{"article":{"annotation":{"article_id":"multi-agent-session-alternation-tests-machine-learning-engineering","available_at":"2026-10-13T00:52:16.046437Z","geography_revision":"ef6ee0010dda5fe7b240fd514749b43e214c57da73222d2e40ec83b276dd0977","id":"cbbb7175821fee0a13e298e7658e769e66dc065fcb79ddbe76a37e62f8a3318e","input_sha256":"6512fc4c0224e8bbcd7e6c2fe728a9633e4e5132f1585ebabdad6bd6aba48848","limitations":["Rule-derived subjects are candidates, not verified editorial labels or calibrated model probabilities.","Coverage is distinct from event location. Unknown country/city/coordinates remain absent; no automatic geocoding.","Article geometries keep their stated precision and methodology; reproducing them is not independent coordinate verification."],"method":"controlled-rules-v1","places":[{"admin1":null,"city":null,"coordinate_note":null,"coordinate_reference_system":"OGC:CRS84","coordinate_source":null,"country":"CN","evidence":[{"field":"country","language":"en","note":"Explicit preserved editorial country/region; no city or national centroid inferred","source_urls":["https://arxiv.org/html/2610.13494v1"]}],"geometry":null,"id":"iso3166-1:CN","kind":"country-or-area","labels":{"ar":"الصين","en":"China","es":"China","fr":"Chine"},"match_status":"candidate","precision":"regional","role":"coverage","uncertainty_m":null}],"rules_sha256":"fb5ad3c27c91cb840f7a0a05f7f64a5cb26ad0374409ff9457f277c7092e22b4","schema_version":1,"subjects":[{"concept_id":"medtop:13000000","confidence":null,"confidence_method":null,"evidence":[{"field":"category+topics","language":"en","note":"Preserved editorial category/topic; broad candidate only","source_urls":["https://arxiv.org/html/2610.13494v1"]}],"method":"controlled-rules-v1","rule_id":"technology-broad","status":"candidate"}],"taxonomy_revision":"1f2db31d811fca07e99b16d402c10e2cc2965c9df1c7b41d9be348b40a24e7fb","taxonomy_version":"IPTC Media Topics 2026-Q2"},"api_url":"https://newsverum.com/api/v1/articles/multi-agent-session-alternation-tests-machine-learning-engineering?lang=fr","category":"technology","content":{"key_points":[],"paragraphs":["Des chercheurs associés à Tsinghua, Zhejiang et NVIDIA ont présenté Humanize MLE Agents, une méthode qui alterne les sessions de deux agents sur des tâches longues d’ingénierie en apprentissage automatique. Chaque session s’arrête après un nombre plafonné d’expériences; le contexte est réinitialisé entre les sessions, tandis que le code, les modèles et les résultats antérieurs restent dans un espace de travail commun.","Sur 75 tâches Kaggle de MLE-bench, les auteurs rapportent un taux de 78,2 % pour obtenir au moins une médaille et indiquent que le workflow a dépassé la meilleure référence open source avec un budget temporel inférieur. Il s’agit d’un résultat propre à ce jeu de concours et à ces configurations d’agents; il ne garantit pas le même gain avec d’autres versions de modèles ou dans des projets réels."],"sections":[]},"corrections":[],"country":"CN","data":{"blocks":[{"columns":[{"key":"metric","label":{"ar":"المقياس","de":"Messgröße","en":"Measure","es":"Medida","fr":"Mesure","pt":"Medida","tr":"Ölçüt"},"language":"en","type":"text"},{"key":"value","label":{"ar":"القيمة المعلنة","de":"Berichteter Wert","en":"Reported value","es":"Valor comunicado","fr":"Valeur rapportée","pt":"Valor informado","tr":"Bildirilen değer"},"type":"number","unit":"percent"},{"key":"scope","label":{"ar":"نطاق التجربة","de":"Versuchsumfang","en":"Experimental scope","es":"Alcance experimental","fr":"Périmètre expérimental","pt":"Escopo experimental","tr":"Deney kapsamı"},"language":"en","type":"text"}],"description":{"ar":"75 مهمة Kaggle؛ تقييم MLE-bench","de":"75 Kaggle-Aufgaben; MLE-bench-Auswertung","en":"75 Kaggle tasks; MLE-bench evaluation","es":"75 tareas de Kaggle; evaluación MLE-bench","fr":"75 tâches Kaggle; évaluation MLE-bench","pt":"75 tarefas do Kaggle; avaliação MLE-bench","tr":"75 Kaggle görevi; MLE-bench değerlendirmesi"},"id":"reported-result-1","methodology":{"ar":"القيم من