研究人员开发了ResilPhase，一个旨在加速扩散模型推理速度而不牺牲质量的新框架。现有方法在更高加速比下，由于离散外插和数值不稳定的问题，性能常常会下降。ResilPhase通过将加速重新构建为ODE空间中稳定的宏轨迹外插，并将预测与模型的全局漂移对齐来解决这个问题。它采用无导数外插器和有界相位映射来减轻噪声并抑制误差增长，在FLUX.1-dev和HunyuanVideo上展示了最先进的保真度。 AI
影响 该框架可以显著降低扩散模型的推理延迟，使其在实时应用中更具实用性，并降低计算成本。
排序理由 该集群包含一篇详细介绍AI模型加速新技术的论文。
在 Hugging Face Daily Papers 阅读 →
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
arXiv:2606.26769v1 Announce Type: new Abstract: The adoption of powerful diffusion models is hindered by their significant inference latency. Recent ``cache-then-forecast'' schemes alleviate this issue by accelerating DiTs using derivative-based polynomials, but they suffer from …
The adoption of powerful diffusion models is hindered by their significant inference latency. Recent ``cache-then-forecast'' schemes alleviate this issue by accelerating DiTs using derivative-based polynomials, but they suffer from severe quality degradation at high acceleration …
The adoption of powerful diffusion models is hindered by their significant inference latency. Recent ``cache-then-forecast'' schemes alleviate this issue by accelerating DiTs using derivative-based polynomials, but they suffer from severe quality degradation at high acceleration …