{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-07-23T06:40:50.084Z","headline":"NexForge：通过需求驱动任务合成扩展LLM智能体能力","description":"NexForge提出需求驱动框架，无需领域特定基础设施即可自动合成多样化可执行智能体任务与专家轨迹。该框架生成3.6K终端和2K办公任务，将Qwen3.5-35B-A3B Base在Terminal-Bench 2.0上从22.5%提升至52.0%，GDPval Elo从813提升至1338。","url":"https://www.aioga.com/news/cmrvfcrfv01kebihbglmvfomx/","mainEntityOfPage":"https://www.aioga.com/news/cmrvfcrfv01kebihbglmvfomx/","datePublished":"2026-07-17T00:00:00.000Z","dateModified":"2026-07-17T00:00:00.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://arxiv.org/abs/2607.14186","https://aihot.virxact.com/items/cmrvfcrfv01kebihbglmvfomx"],"canonicalUrl":"https://www.aioga.com/news/cmrvfcrfv01kebihbglmvfomx/","directAnswer":{"@type":"Answer","text":"NexForge提出需求驱动的任务合成框架，可在无需领域特定基础设施的条件下，自动生成多样化、可执行的智能体任务及专家轨迹。","url":"https://www.aioga.com/news/cmrvfcrfv01kebihbglmvfomx/","dateCreated":"2026-07-17T00:00:00.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"arXiv source article","url":"https://arxiv.org/abs/2607.14186","datePublished":"2026-07-17T00:00:00.000Z","provider":{"@type":"Organization","name":"arXiv","url":"https://arxiv.org/abs/2607.14186"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmrvfcrfv01kebihbglmvfomx","datePublished":"2026-07-17T00:00:00.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmrvfcrfv01kebihbglmvfomx"}}],"aggregationSource":"HuggingFace Daily Papers（社区热门论文）","originalPublisher":{"name":"arXiv","url":"https://arxiv.org/abs/2607.14186"},"article":{"id":"cmrvfcrfv01kebihbglmvfomx","slug":"cmrvfcrfv01kebihbglmvfomx","url":"https://www.aioga.com/news/cmrvfcrfv01kebihbglmvfomx/","title":"NexForge：通过需求驱动任务合成扩展LLM智能体能力","title_en":"NexForge： Scaling Agent Capabilities through Requirement-Driven Task Synthesis for LLMs","summary":"NexForge提出需求驱动框架，无需领域特定基础设施即可自动合成多样化可执行智能体任务与专家轨迹。该框架生成3.6K终端和2K办公任务，将Qwen3.5-35B-A3B Base在Terminal-Bench 2.0上从22.5%提升至52.0%，GDPval Elo从813提升至1338。","source":"HuggingFace Daily Papers（社区热门论文）","sourceUrl":"https://arxiv.org/abs/2607.14186","aiHotUrl":"https://aihot.virxact.com/items/cmrvfcrfv01kebihbglmvfomx","publishedAt":"2026-07-17T00:00:00.000Z","category":"论文研究","score":75,"selected":true,"articleBody":["arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.","Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs ：https://info.arxiv.org/labs/index.html."],"articleImages":[{"sourceUrl":"https://arxiv.org/static/base/1.0.1/images/funders/simons-foundation.png","alt":"Simons Foundation","afterParagraph":1,"url":"/media/articles/cmrvfcrfv01kebihbglmvfomx/e2d7f38d62f5ca91.png"},{"sourceUrl":"https://arxiv.org/static/base/1.0.1/images/funders/simons-foundation-international.png","alt":"Simons Foundation International","afterParagraph":1,"url":"/media/articles/cmrvfcrfv01kebihbglmvfomx/1d56e29c5557cbdc.png"},{"sourceUrl":"https://arxiv.org/static/base/1.0.1/images/funders/schmidt-sciences.png","alt":"Schmidt Sciences","afterParagraph":1,"url":"/media/articles/cmrvfcrfv01kebihbglmvfomx/8e18212b8219c104.png"}],"mediaStatus":"ok","articleBodyZh":["arXivLabs 是一个框架，允许合作者直接在我们的网站上开发和分享新的 arXiv 功能。","有一个可以为 arXiv 社区增加价值的项目想法吗？了解更多关于 arXivLabs 的信息：https://info.arxiv.org/labs/index.html。"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"NexForge提出需求驱动的任务合成框架，可在无需领域特定基础设施的条件下，自动生成多样化、可执行的智能体任务及专家轨迹。","background":"该研究属于论文研究，来源为HuggingFace Daily Papers社区热门论文。