[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-tcsexam-lean-verified-tcs-benchmark":3,"topics-all":39,"news-related-ca1e87fb-2e21-4af2-b958-046a7efbefb0":57},{"id":4,"title":5,"summary":6,"content":7,"original_url":8,"source_id":9,"tags":10,"translations":25,"news_slug":32,"published_at":33,"created_at":34,"modified_at":35,"is_published":36,"publish_type":37,"image_url":15,"view_count":38},"ca1e87fb-2e21-4af2-b958-046a7efbefb0","姚期智署名TCSExam:模型考理论CS,跑分跑出新构造","TCSExam给大模型出了张理论CS考卷:100道专家审定任务,横跨Lean定理证明、求解器工程、组合构造与公开猜想,全部由Lean内核或可执行检查器判分。8个agent应考,GPT-6 Astra 71.9分领跑Full榜;Kimi K3与Astra交出630边图构造与距离6量子纠错码,跑分跑出可验证的研究产出。","跑基准通常只产出一个排行榜。11月5日出现在arXiv的TCSExam（[2611.07962](https:\u002F\u002Farxiv.org\u002Fabs\u002F2611.07962)）不太一样：这张考卷除了给分，还真的跑出了两件经形式验证的研究产出——一个630条边的图构造，和一个更强的量子纠错码。论文52页，作者名单里有图灵奖得主姚期智（Andrew Chi-Chih Yao），代码开在MoonshotAI组织下。\n\n## 一张机器判卷的理论CS考卷\n\nTCSExam-Full收录100道专家审定任务，分四条赛道：定理证明20题，要求给出已发表结果的完整Lean证明；求解器工程30题，要求实现算法并在隐藏实例上评测；构造20题，要求交出性质经形式验证的对象或界；公开猜想30题，接受完整的Lean证明或反驳。判卷不靠人工裁量：提交由Lean内核（含公理检查）或可执行任务检查器判定。另有20题的Lite版，每条赛道抽5题，在同等单题预算下把尝试次数砍掉80%；Full与Lite共有的八个agent排名高度一致（Spearman ρ=0.93），支撑Lite当一张紧凑但有区分度的卷子用。\n\n综合分TCSExam Index按30\u002F30\u002F30\u002F10加权四条赛道，构造赛道得分可以超过1，所以指数以点计、允许超过100。评测框架默认用Kimi Code当agent，也支持Claude Code、Codex和Grok Build，由Harbor 0.22.0负责跑评测、收产物。\n\n## 成绩单与两个意外收获\n\n榜单本身：在TCSExam上，GPT-6 Astra以71.9分领跑Full，Opus 5.5以72.3领先Lite。更有意思的是\"考出来的东西\"：Kimi K3在124个顶点、禁止三角形和四边环的约束下构造出630条边的图，比Goedgebeur等人2025年给出的629边构造又多了一条边——这是极值图论下界的一次具体推进；GPT-6 Astra则构造出[[25,7,6]]量子纠错码，同样用25个物理比特保护7个逻辑比特，把检错距离从Grassl码表参考码的5提到6，同样规模下能多检出一类错误。项目站对这些成果的措辞很克制：\"据我们所知，改进了此前已知的最佳结果。\"论文还报告了公开猜想尝试中若干经独立验证的特殊情形与界。\n\n## 为什么值得在意\n\n\"AI能不能做理论研究\"这场争论，缺的从来不是观点，是可核对的证据。TCSExam把判卷权交给Lean内核：证明要么过内核，要么不过，没有评委滤镜。求解器赛道的题目源自PACE、DIMACS、ROADEF\u002FEURO等真实研究社区的基准，仓库配套Docker构建配方与评测工具，代码Apache-2.0、题面CC BY 4.0，整体可复现（[GitHub](https:\u002F\u002Fgithub.com\u002FMoonshotAI\u002FTCSExam)）。当\"做出了多少研究\"第一次变成逐题可验证的事实，下一个问题才是：跑分之外，这些模型还能攒出多少真定理。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2611.07962","7437aeb9-930c-4866-a2e9-48003c1a792b",[11,16,19,22],{"id":12,"name":13,"slug":13,"description":14,"color":15},"9112951a-2abb-4214-b63a-385ec7afb2ba","ai-for-science","AI for Science 专题：追踪 AI 在生命科学、化学材料、物理世界模型等科学方向的关键突破",null,{"id":17,"name":18,"slug":18,"description":15,"color":15},"120fa59a-ff6f-4537-9bf5-f818df636a0e","benchmark",{"id":20,"name":21,"slug":21,"description":15,"color":15},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":23,"name":24,"slug":24,"description":15,"color":15},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source",[26],{"id":27,"lang":28,"title":29,"summary":30,"content":31},"f3bb0194-5132-4702-a3e1-fdd2f8492e93","en","TCSExam: LLMs Take a 100-Task TCS Exam, Graded by Lean","TCSExam grades agents on 100 expert-reviewed, Lean-verified TCS tasks. GPT-6 Astra leads Full at 71.9; Kimi K3 produced a verified 630-edge graph.","Benchmarks usually output nothing but a leaderboard. TCSExam, posted to arXiv on November 5 ([2611.07962](https:\u002F\u002Farxiv.org\u002Fabs\u002F2611.07962)), is different: this exam produced two formally verified research artifacts on top of scores — a 630-edge graph construction and a stronger quantum error-correcting code. The 52-page paper counts Turing Award winner Andrew Chi-Chih Yao among its authors, and the code lives under the MoonshotAI organization.\n\n## A machine-graded exam for TCS research\n\nTCSExam-Full contains 100 expert-reviewed tasks across four tracks: 20 theorem-proving tasks require a complete Lean proof of a published result; 30 solver-engineering tasks require an algorithm implementation judged on hidden instances; 20 construction tasks require an object or bound with formally verified properties; 30 open-conjecture tasks accept a complete Lean proof or refutation. Grading needs no human discretion: submissions are checked by the Lean kernel (with axiom checks) or an executable task checker. A 20-task Lite version picks five tasks per track and cuts the number of attempts by 80% at the same per-attempt budget; rankings align strongly across the eight agents shared by both suites (Spearman ρ=0.93), supporting Lite as a compact yet discriminative benchmark.