Deep-Attention Text Classifier in Tensorflow

This repository contains code for text classification using attention mechanism in Tensorflow with tensorboard visualization.

Requirements

• Python 3.6
  
• Tensorflow 1.2.1
  
• Numpy
  

Project Module

• utility_dir: storage module for data, vocab files, saved models, tensorboard logs, outputs.

• pre_processing_module: code for pre-processing text file which includes sampling infrequent words, creation of training vocab and classes in form of pickle dictionary.

• implementation_module: code for model architecture, data reader, training pipeline and test pipeline.

• settings_module: code to set directory paths (data path, vocab path, model path etc.), set model parameters (hidden dim, attention dim, regularization, dropout etc.), set vocab dictionary.

• run_module: wrapper code to execute end-to-end train and test pipeline.

• viz_module: code to generate embedding visualization via tensorboard.

• utility_code: other utility codes

How to run

• train: python -m global_module.run_module.run_train

• test: python -m global_module.run_module.run_test

• visualize tensorboard: tensorboard --logdir=PATH-TO-LOG-DIR

Data Sample

• Utterance file

  • it is hard to resist
  • But something seems to be missing .
  • A movie of technical skill and rare depth of intellect and feeling .
  • Brosnan is more feral in this film than I 've seen him before
  • . . .

• Utterance label

  • neg
  • neg
  • pos
  • neg
  • . . .

How to change model parameters

Go to set_params.py here.

Model Graph

Loss and Accuracy Plots

#Histogram

Embedding Visualization