现有材料仅提供摘要数据，正文摘录为arXivLabs通用介绍。","viewpoint":"Aioga判断，NexForge的核心关注点是以需求驱动方式扩展训练任务供给；其价值仍需结合完整论文的方法、实验设置与复现结果评估。","implications":"摘要称，框架生成3.6K终端任务和2K办公任务，并使Qwen3.5-35B-A3B Base在两项所列评测中的成绩提高，值得关注。","nextStep":"建议后续核查完整论文，重点确认任务合成流程、专家轨迹来源、基准评测条件、对照实验及结果可复现性，避免仅凭摘要作出推广判断。","evidenceRefs":["title","summary","articleBody","source"],"status":"published","aiGenerated":true,"autoApproved":true,"generatedBy":"aioga-editorial:gpt-5.6-sol","reviewedBy":"aioga-editorial-review:gpt-5.6-sol","generatedAt":"2026-07-22T21:31:58.813Z","sourceHash":"0b69ee288b3035e1","review":{"approved":true,"groundedness":97,"clarity":93,"duplicationRisk":18,"blockingIssues":[],"notes":["候选内容准确区分了摘要信息与正文摘录的局限，并将“Aioga判断”明确标示为观点。","“两项所列评测中的成绩提高”有来源数据支持；如需提升信息量，可补充Terminal-Bench 2.0从22.5%升至52.0%、GDPval Elo从813升至1338的具体数值，但这不是阻断问题。"]},"validation":{"passed":true,"mode":"ai-auto","revisions":0,"checks":["schema","length","source-attribution","low-source-overlap","no-html","independent-ai-review"]}},"tags":["论文研究","HuggingFace Daily Papers（社区热门论文）"],"translations":{"zh-CN":{"title":"NexForge：通过需求驱动任务合成扩展LLM智能体能力","summary":"NexForge提出需求驱动框架，无需领域特定基础设施即可自动合成多样化可执行智能体任务与专家轨迹。该框架生成3.6K终端和2K办公任务，将Qwen3.5-35B-A3B Base在Terminal-Bench 2.0上从22.5%提升至52.0%，GDPval Elo从813提升至1338。","category":"论文研究","source":"arXiv","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge：通过需求驱动任务合成扩展LLM智能体能力 - Aioga AI资讯","description":"NexForge提出需求驱动框架，无需领域特定基础设施即可自动合成多样化可执行智能体任务与专家轨迹。该框架生成3.6K终端和2K办公任务，将Qwen3.5-35B-A3B Base在Terminal-Bench 2.0上从22.5%提升至52.0%，GDPval Elo从813提升至1338。","url":"https://www.aioga.com/news/cmrvfcrfv01kebihbglmvfomx/"},"en":{"title":"NexForge: Expanding LLM Agent Capabilities Through Demand-Driven Task Synthesis","summary":"NexForge proposes a demand-driven framework that can automatically synthesize diverse executable agent tasks and expert trajectories without domain-specific infrastructure. The framework generates 3.6K terminal and 2K office tasks, improving Qwen3.5-35B-A3B Base from 22.5% to 52.0% on Terminal-Bench 2.0, and raising GDPval Elo from 813 to 1338.","category":"Research","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: Expanding LLM Agent Capabilities Through Demand-Driven Task Synthesis - Aioga AI News","description":"NexForge proposes a demand-driven framework that can automatically synthesize diverse executable agent tasks and expert trajectories without domain-specific infrastructure. The fra...","url":"https://www.aioga.com/en/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:06:28.096Z"},"ja":{"title":"NexForge：需要駆動型タスク合成によるLLMエージェント能力の拡張","summary":"NexForgeは要求駆動型フレームワークを提案し、特定のドメインに依存する基盤がなくても、多様な実行可能なエージェントタスクと専門家の軌跡を自動的に合成できます。このフレームワークは3.6Kの端末タスクと2Kのオフィスタスクを生成し、Qwen3.5-35B-A3B BaseをTerminal-Bench 2.0上で22.5%から52.0%に向上させ、GDPval Eloを813から1338に引き上げました。","category":"論文研究","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge：需要駆動型タスク合成によるLLMエージェント能力の拡張 - Aioga AIニュース","description":"NexForgeは要求駆動型フレームワークを提案し、特定のドメインに依存する基盤がなくても、多様な実行可能なエージェントタスクと専門家の軌跡を自動的に合成できます。このフレームワークは3.6Kの端末タスクと2Kのオフィスタスクを生成し、Qwen3.5-35B-A3B BaseをTerminal-Bench 2.0上で22.5%から52.0%に向上させ、GDP...","url":"https://www.aioga.com/ja/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:06:44.866Z"},"ko":{"title":"NexForge：수요 주도 작업 합성을 통해 LLM 에이전트의 능력 확장","summary":"NexForge는 요구 기반 프레임워크를 제안하며, 도메인 특화 인프라 없이도 다양한 실행 가능한 인공지능 에이전트 작업과 전문가 경로를 자동으로 합성할 수 있습니다. 이 프레임워크는 3.6K 단말기 및 2K 사무 작업을 생성하며, Terminal-Bench 2.0에서 Qwen3.5-35B-A3B Base를 22.5%에서 52.0%로 향상시키고, GDPval Elo를 813에서 1338로 올렸습니다.","category":"연구","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge：수요 주도 작업 합성을 통해 LLM 에이전트의 능력 확장 - Aioga AI 뉴스","description":"NexForge는 요구 기반 프레임워크를 제안하며, 도메인 특화 인프라 없이도 다양한 실행 가능한 인공지능 에이전트 작업과 전문가 경로를 자동으로 합성할 수 있습니다. 