تجارب المعيار التي حددتها الورقة؛ وليست قياسات ميدانية أو ضمانات أداء عامة.","de":"Die Werte stammen aus den im Artikel beschriebenen Benchmark-Experimenten; sie sind weder Feldmessungen noch allgemeine Leistungsgarantien.","en":"Values come from the paper's stated benchmark experiments; they are not field measurements or general performance guarantees.","es":"Las cifras proceden de los experimentos de benchmark descritos en el artículo; no son mediciones de campo ni una garantía general de rendimiento.","fr":"Les valeurs proviennent des expériences de benchmark décrites dans l’article; elles ne constituent ni des mesures de terrain ni une garantie générale de performance.","pt":"Os valores vêm dos experimentos de benchmark descritos no artigo; não são medições de campo nem garantias gerais de desempenho.","tr":"Rakamlar makalenin tanımladığı kıyaslama deneylerine aittir; saha ölçümü veya genel performans garantisi değildir."},"rowTranslations":[{"metric":{"ar":"معدل الفوز بأي ميدالية في مهام Kaggle ضمن MLE-bench","de":"Any-Medal-Rate bei Kaggle-Aufgaben in MLE-bench","en":"Any-medal rate on MLE-bench Kaggle tasks","es":"Tasa de obtención de alguna medalla en tareas Kaggle de MLE-bench","fr":"Taux de médaille obtenue sur les tâches Kaggle de MLE-bench","pt":"Taxa de obtenção de alguma medalha em tarefas Kaggle do MLE-bench","tr":"MLE-bench Kaggle görevlerinde herhangi madalya oranı"},"scope":{"ar":"75 مهمة Kaggle؛ تقييم MLE-bench","de":"75 Kaggle-Aufgaben; MLE-bench-Auswertung","en":"75 Kaggle tasks; MLE-bench evaluation","es":"75 tareas de Kaggle; evaluación MLE-bench","fr":"75 tâches Kaggle; évaluation MLE-bench","pt":"75 tarefas do Kaggle; avaliação MLE-bench","tr":"75 Kaggle görevi; MLE-bench değerlendirmesi"}}],"rows":[{"metric":"Any-medal rate on MLE-bench Kaggle tasks","scope":"75 Kaggle tasks; MLE-bench evaluation","value":78.2}],"sourceUrls":["https://arxiv.org/html/2610.13494v1"],"title":{"ar":"المقياس","de":"Messgröße","en":"Measure","es":"Medida","fr":"Mesure","pt":"Medida","tr":"Ölçüt"},"type":"table","updatedAt":"2026-10-13T00:12:48Z"},{"columns":[{"key":"metric","label":{"ar":"المقياس","de":"Messgröße","en":"Measure","es":"Medida","fr":"Mesure","pt":"Medida","tr":"Ölçüt"},"language":"en","type":"text"},{"key":"value","label":{"ar":"القيمة المعلنة","de":"Berichteter Wert","en":"Reported value","es":"Valor comunicado","fr":"Valeur rapportée","pt":"Valor informado","tr":"Bildirilen değer"},"type":"number","unit":"count"},{"key":"scope","label":{"ar":"نطاق التجربة","de":"Versuchsumfang","en":"Experimental scope","es":"Alcance experimental","fr":"Périmètre expérimental","pt":"Escopo experimental","tr":"Deney kapsamı"},"language":"en","type":"text"}],"description":{"ar":"مجموعة مهام Kaggle المذكورة في تقييم MLE-bench","de":"Im MLE-bench-Test ausgewerteter Kaggle-Aufgabensatz","en":"Kaggle task set reported in the MLE-bench evaluation","es":"Conjunto de tareas Kaggle informado en la evaluación MLE-bench","fr":"Ensemble de tâches Kaggle indiqué dans l’évaluation MLE-bench","pt":"Conjunto de tarefas Kaggle informado na avaliação do MLE-bench","tr":"MLE-bench değerlendirmesinde raporlanan Kaggle görev kümesi"},"id":"result-75","methodology":{"ar":"القيم من تجارب المعيار التي حددتها الورقة؛ وليست قياسات ميدانية أو ضمانات أداء عامة.","de":"Die Werte stammen aus den im Artikel beschriebenen Benchmark-Experimenten; sie sind weder Feldmessungen noch allgemeine Leistungsgarantien.","en":"Values come from the paper's stated benchmark experiments; they are not field measurements or general performance guarantees.","es":"Las cifras proceden de los experimentos de benchmark descritos en el artículo; no son mediciones de campo ni una garantía general de rendimiento.","fr":"Les valeurs proviennent des expériences de benchmark décrites dans l’article; elles ne constituent ni des mesures de terrain ni une garantie générale de performance.","pt":"Os valores vêm dos experimentos de benchmark descritos no artigo; não são medições de campo nem garantias gerais de desempenho.","tr":"Rakamlar makalenin tanımladığı kıyaslama deneylerine aittir; saha ölçümü veya genel performans garantisi değildir."