Official code for the manuscript "Prompt Customization for Continual Learning"

For any problems, please feel free to contect me at [email protected]

trained model could be found in "https://pan.baidu.com/s/1vZIpDEgYh23lla59WQzfOQ?pwd=uigy

提取码：uigy"

Environment

The system I used and tested in

• Ubuntu 20.04.4 LTS
• NVIDIA GeForce a100
• Python 3.8

Usage

First, install the packages below:

pytorch==1.12.1
torchvision==0.13.1
timm==0.6.7
pillow==9.2.0
matplotlib==3.5.3

These packages can be installed easily by

pip install -r requirements.txt

Data preparation

If you already have CIFAR-100 datasets, pass your dataset path to --data-path.

If the dataset isn't ready, change the download argument in continual_dataloader.py as follows

datasets.CIFAR100(download=True)

Train

To train a model on CIFAR-100, set the --data-path (path to dataset) and --output-dir (result logging directory) and run the main.py

Evaluation

To evaluate a trained model:

set the--use_env in main.py as --eval
And then run the main.py
Or you can directly evaluate the model by our provided trained model in "https://pan.baidu.com/s/1vZIpDEgYh23lla59WQzfOQ?pwd=uigy.

Thanks for your concerning. This repo is based on the DualPrompt Implementation.