Adopted match-LSTM and Answer-Pointer model developed by Wang and Jiang to implement a machine comprehension system.
Used the Stanford Question Answer Dataset(SQuAD) for this purpose, which is composed of around 100K question-answer pairs, along with a context paragraph.
Python 3
Create the following folders in your code directory:
- Folder 1: \dwr
- Folder 2: \squad
- Folder 3: \data\squadDownload the dataset and glove vectors:
Download the dataset and glove vectors:
- Glove Word Vectors: http://nlp.stanford.edu/data/glove.6B.zip (folder 1)
- SQuAD dataset: https://rajpurkar.github.io/SQuAD-explorer/ (folder 2)
Order to run the code:
- preproc1.py
- preproc2.py
- model.py