},"rowTranslations":[{"metric":{"ar":"عدد مهام Kaggle في MLE-bench","de":"Zahl der Kaggle-Aufgaben in MLE-bench","en":"Number of MLE-bench Kaggle tasks","es":"Número de tareas Kaggle de MLE-bench","fr":"Nombre de tâches Kaggle de MLE-bench","pt":"Número de tarefas Kaggle do MLE-bench","tr":"MLE-bench Kaggle görev sayısı"},"scope":{"ar":"مجموعة مهام Kaggle المذكورة في تقييم MLE-bench","de":"Im MLE-bench-Test ausgewerteter Kaggle-Aufgabensatz","en":"Kaggle task set reported in the MLE-bench evaluation","es":"Conjunto de tareas Kaggle informado en la evaluación MLE-bench","fr":"Ensemble de tâches Kaggle indiqué dans l’évaluation MLE-bench","pt":"Conjunto de tarefas Kaggle informado na avaliação do MLE-bench","tr":"MLE-bench değerlendirmesinde raporlanan Kaggle görev kümesi"}}],"rows":[{"metric":"Number of MLE-bench Kaggle tasks","scope":"Kaggle task set reported in the MLE-bench evaluation","value":75}],"sourceUrls":["https://arxiv.org/html/2610.13494v1"],"title":{"ar":"المقياس","de":"Messgröße","en":"Measure","es":"Medida","fr":"Mesure","pt":"Medida","tr":"Ölçüt"},"type":"table","updatedAt":"2026-10-13T00:12:48Z"}],"version":1},"data_downloads":[{"formats":{"csv":"https://newsverum.com/api/v1/articles/multi-agent-session-alternation-tests-machine-learning-engineering/data/reported-result-1?lang=fr&format=csv","json":"https://newsverum.com/api/v1/articles/multi-agent-session-alternation-tests-machine-learning-engineering/data/reported-result-1?lang=fr&format=json"},"id":"reported-result-1"},{"formats":{"csv":"https://newsverum.com/api/v1/articles/multi-agent-session-alternation-tests-machine-learning-engineering/data/result-75?lang=fr&format=csv","json":"https://newsverum.com/api/v1/articles/multi-agent-session-alternation-tests-machine-learning-engineering/data/result-75?lang=fr&format=json"},"id":"result-75"}],"data_types":["table"],"editorial_illustration":{"alt":"Illustration conceptuelle de la technologie, des processeurs et des informations connectées.","asset_id":"newsverum-v1-technology","avif_variants":[{"height":320,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-480.avif","width":480},{"height":480,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-720.avif","width":720},{"height":640,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-960.avif","width":960},{"height":800,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1200.avif","width":1200},{"height":1024,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1536.avif","width":1536}],"credit":"Newsverum","documentary":false,"height":1024,"kind":"symbolic-editorial-illustration","language":"fr","scope":"topic","url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1536.webp","variants":[{"height":320,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-480.webp","width":480},{"height":640,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-960.webp","width":960},{"height":1024,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1536.webp","width":1536}],"width":1536},"event_id":"event-5feb6b74-59c2-5e3f-984a-af92c529aa84","event_period":"2026-10-09","event_period_basis":"event-or-reference-period","id":"multi-agent-session-alternation-tests-machine-learning-engineering","image":null,"importance":0.6,"language":"fr","modified_at":"2026-10-13T00:51:41+00:00","publication_basis":"publication-record","published_at":"2026-10-13T00:51:41+00:00","reading_minutes":1,"region":"world","section_illustration":{"alt":"Illustration conceptuelle de la technologie, des processeurs et des informations