[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-reward-lightning-video-generation-distillation":3,"topics-all":33,"news-related-b5909ee4-586c-494a-9353-4d10dee93227":52},{"id":4,"title":5,"summary":6,"content":6,"original_url":7,"source_id":8,"tags":9,"translations":20,"news_slug":26,"published_at":27,"created_at":28,"modified_at":29,"is_published":30,"publish_type":31,"image_url":13,"view_count":32},"b5909ee4-586c-494a-9353-4d10dee93227","Reward Lightning:把「打分器」和「蒸馏器」焊进同一根骨干,1-4 步视频生成的同源解法","视频扩散一直卡在一对老矛盾上:用 RLHF 让画面更对,就要再叠一个奖励模型;想把 50 步去噪压到 4 步以内,就得做蒸馏。两套目标在两张表征空间里互相拉扯,改一个就崩另一个。\n\nECCV 2026 收录的 Reward Lightning (arXiv:2607.03960) 把打分和蒸馏焊进同一根骨干。核心是潜空间奖励模型 LRM——直接在扩散的潜空间里给视频打分,不再绕回像素空间;基于 LRM 的同源偏好蒸馏 HPD 让判别和生成共享权重,从根本上消除梯度冲突。\n\n实测只需 1-4 步就能生成高保真视频,平均 VBench 提升 2.1%,文本对齐、运动质量、视觉质量三个子项均领先现有方法;LRM 本身也比像素级和潜空间级奖励基线分别高 11.0% 和 14.7%。这种对齐+加速同源的设计,大概率会成为 Wan、LongCat-Video、Cosmos 等下一代视频扩散蒸馏的标配范式。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.03960v1","7437aeb9-930c-4866-a2e9-48003c1a792b",[10,14,17],{"id":11,"name":12,"slug":12,"description":13,"color":13},"7b67033c-19e6-4052-a626-e681bba64c7a","diffusion",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"0ef8513a-0a26-42f0-b6f9-5b6dadded45c","efficiency",{"id":18,"name":19,"slug":19,"description":13,"color":13},"ebe5dcd1-46b1-4298-b8c2-8e0e2f456e56","video-generation",[21],{"id":22,"lang":23,"title":24,"summary":25,"content":25},"f4564cbf-ae3e-4b89-ba0c-68b3351039e2","en","Reward Lightning: scorer and distiller share one backbone","Video diffusion has long been stuck on an old pair of contradictions: to make frames \"look more right\" with RLHF, you have to add a reward model; to compress 50 denoising steps to 4, you have to distill. The two goals pull against each other in two different representation spaces — change one and the other breaks. ECCV 2026's accepted Reward Lightning (arXiv:2607.03960) welds scoring and distilling into the same backbone. The core is a latent-space reward model (LRM) that scores videos directly in the diffusion's latent space, no longer detouring back to pixel space; same-source preference distillation (HPD) based on LRM lets discrimination and generation share weights, fundamentally eliminating gradient conflict. With just 1-4 steps, the method generates high-fidelity video, with VBench improving 2.1% on average and leading the existing methods on all three sub-metrics of text alignment, motion quality, and visual quality; LRM itself also beats the pixel-level and latent-space-level reward baselines by 11.0% and 14.7% respectively. This alignment + acceleration same-source design will almost certainly become the standard paradigm for the next generation of video diffusion distillation in Wan, LongCat-Video, Cosmos, and the like.","reward-lightning-video-generation-distillation","2026-07-20T00:15:00Z","2026-07-19T16:10:25.138465Z","2026-08-19T02:08:40.142862Z",true,"agent",215,[34,43],{"slug":35,"tag_slug":35,"title_zh":36,"title_en":37,"intro_zh":38,"intro_en":39,"id":40,"is_active":30,"created_at":41,"modified_at":42},"ai-for-science","AI for Science 2026：从 UniPert 到 GPT-Rosalind 的硬核进化","AI for Science 2026: from UniPert to GPT-Rosalind","生命科学、化学材料、物理世界模型——AI 正在从\"语言工具\"变成\"实验伙伴\"。