[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-atas-world-state-agentic-video-generation":3,"topics-all":38,"news-related-d0d8e772-57f7-47a0-8fd8-dce09f3bd1e5":57},{"id":4,"title":5,"summary":6,"content":7,"original_url":8,"source_id":9,"tags":10,"translations":24,"news_slug":31,"published_at":32,"created_at":33,"modified_at":34,"is_published":35,"publish_type":36,"image_url":14,"view_count":37},"d0d8e772-57f7-47a0-8fd8-dce09f3bd1e5","浙大ATAS登顶HF论文榜:给AI视频装上持久世界记忆","浙大联合 Monash 与 UC Berkeley 等团队开源 ATAS 框架:把剧本编译成带场景、机位与事件时间的分镜请求,用场景与事件两套记忆维护持久世界状态,在四个视频生成基座上全部取得跨镜一致性评测最高综合分,人类盲评六项标准全数第一,MIT 协议开源。","长视频生成最尴尬的一幕:每个镜头单独看都不错,连成一部片子就穿帮——角色换了衣服,房间换了布局,昨天发生过的事今天演成了另一版本。单镜质量卷到 0.99 的今天,瓶颈早就不是「画得像不像」,而是「记不记得住」。\n\n此前的多镜流水线靠在镜头间传 prompt 或最近几帧维持连贯。浙大联合 Monash、UC Berkeley BAIR 与阿德莱德大学的团队指出:传 prompt 给不出共享空间参照系,传几帧也给不出按时间索引的事件记录——回到同一场景,模型不知道房间长什么样;重演旧事件,找不到那段动态证据。他们的答案叫 ATAS(Any Time, Any Space),10 月 14 日挂上 arXiv,次日登顶 Hugging Face Daily Papers,目前 271 赞。\n\n## 框架:编译剧本,维护世界\n\nATAS 的核心对象是一个持久世界状态。一个「世界状态编译器」先把剧本递归拆分成带边界的分镜请求——每条请求显式携带场景、空间锚点、机位、时间段与可选事件,描述部分才是给视频模型的 prompt。生成前,请求向两套记忆检索证据:场景记忆存放已配准的静态几何,以点云形式挂在一个共享坐标系里;事件记忆存放按场景、事件、时间索引的动态观测。生成后,新镜头要经过重建、配准与审查三道关卡,全部通过才写回世界状态。\n\n一个耐人寻味的工程设计:成片与记忆解耦——镜头可以留在电影里,但如果观测质量不过关,它不会成为后续镜头的证据。脏数据不进记忆,这是数据库工程师都懂的纪律。\n\n## 跑分:四个基座全面领先\n\n团队在 MiniMax-H3、Wan 2.7、Wan 3.0、HappyHorse 1.0 四个基座上,把 ATAS 与 ReCA 等五种方法同场对比,综合分全部第一:Wan 3.0 上 0.734 对 ReCA 的 0.648,MiniMax-H3 上 0.702 对 0.659,HappyHorse 1.0 上 0.694 对 0.656,Wan 2.7 上 0.656 对 0.612。优势最大的是跨镜一致性指标:Wan 3.0 上 ST-Bench 0.719 对 0.602,ViStory-Cross 0.584 对 0.481。\n\n两个针对性协议更能说明问题:空间重返测试(回到去过的场景与目标图比对)中,ATAS 的 PSNRC 达 15.968,优于 ViewCrafter 的 15.255 与全部 agent 基线;事件重摄测试中事件身份、时间顺序、视角实现三项领先,ReCA 仅在动作与状态一项略高(0.700 对 0.688)。人类盲评上,ATAS 六项标准均值全部第一:剧本忠实度 4.40、角色一致性 4.22;最强基线 ReCA 的区间是 2.62 到 3.67。\n\n## 代价:省了重试,花了算力\n\n记忆不是白来的。消融显示,从无记忆到完整世界状态,单故事墙钟时间从 51.9 分钟涨到 64.9 分钟,吞吐从每小时 7.28 镜降到 6.21 镜,每镜记忆开销 98 秒;换来的是每镜重试从 1.00 次降到 0.54 次,审查拒绝率从 52.6% 降到 36.8%。另一个细节:状态准入检查的松紧存在甜蜜点——默认设置准入 67.4% 时三项指标最好,过松(90.0%)与过严(36.0%)都掉分。放太多脏数据与拒掉好数据,代价对称。\n\n代码与结果已按 MIT 协议开源([arXiv:2610.20571](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.20571),GitHub:ziplab\u002FATAS)。\n\n所以呢?当单镜生成接近饱和,竞争正在转向「状态管理」——谁能维护一个可信的共享世界,谁才能把 AI 视频从素材生成器变成叙事工具。ATAS 把数据库式的准入纪律带进了生成管线,这个思路比跑分本身更值得抄。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.20571","7437aeb9-930c-4866-a2e9-48003c1a792b",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"120fa59a-ff6f-4537-9bf5-f818df636a0e","benchmark",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"499f4b56-819d-49a3-9609-33e775143b86","multimodal",{"id":19,"name":20,"slug":20,"description":14,"color":14},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source",{"id":22,"name":23,"slug":23,"description":14,"color":14},"ebe5dcd1-46b1-4298-b8c2-8e0e2f456e56","video-generation",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"1868f6b9-06b3-4d60-b541-c4b3d1136fe8","en","ZJU's ATAS Tops HF Papers: Persistent World Memory for AI Video","ZJU's ATAS tops HF Papers: persistent scene and event memories keep AI video consistent across shots, leading all four backbones tested. MIT licensed.","The most awkward moment in long-form AI video: every shot looks fine on its own, but stitch them into a film and things break — the character's outfit changes, the room rearranges itself, yesterday's event replays as a different version. With single-shot quality already pushing 0.99, the bottleneck is no longer \"can it render well\" but \"can it remember.\"\n\nPrevious multi-shot pipelines pass prompts or recent frames between shots. The ATAS team — Zhejiang University with Monash, UC Berkeley BAIR and Adelaide — argues this is not enough: passing prompts provides no shared spatial frame for revisiting a scene, and passing frames gives no time-indexed record for replaying an earlier event. Their answer, ATAS (Any Time, Any Space), landed on arXiv (2610.20571) on October 14 and reached the top of Hugging Face Daily Papers the next day, now at 271 upvotes.\n\n## The framework: compile the screenplay, keep the world\n\nThe central object is a persistent world state. A world-state compiler recursively splits the screenplay into bounded shot requests — each carrying an explicit scene, spatial anchor, camera, time interval and optional event; only the description becomes the prompt for the video model. Before generation, each request retrieves evidence from two memories: scene memory holds registered static geometry as point clouds in one shared coordinate frame, while event memory holds dynamic observations indexed by scene, event and time. After generation, the new clip must pass reconstruction, registration and review before it is written back into the world.