Can I read Probing Hyperbole in Language Models on EtoBox?
Probing Hyperbole in Language Models by Sulieman Awad is a document available to read on EtoBox.
What is Probing Hyperbole in Language Models about?
This study investigates the encoding of hyperbolic information in pre-trained language models (PLMs) through probing experiments, revealing that hyperbole is only partially represented, primarily in the final layers. The authors also develop an operational taxonomy for hyperbole and analyze errors in model predictions to enhance hyperbole identification in NLP. The findings suggest that hyperbolic information may be better captured using sentence-level representations due to its pragmatic nature.
- Author
- Sulieman Awad
- Language
- EN