4-bit量化让扩散模型保持高清细节，不靠额外计算
Post-Training Quantization via Residual Truncation and Zero Suppression for Diffusion Models
- 用残差截断+零位压缩，保留小数值细节
- 4比特下FID达6.98，优于需高精度分支的方法
- 适合部署高清图像生成模型的轻量化场景
扩散模型生成质量高，但计算开销大。尽管8比特异常值感知后训练量化（PTQ）可匹配全精度性能，但扩展到4比特仍具挑战。4比特量化中更大的步长会放大密集低幅值激活的舍入误差，导致细微纹理丢失。我们假设不仅异常值，小激活也对纹理保真度至关重要。为此，提出基于残差截断与零位抑制的4比特PTQ方法QuaRTZ。QuaRTZ采用8比特最小-最大量化处理异常值，并通过前导零压缩降至4比特以保留最低有效位，从而维持纹理细节。该方法通过平衡异常值保护与最低有效位精度，降低舍入误差并提升量化效率。理论推导与实证评估表明QuaRTZ在多种激活分布下具有普适性。显著的是，4比特QuaRTZ在FLUX.1-schnell上达到FID 6.98，优于需辅助FP16分支的SVDQuant。
原文摘要 · Abstract (English)
Diffusion models achieve high-quality image generation but face deployment challenges due to their high computational requirements. Although 8-bit outlier-aware post-training quantization (PTQ) matches full-precision performance, extending PTQ to 4 bits remains challenging. Larger step sizes in 4-bit quantization amplify rounding errors in dense, low-magnitude activations, leading to the loss of fine-grained textures. We hypothesize that not only outliers but also small activations are critical for texture fidelity. To this end, we propose Quantization via Residual Truncation and Zero Suppression (QuaRTZ), a 4-bit PTQ scheme for diffusion models. QuaRTZ applies 8-bit min-max quantization for outlier handling and compresses to 4 bits via leading-zero suppression to retain LSBs, thereby preserving texture details. Our approach reduces rounding errors and improves quantization efficiency by balancing outlier preservation and LSB precision. Both theoretical derivations and empirical evaluations demonstrate the generalizability of QuaRTZ across diverse activation distributions. Notably, 4-bit QuaRTZ achieves an FID of 6.98 on FLUX.1-schnell, outperforming SVDQuant that requires auxiliary FP16 branches.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。