Stream-R1: Reliability-Perplexity Aware Reward Distillation for Streaming Video Generation
From Daily Paper Cast by Jingwen Liang, Gengyu Wang
May 8, 2026 · 22 min · Episode 1842
About this episode
This episode discusses the Stream-R1 framework for improving streaming video generation through reliability and perplexity aware reward distillation.
🤗 Upvotes: 109 | cs.CV Authors: Bin Wu, Mengqi Huang, Shaojin Wu, Weinan Jia, Yuxin Wang, Zhendong Mao, Yongdong Zhang Title: Stream-R1: Reliability-Perplexity Aware Reward Distillation for Streaming Video Generation Arxiv: http://arxiv.org/abs/2605.03849v1 Abstract: Distillation-based acceleration has become foundational for making autoregressive streaming video diffusion models practical, with distribution matching distillation (DMD) as the de facto choice. Existing methods, however, train the student to match the teacher's output indiscriminately, treating every rollout, frame, and pixel as equally reliable supervision. We argue that this caps distilled quality, since it overlooks two complementary axes of variance in DMD supervision: Inter-Reliability across student rollouts whose supervision varies in reliability, and Intra-Perplexity across spatial regions and temporal frames that contribute unequally to where quality can still be improved. The objective thus conflates two questions under a uniform weight: whether to learn from each rollout, and where to concentrate optimization within it. To address this, we propose Stream-R1, a Reliability-Perplexity Aware Reward…
People in this episode
Hosts: Jingwen Liang, Gengyu Wang
Topics covered
- video generation
- reward distillation
- autoregressive models
- machine learning
- computer vision
Keywords
- streaming video
- distillation
- autoregressive models
- machine learning
- computer vision
- reward mechanism
- video diffusion
Mentioned in this episode
Organizations: Arxiv
More episodes of Daily Paper Cast
- EvoArena: Tracking Memory Evolution for Robust LLM Agents in Dynamic Environments · June 13, 2026 · 24 min
- MiniMax Sparse Attention · June 13, 2026 · 26 min
- SpatialClaw: Rethinking Action Interface for Agentic Spatial Reasoning · June 13, 2026 · 23 min
- InterleaveThinker: Reinforcing Agentic Interleaved Generation · June 13, 2026 · 21 min
- FORT-Searcher: Synthesizing Shortcut-Resistant Search Tasks for Training Deep Search Agents · June 13, 2026 · 23 min
- Robust-U1: Can MLLMs Self-Recover Corrupted Visual Content for Robust Understanding? · June 13, 2026 · 20 min
Explore listener stats, chart rankings, contacts and more on the Daily Paper Cast podcast page.