\n\nOne design choice deserves attention: film delivery is decoupled from memory admission. A clip can stay in the film even if its observation fails validation and never becomes evidence for later shots. Dirty data stays out of memory — a discipline database engineers know well.\n\n## Results: first place on all four backbones\n\nThe team compared ATAS against I2V Extension, VGoT, Mora, MovieAgent and ReCA on four backbones: MiniMax-H3, Wan 2.7, Wan 3.0 and HappyHorse 1.0. ATAS takes the best average on every backbone: 0.734 vs ReCA's 0.648 on Wan 3.0, 0.702 vs 0.659 on MiniMax-H3, 0.694 vs 0.656 on HappyHorse 1.0, and 0.656 vs 0.612 on Wan 2.7. The widest gap is on cross-shot consistency: ST-Bench 0.719 vs 0.602 and ViStory-Cross 0.584 vs 0.481 on Wan 3.0.\n\nTwo targeted protocols make the point sharper. In the spatial revisit test (return to an earlier location, compare against a target image), ATAS reaches 15.968 PSNRC, ahead of ViewCrafter at 15.255 and every agent baseline. In event recapture, ATAS leads event identity, temporal order and viewpoint realization, while ReCA stays slightly ahead on action and state (0.700 vs 0.688). In human evaluation, ATAS has the highest mean on all six criteria — script faithfulness 4.40, character consistency 4.22 — against ReCA's range of 2.62 to 3.67.\n\n## The cost: fewer retries, more compute\n\nMemory is not free. Ablations show that moving from no memory to the full world state raises per-story wall-clock time from 51.9 to 64.9 minutes and drops throughput from 7.28 to 6.21 shots per hour, with 98 seconds of memory overhead per shot; in exchange, retries fall from 1.00 to 0.54 per shot and the film-review rejection rate drops from 52.6% to 36.8%. Another detail: the state admission check has a sweet spot — the default setting admits 67.4% of candidates and scores best on all three measures, while permissive (90.0%) and strict (36.0%) both lose quality. Admitting too much dirty data and rejecting too much good data cost symmetrically.\n\nCode and results are open-sourced under MIT ([arXiv:2610.20571](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.20571), [GitHub:ziplab\u002FATAS](https:\u002F\u002Fgithub.com\u002Fziplab\u002FATAS)).\n\nSo what? As single-shot generation saturates, competition shifts to state management — whoever maintains a trustworthy shared world turns AI video from a clip generator into a narrative tool. ATAS brings database-style admission discipline into the generation pipeline, and that idea is more worth copying than the scores themselves.","atas-world-state-agentic-video-generation","2026-10-15T15:15:45Z","2026-10-15T15:16:16.150967Z","2026-10-15T15:16:16.150982Z",true,"agent",179,[39,48],{"slug":40,"tag_slug":40,"title_zh":41,"title_en":42,"intro_zh":43,"intro_en":44,"id":45,"is_active":35,"created_at":46,"modified_at":47},"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":49,"tag_slug":49,"title_zh":50,"title_en":51,"intro_zh":52,"intro_en":53,"id":54,"is_active":35,"created_at":55,"modified_at":56},"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":58},[59,64,69,74,79,84],{"id":60,"title":61,"news_slug":62,"published_at":63},"cf7aeaaa-7179-4de2-b177-9adf54668d61","StreamEval：37 模型实测，实时视频最缺听指令","streameval-interactive-streaming-video-benchmark","2026-10-22T11:08:09+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"abd2ad4a-d732-41f0-b1a1-5711103ef7a6","Orbia 基准:生成视频还构不成一致的 3D 世界","orbia-3d-consistency-benchmark","2026-10-18T15:12:16+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"ba0ed7bf-3de3-4f92-98fe-a50d6ac274d0","WROP 开源:用 150 个物体恒存任务给世界模型补认知课","wrop-object-permanence-world-models","2026-09-25T17:08:02+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"b95b93e8-294a-4c5b-b53d-ce6ea07c1519","SemComp-Bench 登顶 Hugging Face 日榜:视频生成开始考「任务做没做成」","semcomp-bench-video-task-completion","2026-08-20T13:30:00+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"9fa15063-58b1-4b87-a0cb-0e19ffc4dc6c","4B 参数横扫四大具身基准：开悟世界模型让小模型重新定义 SOTA","kaiwu-4b-world-model-sensetime-72x","2026-06-12T04:00:00+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"bef7a773-36fb-406b-949d-8d728b97b242","Matrix-Game 4.0 开源:单卡 60FPS 的实时可玩游戏世界","matrix-game-4-0-open-source-60fps","2026-10-21T13:09:32+00:00"]