MoE-Adapters4CL
論文「Mixture-of-Expertsアダプターによる視覚言語モデルの継続学習の強化」CVPR2024のコード。目次
概要
Continual learning can empower vision-language models to continuously acquire new knowledge, without the need for access to the entire historical dataset. However, mitigating the performance degradation in large-scale models is non-trivial due to (i) parameter shifts throughout lifelong learning and (ii) significant computational burdens associated with full-model tuning. In this work, we present a parameter-efficient continual learning framework to alleviate long-term forgetting in incremental learning with vision-language models. Our approach involves the dynamic expansion of a pre-trained CLIP model, through the integration of Mixture-of-Experts (MoE) adapters in response to new tasks. To preserve the zero-shot recognition capability of vision-language models, we further introduce a Distribution Discriminative Auto-Selector (DDAS) that automatically routes in-distribution and out-of-distribution inputs to the MoE Adapter and the original CLIP, respectively. Through extensive experiments across various settings, our proposed method consistently outperforms previous state-of-the-art approaches while concurrently reducing parameter training burdens by 60%.アプローチ
___
インストール
conda create -n MoE_Adapters4CL python=3.9
conda activate MoE_Adapters4CL
conda install pytorch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 pytorch-cuda=11.8 -c pytorch -c nvidia
cd cil
pip install -r requirements.txtデータ準備
対象データセット: Aircraft, Caltech101, CIFAR10, CIFAR100, DTD, EuroSAT, Flowers, Food, MNIST, OxfordPet, StanfordCars, SUN397, TinyImagenet。Caltech101で問題がある場合は、issue#6を参照してください。
詳細はZSCLのdatasets.mdを参照してください。素晴らしい仕事に感謝します！
モデルckpt
| | モデル | リンク | |------------------|----------------------------------------------------------------------|---------------------------------------------------------------------- | | full_shot_order1 | full_shot_order1_1000iters.pth | Baidu Disk / Google Drive | | few_shot_order1 | few_shot_order1_1000iters.pth | Baidu Disk / Google Drive |MTCL
テスト段階
例:- チェックポイントをMoE-Adapters4CL/ckptに移動
- ``
cd MoE-Adapters4CL/mtil` - スクリプトを実行 `
bash srcipts/test/Full_Shot_order1.sh`
トレーニング段階
例:- チェックポイントをMoE-Adapters4CL/ckptに移動
- `
cd MoE-Adapters4CL/mtil` - スクリプトを実行 `
bash srcipts/train/train_full_shot_router11_experts22_1000iters.sh`
クラス増分学習
トレーニング段階
例:- `
cd cil` - `
bash run_cifar100-2-2.sh``
引用
@inproceedings{yu2024boosting,
title={Boosting continual learning of vision-language models via mixture-of-experts adapters},
author={Yu, Jiazuo and Zhuge, Yunzhi and Zhang, Lu and Hu, Ping and Wang, Dong and Lu, Huchuan and He, You},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={23219--23230},
year={2024}
}謝辞
本リポジトリは wise-ft、Continual-CLIP、および ZSCL を基に構築されています。著者の皆様がコードを共有してくださったことに感謝します。--- Tranlated By Open Ai Tx | Last indexed: 2025-12-04 ---