SOAR improves NVFP4 post-training quantization accuracy for LLMs by analytically solving joint scale optimization and searching decoupled scales.
Resq: Mixed-precision quantization of large language models with low-rank residuals
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
SOAR: Scale Optimization for Accurate Reconstruction in NVFP4 Quantization
SOAR improves NVFP4 post-training quantization accuracy for LLMs by analytically solving joint scale optimization and searching decoupled scales.