Disclaimer

This repository holds the code used in the CS224N final project in Winter 2018. The code is not longer being actively developed/maintained. Use at your own risk.

For future CS224N students, please read the Honor Code and make sure there is no violations of policies before proceding with reading/using this code base.

cs224n-win18-squad

Code for the Default Final Project (SQuAD) for CS224n, Winter 2018

Note: this code is adapted in part from the Neural Language Correction code by the Stanford Machine Learning Group.

1. Run ensemble models: python code/main.py --enable_cnn=1 --mode=official_eval --enable_ensemble_model=1 --ensemble_model_names="binbin-cnn;binbin-cnn;binbin-cnn" --json_in_path=data/tiny-dev.json --enable_beam_search=1 --context_len=400 --question_len=27

*Also note the default param of --beam_search_size=5 --ensemble_schema='sum' --sum_weights=''