https://korshunov.ai/en/article/12161-microsoft-releases-bitnet-0-6b-embedding-model-on-hugging-face/korshunov.aien2026-07-15T12:31:03.400901+00:00Microsoft releases BitNet 0.6B embedding model on Hugging Face https://korshunov.ai/ru/article/12161-microsoft-vypustila-model-vstraivaniia-bitnet-0-6b-na-hugging-face/korshunov.airu2026-07-15T12:31:03.400901+00:00Microsoft выпустила модель встраивания BitNet 0.6B на Hugging Face https://korshunov.ai/es/article/12161-microsoft-lanza-el-modelo-de-incrustacion-bitnet-0-6b-en-hugging-face/korshunov.aies2026-07-15T12:31:03.400901+00:00Microsoft lanza el modelo de incrustación BitNet 0.6B en Hugging Face https://korshunov.ai/zh-hans/article/12161-wei-ruan-zai-hugging-face-shang-fa-bu-liao-bitnet-0-6b-qian-ru-mo-xing/korshunov.aizh-cn2026-07-15T12:31:03.400901+00:00微软在 Hugging Face 上发布了 BitNet 0.6B 嵌入模型 https://korshunov.ai/pt-br/article/12161-microsoft-lanca-modelo-de-embedding-bitnet-0-6b-no-hugging-face/korshunov.aipt2026-07-15T12:31:03.400901+00:00Microsoft lança modelo de embedding BitNet 0.6B no Hugging Face https://korshunov.ai/hi/article/12161-maaikrosonphtt-ne-hugging-face-pr-bitnet-0-6b-embedding-monddl-jaarii-kiyaa/korshunov.aihi2026-07-15T12:31:03.400901+00:00माइक्रोसॉफ्ट ने Hugging Face पर BitNet 0.6B एम्बेडिंग मॉडल जारी किया https://korshunov.ai/ja/article/12161-microsoft-hugging-facenibitnet-0-6bmai-meip-mimoteruwogong-kai/korshunov.aija2026-07-15T12:31:03.400901+00:00Microsoft、Hugging FaceにBitNet 0.6B埋め込みモデルを公開 https://korshunov.ai/ko/article/12161-microsoft-hugging-facee-bitnet-0-6b-imbeding-model-culsi/korshunov.aiko2026-07-15T12:31:03.400901+00:00Microsoft, Hugging Face에 BitNet 0.6B 임베딩 모델 출시 https://korshunov.ai/fr/article/12161-microsoft-publie-le-modele-d-embedding-bitnet-0-6b-sur-hugging-face/korshunov.aifr2026-07-15T12:31:03.400901+00:00Microsoft publie le modèle d'embedding BitNet 0.6B sur Hugging Face https://korshunov.ai/id/article/12161-microsoft-merilis-model-embedding-bitnet-0-6b-di-hugging-face/korshunov.aiid2026-07-15T12:31:03.400901+00:00Microsoft merilis model embedding BitNet 0.6B di Hugging Face https://korshunov.ai/ar/article/12161-mykrwswft-ttlq-nmwdhj-ltdmyn-bitnet-0-6b-l-hwjyng-fys/korshunov.aiar2026-07-15T12:31:03.400901+00:00مايكروسوفت تطلق نموذج التضمين BitNet 0.6B على هوجينغ فيس https://korshunov.ai/en/article/12162-llama-cpp-b10017-raises-sycl-usm-minimum-buffer-size-to-4-gib/korshunov.aien2026-07-15T12:27:50.318927+00:00llama.cpp b10017 raises SYCL USM minimum buffer size to 4 GiB https://korshunov.ai/ko/article/12162-llama-cpp-b10017-sycl-usm-coeso-beopeo-keugireul-4-gibro-sanghyang-jojeong/korshunov.aiko2026-07-15T12:27:50.318927+00:00llama.cpp b10017, SYCL USM 최소 버퍼 크기를 4 GiB로 상향 조정 https://korshunov.ai/fr/article/12162-llama-cpp-b10017-augmente-la-taille-minimale-du-tampon-usm-sycl-a-4-gib/korshunov.aifr2026-07-15T12:27:50.318927+00:00llama.cpp b10017 augmente la taille minimale du tampon USM SYCL à 4 GiB https://korshunov.ai/id/article/12162-llama-cpp-b10017-menaikkan-ukuran-buffer-minimum-usm-sycl-menjadi-4-gib/korshunov.aiid2026-07-15T12:27:50.318927+00:00llama.cpp b10017 menaikkan ukuran buffer minimum USM SYCL menjadi 4 GiB https://korshunov.ai/ru/article/12162-llama-cpp-b10017-uvelichivaet-minimalnyi-razmer-bufera-sycl-usm-do-4-gib/korshunov.airu2026-07-15T12:27:50.318927+00:00llama.cpp b10017 увеличивает минимальный размер буфера SYCL USM до 4 ГиБ https://korshunov.ai/es/article/12162-llama-cpp-b10017-eleva-el-tamano-minimo-del-bufer-sycl-usm-a-4-gib/korshunov.aies2026-07-15T12:27:50.318927+00:00llama.cpp b10017 eleva el tamaño mínimo del búfer SYCL USM a 4 GiB https://korshunov.ai/zh-hans/article/12162-llama-cpp-b10017-jiang-sycl-usm-zui-xiao-huan-chong-qu-da-xiao-ti-sheng-zhi-4/korshunov.aizh-cn2026-07-15T12:27:50.318927+00:00llama.cpp b10017 将 SYCL USM 最小缓冲区大小提升至 4 GiB https://korshunov.ai/pt-br/article/12162-llama-cpp-b10017-aumenta-o-tamanho-minimo-do-buffer-sycl-usm-para-4-gib/korshunov.aipt2026-07-15T12:27:50.318927+00:00llama.cpp b10017 aumenta o tamanho mínimo do buffer SYCL USM para 4 GiB https://korshunov.ai/hi/article/12162-llama-cpp-b10017-ne-sycl-usm-nyuuntm-bphr-aakaar-ko-4-gib-tk-bddhaa-diyaa/korshunov.aihi2026-07-15T12:27:50.318927+00:00llama.cpp b10017 ने SYCL USM न्यूनतम बफ़र आकार को 4 GiB तक बढ़ा दिया https://korshunov.ai/ja/article/12162-llama-cpp-b10017kasycl-usmnozui-xiao-hatsuhuasaisuwo4-gibniyin-kishang-ke/korshunov.aija2026-07-15T12:27:50.318927+00:00llama.cpp b10017がSYCL USMの最小バッファサイズを4 GiBに引き上げ https://korshunov.ai/ar/article/12162-yrf-llama-cpp-b10017-lhd-ldn-lhjm-mkhzn-usm-l-sycl-l-4-gib/korshunov.aiar2026-07-15T12:27:50.318927+00:00يرفع llama.cpp b10017 الحد الأدنى لحجم مخزن USM لـ SYCL إلى 4 GiB https://korshunov.ai/hi/article/12071-maanv-llms-kii-tulnaa-men-adhik-khojii-arthpuurnn-nevigeshn-prdrshit-krte-hain/korshunov.aihi2026-07-15T12:19:12.336570+00:00मानव LLMs की तुलना में अधिक खोजी अर्थपूर्ण नेविगेशन प्रदर्शित करते हैं https://korshunov.ai/ja/article/12071-ren-jian-hallmyorimotan-suo-de-nayi-wei-nahike-shiyonwoshi-su/korshunov.aija2026-07-15T12:19:12.336570+00:00人間はLLMよりも探索的な意味ナビゲーションを示す https://korshunov.ai/ko/article/12071-inganeun-llmboda-tamsaegjeog-yimi-tamsaegeul-deo-manhi-boinda/korshunov.aiko2026-07-15T12:19:12.336570+00:00인간은 LLM보다 탐색적 의미 탐색을 더 많이 보인다 https://korshunov.ai/fr/article/12071-les-humains-presentent-une-navigation-semantique-plus-exploratoire-que-les-llm/korshunov.aifr2026-07-15T12:19:12.336570+00:00Les humains présentent une navigation sémantique plus exploratoire que les LLM https://korshunov.ai/id/article/12071-manusia-menunjukkan-navigasi-semantik-yang-lebih-eksploratif-daripada-llm/korshunov.aiid2026-07-15T12:19:12.336570+00:00Manusia menunjukkan navigasi semantik yang lebih eksploratif daripada LLM https://korshunov.ai/en/article/12071-humans-exhibit-more-exploratory-semantic-navigation-than-llms/korshunov.aien2026-07-15T12:19:12.336570+00:00Humans exhibit more exploratory semantic navigation than LLMs https://korshunov.ai/ru/article/12071-liudi-demonstriruiut-bolee-issledovatelskuiu-semanticheskuiu-navigatsiiu-chem/korshunov.airu2026-07-15T12:19:12.336570+00:00Люди демонстрируют более исследовательскую семантическую навигацию, чем LLM https://korshunov.ai/es/article/12071-los-humanos-exhiben-una-navegacion-semantica-mas-exploratoria-que-los-llm/korshunov.aies2026-07-15T12:19:12.336570+00:00Los humanos exhiben una navegación semántica más exploratoria que los LLM https://korshunov.ai/zh-hans/article/12071-ren-lei-biao-xian-chu-bi-da-yu-yan-mo-xing-geng-ju-tan-suo-xing-de-yu-yi-dao/korshunov.aizh-cn2026-07-15T12:19:12.336570+00:00人类表现出比大语言模型更具探索性的语义导航 https://korshunov.ai/pt-br/article/12071-humanos-exibem-navegacao-semantica-mais-exploratoria-do-que-llms/korshunov.aipt2026-07-15T12:19:12.336570+00:00Humanos exibem navegação semântica mais exploratória do que LLMs https://korshunov.ai/ar/article/12071-yuzhr-lbshr-tnqlan-dllyan-stkshfyan-kthr-mn-nmdhj-llgt-lkbyr-llms/korshunov.aiar2026-07-15T12:19:12.336570+00:00يُظهر البشر تنقلاً دلاليًا استكشافيًا أكثر من نماذج اللغات الكبيرة (LLMs) https://korshunov.ai/es/article/12070-un-marco-multiagente-afinado-detecta-rasgos-ocean-en-narrativas-vitales/korshunov.aies2026-07-15T12:11:40.929353+00:00Un marco multiagente afinado detecta rasgos OCEAN en narrativas vitales https://korshunov.ai/zh-hans/article/12070-wei-diao-duo-zhi-neng-ti-kuang-jia-jian-ce-sheng-huo-xu-shi-zhong-de-oceante-zhi/korshunov.aizh-cn2026-07-15T12:11:40.929353+00:00微调多智能体框架检测生活叙事中的OCEAN特质 https://korshunov.ai/pt-br/article/12070-framework-multiagente-ajustado-detecta-tracos-ocean-em-narrativas-de-vida/korshunov.aipt2026-07-15T12:11:40.929353+00:00Framework multiagente ajustado detecta traços OCEAN em narrativas de vida https://korshunov.ai/hi/article/12070-phaain-ttyuundd-mlttii-ejentt-phremvrk-jiivn-kthaaon-men-ocean-lkssnnon-kaa-ptaa/korshunov.aihi2026-07-15T12:11:40.929353+00:00फाइन-ट्यूंड मल्टी-एजेंट फ्रेमवर्क जीवन कथाओं में OCEAN लक्षणों का पता लगाता है https://korshunov.ai/en/article/12070-fine-tuned-multi-agent-framework-detects-ocean-traits-in-life-narratives/korshunov.aien2026-07-15T12:11:40.929353+00:00Fine-tuned multi-agent framework detects OCEAN traits in life narratives https://korshunov.ai/ru/article/12070-doobuchennaia-mnogoagentnaia-sistema-vyiavliaet-cherty-ocean-v-zhiznennykh/korshunov.airu2026-07-15T12:11:40.929353+00:00Дообученная многоагентная система выявляет черты OCEAN в жизненных нарративах https://korshunov.ai/ja/article/12070-huainchiyu-ninkusaretamaruchie-shientohure-muwa-kukaren-sheng-nowu-yu-xing-chu/korshunov.aija2026-07-15T12:11:40.929353+00:00ファインチューニングされたマルチエージェントフレームワークが人生の物語からOCEAN特性を検出 https://korshunov.ai/ko/article/12070-paintyuningdoen-meolti-eijeonteu-peureimweokeuga-insaeng-seosaeseo-ocean-gamji/korshunov.aiko2026-07-15T12:11:40.929353+00:00파인튜닝된 멀티 에이전트 프레임워크가 인생 서사에서 OCEAN 특성 감지 https://korshunov.ai/fr/article/12070-un-cadre-multi-agents-affine-detecte-les-traits-ocean-dans-les-recits-de-vie/korshunov.aifr2026-07-15T12:11:40.929353+00:00Un cadre multi-agents affiné détecte les traits OCEAN dans les récits de vie https://korshunov.ai/id/article/12070-kerangka-kerja-multi-agen-yang-dituning-halus-mendeteksi-sifat-ocean-dalam/korshunov.aiid2026-07-15T12:11:40.929353+00:00Kerangka kerja multi-agen yang dituning halus mendeteksi sifat OCEAN dalam narasi kehidupan https://korshunov.ai/ar/article/12070-tr-ml-mt-dd-lwkl-mudqwq-ykshf-smt-ocean-fy-srdyt-lhy/korshunov.aiar2026-07-15T12:11:40.929353+00:00إطار عمل متعدد الوكلاء مُدقّق يكشف سمات OCEAN في سرديات الحياة https://korshunov.ai/pt-br/article/12069-rcwt-mede-o-deslocamento-de-orcamento-da-tarefa-a-partir-do-conteudo-de-em-llm/korshunov.aipt2026-07-15T12:04:16.382242+00:00RCWT mede o deslocamento de orçamento da tarefa a partir do conteúdo de coordenação em chamadas LLM https://korshunov.ai/hi/article/12069-rcwt-llm-konls-men-smnvy-saamgrii-se-kaary-bjtt-visthaapn-ko-maaptaa-hai/korshunov.aihi2026-07-15T12:04:16.382242+00:00RCWT LLM कॉल्स में समन्वय सामग्री से कार्य-बजट विस्थापन को मापता है https://korshunov.ai/en/article/12069-rcwt-measures-task-budget-displacement-from-coordination-content-in-llm-calls/korshunov.aien2026-07-15T12:04:16.382242+00:00RCWT measures task-budget displacement from coordination content in LLM calls https://korshunov.ai/ru/article/12069-rcwt-izmeriaet-smeshchenie-biudzheta-zadachi-iz-za-koordinatsionnogo-kontenta-v/korshunov.airu2026-07-15T12:04:16.382242+00:00RCWT измеряет смещение бюджета задачи из-за координационного контента в вызовах LLM https://korshunov.ai/es/article/12069-rcwt-mide-el-desplazamiento-de-la-tarea-presupuesto-por-contenido-de-en-llamadas/korshunov.aies2026-07-15T12:04:16.382242+00:00RCWT mide el desplazamiento de la tarea-presupuesto por contenido de coordinación en llamadas LLM https://korshunov.ai/zh-hans/article/12069-rcwt-heng-liang-duo-zhi-neng-ti-he-zeng-qiang-ji-yi-llm-diao-yong-zhong-xie-diao/korshunov.aizh-cn2026-07-15T12:04:16.382242+00:00RCWT 衡量多智能体和增强记忆 LLM 调用中协调内容导致的任务-预算位移 https://korshunov.ai/ja/article/12069-rcwthallmhu-hichu-shiniokerudiao-zheng-kontentsukarasheng-shirutasukuyu-suan-tuo/korshunov.aija2026-07-15T12:04:16.382242+00:00RCWTはLLM呼び出しにおける調整コンテンツから生じるタスク予算の逸脱を測定する https://korshunov.ai/ko/article/12069-rcwtneun-llm-hoculeseo-jojeong-kontenceue-gibanhan-jageob-yesan-idongseongeul/korshunov.aiko2026-07-15T12:04:16.382242+00:00RCWT는 LLM 호출에서 조정 콘텐츠에 기반한 작업-예산 이동성을 측정합니다 https://korshunov.ai/fr/article/12069-le-rcwt-mesure-le-deplacement-du-budget-tache-a-partir-du-contenu-de-dans-les/korshunov.aifr2026-07-15T12:04:16.382242+00:00Le RCWT mesure le déplacement du budget-tâche à partir du contenu de coordination dans les appels aux LLM https://korshunov.ai/id/article/12069-rcwt-mengukur-pergeseran-tugas-anggaran-dari-konten-koordinasi-dalam-panggilan/korshunov.aiid2026-07-15T12:04:16.382242+00:00RCWT mengukur pergeseran tugas-anggaran dari konten koordinasi dalam panggilan LLM https://korshunov.ai/ar/article/12069-yqys-rcwt-zh-myzny-lmhm-mn-mhtw-ltnsyq-fy-std-t-nmdhj-llgt-lkbyr/korshunov.aiar2026-07-15T12:04:16.382242+00:00يقيس RCWT إزاحة ميزانية المهمة من محتوى التنسيق في استدعاءات نماذج اللغات الكبيرة https://korshunov.ai/zh-hans/article/12068-fin-analystzai-finmmeval-2026ren-wu-3zhong-yi-13-51-de-tslahui-bao-lu-wei-lie-di/korshunov.aizh-cn2026-07-15T11:45:45.612199+00:00Fin-Analyst在FinMMEval 2026任务3中以+13.51%的TSLA回报率位列第一 https://korshunov.ai/en/article/12068-fin-analyst-ranks-first-in-finmmeval-2026-task-3-with-13-51-tsla-return/korshunov.aien2026-07-15T11:45:45.612199+00:00Fin-Analyst ranks first in FinMMEval 2026 Task 3 with +13.51% TSLA return https://korshunov.ai/ru/article/12068-fin-analyst-zanial-pervoe-mesto-v-finmmeval-2026-task-3-s-dokhodnostiu-tsla-13/korshunov.airu2026-07-15T11:45:45.612199+00:00Fin-Analyst занял первое место в FinMMEval 2026 Task 3 с доходностью TSLA +13.51% https://korshunov.ai/es/article/12068-fin-analyst-ocupa-el-primer-lugar-en-finmmeval-2026-tarea-3-con-un-rendimiento/korshunov.aies2026-07-15T11:45:45.612199+00:00Fin-Analyst ocupa el primer lugar en FinMMEval 2026 Tarea 3 con un rendimiento de TSLA del +13.51% https://korshunov.ai/pt-br/article/12068-fin-analyst-ocupa-o-primeiro-lugar-no-finmmeval-2026-tarefa-3-com-retorno-de-13/korshunov.aipt2026-07-15T11:45:45.612199+00:00Fin-Analyst ocupa o primeiro lugar no FinMMEval 2026 Tarefa 3 com retorno de +13.51% na TSLA https://korshunov.ai/hi/article/12068-fin-analyst-ne-13-51-tsla-rittrn-ke-saath-finmmeval-2026-ttaask-3-men-phle-pr/korshunov.aihi2026-07-15T11:45:45.612199+00:00Fin-Analyst ने +13.51% TSLA रिटर्न के साथ FinMMEval 2026 टास्क 3 में पहले स्थान पर है https://korshunov.ai/ja/article/12068-fin-analystkafinmmeval-2026-tasuku3tetslarita-n-13-51-teshou-wei/korshunov.aija2026-07-15T11:45:45.612199+00:00Fin-AnalystがFinMMEval 2026 タスク3でTSLAリターン+13.51%で首位 https://korshunov.ai/ko/article/12068-fin-analyst-finmmeval-2026-jageob3eseo-tsla-suigryul-13-51-ro-1wi/korshunov.aiko2026-07-15T11:45:45.612199+00:00Fin-Analyst, FinMMEval 2026 작업3에서 TSLA 수익률 +13.51%로 1위 https://korshunov.ai/fr/article/12068-fin-analyst-premier-dans-finmmeval-2026-tache-3-avec-un-rendement-tsla-de-1351/korshunov.aifr2026-07-15T11:45:45.612199+00:00Fin-Analyst premier dans FinMMEval 2026 Tâche 3 avec un rendement TSLA de +13,51 % https://korshunov.ai/id/article/12068-fin-analyst-peringkat-pertama-di-finmmeval-2026-tugas-3-dengan-return-tsla-1351/korshunov.aiid2026-07-15T11:45:45.612199+00:00Fin-Analyst peringkat pertama di FinMMEval 2026 Tugas 3 dengan return TSLA +13,51% https://korshunov.ai/ar/article/12068-fin-analyst-yhtl-lmrtb-lwl-fy-finmmeval-2026-lmhm-3-b-yd-tsla-bnsb-13-51/korshunov.aiar2026-07-15T11:45:45.612199+00:00Fin-Analyst يحتل المرتبة الأولى في FinMMEval 2026 المهمة 3 بعائد TSLA بنسبة +13.51% https://korshunov.ai/ru/article/12067-spekuliatsiia-s-pamiatiu-bezpoternoe-uskorenie-dlia-agentov-llm/korshunov.airu2026-07-15T11:37:24.915461+00:00Спекуляция с памятью: безпотерьное ускорение для агентов LLM https://korshunov.ai/en/article/12067-speculate-with-memory-lossless-acceleration-for-llm-agents/korshunov.aien2026-07-15T11:37:24.915461+00:00Speculate with Memory: Lossless Acceleration for LLM Agents https://korshunov.ai/es/article/12067-especulacion-con-memoria-aceleracion-sin-perdidas-para-agentes-llm/korshunov.aies2026-07-15T11:37:24.915461+00:00Especulación con memoria: aceleración sin pérdidas para agentes LLM https://korshunov.ai/zh-hans/article/12067-ji-yu-ji-yi-de-tui-ce-llmdai-li-de-wu-sun-jia-su/korshunov.aizh-cn2026-07-15T11:37:24.915461+00:00基于记忆的推测：LLM代理的无损加速 https://korshunov.ai/pt-br/article/12067-especulacao-com-memoria-aceleracao-sem-perdas-para-agentes-llm/korshunov.aipt2026-07-15T11:37:24.915461+00:00Especulação com Memória: Aceleração sem Perdas para Agentes LLM https://korshunov.ai/hi/article/12067-memorii-ke-saath-anumaan-llm-ejentts-ke-lie-haanirhit-tvrnn/korshunov.aihi2026-07-15T11:37:24.915461+00:00मेमोरी के साथ अनुमान: LLM एजेंट्स के लिए हानिरहित त्वरण https://korshunov.ai/ja/article/12067-memoritetui-ce-llme-shientonowu-sun-shi-jia-su/korshunov.aija2026-07-15T11:37:24.915461+00:00メモリで推測：LLMエージェントの無損失加速 https://korshunov.ai/ko/article/12067-memoriro-cuceug-llm-eijeonteuyi-musonsil-gasog/korshunov.aiko2026-07-15T11:37:24.915461+00:00메모리로 추측: LLM 에이전트의 무손실 가속 https://korshunov.ai/fr/article/12067-speculer-avec-la-memoire-acceleration-sans-perte-pour-les-agents-llm/korshunov.aifr2026-07-15T11:37:24.915461+00:00Spéculer avec la mémoire : Accélération sans perte pour les agents LLM https://korshunov.ai/id/article/12067-berspekulasi-dengan-memori-akselerasi-tanpa-kehilangan-untuk-agen-llm/korshunov.aiid2026-07-15T11:37:24.915461+00:00Berspekulasi dengan Memori: Akselerasi Tanpa Kehilangan untuk Agen LLM https://korshunov.ai/ar/article/12067-ltkhn-bldhkr-tsry-bdwn-fqdn-lwkl-llm/korshunov.aiar2026-07-15T11:37:24.915461+00:00التكهن بالذاكرة: تسريع بدون فقدان لوكلاء LLM https://korshunov.ai/en/article/12066-finresearchbench-ii-introduces-consensus-derived-gold-rubrics-to-automate-report/korshunov.aien2026-07-15T11:29:09.213571+00:00FinResearchBench II introduces consensus-derived gold rubrics to automate financial report quality evaluation https://korshunov.ai/ru/article/12066-finresearchbench-ii-vvodit-konsensusnye-zolotye-rubriki-dlia-avtomatizatsii/korshunov.airu2026-07-15T11:29:09.213571+00:00FinResearchBench II вводит консенсусные золотые рубрики для автоматизации оценки качества финансовых отчетов https://korshunov.ai/es/article/12066-finresearchbench-ii-introduce-rubricas-doradas-derivadas-de-consenso-para-la-de/korshunov.aies2026-07-15T11:29:09.213571+00:00FinResearchBench II introduce rúbricas doradas derivadas de consenso para automatizar la evaluación de la calidad de informes financieros https://korshunov.ai/zh-hans/article/12066-finresearchbench-ii-yin-ru-gong-shi-pai-sheng-huang-jin-ping-fen-biao-zhun-yi-zi/korshunov.aizh-cn2026-07-15T11:29:09.213571+00:00FinResearchBench II 引入共识派生黄金评分标准，以自动化金融报告质量评估 https://korshunov.ai/pt-br/article/12066-finresearchbench-ii-introduz-rubricas-douradas-derivadas-de-consenso-para-a-da/korshunov.aipt2026-07-15T11:29:09.213571+00:00FinResearchBench II introduz rubricas douradas derivadas de consenso para automatizar a avaliação da qualidade de relatórios financeiros https://korshunov.ai/hi/article/12066-finresearchbench-ii-vittiiy-riportt-gunnvttaa-muulyaankn-ko-svcaalit-krne-ke-lie/korshunov.aihi2026-07-15T11:29:09.213571+00:00FinResearchBench II वित्तीय रिपोर्ट गुणवत्ता मूल्यांकन को स्वचालित करने के लिए सहमति-निष्पन्न गोल्ड रूब्रिक्स पेश करता है https://korshunov.ai/ja/article/12066-finresearchbench-ii-kakonsensasuyou-lai-noko-rutoji-zhun-wodao-ru-shi-jin-rong/korshunov.aija2026-07-15T11:29:09.213571+00:00FinResearchBench II がコンセンサス由来のゴールド基準を導入し、金融レポートの品質評価を自動化 https://korshunov.ai/ko/article/12066-finresearchbench-ii-ga-keonsenseoseu-giban-goldeu-gijuneul-doibhayeo-geumyung/korshunov.aiko2026-07-15T11:29:09.213571+00:00FinResearchBench