自强化训练让视频扩散模型生成更准，推理时像真的一样。
Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion
- 训练时用自己生成的帧当上下文，模拟真实推理过程。
- 单卡实时生成视频，延迟低于1秒，质量媲美慢得多的非因果模型。
- 适合做视频生成、扩散模型优化的研究者和开发者。
我们提出Self Forcing，一种自回归视频扩散模型的新训练范式，解决长期存在的暴露偏差问题：训练时使用真实上下文，推理时却依赖自身生成的不完美输出。与以往基于真实帧去噪不同，Self Forcing在训练中通过带键值缓存的自回归滚动生成，使每帧的生成依赖先前自生成的输出。该策略引入视频级整体损失，直接评估整段视频质量，而非仅依赖逐帧目标。为提升效率，采用几步扩散模型结合随机梯度截断策略，在计算成本与性能间取得平衡。进一步设计滚动键值缓存机制，实现高效自回归视频外推。大量实验表明，该方法可在单张GPU上实现亚秒级延迟的实时流式视频生成，生成质量达到甚至超过显著更慢的非因果扩散模型。
原文摘要 · Abstract (English)
We introduce Self Forcing, a novel training paradigm for autoregressive video diffusion models. It addresses the longstanding issue of exposure bias, where models trained on ground-truth context must generate sequences conditioned on their own imperfect outputs during inference. Unlike prior methods that denoise future frames based on ground-truth context frames, Self Forcing conditions each frame's generation on previously self-generated outputs by performing autoregressive rollout with key-value (KV) caching during training. This strategy enables supervision through a holistic loss at the video level that directly evaluates the quality of the entire generated sequence, rather than relying solely on traditional frame-wise objectives. To ensure training efficiency, we employ a few-step diffusion model along with a stochastic gradient truncation strategy, effectively balancing computational cost and performance. We further introduce a rolling KV cache mechanism that enables efficient autoregressive video extrapolation. Extensive experiments demonstrate that our approach achieves real-time streaming video generation with sub-second latency on a single GPU, while matching or even surpassing the generation quality of significantly slower and non-causal diffusion models. Project website: http://self-forcing.github.io/
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。