. **Make sure to include a newline after `<|assistant|>`, this can affect generation quality quite a bit.**
We have included a [chat template](https://huggingface.co/docs/transformers/main/en/chat_templating) in the tokenizer implementing this template.

## Intended uses & limitations

The model was initially fine-tuned on a filtered and preprocessed of the [Tulu V2 mix dataset](https://huggingface.co/datasets/allenai/tulu-v2-sft-mixture), which contains a diverse range of human created instructions and synthetic dialogues generated primarily by other LLMs. 
We then further trained the model with a [Jax RM trainer](https://github.com/hamishivi/EasyLM/blob/main/EasyLM/models/llama/llama_train_rm.py) built on [EasyLM](https://github.com/young-geng/EasyLM) on the dataset mentioned above.
This model is meant as a research artefact.

### Training hyperparameters

The following hyperparameters were used during PPO training:
- learning_rate: 1e-06
- total_train_batch_size: 512
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear cooldown to 1e-05.
- lr_scheduler_warmup_ratio: 0.03
- num_epochs: 1.0

## Citation

If you find Tulu 2.5 is useful in your work, please cite it with:

```
@misc{ivison2024unpacking,
      title={{Unpacking DPO and PPO: Disentangling Best Practices for Learning from Preference Feedback}}, 
      author={{Hamish Ivison and Yizhong Wang and Jiacheng Liu and Ellen Wu and Valentina Pyatkin and Nathan Lambert and Yejin Choi and Noah A. Smith and Hannaneh Hajishirzi}}
      year={2024},
      eprint={2406.09279},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}
```