V-PETL Bench: A Unified Visual Parameter-Efficient Transfer Learning Benchmark
Abstract
Parameter-efficient transfer learning (PETL) methods show promise in adapting a pre-trained model to various downstream tasks while training only a few parameters. In the computer vision (CV) domain, numerous PETL algorithms have been proposed, but their direct employment or comparison remains inconvenient. To address this challenge, we construct a Unified Visual PETL Benchmark (V-PETL Bench) for the CV domain by selecting 30 diverse, challenging, and comprehensive datasets from image recognition, video action recognition, and dense prediction tasks. On these datasets, we systematically evaluate 25 dominant PETL algorithms and open-source a modular and extensible codebase for fair evaluation of these algorithms. V-PETL Bench runs on NVIDIA A800 GPUs and requires approximately 310 GPU days. We release all the benchmark, making it more efficient and friendly to researchers. Additionally, V-PETL Bench will be continuously updated for new PETL algorithms and CV tasks.
Cite
Text
Xin et al. "V-PETL Bench: A Unified Visual Parameter-Efficient Transfer Learning Benchmark." Neural Information Processing Systems, 2024. doi:10.52202/079017-2560Markdown
[Xin et al. "V-PETL Bench: A Unified Visual Parameter-Efficient Transfer Learning Benchmark." Neural Information Processing Systems, 2024.](https://mlanthology.org/neurips/2024/xin2024neurips-vpetl/) doi:10.52202/079017-2560BibTeX
@inproceedings{xin2024neurips-vpetl,
title = {{V-PETL Bench: A Unified Visual Parameter-Efficient Transfer Learning Benchmark}},
author = {Xin, Yi and Luo, Siqi and Liu, Xuyang and Du, Yuntao and Zhou, Haodi and Cheng, Xinyu and Lee, Christina and Du, Junlong and Wang, Haozhe and Chen, Mingcai and Liu, Ting and Hu, Guimin and Wan, Zhongwei and Zhang, Rongchao and Li, Aoxue and Yi, Mingyang and Liu, Xiaohong},
booktitle = {Neural Information Processing Systems},
year = {2024},
doi = {10.52202/079017-2560},
url = {https://mlanthology.org/neurips/2024/xin2024neurips-vpetl/}
}