README.md

Matching

Dependencies

To train the models, please install all required dependencies in a conda environment:

conda create --name matching --file=environment.yml
conda activate matching

The training was performed on a high performance computing cluster. To train the model efficiently, please ensure that CUDA is installed and available on your system.

You can train the model on a CPU, but it will take significantly longer.

Please also update the .env.example file environment configuration and rename it to .env.

Data

The data used for training the model is not included in this repository due to file size restrictions. You can create a database dump by running the data loading cells of the generate.ipynb notebook.

The data should be located in the ./data/ directory.

Training and Logging

To train the model, either run the train.py script using

python3 -u train.py

or schedule a job on the cluster. To schedule a job on the cluster, connect to it, activate the conda environment and run

sbatch train.slurm

Training the model using train.py does not use pretraining.

To use pretraining, please use the pretrain.py file or schedule a job on the cluster using pretrain.slurm.

Both scripts spawn five different runs using different seeds. Please refer to the respective help messages of the scripts to see all available arguments. You can display the help message by running

python3 train.py --help
python3 pretrain.py --help

We use wandb to log the run values and later evaluate them. Please create an account and log in to wandb using

    wandb login

Please change the wandb project names and entity in matching/config.py to match your project names.