Stream-R1: Reliability-Perplexity Aware Reward Distillation for Streaming Video Generation

• URL: http://arxiv.org/abs/2605.03849v1
• Date: Tue, 05 May 2026 15:15:30 GMT
• Title: Stream-R1: Reliability-Perplexity Aware Reward Distillation for Streaming Video Generation
• Abstract summary: Distillation-based acceleration has become foundational for making autore streaming video diffusion models practical.<n>Existing methods treat every rollout, frame, and pixel as equally reliable supervision.<n>We propose Stream-R1, a Reliability-Perplexity Aware Reward Distillation framework.
• Score: 52.10062792673558
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• 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 Distillation framework that adaptively reweights the distillation objective at both rollout and spatiotemporal-element levels through a single shared reward-guided mechanism. At the Inter-Reliability level, Stream-R1 rescales each rollout's loss by an exponential of a pretrained video reward score, so that rollouts with reliable supervision dominate optimization. At the Intra-Perplexity level, it back-propagates the same reward model to extract per-pixel gradient saliency, which is factored into spatial and temporal weights that concentrate optimization pressure on regions and frames where refinement yields the largest expected gain. An adaptive balancing mechanism prevents any single quality axis from dominating across visual quality, motion quality, and text alignment. Stream-R1 attains consistent improvements on all three dimensions over distillation baselines on standard streaming video generation benchmarks, without architectural modification or additional inference cost.

Related papers

• FIS-DiT: Breaking the Few-Step Video Inference Barrier via Training-Free Frame Interleaved Sparsity [23.184639887235218]
  Video Diffusion Transformers (DiTs) can be substantially reduced through model, but per-step inference latency remains a critical bottleneck.<n>We propose Frame Inter Sparsity DiT (FIS-DiT), a training-free and operator-agnostic framework that shifts the optimization focus from the temporal trajectory to the latent frame position.<n>FIS-DiT consistently achieves 2.11--2.41$times$ speedup with negligible degradation across VBench-Q and CLIP metrics.
  arXiv  Detail & Related papers  (2026-05-12T09:49:36Z)
• Salt: Self-Consistent Distribution Matching with Cache-Aware Training for Fast Video Generation [27.698320788533405]
  Trajectory-style consistency distillation (DMD) can recover sharp, mode-seeking samples, but its local training signals do not explicitly regularize how denoising updates compose across timesteps.<n>We propose Self-Consistent Distribution Matching Distillation (SC-DMD), which explicitly regularizes the endpoint-consistent composition of consecutive denoising updates.
  arXiv  Detail & Related papers  (2026-04-03T15:43:21Z)
• DUO-VSR: Dual-Stream Distillation for One-Step Video Super-Resolution [42.8144221191427]
  Diffusion-based video super-resolution (VSR) has recently achieved remarkable fidelity but still suffers from prohibitive sampling costs.<n>We propose DUO-VSR, a three-stage framework that unifies distribution matching and adversarial supervision for one-step VSR.
  arXiv  Detail & Related papers  (2026-03-23T17:56:17Z)
• Astrolabe: Steering Forward-Process Reinforcement Learning for Distilled Autoregressive Video Models [58.3184497327891]
  Distilled autoregressive (AR) video models enable efficient streaming generation but frequently misalign with human visual preferences.<n>We present Astrolabe, an efficient online reinforcement learning framework tailored for distilled AR models.
  arXiv  Detail & Related papers  (2026-03-17T18:32:18Z)
• Contextual Rollout Bandits for Reinforcement Learning with Verifiable Rewards [69.74686029941881]
  Reinforcement Learning with Verifiable Rewards (RLVR) is an effective paradigm for improving the reasoning capabilities of large language models.<n>We propose a unified neural scheduling framework that adaptively selects high-value rollouts throughout training.<n>Experiments on six mathematical reasoning benchmarks demonstrate consistent gains in performance and training efficiency across multiple RLVR optimization methods.
  arXiv  Detail & Related papers  (2026-02-09T10:51:58Z)
• Reward Forcing: Efficient Streaming Video Generation with Rewarded Distribution Matching Distillation [69.57572900337176]
  We introduce Reward Forcing, a novel framework for efficient streaming video generation.<n> EMA-Sink tokens capture both long-term context and recent dynamics, preventing initial frame copying.<n>Re-DMD biases the model's output distribution toward high-reward regions by prioritizing samples with greater dynamics rated by a vision-language model.
  arXiv  Detail & Related papers  (2025-12-04T11:12:13Z)
• Rethinking Diffusion Model-Based Video Super-Resolution: Leveraging Dense Guidance from Aligned Features [51.5076190312734]
  Video Super-Resolution approaches suffer from error accumulation, spatial artifacts, and a trade-off between perceptual quality and fidelity.<n>We propose a novelly Guided diffusion model with Aligned Features for Video Super-Resolution (DGAF-VSR)<n>Experiments on synthetic and real-world datasets demonstrate that DGAF-VSR surpasses state-of-the-art methods in key aspects of VSR.
  arXiv  Detail & Related papers  (2025-11-21T03:40:45Z)
• Large Scale Diffusion Distillation via Score-Regularized Continuous-Time Consistency [60.74505433956616]
  continuous-time consistency model (sCM) is theoretically principled and empirically powerful for accelerating academic-scale diffusion.<n>We first develop a parallelism-compatible FlashAttention-2 JVP kernel, enabling sCM training on models with over 10 billion parameters and high-dimensional video tasks.<n>We propose the score-regularized continuous-time consistency model (rCM), which incorporates score distillation as a long-skip regularizer.
  arXiv  Detail & Related papers  (2025-10-09T16:45:30Z)
• POSE: Phased One-Step Adversarial Equilibrium for Video Diffusion Models [18.761042377485367]
  POSE (Phased One-Step Equilibrium) is a distillation framework that reduces the sampling steps of large-scale video diffusion models.<n>We show that POSE outperforms other acceleration methods on VBench-I2V by average 7.15% in semantic alignment, temporal conference and frame quality.
  arXiv  Detail & Related papers  (2025-08-28T17:20:01Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.