Deep Learning for Language Workshop prepared for the AI For Social Good Summer Lab, 2018
Contains a notebook and code for developing a sentiment analysis model on a subset of IMDB movie review data documented here
Pang, B., & Lee, L. (2004, July). A sentimental education: Sentiment analysis using subjectivity summarization based on minimum cuts. In Proceedings of the 42nd annual meeting on Association for Computational Linguistics (p. 271). Association for Computational Linguistics.
- Python 3.6
- TensorFlow 1.8
- Jupyter Notebook
Download and extract the pretrained model here.
- train a model from scratch
python sentiment_rnn.py train --name experiment --base-dir /my/base/dir
- continue training a model
python sentiment_rnn.py train --base-dir /my/base/dir/experiment --continue-epoch 2
- runtime from an epoch
python sentiment_rnn.py runtime --base-dir /my/base/dir/experiment --continue-epoch 2