[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-confidence-based-llm-routing-multilingual":3,"topics-all":38,"news-related-98bf8c61-e559-4f96-9c3a-e7049b493f6c":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},"98bf8c61-e559-4f96-9c3a-e7049b493f6c","小模型自信≠正确:4.2万次实测拆解LLM路由","arXiv新研究对「小模型先答、拿不准再升级大模型」的路由范式做多语言体检:3个模型家族、4种语言、72组配置、4.2万次预测。置信度路由在阅读理解与分类任务有效,但在知识推理上AUC仅0.510接近随机;语言间升级率差距最大30个百分点,等额配额几乎零代价抹平差距。","小模型先答,拿不准再升级到大模型——这套「小到大路由」(small-to-large routing)是推理降本的标配思路。arXiv 10 月 16 日提交的一篇论文把它放进多语言场景做了次系统体检:3 个模型家族、4 种语言、3 类任务、72 组配置、42,288 次预测,结论是置信度路由强烈依赖任务类型,在知识推理类任务上接近失效。([arXiv 2610.23249](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.23249))\n\n## 实验怎么做的\n\n研究团队(来自 Bangladesh University、Georgia Southern University 等)选了 Qwen3、Gemma 3、Llama 3.2 三个指令模型家族,每族配一对 1B–1.7B 小模型与 3B–4B 大模型,只在家族内部做「小升大」;语言覆盖英语、孟加拉语、印地语、乌尔都语;任务用 Belebele(阅读理解)、MMLU-ProX-Lite(知识推理)、SIB-200(主题分类)。置信度信号取候选答案前两名的概率差,对比四种路由策略:全局置信度、随机路由、等额配额、oracle 上界。\n\n## 两个硬结论\n\n**一,任务类型决定路由生死。** 阅读理解和主题分类上,置信度能有效识别值得升级的样本(平均 AUC 0.612 \u002F 0.705),50% 升级率下比随机路由多拿 1.40–4.67 和 2.54–5.63 个百分点;但在 MMLU-ProX-Lite 上平均 AUC 只有 0.510,接近随机——原因写得很直白:小模型在最高置信度区间仍大量答错(Qwen3 错 63.45%,Gemma 3 错 76.89%,Llama 3.2 错 86.17%),且这些高置信错误只有 15.59–21.89% 能被同族大模型纠正;两个模型在同一道推理题上一起答错的比率高达 63.79–67.33%。\n\n**二,「谁为效率买单」没有固定答案。** 全局置信度路由在 50% 预算下,不同语言拿到的升级率差距最大达 30.33 个百分点(Qwen3 × Belebele);英语在 9 组设置里有 5 组是被升级最少的语言(平均升级率 43.98%),换成等额配额把它拉回 50%,英语的路由准确率反而涨了 1.35 个点——消除语言分配差距,几乎不用付聚合精度代价。\n\n## 所以呢\n\n工程上有两个直接提醒:把置信度当任务无关的路由信号会翻车,上线前先按任务类型做 AUC 体检;多语言产品里加一层等额配额约束,是几乎免费的公平性保险。效率收益也是真的:25% 升级率下,墙钟时间降 27.5%–58.2%、能耗降 26.3%–62.6%。当然,实验只覆盖 1B–4B 开源小模型和选择题任务,更大模型、MoE、生成式场景还有待验证——代码已开在 GitHub。([GitHub](https:\u002F\u002Fgithub.com\u002Fsabbir5622r\u002Fmultilingual-llm-routing))","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.23249","7437aeb9-930c-4866-a2e9-48003c1a792b",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"120fa59a-ff6f-4537-9bf5-f818df636a0e","benchmark",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"0ef8513a-0a26-42f0-b6f9-5b6dadded45c","efficiency",{"id":19,"name":20,"slug":20,"description":14,"color":14},"0a93ec8e-ea39-4693-81de-563ca8c173f7","inference",{"id":22,"name":23,"slug":23,"description":14,"color":14},"01598627-1ea6-4b27-a5d8-874971571a71","llm",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"b3580e1e-015c-4864-b7f8-660b8401fe2f","en","Small-to-large LLM routing: confidence fails on reasoning","72-config study of small-to-large LLM routing: confidence aids reading\u002Fclassification, nears chance on reasoning; language gaps hit 30 points.","Let the small model answer first, escalate to the large model only when it wavers — small-to-large routing is a standard recipe for cutting inference cost. A paper submitted to arXiv on October 16 put the idea through a systematic multilingual checkup: 3 model families, 4 languages, 3 task types, 72 configurations, and 42,288 predictions. The verdict: confidence-based routing is strongly task-dependent, and it nearly fails on knowledge-and-reasoning tasks. ([arXiv 2610.23249](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.23249))\n\n## How the experiment was built\n\nThe team (Bangladesh University, Georgia Southern University, and collaborators) picked three instruction-tuned families — Qwen3, Gemma 3, and Llama 3.2 — each with a 1B–1.7B small model paired with a 3B–4B large model, escalating only within the same family. Languages cover English, Bangla, Hindi, and Urdu; tasks are Belebele (reading comprehension), MMLU-ProX-Lite (knowledge and reasoning), and SIB-200 (topic classification). The confidence signal is the probability margin between the top two candidate answers, and four routing policies are compared: global confidence, random routing, equal-quota, and an oracle upper bound.