Updating only normalization affine parameters under the target fake-quantization graph recovers severely degraded low-bit quantized models while training less than 1.43% of the parameters.
Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Low-Dimensional High-Leverage Subspace Optimization: Beyond Full-Parameter Coupled Training for Neural Network Quantization
Updating only normalization affine parameters under the target fake-quantization graph recovers severely degraded low-bit quantized models while training less than 1.43% of the parameters.