#(In Progress)

What are we Transferring Anyway?: Using Attention Weights to Interpret domain choice in Transfer Learning

Implemented

• Data Collection and Cleaning
• Modified Word Language Model
• LSTM Classifier
• Self-Attention Embedding
• Key-Value Attention
• Language Model Pretraining
• Bayesian Optimization
• Classification Tuning
• Attention Classification Tuning

To Do

• Model Comparison

Data

Language Model Datasets

• Wikitext-2
• Gigaword
• Penn Tree Bank

Text Classification Datasets

• IMDB Sentiment Classification
• MPQA Subjectivity Classification

Word Vectors

• CharNGram
• Google News Word2Vec
• GloVe