Skip to content

Official implementation of Transformer Neural Processes

License

Notifications You must be signed in to change notification settings

tung-nd/TNP-pytorch

Repository files navigation

Transformer Neural Processes: Uncertainty-Aware Meta Learning Via Sequence Modeling

This is the official implementation of the paper Transformer Neural Processes: Uncertainty-Aware Meta Learning Via Sequence Modeling in Pytorch. We propose Transformer Neural Processes (TNPs), a new member of the Neural Processes family that casts uncertainty-aware meta learning as a sequence modeling problem. We learn TNPs via an autoregressive likelihood-based objective and instantiate it with a novel transformer-based architecture. TNPs achieve state-ofthe-art performance on various benchmark problems, outperforming all previous NP variants on meta regression, image completion, contextual multi-armed bandits, and Bayesian optimization.

Install

First, clone the repository:

git clone https://github.com/tung-nd/TNP-pytorch.git

Then install the dependencies as listed in env.yml and activate the environment:

conda env create -f env.yml
conda activate tnp

Usage

Please check the directory of each task for specific usage.

Citation

If you find this repo useful in your research, please consider citing our paper:

@article{nguyen2022transformer,
  title={Transformer neural processes: Uncertainty-aware meta learning via sequence modeling},
  author={Nguyen, Tung and Grover, Aditya},
  journal={arXiv preprint arXiv:2207.04179},
  year={2022}
}

Acknowledgement

The implementation of the baselines is borrowed from the official code base of Bootstrapping Neural Processes.

About

Official implementation of Transformer Neural Processes

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages