{"article":{"annotation":{"article_id":"swe-critix-trains-agent-critics-on-code-repair-trajectories","available_at":"2026-10-13T00:52:26.287290Z","geography_revision":"ef6ee0010dda5fe7b240fd514749b43e214c57da73222d2e40ec83b276dd0977","id":"4cbe9400dd42c8f6c08edadb3e5ca6d84dacdaf6afd80c3b2b11df4f7aa126df","input_sha256":"7f01657d4e2739335df074a4f38139b813c547bc9e39d9789e180cd2682a6b13","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":"HK","evidence":[{"field":"country","language":"en","note":"Explicit preserved editorial country/region; no city or national centroid inferred","source_urls":["https://arxiv.org/html/2610.13356v1"]}],"geometry":null,"id":"iso3166-1:HK","kind":"country-or-area","labels":{"ar":"الصين، منطقة هونغ كونغ الإدارية الخاصة","en":"China, Hong Kong Special Administrative Region","es":"China, región administrativa especial de Hong Kong","fr":"Chine, région administrative spéciale de Hong Kong"},"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/2610.13356v1"]}],"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/swe-critix-trains-agent-critics-on-code-repair-trajectories?lang=fr","category":"technology","content":{"key_points":[],"paragraphs":["Des chercheurs associés à Huawei et à l’université Monash en Malaisie ont présenté SWE-Critix, une méthode d’entraînement de vérificateurs qui jugent si la trajectoire d’exécution d’un agent de programmation a accompli sa tâche. Un modèle enseignant compare des trajectoires réussies et échouées pour un même problème afin de produire des explications d’entraînement, avant un ajustement supervisé et un apprentissage par renforcement fondé sur les résultats des tests.","Le système final a atteint une précision globale de 78 % dans l’évaluation CodeForge Preview de l’article. Les auteurs l’ont mesurée sur un jeu de test issu de la même source de données; ils indiquent que la récompense des tests a aidé à distinguer les réparations réussies, tandis que le rappel restait faible pour certaines catégories d’erreurs. Ce résultat préliminaire de benchmark ne démontre pas la réussite sur des dépôts réels."],"sections":[]},"corrections":[],"country":"HK","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":"النسبة المئوية","de":"Prozent","en":"Percent","es":"Porcentaje","fr":"Pourcentage","pt":"Percentual","tr":"Yüzde"},"type":"number","unit":"percent"},{"key":"scope","label":{"ar":"النطاق","de":"Umfang","en":"Scope","es":"Alcance","fr":"Périmètre","pt":"Escopo","tr":"Kapsam"},"language":"en","type":"text"}],"description":{"ar":"CodeForge Preview؛ مسارات اختبار مشتقة من المصدر نفسه؛ الدقة الإجمالية وفق المؤلفين","de":"CodeForge Preview; Testverläufe aus derselben Quelle; von den Autoren berichtete Gesamtgenauigkeit","en":"CodeForge Preview; test trajectories derived from the same source; overall accuracy reported by authors","es":"CodeForge Preview; trayectorias de prueba derivadas de la misma fuente; precisión general comunicada por los autores","fr":"CodeForge Preview; trajectoires de test issues de la même source; précision globale rapportée par les auteurs","pt":"CodeForge Preview; trajetórias de teste derivadas da mesma fonte; acurácia geral informada pelos autores","tr":"CodeForge Preview; aynı kaynaktan türetilen test izleri; yazarların bildirdiği genel doğruluk"},"id":"reported-result-1","methodology":{"ar":"تقتصر القيمة على تجربة المعيار المذكورة في الورقة؛ وليست نتيجة ميدانية أو ضماناً عاماً للأداء.","de":"Der Wert ist auf das im Artikel beschriebene Benchmark-Experiment beschränkt; er ist kein Feldergebnis und keine allgemeine Leistungsgarantie.","en":"The value is limited to the benchmark experiment reported in the paper; it is not a field result or general performance guarantee.","es":"El valor se limita al experimento de benchmark descrito en el artículo; no es un resultado de campo ni una garantía general.","fr":"La valeur se limite à l’expérience de benchmark décrite dans l’article; ce n’est ni un résultat de terrain ni une garantie générale.","pt":"O valor se limita ao experimento de benchmark descrito no artigo; não é um resultado de campo nem uma garantia geral.","tr":"Değer, makalenin belirttiği benchmark deneyiyle sınırlıdır; saha sonucu veya genel performans garantisi değildir."},"rowTranslations":[{"metric":{"ar":"الدقة الإجمالية لمسارات اختبار CodeForge Preview","de":"Gesamtgenauigkeit bei CodeForge-Preview-Testverläufen","en":"Overall accuracy on CodeForge Preview test trajectories","es":"Precisión general en trayectorias de prueba de CodeForge Preview","fr":"Précision globale sur les trajectoires de test CodeForge Preview","pt":"Acurácia geral nas trajetórias de teste do CodeForge Preview","tr":"CodeForge Preview test izlerinde genel doğruluk"},"scope":{"ar":"CodeForge Preview؛ مسارات اختبار مشتقة من المصدر نفسه؛ الدقة الإجمالية وفق المؤلفين","de":"CodeForge Preview; Testverläufe aus derselben Quelle; von den Autoren berichtete Gesamtgenauigkeit","en":"CodeForge Preview; test trajectories derived from the same source; overall