Single-Loop-bilevel-Optimizer-under-Unbounded-Smoothness

This repository contains PyTorch codes for ICML2024 paper "A Nearly Optimal Single Loop Algorithm for Stochastic Bilevel Optimization under Unbounded Smoothness" Xiaochuan Gong, Jie Hao, Mingrui Liu.

Requirements

PyTorch 2.0

Run experiments

Please check data hyper-cleaning for details of data hyper-cleaning and hyper-representation for meta-learning experiment.

Citation

If you found this repository helpful, please cite our paper:

@inproceedings{gong2024nearly,
title={A Nearly Optimal Single Loop Algorithm for Stochastic Bilevel Optimization under Unbounded Smoothness},
author={Xiaochuan Gong, Jie Hao, Mingrui Liu},
booktitle={41th International Conference on Machine Learning},
year={2024}
}