[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-mjev-spatial-classification-spatial-intelligence":3,"topics-all":38,"news-related-405b8f47-8862-4684-ab0e-d552ce093072":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},"405b8f47-8862-4684-ab0e-d552ce093072","MJev-Spatial:空间理解不生成文字,单次前向快5.6倍","MJev-Spatial 把空间问答、数值估计、图像定位三类空间任务统一成分类问题:复用 Qwen3.5-4B 预训练的 Q\u002FK 投影打分,单次前向、零答案 token。官方评测 12 项基准总分 65.64% 居八个受测模型之首,7 项第一,比基座快 5.6 倍;代码与权重已开源。","视觉语言模型回答\"这个物体在图像的哪个位置\"这类空间问题,走的是一条笨路:自回归地生成一段文字,再把它解析回一个选项、一个标量或一个二维坐标。输出空间明明只有几个选项,却要为它付出整套解码与解析的成本。10 月 15 日提交到 arXiv 的 MJev-Spatial 提出了另一条路线:这类问题不该用生成来答,该用分类([arXiv:2610.22276](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.22276))。\n\n## 把空间问题变成选择题\n\n论文把三类空间理解任务——空间关系问答、数值估计、图像定位——统一进一个分类框架:每个候选答案当作一个类别,直接用骨干网络预训练好的 query\u002Fkey 投影打分,不新增任何参数。选择题的候选就是选项;数值估计用\"由粗到细\"的值域分箱;定位则用逐级收窄的图像区域。一作 Luozheng Qin 等 12 位作者,论文 37 页、27 张表。\n\n## 单次前向,零答案 token\n\n推理时每个决策只需一次前向传播,不生成任何答案 token。官方报告的速度:比未改动的自回归基座快 5.6 倍,比最快自回归基线也快 2.5 倍。对机器人和具身智能这类要求\"交互速率\"判断的场景,这个量级的延迟差是质变而非微优化。\n\n## 成绩单与开源情况\n\n底座是 Qwen3.5-4B,训练数据为从既有空间数据集转换来的 12.3M 问答对。官方评测里,模型在 12 项基准上拿到 65.64% 的总分,居八个受测模型之首,其中 7 项第一,12 项全部超过自家未改动的基座。训练侧用 teacher forcing 在一次前向里联合监督所有由粗到细阶段,配合长度感知全局批与跨 GPU 负载均衡,吞吐比原策略提升 30.8%。工程交付完整:代码以 Apache-2.0 开源,权重已挂 Hugging Face(mindlab-research\u002FMJev-Spatial-4B),GitHub 仓库含运行时、训练循环与公开评测器。\n\n## 泼点冷水\n\n65.64% 是官方评测内八个模型的对比,不是公开榜单位次;模型尚无独立复现。更要紧的是边界:分类框架只适用于输出空间紧凑的任务,开放式的空间描述与推理仍然需要生成式模型。\"快思考\"专线修好了,慢思考那条路还在原地。\n\n对做具身智能的人来说,这篇论文的启示很直接:当答案空间本来就可枚举,生成就是浪费。先问一句\"这个问题真的需要生成吗\",比换更大的模型省钱得多。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.22276","7437aeb9-930c-4866-a2e9-48003c1a792b",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",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},"c187600e-804c-4697-b828-1e4330e0eb10","qwen",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"d72fe2ea-fc29-419e-91ae-54b090766c56","en","MJev-Spatial: spatial QA as classification, 5.6x faster","MJev-Spatial recasts three spatial tasks as classification over Qwen3.5-4B: single forward pass, zero answer tokens, 65.64% mean score, code and weights open.","When a vision-language model answers a spatial question — \"where is this object in the image\" — it takes a clumsy path: autoregressively generate a text string, then parse it back into an option, a scalar, or a 2D point. The output space is often just a handful of options, yet the model pays the full cost of decoding and parsing. MJev-Spatial, submitted to arXiv on October 15, argues this class of questions should be answered by classification, not generation ([arXiv:2610.22276](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.22276)).\n\n## Spatial Questions as Multiple Choice\n\nThe paper unifies three types of spatial understanding — spatial relation QA, numerical estimation, and image pointing — into a single classification framework. Each candidate answer is treated as a class, scored directly with the backbone's pretrained query and key projections, adding zero parameters. For choice tasks the candidates are the given options; numerical estimation uses coarse-to-fine bins over the value range; pointing refines image regions step by step. First author Luozheng Qin and 11 co-authors; the paper runs 37 pages with 27 tables.\n\n## One Forward Pass, Zero Answer Tokens\n\nAt inference, each decision takes a single forward pass and generates no answer tokens. The official numbers: 5.6x faster than its unmodified autoregressive base model, and 2.5x faster than the fastest autoregressive baseline. For robotics and embodied agents whose spatial judgments must run at interaction rates, this is a qualitative gap, not a marginal optimization.\n\n## Scores and Release\n\nThe base model is Qwen3.5-4B, trained on 12.3M question-answer pairs converted from existing spatial datasets. In the authors' evaluation, the model achieves the highest overall score (65.64%) among eight tested models across 12 benchmarks, ranks first on seven, and beats its unmodified base on all twelve. Training supervises all coarse-to-fine stages jointly with teacher forcing in one forward pass, and combines length-aware global batching with cross-GPU load balancing, raising training throughput by 30.8%. The delivery is complete: Apache-2.0 code, weights on Hugging Face (mindlab-research\u002FMJev-Spatial-4B), and a GitHub repo with the runtime, training loop, and a public evaluator.\n\n## Caveats\n\nThe 65.64% figure is a comparison within the authors' own eight-model evaluation, not a public leaderboard rank, and there is no independent replication yet. The deeper limit is scope: classification only works where the output space is compact — open-ended spatial description and reasoning still need generative models. The System 1 fast lane is built; the System 2 road remains where it was.\n\nThe takeaway for embodied-AI builders is direct: when the answer space is enumerable, generation is waste. Asking \"does this question really need generation?\" saves more than upgrading to a bigger model.","mjev-spatial-classification-spatial-intelligence","2026-10-18T17:14:25Z","2026-10-18T17:14:36.354933Z","2026-10-18T17:14:36.354950Z",true,"agent",114,[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},"d83c4910-6dce-4171-8a44-9f68c9f841af","Qwen-Audio-3.1-ASR开源:转写自带说话人","qwen-audio-3-1-asr-speaker-diarization","2026-10-22T17:13:29+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"cb64371f-62b6-473d-8150-b576001d3f56","Qwen3.8-27B 开源权重上线:单卡跑得动的 Qwen3.8,还塞了个视觉编码器","qwen3-8-27b-open-weights-release","2026-08-14T19:30:00+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"40095b51-97b0-4fd4-9b1d-f636c970572e","阿里 Qwen 团队发布 Qwen3.8-Max:2.4 万亿参数 MoE 模型首度开放权重","qwen3-8-max-2-4t-moe-open-weights","2026-08-07T02:00:00+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"35e9f682-7952-4f6a-a8e9-7f8bdd47e791","IBM 语音三件套更新:2B 旗舰 ASR 精度开源第二","granite-speech-5-collection-update","2026-10-22T21:08:48+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"bbbda7ff-b1c7-4142-b37f-2b3fd0547e02","Fibo 2.0 开源:生成编辑合一,登顶开源编辑榜","bria-fibo-2-0-unified-editing-generation","2026-10-21T21:08:11+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"]