II 가 컨센서스 기반 골드 기준을 도입하여 금융 보고서 품질 평가를 자동화하다 https://korshunov.ai/fr/article/12066-finresearchbench-ii-introduit-des-rubriques-d-or-derivees-du-consensus-pour-l-de/korshunov.aifr2026-07-15T11:29:09.213571+00:00FinResearchBench II introduit des rubriques d'or dérivées du consensus pour automatiser l'évaluation de la qualité des rapports financiers https://korshunov.ai/id/article/12066-finresearchbench-ii-memperkenalkan-rubrik-emas-berbasis-konsensus-untuk-evaluasi/korshunov.aiid2026-07-15T11:29:09.213571+00:00FinResearchBench II memperkenalkan rubrik emas berbasis konsensus untuk otomatisasi evaluasi kualitas laporan keuangan https://korshunov.ai/ar/article/12066-finresearchbench-ii-yqdm-m-yyr-dhhby-mstmd-mn-ljm-ltmt-tqyym-jwd-ltqryr-lmly/korshunov.aiar2026-07-15T11:29:09.213571+00:00FinResearchBench II يقدم معايير ذهبية مستمدة من الإجماع لأتمتة تقييم جودة التقارير المالية https://korshunov.ai/ru/article/12065-lokalnyi-issledovatelskii-agent-na-4b-ekspozitsiia-ogranichivaet-dostovernost/korshunov.airu2026-07-15T11:20:40.453696+00:00Локальный исследовательский агент на 4B: экспозиция ограничивает достоверность, поиск ограничивает охват https://korshunov.ai/zh-hans/article/12065-duan-ce-4byan-jiu-dai-li-pu-guang-liang-xian-zhi-zhong-shi-du-jian-suo-liang-zhi/korshunov.aizh-cn2026-07-15T11:20:40.453696+00:00端侧4B研究代理：曝光量限制忠实度，检索量限制覆盖率 https://korshunov.ai/en/article/12065-on-device-4b-research-agent-exposure-bounds-faithfulness-retrieval-bounds/korshunov.aien2026-07-15T11:20:40.453696+00:00On-device 4B research agent: exposure bounds faithfulness, retrieval bounds coverage https://korshunov.ai/es/article/12065-agente-de-investigacion-de-4b-en-el-dispositivo-la-exposicion-limita-la-la-la/korshunov.aies2026-07-15T11:20:40.453696+00:00Agente de investigación de 4B en el dispositivo: la exposición limita la fidelidad, la recuperación limita la cobertura https://korshunov.ai/pt-br/article/12065-agente-de-pesquisa-de-4b-no-dispositivo-a-exposicao-limita-a-fidelidade-a-limita/korshunov.aipt2026-07-15T11:20:40.453696+00:00Agente de pesquisa de 4B no dispositivo: a exposição limita a fidelidade, a recuperação limita a cobertura https://korshunov.ai/hi/article/12065-ddivaais-saaidd-4b-risrc-ejentt-ekspojr-vishvsniiytaa-ko-siimit-krtaa-hai-kvrej/korshunov.aihi2026-07-15T11:20:40.453696+00:00डिवाइस-साइड 4B रिसर्च एजेंट: एक्सपोज़र विश्वसनीयता को सीमित करता है, रिकवरी कवरेज को सीमित करता है https://korshunov.ai/ja/article/12065-ontehaisu4byan-jiu-e-shiento-pu-lu-kazhong-shi-du-wozhi-yue-shi-jian-suo-kawang/korshunov.aija2026-07-15T11:20:40.453696+00:00オンデバイス4B研究エージェント：曝露が忠実度を制約し、検索が網羅性を制約する https://korshunov.ai/ko/article/12065-ondibaiseu-4b-yeongu-eijeonteu-noculi-cungsildoreul-jehanhago-geomsaegi-jehanham/korshunov.aiko2026-07-15T11:20:40.453696+00:00온디바이스 4B 연구 에이전트: 노출이 충실도를 제한하고 검색이 커버리지를 제한함 https://korshunov.ai/fr/article/12065-agent-de-recherche-4b-sur-appareil-l-exposition-lie-la-fidelite-la-recuperation/korshunov.aifr2026-07-15T11:20:40.453696+00:00Agent de recherche 4B sur appareil : l'exposition lie la fidélité, la récupération lie la couverture https://korshunov.ai/id/article/12065-agen-riset-4b-di-perangkat-eksposur-membatasi-kesetiaan-pengambilan-membatasi/korshunov.aiid2026-07-15T11:20:40.453696+00:00Agen riset 4B di perangkat: eksposur membatasi kesetiaan, pengambilan membatasi cakupan https://korshunov.ai/ar/article/12065-wkyl-bhth-4b-l-ljhz-lt-rd-yhdd-ldq-wlstrj-yhdd-ltgty/korshunov.aiar2026-07-15T11:20:40.453696+00:00وكيل بحث 4B على الجهاز: التعرض يحدد الدقة، والاسترجاع يحدد التغطية https://korshunov.ai/es/article/12064-code-mue-mide-la-incertidumbre-de-los-llm-de-codigo-mediante-grafos-semanticos/korshunov.aies2026-07-15T11:12:37.770068+00:00Code-MUE mide la incertidumbre de los LLM de código mediante grafos semánticos basados en ejecución https://korshunov.ai/zh-hans/article/12064-code-mue-tong-guo-ji-yu-zhi-xing-de-yu-yi-tu-heng-liang-dai-ma-llm-de-bu-que/korshunov.aizh-cn2026-07-15T11:12:37.770068+00:00Code-MUE 通过基于执行的语义图衡量代码 LLM 的不确定性 https://korshunov.ai/pt-br/article/12064-code-mue-mede-a-incerteza-de-llms-de-codigo-por-meio-de-grafos-semanticos-em/korshunov.aipt2026-07-15T11:12:37.770068+00:00Code-MUE mede a incerteza de LLMs de código por meio de grafos semânticos baseados em execução https://korshunov.ai/hi/article/12064-code-mue-nisspaadn-aadhaarit-arthpuurnn-graaph-ke-maadhym-se-kodd-llm-kii-ko-hai/korshunov.aihi2026-07-15T11:12:37.770068+00:00Code-MUE निष्पादन-आधारित अर्थपूर्ण ग्राफ़ के माध्यम से कोड LLM की अनिश्चितता को मापता है https://korshunov.ai/ru/article/12064-code-mue-izmeriaet-neopredelennost-kodovykh-llm-cherez-semanticheskie-grafy-na/korshunov.airu2026-07-15T11:12:37.770068+00:00Code-MUE измеряет неопределённость кодовых LLM через семантические графы на основе выполнения https://korshunov.ai/en/article/12064-code-mue-measures-code-llm-uncertainty-via-execution-based-semantic-graphs/korshunov.aien2026-07-15T11:12:37.770068+00:00Code-MUE measures code LLM uncertainty via execution-based semantic graphs https://korshunov.ai/ja/article/12064-code-muehashi-xing-he-sunosemanteitsukukurahuwotong-shiteko-tollmnobu-que-shi/korshunov.aija2026-07-15T11:12:37.770068+00:00Code-MUEは実行ベースのセマンティックグラフを通じてコードLLMの不確実性を測定する https://korshunov.ai/ko/article/12064-code-mueneun-silhaeng-giban-yimi-geuraepeureul-tonghae-kodeu-llmyi/korshunov.aiko2026-07-15T11:12:37.770068+00:00Code-MUE는 실행 기반 의미 그래프를 통해 코드 LLM의 불확실성을 측정합니다 https://korshunov.ai/fr/article/12064-code-mue-mesure-l-incertitude-des-llm-de-code-via-des-graphes-semantiques-bases/korshunov.aifr2026-07-15T11:12:37.770068+00:00Code-MUE mesure l'incertitude des LLM de code via des graphes sémantiques basés sur l'exécution https://korshunov.ai/id/article/12064-code-mue-mengukur-ketidakpastian-llm-kode-melalui-grafik-semantik-berbasis/korshunov.aiid2026-07-15T11:12:37.770068+00:00Code-MUE mengukur ketidakpastian LLM kode melalui grafik semantik berbasis eksekusi https://korshunov.ai/ar/article/12064-yqys-code-mue-dm-lyqyn-fy-nmdhj-llgt-lkbyr-llkwd-br-rswm-byny-dlly-qym-l-ltnfydh/korshunov.aiar2026-07-15T11:12:37.770068+00:00يقيس Code-MUE عدم اليقين في نماذج اللغات الكبيرة للكود عبر رسوم بيانية دلالية قائمة على التنفيذ https://korshunov.ai/en/article/12160-user-creates-html-physics-shooter-using-qwen-3-6-27b-and-mtp/korshunov.aien2026-07-15T11:03:22.265086+00:00User creates HTML physics shooter using Qwen 3.6 27B and MTP https://korshunov.ai/ru/article/12160-polzovatel-sozdal-html-shuter-s-fizikoi-ispolzuia-qwen-3-6-27b-i-mtp/korshunov.airu2026-07-15T11:03:22.265086+00:00Пользователь создал HTML-шутер с физикой, используя Qwen 3.6 27B и MTP https://korshunov.ai/es/article/12160-un-usuario-crea-un-shooter-de-fisica-html-usando-qwen-3-6-27b-y-mtp/korshunov.aies2026-07-15T11:03:22.265086+00:00Un usuario crea un shooter de física HTML usando Qwen 3.6 27B y MTP https://korshunov.ai/zh-hans/article/12160-yong-hu-shi-yong-qwen-3-6-27b-he-mtp-chuang-jian-html-wu-li-she-ji-you-xi/korshunov.aizh-cn2026-07-15T11:03:22.265086+00:00用户使用 Qwen 3.6 27B 和 MTP 创建 HTML 物理射击游戏 https://korshunov.ai/pt-br/article/12160-usuario-cria-shooter-de-fisica-em-html-usando-qwen-3-6-27b-e-mtp/korshunov.aipt2026-07-15T11:03:22.265086+00:00Usuário cria shooter de física em HTML usando Qwen 3.6 27B e MTP https://korshunov.ai/hi/article/12160-upyogkrtaa-ne-qwen-3-6-27b-aur-mtp-kaa-upyog-krke-html-bhautikii-shuuttr-bnaayaa/korshunov.aihi2026-07-15T11:03:22.265086+00:00उपयोगकर्ता ने Qwen 3.6 27B और MTP का उपयोग करके HTML भौतिकी शूटर बनाया https://korshunov.ai/ja/article/12160-yu-sa-kaqwen-3-6-27btomtpwoshi-yong-shitehtmlwu-li-shiyu-ta-wozuo-cheng/korshunov.aija2026-07-15T11:03:22.265086+00:00ユーザーがQwen 3.6-27BとMTPを使用してHTML物理シューターを作成 https://korshunov.ai/ko/article/12160-sayongjaga-qwen-3-6-27b-mic-mtpreul-sayonghayeo-html-mulri-syuteo-geim/korshunov.aiko2026-07-15T11:03:22.265086+00:00사용자가 Qwen 3.6-27B 및 MTP를 사용하여 HTML 물리 슈터 게임 생성 https://korshunov.ai/fr/article/12160-un-utilisateur-cree-un-jeu-de-tir-physique-html-en-utilisant-qwen-3-6-27b-et-mtp/korshunov.aifr2026-07-15T11:03:22.265086+00:00Un utilisateur crée un jeu de tir physique HTML en utilisant Qwen 3.6-27B et MTP https://korshunov.ai/id/article/12160-pengguna-membuat-penembak-fisika-html-menggunakan-qwen-3-6-27b-dan-mtp/korshunov.aiid2026-07-15T11:03:22.265086+00:00Pengguna membuat penembak fisika HTML menggunakan Qwen 3.6-27B dan MTP https://korshunov.ai/ar/article/12160-mstkhdm-ynshy-l-b-tlq-nr-fyzyyy-bstkhdm-html-m-qwen-3-6-27b-wmtp/korshunov.aiar2026-07-15T11:03:22.265086+00:00مستخدم ينشئ لعبة إطلاق نار فيزيائية باستخدام HTML مع Qwen 3.6-27B وMTP https://korshunov.ai/ar/article/12159-tyntsin-ttlq-hy-embodied-rxbrain-1-0-nmwdhj-ssy-mt-dd-lwsyt-mwhd-lldrk-ljsdy/korshunov.aiar2026-07-15T10:57:21.701572+00:00تيانتسِن تطلق Hy-Embodied-RxBrain-1.0، نموذج أساسي متعدد الوسائط موحد للإدراك الجسدي https://korshunov.ai/en/article/12159-tencent-releases-hy-embodied-rxbrain-1-0-a-unified-multimodal-foundation-model/korshunov.aien2026-07-15T10:57:21.701572+00:00Tencent releases Hy-Embodied-RxBrain-1.0, a unified multimodal foundation model for embodied cognition https://korshunov.ai/ru/article/12159-tencent-vypustila-hy-embodied-rxbrain-1-0-edinuiu-multimodalnuiu-bazovuiu-model/korshunov.airu2026-07-15T10:57:21.701572+00:00Tencent выпустила Hy-Embodied-RxBrain-1.0, единую мультимодальную базовую модель для воплощённого познания https://korshunov.ai/es/article/12159-tencent-lanza-hy-embodied-rxbrain-1-0-un-modelo-base-multimodal-unificado-para/korshunov.aies2026-07-15T10:57:21.701572+00:00Tencent lanza Hy-Embodied-RxBrain-1.0, un modelo base multimodal unificado para la cognición encarnada https://korshunov.ai/zh-hans/article/12159-teng-xun-fa-bu-hy-embodied-rxbrain-1-0-yi-chong-yong-yu-ju-shen-ren-zhi-de-tong/korshunov.aizh-cn2026-07-15T10:57:21.701572+00:00腾讯发布 Hy-Embodied-RxBrain-1.0，一种用于具身认知的统一多模态基础模型 https://korshunov.ai/pt-br/article/12159-tencent-lanca-hy-embodied-rxbrain-1-0-um-modelo-base-multimodal-unificado-para/korshunov.aipt2026-07-15T10:57:21.701572+00:00Tencent lança Hy-Embodied-RxBrain-1.0, um modelo base multimodal unificado para cognição encarnada https://korshunov.ai/hi/article/12159-tencent-ne-hy-embodied-rxbrain-1-0-jaarii-kiyaa-jo-embonddedd-kongnishn-ke-lie/korshunov.aihi2026-07-15T10:57:21.701572+00:00Tencent ने Hy-Embodied-RxBrain-1.0 जारी किया, जो एम्बॉडेड कॉग्निशन के लिए एक एकीकृत मल्टीमोडल फाउंडेशन मॉडल है https://korshunov.ai/ja/article/12159-tencentkaju-xian-hua-ren-zhi-notamenotong-yi-maruchimo-taruji-pan-moteru-hy-1-0/korshunov.aija2026-07-15T10:57:21.701572+00:00Tencentが具現化認知のための統一マルチモーダル基盤モデル「Hy-Embodied-RxBrain-1.0」をリリース https://korshunov.ai/ko/article/12159-tencent-sincejeog-injireul-wihan-tonghab-meoltimodal-paundeisyeon-model-hy-1-0/korshunov.aiko2026-07-15T10:57:21.701572+00:00Tencent, 신체적 인지를 위한 통합 멀티모달 파운데이션 모델 Hy-Embodied-RxBrain-1.0 출시 https://korshunov.ai/fr/article/12159-tencent-publie-hy-embodied-rxbrain-1-0-un-modele-fondamental-multimodal-unifie/korshunov.aifr2026-07-15T10:57:21.701572+00:00Tencent publie Hy-Embodied-RxBrain-1.0, un modèle fondamental multimodal unifié pour la cognition incarnée https://korshunov.ai/id/article/12159-tencent-merilis-hy-embodied-rxbrain-1-0-model-dasar-multimodal-terpadu-untuk/korshunov.aiid2026-07-15T10:57:21.701572+00:00Tencent merilis Hy-Embodied-RxBrain-1.0, model dasar multimodal terpadu untuk kognisi berwujud https://korshunov.ai/en/article/12063-llama-3-1-70b-instruct-uses-shared-countdown-subcircuit-for-token-counting/korshunov.aien2026-07-15T10:49:23.813293+00:00Llama-3.1-70B-Instruct uses shared countdown subcircuit for token counting https://korshunov.ai/fr/article/12063-llama-3-1-70b-instruct-utilise-un-sous-circuit-de-decompte-partage-pour-le-des/korshunov.aifr2026-07-15T10:49:23.813293+00:00Llama-3.1-70B-Instruct utilise un sous-circuit de décompte partagé pour le comptage des tokens https://korshunov.ai/id/article/12063-llama-3-1-70b-instruct-menggunakan-sub-sirkuit-hitung-mundur-bersama-untuk-token/korshunov.aiid2026-07-15T10:49:23.813293+00:00Llama-3.1-70B-Instruct menggunakan sub-sirkuit hitung mundur bersama untuk penghitungan token https://korshunov.ai/ar/article/12063-ystkhdm-llama-3-1-70b-instruct-dyr-fr-y-mshtrk-ll-d-ltnzly-l-d-lrmwz/korshunov.aiar2026-07-15T10:49:23.813293+00:00يستخدم Llama-3.1-70B-Instruct دائرة فرعية مشتركة للعد التنازلي لعد الرموز https://korshunov.ai/zh-hans/article/12063-llama-3-1-70b-instruct-shi-yong-gong-xiang-dao-ji-shi-zi-dian-lu-jin-xing-ling/korshunov.aizh-cn2026-07-15T10:49:23.813293+00:00Llama-3.1-70B-Instruct 使用共享倒计时子电路进行令牌计数 https://korshunov.ai/ru/article/12063-llama-3-1-70b-instruct-ispolzuet-obshchuiu-podtsep-obratnogo-otscheta-dlia/korshunov.airu2026-07-15T10:49:23.813293+00:00Llama-3.1-70B-Instruct использует общую подцепь обратного отсчёта для подсчёта токенов https://korshunov.ai/es/article/12063-llama-3-1-70b-instruct-utiliza-un-subcircuito-de-cuenta-regresiva-compartido-el/korshunov.aies2026-07-15T10:49:23.813293+00:00Llama-3.1-70B-Instruct utiliza un subcircuito de cuenta regresiva compartido para el conteo de tokens https://korshunov.ai/pt-br/article/12063-llama-3-1-70b-instruct-usa-subcircuito-de-contagem-regressiva-compartilhado-para/korshunov.aipt2026-07-15T10:49:23.813293+00:00Llama-3.1-70B-Instruct usa subcircuito de contagem regressiva compartilhado para contagem de tokens https://korshunov.ai/hi/article/12063-llama-3-1-70b-instruct-ttokn-gintii-ke-lie-saajhaa-kaaunttddaaun-sbsrkitt-kaa/korshunov.aihi2026-07-15T10:49:23.813293+00:00Llama-3.1-70B-Instruct टोकन गिनती के लिए साझा काउंटडाउन सबसर्किट का उपयोग करता है https://korshunov.ai/ja/article/12063-llama-3-1-70b-instructhato-kunshu-eshang-kenigong-you-nokauntotaunbu-fen-hui-lu/korshunov.aija2026-07-15T10:49:23.813293+00:00Llama-3.1-70B-Instructはトークン数え上げに共有のカウントダウン部分回路を使用 https://korshunov.ai/ko/article/12063-llama-3-1-70b-instructneun-tokeun-kauntingeul-wihae-gongyu-kaunteudaun-sayong/korshunov.aiko2026-07-15T10:49:23.813293+00:00Llama-3.1-70B-Instruct는 토큰 카운팅을 위해 공유 카운트다운 서브서킷 사용 https://korshunov.ai/ar/article/12062-nmdhj-tdmyn-lswt-llg-tfshl-fy-trmyz-mfhym-lswt-lmnfy/korshunov.aiar2026-07-15T10:44:02.190172+00:00نماذج تضمين الصوت-اللغة تفشل في ترميز مفاهيم الصوت المنفي https://korshunov.ai/en/article/12062-audio-language-embedding-models-fail-to-encode-negated-sound-concepts/korshunov.aien2026-07-15T10:44:02.190172+00:00Audio-language embedding models fail to encode negated sound concepts https://korshunov.ai/ru/article/12062-modeli-audio-iazykovykh-embeddingov-ne-sposobny-kodirovat-otritsatelnye-zvukovye/korshunov.airu2026-07-15T10:44:02.190172+00:00Модели аудио-языковых эмбеддингов не способны кодировать отрицательные звуковые концепции https://korshunov.ai/es/article/12062-los-modelos-de-incrustacion-audio-idioma-fallan-al-codificar-conceptos-sonoros/korshunov.aies2026-07-15T10:44:02.190172+00:00Los modelos de incrustación audio-idioma fallan al codificar conceptos sonoros negados https://korshunov.ai/pt-br/article/12062-modelos-de-embedding-audio-idioma-falham-ao-codificar-conceitos-sonoros-negados/korshunov.aipt2026-07-15T10:44:02.190172+00:00Modelos de embedding áudio-idioma falham ao codificar conceitos sonoros negados https://korshunov.ai/hi/article/12062-onddiyo-bhaassaa-embedding-monddl-nissedhit-dhvni-avdhaarnnaaon-ko-enkodd-krne/korshunov.aihi2026-07-15T10:44:02.190172+00:00ऑडियो-भाषा एम्बेडिंग मॉडल निषेधित ध्वनि अवधारणाओं को एन्कोड करने में विफल रहते हैं https://korshunov.ai/ja/article/12062-yin-sheng-yan-yu-mai-meip-mimoteruhafou-ding-saretayin-nogai-nian-wofu-hao-hua/korshunov.aija2026-07-15T10:44:02.190172+00:00音声言語埋め込みモデルは否定された音の概念を符号化できない https://korshunov.ai/ko/article/12062-odio-eoneo-imbeding-modeleun-bujeongdoen-sori-gaenyeomeul-inkodinghaji-mosham/korshunov.aiko2026-07-15T10:44:02.190172+00:00오디오-언어 임베딩 모델은 부정된 소리 개념을 인코딩하지 못함 https://korshunov.ai/fr/article/12062-les-modeles-d-embedding-audio-langue-echouent-a-encoder-les-concepts-sonores/korshunov.aifr2026-07-15T10:44:02.190172+00:00Les modèles d'embedding audio-langue échouent à encoder les concepts sonores négés https://korshunov.ai/id/article/12062-model-embedding-audio-bahasa-gagal-mengodekan-konsep-suara-yang-dinegasi/korshunov.aiid2026-07-15T10:44:02.190172+00:00Model embedding audio-bahasa gagal mengodekan konsep suara yang dinegasi https://korshunov.ai/zh-hans/article/12062-yin-pin-yu-yan-qian-ru-mo-xing-wu-fa-bian-ma-fou-ding-sheng-yin-gai-nian/korshunov.aizh-cn2026-07-15T10:44:02.190172+00:00音频-语言嵌入模型无法编码否定声音概念 https://korshunov.ai/hi/article/12061-lakequest-ddettaa-leks-men-graaunddedd-prshnottr-ke-lie-ek-tiin-ddomen-bencmaark/korshunov.aihi2026-07-15T10:36:50.172710+00:00LakeQuest: डेटा लेक्स में ग्राउंडेड प्रश्नोत्तर के लिए एक तीन-डोमेन बेंचमार्क https://korshunov.ai/en/article/12061-lakequest-a-three-domain-benchmark-for-grounded-question-answering-across-data/korshunov.aien2026-07-15T10:36:50.172710+00:00LakeQuest: A Three-Domain Benchmark for Grounded Question Answering across Data Lakes https://korshunov.ai/ru/article/12061-lakequest-trekhdomennyi-benchmark-dlia-obosnovannogo-otveta-na-voprosy-v-ozerakh/korshunov.airu2026-07-15T10:36:50.172710+00:00LakeQuest: Трёхдоменный бенчмарк для обоснованного ответа на вопросы в озёрах данных https://korshunov.ai/es/article/12061-lakequest-una-evaluacion-de-tres-dominios-para-respuesta-a-preguntas-en-lagos-de/korshunov.aies2026-07-15T10:36:50.172710+00:00LakeQuest: Una evaluación de tres dominios para respuesta a preguntas fundamentada en lagos de datos https://korshunov.ai/zh-hans/article/12061-lakequest-mian-xiang-shu-ju-hu-zhong-jie-di-wen-da-de-san-yu-ji-zhun/korshunov.aizh-cn2026-07-15T10:36:50.172710+00:00LakeQuest：面向数据湖中接地问答的三域基准 https://korshunov.ai/pt-br/article/12061-lakequest-uma-avaliacao-de-tres-dominios-para-resposta-a-perguntas-fundamentada/korshunov.aipt2026-07-15T10:36:50.172710+00:00LakeQuest: Uma avaliação de três domínios para resposta a perguntas fundamentada em lagos de dados https://korshunov.ai/ja/article/12061-lakequest-te-tareikusuniwataru-grounded-nazhi-wen-ying-da-ku/korshunov.aija2026-07-15T10:36:50.172710+00:00LakeQuest: データレイクスにわたる grounded な質問応答のための3ドメインベンチマーク https://korshunov.ai/ko/article/12061-lakequest-deiteo-reikeu-gan-gocag-jilmun-dabbyeoneul-wihan-3domein-bencimakeu/korshunov.aiko2026-07-15T10:36:50.172710+00:00LakeQuest: 데이터 레이크 간 고착 질문 답변을 위한 3도메인 벤치마크 https://korshunov.ai/fr/article/12061-lakequest-un-benchmark-a-trois-domaines-pour-la-reponse-aux-questions-ancree-les/korshunov.aifr2026-07-15T10:36:50.172710+00:00LakeQuest : Un benchmark à trois domaines pour la réponse aux questions ancrée dans les lacs de données https://korshunov.ai/id/article/12061-lakequest-benchmark-tiga-domain-untuk-jawaban-pertanyaan-berlandaskan-di-seluruh/korshunov.aiid2026-07-15T10:36:50.172710+00:00LakeQuest: Benchmark Tiga-Domain untuk Jawaban Pertanyaan