- python 3.6.13
- pytorch 1.7.0
- numpy 1.19.5
- matplotlib 3.1.2
├── data
│ ├── recipe
│ │ └── total_items.pkl
│ ├── clustered_ESC_CIFAR_TEST.pkl
│ └── clustered_ESC_CIFAR_TRAIN.pkl
├── src
│ ├── utils
│ │ ├── preprocess_ESC_CIFAR.ipynb
│ │ ├── preprocess.py
│ │ ├── Xmodal_dataloader.py
│ │ ├── Xmodal_dataloader_t2i.py
│ │ └── Xmodal_dataloader_v2.py
│ ├── data_utils.py
│ ├── losses.py
│ ├── models.py
│ ├── tools.py
│ ├── train_croma.sh
│ ├── train_proto.py
│ └── train.sh
└── requirements.txt
$ ./train.sh
$ python train_proto.py --mode a2i --train_mode test --load_checkpoint checkpoint.pt
$ ./train_croma.sh
Our code is based on the paper Cross-Modal Generalization: Learning in Low Resource Modalities via Meta-Alignment
and the implementation https://github.com/peter-yh-wu/xmodal