How can we train semantic parsers if neither question-parse nor question-answer pairs can be collected?
StatNLP HD Blog
This is the blog of the Statistical NLP Group at the Department of Computational Linguistics, Heidelberg University.
Our research addresses various aspects of the problem of the confusion of languages, by means of statistical learning for natural language processing.
We blog about pitfalls in methodologies, recent advances and important problems in this field of research.
Discussing good, bad and ugly practices of reinforcement learning in neural machine translation.
This post explains the need for the score function gradient estimator trick and how it works.