[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-lala-world-language-action-latent-world-model":3,"topics-all":38,"news-related-673fb0b6-cf7f-419e-ae5b-e85c95317c7f":67},{"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},"673fb0b6-cf7f-419e-ae5b-e85c95317c7f","LaLa-World:语言当动作接口,训练成本砍到1\u002F6","NAVER 发布潜在世界模型 LaLa-World:把动作与视觉变化写成自然语言,用语言模型当预测器,跨仿真、机器人、图像、视频四类数据联合训练。对照视频生成路线,训练成本约 1\u002F6、规划快 13.9 倍;微调出的机器人策略与 Cosmos Policy 相当,推理最高快 5.5 倍。权重尚未放出。","世界模型赛道过去两年的主流叙事,是用视频生成模型学会\"想象\":先在海量视频上训练一个生成模型,再拿它做规划和策略学习。这条路线的代价众所周知——生成像素太贵,训练烧算力,推理时在像素空间里滚动一步都要付出完整的前向开销。NAVER AI Lab 本周联合 Georgia Tech、Brown 大学和 Mila 发布的 LaLa-World 给出了反方向样本:不生成像素,只预测潜在表示,而最难统一的\"动作\"这一维,直接写成自然语言。\n\n## 核心设计:动作也是文字\n\nLaLa-World 是语言条件的潜在世界模型,论文编号 [arXiv:2610.27851](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.27851)。两个设计决定了它能把异构数据揉进同一个模型:第一,数值动作和视觉变化统一用语言描述——机械臂的动作增量按归一化规格渲染成文本,图像与视频数据则用变化描述和编辑指令表达,于是同一个未来预测器可以同时吃仿真轨迹(Wall、PointMaze、Push-T、Meta-World 等)、机器人数据集 DROID、OpenVid-1M 帧序列和 OmniEdit 编辑对;第二,预测器直接建在语言模型上——一个 24 层的 Qwen3.5-0.8B,配上冻结的视觉编码器(DINOv2、DINOv3、SigLIP2 三种配置),输入当前视觉表示和文字版动作,输出对未来的预测。规划阶段在潜在空间跑 MPC:候选动作块写成训练时的文本格式,由预测器滚动展开,再对照编码后的目标打分,全程不解码任何像素。\n\n## 成本账:对视频生成路线的六比一\n\n论文给出一组直接对照:与匹配训练的 RAEv2 视频生成世界模型相比,LaLa-World 以约六分之一的训练成本达到相当的规划性能,规划速度快 13.9 倍;平均 MPC 规划性能也高于按域、按任务分别训练的潜在世界模型,加入图像视频数据后,性能与对动作措辞的鲁棒性进一步提升。训练配置并不夸张:20 epochs、每轮 100 万样本,全局 batch 512,8 张 H100 跑约两天。往机器人策略微调出的 LaLa-Policy,动作头采用 Isaac-GR00T 的 flow-matching DiT,按 Cosmos Policy 协议在 LIBERO、RoboCasa 两个仿真基准上与 NVIDIA 的 Cosmos Policy 相当,推理最快快 5.5 倍。\n\n## 冷水:论文先行,权重未放\n\n[GitHub 仓库](https:\u002F\u002Fgithub.com\u002Fjunha1125\u002FLaLa-World)目前只有一个 commit:环境搭建与数据准备脚本已经可用,训练与评估代码标注\"即将更新\",三个视觉编码器版本的 checkpoint 全部\"即将放出\"。也就是说,上述倍数目前只有论文自报这一个发布方,第三方复现暂时无从做起。另外,Qwen3.5-0.8B 当预测器虽然便宜,但 0.8B 的容量上限能否撑住更复杂的真实机器人场景是个问号——LIBERO 和 RoboCasa 都是仿真基准,论文没有真实硬件结果。\n\n这篇论文真正值得记住的不是倍数,而是\"自然语言当异构数据胶水\"这个工程判断:每家机器人实验室的动作空间都不一样,文本反而成了唯一通用的接口。世界模型\"生成式对潜在表示\"的路线之争远未结束,但 LaLa-World 至少把成本牌拍在了桌上。权重放出之前,先别急着站队。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.27851","b0060acd-5f64-427d-9673-5caea2940874",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"0ef8513a-0a26-42f0-b6f9-5b6dadded45c","efficiency",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":19,"name":20,"slug":20,"description":14,"color":14},"499f4b56-819d-49a3-9609-33e775143b86","multimodal",{"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},"13cd8865-c4c8-4d03-a206-99ef8c143138","en","LaLa-World: language as action interface, 1\u002F6 training cost","NAVER's LaLa-World writes actions as text to unify training data: 1\u002F6 training cost, 13.9x faster planning; policy up to 5.5x faster than Cosmos Policy.","The dominant recipe for world models over the past two years has been video generation: train a generative model on massive video corpora, then use it for planning and policy learning. The cost is well known — generating pixels is expensive at train time and slow at inference, because every rollout step pays for full pixel-space computation. LaLa-World, released this week by NAVER AI Lab together with Georgia Tech, Brown and Mila, takes the opposite direction: never generate pixels, only predict latent representations — and it writes the hardest-to-unify dimension, actions, in natural language.\n\n## Core design: actions are text\n\nLaLa-World is a language-conditioned latent world model, paper no. [arXiv:2610.27851](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.27851). Two design choices let it fold heterogeneous data into one model. First, numerical actions and visual changes are both described in language: robot action deltas are rendered into text following normalization specs, while image and video data contribute change captions and edit instructions. A single future predictor can therefore consume simulation trajectories (Wall, PointMaze, Push-T, Meta-World and more), the DROID robot dataset, OpenVid-1M frame sequences and OmniEdit pairs. Second, the predictor is built on a language model: a 24-layer Qwen3.5-0.8B paired with a frozen vision encoder (DINOv2, DINOv3 or SigLIP2), taking the current visual representation plus a text-form action and predicting the future representation. Planning runs MPC in latent space: candidate action chunks are written in the training text format, rolled out by the predictor, and scored against the encoded goal — no pixels decoded.