---
license: mit
language:
- en
tags:
- biology
- medical
- protein-language-model
base_model:
- westlake-repl/SaProt_650M_AF2
- meta-llama/Llama-3.1-8B-Instruct
---

# Evolla-10B-DPO

*A frontier protein-language generative model — because proteins deserve better small talk.*

<a href="http://www.chat-protein.com/"><img src="https://img.shields.io/badge/Demo-Live_Chat-blue" style="max-width: 100%;" alt="Live Demo"></a>
<a href="https://doi.org/10.1101/2025.01.05.630192"><img src="https://img.shields.io/badge/Paper-bioRxiv-green" style="max-width: 100%;" alt="Paper on bioRxiv"></a>
<a href="https://github.com/westlake-repl/Evolla"><img src="https://img.shields.io/badge/GitHub-Original_Repo-black" style="max-width: 100%;" alt="GitHub Repository"></a>
<a href="https://x.com/duguyuan/status/1876446845951492221"><img src="https://img.shields.io/badge/Post-X-black" style="max-width: 100%;" alt="Post on X"></a>

## Model Description

**Evolla** is an advanced 80-billion-parameter (with 10B variants) protein-language generative model designed to decode the molecular language of proteins. It integrates information from protein sequences, structures, and user queries to generate precise and contextually nuanced insights into protein function.

This specific repository contains the **10B parameter model**, refined with **Direct Preference Optimization (DPO)** to improve response quality and relevance based on preference signals. 

**Note:** This set of model parameters is designed to be used with our [original GitHub repository](https://github.com/westlake-repl/Evolla). If you want to use Evolla directly with the standard 🤗 Transformers library, please check out the [Official Evolla Documentation in Transformers](https://huggingface.co/docs/transformers/model_doc/evolla) and use [Evolla-10B-hf](https://huggingface.co/westlake-repl/Evolla-10B-hf) or [Evolla-10B-DPO-hf](https://huggingface.co/westlake-repl/Evolla-10B-DPO-hf).

## Usage with Original Repository

To use this model, you should clone our official repository and set up the environment as follows:

**1. Clone the repository and install dependencies:**
```bash
git clone https://github.com/westlake-repl/Evolla.git
cd Evolla
conda create -n Evolla python=3.10
conda activate Evolla
bash environment.sh
```

**2. Download the model weights:**
```bash
mkdir -p ckpt/huggingface
cd ckpt/huggingface
git lfs install
git clone https://huggingface.co/westlake-repl/Evolla-10B
git clone https://huggingface.co/westlake-repl/Evolla-10B-DPO
git clone https://huggingface.co/westlake-repl/SaProt_650M_AF2
git clone https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct
cd ../..
```

**3. Run Inference:**
You can run the model using the provided inference script. Prepare your inputs in a TSV file (e.g., `examples/inputs.tsv`) and run:
```bash
python scripts/inference.py \
    --config_path config/Evolla_10B.yaml \
    --input_path examples/inputs.tsv
```

For more detailed instructions, please refer to the [Evolla GitHub Repository](https://github.com/westlake-repl/Evolla).

## Citation

If you find Evolla useful in your research, please cite our paper:

```bibtex
@article{zhou2025decoding,
  title={Decoding the molecular language of proteins with evolla},
  author={Zhou, Xibin and Han, Chenchen and Zhang, Yingqi and Du, Huan and Tian, Jiayuan and Su, Jin and Liu, Renju and Zhuang, Kai and Jiang, Shiyu and Gitter, Anthony and others},
  journal={bioRxiv},
  pages={2025--01},
  year={2025},
  publisher={Cold Spring Harbor Laboratory}
}
```