本专题收录 AI 在三大科学方向的关键节点：UniPert 统一基因与化学扰动空间、GPT-Rosalind 端到端生命科学推理、达摩院 AI 智能体 28 小时找到 4 种超导新材料、Anthropic Claude Science 把工作台做成标准品。","From language tool to lab partner — AI is reshaping life sciences, chemistry\u002Fmaterials, and physical world models. This topic covers the key milestones: UniPert unifying genetic-chemical perturbation spaces, GPT-Rosalind's end-to-end life-sciences reasoning, DAMO's AI agent discovering 4 superconducting materials in 28 hours, and Anthropic's Claude Science workbench going mainstream.","988a4300-5fab-41c4-b5d8-63711a2dc757","2026-09-10T01:34:15.296649Z","2026-09-10T01:34:15.296663Z",{"slug":44,"tag_slug":44,"title_zh":45,"title_en":46,"intro_zh":47,"intro_en":48,"id":49,"is_active":30,"created_at":50,"modified_at":51},"h3-series","MiniMax H3 系列：从开源权重到 35 倍吞吐","MiniMax H3 Series: from open weights to 35x throughput","MiniMax H3 自 2026 年 8 月开源以来节奏密集：官方把生成、参考与编辑收回一个模型；ComfyUI 当天压进 RTX 3060；摩尔线程 3 小时完成国产 GPU 适配；fal 后训练版把吞吐拉到 35 倍；FastH3 蒸馏再砍推理成本。本专题持续追踪 H3 的发布—开源—蒸馏—部署全链路。","Since MiniMax open-sourced H3 in August 2026 the pace has been relentless: one unified omni-modal model, same-day ComfyUI support down to an RTX 3060, a 3-hour Day-0 port to Moore Threads GPUs, fal's post-trained H3 Max at 35x throughput, and FastH3 distillation cutting inference cost further. This topic tracks the full H3 chain — release, open weights, distillation, deployment.","83ef0daa-3c31-4cb3-86ed-e5ee58654d5f","2026-09-08T07:33:19.942193Z","2026-09-08T07:33:19.942209Z",{"items":53},[54,59,64,69,74,79],{"id":55,"title":56,"news_slug":57,"published_at":58},"caed836e-2168-418f-b5c1-bde3ce962e66","Mask Forcing 往蒸馏 rollout 里掺干净 token:修视频生成的模式坍缩,指令遵循最高涨 6.5 分","mask-forcing-video-diffusion-distillation","2026-09-09T23:08:37+00:00",{"id":60,"title":61,"news_slug":62,"published_at":63},"02c8b500-ec11-44a6-8c58-6e880563dad8","FastH3 开源:4 步蒸馏版 MiniMax H3,B200 单卡最高提速 14 倍","fasth3-4-step-distilled-minimax-h3","2026-08-30T21:30:00+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"18d2aa73-7244-4b10-b611-46475e17327e","ForgeWM开源:一步去噪72FPS的可玩世界模型,8张卡复现全流程","forgewm-few-step-playable-world-model","2026-08-24T21:10:00+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"0599b775-ac17-49d2-aebd-a16f531c7168","腾讯混元 MeanFlowNFT：把 RL 接进「平均速度生成器」，Wan 2.1 4 步反超 50 步 LongCat-Video RL","tencent-hunyuan-meanflownft","2026-07-16T12:00:00+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"0e44f256-e66e-495c-82e3-aae4dd5e2374","LiveEdit 把扩散视频编辑推到 12.66 FPS：清华让 AR 实时编辑走出 PPT","liveedit-ar-video-editing","2026-07-01T06:15:00+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"2f01f1ec-b078-4aca-afa2-654dc48cc784","Video-Mirai：自回归视频扩散的「远见」机制，零推理成本打破长程漂移","video-mirai-foresight-ar-diffusion-zero-cost","2026-06-08T12:15:00+00:00"]