search

TalSchuster / CrossLingualContextualEmb

Cross-Lingual Alignment of Contextual Word Embeddings
MIT License
98 stars 9 forks source link

Few-shot training #8

Closed niless closed 5 years ago

niless commented 5 years ago

Is it possible to train anchored LM for few-shot parsing?

TalSchuster commented 5 years ago

Hi @niless I'm not sure that I understand your question. The language models are to compute contextual embeddings. These embeddings could be used for any downstream task including few-shot parsing. In the paper, we've demonstrated few-shot parsing results for Kazakh from Turkish.

Feel free to shoot me an email with more details if this doesn't answer your question