이 프레임워크는 3.6K 단말기 및 2K 사무 작업을 생성하며, Terminal-Bench 2.0에서 Qwen3.5-35B-A3B Base를 22.5%에서...","url":"https://www.aioga.com/ko/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:07:34.909Z"},"es":{"title":"NexForge: Ampliando las capacidades de los agentes LLM mediante la síntesis de tareas impulsada por la demanda","summary":"NexForge propone un marco impulsado por la demanda, que puede sintetizar automáticamente tareas inteligentes ejecutables diversas y trayectorias de expertos sin necesidad de infraestructuras específicas de dominio. Este marco genera 3.6K tareas terminales y 2K tareas de oficina, aumentando el Qwen3.5-35B-A3B Base en Terminal-Bench 2.0 del 22.5% al 52.0%, y el GDPval Elo de 813 a 1338.","category":"Investigación","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: Ampliando las capacidades de los agentes LLM mediante la síntesis de tareas impulsada por la demanda - Aioga Noticias de IA","description":"NexForge propone un marco impulsado por la demanda, que puede sintetizar automáticamente tareas inteligentes ejecutables diversas y trayectorias de expertos sin necesidad de infrae...","url":"https://www.aioga.com/es/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:07:32.980Z"},"fr":{"title":"NexForge : Étendre les capacités des agents LLM grâce à la synthèse de tâches basée sur la demande","summary":"NexForge propose un cadre piloté par les besoins, capable de synthétiser automatiquement des tâches diversifiées exécutables par des agents intelligents et des trajectoires d'experts sans infrastructure spécifique au domaine. Ce cadre génère 3,6K tâches terminales et 2K tâches de bureau, améliorant Qwen3.5-35B-A3B Base sur Terminal-Bench 2.0 de 22,5% à 52,0%, et le Elo GDPval de 813 à 1338.","category":"Recherche","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge : Étendre les capacités des agents LLM grâce à la synthèse de tâches basée sur la demande - Aioga Actualités IA","description":"NexForge propose un cadre piloté par les besoins, capable de synthétiser automatiquement des tâches diversifiées exécutables par des agents intelligents et des trajectoires d'exper...","url":"https://www.aioga.com/fr/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:08:17.474Z"},"de":{"title":"NexForge: Erweiterung der LLM-Agentenfähigkeiten durch bedarfsgetriebene Aufgaben-Synthese","summary":"NexForge stellt einen bedarfsorientierten Rahmen vor, der die automatische Synthese vielfältiger ausführbarer Agentenaufgaben und Expertenpfade ermöglicht, ohne dass domainspezifische Infrastruktur erforderlich ist. Dieser Rahmen erzeugt 3,6K Terminal- und 2K Büroaufgaben und verbessert Qwen3.5-35B-A3B Base auf Terminal-Bench 2.0 von 22,5 % auf 52,0 %, während der GDPval Elo von 813 auf 1338 steigt.","category":"论文研究","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: Erweiterung der LLM-Agentenfähigkeiten durch bedarfsgetriebene Aufgaben-Synthese - Aioga KI-News","description":"NexForge stellt einen bedarfsorientierten Rahmen vor, der die automatische Synthese vielfältiger ausführbarer Agentenaufgaben und Expertenpfade ermöglicht, ohne dass domainspezifis...","url":"https://www.aioga.com/de/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:08:21.025Z"},"pt-BR":{"title":"NexForge: Expandindo a capacidade de agentes LLM através da síntese de tarefas orientada por demanda","summary":"A NexForge propôs uma estrutura orientada por demanda, que permite sintetizar