A proof of concept bi-directional LSTM text classifier written in TensorFlow.
This repo explores sentiment classification on the Cornell Movie Review Dataset and on the Stanford Large Movie Review Dataset.
The purpose of this repo is to explore the stacked, bidirectional LSTM architecture on text input.
The preferred method for starting this notebook is to create a new environment in conda
:
$ conda create --name TextClassification
$ source activate TextClassification
$ conda install pip
$ pip install -r requirements.txt
$ jupyter notebook