Berlandaskan di Seluruh Data Lake https://korshunov.ai/ar/article/12061-lakequest-m-yr-thlthy-lntqt-lljb-l-lsyl-lmrtbt-br-bhyrt-lbynt/korshunov.aiar2026-07-15T10:36:50.172710+00:00LakeQuest: معيار ثلاثي النطاقات للإجابة على الأسئلة المرتبطة عبر بحيرات البيانات https://korshunov.ai/fr/article/12060-selection-de-preuves-optimisee-par-qubo-pour-la-reponse-aux-questions-augmentee/korshunov.aifr2026-07-15T10:29:35.064577+00:00Sélection de preuves optimisée par QUBO pour la réponse aux questions augmentée par la récupération avec des solveurs non conventionnels https://korshunov.ai/id/article/12060-seleksi-bukti-dioptimalkan-dengan-qubo-untuk-jawaban-pertanyaan-berbasis-dengan/korshunov.aiid2026-07-15T10:29:35.064577+00:00Seleksi Bukti Dioptimalkan dengan QUBO untuk Jawaban Pertanyaan Berbasis Retrieval dengan Solver Tidak Konvensional https://korshunov.ai/hi/article/12060-anaupcaarik-sonlvrs-ke-saath-punrpraapti-vrdhit-prshnottr-ke-lie-qubo-anukuulit/korshunov.aihi2026-07-15T10:29:35.064577+00:00अनौपचारिक सॉल्वर्स के साथ पुनर्प्राप्ति-वर्धित प्रश्नोत्तर के लिए QUBO-अनुकूलित साक्ष्य चयन https://korshunov.ai/ja/article/12060-fei-chuan-tong-de-nasoruha-woyong-itajian-suo-kuo-zhang-xing-zhi-wen-ying-da-shi/korshunov.aija2026-07-15T10:29:35.064577+00:00非伝統的なソルバーを用いた検索拡張型質問応答におけるQUBO最適化による証拠選択 https://korshunov.ai/ko/article/12060-bijeontongjeog-solbeoreul-hwalyonghan-geomsaeg-jeunggang-jilmun-dabbyeoneul-qubo/korshunov.aiko2026-07-15T10:29:35.064577+00:00비전통적 솔버를 활용한 검색 증강 질문 답변을 위한 QUBO 최적화 증거 선택 https://korshunov.ai/en/article/12060-qubo-optimized-evidence-selection-for-retrieval-augmented-question-answering/korshunov.aien2026-07-15T10:29:35.064577+00:00QUBO-Optimized Evidence Selection for Retrieval-Augmented Question Answering with Unconventional Solvers https://korshunov.ai/ru/article/12060-optimizatsiia-otbora-dokazatelstv-po-qubo-dlia-voprosno-otvetnykh-sistem-s-i/korshunov.airu2026-07-15T10:29:35.064577+00:00Оптимизация отбора доказательств по QUBO для вопросно-ответных систем с извлечением информации и нестандартными решателями https://korshunov.ai/es/article/12060-seleccion-de-evidencia-optimizada-por-qubo-para-respuesta-a-preguntas-con-y-no/korshunov.aies2026-07-15T10:29:35.064577+00:00Selección de evidencia optimizada por QUBO para respuesta a preguntas con recuperación aumentada y solucionadores no convencionales https://korshunov.ai/zh-hans/article/12060-shi-yong-fei-chuan-tong-qiu-jie-qi-jin-xing-jian-suo-zeng-qiang-wen-da-de-hua-ju/korshunov.aizh-cn2026-07-15T10:29:35.064577+00:00使用非传统求解器进行检索增强问答的QUBO优化证据选择 https://korshunov.ai/pt-br/article/12060-selecao-de-evidencia-otimizada-por-qubo-para-resposta-a-perguntas-com-aumentada/korshunov.aipt2026-07-15T10:29:35.064577+00:00Seleção de Evidência Otimizada por QUBO para Resposta a Perguntas com Recuperação Aumentada e Solucionadores Não Convencionais https://korshunov.ai/ar/article/12060-khtyr-ldl-lmuhsawn-bwst-qubo-lljb-l-lsyl-lm-zz-blstrj-bstkhdm-hllt-gyr-tqlydy/korshunov.aiar2026-07-15T10:29:35.064577+00:00اختيار الأدلة المُحسَّن بواسطة QUBO للإجابة على الأسئلة المعززة بالاسترجاع باستخدام حلالات غير تقليدية https://korshunov.ai/hi/article/12059-baanglaa-kes-sttddii-svaasthy-bhraamk-jaankaarii-kaa-ptaa-lgaane-ke-lie-slms-ko/korshunov.aihi2026-07-15T10:21:43.076401+00:00बांग्ला केस स्टडी स्वास्थ्य भ्रामक जानकारी का पता लगाने के लिए SLMs को सांस्कृतिक-संवेदनशील NLP के साथ एकीकृत करती है https://korshunov.ai/ja/article/12059-henkaruyu-noshi-li-yan-jiu-slmtowen-hua-de-nipei-lu-shitanlpwojian-kang-wu-qing/korshunov.aija2026-07-15T10:21:43.076401+00:00ベンガル語の事例研究: SLMと文化的に配慮したNLPを健康誤情報検出に統合 https://korshunov.ai/ko/article/12059-benggoleo-sarye-yeongu-geongang-ojeongbo-gamjireul-wihae-slmgwa-munhwajeogeuro/korshunov.aiko2026-07-15T10:21:43.076401+00:00벵골어 사례 연구: 건강 오정보 감지를 위해 SLM과 문화적으로 민감한 NLP 통합 https://korshunov.ai/fr/article/12059-etude-de-cas-bengali-integration-des-slm-et-d-une-nlp-culturellement-sensible-la/korshunov.aifr2026-07-15T10:21:43.076401+00:00Étude de cas bengali : intégration des SLM et d'une NLP culturellement sensible pour la détection de la désinformation sanitaire https://korshunov.ai/id/article/12059-studi-kasus-bengali-mengintegrasikan-slm-dengan-nlp-yang-sensitif-secara-budaya/korshunov.aiid2026-07-15T10:21:43.076401+00:00Studi kasus Bengali mengintegrasikan SLM dengan NLP yang sensitif secara budaya untuk deteksi misinformasi kesehatan https://korshunov.ai/ar/article/12059-drs-hl-bngly-tdmj-lnmdhj-llgwy-lsgyr-slms-m-m-lj-llg-ltby-y-lhss-thqfyan-llkshf/korshunov.aiar2026-07-15T10:21:43.076401+00:00دراسة حالة بنغالية تدمج النماذج اللغوية الصغيرة (SLMs) مع معالجة اللغة الطبيعية الحساسة ثقافياً للكشف عن المعلومات المضللة الصحية https://korshunov.ai/en/article/12059-bangla-case-study-integrates-slms-with-culturally-sensitive-nlp-for-health/korshunov.aien2026-07-15T10:21:43.076401+00:00Bangla case study integrates SLMs with culturally-sensitive NLP for health misinformation detection https://korshunov.ai/ru/article/12059-issledovanie-na-bengalskom-iazyke-integriruet-slm-s-kulturno-chuvstvitelnym-nlp/korshunov.airu2026-07-15T10:21:43.076401+00:00Исследование на бенгальском языке интегрирует SLM с культурно-чувствительным NLP для обнаружения дезинформации о здоровье https://korshunov.ai/es/article/12059-estudio-de-caso-en-bengali-integra-slms-con-nlp-culturalmente-sensible-para-la/korshunov.aies2026-07-15T10:21:43.076401+00:00Estudio de caso en bengalí integra SLMs con NLP culturalmente sensible para la detección de desinformación sobre salud https://korshunov.ai/zh-hans/article/12059-meng-jia-la-yu-an-li-yan-jiu-jiang-slmyu-wen-hua-min-gan-xing-nlpzheng-he-yong/korshunov.aizh-cn2026-07-15T10:21:43.076401+00:00孟加拉语案例研究将SLM与文化敏感型NLP整合，用于健康虚假信息检测 https://korshunov.ai/pt-br/article/12059-estudo-de-caso-em-bangla-integra-slms-com-nlp-culturalmente-sensivel-para-de/korshunov.aipt2026-07-15T10:21:43.076401+00:00Estudo de caso em bangla integra SLMs com NLP culturalmente sensível para detecção de desinformação sobre saúde https://korshunov.ai/pt-br/article/12058-pcc-sql-impoe-controle-de-acesso-em-nivel-de-coluna-no-text-to-sql-via-restrita/korshunov.aipt2026-07-15T10:14:02.095952+00:00PCC-SQL impõe controle de acesso em nível de coluna no Text-to-SQL via decodificação restrita condicionada por política https://korshunov.ai/hi/article/12058-pcc-sql-niiti-shrtbndhit-prtibddh-ddikodding-ke-maadhym-se-ttekstt-ttuu-eskyuuel/korshunov.aihi2026-07-15T10:14:02.095952+00:00PCC-SQL नीति-शर्तबंधित प्रतिबद्ध डिकोडिंग के माध्यम से टेक्स्ट-टू-एसक्यूएल में कॉलम-स्तर एक्सेस नियंत्रण लागू करता है https://korshunov.ai/ja/article/12058-pcc-sqlhahorishi-tiao-jian-fu-kizhi-yue-teko-teinkuniyoritext-to-sqltelie-yu-zhi/korshunov.aija2026-07-15T10:14:02.095952+00:00PCC-SQLはポリシー条件付き制約デコーディングによりText-to-SQLで列レベルのアクセス制御を強制する https://korshunov.ai/ko/article/12058-pcc-sqleun-jeongcaeg-jogeonbu-jeyag-dikodingeul-tonghae-text-to-sqleseo-yeol/korshunov.aiko2026-07-15T10:14:02.095952+00:00PCC-SQL은 정책 조건부 제약 디코딩을 통해 Text-to-SQL에서 열 수준 액세스 제어를 강제합니다 https://korshunov.ai/fr/article/12058-pcc-sql-impose-un-controle-d-acces-au-niveau-des-colonnes-dans-text-to-sql-via/korshunov.aifr2026-07-15T10:14:02.095952+00:00PCC-SQL impose un contrôle d'accès au niveau des colonnes dans Text-to-SQL via un décodage contraint conditionné par politique https://korshunov.ai/id/article/12058-pcc-sql-menegakkan-kontrol-akses-tingkat-kolom-dalam-text-to-sql-melalui/korshunov.aiid2026-07-15T10:14:02.095952+00:00PCC-SQL menegakkan kontrol akses tingkat kolom dalam Text-to-SQL melalui decoding terkondisi kebijakan https://korshunov.ai/en/article/12058-pcc-sql-enforces-column-level-access-control-in-text-to-sql-via-policy-decoding/korshunov.aien2026-07-15T10:14:02.095952+00:00PCC-SQL enforces column-level access control in Text-to-SQL via policy-conditioned constrained decoding https://korshunov.ai/ru/article/12058-pcc-sql-obespechivaet-kontrol-dostupa-na-urovne-stolbtsov-v-text-to-sql-s/korshunov.airu2026-07-15T10:14:02.095952+00:00PCC-SQL обеспечивает контроль доступа на уровне столбцов в Text-to-SQL с помощью ограниченного декодирования, управляемого политикой https://korshunov.ai/zh-hans/article/12058-pcc-sql-tong-guo-ce-lue-yue-shu-jie-ma-zai-text-to-sql-zhong-shi-shi-lie-ji-fang/korshunov.aizh-cn2026-07-15T10:14:02.095952+00:00PCC-SQL 通过策略约束解码在 Text-to-SQL 中实施列级访问控制 https://korshunov.ai/es/article/12058-pcc-sql-aplica-control-de-acceso-a-nivel-de-columna-en-text-to-sql-mediante-por/korshunov.aies2026-07-15T10:14:02.095952+00:00PCC-SQL aplica control de acceso a nivel de columna en Text-to-SQL mediante decodificación restringida condicionada por política https://korshunov.ai/ar/article/12058-yfrd-pcc-sql-lthkm-fy-lwswl-l-mstw-l-md-fy-text-to-sql-br-fk-ltshfyr-lmqyd-blsys/korshunov.aiar2026-07-15T10:14:02.095952+00:00يفرض PCC-SQL التحكم في الوصول على مستوى الأعمدة في Text-to-SQL عبر فك التشفير المقيد المشروط بالسياسة https://korshunov.ai/ar/article/12057-tsnyf-mt-dd-lb-d-lm-lwmt-lsrtn-lmdll-l-rdyt/korshunov.aiar2026-07-15T10:07:16.078027+00:00تصنيف متعدد الأبعاد لمعلومات السرطان المضللة على رديت https://korshunov.ai/en/article/12057-multi-dimensional-taxonomy-of-cancer-misinformation-on-reddit/korshunov.aien2026-07-15T10:07:16.078027+00:00Multi-Dimensional Taxonomy of Cancer Misinformation on Reddit https://korshunov.ai/fr/article/12057-taxonomie-multidimensionnelle-de-la-desinformation-sur-le-cancer-sur-reddit/korshunov.aifr2026-07-15T10:07:16.078027+00:00Taxonomie multidimensionnelle de la désinformation sur le cancer sur Reddit https://korshunov.ai/id/article/12057-taksonomi-multi-dimensi-misinformasi-kanker-di-reddit/korshunov.aiid2026-07-15T10:07:16.078027+00:00Taksonomi Multi-Dimensi Misinformasi Kanker di Reddit https://korshunov.ai/zh-hans/article/12057-redditshang-guan-yu-yan-zheng-de-duo-wei-du-xu-jia-xin-xi-fen-lei-ti-xi/korshunov.aizh-cn2026-07-15T10:07:16.078027+00:00Reddit上关于癌症的多维度虚假信息分类体系 https://korshunov.ai/ru/article/12057-mnogomernaia-taksonomiia-dezinformatsii-o-rake-v-reddit/korshunov.airu2026-07-15T10:07:16.078027+00:00Многомерная таксономия дезинформации о раке в Reddit https://korshunov.ai/es/article/12057-taxonomia-multidimensional-de-la-desinformacion-sobre-el-cancer-en-reddit/korshunov.aies2026-07-15T10:07:16.078027+00:00Taxonomía multidimensional de la desinformación sobre el cáncer en Reddit https://korshunov.ai/pt-br/article/12057-taxonomia-multidimensional-da-desinformacao-sobre-cancer-no-reddit/korshunov.aipt2026-07-15T10:07:16.078027+00:00Taxonomia multidimensional da desinformação sobre câncer no Reddit https://korshunov.ai/hi/article/12057-redditt-pr-kainsr-kii-glt-jaankaarii-ke-lie-bhu-aayaamii-vrgiikrnn/korshunov.aihi2026-07-15T10:07:16.078027+00:00रेडिट पर कैंसर की गलत जानकारी के लिए बहु-आयामी वर्गीकरण https://korshunov.ai/ja/article/12057-redditniokerukanwu-qing-bao-kaduo-jiao-de-fen-lei-ti-xi/korshunov.aija2026-07-15T10:07:16.078027+00:00Redditにおけるがん誤情報が多角的分類体系 https://korshunov.ai/ko/article/12057-redisyi-am-heowi-jeongboe-daehan-dacaweon-bunryu-cegye/korshunov.aiko2026-07-15T10:07:16.078027+00:00레딧의 암 허위 정보에 대한 다차원 분류 체계 https://korshunov.ai/ru/article/12157-llama-cpp-dobavliaet-podderzhku-sycl-flash-attention-i-xielu-dlia-gpu-intel/korshunov.airu2026-07-15T10:00:48.252309+00:00llama.cpp добавляет поддержку SYCL Flash Attention и XIELU для GPU Intel https://korshunov.ai/es/article/12157-llama-cpp-anade-flash-attention-sycl-y-soporte-xielu-para-gpus-intel/korshunov.aies2026-07-15T10:00:48.252309+00:00llama.cpp añade Flash Attention SYCL y soporte XIELU para GPUs Intel https://korshunov.ai/zh-hans/article/12157-llama-cpp-wei-intel-gpu-tian-jia-sycl-flash-attention-he-xielu-zhi-chi/korshunov.aizh-cn2026-07-15T10:00:48.252309+00:00llama.cpp 为 Intel GPU 添加 SYCL Flash Attention 和 XIELU 支持 https://korshunov.ai/pt-br/article/12157-llama-cpp-adiciona-flash-attention-sycl-e-suporte-xielu-para-gpus-intel/korshunov.aipt2026-07-15T10:00:48.252309+00:00llama.cpp adiciona Flash Attention SYCL e suporte XIELU para GPUs Intel https://korshunov.ai/hi/article/12157-llama-cpp-ne-intel-gpu-ke-lie-sycl-flash-attention-aur-xielu-smrthn-joddaa-hai/korshunov.aihi2026-07-15T10:00:48.252309+00:00llama.cpp ने Intel GPU के लिए SYCL Flash Attention और XIELU समर्थन जोड़ा है https://korshunov.ai/ja/article/12157-llama-cppkaintel-gpuxiang-kenisycl-flash-attentiontoxielusaho-towozhui-jia/korshunov.aija2026-07-15T10:00:48.252309+00:00llama.cppがIntel GPU向けにSYCL Flash AttentionとXIELUサポートを追加 https://korshunov.ai/ko/article/12157-llama-cpp-intel-gpuyong-sycl-flash-attention-mic-xielu-jiweon-cuga/korshunov.aiko2026-07-15T10:00:48.252309+00:00llama.cpp, Intel GPU용 SYCL Flash Attention 및 XIELU 지원 추가 https://korshunov.ai/fr/article/12157-llama-cpp-ajoute-le-support-de-sycl-flash-attention-et-xielu-pour-les-gpu-intel/korshunov.aifr2026-07-15T10:00:48.252309+00:00llama.cpp ajoute le support de SYCL Flash Attention et XIELU pour les GPU Intel https://korshunov.ai/id/article/12157-llama-cpp-menambahkan-dukungan-sycl-flash-attention-dan-xielu-untuk-gpu-intel/korshunov.aiid2026-07-15T10:00:48.252309+00:00llama.cpp menambahkan dukungan SYCL Flash Attention dan XIELU untuk GPU Intel https://korshunov.ai/ar/article/12157-tdyf-llama-cpp-d-m-sycl-flash-attention-w-xielu-lbtqt-intel-lrswmy/korshunov.aiar2026-07-15T10:00:48.252309+00:00تضيف llama.cpp دعم SYCL Flash Attention و XIELU لبطاقات Intel الرسومية https://korshunov.ai/en/article/12157-llama-cpp-adds-sycl-flash-attention-and-xielu-support-for-intel-gpus/korshunov.aien2026-07-15T10:00:48.252309+00:00llama.cpp adds SYCL Flash Attention and XIELU support for Intel GPUs https://korshunov.ai/ru/article/12158-polzovatel-sravnivaet-apgreid-s-dvumia-rtx-3060-s-ispolzovaniem-3090-ili-dvukh/korshunov.airu2026-07-15T09:52:55.959039+00:00Пользователь сравнивает апгрейд с двумя RTX 3060 с использованием 3090 или двух RTX 5060 Ti https://korshunov.ai/es/article/12158-un-usuario-compara-actualizar-dos-rtx-3060-con-una-3090-usada-o-dos-rtx-5060-ti/korshunov.aies2026-07-15T09:52:55.959039+00:00Un usuario compara actualizar dos RTX 3060 con una 3090 usada o dos RTX 5060 Ti https://korshunov.ai/zh-hans/article/12158-yong-hu-bi-jiao-sheng-ji-shuang-rtx-3060-yu-er-shou-3090-huo-liang-zhang-rtx-ti/korshunov.aizh-cn2026-07-15T09:52:55.959039+00:00用户比较升级双 RTX 3060 与二手 3090 或两张 RTX 5060 Ti https://korshunov.ai/pt-br/article/12158-usuario-compara-atualizacao-de-duas-rtx-3060-com-uma-3090-usada-ou-duas-rtx-5060/korshunov.aipt2026-07-15T09:52:55.959039+00:00Usuário compara atualização de duas RTX 3060 com uma 3090 usada ou duas RTX 5060 Ti https://korshunov.ai/hi/article/12158-upyogkrtaa-do-rtx-3060-ko-apgredd-krne-kaa-upyog-kii-gii-3090-yaa-do-rtx-5060-ti/korshunov.aihi2026-07-15T09:52:55.959039+00:00उपयोगकर्ता दो RTX 3060 को अपग्रेड करने का उपयोग की गई 3090 या दो RTX 5060 Ti से तुलना कर रहा है https://korshunov.ai/ko/article/12158-dyueol-rtx-3060-eobgeureideureul-junggo-3090-ddoneun-du-gaeyi-rtx-5060-tiwa/korshunov.aiko2026-07-15T09:52:55.959039+00:00듀얼 RTX 3060 업그레이드를 중고 3090 또는 두 개의 RTX 5060 Ti와 비교하는 사용자 https://korshunov.ai/fr/article/12158-un-utilisateur-compare-la-mise-a-niveau-d-un-double-rtx-3060-avec-un-3090-d-ou/korshunov.aifr2026-07-15T09:52:55.959039+00:00Un utilisateur compare la mise à niveau d'un double RTX 3060 avec un 3090 d'occasion ou deux RTX 5060 Ti https://korshunov.ai/id/article/12158-pengguna-membandingkan-peningkatan-ganda-rtx-3060-dengan-3090-bekas-atau-dua-rtx/korshunov.aiid2026-07-15T09:52:55.959039+00:00Pengguna membandingkan peningkatan ganda RTX 3060 dengan 3090 bekas atau dua RTX 5060 Ti https://korshunov.ai/ar/article/12158-mqrn-mstkhdm-ltrqy-dd-rtx-3060-mzdwj-b-3090-mst-ml-w-thnyn-mn-rtx-5060-ti/korshunov.aiar2026-07-15T09:52:55.959039+00:00مقارنة مستخدم لترقية إعداد RTX 3060 مزدوج بـ 3090 مستعمل أو اثنين من RTX 5060 Ti https://korshunov.ai/ja/article/12158-teyuarurtx-3060noatsuhukure-towo-zhong-gu-no3090mataha2mei-nortx-5060-titobi-sa/korshunov.aija2026-07-15T09:52:55.959039+00:00デュアルRTX 3060のアップグレードを、中古の3090または2枚のRTX 5060 Tiと比較するユーザー https://korshunov.ai/en/article/12158-user-compares-upgrading-dual-rtx-3060-with-used-3090-or-two-rtx-5060-ti/korshunov.aien2026-07-15T09:52:55.959039+00:00User compares upgrading dual RTX 3060 with used 3090 or two RTX 5060 Ti https://korshunov.ai/ja/article/12156-llama-cpp-b10016kaxmxenshinniyorusycl-flash-attentionwozhui-jia/korshunov.aija2026-07-15T09:45:37.151387+00:00llama.cpp b10016がXMXエンジンによるSYCL Flash Attentionを追加 https://korshunov.ai/ko/article/12156-llama-cpp-b10016-xmx-enjineul-tonghan-sycl-flash-attention-cuga/korshunov.aiko2026-07-15T09:45:37.151387+00:00llama.cpp b10016, XMX 엔진을 통한 SYCL Flash Attention 추가 https://korshunov.ai/fr/article/12156-llama-cpp-b10016-ajoute-flash-attention-sycl-avec-le-moteur-xmx/korshunov.aifr2026-07-15T09:45:37.151387+00:00llama.cpp b10016 ajoute Flash Attention SYCL avec le moteur XMX https://korshunov.ai/id/article/12156-llama-cpp-b10016-menambahkan-flash-attention-sycl-dengan-mesin-xmx/korshunov.aiid2026-07-15T09:45:37.151387+00:00llama.cpp b10016 menambahkan Flash Attention SYCL dengan mesin XMX https://korshunov.ai/ar/article/12156-ydyf-llama-cpp-b10016-d-m-flash-attention-br-sycl-bstkhdm-mhrk-xmx/korshunov.aiar2026-07-15T09:45:37.151387+00:00يضيف llama.cpp b10016 دعم Flash Attention عبر SYCL باستخدام محرك XMX https://korshunov.ai/en/article/12156-llama-cpp-b10016-adds-sycl-flash-attention-with-xmx-engine/korshunov.aien2026-07-15T09:45:37.151387+00:00llama.cpp b10016 adds SYCL Flash Attention with XMX engine https://korshunov.ai/ru/article/12156-llama-cpp-b10016-dobavliaet-sycl-flash-attention-s-dvizhkom-xmx/korshunov.airu2026-07-15T09:45:37.151387+00:00llama.cpp b10016 добавляет SYCL Flash Attention с движком XMX https://korshunov.ai/es/article/12156-llama-cpp-b10016-anade-flash-attention-sycl-con-motor-xmx/korshunov.aies2026-07-15T09:45:37.151387+00:00llama.cpp b10016 añade Flash Attention SYCL con motor XMX https://korshunov.ai/zh-hans/article/12156-llama-cpp-b10016-tian-jia-dai-you-xmx-yin-qing-de-sycl-flash-attention/korshunov.aizh-cn2026-07-15T09:45:37.151387+00:00llama.cpp b10016 添加带有 XMX 引擎的 SYCL Flash Attention https://korshunov.ai/pt-br/article/12156-llama-cpp-b10016-adiciona-flash-attention-sycl-com-motor-xmx/korshunov.aipt2026-07-15T09:45:37.151387+00:00llama.cpp b10016 adiciona Flash Attention SYCL com motor XMX https://korshunov.ai/hi/article/12156-llama-cpp-b10016-ne-xmx-injn-ke-saath-sycl-flash-attention-joddaa/korshunov.aihi2026-07-15T09:45:37.151387+00:00llama.cpp b10016 ने XMX इंजन के साथ SYCL Flash Attention जोड़ा https://korshunov.ai/en/article/12056-ring-zero-scales-zero-rl-to-1t-parameters-for-emergent-reasoning/korshunov.aien2026-07-15T09:35:46.299890+00:00Ring-Zero scales zero RL to 1T parameters for emergent reasoning https://korshunov.ai/ko/article/12056-ring-zero-jero-rleul-1t-paramiteoro-hwagjanghayeo-dolbaljeog-curon-guhyeon/korshunov.aiko2026-07-15T09:35:46.299890+00:00Ring-Zero, 