[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-mit-adaptive-draft-speculative-train-2-3x":3,"topics-all":36,"news-related-4556940e-6456-43ce-b9b4-a0a7fa7a5865":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},"4556940e-6456-43ce-b9b4-a0a7fa7a5865","MIT 新方法：自适应草稿模型将推理 LLM 训练速度提升 2-3 倍","推理大模型（Reasoning LLM）因能自我纠错、多步推理，在复杂编程和规划任务上表现优异，但训练成本极高。MIT 与 NVIDIA、ETH Zurich 等机构的研究人员发现，训练过程中高达 85% 的时间被「rollout」阶段消耗——模型生成多个候选答案以供评分，这个阶段只有少部分 GPU 在工作，其余大量算力处于空闲状态。\n\n**核心突破：自适应草稿模型**\n\n研究团队设计了一套自适应系统，在检测到部分 GPU 空闲时，自动启用一个更小更快的「草稿模型」（Drafter），由它预测大模型的输出，再由大模型验证。草稿模型只在需要时介入，不产生额外开销。\n\n关键在于「自适应」——草稿模型根据实时负载动态调度，而非固定比例分配，这解决了传统投机解码（Speculative Decoding）中草稿命中率不稳的问题。实验表明，该方法在保持精度的前提下，将训练速度提升 2-3 倍。\n\n**为何重要**\n\n训练一个前沿推理模型的电费已达数百万美元量级，这种「lossless」优化意味着不需要在精度和效率之间妥协。更重要的是，作为副产物训练出的轻量草稿模型本身也可直接用于推理加速，一石二鸟。\n\n**行业启示**\n\n随着 LLM 能力竞争逐渐触及算力瓶颈，训练效率优化正从边缘走向舞台中央。FlashAttention 解决了注意力计算的内存问题，MTP（Majorization-Threatens-Posterior）改进了投机解码，而 MIT 这项工作的价值在于证明了「GPU 空闲」这座冰山水下还有可观的优化空间。2026 年，LLM 的竞争或许不只在模型本身，还在谁能把算力压榨得更彻底。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.16665","4613a0c2-8d14-4485-b855-f8fad33c4527",[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},"0ef8513a-0a26-42f0-b6f9-5b6dadded45c","efficiency",{"id":18,"name":19,"slug":19,"description":13,"color":13},"0a93ec8e-ea39-4693-81de-563ca8c173f7","inference",{"id":21,"name":22,"slug":22,"description":13,"color":13},"01598627-1ea6-4b27-a5d8-874971571a71","llm",[24],{"id":25,"lang":26,"title":27,"summary":28,"content":13},"b4c9d482-811c-4739-9590-0867aece865b","en","MIT: adaptive draft models train reasoning LLMs 2-3x faster","Reasoning LLMs (RLLMs), with their ability to self-correct and perform multi-step reasoning, perform excellently on complex programming and planning tasks, but their training cost is extremely high. Researchers from MIT, NVIDIA, ETH Zurich, and others found that up to 85% of training time is consumed by the \"rollout\" stage — the model generates multiple candidate answers for scoring, where only a small fraction of GPUs are working, while the rest sit idle.\n\n**Core breakthrough: adaptive draft model**\n\nThe team designed an adaptive system that, upon detecting some GPUs sitting idle, automatically enables a smaller, faster \"Drafter\" model to predict the large model's output, which the large model then verifies. The draft model intervenes only when needed, with no extra overhead.\n\nThe key is \"adaptive\" — the draft model is dynamically scheduled based on real-time load, not fixed-ratio allocation, which solves the unstable draft-hit-rate problem of traditional speculative decoding. Experiments show that, while preserving accuracy, this method boosts training speed 2-3×.\n\n**Why it matters**\n\nTraining a frontier reasoning model already costs millions of dollars in electricity. This kind of \"lossless\" optimization means there's no need to compromise between accuracy and efficiency. Even more importantly, the lightweight draft model trained as a byproduct can itself be directly used for inference acceleration, killing two birds with one stone.\n\n**Industry implications**\n\nAs LLM capability competition gradually hits the compute ceiling, training-efficiency optimization is moving from the margins to center stage. FlashAttention solved attention-compute memory issues, MTP (Majorization-Threatens-Posterior) improved speculative decoding, and MIT's work here proves that beneath the \"GPU idle\" tip of the iceberg there's still substantial optimization space. In 2026, LLM competition may not just be about the model itself, but about who can squeeze the most out of compute.","mit-adaptive-draft-speculative-train-2-3x","2026-05-15T02:05:00Z","2026-05-15T10:05:15.339884Z","2026-08-19T02:08:40.142862Z",true,"agent",242,[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},"748a4486-34e7-4215-b515-7eb56b3258c5","TwELL：Sakana AI与NVIDIA联合提出稀疏LLM推理加速20%，解决GPU批处理落地难题","sakana-nvidia-twell-20pct-sparse-batch-gemm","2026-05-30T08:20:00+00:00",{"id":63,"title":64,"news_slug":65,"published_at":66},"2fa66657-afbb-4f03-849a-f420f42cf2ab","Prompt Caching：LLM推理成本削减90%的隐藏利器","prompt-caching-90pct-token-cost","2026-05-26T01:10:00+00:00",{"id":68,"title":69,"news_slug":70,"published_at":71},"73f4d31e-745a-4bba-8a0f-38e7564966de","Sakana AI 提出 99% 稀疏性Transformer：在前馈层动刀革新LLM效率","sakana-99pct-sparse-ffn-transformer","2026-05-16T19:04:00+00:00",{"id":73,"title":74,"news_slug":75,"published_at":76},"137ce22e-389d-47fb-8219-42ca53d6e916","Qwen 3.6 27B 重磅更新：MTP 技术让本地推理提速 2.5 倍","qwen-3-6-27b-mtp-local-2-5x","2026-05-16T01:01:00+00:00",{"id":78,"title":79,"news_slug":80,"published_at":81},"e4fd45e9-e0fd-4839-973e-909a442ce5ff","DeepSeek V3.2稀疏注意力：如何将长上下文推理成本砍半","deepseek-v3-2-dsa-sparse-attention-50pct-cost-cut","2026-05-01T10:15:00+00:00",{"id":83,"title":84,"news_slug":85,"published_at":86},"810251ee-b8bf-4fef-a9d8-e167c22ae4c5","BoostLoRA：梯度增强让低秩适配器「自我进化」，小参数也能有大表达","boostlora-gradient-boosting-lora-residual","2026-05-01T05:10:00+00:00"]