QuantFactory/wavecoder-ultra-6.7b-GGUF
1503
1 2---3 4license: mit5license_link: https://huggingface.co/microsoft/wavecoder-ultra-6.7b/blob/main/LICENSE6language:7 - en8library_name: transformers9datasets:10 - humaneval11pipeline_tag: text-generation12tags:13 - code14metrics:15 - code_eval16 17---18 1920 21# QuantFactory/wavecoder-ultra-6.7b-GGUF22This is quantized version of [microsoft/wavecoder-ultra-6.7b](https://huggingface.co/microsoft/wavecoder-ultra-6.7b) created using llama.cpp23 24# Original Model Card25 26 27<h1 align="center">28🌊 WaveCoder: Widespread And Versatile Enhanced Code LLM29</h1>30 31<p align="center">32 <a href="https://arxiv.org/abs/2312.14187"><b>[📜 Paper]</b></a> •33 <!-- <a href=""><b>[🤗 HF Models]</b></a> • -->34 <a href="https://github.com/microsoft/WaveCoder"><b>[🐱 GitHub]</b></a>35 <br>36 <a href="https://twitter.com/TeamCodeLLM_AI"><b>[🐦 Twitter]</b></a> •37 <a href="https://www.reddit.com/r/LocalLLaMA/comments/19a1scy/wavecoderultra67b_claims_to_be_the_2nd_best_model/"><b>[💬 Reddit]</b></a> •38 <a href="https://www.analyticsvidhya.com/blog/2024/01/microsofts-wavecoder-and-codeocean-revolutionize-instruction-tuning/">[🍀 Unofficial Blog]</a>39 <!-- <a href="#-quick-start">Quick Start</a> • -->40 <!-- <a href="#%EF%B8%8F-citation">Citation</a> -->41</p>42 43<p align="center">44Repo for "<a href="https://arxiv.org/abs/2312.14187" target="_blank">WaveCoder: Widespread And Versatile Enhanced Instruction Tuning with Refined Data Generation</a>" 45</p>46 47## 🔥 News48 49- [2024/04/10] 🔥🔥🔥 WaveCoder repo, models released at [🤗 HuggingFace](https://huggingface.co/microsoft/wavecoder-ultra-6.7b)!50- [2023/12/26] WaveCoder paper released.51 52## 💡 Introduction53 54WaveCoder 🌊 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.55 56| Model | HumanEval | MBPP(500) | HumanEval<br>Fix(Avg.) | HumanEval<br>Explain(Avg.) |57| -------------------------------------------------------------------------------- | --------- | --------- | ---------------------- | -------------------------- |58| GPT-4 | 85.4 | - | 47.8 | 52.1 |59| [🌊 WaveCoder-DS-6.7B](https://huggingface.co/microsoft/wavecoder-ds-6.7b) | 65.8 | 63.0 | 49.5 | 40.8 |60| [🌊 WaveCoder-Pro-6.7B](https://huggingface.co/microsoft/wavecoder-pro-6.7b) | 74.4 | 63.4 | 52.1 | 43.0 |61| [🌊 WaveCoder-Ultra-6.7B](https://huggingface.co/microsoft/wavecoder-ultra-6.7b) | 79.9 | 64.6 | 52.3 | 45.7 |62 63## 🪁 Evaluation64 65Please refer to WaveCoder's [GitHub repo](https://github.com/microsoft/WaveCoder) for inference, evaluation, and training code.66 67```python68# Load model directly69from transformers import AutoTokenizer, AutoModelForCausalLM70tokenizer = AutoTokenizer.from_pretrained("microsoft/wavecoder-ultra-6.7b")71model = AutoModelForCausalLM.from_pretrained("microsoft/wavecoder-ultra-6.7b")72```73 74## 📖 License75 76This 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).77 78## ☕️ Citation79 80If you find this repository helpful, please consider citing our paper:81 82```83@article{yu2023wavecoder,84 title={Wavecoder: Widespread and versatile enhanced instruction tuning with refined data generation},85 author={Yu, Zhaojian and Zhang, Xin and Shang, Ning and Huang, Yangyu and Xu, Can and Zhao, Yishujie and Hu, Wenxiang and Yin, Qiufeng},86 journal={arXiv preprint arXiv:2312.14187},87 year={2023}88}89```90 91## Note92 93WaveCoder 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.94 95