automaticamente tarefas diversificadas de agentes executáveis e trajetórias de especialistas sem a necessidade de infraestrutura específica do domínio. Essa estrutura gerou 3,6 mil tarefas de terminal e 2 mil tarefas de escritório, aumentando o Qwen3.5-35B-A3B Base no Terminal-Bench 2.0 de 22,5% para 52,0%, e o Elo do GDPval de 813 para 1338.","category":"论文研究","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: Expandindo a capacidade de agentes LLM através da síntese de tarefas orientada por demanda - Aioga Notícias de IA","description":"A NexForge propôs uma estrutura orientada por demanda, que permite sintetizar automaticamente tarefas diversificadas de agentes executáveis e trajetórias de especialistas sem a nec...","url":"https://www.aioga.com/pt-BR/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:09:04.380Z"},"ru":{"title":"NexForge: Расширение возможностей интеллектуальных агентов LLM через синтез задач, управляемый требованиями","summary":"NexForge предложила фреймворк, основанный на запросах, который позволяет автоматически синтезировать разнообразные исполняемые задачи для агентов и экспертные траектории без необходимости в специализированной инфраструктуре. Этот фреймворк сгенерировал 3,6 тыс. терминальных и 2 тыс. офисных задач, повысив Qwen3.5-35B-A3B Base на Terminal-Bench 2.0 с 22,5% до 52,0%, а Elo по GDPval — с 813 до 1338.","category":"论文研究","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: Расширение возможностей интеллектуальных агентов LLM через синтез задач, управляемый требованиями - Aioga Новости ИИ","description":"NexForge предложила фреймворк, основанный на запросах, который позволяет автоматически синтезировать разнообразные исполняемые задачи для агентов и экспертные траектории без необхо...","url":"https://www.aioga.com/ru/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:09:07.225Z"},"ar":{"title":"NexForge: توسيع قدرات وكيل الذكاء الاصطناعي الكبير من خلال تجميع المهام القائمة على الطلب","summary":"اقترحت NexForge إطار عمل مدفوع بالاحتياجات، يمكنه تلقائيًا توليد مهام متنوعة قابلة للتنفيذ من قبل الوكلاء ومسارات الخبراء دون الحاجة إلى بنية تحتية خاصة بالمجال. يولد هذا الإطار 3.6 ألف مهمة نهائية و2 ألف مهمة مكتبية، وعمل على تحسيـن أداء Qwen3.5-35B-A3B Base على Terminal-Bench 2.0 من 22.5٪ إلى 52.0٪، وزاد مؤشر GDPval Elo من 813 إلى 1338.","category":"论文研究","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: توسيع قدرات وكيل الذكاء الاصطناعي الكبير من خلال تجميع المهام القائمة على الطلب - Aioga أخبار الذكاء الاصطناعي","description":"اقترحت NexForge إطار عمل مدفوع بالاحتياجات، يمكنه تلقائيًا توليد مهام متنوعة قابلة للتنفيذ من قبل الوكلاء ومسارات الخبراء دون الحاجة إلى بنية تحتية خاصة بالمجال. يولد هذا الإطار 3....","url":"https://www.aioga.com/ar/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:09:53.063Z"},"hi":{"title":"NexForge: डिमांड-ड्रिवन टास्क संश्लेषण के माध्यम से LLM एजेंट की क्षमताओं का विस्तार","summary":"NexForge ने एक आवश्यकता-चालित ढांचा प्रस्तुत किया, जो किसी विशिष्ट क्षेत्र के आधारभूत संरचना की आवश्यकता के बिना विविध निष्पादन योग्य बौद्धिक कार्यों और विशेषज्ञ मार्गों को स्वचालित रूप से संश्लेषित कर सकता है। इस ढांचे ने 3.6K टर्मिनल और 2K कार्यालय कार्य उत्पन्न किए, और Qwen3.5-35B-A3B बेस को Terminal-Bench 2.0 पर 22.5% से बढ़ाकर 52.0% कर दिया, जबकि GDPval Elo 813 से बढ़कर 1338 हो गया।","category":"论文研究","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: डिमांड-ड्रिवन टास्क संश्लेषण के माध्यम से LLM एजेंट की क्षमताओं का विस्तार - Aioga AI समाचार","description":"NexForge ने एक आवश्यकता-चालित