huggingface/optimum
A Hugging Face library that accelerates ML model inference and training through hardware-specific optimizations and quantization.
Not currently ranked — collecting fresh signals.
star history
Optimum extends popular ML libraries (Transformers, Diffusers, TIMM, Sentence-Transformers) with optimization tools for efficient model deployment. It provides hardware-specific acceleration for Intel, GraphCore, and Habana processors, supports quantization techniques, ONNX export, and ONNX Runtime execution. The library aims to maximize inference and training efficiency while maintaining ease of use through a unified API.
Frequently asked
- What is huggingface/optimum?
- A Hugging Face library that accelerates ML model inference and training through hardware-specific optimizations and quantization.
- Is optimum open source?
- Yes — huggingface/optimum is open source, released under the Apache-2.0 license.
- What language is optimum written in?
- huggingface/optimum is primarily written in Python.
- How popular is optimum?
- huggingface/optimum has 3.4k stars on GitHub.
- Where can I find optimum?
- huggingface/optimum is on GitHub at https://github.com/huggingface/optimum.