Home

Awesome

Compositional Cosine Graph Embeddings (Co-CGE)

This is the official PyTorch code of the T-PAMI 2022 paper Learning Graph Embeddings for Open World Compositional Zero-shot Learning.

Note: Most of the code is borrowed from https://github.com/ExplainableML/czsl.

<p align="center"> <img src="utils/img.png" /> </p>

Setup

  1. Clone the repo

  2. We recommend using Anaconda for environment setup. To create the environment and activate it, please run:

    conda env create --file environment.yml
    conda activate czsl
  1. Go to the cloned repo and open a terminal. Download the datasets and embeddings, specifying the desired path (e.g. DATA_ROOT in the example):
    bash ./utils/download_data.sh DATA_ROOT
    mkdir logs

Training

To train Co-CGE for OW-CZSL, the command is simply:

    python train.py --config configs/co-cge/CONFIG_FILE --fast_eval

where CONFIG_FILE is the path to the configuration file of the model. We suggest to use the fast_eval flag to speed up the test phase.

Note that the folder configs contains configuration files for Co-CGE and all other methods, i.e. CGE in configs/cge, CompCos in configs/compcos, and the other methods in configs/baselines.

To train for OW-CZSL a non-open world method, just add --open_world after the command. E.g. for running SymNet in the open world scenario on Mit-States, the command is:

    python train.py --config configs/baselines/mit/symnet.yml --open_world

Note 1: Not all methods are compatible with the fast_eval (e.g. SymNet is one) Note 2: To create a new config, all the available arguments are indicated in flags.py.

Test

Closed World. To test a model, the code is simple:

    python test.py --logpath LOG_DIR

where LOG_DIR is the directory containing the logs of a model.

Open World. To test a model in the open world setting, run:

    python test.py --logpath LOG_DIR --open_world --fast_eval

To test a model trained for OW-CZSL in the closed setting, run:

    python test.py --logpath LOG_DIR --open_world --fast_eval --closed_eval

References

If you use this code, please cite

@inproceedings{mancini2022learning,
  title={Learning Graph Embeddings for Open World Compositional Zero-shot Learning},
  author={Mancini, M and Naeem, MF and Xian, Y and Akata, Zeynep},
  booktitle={IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year={2022},
  organization={IEEE}
}