创建

登录 ReadmeX

或

了解我们如何处理你的信息: 隐私政策

ReadmeX

陪你读懂 AI 动态

今天,想先了解什么?

选一份动态总结,或直接聊聊你的问题。

社区和关注动态总结只属于你自己。

资讯

Apple 提出 NTM:四步采样并保留精确似然

AI 总结

苹果机器学习研究团队在 NeurIPS 发表论文,提出 Normalizing Trajectory Models(NTM):把扩散模型每一步反向过程建模为具备精确似然训练的条件归一化流,并在整条轨迹上叠加深层并行预测器,可从头训练,也可从预训练的流匹配模型初始化。其精确轨迹似然还支持自蒸馏——用模型自身诱导的得分函数训练轻量去噪器,四步即可生成高质量样本。在文生图基准上,NTM 仅用四个采样步就达到或超过强基线,同时唯一地保留了对整条生成轨迹的精确似然。

深度见解

此前的少步生成方法往往要放弃似然框架,NTM 显示精确似然与少步高质量采样可以兼得。

content type paperpublished October 2026

Normalizing Trajectory Models

AuthorsJiatao Gu†, Tianrong Chen, Ying Shen‡**, David Berthelot, Shuangfei Zhai, Josh Susskind

View publication

Diffusion-based models decompose sampling into many small Gaussian denoising steps, an assumption that breaks down when generation is compressed to a few coarse transitions. Existing few-step methods address this through distillation, consistency training, or adversarial objectives, but sacrifice the likelihood framework in the process. We introduce Normalizing Trajectory Models (NTM), which models each reverse step as an expressive conditional normalizing flow with exact likelihood training. Architecturally, NTM combines shallow invertible blocks within each step with a deep parallel predictor across the trajectory, forming an end-to-end network trainable from scratch or initializable from pretrained flow-matching models. Its exact trajectory likelihood further enables self-distillation: a lightweight denoiser trained on the score function induced by the model itself produces high-quality samples in four steps. On text-to-image benchmarks, NTM matches or outperforms strong image generation baselines in just four sampling steps while uniquely retaining exact likelihood over the generative trajectory.

  • † University of Pennsylvania
  • ‡ UIUC
  • ** Work done while at Apple

#Apple#Normalizing Trajectory Models

原文约 2 分钟读完阅读原文
群聊

还没有评论,来说说你的看法。