google-research/big_vision
A Google Research codebase for training large-scale vision models at scale using Cloud TPU VMs or GPU machines.
Not currently ranked — collecting fresh signals.
star history
big_vision provides infrastructure for training vision models including Vision Transformers, SigLIP, MLP-Mixer, and LiT. The codebase uses Jax/Flax libraries and TensorFlow data pipelines to enable large-scale distributed training across up to 2048 TPU cores. It serves as both a publication platform for Google research projects and a starting point for large-scale vision experiments.
Frequently asked
- What is google-research/big_vision?
- A Google Research codebase for training large-scale vision models at scale using Cloud TPU VMs or GPU machines.
- Is big_vision open source?
- Yes — google-research/big_vision is open source, released under the Apache-2.0 license.
- What language is big_vision written in?
- google-research/big_vision is primarily written in Jupyter Notebook.
- How popular is big_vision?
- google-research/big_vision has 3.5k stars on GitHub.
- Where can I find big_vision?
- google-research/big_vision is on GitHub at https://github.com/google-research/big_vision.