\n\nThe composite TCSExam Index weights the four tracks at 30\u002F30\u002F30\u002F10; construction scores can exceed one, so the Index is measured in points and may exceed 100. The harness runs Kimi Code as the default agent, also supports Claude Code, Codex and Grok Build, and uses Harbor 0.22.0 to run evaluations and collect outputs.\n\n## The scoreboard and two surprise results\n\nOn the TCSExam leaderboard, GPT-6 Astra leads Full at 71.9 and Opus 5.5 leads Lite at 72.3. More interesting is what the runs produced: Kimi K3 constructed a 630-edge graph on 124 vertices with no triangles or four-edge cycles, one edge beyond the 629-edge construction from Goedgebeur et al. (2025) — a concrete advance on a lower bound in extremal graph theory. GPT-6 Astra produced a [[25,7,6]] quantum error-correcting code: the same 25 physical qubits protecting 7 logical qubits, with detection distance raised from 5 (the Grassl table reference code) to 6. The project site hedges the claims carefully: \"these constructions improve on the best previously known results, to the best of our knowledge.\" The paper also reports independently verified special cases and bounds from open-conjecture attempts.\n\n## Why it matters\n\nThe debate over whether AI can do real research has never lacked opinions — it lacked checkable evidence. TCSExam hands grading to the Lean kernel: a proof either passes the kernel or it does not, with no reviewer filter. Solver tasks build on benchmarks from real research communities (PACE, DIMACS, ROADEF\u002FEURO), and the repository ships Docker build recipes and evaluation tooling; code is Apache-2.0 and task statements CC BY 4.0, so the whole thing is reproducible ([GitHub](https:\u002F\u002Fgithub.com\u002FMoonshotAI\u002FTCSExam)). Once \"how much research got done\" becomes a task-by-task verifiable fact, the next question is how many real theorems these models can stack up beyond the score.","tcsexam-lean-verified-tcs-benchmark","2026-11-06T19:10:36Z","2026-11-06T19:10:37.053451Z","2026-11-06T19:10:37.053464Z",true,"agent",111,[40,48],{"slug":13,"tag_slug":13,"title_zh":41,"title_en":42,"intro_zh":43,"intro_en":44,"id":45,"is_active":36,"created_at":46,"modified_at":47},"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":36,"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},"ec2c558c-502d-43a5-9494-c766dfd515e9","EurekAgent：把科学发现的瓶颈从「工作流」拽到「环境」，11 美元跑出 26 圆 packing 新 SOTA","eurekagent-environment-engineering-11-usd","2026-06-11T17:56:35+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"158f2181-ecf3-4e5c-ac5a-16e5c45c9d9a","千篇论文建成AI实验室:前沿模型74%胜过人类作者","scientist-world-1000-research-environments","2026-11-08T21:08:04+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"4193fdcb-58a9-484e-9bd7-e4071c6588bd","Ripple-65B:65B稠密模型80.52分压过975B","ricursive-ripple-65b-dense-open-weight","2026-11-05T05:30:00+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"90eaefa3-cbb1-440a-9c2f-97c594d0bf78","CUA-Master:教学视频喂出OSWorld 80.2","cua-master-open-world-flywheel","2026-11-03T21:08:37+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"2f9ad534-b737-490e-b97e-bf01f2b6d26b","交错生成新基准:闭源旗舰文本赢,图像一致性输给开源","intreal-interleaved-generation-benchmark","2026-11-01T05:27:08+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"6f2c5b92-a390-4b36-9eff-534daf04f4aa","WorkFactory:先修任务再微调,879涨到1565","workfactory-workpilot-agent-training","2026-10-31T23:10:20+00:00"]