ढांचा प्रस्तुत किया, जो किसी विशिष्ट क्षेत्र के आधारभूत संरचना की आवश्यकता के बिना विविध निष्पादन योग्य बौद्धिक कार्यों और विशेषज्ञ मार्गों को स्वचालि...","url":"https://www.aioga.com/hi/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:09:56.950Z"},"it":{"title":"NexForge: Estendere le capacità degli agenti LLM attraverso la sintesi dei compiti guidata dalla domanda","summary":"NexForge propone un framework guidato dalla domanda, in grado di sintetizzare automaticamente compiti e traiettorie di esperti diversificati senza infrastrutture specifiche di dominio. Il framework genera 3,6K compiti terminali e 2K compiti d'ufficio, portando Qwen3.5-35B-A3B Base da 22,5% a 52,0% su Terminal-Bench 2.0, e l'Elo di GDPval da 813 a 1338.","category":"论文研究","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: Estendere le capacità degli agenti LLM attraverso la sintesi dei compiti guidata dalla domanda - Aioga Notizie IA","description":"NexForge propone un framework guidato dalla domanda, in grado di sintetizzare automaticamente compiti e traiettorie di esperti diversificati senza infrastrutture specifiche di domi...","url":"https://www.aioga.com/it/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:10:48.783Z"},"nl":{"title":"NexForge: Het uitbreiden van de capaciteit van LLM-agenten door taaksynthese op basis van vraag","summary":"NexForge stelt een vraaggestuurd kader voor, waarmee gevarieerde uitvoerbare agenttaken en experttrajecten automatisch kunnen worden samengesteld zonder specifieke domeininfrastructuur. Het kader genereert 3,6K terminal- en 2K kantoortaken, en verhoogt Qwen3.5-35B-A3B Base op Terminal-Bench 2.0 van 22,5% naar 52,0%, terwijl de GDPval Elo stijgt van 813 naar 1338.","category":"论文研究","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: Het uitbreiden van de capaciteit van LLM-agenten door taaksynthese op basis van vraag - Aioga AI-nieuws","description":"NexForge stelt een vraaggestuurd kader voor, waarmee gevarieerde uitvoerbare agenttaken en experttrajecten automatisch kunnen worden samengesteld zonder specifieke domeininfrastruc...","url":"https://www.aioga.com/nl/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:10:40.645Z"},"tr":{"title":"NexForge: Talep odaklı görev sentezi ile LLM ajanlarının yeteneklerini genişletme","summary":"NexForge, alan spesifik altyapıya ihtiyaç duymadan çeşitli yürütülebilir akıllı ajan görevleri ve uzman yolları otomatik olarak sentezleyebilen ihtiyaç odaklı bir çerçeve öneriyor. Bu çerçeve, 3.6K terminal ve 2K ofis görevini üretir ve Qwen3.5-35B-A3B Base'i Terminal-Bench 2.0 üzerinde %22,5'ten %52,0'ye yükseltir, GDPval Elo'yu 813'ten 1338'e çıkarır.","category":"论文研究","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: Talep odaklı görev sentezi ile LLM ajanlarının yeteneklerini genişletme - Aioga AI Haberleri","description":"NexForge, alan spesifik altyapıya ihtiyaç duymadan çeşitli yürütülebilir akıllı ajan görevleri ve uzman yolları otomatik olarak sentezleyebilen ihtiyaç odaklı bir çerçeve öneriyor....","url":"https://www.aioga.com/tr/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:11:29.693Z"},"vi":{"title":"NexForge: Mở rộng khả năng của tác nhân LLM thông qua tổng hợp nhiệm vụ dựa trên nhu cầu","summary":"NexForge đề xuất khung dựa trên nhu cầu, không cần cơ sở hạ tầng chuyên ngành vẫn có thể tự động tổng hợp các nhiệm vụ tác nhân thông minh và lộ trình chuyên gia đa dạng. Khung này tạo ra 3,6 nghìn nhiệm vụ đầu cuối và 2 nghìn nhiệm vụ văn phòng, nâng Qwen3.5-35B-A3B