Removing tokens that would not survive re-tokenization at generation time fixes steganographic extraction errors and slightly improves watermark detectability.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Addressing Tokenization Inconsistency in Steganography and Watermarking Based on Large Language Models
Removing tokens that would not survive re-tokenization at generation time fixes steganographic extraction errors and slightly improves watermark detectability.