[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-compute-aligned-training-pass-n-vote-michigan":3,"topics-all":36,"news-related-86c3465b-27f0-4f8c-83d1-9a8e1fa171c8":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},"86c3465b-27f0-4f8c-83d1-9a8e1fa171c8","Compute Aligned Training：让模型学会协同推理的新训练范式","大模型推理时，用多次采样 + 投票的策略已经非常普遍——Best-of-N、Majority Vote、Pass@N，这些名字对关注 LLM 进展的人来说都不陌生。但一个长期被忽视的问题是：训练目标与推理策略之间的根本性错位。标准微调（SFT）或强化学习（RL）都在优化单次采样的正确率，却从未考虑模型最终会以“集成”方式被使用。最近一篇来自密歇根大学的论文提出了 Compute Aligned Training（CAT），尝试从训练阶段就解决这个问题。\n\n问题在哪？\n\n以 Pass@100 为例：假设模型已有 50% 的概率答对，继续把答对概率从 50% 提到 90%，对 Pass@100 几乎没有额外增益——成功早已是定局。但标准训练会继续把梯度压向这个方向，造成“有效过训练”。反过来，Majority Vote 是一个竞争过程，p=0.35 的答案只要比最强对手 p=0.1 更高就能获胜，而标准训练却会推着模型把概率推向 1.0，追求“压倒性胜利”。这是两种完全不同的优化逻辑，但今天的模型训练对它们视而不见。\n\nCAT 怎么做？\n\n核心思路很优雅：把推理时的策略（如 Best-of-N、Majority Vote）形式化为作用在基础策略上的算子，再以这个算子作用后的分布作为训练目标。具体而言，CAT 为 SFT 和 RL 分别推导了新的梯度重加权机制，根据单次采样对最终聚合结果的“边际贡献”来分配梯度权重，而不是简单地对答案概率本身做梯度下降。\n\n论文在三类场景验证了效果：LLM 的 SFT 和 RL 场景下 CAT 显著提升了 Pass@N 和 Majority Vote 的表现；超越 LLM，CAT 在蛋白质语言模型（PLM）上同样有效；在策略泛化性上，CAT 框架可以兼容任意推理策略，不局限于某一种。\n\n这意味着什么？\n\nTest-time scaling（推理时缩放）已经是提升模型能力的重要方向，但之前的努力主要在推理端——更长的 CoT、更多的采样、更复杂的验证链。CAT 的价值在于指出：如果训练阶段就能让模型“理解”它将以集成方式被使用，推理端的投入将被放大，效率也会更高。\n\n这不是一个复杂的工程技巧，而是一个概念层面的对齐——让模型从“追求每次都答对”变成“追求集成后答对”。随着开源模型越来越多支持本地推理和批量采样，这种训练范式的影响可能会比很多人预期的更早到来。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2604.24957","7437aeb9-930c-4866-a2e9-48003c1a792b",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"7ac06d8e-b074-4147-abfc-ffaa4c6b8744","ai-efficiency",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"0a93ec8e-ea39-4693-81de-563ca8c173f7","inference",{"id":18,"name":19,"slug":19,"description":13,"color":13},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":21,"name":22,"slug":22,"description":13,"color":13},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source",[24],{"id":25,"lang":26,"title":27,"summary":28,"content":13},"4fc4ed68-cd31-41b4-8ead-a3a368b33065","en","Compute Aligned Training: models that reason together","For large-model inference, multi-sampling + voting strategies are already common — Best-of-N, Majority Vote, Pass@N — all familiar to anyone following LLM progress. But one long-overlooked issue is a fundamental misalignment between training objectives and inference strategies. Standard fine-tuning (SFT) and reinforcement learning (RL) both optimize for single-sample accuracy, never considering that the model will ultimately be used in an \"ensemble\" fashion. A recent paper from the University of Michigan proposes Compute Aligned Training (CAT), attempting to address this from the training stage.\n\n**Where's the problem?**\n\nTake Pass@100 as an example. Suppose the model already has a 50% chance of being correct; continuing to push that probability from 50% to 90% yields almost no additional Pass@100 gain — success was already locked in. But standard training continues pushing the gradient in that direction, causing \"effective over-training.\" Conversely, Majority Vote is a competitive process: an answer with p=0.35 wins as long as it's higher than the strongest competitor's p=0.1, but standard training would push the model toward p=1.0, pursuing \"overwhelming victory.