제로 RL을 1T 파라미터로 확장하여 돌발적 추론 구현 https://korshunov.ai/fr/article/12056-ring-zero-met-a-l-echelle-le-rl-zero-jusqu-a-1t-de-parametres-pour-un-emergent/korshunov.aifr2026-07-15T09:35:46.299890+00:00Ring-Zero met à l'échelle le RL zéro jusqu'à 1T de paramètres pour un raisonnement émergent https://korshunov.ai/id/article/12056-ring-zero-menskalakan-rl-nol-ke-1t-parameter-untuk-penalaran-yang-muncul/korshunov.aiid2026-07-15T09:35:46.299890+00:00Ring-Zero menskalakan RL nol ke 1T parameter untuk penalaran yang muncul https://korshunov.ai/ar/article/12056-ring-zero-yws-lt-lm-lm-zz-lsfry-l-1-trylywn-m-lm-llstdll-lnshy/korshunov.aiar2026-07-15T09:35:46.299890+00:00Ring-Zero يوسع التعلم المعزز الصفري إلى 1 تريليون معلمة للاستدلال الناشئ https://korshunov.ai/ru/article/12056-ring-zero-masshtabiruet-nulevoe-obuchenie-s-podkrepleniem-do-1-trln-parametrov/korshunov.airu2026-07-15T09:35:46.299890+00:00Ring-Zero масштабирует нулевое обучение с подкреплением до 1 трлн параметров для эмерджентного рассуждения https://korshunov.ai/es/article/12056-ring-zero-escala-el-rl-cero-a-1t-parametros-para-razonamiento-emergente/korshunov.aies2026-07-15T09:35:46.299890+00:00Ring-Zero escala el RL cero a 1T parámetros para razonamiento emergente https://korshunov.ai/zh-hans/article/12056-ring-zero-jiang-ling-yang-ben-qiang-hua-xue-xi-kuo-zhan-zhi-1t-can-shu-yi-shi-li/korshunov.aizh-cn2026-07-15T09:35:46.299890+00:00Ring-Zero 将零样本强化学习扩展至 1T 参数以实现涌现推理 https://korshunov.ai/pt-br/article/12056-ring-zero-escala-rl-zero-para-1t-parametros-para-raciocinio-emergente/korshunov.aipt2026-07-15T09:35:46.299890+00:00Ring-Zero escala RL zero para 1T parâmetros para raciocínio emergente https://korshunov.ai/hi/article/12056-ring-zero-utpnn-trk-ke-lie-1t-pairaamiittr-tk-shuuny-rl-ko-skel-krtaa-hai/korshunov.aihi2026-07-15T09:35:46.299890+00:00Ring-Zero उत्पन्न तर्क के लिए 1T पैरामीटर तक शून्य RL को स्केल करता है https://korshunov.ai/ja/article/12056-ring-zerohaserorlwo1tharame-tanisuke-rinkushi-tu-fa-de-natui-lun-woshi-xian/korshunov.aija2026-07-15T09:35:46.299890+00:00Ring-ZeroはゼロRLを1Tパラメータにスケーリングし、突発的な推論を実現 https://korshunov.ai/ru/article/12055-wikistar-raskryvaet-skrytuiu-nauchnuiu-istoriiu-statei-vikipedii/korshunov.airu2026-07-15T09:27:57.984468+00:00WikiSTAR раскрывает скрытую научную историю статей Википедии https://korshunov.ai/es/article/12055-wikistar-revela-la-historia-cientifica-oculta-de-los-articulos-de-wikipedia/korshunov.aies2026-07-15T09:27:57.984468+00:00WikiSTAR revela la historia científica oculta de los artículos de Wikipedia https://korshunov.ai/zh-hans/article/12055-wikistarjie-shi-wei-ji-bai-ke-wen-zhang-yin-cang-de-ke-xue-li-shi/korshunov.aizh-cn2026-07-15T09:27:57.984468+00:00WikiSTAR揭示维基百科文章隐藏的科学历史 https://korshunov.ai/pt-br/article/12055-wikistar-revela-a-historia-cientifica-oculta-dos-artigos-da-wikipedia/korshunov.aipt2026-07-15T09:27:57.984468+00:00WikiSTAR revela a história científica oculta dos artigos da Wikipedia https://korshunov.ai/hi/article/12055-wikistar-vikipiidia-lekhon-kii-chipii-huii-vaijnyaanik-itihaas-ko-ujaagr-krtaa/korshunov.aihi2026-07-15T09:27:57.984468+00:00WikiStar वikipीdia लेखों की छिपी हुई वैज्ञानिक इतिहास को उजागर करता है https://korshunov.ai/ja/article/12055-wikistarkawikipediaji-shi-noyin-saretake-xue-shi-woming-rakanisuru/korshunov.aija2026-07-15T09:27:57.984468+00:00WikiSTARがWikipedia記事の隠された科学史を明らかにする https://korshunov.ai/ko/article/12055-wikistar-wikibaeggwa-munseoyi-sumgyeojin-gwahagsa-balghyeonae/korshunov.aiko2026-07-15T09:27:57.984468+00:00WikiSTAR, 위키백과 문서의 숨겨진 과학사 밝혀내 https://korshunov.ai/fr/article/12055-wikistar-revele-l-histoire-scientifique-cachee-des-articles-de-wikipedia/korshunov.aifr2026-07-15T09:27:57.984468+00:00WikiSTAR révèle l'histoire scientifique cachée des articles de Wikipédia https://korshunov.ai/id/article/12055-wikistar-mengungkap-sejarah-ilmiah-tersembunyi-artikel-wikipedia/korshunov.aiid2026-07-15T09:27:57.984468+00:00WikiSTAR mengungkap sejarah ilmiah tersembunyi artikel Wikipedia https://korshunov.ai/ar/article/12055-wikistar-tkshf-ltrykh-l-lmy-lmkhfy-lmqlt-wykybydy/korshunov.aiar2026-07-15T09:27:57.984468+00:00WikiSTAR تكشف التاريخ العلمي المخفي لمقالات ويكيبيديا https://korshunov.ai/en/article/12055-wikistar-reveals-hidden-scientific-history-of-wikipedia-articles/korshunov.aien2026-07-15T09:27:57.984468+00:00WikiSTAR reveals hidden scientific history of Wikipedia articles https://korshunov.ai/en/article/12052-jolt-achieves-near-lossless-kv-cache-compression-via-joint-tucker-decomposition/korshunov.aien2026-07-15T09:22:18.016998+00:00JoLT achieves near-lossless KV cache compression via joint Tucker decomposition and JL-residual allocation https://korshunov.ai/ru/article/12052-jolt-dostigaet-pochti-bezpoternoi-kompressii-kv-kesha-posredstvom-sovmestnogo-i/korshunov.airu2026-07-15T09:22:18.016998+00:00JoLT достигает почти безпотерьной компрессии KV-кэша посредством совместного тукеровского разложения и распределения остатков JL https://korshunov.ai/es/article/12052-jolt-logra-una-compresion-de-cache-kv-casi-sin-perdidas-mediante-descomposicion/korshunov.aies2026-07-15T09:22:18.016998+00:00JoLT logra una compresión de caché KV casi sin pérdidas mediante descomposición de Tucker conjunta y asignación de residuos JL https://korshunov.ai/zh-hans/article/12052-jolttong-guo-lian-he-tuckerfen-jie-he-jlcan-chai-fen-pei-shi-xian-jin-hu-wu-sun/korshunov.aizh-cn2026-07-15T09:22:18.016998+00:00JoLT通过联合Tucker分解和JL残差分配实现近乎无损的KV缓存压缩 https://korshunov.ai/pt-br/article/12052-jolt-alcanca-compressao-de-cache-kv-quase-sem-perdas-via-decomposicao-de-tucker/korshunov.aipt2026-07-15T09:22:18.016998+00:00JoLT alcança compressão de cache KV quase sem perdas via decomposição de Tucker conjunta e alocação de resíduos JL https://korshunov.ai/hi/article/12052-jolt-snyukt-ttkr-apghttn-aur-jl-avshess-aavnttn-ke-maadhym-se-lgbhg-haanirhit-kv/korshunov.aihi2026-07-15T09:22:18.016998+00:00JoLT संयुक्त टकर अपघटन और JL-अवशेष आवंटन के माध्यम से लगभग हानिरहित KV कैश संपीड़न प्राप्त करता है https://korshunov.ai/ja/article/12052-jolthajie-he-taruka-fen-jie-tojlcan-chai-ge-ridang-teniyori-hohowu-sun-shi-suo/korshunov.aija2026-07-15T09:22:18.016998+00:00JoLTは結合タルカー分解とJL残差割り当てにより、ほぼ無損失のKVキャッシュ圧縮を実現 https://korshunov.ai/ko/article/12052-joltneun-gyeolhab-teokeo-bunhaewa-jl-janryu-haldangeul-tonghae-geoyi-musonsil-kv/korshunov.aiko2026-07-15T09:22:18.016998+00:00JoLT는 결합 터커 분해와 JL 잔류 할당을 통해 거의 무손실 KV 캐시 압축 달성 https://korshunov.ai/fr/article/12052-jolt-realise-une-compression-de-cache-kv-quasi-sans-perte-via-une-decomposition/korshunov.aifr2026-07-15T09:22:18.016998+00:00JoLT réalise une compression de cache KV quasi sans perte via une décomposition de Tucker conjointe et une allocation de résidus JL https://korshunov.ai/id/article/12052-jolt-mencapai-kompresi-cache-kv-hampir-tanpa-kehilangan-melalui-dekomposisi-dan/korshunov.aiid2026-07-15T09:22:18.016998+00:00JoLT mencapai kompresi cache KV hampir tanpa kehilangan melalui dekomposisi Tucker gabungan dan alokasi residual JL https://korshunov.ai/ar/article/12052-thqq-jolt-dgtan-shbh-dym-lfqd-lkhzn-lmftyh-wlqym-br-lthll-lmshtrk-ltkr-wkhsm-jl/korshunov.aiar2026-07-15T09:22:18.016998+00:00تحقق JoLT ضغطًا شبه عديم الفقد لخزانة المفاتيح والقيم عبر التحلل المشترك لتكر وخصم JL المتبقي https://korshunov.ai/en/article/12150-exllamav3-v1-0-0-production-release-brings-major-performance-upgrades/korshunov.aien2026-07-15T09:10:51.289727+00:00ExLlamaV3 v1.0.0 production release brings major performance upgrades https://korshunov.ai/ru/article/12150-vykhod-exllamav3-v1-0-0-v-proizvodstvennuiu-versiiu-prinosit-znachitelnoe/korshunov.airu2026-07-15T09:10:51.289727+00:00Выход ExLlamaV3 v1.0.0 в производственную версию приносит значительное улучшение производительности https://korshunov.ai/es/article/12150-la-version-de-produccion-exllamav3-v1-0-0-trae-importantes-mejoras-de/korshunov.aies2026-07-15T09:10:51.289727+00:00La versión de producción ExLlamaV3 v1.0.0 trae importantes mejoras de rendimiento https://korshunov.ai/zh-hans/article/12150-exllamav3-v1-0-0-sheng-chan-ban-ben-fa-bu-dai-lai-zhong-da-xing-neng-ti-sheng/korshunov.aizh-cn2026-07-15T09:10:51.289727+00:00ExLlamaV3 v1.0.0 生产版本发布，带来重大性能提升 https://korshunov.ai/pt-br/article/12150-lancamento-de-producao-do-exllamav3-v1-0-0-traz-grandes-melhorias-de-desempenho/korshunov.aipt2026-07-15T09:10:51.289727+00:00Lançamento de produção do ExLlamaV3 v1.0.0 traz grandes melhorias de desempenho https://korshunov.ai/hi/article/12150-exllamav3-v1-0-0-utpaadn-riliij-prmukh-prdrshn-sudhaar-laatii-hai/korshunov.aihi2026-07-15T09:10:51.289727+00:00ExLlamaV3 v1.0.0 उत्पादन रिलीज़ प्रमुख प्रदर्शन सुधार लाती है https://korshunov.ai/ja/article/12150-exllamav3-v1-0-0-noben-fan-riri-suniyorizhu-yao-nahahuo-mansunoxiang-shang/korshunov.aija2026-07-15T09:10:51.289727+00:00ExLlamaV3 v1.0.0 の本番リリースにより主要なパフォーマンスの向上がもたらされる https://korshunov.ai/ko/article/12150-exllamav3-v1-0-0-peurodeogsyeon-rilriseuro-juyo-seongneung-hyangsang-jegong/korshunov.aiko2026-07-15T09:10:51.289727+00:00ExLlamaV3 v1.0.0 프로덕션 릴리스로 주요 성능 향상 제공 https://korshunov.ai/fr/article/12150-la-version-de-production-exllamav3-v1-0-0-apporte-des-ameliorations-majeures-de/korshunov.aifr2026-07-15T09:10:51.289727+00:00La version de production ExLlamaV3 v1.0.0 apporte des améliorations majeures de performance https://korshunov.ai/id/article/12150-rilis-produksi-exllamav3-v1-0-0-membawa-peningkatan-kinerja-utama/korshunov.aiid2026-07-15T09:10:51.289727+00:00Rilis produksi ExLlamaV3 v1.0.0 membawa peningkatan kinerja utama https://korshunov.ai/ar/article/12150-sdr-exllamav3-v1-0-0-llntj-yjlb-thsynt-kbyr-fy-ld/korshunov.aiar2026-07-15T09:10:51.289727+00:00إصدار ExLlamaV3 v1.0.0 للإنتاج يجلب تحسينات كبيرة في الأداء https://korshunov.ai/id/article/12155-eksperimen-dengan-llm-difusi-llada2-1-sumi-di-apple-silicon/korshunov.aiid2026-07-15T08:59:48.137269+00:00Eksperimen dengan LLM difusi (LLaDA2.1 & Sumi) di Apple Silicon https://korshunov.ai/en/article/12155-experiments-with-diffusion-llms-llada2-1-sumi-on-apple-silicon/korshunov.aien2026-07-15T08:59:48.137269+00:00Experiments with diffusion LLMs (LLaDA2.1 & Sumi) on Apple Silicon https://korshunov.ai/ru/article/12155-eksperimenty-s-diffuzionnymi-llm-llada2-1-i-sumi-na-apple-silicon/korshunov.airu2026-07-15T08:59:48.137269+00:00Эксперименты с диффузионными LLM (LLaDA2.1 и Sumi) на Apple Silicon https://korshunov.ai/es/article/12155-experimentos-con-llms-de-difusion-llada2-1-y-sumi-en-apple-silicon/korshunov.aies2026-07-15T08:59:48.137269+00:00Experimentos con LLMs de difusión (LLaDA2.1 y Sumi) en Apple Silicon https://korshunov.ai/zh-hans/article/12155-zai-apple-silicon-shang-dui-kuo-san-llm-llada2-1-he-sumi-de-shi-yan/korshunov.aizh-cn2026-07-15T08:59:48.137269+00:00在 Apple Silicon 上对扩散 LLM（LLaDA2.1 和 Sumi）的实验 https://korshunov.ai/pt-br/article/12155-experimentos-com-llms-de-difusao-llada2-1-e-sumi-no-apple-silicon/korshunov.aipt2026-07-15T08:59:48.137269+00:00Experimentos com LLMs de difusão (LLaDA2.1 e Sumi) no Apple Silicon https://korshunov.ai/hi/article/12155-apple-silicon-pr-ddiphyuujn-llms-llada2-1-aur-sumi-ke-saath-pryog/korshunov.aihi2026-07-15T08:59:48.137269+00:00Apple Silicon पर डिफ्यूज़न LLMs (LLaDA2.1 और Sumi) के साथ प्रयोग https://korshunov.ai/ja/article/12155-apple-silicontenokuo-san-llm-llada2-1-sumi-noshi-yan/korshunov.aija2026-07-15T08:59:48.137269+00:00Apple Siliconでの拡散LLM（LLaDA2.1 & Sumi）の実験 https://korshunov.ai/ko/article/12155-apple-siliconeseoyi-hwagsan-llm-llada2-1-sumi-silheom/korshunov.aiko2026-07-15T08:59:48.137269+00:00Apple Silicon에서의 확산 LLM(LLaDA2.1 & Sumi) 실험 https://korshunov.ai/fr/article/12155-experiences-avec-des-llm-de-diffusion-llada2-1-sumi-sur-apple-silicon/korshunov.aifr2026-07-15T08:59:48.137269+00:00Expériences avec des LLM de diffusion (LLaDA2.1 & Sumi) sur Apple Silicon https://korshunov.ai/ar/article/12155-tjrb-m-nmdhj-llm-lntshry-llada2-1-w-sumi-l-apple-silicon/korshunov.aiar2026-07-15T08:59:48.137269+00:00تجارب مع نماذج LLM الانتشارية (LLaDA2.1 و Sumi) على Apple Silicon https://korshunov.ai/en/article/12153-zerogpu-users-report-mcp-connection-issues/korshunov.aien2026-07-15T08:48:54.078071+00:00ZeroGPU users report MCP connection issues https://korshunov.ai/ru/article/12153-polzovateli-zerogpu-soobshchaiut-o-problemakh-s-podkliucheniem-mcp/korshunov.airu2026-07-15T08:48:54.078071+00:00Пользователи ZeroGPU сообщают о проблемах с подключением MCP https://korshunov.ai/es/article/12153-usuarios-de-zerogpu-reportan-problemas-de-conexion-con-mcp/korshunov.aies2026-07-15T08:48:54.078071+00:00Usuarios de ZeroGPU reportan problemas de conexión con MCP https://korshunov.ai/zh-hans/article/12153-zerogpuyong-hu-bao-gao-mcplian-jie-wen-ti/korshunov.aizh-cn2026-07-15T08:48:54.078071+00:00ZeroGPU用户报告MCP连接问题 https://korshunov.ai/pt-br/article/12153-usuarios-do-zerogpu-relatam-problemas-de-conexao-com-mcp/korshunov.aipt2026-07-15T08:48:54.078071+00:00Usuários do ZeroGPU relatam problemas de conexão com MCP https://korshunov.ai/hi/article/12153-zerogpu-upyogkrtaaon-ne-mcp-knekshn-smsyaaen-riportt-kiin/korshunov.aihi2026-07-15T08:48:54.078071+00:00ZeroGPU उपयोगकर्ताओं ने MCP कनेक्शन समस्याएं रिपोर्ट कीं https://korshunov.ai/ja/article/12153-zerogpuyu-sa-kamcpjie-sok-nowen-ti-wobao-gao/korshunov.aija2026-07-15T08:48:54.078071+00:00ZeroGPUユーザーがMCP接続の問題を報告 https://korshunov.ai/ko/article/12153-zerogpu-sayongjaga-mcp-yeongyeol-munjereul-bogo/korshunov.aiko2026-07-15T08:48:54.078071+00:00ZeroGPU 사용자가 MCP 연결 문제를 보고 https://korshunov.ai/fr/article/12153-les-utilisateurs-de-zerogpu-signalent-des-problemes-de-connexion-mcp/korshunov.aifr2026-07-15T08:48:54.078071+00:00Les utilisateurs de ZeroGPU signalent des problèmes de connexion MCP https://korshunov.ai/id/article/12153-pengguna-zerogpu-melaporkan-masalah-koneksi-mcp/korshunov.aiid2026-07-15T08:48:54.078071+00:00Pengguna ZeroGPU melaporkan masalah koneksi MCP https://korshunov.ai/ar/article/12153-mstkhdmw-zerogpu-yblgwn-n-mshkl-fy-tsl-mcp/korshunov.aiar2026-07-15T08:48:54.078071+00:00مستخدمو ZeroGPU يبلغون عن مشاكل في اتصال MCP https://korshunov.ai/en/article/12152-hugging-face-removes-free-docker-sdk-tier-from-spaces/korshunov.aien2026-07-15T08:43:56.286139+00:00Hugging Face removes free Docker SDK tier from Spaces https://korshunov.ai/ko/article/12152-hugging-face-spacesyi-muryo-docker-sdk-tieo-jegeo/korshunov.aiko2026-07-15T08:43:56.286139+00:00Hugging Face, Spaces의 무료 Docker SDK 티어 제거 https://korshunov.ai/ru/article/12152-hugging-face-otkliuchaet-besplatnyi-uroven-docker-sdk-dlia-spaces/korshunov.airu2026-07-15T08:43:56.286139+00:00Hugging Face отключает бесплатный уровень Docker SDK для Spaces https://korshunov.ai/es/article/12152-hugging-face-elimina-la-capa-gratuita-de-docker-sdk-en-spaces/korshunov.aies2026-07-15T08:43:56.286139+00:00Hugging Face elimina la capa gratuita de Docker SDK en Spaces https://korshunov.ai/zh-hans/article/12152-hugging-face-yi-chu-spaces-de-mian-fei-docker-sdk-ceng-ji/korshunov.aizh-cn2026-07-15T08:43:56.286139+00:00Hugging Face 移除 Spaces 的免费 Docker SDK 层级 https://korshunov.ai/pt-br/article/12152-hugging-face-remove-a-camada-gratuita-do-docker-sdk-dos-spaces/korshunov.aipt2026-07-15T08:43:56.286139+00:00Hugging Face remove a camada gratuita do Docker SDK dos Spaces https://korshunov.ai/hi/article/12152-hugging-face-ne-spaces-se-mupht-docker-sdk-str-httaa-diyaa/korshunov.aihi2026-07-15T08:43:56.286139+00:00Hugging Face ने Spaces से मुफ्त Docker SDK स्तर हटा दिया https://korshunov.ai/ja/article/12152-hugging-facekaspacesnowu-liao-docker-sdkteiawoxue-chu/korshunov.aija2026-07-15T08:43:56.286139+00:00Hugging FaceがSpacesの無料Docker SDKティアを削除 https://korshunov.ai/fr/article/12152-hugging-face-supprime-le-niveau-gratuit-du-sdk-docker-pour-spaces/korshunov.aifr2026-07-15T08:43:56.286139+00:00Hugging Face supprime le niveau gratuit du SDK Docker pour Spaces https://korshunov.ai/id/article/12152-hugging-face-menghapus-tier-docker-sdk-gratis-untuk-spaces/korshunov.aiid2026-07-15T08:43:56.286139+00:00Hugging Face menghapus tier Docker SDK gratis untuk Spaces https://korshunov.ai/ar/article/12152-hugging-face-tzyl-tbq-docker-sdk-lmjny-mn-lmsht/korshunov.aiar2026-07-15T08:43:56.286139+00:00Hugging Face تزيل طبقة Docker SDK المجانية من المساحات https://korshunov.ai/en/article/12151-inquiry-about-multi-gpu-deployment-requirements-on-hugging-face-spaces/korshunov.aien2026-07-15T08:40:28.047973+00:00Inquiry about multi-GPU deployment requirements on Hugging Face Spaces https://korshunov.ai/ru/article/12151-zapros-o-trebovaniiakh-k-razvertyvaniiu-na-neskolkikh-gpu-v-hugging-face-spaces/korshunov.airu2026-07-15T08:40:28.047973+00:00Запрос о требованиях к развертыванию на нескольких GPU в Hugging Face Spaces https://korshunov.ai/es/article/12151-consulta-sobre-los-requisitos-de-implementacion-multi-gpu-en-hugging-face-spaces/korshunov.aies2026-07-15T08:40:28.047973+00:00Consulta sobre los requisitos de implementación multi-GPU en Hugging Face Spaces https://korshunov.ai/zh-hans/article/12151-guan-yu-zai-hugging-face-spaces-shang-bu-shu-duo-gpu-xu-qiu-de-zi-xun/korshunov.aizh-cn2026-07-15T08:40:28.047973+00:00关于在 Hugging Face Spaces 上部署多 GPU 需求的咨询 https://korshunov.ai/pt-br/article/12151-consulta-sobre-requisitos-de-implantacao-multi-gpu-no-hugging-face-spaces/korshunov.aipt2026-07-15T08:40:28.047973+00:00Consulta sobre requisitos de implantação multi-GPU no Hugging Face Spaces https://korshunov.ai/hi/article/12151-hugging-face-spaces-pr-mlttii-gpu-ddiplonymentt-kii-aavshyktaaon-ke-baare-men/korshunov.aihi2026-07-15T08:40:28.047973+00:00Hugging Face Spaces पर मल्टी-GPU डिप्लॉयमेंट की आवश्यकताओं के बारे में पूछताछ https://korshunov.ai/ja/article/12151-hugging-face-spacesniokerumaruchigputehuroimentonoyao-jian-niguan-suruwen-ihe/korshunov.aija2026-07-15T08:40:28.047973+00:00Hugging Face SpacesにおけるマルチGPUデプロイメントの要件に関する問い合わせ https://korshunov.ai/ko/article/12151-hugging-face-spaceseseoyi-meolti-gpu-baepo-yogu-sahang-munyi/korshunov.aiko2026-07-15T08:40:28.047973+00:00Hugging Face Spaces에서의 멀티 GPU 배포 요구 사항 문의 https://korshunov.ai/fr/article/12151-demande-concernant-les-exigences-de-deploiement-multi-gpu-sur-hugging-face/korshunov.aifr2026-07-15T08:40:28.047973+00:00Demande concernant les exigences de déploiement multi-GPU sur Hugging Face Spaces https://korshunov.ai/id/article/12151-pertanyaan-mengenai-persyaratan-penyebaran-multi-gpu-di-hugging-face-spaces/korshunov.aiid2026-07-15T08:40:28.047973+00:00Pertanyaan mengenai persyaratan penyebaran multi-GPU di Hugging Face Spaces https://korshunov.ai/ar/article/12151-stfsr-hwl-mttlbt-nshr-mt-dd-gpus-l-msht-hugging-face/korshunov.aiar2026-07-15T08:40:28.047973+00:00استفسار حول متطلبات نشر متعدد GPUs على مساحات Hugging Face https://korshunov.ai/en/article/12054-language-identification-with-succinct-machine-independent-traces/korshunov.aien2026-07-15T08:28:44.140816+00:00Language Identification with Succinct Machine-Independent Traces