General Purpose Sentence Representations

Implementation of the paper 'Learning General Purpose Distributed Sentence Representations via Large Scale Multi-Task Learning' (link) for ELL881 Fundamentals of Deep Learning.

Description of Files

1. preprocess_data.ipynb : Preprocessing code for training and evaluation data for each task
2. NLI_model.ipynb : Natural Language Inference model tensorflow code
3. Constituency_parsing.py : Constituency parsing model tensorflow code
4. nmt.py : Neural Machine Translation model tensorflow code
5. NNs.ipynb : code for finding 5 nearest neighbours of each word in the vocabulary from the learned encoder representation