ROSAQ uses PCA projection to find salient transformer weight channels, keeps those in FP16, quantizes the rest to INT3/4, and reports modest gains over previous quantization methods.
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
-
ROSAQ: Rotation-based Saliency-Aware Weight Quantization for Efficiently Compressing Large Language Models
ROSAQ uses PCA projection to find salient transformer weight channels, keeps those in FP16, quantizes the rest to INT3/4, and reports modest gains over previous quantization methods.