https://korshunov.ai/ru/article/12054-identifikatsiia-iazykov-s-pomoshchiu-lakonichnykh-mashinno-nezavisimykh-trass/korshunov.airu2026-07-15T08:28:44.140816+00:00Идентификация языков с помощью лаконичных машинно-независимых трасс https://korshunov.ai/es/article/12054-identificacion-de-idiomas-con-trazas-sucintas-e-independientes-de-la-maquina/korshunov.aies2026-07-15T08:28:44.140816+00:00Identificación de idiomas con trazas sucintas e independientes de la máquina https://korshunov.ai/zh-hans/article/12054-shi-yong-jian-ji-de-ji-qi-wu-guan-ji-xian-jin-xing-yu-yan-shi-bie/korshunov.aizh-cn2026-07-15T08:28:44.140816+00:00使用简洁的机器无关迹线进行语言识别 https://korshunov.ai/pt-br/article/12054-identificacao-de-idiomas-com-tracos-sucintos-e-independentes-da-maquina/korshunov.aipt2026-07-15T08:28:44.140816+00:00Identificação de idiomas com traços sucintos e independentes da máquina https://korshunov.ai/hi/article/12054-snkssipt-mshiin-svtntr-ttres-ke-saath-bhaassaa-phcaan/korshunov.aihi2026-07-15T08:28:44.140816+00:00संक्षिप्त मशीन-स्वतंत्र ट्रेस के साथ भाषा पहचान https://korshunov.ai/ja/article/12054-jian-jie-naji-jie-fei-yi-cun-tore-suniyoruyan-yu-shi-bie/korshunov.aija2026-07-15T08:28:44.140816+00:00簡潔な機械非依存トレースによる言語識別 https://korshunov.ai/ko/article/12054-gangyeolhan-gigye-dogribjeog-teureiseureul-tonghan-eoneo-sigbyeol/korshunov.aiko2026-07-15T08:28:44.140816+00:00간결한 기계 독립적 트레이스를 통한 언어 식별 https://korshunov.ai/fr/article/12054-identification-de-langue-avec-des-traces-succinctes-et-independantes-de-la/korshunov.aifr2026-07-15T08:28:44.140816+00:00Identification de langue avec des traces succinctes et indépendantes de la machine https://korshunov.ai/id/article/12054-identifikasi-bahasa-dengan-jejak-ringkas-yang-independen-dari-mesin/korshunov.aiid2026-07-15T08:28:44.140816+00:00Identifikasi Bahasa dengan Jejak Ringkas yang Independen dari Mesin https://korshunov.ai/ar/article/12054-thdyd-llg-bstkhdm-thr-mwjz-wmstql-n-ll/korshunov.aiar2026-07-15T08:28:44.140816+00:00تحديد اللغة باستخدام آثار موجزة ومستقلة عن الآلة https://korshunov.ai/en/article/12053-function-aware-fill-in-the-middle-mid-training-improves-coding-agent-foundation/korshunov.aien2026-07-15T08:24:09.020196+00:00Function-Aware Fill-in-the-Middle mid-training improves coding agent foundation models https://korshunov.ai/ru/article/12053-osoznannoe-zapolnenie-serediny-s-uchetom-funktsii-na-etape-promezhutochnogo-dlia/korshunov.airu2026-07-15T08:24:09.020196+00:00Осознанное заполнение середины с учётом функций на этапе промежуточного обучения улучшает базовые модели для кодовых агентов https://korshunov.ai/es/article/12053-el-relleno-intermedio-consciente-de-funciones-en-el-entrenamiento-intermedio-los/korshunov.aies2026-07-15T08:24:09.020196+00:00El relleno intermedio consciente de funciones en el entrenamiento intermedio mejora los modelos base para agentes de código https://korshunov.ai/zh-hans/article/12053-mian-xiang-han-shu-de-zhong-jian-tian-chong-zhong-qi-xun-lian-ke-ti-sheng-dai-ma/korshunov.aizh-cn2026-07-15T08:24:09.020196+00:00面向函数的中间填充中期训练可提升代码代理基础模型 https://korshunov.ai/pt-br/article/12053-preenchimento-do-meio-ciente-de-funcoes-no-treinamento-intermediario-melhora-de/korshunov.aipt2026-07-15T08:24:09.020196+00:00Preenchimento do meio ciente de funções no treinamento intermediário melhora modelos base para agentes de código https://korshunov.ai/hi/article/12053-phnkshn-aveyr-phil-in-d-middl-midd-ttrening-kodding-ejentt-phaaunddeshn-monddls/korshunov.aihi2026-07-15T08:24:09.020196+00:00फ़ंक्शन-अवेयर फ़िल-इन-द-मिडल मिड-ट्रेनिंग कोडिंग एजेंट फाउंडेशन मॉडल्स को बेहतर बनाती है https://korshunov.ai/ja/article/12053-guan-shu-ren-shi-xing-huirinku-in-sa-mitorunozhong-jian-tore-ninkukako-teinkue/korshunov.aija2026-07-15T08:24:09.020196+00:00関数認識型フィリング・イン・ザ・ミドルの中間トレーニングがコーディングエージェント基盤モデルを改善 https://korshunov.ai/ko/article/12053-hamsu-insig-junggan-caeugi-mid-training-ga-koding-eijeonteu-paundeisyeon/korshunov.aiko2026-07-15T08:24:09.020196+00:00함수 인식 중간 채우기(mid-training)가 코딩 에이전트 파운데이션 모델을 개선합니다 https://korshunov.ai/fr/article/12053-le-mid-training-aware-des-fonctions-fim-ameliore-les-modeles-de-base-pour-agents/korshunov.aifr2026-07-15T08:24:09.020196+00:00Le mid-training aware des fonctions (FIM) améliore les modèles de base pour agents de codage https://korshunov.ai/id/article/12053-mid-training-fill-in-the-middle-yang-sadar-fungsi-meningkatkan-model-fondasi/korshunov.aiid2026-07-15T08:24:09.020196+00:00Mid-training Fill-in-the-Middle yang Sadar Fungsi Meningkatkan Model Fondasi Agen Pemrograman https://korshunov.ai/ar/article/12053-ltdryb-lwsyt-lml-lwst-lw-y-blwzyf-yhswn-nmdhj-lss-lwkl-lbrmj/korshunov.aiar2026-07-15T08:24:09.020196+00:00التدريب الوسيط لملء الوسط الواعي بالوظائف يحسّن نماذج الأساس لوكلاء البرمجة https://korshunov.ai/ja/article/12051-fan-yi-he-sunohuainchiyu-ninkukadi-zi-yuan-yan-yu-teshi-xian-ke-neng-suyan-jiu/korshunov.aija2026-07-15T08:01:51.870451+00:00翻訳ベースのファインチューニングが低資源言語で実現可能であることを示す研究 https://korshunov.ai/ko/article/12051-beonyeog-giban-paintyuningi-jeojaweon-eoneoe-silhyeon-ganeunghameul-boyeojuneun/korshunov.aiko2026-07-15T08:01:51.870451+00:00번역 기반 파인튜닝이 저자원 언어에 실현 가능함을 보여주는 연구 https://korshunov.ai/fr/article/12051-une-etude-montre-que-le-fine-tuning-base-sur-la-traduction-est-realisable-pour/korshunov.aifr2026-07-15T08:01:51.870451+00:00Une étude montre que le fine-tuning basé sur la traduction est réalisable pour les langues peu dotées https://korshunov.ai/id/article/12051-studi-menemukan-bahwa-penalaan-halus-berbasis-terjemahan-layak-untuk-bahasa-daya/korshunov.aiid2026-07-15T08:01:51.870451+00:00Studi menemukan bahwa penalaan halus berbasis terjemahan layak untuk bahasa dengan sumber daya rendah https://korshunov.ai/ar/article/12051-drs-tjd-n-ldbt-ldqyq-lqym-l-ltrjm-mmkn-llgt-mnkhfd-lmwrd/korshunov.aiar2026-07-15T08:01:51.870451+00:00دراسة تجد أن الضبط الدقيق القائم على الترجمة ممكن للغات منخفضة الموارد https://korshunov.ai/hi/article/12051-adhyyn-paataa-hai-ki-anuvaad-aadhaarit-phaain-ttyuuning-km-snsaadhn-vaale-ke-lie/korshunov.aihi2026-07-15T08:01:51.870451+00:00अध्ययन पाता है कि अनुवाद-आधारित फाइन-ट्यूनिंग कम संसाधन वाले भाषाओं के लिए संभव है https://korshunov.ai/en/article/12051-study-finds-translation-based-fine-tuning-feasible-for-low-resource-languages/korshunov.aien2026-07-15T08:01:51.870451+00:00Study finds translation-based fine-tuning feasible for low-resource languages https://korshunov.ai/ru/article/12051-issledovanie-pokazyvaet-chto-doobuchenie-na-osnove-perevoda-vypolnimo-dlia/korshunov.airu2026-07-15T08:01:51.870451+00:00Исследование показывает, что дообучение на основе перевода выполнимо для малоресурсных языков https://korshunov.ai/es/article/12051-un-estudio-encuentra-que-el-ajuste-fino-basado-en-traduccion-es-viable-para-con/korshunov.aies2026-07-15T08:01:51.870451+00:00Un estudio encuentra que el ajuste fino basado en traducción es viable para idiomas con pocos recursos https://korshunov.ai/zh-hans/article/12051-yan-jiu-fa-xian-ji-yu-fan-yi-de-wei-diao-dui-di-zi-yuan-yu-yan-ke-xing/korshunov.aizh-cn2026-07-15T08:01:51.870451+00:00研究发现基于翻译的微调对低资源语言可行 https://korshunov.ai/pt-br/article/12051-estudo-encontra-o-ajuste-fino-baseado-em-traducao-viavel-para-idiomas-com-poucos/korshunov.aipt2026-07-15T08:01:51.870451+00:00Estudo encontra o ajuste fino baseado em tradução viável para idiomas com poucos recursos https://korshunov.ai/en/article/12050-knowact-guiclaw-improves-gui-agent-accuracy-and-cross-platform-adaptability/korshunov.aien2026-07-15T07:53:58.776800+00:00KnowAct-GUIClaw improves GUI agent accuracy and cross-platform adaptability https://korshunov.ai/ru/article/12050-knowact-guiclaw-povyshaet-tochnost-gui-agenta-i-krossplatformennuiu-adaptivnost/korshunov.airu2026-07-15T07:53:58.776800+00:00KnowAct-GUIClaw повышает точность GUI-агента и кроссплатформенную адаптивность https://korshunov.ai/es/article/12050-knowact-guiclaw-mejora-la-precision-del-agente-gui-y-la-adaptabilidad/korshunov.aies2026-07-15T07:53:58.776800+00:00KnowAct-GUIClaw mejora la precisión del agente GUI y la adaptabilidad multiplataforma https://korshunov.ai/zh-hans/article/12050-knowact-guiclaw-ti-sheng-gui-dai-li-de-zhun-que-xing-he-kua-ping-tai-gua-ying/korshunov.aizh-cn2026-07-15T07:53:58.776800+00:00KnowAct-GUIClaw 提升 GUI 代理的准确性和跨平台适应性 https://korshunov.ai/pt-br/article/12050-knowact-guiclaw-melhora-a-precisao-do-agente-gui-e-a-adaptabilidade/korshunov.aipt2026-07-15T07:53:58.776800+00:00KnowAct-GUIClaw melhora a precisão do agente GUI e a adaptabilidade multiplataforma https://korshunov.ai/hi/article/12050-knowact-guiclaw-gui-ejentt-kii-sttiiktaa-aur-krons-plettphonrm-anukuuln-ko-hai/korshunov.aihi2026-07-15T07:53:58.776800+00:00KnowAct-GUIClaw GUI एजेंट की सटीकता और क्रॉस-प्लेटफ़ॉर्म अनुकूलन को बढ़ाता है https://korshunov.ai/ja/article/12050-knowact-guiclawkaguie-shientonojing-du-tokurosuhuratsutohuo-mushi-ying-xing/korshunov.aija2026-07-15T07:53:58.776800+00:00KnowAct-GUIClawがGUIエージェントの精度とクロスプラットフォーム適応性を向上 https://korshunov.ai/ko/article/12050-knowact-guiclawga-gui-eijeonteuyi-jeonghwagseonggwa-keuroseu-peulraespom/korshunov.aiko2026-07-15T07:53:58.776800+00:00KnowAct-GUIClaw가 GUI 에이전트의 정확성과 크로스 플랫폼 적응성을 향상시킵니다 https://korshunov.ai/fr/article/12050-knowact-guiclaw-ameliore-la-precision-de-l-agent-gui-et-l-adaptabilite/korshunov.aifr2026-07-15T07:53:58.776800+00:00KnowAct-GUIClaw améliore la précision de l'agent GUI et l'adaptabilité interplateforme https://korshunov.ai/id/article/12050-knowact-guiclaw-meningkatkan-akurasi-agen-gui-dan-adaptabilitas-lintas-platform/korshunov.aiid2026-07-15T07:53:58.776800+00:00KnowAct-GUIClaw meningkatkan akurasi agen GUI dan adaptabilitas lintas-platform https://korshunov.ai/ar/article/12050-knowact-guiclaw-yhswn-dq-wkyl-wjh-lmstkhdm-lrswmy-wqdrth-l-ltkyf-br-lmnst/korshunov.aiar2026-07-15T07:53:58.776800+00:00KnowAct-GUIClaw يحسّن دقة وكيل واجهة المستخدم الرسومية وقدرته على التكيف عبر المنصات https://korshunov.ai/en/article/12049-study-finds-induced-anger-reduces-llm-penalty-sensitivity-in-sequential/korshunov.aien2026-07-15T07:44:29.951610+00:00Study finds induced anger reduces LLM penalty sensitivity in sequential decisions https://korshunov.ai/ru/article/12049-issledovanie-vyiavliaet-vyzvannyi-gnev-snizhaet-chuvstvitelnost-llm-k-shtrafam-v/korshunov.airu2026-07-15T07:44:29.951610+00:00Исследование выявляет: вызванный гнев снижает чувствительность LLM к штрафам в последовательных решениях https://korshunov.ai/es/article/12049-un-estudio-encuentra-que-la-ira-inducida-reduce-la-sensibilidad-a-las-de-los-llm/korshunov.aies2026-07-15T07:44:29.951610+00:00Un estudio encuentra que la ira inducida reduce la sensibilidad a las penalizaciones de los LLM en decisiones secuenciales https://korshunov.ai/zh-hans/article/12049-yan-jiu-fa-xian-you-dao-de-fen-nu-jiang-di-liao-da-xing-yu-yan-mo-xing-zai-xu-ce/korshunov.aizh-cn2026-07-15T07:44:29.951610+00:00研究发现诱导的愤怒降低了大型语言模型在序列决策中对惩罚的敏感度 https://korshunov.ai/pt-br/article/12049-estudo-descobre-que-a-raiva-induzida-reduz-a-sensibilidade-a-penalidades-de-llms/korshunov.aipt2026-07-15T07:44:29.951610+00:00Estudo descobre que a raiva induzida reduz a sensibilidade a penalidades de LLMs em decisões sequenciais https://korshunov.ai/hi/article/12049-adhyyn-paataa-hai-ki-prerit-gusse-se-krmik-nirnnyon-men-llm-kii-dndd-km-ho-hai/korshunov.aihi2026-07-15T07:44:29.951610+00:00अध्ययन पाता है कि प्रेरित गुस्से से क्रमिक निर्णयों में LLM की दंड संवेदनशीलता कम हो जाती है https://korshunov.ai/ja/article/12049-you-fa-saretanu-rikallmnozhu-ci-yi-si-jue-ding-niokeruhenaruteigan-du-wodi-xia/korshunov.aija2026-07-15T07:44:29.951610+00:00誘発された怒りがLLMの逐次意思決定におけるペナルティ感度を低下させることを発見 https://korshunov.ai/ko/article/12049-yudodoen-bunnoga-suncajeog-yisagyeoljeongeseo-llmyi-peneolti-mingamdoreul-yeongu/korshunov.aiko2026-07-15T07:44:29.951610+00:00유도된 분노가 순차적 의사결정에서 LLM의 페널티 민감도를 감소시킨다는 연구 https://korshunov.ai/fr/article/12049-une-etude-revele-que-la-colere-induite-reduit-la-sensibilite-aux-penalites-des/korshunov.aifr2026-07-15T07:44:29.951610+00:00Une étude révèle que la colère induite réduit la sensibilité aux pénalités des LLM dans les décisions séquentielles https://korshunov.ai/id/article/12049-studi-menemukan-bahwa-kemarahan-yang-diinduksi-mengurangi-sensitivitas-penalti/korshunov.aiid2026-07-15T07:44:29.951610+00:00Studi menemukan bahwa kemarahan yang diinduksi mengurangi sensitivitas penalti LLM dalam keputusan berurutan https://korshunov.ai/ar/article/12049-drs-tjd-n-lgdb-lmuhfawz-yqll-mn-hssy-l-qwbt-fy-qrrt-nmdhj-llgt-lkbyr-lmtslsl/korshunov.aiar2026-07-15T07:44:29.951610+00:00دراسة تجد أن الغضب المُحفَّز يقلل من حساسية العقوبات في قرارات نماذج اللغات الكبيرة المتسلسلة https://korshunov.ai/en/article/12048-grpo-fails-to-improve-small-web-agents-unless-tasks-have-headroom/korshunov.aien2026-07-15T07:38:15.404894+00:00GRPO fails to improve small web agents unless tasks have headroom https://korshunov.ai/ru/article/12048-grpo-ne-uluchshaet-malye-veb-agenty-esli-u-zadach-est-zapas/korshunov.airu2026-07-15T07:38:15.404894+00:00GRPO не улучшает малые веб-агенты, если у задач есть запас https://korshunov.ai/es/article/12048-grpo-no-mejora-a-los-pequenos-agentes-web-a-menos-que-las-tareas-tengan-margen/korshunov.aies2026-07-15T07:38:15.404894+00:00GRPO no mejora a los pequeños agentes web a menos que las tareas tengan margen https://korshunov.ai/zh-hans/article/12048-chu-fei-ren-wu-you-ti-sheng-kong-jian-fou-ze-grpowu-fa-gai-jin-xiao-xing-webdai/korshunov.aizh-cn2026-07-15T07:38:15.404894+00:00除非任务有提升空间，否则GRPO无法改进小型Web代理 https://korshunov.ai/pt-br/article/12048-grpo-falha-em-melhorar-pequenos-agentes-web-a-menos-que-as-tarefas-tenham-margem/korshunov.aipt2026-07-15T07:38:15.404894+00:00GRPO falha em melhorar pequenos agentes web a menos que as tarefas tenham margem https://korshunov.ai/hi/article/12048-grpo-tb-tk-chotte-veb-ejentts-men-sudhaar-nhiin-krtaa-jb-tk-kaaryon-men-jgh-n-ho/korshunov.aihi2026-07-15T07:38:15.404894+00:00GRPO तब तक छोटे वेब एजेंट्स में सुधार नहीं करता जब तक कार्यों में जगह न हो https://korshunov.ai/ja/article/12048-tasukuniyu-li-kanaixian-rigrpohaxiao-xing-uehue-shientonoxing-neng-woxiang-shang/korshunov.aija2026-07-15T07:38:15.404894+00:00タスクに余力がない限りGRPOは小型ウェブエージェントの性能を向上させない https://korshunov.ai/ko/article/12048-jageobe-yeoyuga-eobsneun-han-grponeun-jageun-web-eijeonteuyi-seongneungeul/korshunov.aiko2026-07-15T07:38:15.404894+00:00작업에 여유가 없는 한 GRPO는 작은 웹 에이전트의 성능을 개선하지 못함 https://korshunov.ai/fr/article/12048-grpo-echoue-a-ameliorer-les-petits-agents-web-sauf-si-les-taches-ont-une-marge/korshunov.aifr2026-07-15T07:38:15.404894+00:00GRPO échoue à améliorer les petits agents web sauf si les tâches ont une marge de progression https://korshunov.ai/id/article/12048-grpo-gagal-meningkatkan-agen-web-kecil-kecuali-tugas-memiliki-ruang-untuk/korshunov.aiid2026-07-15T07:38:15.404894+00:00GRPO gagal meningkatkan agen web kecil kecuali tugas memiliki ruang untuk berkembang https://korshunov.ai/ar/article/12048-yfshl-grpo-fy-thsyn-wkl-lwyb-lsgyr-m-lm-tkn-lmhm-ldyh-hmsh-llttwyr/korshunov.aiar2026-07-15T07:38:15.404894+00:00يفشل GRPO في تحسين وكلاء الويب الصغيرة ما لم تكن المهام لديها هامش للتطوير https://korshunov.ai/en/article/12047-formalizing-matched-comparisons-for-rigorous-extractable-memorization-claims/korshunov.aien2026-07-15T07:29:23.903751+00:00Formalizing matched comparisons for rigorous extractable memorization claims https://korshunov.ai/ru/article/12047-formalizatsiia-sopostavlennykh-sravnenii-dlia-strogikh-utverzhdenii-o/korshunov.airu2026-07-15T07:29:23.903751+00:00Формализация сопоставленных сравнений для строгих утверждений о извлекаемой запоминаемости https://korshunov.ai/es/article/12047-formalizando-comparaciones-emparejadas-para-afirmaciones-rigurosas-de-extraible/korshunov.aies2026-07-15T07:29:23.903751+00:00Formalizando comparaciones emparejadas para afirmaciones rigurosas de memorización extraíble https://korshunov.ai/zh-hans/article/12047-xing-shi-hua-pi-pei-bi-jiao-yi-zhi-chi-yan-ge-de-ke-ti-qu-ji-yi-sheng-ming/korshunov.aizh-cn2026-07-15T07:29:23.903751+00:00形式化匹配比较以支持严格的可提取记忆声明 https://korshunov.ai/pt-br/article/12047-formalizando-comparacoes-pareadas-para-reivindicacoes-rigorosas-de-memorizacao/korshunov.aipt2026-07-15T07:29:23.903751+00:00Formalizando comparações pareadas para reivindicações rigorosas de memorização extraível https://korshunov.ai/hi/article/12047-ktthor-vyutpnn-smrti-daavon-ke-lie-mel-khaate-tulnon-kaa-aupcaarikiikrnn/korshunov.aihi2026-07-15T07:29:23.903751+00:00कठोर व्युत्पन्न स्मृति दावों के लिए मेल खाते तुलनों का औपचारिकीकरण https://korshunov.ai/ja/article/12047-yan-mi-nachou-chu-ke-neng-ji-yi-nozhu-zhang-notamenoyi-zhi-bi-jiao-noxing-shi/korshunov.aija2026-07-15T07:29:23.903751+00:00厳密な抽出可能記憶の主張のための一致比較の形式化 https://korshunov.ai/ko/article/12047-eomgyeoghan-cucul-ganeung-gieog-jujangeul-wihan-ilci-bigyoyi-hyeongsighwa/korshunov.aiko2026-07-15T07:29:23.903751+00:00엄격한 추출 가능 기억 주장을 위한 일치 비교의 형식화 https://korshunov.ai/fr/article/12047-formaliser-les-comparaisons-appariees-pour-des-affirmations-rigoureuses-sur-la/korshunov.aifr2026-07-15T07:29:23.903751+00:00Formaliser les comparaisons appariées pour des affirmations rigoureuses sur la mémorisation extractible https://korshunov.ai/id/article/12047-memformalkan-perbandingan-yang-cocok-untuk-klaim-memorisasi-yang-dapat-secara/korshunov.aiid2026-07-15T07:29:23.903751+00:00Memformalkan perbandingan yang cocok untuk klaim memorisasi yang dapat diekstraksi secara ketat https://korshunov.ai/ar/article/12047-twhyd-lmqrnt-lmttbq-ld-t-lhfz-lqbl-llstkhrj-bdq/korshunov.aiar2026-07-15T07:29:23.903751+00:00توحيد المقارنات المتطابقة لادعاءات الحفظ القابلة للاستخراج بدقة https://korshunov.ai/en/article/12046-serin-decomposes-multimodal-fusion-for-sentiment-analysis/korshunov.aien2026-07-15T07:19:29.233985+00:00SeRIn decomposes multimodal fusion for sentiment analysis https://korshunov.ai/ko/article/12046-serineun-gamjeong-bunseogeul-wihan-meoltimodal-yunghabeul-bunhaehabnida/korshunov.aiko2026-07-15T07:19:29.233985+00:00SeRIn은 감정 분석을 위한 멀티모달 융합을 분해합니다 https://korshunov.ai/ru/article/12046-serin-razdeliaet-multimodalnuiu-integratsiiu-dlia-analiza-tonalnosti/korshunov.airu2026-07-15T07:19:29.233985+00:00SeRIn разделяет мультимодальную интеграцию для анализа тональности https://korshunov.ai/es/article/12046-serin-descompone-la-fusion-multimodal-para-el-analisis-de-sentimiento/korshunov.aies2026-07-15T07:19:29.233985+00:00SeRIn descompone la fusión multimodal para el análisis de sentimiento https://korshunov.ai/zh-hans/article/12046-serin-fen-jie-duo-mo-tai-rong-he-yi-jin-xing-qing-gan-fen-xi/korshunov.aizh-cn2026-07-15T07:19:29.233985+00:00SeRIn 分解多模态融合以进行情感分析 https://korshunov.ai/pt-br/article/12046-serin-decompoe-a-fusao-multimodal-para-analise-de-sentimento/korshunov.aipt2026-07-15T07:19:29.233985+00:00SeRIn decompõe a fusão multimodal para análise de sentimento