Bias-Aware FP8 Format for Efficient Weight Representation in Post-Training Quantization
Open Access
- 1 January 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 121961-121977
Abstract
No abstract availableFunding Information
- European Union’s Horizon Europe research and innovation programme under the Twinning scheme (101160293)
- U.K. Research and Innovation (UKRI) through U.K. Government’s Horizon Europe (10116105)
- Ministry of Science, Technological Development and Innovation of Serbia (451-03-34/2026-03/200102)
This publication has 17 references indexed in Scilit:
- AIDA4Edge: Twinning for Excellence in Adaptive Edge Artificial Intelligence2025
- Shedding the Bits: Pushing the Boundaries of Quantization with Minifloats on FPGAs2024
- Novel adaptive quantization methodology for 8-bit floating-point DNN trainingDesign Automation for Embedded Systems, 2024
- Floating Point and Fixed Point 32-bits Quantizers for Quantization of Weights of Neural Networks2021
- Iterative algorithm for designing asymptotically optimal uniform scalar quantisation of the one‐sided Rayleigh densityIET Communications, 2021
- Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference2018
- Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification2015
- Variance Mismatch Analysis of Unrestricted Polar Quantization for Gaussian SourceIEEE Signal Processing Letters, 2014
- IEEE Standard for Binary Floating-Point Arithmetic2008
- On the support of MSE-optimal, fixed-rate, scalar quantizersIEEE Transactions on Information Theory, 2001