accuracy reported by authors","es":"CodeForge Preview; trayectorias de prueba derivadas de la misma fuente; precisión general comunicada por los autores","fr":"CodeForge Preview; trajectoires de test issues de la même source; précision globale rapportée par les auteurs","pt":"CodeForge Preview; trajetórias de teste derivadas da mesma fonte; acurácia geral informada pelos autores","tr":"CodeForge Preview; aynı kaynaktan türetilen test izleri; yazarların bildirdiği genel doğruluk"}}],"rows":[{"metric":"Overall accuracy on CodeForge Preview test trajectories","scope":"CodeForge Preview; test trajectories derived from the same source; overall accuracy reported by authors","value":78}],"sourceUrls":["https://arxiv.org/html/2610.13356v1"],"title":{"ar":"نتيجة البحث","de":"Forschungsergebnis","en":"Research result","es":"Resultado del estudio","fr":"Résultat de l’étude","pt":"Resultado da pesquisa","tr":"Araştırma sonucu"},"type":"table","updatedAt":"2026-10-13T00:31:08Z"}],"version":1},"data_downloads":[{"formats":{"csv":"https://newsverum.com/api/v1/articles/swe-critix-trains-agent-critics-on-code-repair-trajectories/data/reported-result-1?lang=fr&format=csv","json":"https://newsverum.com/api/v1/articles/swe-critix-trains-agent-critics-on-code-repair-trajectories/data/reported-result-1?lang=fr&format=json"},"id":"reported-result-1"}],"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-a97c71e5-07d3-531c-8218-919476d4e033","event_period":"2026-10-09","event_period_basis":"event-or-reference-period","id":"swe-critix-trains-agent-critics-on-code-repair-trajectories","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":"swe-critix-trains-agent-critics-on-code-repair-trajectories","sources":[{"checkedAt":"2026-10-13","name":"arXiv preprint by researchers affiliated with Huawei and Monash University Malaysia","publishedAt":"2026-10-09","url":"https://arxiv.org/html/2610.13356v1"}],"summary":"La méthode entraîne un agent de code à critiquer des trajectoires de réparation à l’aide d’un raisonnement différentiel et de récompenses de tests.","text":"La méthode entraîne un agent de code à critiquer des trajectoires de réparation à l’aide d’un raisonnement différentiel et de récompenses de tests.\n\nDes chercheurs associés à Huawei et à l’université Monash en Malaisie ont présenté SWE-Critix, une méthode d’entraînement de vérificateurs qui jugent si la trajectoire d’exécution d’un agent de programmation a accompli sa tâche. Un modèle enseignant compare des trajectoires réussies et échouées pour un même problème afin de produire des explications d’entraînement, avant un ajustement supervisé et un apprentissage par renforcement fondé sur les résultats des tests.\n\nLe système final a atteint une précision globale de 78 % dans l’évaluation CodeForge Preview de l’article. Les auteurs l’ont mesurée sur un jeu de test issu de la même source de données; ils indiquent que la récompense des tests a aidé à distinguer les réparations réussies, tandis que le rappel restait faible pour certaines catégories d’erreurs. Ce résultat préliminaire de benchmark ne démontre pas la réussite sur des dépôts réels.","title":"SWE-Critix entraîne des agents critiques pour la réparation de code","topics":[],"translations":[{"api_url":"https://newsverum.com/api/v1/articles/swe-critix-trains-agent-critics-on-code-repair-trajectories?lang=tr","language":"tr","url":"https://newsverum.com/tr/news/swe-critix-trains-agent-critics-on-code-repair-trajectories"},{"api_url":"https://newsverum.com/api/v1/articles/swe-critix-trains-agent-critics-on-code-repair-trajectories?lang=en","language":"en","url":"https://newsverum.com/en/news/swe-critix-trains-agent-critics-on-code-repair-trajectories"},{"api_url":"https://newsverum.com/api/v1/articles/swe-critix-trains-agent-critics-on-code-repair-trajectories?lang=ar","language":"ar","url":"https://newsverum.com/ar/news/swe-critix-trains-agent-critics-on-code-repair-trajectories"},{"api_url":"https://newsverum.com/api/v1/articles/swe-critix-trains-agent-critics-on-code-repair-trajectories?lang=fr","language":"fr","url":"https://newsverum.com/fr/news/swe-critix-trains-agent-critics-on-code-repair-trajectories"},{"api_url":"https://newsverum.com/api/v1/articles/swe-critix-trains-agent-critics-on-code-repair-trajectories?lang=es","language":"es","url":"https://newsverum.com/es/news/swe-critix-trains-agent-critics-on-code-repair-trajectories"},{"api_url":"https://newsverum.com/api/v1/articles/swe-critix-trains-agent-critics-on-code-repair-trajectories?lang=de","language":"de","url":"https://newsverum.com/de/news/swe-critix-trains-agent-critics-on-code-repair-trajectories"},{"api_url":"https://newsverum.com/api/v1/articles/swe-critix-trains-agent-critics-on-code-repair-trajectories?lang=pt","language":"pt","url":"https://newsverum.com/pt/news/swe-critix-trains-agent-critics-on-code-repair-trajectories"}],"url":"https://newsverum.com/fr/news/swe-critix-trains-agent-critics-on-code-repair-trajectories"},"version":"1"}