When Less is More: 8-bit Quantization Improves Continual Learning in Large Language Models

Documenti analoghi

• VPTQ: Extreme Low-bit Vector Post-Training Quantization for Large Language Models
  di: Liu, Yifei, et al.
  Pubblicazione: (2024)
• Quantize More, Lose Less: Autoregressive Generation from Residually Quantized Speech Representations
  di: Han, Yichen, et al.
  Pubblicazione: (2025)
• Post Training Quantization of Large Language Models with Microscaling Formats
  di: Sharify, Sayeh, et al.
  Pubblicazione: (2024)
• Less is More: Undertraining Experts Improves Model Upcycling
  di: Horoi, Stefan, et al.
  Pubblicazione: (2025)
• Continual Quantization-Aware Pre-Training: When to transition from 16-bit to 1.58-bit pre-training for BitNet language models?
  di: Nielsen, Jacob, et al.
  Pubblicazione: (2025)