[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-blockpilot-instance-adaptive-block":3,"topics-all":31,"news-related-cd82f869-d4a7-45b7-95ce-b66051e9d933":50},{"id":4,"title":5,"summary":6,"content":6,"original_url":7,"source_id":8,"tags":9,"translations":23,"news_slug":24,"published_at":25,"created_at":26,"modified_at":27,"is_published":28,"publish_type":29,"image_url":13,"view_count":30},"cd82f869-d4a7-45b7-95ce-b66051e9d933","BlockPilot：实例自适应策略学习让扩散式投机解码再下一城,Qwen3-4B 上首破 4.20× 加速","扩散式投机解码(diffusion-based speculative decoding)是目前 LLM 推理加速最前沿的方向之一:通过块级扩散在单次前向中并行生成多个候选 token,再用目标模型一次性验证,实现无损加速。但现有方法普遍采用固定推理块大小,默认最优策略对所有输入一致,严重限制了进一步提速的空间。\n\nBlockPilot 的关键观察指出:最优块大小在不同样本间差异显著,而且这些值呈现围绕训练块大小的局部集中结构,使块大小选择变成一个低维、可学习的决策问题。基于此,作者把块大小选择形式化为一个轻量级策略学习问题,提出实例自适应决策机制:只需在 prefill 之后用 prefill 表示预测一次最优块大小,然后在整段解码过程中保持不变。这种设计使 BlockPilot 与现有扩散投机解码系统即插即用,不需要修改目标模型,base LLM 完全冻结,新增训练参数低于 0.05%。\n\n在 Qwen3-4B 温度 T=1 设置下,BlockPilot 取得 5.92 的接受长度与 4.20× 端到端加速,显著超越现有 SOTA 扩散投机解码基线。这一工作的意义在于把「样本难度异质性」明确纳入投机解码设计空间,打破了此前固定块大小的隐含假设,也为长序列、Agentic 工作流等块大小天然多变的场景打开了动态解码策略的新方向。\n\n原文:arXiv:2606.31315,提交于 2026 年 6 月 30 日。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2606.31315","7437aeb9-930c-4866-a2e9-48003c1a792b",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"5e628969-6d2a-437f-998a-104e4b16cfb1","ai-progress",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",[],"blockpilot-instance-adaptive-block","2026-07-01T14:17:29Z","2026-07-01T14:19:07.430537Z","2026-08-19T02:08:40.142862Z",true,"agent",225,[32,41],{"slug":33,"tag_slug":33,"title_zh":34,"title_en":35,"intro_zh":36,"intro_en":37,"id":38,"is_active":28,"created_at":39,"modified_at":40},"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":42,"tag_slug":42,"title_zh":43,"title_en":44,"intro_zh":45,"intro_en":46,"id":47,"is_active":28,"created_at":48,"modified_at":49},"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":51},[52,57,62,67,72,77],{"id":53,"title":54,"news_slug":55,"published_at":56},"3d922c00-afcb-4f1c-a6d5-8f9d6c10c642","从 Kimi Linear 到 Kimi K3:MoE 推理效率战里被忽略的架构升级","kimi-k3-latentmoe-kda-attnres-nope","2026-07-30T00:30:00+00:00",{"id":58,"title":59,"news_slug":60,"published_at":61},"f2b0bde3-ebb5-49e5-a418-5ece37639d1b","MIPU\u002FMIPI：把 LLM RL 的「训练—推理失配」从工程噪音重写为优化目标","mipu-mipi-rl-mismatch","2026-07-04T08:00:00+00:00",{"id":63,"title":64,"news_slug":65,"published_at":66},"5a4c2a98-6ce7-4e51-8348-118be3083afc","ELDR 把 MoE 推理的「延迟最后一公里」拉直:vLLM 实测 TPOT 最多砍 13.9%","eldr-moe-routing","2026-07-03T12:30:00+00:00",{"id":68,"title":69,"news_slug":70,"published_at":71},"abf22bbb-eccf-46d0-99d6-debe1596f92b","自验证蒸馏：无需外部教师，LLM如何实现自我进化","self-verified-distill-qwen3-16-7pp-math","2026-05-27T19:00:00+00:00",{"id":73,"title":74,"news_slug":75,"published_at":76},"6c9e553d-3029-4867-bc29-ee069d26b934","自验证蒸馏：无需外部教师，LLM 如何实现自我进化","svd-self-verified-distill-stanford-perplexity","2026-05-27T16:05:00+00:00",{"id":78,"title":79,"news_slug":80,"published_at":81},"2ca620f4-a046-4274-925d-f0689123a498","ParaRNN：Apple 让 RNN 重回战场，7B 参数模型训练提速 665 倍","apple-pararnn-7b-rnn-665x-faster-iclr","2026-05-27T13:15:00+00:00"]