WaveCoder

by Microsoft

About the model:

Not absolutely clear that base rather than instruct-tuned model was used, but can be assumed.

Model type:

Code

Model performance class:

Full

Link to the model:

https://huggingface.co/microsoft/wavecoder-ultra-6.7b

Base models:

Deepseek-Coder-6.7B-Base

End model:

WaveCoder-Ultra-6.7B

End model license:

MIT License

About the organisation:

Major technology company.

Link to the organisation:

https://huggingface.co/microsoft

Model release date:

July 2024

Availability

Base Model Data

Data sources of base model described only at the highest level.

https://arxiv.org/pdf/2401.14196

End User Model Data

End model makes heavy use of synthetic data which does not seem to have been published. Some non-synthetic data was used which is not described in detail.

https://arxiv.org/pdf/2312.14187

Base Model Weights

Weights made available on HuggingFace.

https://huggingface.co/deepseek-ai/deepseek-coder-6.7b-base

End User Model Weights

Weights made available on HuggingFace.

https://huggingface.co/microsoft/wavecoder-ultra-6.7b

Training Code

Training script and data processing architecture made available on GitHub. Fine-tuning script not made available.

https://github.com/microsoft/WaveCoder

Documentation

Code Documentation

Codebase is well-documented.

https://github.com/microsoft/WaveCoder

Hardware Architecture

Hardware architecture described in paper.

https://arxiv.org/pdf/2312.14187

Preprint

Preprint made available on arXiv.

https://arxiv.org/pdf/2312.14187

Paper

Paper published in ACL.

https://aclanthology.org/2024.acl-long.280/

Modelcard

Model card primarily contains information regarding inference and licensing.

https://huggingface.co/microsoft/wavecoder-ultra-6.7b

Datasheet

No datasheet found.

Access

Licenses

Code licensed under MIT license, and OSI-approved license. (Base?) model licensed under DeepSeek's license.

https://huggingface.co/microsoft/wavecoder-ultra-6.7b#%F0%9F%93%96-license

Is this information not up to date?

Contribute here ->

Supported by the Centre for Language Studies and the Dutch Research Council. Website design & development © 2024 by BSTN. This version of the index generated 09 April 2026, website content last updated 11 March 2026.