\n\n## Two hard findings\n\n**First, task type decides whether routing lives or dies.** On reading comprehension and topic classification, confidence identifies escalations worth making (mean correction-detection AUC of 0.612 and 0.705); at a 50% escalation budget it beats random routing by 1.40–4.67 and 2.54–5.63 percentage points. But on MMLU-ProX-Lite the mean AUC is just 0.510, essentially chance. The stated reason is blunt: in the highest-confidence quintile, small models still get reasoning questions wrong at high rates (Qwen3 63.45%, Gemma 3 76.89%, Llama 3.2 86.17% errors), only 15.59–21.89% of those confident errors get corrected by the same-family large model, and both models fail together on 63.79–67.33% of examples.\n\n**Second, \"who pays for efficient inference\" has no fixed answer.** Under global confidence routing at a nominal 50% budget, the gap between the most- and least-escalated language reaches 30.33 percentage points (Qwen3 on Belebele). English is the least-escalated language in 5 of 9 settings (mean escalation 43.98%); switching to equal-quota routing pulls it back to 50% and actually raises English routed accuracy by 1.35 points — removing the language allocation gap costs almost nothing in aggregate accuracy.\n\n## So what\n\nTwo practical takeaways: treating confidence as a task-agnostic routing signal will bite you — run an AUC check per task type before shipping; and in multilingual products, an equal-quota constraint is a nearly free fairness insurance. The efficiency gains are real: at 25% escalation, wall time drops 27.5%–58.2% and energy drops 26.3%–62.6%. Caveats: the study covers only 1B–4B open-weight models and multiple-choice tasks — larger models, MoE, and free-form generation remain untested. Code is on GitHub. ([GitHub](https:\u002F\u002Fgithub.com\u002Fsabbir5622r\u002Fmultilingual-llm-routing))","confidence-based-llm-routing-multilingual","2026-10-24T19:30:00Z","2026-10-24T19:13:39.654532Z","2026-10-24T19:13:39.654551Z",true,"agent",247,[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},"a550dbc5-240b-4747-b625-9aa874aad444","工具拉满的AI停在0.48倍:会提问的做到1.13倍","kernelseek-question-driven-cuda-kernel-agent","2026-11-03T09:05:00+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"932bddfc-f9c2-43c9-9eec-676df0a7bf23","解码长度预测撞上天花板:误差大头是采样噪声,不是模型","prefill-decode-length-prediction-limits","2026-11-01T23:09:55+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"c2e41947-c219-4fbe-83f2-5dd9e4501e73","信号处理审计:推理链置信度评分其实量的是长度","token-confidence-chain-length-audit","2026-10-28T11:12:21+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"13ed0c94-9bea-440c-93c4-771c38ba9834","sink 消失后，KV 驱逐改看 value 几何","valuediff-value-geometric-kv-eviction","2026-09-22T17:05:00+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"7cacddc6-fa02-4de9-84a2-c3320e225571","因果归因剪枝 CAP：让 LLM 推理能力不再随稀疏化而流失","cap-causal-attribution-pruning-arc-61pct","2026-06-20T22:14:08.915874+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"9a1e1c85-60eb-47c6-92b5-bace1746e217","大模型竞争进入下半场：从「比参数」到「比部署」——2026年5月技术格局观察","llm-2nd-half-deploy-vs-params-may-2026","2026-05-25T05:15:00+00:00"]