Quantization Variation: A New Perspective on Training Transformers with Low-Bit Precision

Documenti analoghi

• LLM-FP4: 4-Bit Floating-Point Quantized Transformers
  di: Liu, Shih-yang, et al.
  Pubblicazione: (2023)
• Efficient and Robust Quantization-aware Training via Adaptive Coreset Selection
  di: Huang, Xijie, et al.
  Pubblicazione: (2023)
• MARR: Module-Adaptive Residual Reconstruction for Low-Bit Post-Training Quantization
  di: Su, Le, et al.
  Pubblicazione: (2026)
• Progressive Fine-to-Coarse Reconstruction for Accurate Low-Bit Post-Training Quantization in Vision Transformers
  di: Ding, Rui, et al.
  Pubblicazione: (2024)
• MimiQ: Low-Bit Data-Free Quantization of Vision Transformers with Encouraging Inter-Head Attention Similarity
  di: Choi, Kanghyun, et al.
  Pubblicazione: (2024)