https://korshunov.ai/hi/article/12046-serin-bhaavnaa-vishlessnn-ke-lie-bhuaayaamii-phyuujn-ko-vighttit-krtaa-hai/korshunov.aihi2026-07-15T07:19:29.233985+00:00SeRIn भावना विश्लेषण के लिए बहुआयामी फ्यूजन को विघटित करता है https://korshunov.ai/ja/article/12046-serinhagan-qing-fen-xi-notamenomaruchimo-tarurong-he-wofen-jie-suru/korshunov.aija2026-07-15T07:19:29.233985+00:00SeRInは感情分析のためのマルチモーダル融合を分解する https://korshunov.ai/fr/article/12046-serin-decompose-la-fusion-multimodale-pour-l-analyse-des-sentiments/korshunov.aifr2026-07-15T07:19:29.233985+00:00SeRIn décompose la fusion multimodale pour l'analyse des sentiments https://korshunov.ai/id/article/12046-serin-mengurai-fusi-multimodal-untuk-analisis-sentimen/korshunov.aiid2026-07-15T07:19:29.233985+00:00SeRIn mengurai fusi multimodal untuk analisis sentimen https://korshunov.ai/ar/article/12046-yqwm-serin-bthlyl-lndmj-mt-dd-lwsyt-lthlyl-lmsh-r/korshunov.aiar2026-07-15T07:19:29.233985+00:00يقوم SeRIn بتحليل الاندماج متعدد الوسائط لتحليل المشاعر https://korshunov.ai/en/article/12149-how-to-run-prism-bonsai-27b/korshunov.aien2026-07-15T07:13:28.861871+00:00How to run Prism Bonsai 27B https://korshunov.ai/ru/article/12149-kak-zapustit-prism-bonsai-27b/korshunov.airu2026-07-15T07:13:28.861871+00:00Как запустить Prism Bonsai 27B https://korshunov.ai/es/article/12149-como-ejecutar-prism-bonsai-27b/korshunov.aies2026-07-15T07:13:28.861871+00:00Cómo ejecutar Prism Bonsai 27B https://korshunov.ai/zh-hans/article/12149-ru-he-yun-xing-prism-bonsai-27b/korshunov.aizh-cn2026-07-15T07:13:28.861871+00:00如何运行 Prism Bonsai 27B https://korshunov.ai/pt-br/article/12149-como-executar-o-prism-bonsai-27b/korshunov.aipt2026-07-15T07:13:28.861871+00:00Como executar o Prism Bonsai 27B https://korshunov.ai/hi/article/12149-prism-bonsai-27b-ko-kaise-claaen/korshunov.aihi2026-07-15T07:13:28.861871+00:00Prism Bonsai 27B को कैसे चलाएं https://korshunov.ai/ja/article/12149-prism-bonsai-27bnoshi-xing-fang-fa/korshunov.aija2026-07-15T07:13:28.861871+00:00Prism Bonsai 27Bの実行方法 https://korshunov.ai/ko/article/12149-prism-bonsai-27b-silhaeng-bangbeob/korshunov.aiko2026-07-15T07:13:28.861871+00:00Prism Bonsai 27B 실행 방법 https://korshunov.ai/fr/article/12149-comment-executer-prism-bonsai-27b/korshunov.aifr2026-07-15T07:13:28.861871+00:00Comment exécuter Prism Bonsai 27B https://korshunov.ai/id/article/12149-cara-menjalankan-prism-bonsai-27b/korshunov.aiid2026-07-15T07:13:28.861871+00:00Cara menjalankan Prism Bonsai 27B https://korshunov.ai/ar/article/12149-kyfy-tshgyl-prism-bonsai-27b/korshunov.aiar2026-07-15T07:13:28.861871+00:00كيفية تشغيل Prism Bonsai 27B https://korshunov.ai/en/article/12045-care-ppo-enables-joint-learning-of-accurate-predictions-and-reliable-confidence/korshunov.aien2026-07-15T07:04:44.176015+00:00CARE-PPO enables joint learning of accurate predictions and reliable confidence in language-based quantitative tasks https://korshunov.ai/ru/article/12045-care-ppo-obespechivaet-sovmestnoe-obuchenie-tochnym-predskazaniiam-i-nadezhnoi-v/korshunov.airu2026-07-15T07:04:44.176015+00:00CARE-PPO обеспечивает совместное обучение точным предсказаниям и надёжной уверенности в языковых количественных задачах https://korshunov.ai/es/article/12045-care-ppo-permite-el-aprendizaje-conjunto-de-predicciones-precisas-y-confianza-en/korshunov.aies2026-07-15T07:04:44.176015+00:00CARE-PPO permite el aprendizaje conjunto de predicciones precisas y confianza confiable en tareas cuantitativas basadas en lenguaje https://korshunov.ai/zh-hans/article/12045-care-ppo-shi-xian-yu-yan-lei-ding-liang-ren-wu-zhong-zhun-que-yu-ce-yu-ke-kao-du/korshunov.aizh-cn2026-07-15T07:04:44.176015+00:00CARE-PPO 实现语言类定量任务中准确预测与可靠置信度的联合学习 https://korshunov.ai/pt-br/article/12045-care-ppo-permite-o-aprendizado-conjunto-de-previsoes-precisas-e-confianca-em-em/korshunov.aipt2026-07-15T07:04:44.176015+00:00CARE-PPO permite o aprendizado conjunto de previsões precisas e confiança confiável em tarefas quantitativas baseadas em linguagem https://korshunov.ai/hi/article/12045-care-ppo-bhaassaa-aadhaarit-maatraatmk-kaaryon-men-sttiik-bhvissyvaanniyon-aur/korshunov.aihi2026-07-15T07:04:44.176015+00:00CARE-PPO भाषा-आधारित मात्रात्मक कार्यों में सटीक भविष्यवाणियों और विश्वसनीय आत्मविश्वास का संयुक्त शिक्षण सक्षम बनाता है https://korshunov.ai/ja/article/12045-care-ppohayan-yu-he-sunoding-liang-de-tasukunioitezheng-que-nayu-ce-toxin-lai-xi/korshunov.aija2026-07-15T07:04:44.176015+00:00CARE-PPOは言語ベースの定量的タスクにおいて正確な予測と信頼できる自信の共同学習を可能にする https://korshunov.ai/ko/article/12045-care-pponeun-eoneo-giban-jeongryangjeog-jageobeseo-jeonghwaghan-yeceuggwa-su-ham/korshunov.aiko2026-07-15T07:04:44.176015+00:00CARE-PPO는 언어 기반 정량적 작업에서 정확한 예측과 신뢰할 수 있는 확신의 공동 학습을 가능하게 함 https://korshunov.ai/fr/article/12045-care-ppo-permet-l-apprentissage-conjoint-de-predictions-precises-et-d-une-fiable/korshunov.aifr2026-07-15T07:04:44.176015+00:00CARE-PPO permet l'apprentissage conjoint de prédictions précises et d'une confiance fiable dans les tâches quantitatives basées sur le langage https://korshunov.ai/id/article/12045-care-ppo-memungkinkan-pembelajaran-gabungan-prediksi-akurat-dan-kepercayaan-yang/korshunov.aiid2026-07-15T07:04:44.176015+00:00CARE-PPO memungkinkan pembelajaran gabungan prediksi akurat dan kepercayaan yang andal dalam tugas kuantitatif berbasis bahasa https://korshunov.ai/ar/article/12045-ytyh-care-ppo-lt-lm-lmshtrk-lltnbwt-ldqyq-wlthq-lmwthwq-fy-lmhm-lkmy-lqym-l-llg/korshunov.aiar2026-07-15T07:04:44.176015+00:00يتيح CARE-PPO التعلم المشترك للتنبؤات الدقيقة والثقة الموثوقة في المهام الكمية القائمة على اللغة https://korshunov.ai/en/article/12044-ecospec-reduces-active-experts-and-speeds-up-decoding-in-sparse-moe-models/korshunov.aien2026-07-15T06:55:13.181969+00:00EcoSpec reduces active experts and speeds up decoding in sparse MoE models https://korshunov.ai/ru/article/12044-ecospec-snizhaet-kolichestvo-aktivnykh-ekspertov-i-uskoriaet-dekodirovanie-v-moe/korshunov.airu2026-07-15T06:55:13.181969+00:00EcoSpec снижает количество активных экспертов и ускоряет декодирование в разреженных моделях MoE https://korshunov.ai/es/article/12044-ecospec-reduce-los-expertos-activos-y-acelera-la-decodificacion-en-modelos-moe/korshunov.aies2026-07-15T06:55:13.181969+00:00EcoSpec reduce los expertos activos y acelera la decodificación en modelos MoE dispersos https://korshunov.ai/zh-hans/article/12044-ecospecjian-shao-xi-shu-moemo-xing-zhong-de-huo-yue-zhuan-jia-shu-liang-bing-jia/korshunov.aizh-cn2026-07-15T06:55:13.181969+00:00EcoSpec减少稀疏MoE模型中的活跃专家数量并加速解码 https://korshunov.ai/pt-br/article/12044-ecospec-reduz-especialistas-ativos-e-acelera-a-decodificacao-em-modelos-moe/korshunov.aipt2026-07-15T06:55:13.181969+00:00EcoSpec reduz especialistas ativos e acelera a decodificação em modelos MoE esparsos https://korshunov.ai/hi/article/12044-ecospec-virl-moe-monddlon-men-skriy-vishessjnyon-ko-km-krtaa-hai-aur-ddikodding/korshunov.aihi2026-07-15T06:55:13.181969+00:00EcoSpec विरल MoE मॉडलों में सक्रिय विशेषज्ञों को कम करता है और डिकोडिंग को तेज़ करता है https://korshunov.ai/ja/article/12044-ecospechashu-namoemoteruniokeruakuteihunaekisuha-towoxue-jian-shi-teko-su-hua/korshunov.aija2026-07-15T06:55:13.181969+00:00EcoSpecは疎なMoEモデルにおけるアクティブなエキスパートを削減し、デコーディングを高速化する https://korshunov.ai/ko/article/12044-ecospeceun-hyiso-moe-modeleseo-hwalseong-jeonmungareul-juligo-dikoding-sogdoreul/korshunov.aiko2026-07-15T06:55:13.181969+00:00EcoSpec은 희소 MoE 모델에서 활성 전문가를 줄이고 디코딩 속도를 높입니다 https://korshunov.ai/fr/article/12044-ecospec-reduit-les-experts-actifs-et-accelere-le-decodage-dans-les-modeles-moe/korshunov.aifr2026-07-15T06:55:13.181969+00:00EcoSpec réduit les experts actifs et accélère le décodage dans les modèles MoE épars https://korshunov.ai/id/article/12044-ecospec-mengurangi-ahli-aktif-dan-mempercepat-dekoding-pada-model-moe-jarang/korshunov.aiid2026-07-15T06:55:13.181969+00:00EcoSpec mengurangi ahli aktif dan mempercepat dekoding pada model MoE jarang https://korshunov.ai/ar/article/12044-yqll-ecospec-mn-lkhbr-lnshtyn-wysr-fk-ltshfyr-fy-nmdhj-moe-lmtnthr/korshunov.aiar2026-07-15T06:55:13.181969+00:00يقلل EcoSpec من الخبراء النشطين ويسرع فك التشفير في نماذج MoE المتناثرة https://korshunov.ai/en/article/12043-esfp-benchmark-reveals-epistemic-stance-flexibility-in-large-language-models/korshunov.aien2026-07-15T06:47:24.784473+00:00ESFP benchmark reveals epistemic stance flexibility in large language models https://korshunov.ai/zh-hans/article/12043-esfpji-zhun-ce-shi-jie-shi-da-yu-yan-mo-xing-zai-ren-shi-lun-li-chang-shang-de/korshunov.aizh-cn2026-07-15T06:47:24.784473+00:00ESFP基准测试揭示大语言模型在认识论立场上的灵活性 https://korshunov.ai/ja/article/12043-esfphenchima-kukada-gui-mo-yan-yu-moteruniokeruren-shi-de-li-chang-norou-ruan/korshunov.aija2026-07-15T06:47:24.784473+00:00ESFPベンチマークが大規模言語モデルにおける認識的立場の柔軟性を明らかにする https://korshunov.ai/ru/article/12043-benchmark-esfp-vyiavliaet-gibkost-epistemicheskoi-pozitsii-v-bolshikh-iazykovykh/korshunov.airu2026-07-15T06:47:24.784473+00:00Бенчмарк ESFP выявляет гибкость эпистемической позиции в больших языковых моделях https://korshunov.ai/es/article/12043-el-benchmark-esfp-revela-flexibilidad-en-la-postura-epistemica-de-los-grandes-de/korshunov.aies2026-07-15T06:47:24.784473+00:00El benchmark ESFP revela flexibilidad en la postura epistémica de los grandes modelos de lenguaje https://korshunov.ai/pt-br/article/12043-benchmark-esfp-revela-flexibilidade-de-postura-epistemica-em-grandes-modelos-de/korshunov.aipt2026-07-15T06:47:24.784473+00:00Benchmark ESFP revela flexibilidade de postura epistêmica em grandes modelos de linguagem https://korshunov.ai/hi/article/12043-esfp-bencmaark-bdde-bhaassaa-monddlon-men-jnyaanvaadii-sthiti-lciilepn-ko-ujaagr/korshunov.aihi2026-07-15T06:47:24.784473+00:00ESFP बेंचमार्क बड़े भाषा मॉडलों में ज्ञानवादी स्थिति लचीलेपन को उजागर करता है https://korshunov.ai/ko/article/12043-esfp-bencimakeuneun-daegyumo-eoneo-modelyi-insigjeog-taedo-yuyeonseongeul/korshunov.aiko2026-07-15T06:47:24.784473+00:00ESFP 벤치마크는 대규모 언어 모델의 인식적 태도 유연성을 드러냅니다 https://korshunov.ai/fr/article/12043-le-benchmark-esfp-revele-la-flexibilite-epistemique-des-grands-modeles-de/korshunov.aifr2026-07-15T06:47:24.784473+00:00Le benchmark ESFP révèle la flexibilité épistémique des grands modèles de langage https://korshunov.ai/id/article/12043-benchmark-esfp-mengungkap-fleksibilitas-sikap-epistemik-dalam-model-bahasa-besar/korshunov.aiid2026-07-15T06:47:24.784473+00:00Benchmark ESFP mengungkap fleksibilitas sikap epistemik dalam model bahasa besar https://korshunov.ai/ar/article/12043-ykshf-mqys-esfp-n-mrwn-lmwqf-lm-rfy-fy-lnmdhj-llgwy-lkbyr/korshunov.aiar2026-07-15T06:47:24.784473+00:00يكشف مقياس ESFP عن مرونة الموقف المعرفي في النماذج اللغوية الكبيرة https://korshunov.ai/en/article/12042-oat-uses-one-class-learning-to-trace-agentic-failure-from-success-flows/korshunov.aien2026-07-15T06:38:22.979794+00:00OAT uses one-class learning to trace agentic failure from success flows https://korshunov.ai/ru/article/12042-oat-ispolzuet-obuchenie-s-odnim-klassom-dlia-otslezhivaniia-agentnykh-sboev-na/korshunov.airu2026-07-15T06:38:22.979794+00:00OAT использует обучение с одним классом для отслеживания агентных сбоев на основе успешных траекторий https://korshunov.ai/es/article/12042-oat-utiliza-aprendizaje-de-una-clase-para-rastrear-fallos-agenticos-a-partir-de/korshunov.aies2026-07-15T06:38:22.979794+00:00OAT utiliza aprendizaje de una clase para rastrear fallos agénticos a partir de flujos exitosos https://korshunov.ai/zh-hans/article/12042-oat-shi-yong-dan-lei-xue-xi-cong-cheng-gong-liu-cheng-zhong-zhui-zong-zhi-neng/korshunov.aizh-cn2026-07-15T06:38:22.979794+00:00OAT 使用单类学习从成功流程中追踪智能体故障 https://korshunov.ai/pt-br/article/12042-oat-usa-aprendizado-de-uma-classe-para-rastrear-falhas-agenticas-a-partir-de-bem/korshunov.aipt2026-07-15T06:38:22.979794+00:00OAT usa aprendizado de uma classe para rastrear falhas agênticas a partir de fluxos bem-sucedidos https://korshunov.ai/hi/article/12042-oat-sphl-prvaahon-se-ejenttik-viphltaa-ko-ttrais-krne-ke-lie-vn-klaas-lrning-kaa/korshunov.aihi2026-07-15T06:38:22.979794+00:00OAT सफल प्रवाहों से एजेंटिक विफलता को ट्रैस करने के लिए वन-क्लास लर्निंग का उपयोग करता है https://korshunov.ai/ja/article/12042-oathawankurasuxue-xi-woyong-ite-cheng-gong-huro-karae-shientonoshi-bai-wozhui-ji/korshunov.aija2026-07-15T06:38:22.979794+00:00OATはワンクラス学習を用いて、成功フローからエージェントの失敗を追跡する https://korshunov.ai/ko/article/12042-oatneun-weonkeulraeseu-hagseubeul-sayonghayeo-seonggong-heureumeseo-eijeonteu/korshunov.aiko2026-07-15T06:38:22.979794+00:00OAT는 원클래스 학습을 사용하여 성공 흐름에서 에이전트 실패를 추적합니다 https://korshunov.ai/fr/article/12042-oat-utilise-l-apprentissage-par-une-seule-classe-pour-retracer-les-echecs-des-a/korshunov.aifr2026-07-15T06:38:22.979794+00:00OAT utilise l'apprentissage par une seule classe pour retracer les échecs des agents à partir de flux de succès https://korshunov.ai/id/article/12042-oat-menggunakan-pembelajaran-satu-kelas-untuk-melacak-kegagalan-agentic-dari/korshunov.aiid2026-07-15T06:38:22.979794+00:00OAT menggunakan pembelajaran satu-kelas untuk melacak kegagalan agentic dari alur keberhasilan https://korshunov.ai/ar/article/12042-ystkhdm-oat-lt-lm-hdy-lfy-lttb-fshl-lwkl-mn-tdfqt-lnjh/korshunov.aiar2026-07-15T06:38:22.979794+00:00يستخدم OAT التعلم أحادي الفئة لتتبع فشل الوكلاء من تدفقات النجاح https://korshunov.ai/en/article/12041-qwen3-30b-a3b-achieves-8-3-on-fukuyamabench-for-mechanistic-reasoning/korshunov.aien2026-07-15T06:29:47.449325+00:00Qwen3-30B-A3B achieves 8.3% on FukuyamaBench for mechanistic reasoning https://korshunov.ai/ru/article/12041-qwen3-30b-a3b-dostigaet-83-na-fukuyamabench-po-mekhanisticheskomu-rassuzhdeniiu/korshunov.airu2026-07-15T06:29:47.449325+00:00Qwen3-30B-A3B достигает 8,3% на FukuyamaBench по механистическому рассуждению https://korshunov.ai/es/article/12041-qwen3-30b-a3b-alcanza-8-3-en-fukuyamabench-para-razonamiento-mecanistico/korshunov.aies2026-07-15T06:29:47.449325+00:00Qwen3-30B-A3B alcanza 8.3% en FukuyamaBench para razonamiento mecanístico https://korshunov.ai/zh-hans/article/12041-qwen3-30b-a3bzai-fukuyamabenchji-zhi-tui-li-shang-da-dao-8-3/korshunov.aizh-cn2026-07-15T06:29:47.449325+00:00Qwen3-30B-A3B在FukuyamaBench机制推理上达到8.3% https://korshunov.ai/pt-br/article/12041-qwen3-30b-a3b-atinge-83-no-fukuyamabench-para-raciocinio-mecanistico/korshunov.aipt2026-07-15T06:29:47.449325+00:00Qwen3-30B-A3B atinge 8,3% no FukuyamaBench para raciocínio mecanístico https://korshunov.ai/hi/article/12041-qwen3-30b-a3b-ne-yaantrik-trk-ke-lie-fukuyamabench-pr-8-3-haasil-kiyaa/korshunov.aihi2026-07-15T06:29:47.449325+00:00Qwen3-30B-A3B ने यांत्रिक तर्क के लिए FukuyamaBench पर 8.3% हासिल किया https://korshunov.ai/ja/article/12041-qwen3-30b-a3bkafukuyamabenchnoji-jie-de-tui-lun-te8-3-woda-cheng/korshunov.aija2026-07-15T06:29:47.449325+00:00Qwen3-30B-A3BがFukuyamaBenchの機械的推論で8.3%を達成 https://korshunov.ai/ko/article/12041-qwen3-30b-a3b-fukuyamabenchyi-gigyejeog-curoneseo-8-3-dalseong/korshunov.aiko2026-07-15T06:29:47.449325+00:00Qwen3-30B-A3B, FukuyamaBench의 기계적 추론에서 8.3% 달성 https://korshunov.ai/fr/article/12041-qwen3-30b-a3b-atteint-83-sur-fukuyamabench-pour-le-raisonnement-mecanistique/korshunov.aifr2026-07-15T06:29:47.449325+00:00Qwen3-30B-A3B atteint 8,3 % sur FukuyamaBench pour le raisonnement mécanistique https://korshunov.ai/id/article/12041-qwen3-30b-a3b-mencapai-83-pada-fukuyamabench-untuk-penalaran-mekanistik/korshunov.aiid2026-07-15T06:29:47.449325+00:00Qwen3-30B-A3B mencapai 8,3% pada FukuyamaBench untuk penalaran mekanistik https://korshunov.ai/ar/article/12041-hqq-qwen3-30b-a3b-nsb-8-3-l-fukuyamabench-llstdll-lmyknyky/korshunov.aiar2026-07-15T06:29:47.449325+00:00حقق Qwen3-30B-A3B نسبة 8.3٪ على FukuyamaBench للاستدلال الميكانيكي https://korshunov.ai/en/article/12040-light-mer-sub-billion-model-matches-sota-in-multimodal-emotion-recognition/korshunov.aien2026-07-15T06:22:48.408429+00:00Light-MER: sub-billion model matches SOTA in multimodal emotion recognition https://korshunov.ai/ru/article/12040-light-mer-submilliardnaia-model-dostigaet-sota-v-multimodalnom-raspoznavanii/korshunov.airu2026-07-15T06:22:48.408429+00:00Light-MER: субмиллиардная модель достигает SOTA в мультимодальном распознавании эмоций https://korshunov.ai/es/article/12040-light-mer-un-modelo-sub-mil-millones-iguala-el-sota-en-reconocimiento-de/korshunov.aies2026-07-15T06:22:48.408429+00:00Light-MER: un modelo sub-mil millones iguala el SOTA en reconocimiento de emociones multimodales https://korshunov.ai/zh-hans/article/12040-light-mer-shi-yi-can-shu-yi-xia-mo-xing-zai-kua-mo-tai-qing-gan-shi-bie-zhong-da/korshunov.aizh-cn2026-07-15T06:22:48.408429+00:00Light-MER：十亿参数以下模型在跨模态情感识别中达到SOTA https://korshunov.ai/pt-br/article/12040-light-mer-modelo-sub-bilhao-iguala-sota-em-reconhecimento-de-emocoes-multimodais/korshunov.aipt2026-07-15T06:22:48.408429+00:00Light-MER: modelo sub-bilhão iguala SOTA em reconhecimento de emoções multimodais https://korshunov.ai/hi/article/12040-light-mer-sb-biliyn-monddl-mlttiimoddl-imoshn-rikgnishn-men-sota-ke-braabr/korshunov.aihi2026-07-15T06:22:48.408429+00:00Light-MER: सब-बिलियन मॉडल मल्टीमोडल इमोशन रिकग्निशन में SOTA के बराबर https://korshunov.ai/ja/article/12040-light-mer-sahu10yi-harame-tamoterukamaruchimo-tarugan-qing-ren-shi-tesotanipi-di/korshunov.aija2026-07-15T06:22:48.408429+00:00Light-MER: サブ10億パラメータモデルがマルチモーダル感情認識でSOTAに匹敵 https://korshunov.ai/ko/article/12040-light-mer-seobeu-10eog-paramiteo-modeli-meoltimodal-gamjeong-insigeseo-sotae/korshunov.aiko2026-07-15T06:22:48.408429+00:00Light-MER: 서브-10억 파라미터 모델이 멀티모달 감정 인식에서 SOTA에 도달 https://korshunov.ai/fr/article/12040-light-mer-un-modele-sous-le-milliard-de-parametres-egale-l-etat-de-l-art-en-d/korshunov.aifr2026-07-15T06:22:48.408429+00:00Light-MER : un modèle sous le milliard de paramètres égale l'état de l'art en reconnaissance d'émotions multimodales https://korshunov.ai/id/article/12040-light-mer-model-sub-miliar-parameter-menyamai-sota-dalam-pengenalan-emosi/korshunov.aiid2026-07-15T06:22:48.408429+00:00Light-MER: model sub-miliar parameter menyamai SOTA dalam pengenalan emosi multimodal https://korshunov.ai/ar/article/12040-light-mer-nmwdhj-dwn-lmlyr-m-lm-yswy-lhl-lsyd-fy-lt-rf-l-lmsh-r-mt-dd-lwsyt/korshunov.aiar2026-07-15T06:22:48.408429+00:00Light-MER: نموذج دون المليار معلمة يساوي الحالة السائدة في التعرف على المشاعر متعددة الوسائط https://korshunov.ai/zh-hans/article/12039-zi-wo-gai-jin-llmdai-li-de-ping-gu-zhi-biao-yu-ji-neng-de-xie-tong-jin-hua/korshunov.aizh-cn2026-07-15T06:14:46.174001+00:00自我改进LLM代理的评估指标与技能的协同进化 https://korshunov.ai/en/article/12039-co-evolving-evaluation-metrics-and-skills-for-self-improving-llm-agents/korshunov.aien2026-07-15T06:14:46.174001+00:00Co-Evolving Evaluation Metrics and Skills for Self-Improving LLM Agents https://korshunov.ai/ru/article/12039-sovmestnaia-evoliutsiia-metrik-otsenki-i-navykov-dlia-samoobuchaiushchikhsia-llm/korshunov.airu2026-07-15T06:14:46.174001+00:00Совместная эволюция метрик оценки