Sentence Generation and Classification with Variational Autoencoder and BERT
Geeling Chau, Anshuman Dewangan, Jin-Long Huang, Keshav Rungta, Margot Wagner
GitHub Repo
Introduction
- Natural-language generation: transformation of structured data into natural language, in this case using AI techniques
- Current SOTA: GPT-3 by OpenAI – 175 billion parameters (May 2020)
- Problem: text generation can tend to be contradictory
- Goal: generate text responses with different levels of contradictoriness
Sentence Generation and Classification with Variational Autoencoder and BERT Geeling Chau, Anshuman Dewangan, Jin-Long Huang, Keshav Rungta, Margot Wagner GitHub Repo