Base trên Terminal-Bench 2.0 từ 22,5% lên 52,0%, GDPval Elo từ 813 lên 1338.","category":"论文研究","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: Mở rộng khả năng của tác nhân LLM thông qua tổng hợp nhiệm vụ dựa trên nhu cầu - Tin tức AI Aioga","description":"NexForge đề xuất khung dựa trên nhu cầu, không cần cơ sở hạ tầng chuyên ngành vẫn có thể tự động tổng hợp các nhiệm vụ tác nhân thông minh và lộ trình chuyên gia đa dạng. Khung này...","url":"https://www.aioga.com/vi/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:11:34.830Z"},"id":{"title":"NexForge: Memperluas kemampuan agen LLM melalui sintesis tugas berbasis permintaan","summary":"NexForge mengajukan kerangka kerja berbasis kebutuhan, yang dapat secara otomatis menyintesis berbagai tugas agen yang dapat dieksekusi dan jejak ahli tanpa membutuhkan infrastruktur khusus domain. Kerangka kerja ini menghasilkan 3.6K tugas terminal dan 2K tugas kantor, meningkatkan Qwen3.5-35B-A3B Base di Terminal-Bench 2.0 dari 22,5% menjadi 52,0%, dan GDPval Elo dari 813 menjadi 1338.","category":"论文研究","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: Memperluas kemampuan agen LLM melalui sintesis tugas berbasis permintaan - Berita AI Aioga","description":"NexForge mengajukan kerangka kerja berbasis kebutuhan, yang dapat secara otomatis menyintesis berbagai tugas agen yang dapat dieksekusi dan jejak ahli tanpa membutuhkan infrastrukt...","url":"https://www.aioga.com/id/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:12:15.283Z"},"th":{"title":"NexForge: ขยายความสามารถของเอเจนต์ LLM ผ่านการสร้างงานตามความต้องการ","summary":"NexForge ได้นำเสนอกรอบงานที่ขับเคลื่อนด้วยความต้องการ โดยไม่จำเป็นต้องใช้โครงสร้างพื้นฐานเฉพาะทาง ก็สามารถสังเคราะห์งานตัวแทนอัจฉริยะที่หลากหลายและเส้นทางผู้เชี่ยวชาญได้โดยอัตโนมัติ กรอบงานนี้สร้างงานปลายทาง 3.6K และงานสำนักงาน 2K ทำให้ Qwen3.5-35B-A3B Base ใน Terminal-Bench 2.0 เพิ่มขึ้นจาก 22.5% เป็น 52.0% GDPval Elo เพิ่มจาก 813 เป็น 1338","category":"论文研究","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: ขยายความสามารถของเอเจนต์ LLM ผ่านการสร้างงานตามความต้องการ - ข่าว AI Aioga","description":"NexForge ได้นำเสนอกรอบงานที่ขับเคลื่อนด้วยความต้องการ โดยไม่จำเป็นต้องใช้โครงสร้างพื้นฐานเฉพาะทาง ก็สามารถสังเคราะห์งานตัวแทนอัจฉริยะที่หลากหลายและเส้นทางผู้เชี่ยวชาญได้โดยอัตโนมัต...","url":"https://www.aioga.com/th/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:12:24.595Z"},"pl":{"title":"NexForge: Rozszerzanie możliwości agentów LLM poprzez syntezę zadań napędzaną potrzebami","summary":"NexForge wprowadza ramy oparte na popycie, które pozwalają automatycznie tworzyć zróżnicowane zadania wykonywalnych agentów i trajektorie ekspertów bez potrzeby specyficznej dla dziedziny infrastruktury podstawowej. Te ramy generują 3,6 tys. zadań terminalowych i 2 tys. biurowych, podnosząc wynik Qwen3.5-35B-A3B Base na Terminal-Bench 2.0 z 22,5% do 52,0%, a Elo GDPval z 813 do 1338.","category":"论文研究","source":"HuggingFace Daily Papers（社区热门论文）","aggregationSource":"HuggingFace Daily Papers（社区热门论文）","pageTitle":"NexForge: Rozszerzanie możliwości agentów LLM poprzez syntezę zadań napędzaną potrzebami - Aioga Wiadomości AI","description":"NexForge wprowadza ramy oparte na popycie, które pozwalają automatycznie tworzyć zróżnicowane zadania wykonywalnych agentów i trajektorie ekspertów bez potrzeby specyficznej dla dz...","url":"https://www.aioga.com/pl/news/cmrvfcrfv01kebihbglmvfomx/","contentTranslated":true,"sourceHash":"48c00c74bfadd84a","translatedAt":"2026-07-22T17:13:13.091Z"}}}}