facebookresearch/hiera
A hierarchical vision transformer model designed for computer vision tasks, published as an ICML 2023 Oral paper.
Not currently ranked — collecting fresh signals.
star history
Hiera is a hierarchical vision transformer architecture that simplifies the design of vision transformers by removing unnecessary components while maintaining strong performance. The model achieves state-of-the-art results on image classification and other vision tasks with faster training and inference speeds. It provides implementations compatible with PyTorch Hub and Hugging Face Hub.
Frequently asked
- What is facebookresearch/hiera?
- A hierarchical vision transformer model designed for computer vision tasks, published as an ICML 2023 Oral paper.
- Is hiera open source?
- Yes — facebookresearch/hiera is open source, released under the Apache-2.0 license.
- What language is hiera written in?
- facebookresearch/hiera is primarily written in Python.
- How popular is hiera?
- facebookresearch/hiera has 1.1k stars on GitHub.
- Where can I find hiera?
- facebookresearch/hiera is on GitHub at https://github.com/facebookresearch/hiera.