Preprint
Aug 2026
ReRound: Reconstructive Rounding to Resolve Midpoint Ambiguity in Calibration-Free LLM Quantization
The ReRound strategy represents a new approach for low-bit quantization and consistently outperforms standard RTN for 3-bit and 4-bit weight quantization across a range of such models, and applies to AI models beyond LLMs.
He-Yen Hsieh, Hsiang-Tsung Kung
· 0 citations