LSG uses an LLM's own distribution difference from a wait-1 baseline to decide read/write timing, and reports strong results on simultaneous translation and streaming ASR.
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
-
Large Language Models Are Read/Write Policy-Makers for Simultaneous Generation
LSG uses an LLM's own distribution difference from a wait-1 baseline to decide read/write timing, and reports strong results on simultaneous translation and streaming ASR.