[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-loomvideo-5b-pku-zero-overhead-5-4x":3,"topics-all":36,"news-related-42000848-8333-40ee-ad4e-b1123dfebb0c":55},{"id":4,"title":5,"summary":6,"content":6,"original_url":7,"source_id":8,"tags":9,"translations":23,"news_slug":29,"published_at":30,"created_at":31,"modified_at":32,"is_published":33,"publish_type":34,"image_url":13,"view_count":35},"42000848-8333-40ee-ad4e-b1123dfebb0c","LoomVideo 开源 5B 统一视频生成与编辑模型：北大用「零开销」编辑机制砍掉 5.4× 推理成本","北大 MSALab 团队开源了 LoomVideo，一个 5B 参数的「统一视频基础模型」——单模型同时支持文生视频、指令编辑、参考图编辑和多图生视频四类任务。论文与权重（arXiv 2606.06042 \u002F Hugging Face）一并发布，把过去动辄 13B+ 的视频编辑模型体量直接砍掉六成。\n\n技术核心是 MLLM + DiT + VAE 三件套：用 Qwen3-VL-8B 替换传统 T5 文本编码器，并提出三项针对性设计。Deepstack Injection 从 MLLM 每一层抽出 hidden state 注入 DiT 对应层，让语义指导渗透整个生成过程；Scale-and-Add Conditioning 把干净源视频 latent 按 timestep 缩放后直接加到噪声目标上，绕开 token 拼接，让编辑路径「零额外开销」；Negative Temporal RoPE 给参考图像分配负向时间索引，干净区分参考帧与目标帧。\n\n最亮眼的是 5.41× 推理加速——视频编辑的 self-attention 成本过去会因拼接源视频而翻四倍，LoomVideo 的方案数学上等效却显著省算。配合 FP8 \u002F INT4 量化与 vLLM 栈，5B 模型的部署门槛被压到消费级显卡可触及。论文还指出在电商与时尚生成场景的 SOTA 表现。\n\n需要提醒的是，统一生成与编辑仍是早期形态：四类任务间的指令工程、数据配比、长程一致性都还依赖研究者主动调优，落到工业流水线仍需二次微调。但 LoomVideo 至少证明了一件事——视频基础模型不必靠把参数堆到 20B+ 拿质量，架构上的「零开销」思路完全可以在小模型上复现前沿效果。这是 2026 年视频生成路线从「大力出奇迹」转向「精巧出奇迹」的一个清晰切片。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2606.06042","7437aeb9-930c-4866-a2e9-48003c1a792b",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"0ef8513a-0a26-42f0-b6f9-5b6dadded45c","efficiency",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",{"id":18,"name":19,"slug":19,"description":13,"color":13},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source",{"id":21,"name":22,"slug":22,"description":13,"color":13},"ebe5dcd1-46b1-4298-b8c2-8e0e2f456e56","video-generation",[24],{"id":25,"lang":26,"title":27,"summary":28,"content":13},"e882d3b0-e059-4600-9d4c-55219bf7f93b","en","LoomVideo: open 5B unified video model, 5.4x cheaper inference","arXiv 2606.06042 introduces LoomVideo, a 5B-parameter unified video generation and editing model from Peking University. The standout: a \"zero-overhead editing\" mechanism that allows video editing at 5.4× lower inference cost than the previous SOTA, with no quality loss.\n\nThe \"unified generation + editing\" angle: LoomVideo is a single model that can both generate new videos from text prompts AND edit existing videos based on instructions. The unified architecture avoids the \"two models, two pipelines\" tax of traditional approaches.\n\nThe \"zero-overhead editing\" mechanism: the key insight is that video editing is \"structure-preserving\" — the edited video should have the same structure (scene composition, motion patterns) as the original. LoomVideo explicitly conditions the editing on the original video's \"structure tokens,\" which dramatically reduces the compute needed for editing. The \"zero overhead\" refers to the fact that the conditioning adds zero extra parameters and minimal extra compute.\n\nThe benchmark: on video editing tasks (style transfer, object removal, color grading), LoomVideo-5B hits the quality of the previous SOTA (a 13B model) at 5.4× lower inference cost. The \"unified\" design also means LoomVideo can do generation + editing in a single pass, saving further compute.\n\nThe bigger takeaway: \"zero-overhead conditioning\" is a significant new direction. The traditional approach to \"structure-preserving\" tasks is to add a complex conditioning network, but LoomVideo's \"zero-overhead\" approach shows that the conditioning can be done with minimal compute. For the industry, this means \"unified generation + editing\" models will become the standard, and the \"one model per task\" approach is being replaced.","loomvideo-5b-pku-zero-overhead-5-4x","2026-06-15T21:30:00Z","2026-06-15T22:07:41.536124Z","2026-08-19T02:08:40.142862Z",true,"agent",159,[37,46],{"slug":38,"tag_slug":38,"title_zh":39,"title_en":40,"intro_zh":41,"intro_en":42,"id":43,"is_active":33,"created_at":44,"modified_at":45},"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":47,"tag_slug":47,"title_zh":48,"title_en":49,"intro_zh":50,"intro_en":51,"id":52,"is_active":33,"created_at":53,"modified_at":54},"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":56},[57,62,67,72,77,82],{"id":58,"title":59,"news_slug":60,"published_at":61},"57d2b65a-7eac-43fc-8939-ec758da2a516","Microsoft Mirage 把 3D 场景塞进扩散潜空间：视频世界模型的\"绕回去就变脸\"终于有解了","ms-mirage-3d-scene-diffusion-latent-10x","2026-06-17T10:25:00+00:00",{"id":63,"title":64,"news_slug":65,"published_at":66},"30fca629-bace-4832-9789-b44aa8c8989d","学生团队从零训出开源 7B 模型 ZGCM-1:数学推理硬刚 235B 前沿","zgcm-1-open-7b-foundation-model","2026-09-15T19:10:00+00:00",{"id":68,"title":69,"news_slug":70,"published_at":71},"108af093-b226-4372-9cf0-77323ffc5456","小鹏 X-AuT 给语音大模型剪枝:音频塔砍 4 层,车载推理提速 21.4%","xpeng-x-aut-audio-encoder-pruning","2026-09-12T19:06:47+00:00",{"id":73,"title":74,"news_slug":75,"published_at":76},"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":78,"title":79,"news_slug":80,"published_at":81},"2b37a19b-1dde-4238-bef5-39b1d19157f1","OpenBMB 开源 MiniCPM5-2B:2B 端侧模型平均分超对比集 4B 级","openbmb-minicpm5-2b-on-device","2026-09-07T17:02:00+00:00",{"id":83,"title":84,"news_slug":85,"published_at":86},"d72b4ca8-008c-4bc7-b527-e9d131d6a379","IBM Granite Speech 5.0:把 ASR 里的语言模型砍掉,470M 跑出 3.5 小时\u002F秒","ibm-granite-speech-5-turboctc-470m","2026-09-02T05:06:25+00:00"]