This code accompanies the paper Low-Shot Learning for Fictional Claim Verification.
Steps to run:
- install conda env using conda_env.yml:
conda env create --file=conda_env_gpu.yml - install spacy lexicon:
python -m spacy download en_core_web_sm - see model training options and train a model:
python train.py --help python clean_data.pyto delete all cached data ONLY for debugging purposes (warning: this will destroy your cached data!)
Reproducibility: Despite our efforts to the contrary, reproducibility is not assured. Results that are slightly different than our own can occur with even small changes such as different CUDA versions or different GPUs (https://pytorch.org/docs/stable/notes/randomness.html for more info).