[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-fastino-gliner2-5-decide-classifier":3,"topics-all":38,"news-related-9103452d-d036-4b0f-bee6-5eda4b7fbf66":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},"9103452d-d036-4b0f-bee6-5eda4b7fbf66","Fastino 340M 分类模型:官方基准胜 4B 生成式","Fastino 发布开放权重分类模型 GLiNER2.5-Decide:340M 参数,意图识别、情感、主题、工单路由用任意标签集一次前向完成,不写提示词、不生成 token。官方 17 域基准平均 60.2%,超过 1B GLiNER2 XL 和 4B 生成式基线,CPU 可跑,内置中文字符级切分。","2026 年 9 月 23 日,Fastino 把 GLiNER 家族的分类分支 GLiNER2.5-Decide 挂上了 Hugging Face:一个 3.4 亿参数的编码器模型,专门干生产环境里最常见的那类脏活——意图识别、情感分析、主题分类、工单路由、审核判级、垃圾邮件过滤。这不是通用对话模型,而是「决策专才」:官方明确说明它不推理、不解释、不回答开放问题。\n\n## 一个模型顶一排分类器\n\n传统做法里,客服意图、紧急度、派发队列是三个独立分类器,或者用提示词驱动一个通用大模型。Decide 把标签集当作调用时传入的参数:同一段文本、同一批候选标签,一次前向出全部结果;单标签任务返回一个字符串,多标签任务返回所有过阈值的标签。模型卡的场景清单覆盖银行转账意图、差旅改签、诊所分诊、评论维度情感、文档类型、工单路由、人工转接判断,直到内容审核与事件严重度评级。\n\n对部署侧最实际的一点:没有提示词模板,没有生成 token。生成式大模型每答一次分类题都要先「想」再「说」,编码器直接输出结构化结果,延迟和成本压在另一个量级。配套库 gliner2 一行 pip 安装,本地推理不需要 GPU,官方定位 CPU 优先、隐私全本地。\n\n## 340M 对 4B:官方基准的说服力与水分\n\n模型卡公布了 fast-decisions 基准:17 个领域、每域 300 条 held-out 样本,所有模型用相同文本和相同候选标签。GLiNER2.5-Decide 平均 60.2%,高于 GLiNER2 XL(10 亿参数,59.6%)、JevK5(57.6%)、SemIf(基于 Qwen3.5-4B 的生成式,56.4%)、GLiFormer large-v1(49.0%)和 Laya Router(46.6%)。340M 编码器排在 4B 生成式基线前面,这个反差就是核心卖点。\n\n但要说满之前得看清:这张表出自 Fastino 自家数据集,对照模型是否调优到位、标签集设计是否对自家架构有利,都没有第三方复现可核对。自报数字按惯例打折看,「领先」只在表内成立。有意思的是这条赛道正在变挤:Convai 的 Laya Router 同样主打路由与安全检查,46.6% 排在这张表末位。\n\n## 边界架构与中文支持\n\nGLiNER2.5 用 boundary 架构:稀疏起点-终点配对而非固定宽度 span 网格,编码窗口内任意长度都能表示。GitHub 仓库(fastino-ai\u002FGLiNER2,Apache 2.0,1.8k star)同时维护 span 与 boundary 两套抽取架构,AutoExtractor 按 architecture 字段自动分发。中文用户有加分项:GLiNER2 支持字符级切分器,中文不降级到英文默认切分。方法学有 EMNLP 2025 论文背书。\n\n## 所以呢\n\n方向很清楚:不是所有 NLP 任务都值得一个大模型。高频、低复杂度、要求延迟的决策,用 340M 专才在 CPU 上跑,比把每次分类发给生成式 API 便宜也更可控。但自建基准永远是厂商自己的尺子——上线前拿自己的标签集跑一遍 held-out,比相信任何 60.2% 都重要。许可证是「other」非标准许可,商用前先读条款。\n\n参考:[fastino\u002FGLiNER2.5-Decide](https:\u002F\u002Fhuggingface.co\u002Ffastino\u002FGLiNER2.5-Decide) · [fast-decisions 数据集](https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002Ffastino\u002Ffast-decisions) · [fastino-ai\u002FGLiNER2](https:\u002F\u002Fgithub.com\u002Ffastino-ai\u002FGLiNER2)","https:\u002F\u002Fhuggingface.co\u002Ffastino\u002FGLiNER2.5-Decide","24d5c6c5-6573-4180-a1fd-f1459842d1af",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"7ac06d8e-b074-4147-abfc-ffaa4c6b8744","ai-efficiency",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"40269b40-7942-4650-9672-ed2e6524d37a","ai-technology",{"id":19,"name":20,"slug":20,"description":14,"color":14},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",{"id":22,"name":23,"slug":23,"description":14,"color":14},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"f8a15b65-21de-4f67-8033-f486dc80d066","en","Fastino's 340M Classifier Beats 4B Generative Baseline","Fastino's GLiNER2.5-Decide: a 340M open-weights classifier, 60.2% on a 17-domain benchmark, beating 1B and 4B rivals in one prompt-free forward pass.","On September 23, 2026, Fastino released GLiNER2.5-Decide on Hugging Face: a 340M-parameter open-weights encoder built for production decision work — intent detection, sentiment, topic classification, ticket routing, moderation, and spam filtering. It is deliberately not a general assistant. The model card states it does not reason, explain, or answer open questions. It is a specialist for operational decisions.\n\n## One model replaces a rack of classifiers\n\nThe traditional setup runs three separate classifiers for support intent, urgency, and routing — or prompts a general-purpose LLM and pays generative latency on every call. Decide takes a different route: the label set is a runtime argument. The same text and candidate labels go through a single forward pass; single-label tasks return one string, multi-label tasks return every label above threshold. The model card's examples cover banking intent, travel rebooking, clinic triage, aspect-level review sentiment, document type, ticket routing, human-handoff gating, moderation, and incident severity rating.