ToxPrune prunes toxic subwords from BPE tokenizers in LLMs to mitigate toxic dialogue responses and improve diversity on both toxic and non-toxic models.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Toxic Subword Pruning for Dialogue Response Generation on Large Language Models
ToxPrune prunes toxic subwords from BPE tokenizers in LLMs to mitigate toxic dialogue responses and improve diversity on both toxic and non-toxic models.