и навыков для самообучающихся LLM-агентов https://korshunov.ai/es/article/12039-co-evolucion-de-metricas-de-evaluacion-y-habilidades-para-agentes-llm-auto/korshunov.aies2026-07-15T06:14:46.174001+00:00Co-evolución de métricas de evaluación y habilidades para agentes LLM auto-mejorables https://korshunov.ai/pt-br/article/12039-co-evolucao-de-metricas-de-avaliacao-e-habilidades-para-agentes-llm-auto/korshunov.aipt2026-07-15T06:14:46.174001+00:00Co-evolução de Métricas de Avaliação e Habilidades para Agentes LLM Auto-Melhoráveis https://korshunov.ai/hi/article/12039-sv-sudhaar-krne-vaale-llm-ejentts-ke-lie-muulyaankn-mettriks-aur-kaushl-kaa-sh/korshunov.aihi2026-07-15T06:14:46.174001+00:00स्व-सुधार करने वाले LLM एजेंट्स के लिए मूल्यांकन मेट्रिक्स और कौशल का सह-विकास https://korshunov.ai/ja/article/12039-zi-ji-gai-shan-xing-llme-shientonotamenoping-jia-zhi-biao-tosukirunogong-jin-hua/korshunov.aija2026-07-15T06:14:46.174001+00:00自己改善型LLMエージェントのための評価指標とスキルの共進化 https://korshunov.ai/ko/article/12039-jagi-gaeseon-llm-eijeonteureul-wihan-pyeongga-jipyowa-seukilyi-gongdong-jinhwa/korshunov.aiko2026-07-15T06:14:46.174001+00:00자기 개선 LLM 에이전트를 위한 평가 지표와 스킬의 공동 진화 https://korshunov.ai/fr/article/12039-co-evolution-des-metriques-d-evaluation-et-des-competences-pour-les-agents-llm/korshunov.aifr2026-07-15T06:14:46.174001+00:00Co-évolution des métriques d'évaluation et des compétences pour les agents LLM auto-améliorants https://korshunov.ai/id/article/12039-ko-evolusi-metrik-evaluasi-dan-keterampilan-untuk-agen-llm-yang-meningkatkan/korshunov.aiid2026-07-15T06:14:46.174001+00:00Ko-evolusi Metrik Evaluasi dan Keterampilan untuk Agen LLM yang Meningkatkan Diri https://korshunov.ai/ar/article/12039-lttwr-lmshtrk-lmqyys-ltqyym-wlmhrt-lwkl-llm-dhty-lthsyn/korshunov.aiar2026-07-15T06:14:46.174001+00:00التطور المشترك لمقاييس التقييم والمهارات لوكلاء LLM ذاتي التحسين https://korshunov.ai/en/article/12148-tencent-releases-hy-embodied-rxbrain-1-0-unified-multimodal-foundation-model/korshunov.aien2026-07-15T06:02:21.712056+00:00Tencent releases Hy-Embodied-RxBrain-1.0 unified multimodal foundation model https://korshunov.ai/id/article/12148-tencent-merilis-model-fondasi-multimodal-terpadu-hy-embodied-rxbrain-1-0/korshunov.aiid2026-07-15T06:02:21.712056+00:00Tencent merilis model fondasi multimodal terpadu Hy-Embodied-RxBrain-1.0 https://korshunov.ai/ru/article/12148-tencent-vypustila-unifitsirovannuiu-multimodalnuiu-bazovuiu-model-hy-embodied-1/korshunov.airu2026-07-15T06:02:21.712056+00:00Tencent выпустила унифицированную мультимодальную базовую модель Hy-Embodied-RxBrain-1.0 https://korshunov.ai/es/article/12148-tencent-lanza-el-modelo-base-multimodal-unificado-hy-embodied-rxbrain-1-0/korshunov.aies2026-07-15T06:02:21.712056+00:00Tencent lanza el modelo base multimodal unificado Hy-Embodied-RxBrain-1.0 https://korshunov.ai/zh-hans/article/12148-teng-xun-fa-bu-hy-embodied-rxbrain-1-0-tong-yi-duo-mo-tai-ji-chu-mo-xing/korshunov.aizh-cn2026-07-15T06:02:21.712056+00:00腾讯发布 Hy-Embodied-RxBrain-1.0 统一多模态基础模型 https://korshunov.ai/pt-br/article/12148-tencent-lanca-o-modelo-base-multimodal-unificado-hy-embodied-rxbrain-1-0/korshunov.aipt2026-07-15T06:02:21.712056+00:00Tencent lança o modelo base multimodal unificado Hy-Embodied-RxBrain-1.0 https://korshunov.ai/hi/article/12148-ttensentt-ne-haai-embonddidd-raiksbren-1-0-ekiikrt-mlttiimoddl-phaaunddeshn/korshunov.aihi2026-07-15T06:02:21.712056+00:00टेंसेंट ने हाइ-एम्बॉडिड-रैक्सब्रेन-1.0 एकीकृत मल्टीमोडल फाउंडेशन मॉडल जारी किया https://korshunov.ai/ja/article/12148-tencentkahy-embodied-rxbrain-1-0tong-yi-maruchimo-taruji-pan-moteruworiri-su/korshunov.aija2026-07-15T06:02:21.712056+00:00TencentがHy-Embodied-RxBrain-1.0統一マルチモーダル基盤モデルをリリース https://korshunov.ai/ko/article/12148-tencent-hy-embodied-rxbrain-1-0-tonghab-meoltimodal-paundeisyeon-model-culsi/korshunov.aiko2026-07-15T06:02:21.712056+00:00Tencent, Hy-Embodied-RxBrain-1.0 통합 멀티모달 파운데이션 모델 출시 https://korshunov.ai/fr/article/12148-tencent-lance-le-modele-de-base-multimodal-unifie-hy-embodied-rxbrain-1-0/korshunov.aifr2026-07-15T06:02:21.712056+00:00Tencent lance le modèle de base multimodal unifié Hy-Embodied-RxBrain-1.0 https://korshunov.ai/ar/article/12148-tyntsin-ttlq-nmwdhj-lss-mt-dd-lwsyt-lmwhd-hy-embodied-rxbrain-1-0/korshunov.aiar2026-07-15T06:02:21.712056+00:00تيانتسِن تطلق نموذج الأساس متعدد الوسائط الموحد Hy-Embodied-RxBrain-1.0 https://korshunov.ai/en/article/12038-one-word-census-reveals-answer-choice-conformity-across-44-language-models/korshunov.aien2026-07-15T05:44:22.138755+00:00One-Word Census reveals answer-choice conformity across 44 language models https://korshunov.ai/ru/article/12038-odnoslovnaia-perepis-vyiavliaet-konformnost-variantov-otvetov-sredi-44-modelei/korshunov.airu2026-07-15T05:44:22.138755+00:00Однословная перепись выявляет конформность вариантов ответов среди 44 языковых моделей https://korshunov.ai/es/article/12038-el-censo-de-una-palabra-revela-la-conformidad-de-las-opciones-de-respuesta-en-44/korshunov.aies2026-07-15T05:44:22.138755+00:00El Censo de una Palabra revela la conformidad de las opciones de respuesta en 44 modelos de lenguaje https://korshunov.ai/zh-hans/article/12038-yi-ci-pu-cha-jie-shi-44ge-yu-yan-mo-xing-zai-da-an-xuan-ze-shang-de-qu-tong-xian/korshunov.aizh-cn2026-07-15T05:44:22.138755+00:00一词普查揭示44个语言模型在答案选择上的趋同现象 https://korshunov.ai/pt-br/article/12038-censo-de-uma-palavra-revela-conformidade-nas-opcoes-de-resposta-entre-44-modelos/korshunov.aipt2026-07-15T05:44:22.138755+00:00Censo de uma Palavra revela conformidade nas opções de resposta entre 44 modelos de linguagem https://korshunov.ai/hi/article/12038-ek-shbd-jngnnnaa-44-bhaassaa-monddlon-men-uttr-viklp-anupaaln-ko-ujaagr-krtii/korshunov.aihi2026-07-15T05:44:22.138755+00:00एक-शब्द जनगणना 44 भाषा मॉडलों में उत्तर-विकल्प अनुपालन को उजागर करती है https://korshunov.ai/ja/article/12038-yi-yu-guo-shi-diao-cha-ka44yan-yu-moteruniokeruhui-da-xuan-ze-zhi-notong-diao/korshunov.aija2026-07-15T05:44:22.138755+00:00一語国勢調査が44言語モデルにおける回答選択肢の同調性を明らかに https://korshunov.ai/ko/article/12038-han-daneo-ingu-josa-44gae-eoneo-model-jeonbanyi-jeongdab-seontaeg-dongjoseong/korshunov.aiko2026-07-15T05:44:22.138755+00:00한 단어 인구 조사, 44개 언어 모델 전반의 정답 선택 동조성 드러내 https://korshunov.ai/fr/article/12038-le-recensement-en-un-mot-revele-la-conformite-aux-choix-de-reponse-chez-44/korshunov.aifr2026-07-15T05:44:22.138755+00:00Le recensement en un mot révèle la conformité aux choix de réponse chez 44 modèles linguistiques https://korshunov.ai/id/article/12038-sensus-satu-kata-mengungkap-keseragaman-pilihan-jawaban-di-44-model-bahasa/korshunov.aiid2026-07-15T05:44:22.138755+00:00Sensus Satu Kata Mengungkap Keseragaman Pilihan Jawaban di 44 Model Bahasa https://korshunov.ai/ar/article/12038-t-dd-bklm-whd-ykshf-n-ttbq-khyrt-ljb-br-44-nmwdhjan-lgwyan/korshunov.aiar2026-07-15T05:44:22.138755+00:00تعداد بكلمة واحدة يكشف عن تطابق خيارات الإجابة عبر 44 نموذجاً لغوياً https://korshunov.ai/en/article/12037-survey-introduces-unified-latency-decomposition-framework-for-accelerating-llms/korshunov.aien2026-07-15T05:34:31.154885+00:00Survey introduces unified latency decomposition framework for accelerating masked diffusion LLMs https://korshunov.ai/ru/article/12037-obzor-predstavliaet-edinuiu-freimvork-dekompozitsii-zaderzhek-dlia-uskoreniia/korshunov.airu2026-07-15T05:34:31.154885+00:00Обзор представляет единую фреймворк декомпозиции задержек для ускорения masked-диффузионных LLM https://korshunov.ai/es/article/12037-encuesta-presenta-un-marco-unificado-de-descomposicion-de-latencia-para-acelerar/korshunov.aies2026-07-15T05:34:31.154885+00:00Encuesta presenta un marco unificado de descomposición de latencia para acelerar LLMs de difusión enmascarada https://korshunov.ai/zh-hans/article/12037-zong-shu-ti-chu-tong-yi-yan-chi-fen-jie-kuang-jia-yi-jia-su-yan-ma-kuo-san-da-yu/korshunov.aizh-cn2026-07-15T05:34:31.154885+00:00综述提出统一延迟分解框架以加速掩码扩散大语言模型 https://korshunov.ai/pt-br/article/12037-pesquisa-apresenta-framework-unificado-de-decomposicao-de-latencia-para-acelerar/korshunov.aipt2026-07-15T05:34:31.154885+00:00Pesquisa apresenta framework unificado de decomposição de latência para acelerar LLMs de difusão mascarada https://korshunov.ai/hi/article/12037-srvekssnn-masked-diffusion-llms-ko-tej-krne-ke-lie-ek-ekiikrt-laittensii-pesh/korshunov.aihi2026-07-15T05:34:31.154885+00:00सर्वेक्षण masked diffusion LLMs को तेज करने के लिए एक एकीकृत लैटेंसी डिकम्पोजिशन फ्रेमवर्क पेश करता है https://korshunov.ai/ja/article/12037-masukitokuo-san-llmnogao-su-hua-notamenotong-yi-reitenshifen-jie-hure-muwa-jie/korshunov.aija2026-07-15T05:34:31.154885+00:00マスキド拡散LLMの高速化のための統一レイテンシ分解フレームワークを紹介する調査 https://korshunov.ai/ko/article/12037-maseukeudoen-hwagsan-llm-gasoghwareul-wihan-tonghab-jiyeon-sigan-bunhae-sogae/korshunov.aiko2026-07-15T05:34:31.154885+00:00마스크된 확산 LLM 가속화를 위한 통합 지연 시간 분해 프레임워크 소개 https://korshunov.ai/fr/article/12037-une-enquete-presente-un-cadre-de-decomposition-unifie-de-la-latence-pour-les-de/korshunov.aifr2026-07-15T05:34:31.154885+00:00Une enquête présente un cadre de décomposition unifié de la latence pour accélérer les LLMs de diffusion masqués https://korshunov.ai/id/article/12037-survei-memperkenalkan-kerangka-dekomposisi-latensi-terpadu-untuk-mempercepat-llm/korshunov.aiid2026-07-15T05:34:31.154885+00:00Survei memperkenalkan kerangka dekomposisi latensi terpadu untuk mempercepat LLM difusi masker https://korshunov.ai/ar/article/12037-msh-yqdm-tr-thlyl-mwhd-llzmn-lstjb-ltsry-nmdhj-llg-lkbyr-lntshry-lmqn/korshunov.aiar2026-07-15T05:34:31.154885+00:00مسح يقدم إطار تحليل موحد للزمن لاستجابة لتسريع نماذج اللغة الكبيرة الانتشارية المقنعة https://korshunov.ai/zh-hans/article/12036-wu-zhi-shi-yu-yan-mo-xing-yi-zhi-can-shu-hua-hui-yi-yi-zhi-chi-ji-yu-zheng-ju-de/korshunov.aizh-cn2026-07-15T05:28:46.023197+00:00无知识语言模型抑制参数化回忆以支持基于证据的推理 https://korshunov.ai/en/article/12036-knowledgeless-language-models-suppress-parametric-recall-for-evidence-grounded/korshunov.aien2026-07-15T05:28:46.023197+00:00Knowledgeless Language Models suppress parametric recall for evidence-grounded reasoning https://korshunov.ai/ru/article/12036-iazykovye-modeli-bez-znanii-podavliaiut-parametricheskoe-vosproizvedenie-dlia-na/korshunov.airu2026-07-15T05:28:46.023197+00:00Языковые модели без знаний подавляют параметрическое воспроизведение для рассуждений, основанных на доказательствах https://korshunov.ai/es/article/12036-los-modelos-de-lenguaje-sin-conocimiento-suprimen-el-recuerdo-parametrico-para/korshunov.aies2026-07-15T05:28:46.023197+00:00Los modelos de lenguaje sin conocimiento suprimen el recuerdo paramétrico para el razonamiento fundamentado en evidencias https://korshunov.ai/pt-br/article/12036-modelos-de-linguagem-sem-conhecimento-suprimem-a-recuperacao-parametrica-para-em/korshunov.aipt2026-07-15T05:28:46.023197+00:00Modelos de linguagem sem conhecimento suprimem a recuperação paramétrica para raciocínio fundamentado em evidências https://korshunov.ai/hi/article/12036-jnyaanhiin-bhaassaa-monddl-prmaann-aadhaarit-trk-ke-lie-pairaamiittrik-smrti-ko/korshunov.aihi2026-07-15T05:28:46.023197+00:00ज्ञानहीन भाषा मॉडल प्रमाण-आधारित तर्क के लिए पैरामीट्रिक स्मृति को दबाते हैं https://korshunov.ai/ja/article/12036-zhi-shi-wochi-tanaiyan-yu-moteruha-zheng-ju-niji-tsukutui-lun-notameniharame-yi/korshunov.aija2026-07-15T05:28:46.023197+00:00知識を持たない言語モデルは、証拠に基づく推論のためにパラメータ記憶を抑制する https://korshunov.ai/ko/article/12036-jisig-eobsneun-eoneo-modeleun-jeunggeo-giban-curoneul-wihae-maegaebyeonsu/korshunov.aiko2026-07-15T05:28:46.023197+00:00지식 없는 언어 모델은 증거 기반 추론을 위해 매개변수 회상을 억제한다 https://korshunov.ai/fr/article/12036-les-modeles-de-langage-sans-connaissances-suppriment-le-rappel-parametrique-au/korshunov.aifr2026-07-15T05:28:46.023197+00:00Les modèles de langage sans connaissances suppriment le rappel paramétrique au profit du raisonnement ancré dans les preuves https://korshunov.ai/id/article/12036-model-bahasa-tanpa-pengetahuan-menekan-recall-parametrik-untuk-penalaran-bukti/korshunov.aiid2026-07-15T05:28:46.023197+00:00Model Bahasa Tanpa Pengetahuan menekan recall parametrik untuk penalaran berbasis bukti https://korshunov.ai/ar/article/12036-nmdhj-llg-lkhly-mn-lm-rf-tkbh-lstrj-lbrmtry-llstdll-lqym-l-ldl/korshunov.aiar2026-07-15T05:28:46.023197+00:00نماذج اللغة الخالية من المعرفة تكبح الاسترجاع البارامتري للاستدلال القائم على الأدلة https://korshunov.ai/en/article/12035-meta-evaluation-shows-augmented-llm-settings-improve-rubric-alignment-for-paper/korshunov.aien2026-07-15T05:20:41.595310+00:00Meta-evaluation shows augmented LLM settings improve rubric alignment for paper reproduction https://korshunov.ai/ru/article/12035-meta-otsenka-pokazyvaet-chto-rasshirennye-nastroiki-llm-uluchshaiut-sootvetstvie/korshunov.airu2026-07-15T05:20:41.595310+00:00Мета-оценка показывает, что расширенные настройки LLM улучшают соответствие рубрикам при воспроизведении статей https://korshunov.ai/es/article/12035-la-metaevaluacion-muestra-que-las-configuraciones-aumentadas-de-llm-mejoran-la/korshunov.aies2026-07-15T05:20:41.595310+00:00La metaevaluación muestra que las configuraciones aumentadas de LLM mejoran la alineación de rúbricas para la reproducción de artículos https://korshunov.ai/zh-hans/article/12035-yuan-ping-gu-xian-shi-zeng-qiang-xing-llmshe-zhi-ti-gao-liao-lun-wen-fu-xian-gui/korshunov.aizh-cn2026-07-15T05:20:41.595310+00:00元评估显示，增强型LLM设置提高了论文复现中量规的对齐度 https://korshunov.ai/pt-br/article/12035-meta-avaliacao-mostra-que-configuracoes-aumentadas-de-llm-melhoram-o-alinhamento/korshunov.aipt2026-07-15T05:20:41.595310+00:00Meta-avaliação mostra que configurações aumentadas de LLM melhoram o alinhamento de rubricas para reprodução de artigos https://korshunov.ai/hi/article/12035-mettaa-muulyaankn-dikhaataa-hai-ki-bddhaae-ge-llm-settings-pepr-punrutpaadn-ke/korshunov.aihi2026-07-15T05:20:41.595310+00:00मेटा-मूल्यांकन दिखाता है कि बढ़ाए गए LLM सेटिंग्स पेपर पुनरुत्पादन के लिए रबरिक एलाइनमेंट को बेहतर बनाते हैं https://korshunov.ai/ja/article/12035-metaping-jia-niyori-llmnokuo-zhang-she-ding-kalun-wen-zai-xian-niokeruru-he-xing/korshunov.aija2026-07-15T05:20:41.595310+00:00メタ評価により、LLMの拡張設定が論文再現におけるルーブリック整合性を向上させることが示された https://korshunov.ai/ko/article/12035-meta-pyeongganeun-jeunggangdoen-llm-seoljeongi-nonmun-jaehyeoneul-wihan-boyeojum/korshunov.aiko2026-07-15T05:20:41.595310+00:00메타 평가는 증강된 LLM 설정이 논문 재현을 위한 기준점 정렬을 개선함을 보여줌 https://korshunov.ai/fr/article/12035-une-meta-evaluation-montre-que-les-parametres-augmentes-des-llm-ameliorent-l-des/korshunov.aifr2026-07-15T05:20:41.595310+00:00Une méta-évaluation montre que les paramètres augmentés des LLM améliorent l'alignement des grilles d'évaluation pour la reproduction d'articles https://korshunov.ai/id/article/12035-meta-evaluasi-menunjukkan-bahwa-pengaturan-llm-yang-diperkuat-meningkatkan-untuk/korshunov.aiid2026-07-15T05:20:41.595310+00:00Meta-evaluasi menunjukkan bahwa pengaturan LLM yang diperkuat meningkatkan keselarasan rubrik untuk reproduksi makalah https://korshunov.ai/ar/article/12035-tuzhr-ltqyymt-lwsfy-n-ddt-nmdhj-llgt-lkbyr-lm-zz-thswn-twfq-m-yyr-ltqyym-lstnskh/korshunov.aiar2026-07-15T05:20:41.595310+00:00تُظهر التقييمات الوصفية أن إعدادات نماذج اللغات الكبيرة المعززة تحسّن توافق معايير التقييم لاستنساخ الأوراق البحثية https://korshunov.ai/en/article/12144-qwen-3-5-122b-heretic-rocmfp4-imatrix-gguf-released/korshunov.aien2026-07-15T05:11:45.965710+00:00Qwen 3.5 122B Heretic ROCmFP4 iMatrix GGUF released https://korshunov.ai/ru/article/12144-vypushchen-qwen-3-5-122b-heretic-rocmfp4-imatrix-gguf/korshunov.airu2026-07-15T05:11:45.965710+00:00Выпущен Qwen 3.5 122B Heretic ROCmFP4 iMatrix GGUF https://korshunov.ai/es/article/12144-se-ha-lanzado-qwen-3-5-122b-heretic-rocmfp4-imatrix-gguf/korshunov.aies2026-07-15T05:11:45.965710+00:00Se ha lanzado Qwen 3.5 122B Heretic ROCmFP4 iMatrix GGUF https://korshunov.ai/zh-hans/article/12144-qwen-3-5-122b-heretic-rocmfp4-imatrix-gguf-yi-fa-bu/korshunov.aizh-cn2026-07-15T05:11:45.965710+00:00Qwen 3.5 122B Heretic ROCmFP4 iMatrix GGUF 已发布 https://korshunov.ai/pt-br/article/12144-qwen-3-5-122b-heretic-rocmfp4-imatrix-gguf-lancado/korshunov.aipt2026-07-15T05:11:45.965710+00:00Qwen 3.5 122B Heretic ROCmFP4 iMatrix GGUF lançado https://korshunov.ai/hi/article/12144-qwen-3-5-122b-heretic-rocmfp4-imatrix-gguf-jaarii-kiyaa-gyaa/korshunov.aihi2026-07-15T05:11:45.965710+00:00Qwen 3.5 122B Heretic ROCmFP4 iMatrix GGUF जारी किया गया https://korshunov.ai/ja/article/12144-qwen-3-5-122b-heretic-rocmfp4-imatrix-gguf-kariri-su/korshunov.aija2026-07-15T05:11:45.965710+00:00Qwen 3.5 122B Heretic ROCmFP4 iMatrix GGUF がリリース https://korshunov.ai/ko/article/12144-qwen-3-5-122b-heretic-rocmfp4-imatrix-gguf-culsi/korshunov.aiko2026-07-15T05:11:45.965710+00:00Qwen 3.5 122B Heretic ROCmFP4 iMatrix GGUF 출시 https://korshunov.ai/fr/article/12144-qwen-3-5-122b-heretic-rocmfp4-imatrix-gguf-publie/korshunov.aifr2026-07-15T05:11:45.965710+00:00Qwen 3.5 122B Heretic ROCmFP4 iMatrix GGUF publié https://korshunov.ai/id/article/12144-qwen-3-5-122b-heretic-rocmfp4-imatrix-gguf-dirilis/korshunov.aiid2026-07-15T05:11:45.965710+00:00Qwen 3.5 122B Heretic ROCmFP4 iMatrix GGUF dirilis https://korshunov.ai/ar/article/12144-sdr-gguf-lkhs-b-qwen-3-5-122b-heretic-rocmfp4-imatrix/korshunov.aiar2026-07-15T05:11:45.965710+00:00إصدار GGUF الخاص بـ Qwen 3.5 122B Heretic ROCmFP4 iMatrix https://korshunov.ai/en/article/12145-user-reports-no-speedup-with-ternary-bonsai-27b-dflash-on-rocm/korshunov.aien2026-07-15T05:05:25.055029+00:00User reports no speedup with Ternary Bonsai 27B DFlash on ROCm https://korshunov.ai/ru/article/12145-polzovateli-ne-otmechaiut-uskoreniia-pri-ispolzovanii-ternary-bonsai-27b-dflash/korshunov.airu2026-07-15T05:05:25.055029+00:00Пользователи не отмечают ускорения при использовании Ternary Bonsai 27B DFlash на ROCm https://korshunov.ai/es/article/12145-usuarios-reportan-sin-aceleracion-con-ternary-bonsai-27b-dflash-en-rocm/korshunov.aies2026-07-15T05:05:25.055029+00:00Usuarios reportan sin aceleración con Ternary Bonsai 27B DFlash en ROCm https://korshunov.ai/zh-hans/article/12145-yong-hu-bao-gao-zai-rocm-shang-shi-yong-ternary-bonsai-27b-dflash-mei-you-jia-su/korshunov.aizh-cn2026-07-15T05:05:25.055029+00:00用户报告在 ROCm 上使用 Ternary Bonsai 27B DFlash 没有加速 https://korshunov.ai/pt-br/article/12145-usuarios-relatam-sem-aceleracao-com-ternary-bonsai-27b-dflash-no-rocm/korshunov.aipt2026-07-15T05:05:25.055029+00:00Usuários relatam sem aceleração com Ternary Bonsai 27B DFlash no ROCm https://korshunov.ai/hi/article/12145-rocm-pr-ternary-bonsai-27b-dflash-ke-saath-koii-gti-vrddhi-nhiin-milne-kii/korshunov.aihi2026-07-15T05:05:25.055029+00:00ROCm पर Ternary Bonsai 27B DFlash के साथ कोई गति वृद्धि नहीं मिलने की रिपोर्ट https://korshunov.ai/ja/article/12145-yu-sa-rocmshang-noternary-bonsai-27b-dflashtesu-du-xiang-shang-nashiwobao-gao/korshunov.aija2026-07-15T05:05:25.055029+00:00ユーザー、ROCm上のTernary