\" These are two completely different optimization logics, but today's model training ignores both.\n\n**How does CAT do it?**\n\nThe core idea is elegant: formalize the inference-time strategy (e.g., Best-of-N, Majority Vote) as an operator acting on the base policy, and use the post-operator distribution as the training target. Concretely, CAT derives new gradient reweighting mechanisms for SFT and RL, assigning gradient weights according to each sample's \"marginal contribution\" to the final aggregated result, rather than naïvely doing gradient descent on the answer probability itself.\n\nThe paper validates the effect in three scenarios: in LLM SFT and RL, CAT significantly improves Pass@N and Majority Vote performance; beyond LLMs, CAT also works on protein language models (PLMs); on policy generalization, the CAT framework can be compatible with any inference strategy, not limited to a specific one.\n\n**What does this mean?**\n\nTest-time scaling is already an important direction for boosting model capability, but previous efforts have focused mainly on the inference side — longer CoT, more samples, more complex verification chains. CAT's value lies in pointing out: if the training stage can let the model \"understand\" that it will be used in ensemble form, the inference-side investment will be amplified, and efficiency will be higher.\n\nThis isn't a complex engineering trick, but a conceptual alignment — shifting the model from \"aim to be right every time\" to \"aim to be right after ensembling.\" As more open-source models support local inference and batch sampling, the impact of this training paradigm may arrive earlier than many expect.","compute-aligned-training-pass-n-vote-michigan","2026-05-22T13:10:00Z","2026-05-22T13:04:21.565264Z","2026-08-19T02:08:40.142862Z",true,"agent",186,[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},"1edd86ea-3eb0-4424-9f57-add15c08d891","HF Hub 接入 SkyPilot：Xet 去重让 20+ 云共享模型数据","hf-skypilot-hf-storage","2026-07-12T06:30:00+00:00",{"id":63,"title":64,"news_slug":65,"published_at":66},"f76e6f4b-1ae5-442a-abfa-823e7226ae81","Hugging Face 把 transformers 跑出 vLLM 原生速度：单 flag 让 235B MoE 直接吃到 EP 红利","hf-vllm-transformers-backend","2026-07-09T04:03:00+00:00",{"id":68,"title":69,"news_slug":70,"published_at":71},"4d436945-18e9-4d69-a4c8-c1e3e975ab33","MiniMax M3发布：稀疏注意力打通百万token上下文，开源模型编程能力逼近闭源前沿","minimax-m3-sparse-attn-million-token-msa","2026-06-04T01:00:00+00:00",{"id":73,"title":74,"news_slug":75,"published_at":76},"068bcc7a-d901-4d3d-924c-eefa7ced4467","TokenSpeed 开源推理引擎发布：剑指 Agentic Workloads 的高效推理","tokenspeed-agentic-inference-engine","2026-05-20T19:10:00+00:00",{"id":78,"title":79,"news_slug":80,"published_at":81},"0565190a-0bcd-492f-934f-0ad2ab32f485","70万参数2.8MB填一张表:Cua开源CUA-S1,单次前向替代23轮LLM","cua-s1-forms-system-one-model","2026-09-20T13:11:48+00:00",{"id":83,"title":84,"news_slug":85,"published_at":86},"b571067a-9fa8-42bf-9431-98f26ac78e03","伯克利把LLM推理搬进SSD:KV缓存压缩15倍","llm-inference-in-flash-cim-ssd","2026-09-19T21:10:00+00:00"]