Search Results for author: Thomas Gueudre

Found 4 papers, 1 papers with code

Truecasing German user-generated conversational text

no code implementations EMNLP (WNUT) 2020 Yulia Grishina, Thomas Gueudre, Ralf Winkler

True-casing, the task of restoring proper case to (generally) lower case input, is important in downstream tasks and for screen display.

Language Modelling

Constraining word alignments with posterior regularization for label transfer

1 code implementation NAACL (ACL) 2022 Thomas Gueudre, Kevin Jose

Unsupervised word alignments offer a lightweight and interpretable method to transfer labels from high- to low-resource languages, as long as semantically related words have the same label across languages.

Sentence

Knowledge Distillation Transfer Sets and their Impact on Downstream NLU Tasks

no code implementations10 Oct 2022 Charith Peris, Lizhen Tan, Thomas Gueudre, Turan Gojayev, Pan Wei, Gokmen Oz

Yet, the generic corpora used to pretrain the teacher and the corpora associated with the downstream target domain are often significantly different, which raises a natural question: should the student be distilled over the generic corpora, so as to learn from high-quality teacher predictions, or over the downstream task corpora to align with finetuning?

domain classification intent-classification +5

Cannot find the paper you are looking for? You can Submit a new open access paper.