BPE-knockout: Pruning Pre-existing BPE Tokenisers with Backwards-compatible Morphological Semi-supervision.

Thomas Bauwens Pieter Delobelle

Published in: NAACL-HLT (2024)

Keyphrases

pruning method
multiscale
structuring elements
mathematical morphology
search space
morphological operations
databases
active learning
database
information systems
case study
expert systems
pruning algorithm
escherichia coli
shape preserving
tree pruning