CoQuant selects optimal high-precision subspaces for mixed-precision LLM quantization via a closed-form weighted PCA that balances weight and activation covariances derived from expected output error.
Kuan Wang, Zhijian Liu, Yujun Lin, Ji Lin, and Song Han
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
-
CoQuant: Joint Weight-Activation Subspace Projection for Mixed-Precision LLMs
CoQuant selects optimal high-precision subspaces for mixed-precision LLM quantization via a closed-form weighted PCA that balances weight and activation covariances derived from expected output error.