\n\n## The cost account: six to one against video generation\n\nThe paper reports a matched comparison: against an RAEv2-based video-generation world model, LaLa-World reaches comparable planning performance with about one-sixth the training cost and 13.9x faster planning. Its average MPC planning performance also exceeds latent world models trained separately per domain or task, and adding image and video data further improves performance and robustness to action phrasing. The training setup is modest: 20 epochs over 1M samples, global batch 512, about two days on 8 H100 GPUs. Fine-tuned into LaLa-Policy — with a flow-matching DiT action head from Isaac-GR00T — it matches NVIDIA's Cosmos Policy on LIBERO and RoboCasa under the Cosmos Policy protocol, with up to 5.5x faster inference.\n\n## Cold water: paper first, weights pending\n\nThe [GitHub repository](https:\u002F\u002Fgithub.com\u002Fjunha1125\u002FLaLa-World) currently holds a single commit: environment setup and data preparation scripts are available, but training and evaluation code is marked \"updating soon\", and all three checkpoint variants are \"coming soon\". Every multiplier above currently rests on the paper as the sole publisher; independent reproduction is not yet possible. And while a Qwen3.5-0.8B predictor is cheap, whether 0.8B capacity holds up on harder real-robot settings is an open question — LIBERO and RoboCasa are both simulation benchmarks, and the paper reports no real-hardware results.\n\nThe point worth remembering is not the multipliers but the engineering judgment that natural language is the glue for heterogeneous data: every robotics lab has a different action space, and text is the one interface they all share. The generative-versus-latent world model debate is far from over, but LaLa-World has put the cost card on the table. Before the weights land, hold off on picking a side.","lala-world-language-action-latent-world-model","2026-10-25T15:13:48Z","2026-10-25T15:13:51.696620Z","2026-10-25T15:13:51.696631Z",true,"agent",492,[39,49,58],{"slug":40,"tag_slug":41,"title_zh":42,"title_en":43,"intro_zh":44,"intro_en":45,"id":46,"is_active":35,"created_at":47,"modified_at":48},"image-generation","text-to-image","图像生成：从文生图到逐框编辑的模型竞赛","Image Generation: from text-to-image to frame-level editing","文生图已是红海，战场正转向编辑与可控性：FLUX.2\u002F3 用开源和量化把生图搬进消费级显卡，Qwen-Image、Seedream 等国产模型紧咬不放，GPT Image 2 则把推理能力带进像素。本专题追踪图像生成与编辑模型的发布、架构演进、加速量化与商业化数据，看「好图」如何变成「可用的图」。","Text-to-image is a crowded race, and the frontier has moved to editing and control: FLUX.2\u002F3 push open weights and quantization onto consumer GPUs, Qwen-Image and Seedream close in from China, and GPT Image 2 brings reasoning into pixels. This topic tracks model releases, architecture shifts, acceleration work, and the business numbers behind \"pretty\" becoming \"usable\".","52de7c59-7864-4569-8ae7-3d656d26a1a1","2026-11-20T07:20:23.534506Z","2026-11-20T07:20:23.534527Z",{"slug":50,"tag_slug":50,"title_zh":51,"title_en":52,"intro_zh":53,"intro_en":54,"id":55,"is_active":35,"created_at":56,"modified_at":57},"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":59,"tag_slug":59,"title_zh":60,"title_en":61,"intro_zh":62,"intro_en":63,"id":64,"is_active":35,"created_at":65,"modified_at":66},"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":68},[69,74,79,84,89,94],{"id":70,"title":71,"news_slug":72,"published_at":73},"6186bff0-6bbb-4970-a0bc-b7bfa93a8e49","像素空间压缩:16个token装下RAG检索到的文档","opticalrag-pixel-space-rag-compression","2026-11-13T03:20:00+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"77a567e5-2209-4981-882b-e2a3077a0012","先猜再听:全双工语音模型把投机解码搬到音频到来之前","speculative-listening-full-duplex-speech-decoding","2026-11-06T09:05:00+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"891e0c6f-0c26-4ca6-bc85-3fca60bdff72","代码写视频:15 个多模态大模型实测,跑通率比画质更决定成败","code2video-compass-mllm-code-video-benchmark","2026-10-25T09:15:00+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"eeb3f139-8df8-40d7-a1e6-82e49b668547","VideoJev:视频问答提速100倍,不生成只打分","videojev-system-one-video-understanding","2026-10-14T15:09:17+00:00",{"id":90,"title":91,"news_slug":92,"published_at":93},"6aca231d-200d-4227-9fa0-f1e6d149b0b0","WanPE:397B 提示词模型上岗,视频生成多了个导演","wanpe-397b-video-prompt-enhancement","2026-09-26T23:07:22+00:00",{"id":95,"title":96,"news_slug":97,"published_at":98},"33bd7c4c-27f8-458c-8404-265134fc6ce8","视频生成缺的不是算力,是记忆:282 篇论文拼出一张全景地图","ar-video-generation-memory-survey","2026-09-24T21:09:28+00:00"]