ReCo coordinates KV-cache compression, reflection-token logit penalties, and confidence-based early stopping under one per-step process reward, reducing tokens and latency while largely preserving accuracy.
InFindings of the Association for Computational Linguistics: ACL 2025, 21539–21564
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Fewer Tokens, Smaller Cache: Reward-Coordinated Efficient Reasoning
ReCo coordinates KV-cache compression, reflection-token logit penalties, and confidence-based early stopping under one per-step process reward, reducing tokens and latency while largely preserving accuracy.