Direction is what you need: Improving Word Embedding Compression in Large Language Models
Tytuł:
Direction is what you need: Improving Word Embedding Compression in Large Language Models
Czasopismo:
PROCEEDINGS OF THE 6TH WORKSHOP ON REPRESENTATION LEARNING FOR NLP (REPL4NLP-2021)
Rok:
2021
Strony:
322–330
Numer DOI:
10.18653/v1/2021.repl4nlp-1.32