connectées.","asset_id":"newsverum-v1-technology","avif_variants":[{"height":320,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-480.avif","width":480},{"height":480,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-720.avif","width":720},{"height":640,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-960.avif","width":960},{"height":800,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1200.avif","width":1200},{"height":1024,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1536.avif","width":1536}],"credit":"Newsverum","documentary":false,"height":1024,"kind":"symbolic-editorial-illustration","language":"fr","scope":"section","url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1536.webp","variants":[{"height":320,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-480.webp","width":480},{"height":640,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-960.webp","width":960},{"height":1024,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1536.webp","width":1536}],"width":1536},"slug":"multi-agent-session-alternation-tests-machine-learning-engineering","sources":[{"checkedAt":"2026-10-13","name":"arXiv preprint by Tsinghua University, Zhejiang University and collaborating researchers","publishedAt":"2026-10-09","url":"https://arxiv.org/html/2610.13494v1"}],"summary":"Humanize MLE Agents fait alterner deux agents, réinitialise le contexte de chaque session et conserve un espace de travail partagé.","text":"Humanize MLE Agents fait alterner deux agents, réinitialise le contexte de chaque session et conserve un espace de travail partagé.\n\nDes chercheurs associés à Tsinghua, Zhejiang et NVIDIA ont présenté Humanize MLE Agents, une méthode qui alterne les sessions de deux agents sur des tâches longues d’ingénierie en apprentissage automatique. Chaque session s’arrête après un nombre plafonné d’expériences; le contexte est réinitialisé entre les sessions, tandis que le code, les modèles et les résultats antérieurs restent dans un espace de travail commun.\n\nSur 75 tâches Kaggle de MLE-bench, les auteurs rapportent un taux de 78,2 % pour obtenir au moins une médaille et indiquent que le workflow a dépassé la meilleure référence open source avec un budget temporel inférieur. Il s’agit d’un résultat propre à ce jeu de concours et à ces configurations d’agents; il ne garantit pas le même gain avec d’autres versions de modèles ou dans des projets réels.","title":"Des chercheurs alternent les sessions d’agents sur des tâches d’apprentissage automatique","topics":[],"translations":[{"api_url":"https://newsverum.com/api/v1/articles/multi-agent-session-alternation-tests-machine-learning-engineering?lang=tr","language":"tr","url":"https://newsverum.com/tr/news/multi-agent-session-alternation-tests-machine-learning-engineering"},{"api_url":"https://newsverum.com/api/v1/articles/multi-agent-session-alternation-tests-machine-learning-engineering?lang=en","language":"en","url":"https://newsverum.com/en/news/multi-agent-session-alternation-tests-machine-learning-engineering"},{"api_url":"https://newsverum.com/api/v1/articles/multi-agent-session-alternation-tests-machine-learning-engineering?lang=ar","language":"ar","url":"https://newsverum.com/ar/news/multi-agent-session-alternation-tests-machine-learning-engineering"},{"api_url":"https://newsverum.com/api/v1/articles/multi-agent-session-alternation-tests-machine-learning-engineering?lang=fr","language":"fr","url":"https://newsverum.com/fr/news/multi-agent-session-alternation-tests-machine-learning-engineering"},{"api_url":"https://newsverum.com/api/v1/articles/multi-agent-session-alternation-tests-machine-learning-engineering?lang=es","language":"es","url":"https://newsverum.com/es/news/multi-agent-session-alternation-tests-machine-learning-engineering"},{"api_url":"https://newsverum.com/api/v1/articles/multi-agent-session-alternation-tests-machine-learning-engineering?lang=de","language":"de","url":"https://newsverum.com/de/news/multi-agent-session-alternation-tests-machine-learning-engineering"},{"api_url":"https://newsverum.com/api/v1/articles/multi-agent-session-alternation-tests-machine-learning-engineering?lang=pt","language":"pt","url":"https://newsverum.com/pt/news/multi-agent-session-alternation-tests-machine-learning-engineering"}],"url":"https://newsverum.com/fr/news/multi-agent-session-alternation-tests-machine-learning-engineering"},"version":"1"}