RL in NMT: The Good, the Bad and the Ugly
Discussing good, bad and ugly practices of reinforcement learning in neural machine translation.
Posts by Year
2020
2019
Translating Middle Egyptian Hieroglyphs
This blog post gives an overview of the paper “Multi-Task Modeling of Phonographic Languages: Translating Middle Egyptian Hieroglyphs”.
Joey NMT - A Minimalist NMT Toolkit for Novices
Introducing Joey NMT, a minimalist neural machine translation framework for novices built on Pytorch.
Response-Based and Counterfactual Learning for Sequence-to-Sequence Tasks in NLP: An Overview
This post presents a summary of my PhD thesis. I explored how to learn from feedback given to model outputs when the collection of direct supervision signals...
The Real Challenge of Real-World Reinforcement Learning: The Human Factor
How can we give RL agents that learn from human feedback a possible advantage to succeed in this difficult learning scenario?
Counterfactual Learning of Semantic Parsers When Even Gold Answers Are Unattainable
How can we train semantic parsers if neither question-parse nor question-answer pairs can be collected?
2018
Taming Wild Reward Functions: The Score Function Gradient Estimator Trick
This post explains the need for the score function gradient estimator trick and how it works.