CNN for Sentence Classification
THis is an implementation with some modifications of the paper Convolutional Neural Networks for Sentence Classification(2014 Kim et al.).
We use pre-trained word embeddings to train the model, which is wiki-news-300d-1M.vec.zip of fastText, an open-sourced library for efficient text classification and representation learning by facebook. Before running the code, you should first download and put it under the word_embeddings directory. The dataset used can be found at CS 11-747 Assignment 1 Topic Classification data. You should also download and put it under the dataset directory.
Using the parameter configuration in the code, the best accuracy achieved on the validation set is 84.11%.
To run the code, you should specify the random seed for numpy and torch module to induce reproducibility, as follows.
python train_valid_test.py --seed 1234