Skip to content

Latest commit

 

History

History
64 lines (34 loc) · 1.49 KB

README.md

File metadata and controls

64 lines (34 loc) · 1.49 KB
Raw

QUESTION ANSWERING USING MODIFIED QANET

  1. A Tensorflow implementation of Google's QANet (Note: This is not an official implementation from the authors of the paper)

    Unofficial Implementation source: https://github.com/NLPLearn/QANet

  2. List of modified files: a) prepro.py : Functions modified - get_embedding(),build_features() Functions created - get_POS_one_hot_vector()

    b) model.py : Functions modified - forward()

    c) layers.py : Functions created - drnn(), bidirlstm(), bidirectional_dynamic_rnn()

    d) demo.py : Functions modified - answer()

    e) config.py : Tuned hyperparameters

  3. List of commands:

To download and preprocess the data, run

download SQuAD and Glove

a) sh download.sh

preprocess the data

b) python config.py --mode prepro

hyper parameters are stored in config.py. To debug/train/test/demo, run

c) python config.py --mode debug/train/test/demo

d) The default directory for the tensorboard log file is train/{model_name}/event

Tensorboard

Run tensorboard for visualisation.

$ tensorboard --logdir=./

  1. List of requirements:
  • Python>=2.7
  • NumPy
  • tqdm
  • TensorFlow>=1.5
  • spacy==2.0.9
  • bottle (only for demo)
  • Cuda 10

Dataset

The dataset used for this task is Stanford Question Answering Dataset. Pretrained GloVe embeddings obtained from common crawl with 840B tokens used for words.