REAL-Q：通过动态梯度下降实现端到端大语言模型量化

AI 摘要

REAL-Q 提出了一种新的后训练量化（PTQ）方法，通过端到端对齐的 Fisher MSE 目标和动态块级梯度下降，解决了现有方法中信息错位的问题。在 LLaMA-3.1 和 Qwen3 模型上，REAL-Q 在 W4A16 下将端到端 KL 散度最多降低 49%。该方法通过细粒度校正和滑动窗口机制，有效缓解了跨层误差传播。

正文节选

REAL-Q: E2E LLM Quantization via Dynamic Gradient Descent Abstract Post-training quantization (PTQ) is essential for deploying large language models (LLMs) under strict resource constraints. State-of-the-art PTQ methods quantize each layer with a single closed-form second-order solver: to remain analytically tractable, they heavily approximate the global loss (dropping cross-channel coupling, pooling output rows into groups), and they then freeze the resulting Hessian across the entire layer, wi

发布时间：2026-09-02 12:00

抓取时间：2026-09-02 12:13

来源机构：arXiv