---
license: mit
license_link: https://huggingface.co/Vezora/WaveCoder-6.7b-Ultra-bf16/blob/main/LICENSE
language:
  - en
library_name: transformers
datasets:
  - humaneval
pipeline_tag: text-generation
tags:
  - code
metrics:
  - code_eval
---


## This is a Re-Upload of Wave-Coder-Ultra in bf16 since original model was uploaded in fp32 and there are none others available. Licensing remains the same as original model. (Half the Size as original, half the download time.)


<h1 align="center">
🌊 WaveCoder: Widespread And Versatile Enhanced Code LLM
</h1>

<p align="center">
  <a href="https://arxiv.org/abs/2312.14187"><b>[📜 Paper]</b></a> •
  <!-- <a href=""><b>[🤗 HF Models]</b></a> • -->
  <a href="https://github.com/microsoft/WaveCoder"><b>[🐱 GitHub]</b></a>
  <br>
  <a href="https://twitter.com/TeamCodeLLM_AI"><b>[🐦 Twitter]</b></a> •
  <a href="https://www.reddit.com/r/LocalLLaMA/comments/19a1scy/wavecoderultra67b_claims_to_be_the_2nd_best_model/"><b>[💬 Reddit]</b></a> •
  <a href="https://www.analyticsvidhya.com/blog/2024/01/microsofts-wavecoder-and-codeocean-revolutionize-instruction-tuning/">[🍀 Unofficial Blog]</a>
  <!-- <a href="#-quick-start">Quick Start</a> • -->
  <!-- <a href="#%EF%B8%8F-citation">Citation</a> -->
</p>

<p align="center">
Repo for "<a href="https://arxiv.org/abs/2312.14187" target="_blank">WaveCoder: Widespread And Versatile Enhanced Instruction Tuning with Refined Data Generation</a>" 
</p>

## 🔥 News

- [2024/04/10] 🔥🔥🔥 WaveCoder repo, models released at [🤗 HuggingFace](https://huggingface.co/microsoft/wavecoder-ultra-6.7b)!
- [2023/12/26] WaveCoder paper released.

## 💡 Introduction

WaveCoder 🌊 is a series of large language models (LLMs) for the coding domain, designed to solve relevant problems in the field of code through instruction-following learning. Its training dataset was generated from a subset of code-search-net data using a generator-discriminator framework based on LLMs that we proposed, covering four general code-related tasks: code generation, code summary, code translation, and code repair.

| Model                                                                            | HumanEval | MBPP(500) | HumanEval<br>Fix(Avg.) | HumanEval<br>Explain(Avg.) |
| -------------------------------------------------------------------------------- | --------- | --------- | ---------------------- | -------------------------- |
| GPT-4                                                                            | 85.4      | -         | 47.8                   | 52.1                       |
| [🌊 WaveCoder-DS-6.7B](https://huggingface.co/microsoft/wavecoder-ds-6.7b)       | 65.8      | 63.0      | 49.5                   | 40.8                       |
| [🌊 WaveCoder-Pro-6.7B](https://huggingface.co/microsoft/wavecoder-pro-6.7b)     | 74.4      | 63.4      | 52.1                   | 43.0                       |
| [🌊 WaveCoder-Ultra-6.7B](https://huggingface.co/microsoft/wavecoder-ultra-6.7b) | 79.9      | 64.6      | 52.3                   | 45.7                       |

## 🪁 Evaluation

Please refer to WaveCoder's [GitHub repo](https://github.com/microsoft/WaveCoder) for inference, evaluation, and training code.

```python
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("microsoft/wavecoder-ultra-6.7b")
model = AutoModelForCausalLM.from_pretrained("microsoft/wavecoder-ultra-6.7b")
```

## 📖 License

This code repository is licensed under the MIT License. The use of DeepSeek Coder models is subject to the its [License](https://github.com/deepseek-ai/DeepSeek-Coder/blob/main/LICENSE-MODEL).

## ☕️ Citation

If you find this repository helpful, please consider citing our paper:

```
@article{yu2023wavecoder,
  title={Wavecoder: Widespread and versatile enhanced instruction tuning with refined data generation},
  author={Yu, Zhaojian and Zhang, Xin and Shang, Ning and Huang, Yangyu and Xu, Can and Zhao, Yishujie and Hu, Wenxiang and Yin, Qiufeng},
  journal={arXiv preprint arXiv:2312.14187},
  year={2023}
}
```

## Note

WaveCoder models are trained on the synthetic data generated by OpenAI models. Please pay attention to OpenAI's [terms of use](https://openai.com/policies/terms-of-use) when using the models and the datasets.