\n\nThe most practical point for deployment: no prompt template, no generated tokens. A generative LLM has to compose an answer for every classification call; an encoder emits structured output directly, landing latency and cost in a different class. The gliner2 library installs with one pip line, runs local inference without a GPU, and the project positions itself as CPU-first with fully local processing.\n\n## 340M vs 4B: what the benchmark shows, and its salt\n\nThe model card publishes the fast-decisions benchmark: 17 domains, 300 held-out examples per domain, identical text and candidate labels for every model. GLiNER2.5-Decide averages 60.2%, ahead of GLiNER2 XL (1B, 59.6%), JevK5 (57.6%), SemIf (a Qwen3.5-4B generative, 56.4%), GLiFormer large-v1 (49.0%), and Laya Router (46.6%). A 340M encoder outscoring a 4B generative baseline is the release's core selling point.\n\nBefore taking that at face value, note the salt: the table comes from Fastino's own dataset, with no third-party reproduction to check whether comparison models were tuned fairly or whether the label-set design favors the home architecture. Treat self-reported numbers with the usual discount. Notably, the lane is crowding: Convai's Laya Router, also pitched at routing and safety checks, sits last in the same table at 46.6%.\n\n## Boundary architecture and Chinese support\n\nGLiNER2.5 uses a boundary architecture: sparse start-end pairing instead of a fixed-width span grid, so any span length inside the encoded window is representable. The GitHub repository (fastino-ai\u002FGLiNER2, Apache 2.0, 1.8k stars) maintains both span and boundary architectures behind one public API, with AutoExtractor dispatching on the saved architecture field. Chinese users get a bonus: GLiNER2 supports a character-level word splitter, so Chinese does not fall back to the English default. The methodology has an EMNLP 2025 System Demonstrations paper behind it.\n\n## So what\n\nThe direction is clear: not every NLP task deserves a large model. High-frequency, low-complexity, latency-sensitive decisions run cheaper and more controllably on a 340M specialist on CPU than on a generative API billed per call. But a vendor-built benchmark is the vendor's own ruler — run your own label set through a held-out pass before adopting; that beats trusting any 60.2%. The model ships under a non-standard \"other\" license, so read the terms before commercial use.\n\nReferences: [fastino\u002FGLiNER2.5-Decide](https:\u002F\u002Fhuggingface.co\u002Ffastino\u002FGLiNER2.5-Decide) · [fast-decisions dataset](https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002Ffastino\u002Ffast-decisions) · [fastino-ai\u002FGLiNER2](https:\u002F\u002Fgithub.com\u002Ffastino-ai\u002FGLiNER2)","fastino-gliner2-5-decide-classifier","2026-09-26T13:09:43Z","2026-09-26T13:10:13.348543Z","2026-09-26T13:10:13.348556Z",true,"agent",786,[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},"ca3472cb-0638-440d-ae0e-f6c3eac95cf7","亚马逊开源2B决策模型:权重、数据、训练脚本全放","amazon-strands-decider-2b","2026-10-04T21:15:00+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"4c4a2a9e-f69b-4985-bd42-97ab2ef4e2ac","Spark-X2.5-4B 开源:4B 跑 1M 上下文,22 项基准打 9B 级 Qwen3.5","spark-x2-5-4b-apache-open-source","2026-09-16T01:30:00+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"550cee5e-18e8-4236-9304-7207ebc221a8","Agnes 3.0 Flash 开源:72 层仅 18 层带 KV 缓存,33B 单卡跑 262k 上下文","agnes-3-0-flash-preview-open-weights","2026-09-13T15:20:00+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"365b770a-2cca-40a0-beb0-1eff823702c0","IBM 开源 Granite Time Series PatchTST-FM-r2:零样本 SOTA,Apache 2.0 商用许可","ibm-granite-patchtst-fm-r2-zero-shot-apache","2026-09-12T11:00:00+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"dc2f4ead-963c-4a8e-bd41-400bebf83bb4","物理、几何、外观一个模型全包:Puffin-World 开源,相机 roll 误差低至 0.26°","puffin-world-native-3d-world-states","2026-09-06T19:09:41+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"454286bd-cb8e-462e-8b33-1b4c77b27262","470M 语音模型 1 秒转写 3.5 小时:IBM 把 ASR 里的语言模型砍掉了","granite-speech-5-turbo-ctc-470m","2026-08-31T15:10:00+00:00"]