LightRot uses grouped local rotation plus outlier alignment to make 4-bit LLaMA inference accurate and cheap, claiming 27.4 TOPS/W on a 28nm accelerator.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AR 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
LightRot: A Light-Weighted Rotation Scheme and Architecture for Accurate Low-Bit Large Language Model Inference
LightRot uses grouped local rotation plus outlier alignment to make 4-bit LLaMA inference accurate and cheap, claiming 27.4 TOPS/W on a 28nm accelerator.