This is an example code for the paper titled Calibrating Transformers via Sparse Gaussian Processes (ICLR 2023)
This code implememts SGPA on CIFAR10 and CoLA datasets.
To use this code: simply run train_cifar.py or train_cola.py
The CoLA dataset can be downloaded here
Dependencies:
- Python - 3.8
- Pytorch - 1.10.2
- numpy - 1.22.4
- einops - 0.4.1
- allennlp - 2.9.3
@inproceedings{chen2023calibrating,
title = {Calibrating Transformers via Sparse Gaussian Processes},
author = {Chen, Wenlong and Li, Yingzhen},
booktitle = {International Conference on Learning Representations},
year = {2023}
}