AMS-QUANT: Adaptive Mantissa Sharing for Floating-point Quantization

Documenti analoghi

• A Convergence Analysis of Adaptive Optimizers under Floating-point Quantization
  di: Tang, Xuan, et al.
  Pubblicazione: (2025)
• Nanoscaling Floating-Point (NxFP): NanoMantissa, Adaptive Microexponents, and Code Recycling for Direct-Cast Compression of Large Language Models
  di: Lo, Yun-Chen, et al.
  Pubblicazione: (2024)
• Achieving binary weight and activation for LLMs using Post-Training Quantization
  di: Song, Siqing, et al.
  Pubblicazione: (2025)
• QUARK: Quantization-Enabled Circuit Sharing for Transformer Acceleration by Exploiting Common Patterns in Nonlinear Operations
  di: Zhao, Zhixiong, et al.
  Pubblicazione: (2025)
• The Hidden Power of Pure 16-bit Floating-Point Neural Networks
  di: Yun, Juyoung, et al.
  Pubblicazione: (2023)