Bonsai 27B DFlashで速度向上なしを報告 https://korshunov.ai/ko/article/12145-sayongja-rocmeseo-ternary-bonsai-27b-dflashyi-sogdo-hyangsang-eobseum-bogo/korshunov.aiko2026-07-15T05:05:25.055029+00:00사용자, ROCm에서 Ternary Bonsai 27B DFlash의 속도 향상 없음 보고 https://korshunov.ai/fr/article/12145-un-utilisateur-signale-aucune-acceleration-avec-ternary-bonsai-27b-dflash-sur/korshunov.aifr2026-07-15T05:05:25.055029+00:00Un utilisateur signale aucune accélération avec Ternary Bonsai 27B DFlash sur ROCm https://korshunov.ai/id/article/12145-pengguna-melaporkan-tidak-ada-percepatan-dengan-ternary-bonsai-27b-dflash-di/korshunov.aiid2026-07-15T05:05:25.055029+00:00Pengguna melaporkan tidak ada percepatan dengan Ternary Bonsai 27B DFlash di ROCm https://korshunov.ai/ar/article/12145-mstkhdm-yblg-n-dm-wjwd-tsry-m-ternary-bonsai-27b-dflash-l-rocm/korshunov.aiar2026-07-15T05:05:25.055029+00:00مستخدم يبلغ عن عدم وجود تسريع مع Ternary Bonsai 27B DFlash على ROCm https://korshunov.ai/en/article/12147-reddit-user-asks-for-model-recommendations-for-384gb-blackwell-setup/korshunov.aien2026-07-15T05:00:42.543680+00:00Reddit user asks for model recommendations for 384GB Blackwell setup https://korshunov.ai/ru/article/12147-polzovatel-reddit-prosit-rekomendatsii-po-modeliam-dlia-konfiguratsii-blackwell/korshunov.airu2026-07-15T05:00:42.543680+00:00Пользователь Reddit просит рекомендации по моделям для конфигурации Blackwell с 384 ГБ https://korshunov.ai/es/article/12147-usuario-de-reddit-solicita-recomendaciones-de-modelos-para-configuracion-de-384/korshunov.aies2026-07-15T05:00:42.543680+00:00Usuario de Reddit solicita recomendaciones de modelos para configuración Blackwell de 384 GB https://korshunov.ai/zh-hans/article/12147-reddityong-hu-xun-wen-384gb-blackwellpei-zhi-de-mo-xing-tui-jian/korshunov.aizh-cn2026-07-15T05:00:42.543680+00:00Reddit用户询问384GB Blackwell配置的模型推荐 https://korshunov.ai/pt-br/article/12147-usuario-do-reddit-pede-recomendacoes-de-modelos-para-configuracao-blackwell-de/korshunov.aipt2026-07-15T05:00:42.543680+00:00Usuário do Reddit pede recomendações de modelos para configuração Blackwell de 384 GB https://korshunov.ai/hi/article/12147-redditt-upyogkrtaa-384gb-blaikvel-settap-ke-lie-monddl-siphaarishen-maangte-hain/korshunov.aihi2026-07-15T05:00:42.543680+00:00रेडिट उपयोगकर्ता 384GB ब्लैकवेल सेटअप के लिए मॉडल सिफारिशें मांगते हैं https://korshunov.ai/ja/article/12147-384gbnoblackwellhuan-jing-xiang-kemoterutui-jiang-woqiu-merureddityu-sa/korshunov.aija2026-07-15T05:00:42.543680+00:00384GBのBlackwell環境向けモデル推奨を求めるRedditユーザー https://korshunov.ai/ko/article/12147-384gb-blackwell-hwangyeongyong-model-cuceoneul-yoceonghaneun-reddit-sayongja/korshunov.aiko2026-07-15T05:00:42.543680+00:00384GB Blackwell 환경용 모델 추천을 요청하는 Reddit 사용자 https://korshunov.ai/fr/article/12147-un-utilisateur-de-reddit-demande-des-recommandations-de-modeles-pour-une-de-384/korshunov.aifr2026-07-15T05:00:42.543680+00:00Un utilisateur de Reddit demande des recommandations de modèles pour une configuration Blackwell de 384 Go https://korshunov.ai/id/article/12147-pengguna-reddit-meminta-rekomendasi-model-untuk-setup-blackwell-384gb/korshunov.aiid2026-07-15T05:00:42.543680+00:00Pengguna Reddit meminta rekomendasi model untuk setup Blackwell 384GB https://korshunov.ai/ar/article/12147-mstkhdm-rdyt-ytlb-twsyt-lnmdhj-l-dd-blackwell-bs-384-jyjbyt/korshunov.aiar2026-07-15T05:00:42.543680+00:00مستخدم رديت يطلب توصيات لنماذج لإعداد Blackwell بسعة 384 جيجابايت https://korshunov.ai/zh-hans/article/12146-llama-cpp-b10015-fa-bu-macos-kleidiai-yi-jin-yong-bing-xiu-fu-opencl/korshunov.aizh-cn2026-07-15T04:51:59.423805+00:00llama.cpp b10015 发布，macOS KleidiAI 已禁用并修复 OpenCL https://korshunov.ai/en/article/12146-llama-cpp-b10015-release-with-macos-kleidiai-disabled-and-opencl-fix/korshunov.aien2026-07-15T04:51:59.423805+00:00llama.cpp b10015 release with macOS KleidiAI disabled and OpenCL fix https://korshunov.ai/ru/article/12146-vypusk-llama-cpp-b10015-s-otkliuchennym-macos-kleidiai-i-ispravleniem-opencl/korshunov.airu2026-07-15T04:51:59.423805+00:00Выпуск llama.cpp b10015 с отключённым macOS KleidiAI и исправлением OpenCL https://korshunov.ai/es/article/12146-lanzamiento-de-llama-cpp-b10015-con-macos-kleidiai-deshabilitado-y-correccion-de/korshunov.aies2026-07-15T04:51:59.423805+00:00Lanzamiento de llama.cpp b10015 con macOS KleidiAI deshabilitado y corrección de OpenCL https://korshunov.ai/pt-br/article/12146-lancamento-do-llama-cpp-b10015-com-macos-kleidiai-desativado-e-correcao-do/korshunov.aipt2026-07-15T04:51:59.423805+00:00Lançamento do llama.cpp b10015 com macOS KleidiAI desativado e correção do OpenCL https://korshunov.ai/hi/article/12146-macos-kleidiai-akssm-aur-opencl-sudhaar-ke-saath-llama-cpp-b10015-riliij/korshunov.aihi2026-07-15T04:51:59.423805+00:00macOS KleidiAI अक्षम और OpenCL सुधार के साथ llama.cpp b10015 रिलीज़ https://korshunov.ai/ja/article/12146-macosnokleidiaikawu-xiao-hua-sare-openclxiu-zheng-kashi-saretallama-cpp-su/korshunov.aija2026-07-15T04:51:59.423805+00:00macOSのKleidiAIが無効化され、OpenCL修正が施されたllama.cpp b10015リリース https://korshunov.ai/ko/article/12146-macos-kleidiai-bihwalseonghwa-mic-opencl-sujeongi-pohamdoen-llama-cpp-b10015/korshunov.aiko2026-07-15T04:51:59.423805+00:00macOS KleidiAI 비활성화 및 OpenCL 수정이 포함된 llama.cpp b10015 릴리스 https://korshunov.ai/fr/article/12146-publication-de-llama-cpp-b10015-avec-kleidiai-desactive-sur-macos-et-correction/korshunov.aifr2026-07-15T04:51:59.423805+00:00Publication de llama.cpp b10015 avec KleidiAI désactivé sur macOS et correction OpenCL https://korshunov.ai/id/article/12146-rilis-llama-cpp-b10015-dengan-kleidiai-macos-dinonaktifkan-dan-perbaikan-opencl/korshunov.aiid2026-07-15T04:51:59.423805+00:00Rilis llama.cpp b10015 dengan KleidiAI macOS dinonaktifkan dan perbaikan OpenCL https://korshunov.ai/ar/article/12146-sdr-llama-cpp-b10015-m-t-tyl-kleidiai-l-macos-wslh-opencl/korshunov.aiar2026-07-15T04:51:59.423805+00:00إصدار llama.cpp b10015 مع تعطيل KleidiAI على macOS وإصلاح OpenCL https://korshunov.ai/en/article/12034-threadmed-qa-dataset-reveals-llm-misconception-correction-degrades-in-multi-turn/korshunov.aien2026-07-15T04:39:41.334775+00:00ThReadMed-QA dataset reveals LLM misconception correction degrades in multi-turn medical conversations https://korshunov.ai/ru/article/12034-nabor-dannykh-threadmed-qa-pokazyvaet-chto-ispravlenie-zabluzhdenii-llm-v/korshunov.airu2026-07-15T04:39:41.334775+00:00Набор данных ThReadMed-QA показывает, что исправление заблуждений LLM ухудшается в многооборотных медицинских беседах https://korshunov.ai/es/article/12034-el-conjunto-de-datos-threadmed-qa-revela-que-la-correccion-de-malentendidos-en/korshunov.aies2026-07-15T04:39:41.334775+00:00El conjunto de datos ThReadMed-QA revela que la corrección de malentendidos en LLM se degrada en conversaciones médicas multironda https://korshunov.ai/zh-hans/article/12034-threadmed-qashu-ju-ji-jie-shi-da-yu-yan-mo-xing-zai-duo-lun-yi-liao-dui-hua-jiu/korshunov.aizh-cn2026-07-15T04:39:41.334775+00:00ThReadMed-QA数据集揭示大语言模型在多轮医疗对话中纠正误解的能力下降 https://korshunov.ai/pt-br/article/12034-conjunto-de-dados-threadmed-qa-revela-que-correcao-de-equivocos-em-llms-se-em/korshunov.aipt2026-07-15T04:39:41.334775+00:00Conjunto de dados ThReadMed-QA revela que correção de equívocos em LLMs se degrada em conversas médicas multironda https://korshunov.ai/hi/article/12034-threadmed-qa-ddettaasett-dikhaataa-hai-ki-bhu-crnn-cikitsaa-snvaadon-men-llm-ke/korshunov.aihi2026-07-15T04:39:41.334775+00:00ThReadMed-QA डेटासेट दिखाता है कि बहु-चरण चिकित्सा संवादों में LLM के भ्रम सुधार की क्षमता कम हो जाती है https://korshunov.ai/ja/article/12034-threadmed-qate-tasetsutoha-maruchita-nyi-liao-hui-hua-niokerullmnowu-gai-nian-li/korshunov.aija2026-07-15T04:39:41.334775+00:00ThReadMed-QAデータセットは、マルチターン医療会話におけるLLMの誤概念修正能力が低下することを明らかにする https://korshunov.ai/ko/article/12034-threadmed-qa-deiteoseseun-dajung-teon-yiryo-daehwaeseo-llmyi-ogaenyeom-gyojeong/korshunov.aiko2026-07-15T04:39:41.334775+00:00ThReadMed-QA 데이터셋은 다중 턴 의료 대화에서 LLM의 오개념 교정 성능이 저하됨을 드러냄 https://korshunov.ai/fr/article/12034-le-jeu-de-donnees-threadmed-qa-revele-que-la-correction-des-idees-fausses-par-se/korshunov.aifr2026-07-15T04:39:41.334775+00:00Le jeu de données ThReadMed-QA révèle que la correction des idées fausses par les LLM se dégrade dans les conversations médicales multi-tours https://korshunov.ai/id/article/12034-dataset-threadmed-qa-mengungkap-degradasi-koreksi-miskonsepsi-llm-dalam-medis/korshunov.aiid2026-07-15T04:39:41.334775+00:00Dataset ThReadMed-QA mengungkap degradasi koreksi miskonsepsi LLM dalam percakapan medis multi-giliran https://korshunov.ai/ar/article/12034-ykshf-mjmw-bynt-threadmed-qa-n-tdhwr-tshyh-lmfhym-lkhty-fy-nmdhj-llgt-lkbyr-khll/korshunov.aiar2026-07-15T04:39:41.334775+00:00يكشف مجموعة بيانات ThReadMed-QA عن تدهور تصحيح المفاهيم الخاطئة في نماذج اللغات الكبيرة خلال المحادثات الطبية متعددة الأدوار https://korshunov.ai/en/article/12033-llm-judges-over-credit-incorrect-answers-without-reference-answers/korshunov.aien2026-07-15T04:29:36.781165+00:00LLM judges over-credit incorrect answers without reference answers https://korshunov.ai/ru/article/12033-sudi-na-osnove-llm-zavyshaiut-otsenku-nepravilnykh-otvetov-bez-etalonnykh/korshunov.airu2026-07-15T04:29:36.781165+00:00Судьи на основе LLM завышают оценку неправильных ответов без эталонных данных https://korshunov.ai/es/article/12033-los-jueces-de-llm-sobrevaloran-las-respuestas-incorrectas-sin-respuestas-de/korshunov.aies2026-07-15T04:29:36.781165+00:00Los jueces de LLM sobrevaloran las respuestas incorrectas sin respuestas de referencia https://korshunov.ai/zh-hans/article/12033-llmcai-pan-zai-wu-can-kao-da-an-shi-guo-du-gei-yu-cuo-wu-da-an-gao-fen/korshunov.aizh-cn2026-07-15T04:29:36.781165+00:00LLM裁判在无参考答案时过度给予错误答案高分 https://korshunov.ai/pt-br/article/12033-juizes-llm-superavaliam-respostas-incorretas-sem-respostas-de-referencia/korshunov.aipt2026-07-15T04:29:36.781165+00:00Juízes LLM superavaliam respostas incorretas sem respostas de referência https://korshunov.ai/hi/article/12033-llm-jj-sndrbh-uttron-ke-binaa-glt-uttron-ko-adhik-kredditt-dete-hain/korshunov.aihi2026-07-15T04:29:36.781165+00:00LLM जज संदर्भ उत्तरों के बिना गलत उत्तरों को अधिक क्रेडिट देते हैं https://korshunov.ai/ja/article/12033-can-zhao-hui-da-nashinollmpan-shi-kawu-da-niguo-sheng-nijia-dian-suru/korshunov.aija2026-07-15T04:29:36.781165+00:00参照回答なしのLLM判事が誤答に過剰に加点する https://korshunov.ai/ko/article/12033-camjo-jeongdabi-eobsneun-llm-pansaga-odabe-gwadohage-gasanjeom-buyeo/korshunov.aiko2026-07-15T04:29:36.781165+00:00참조 정답이 없는 LLM 판사가 오답에 과도하게 가산점 부여 https://korshunov.ai/fr/article/12033-les-juges-llm-surevaluent-les-reponses-incorrectes-en-l-absence-de-references/korshunov.aifr2026-07-15T04:29:36.781165+00:00Les juges LLM surévaluent les réponses incorrectes en l'absence de références https://korshunov.ai/id/article/12033-hakim-llm-memberikan-kredit-berlebihan-pada-jawaban-salah-tanpa-referensi/korshunov.aiid2026-07-15T04:29:36.781165+00:00Hakim LLM memberikan kredit berlebihan pada jawaban salah tanpa referensi https://korshunov.ai/ar/article/12033-hkm-nmdhj-llgt-lkbyr-llm-tmnh-rsydan-mfrtan-lljbt-lkhty-dwn-shrt-mrj-y/korshunov.aiar2026-07-15T04:29:36.781165+00:00أحكام نماذج اللغات الكبيرة (LLM) تمنح رصيداً مفرطاً للإجابات الخاطئة دون إشارات مرجعية https://korshunov.ai/en/article/12032-memops-benchmark-evaluates-lifecycle-memory-operations-in-long-horizon/korshunov.aien2026-07-15T04:22:56.401665+00:00MemOps benchmark evaluates lifecycle memory operations in long-horizon conversations https://korshunov.ai/ru/article/12032-benchmark-memops-otsenivaet-operatsii-s-pamiatiu-v-techenie-zhiznennogo-tsikla/korshunov.airu2026-07-15T04:22:56.401665+00:00Бенчмарк MemOps оценивает операции с памятью в течение жизненного цикла при долгосрочных диалогах https://korshunov.ai/es/article/12032-el-benchmark-memops-evalua-operaciones-de-memoria-en-el-ciclo-de-vida-durante-de/korshunov.aies2026-07-15T04:22:56.401665+00:00El benchmark MemOps evalúa operaciones de memoria en el ciclo de vida durante conversaciones de largo alcance https://korshunov.ai/zh-hans/article/12032-memopsji-zhun-ping-gu-chang-cheng-dui-hua-zhong-de-sheng-ming-zhou-qi-nei-cun/korshunov.aizh-cn2026-07-15T04:22:56.401665+00:00MemOps基准评估长程对话中的生命周期内存操作 https://korshunov.ai/pt-br/article/12032-benchmark-memops-avalia-operacoes-de-memoria-no-ciclo-de-vida-em-conversas-de/korshunov.aipt2026-07-15T04:22:56.401665+00:00Benchmark MemOps avalia operações de memória no ciclo de vida em conversas de longo alcance https://korshunov.ai/hi/article/12032-memops-bencmaark-lnbii-avdhi-kii-baatciit-men-jiivn-ckr-memorii-onpreshn-kaa-hai/korshunov.aihi2026-07-15T04:22:56.401665+00:00MemOps बेंचमार्क लंबी अवधि की बातचीत में जीवन चक्र मेमोरी ऑपरेशन का मूल्यांकन करता है https://korshunov.ai/ja/article/12032-memopshenchima-kuhachang-shi-jian-nohui-hua-niokeruraihusaikurumemoricao-zuo-jia/korshunov.aija2026-07-15T04:22:56.401665+00:00MemOpsベンチマークは長時間の会話におけるライフサイクルメモリ操作を評価する https://korshunov.ai/ko/article/12032-memops-bencimakeuneun-janggi-daehwayi-sumyeong-jugi-memori-yeonsaneul/korshunov.aiko2026-07-15T04:22:56.401665+00:00MemOps 벤치마크는 장기 대화의 수명 주기 메모리 연산을 평가합니다 https://korshunov.ai/fr/article/12032-le-benchmark-memops-evalue-les-operations-de-memoire-de-cycle-de-vie-dans-les-a/korshunov.aifr2026-07-15T04:22:56.401665+00:00Le benchmark MemOps évalue les opérations de mémoire de cycle de vie dans les conversations à long terme https://korshunov.ai/id/article/12032-benchmark-memops-mengevaluasi-operasi-memori-siklus-hidup-dalam-percakapan/korshunov.aiid2026-07-15T04:22:56.401665+00:00Benchmark MemOps mengevaluasi operasi memori siklus hidup dalam percakapan jangka panjang https://korshunov.ai/ar/article/12032-tqywm-mns-memops-mlyt-dhkr-dwr-lhy-fy-lmhdtht-twyl-lmd/korshunov.aiar2026-07-15T04:22:56.401665+00:00تقيّم منصة MemOps عمليات ذاكرة دورة الحياة في المحادثات طويلة المدى https://korshunov.ai/id/article/12083-studi-mengungkap-bahwa-akurasi-agregat-menyembunyikan-pembalikan-prediksi-dari/korshunov.aiid2026-07-15T04:14:20.552130+00:00Studi mengungkap bahwa akurasi agregat menyembunyikan pembalikan prediksi dari konteks yang tidak relevan dengan tugas https://korshunov.ai/en/article/12083-study-reveals-aggregate-accuracy-hides-prediction-flips-from-task-irrelevant/korshunov.aien2026-07-15T04:14:20.552130+00:00Study reveals aggregate accuracy hides prediction flips from task-irrelevant context https://korshunov.ai/ru/article/12083-issledovanie-pokazyvaet-chto-sovokupnaia-tochnost-skryvaet-perekliucheniia-iz-za/korshunov.airu2026-07-15T04:14:20.552130+00:00Исследование показывает, что совокупная точность скрывает переключения предсказаний из-за контекста, не относящегося к задаче https://korshunov.ai/es/article/12083-un-estudio-revela-que-la-precision-agregada-oculta-cambios-de-prediccion-debidos/korshunov.aies2026-07-15T04:14:20.552130+00:00Un estudio revela que la precisión agregada oculta cambios de predicción debidos a contexto irrelevante para la tarea https://korshunov.ai/zh-hans/article/12083-yan-jiu-biao-ming-ju-he-jing-du-yan-gai-liao-you-yu-ren-wu-wu-guan-de-shang-xia/korshunov.aizh-cn2026-07-15T04:14:20.552130+00:00研究表明，聚合精度掩盖了由与任务无关的上下文引起的预测翻转 https://korshunov.ai/pt-br/article/12083-estudo-revela-que-a-precisao-agregada-oculta-mudancas-de-previsao-devido-a-para/korshunov.aipt2026-07-15T04:14:20.552130+00:00Estudo revela que a precisão agregada oculta mudanças de previsão devido a contexto irrelevante para a tarefa https://korshunov.ai/hi/article/12083-adhyyn-se-ptaa-cltaa-hai-ki-sncyii-sttiiktaa-kaary-asnbndhii-sndrbh-se-ke-ulttne/korshunov.aihi2026-07-15T04:14:20.552130+00:00अध्ययन से पता चलता है कि संचयी सटीकता कार्य-असंबंधी संदर्भ से भविष्यवाणी के उलटने को छिपाती है https://korshunov.ai/ja/article/12083-ji-ji-jing-du-katasukuwu-guan-xi-nawen-mo-niyoruyu-ce-nofan-zhuan-woyin-bi-suyan/korshunov.aija2026-07-15T04:14:20.552130+00:00集計精度がタスク無関係な文脈による予測の反転を隠蔽していることを示す研究 https://korshunov.ai/ko/article/12083-jibgye-jeonghwagdoga-jageob-mugwan-munmaegeuro-inhan-yeceug-dwijibgireul-yeongu/korshunov.aiko2026-07-15T04:14:20.552130+00:00집계 정확도가 작업 무관 문맥으로 인한 예측 뒤집기를 숨긴다는 연구 결과 https://korshunov.ai/fr/article/12083-une-etude-revele-que-la-precision-agregee-masque-les-inversions-de-prediction-a/korshunov.aifr2026-07-15T04:14:20.552130+00:00Une étude révèle que la précision agrégée masque les inversions de prédiction dues à un contexte non pertinent https://korshunov.ai/ar/article/12083-drs-tkshf-n-ldq-ljmly-tkhfy-tqlbt-ltnbw-lntj-n-syq-gyr-dhy-sl-blmhm/korshunov.aiar2026-07-15T04:14:20.552130+00:00دراسة تكشف أن الدقة الإجمالية تخفي تقلبات التنبؤ الناتجة عن سياق غير ذي صلة بالمهمة https://korshunov.ai/en/article/12082-palmclaw-native-on-device-agent-framework-for-mobile-phones/korshunov.aien2026-07-15T04:07:59.739942+00:00PalmClaw: Native on-device agent framework for mobile phones https://korshunov.ai/ru/article/12082-palmclaw-nativnaia-freimvork-dlia-agentov-na-ustroistve-dlia-mobilnykh-telefonov/korshunov.airu2026-07-15T04:07:59.739942+00:00PalmClaw: Нативная фреймворк для агентов на устройстве для мобильных телефонов https://korshunov.ai/es/article/12082-palmclaw-marco-de-agente-nativo-para-dispositivos-moviles/korshunov.aies2026-07-15T04:07:59.739942+00:00PalmClaw: Marco de agente nativo para dispositivos móviles https://korshunov.ai/zh-hans/article/12082-palmclaw-mian-xiang-shou-ji-de-ben-di-she-bei-yuan-sheng-zhi-neng-ti-kuang-jia/korshunov.aizh-cn2026-07-15T04:07:59.739942+00:00PalmClaw：面向手机的本地设备原生智能体框架 https://korshunov.ai/pt-br/article/12082-palmclaw-framework-de-agente-nativo-para-dispositivos-moveis/korshunov.aipt2026-07-15T04:07:59.739942+00:00PalmClaw: Framework de agente nativo para dispositivos móveis https://korshunov.ai/hi/article/12082-palmclaw-mobaail-phon-ke-lie-nettiv-ddivaais-saaidd-ejentt-phremvrk/korshunov.aihi2026-07-15T04:07:59.739942+00:00PalmClaw: मोबाइल फोन के लिए नेटिव डिवाइस-साइड एजेंट फ्रेमवर्क https://korshunov.ai/ja/article/12082-palmclaw-mohairuduan-mo-xiang-kenoneiteihuontehaisue-shientohure-muwa-ku/korshunov.aija2026-07-15T04:07:59.739942+00:00PalmClaw: モバイル端末向けのネイティブオンデバイスエージェントフレームワーク https://korshunov.ai/ko/article/12082-palmclaw-mobail-gigireul-wihan-neitibeu-ondibaiseu-eijeonteu-peureimweokeu/korshunov.aiko2026-07-15T04:07:59.739942+00:00PalmClaw: 모바일 기기를 위한 네이티브 온디바이스 에이전트 프레임워크 https://korshunov.ai/fr/article/12082-palmclaw-framework-d-agent-natif-sur-l-appareil-pour-telephones-mobiles/korshunov.aifr2026-07-15T04:07:59.739942+00:00PalmClaw : Framework d'agent natif sur l'appareil pour téléphones mobiles https://korshunov.ai/id/article/12082-palmclaw-kerangka-kerja-agen-on-device-asli-untuk-ponsel/korshunov.aiid2026-07-15T04:07:59.739942+00:00PalmClaw: Kerangka kerja agen on-device asli untuk ponsel https://korshunov.ai/ar/article/12082-palmclaw-tr-ml-wkyl-sly-l-ljhz-llhwtf-lmhmwl/korshunov.aiar2026-07-15T04:07:59.739942+00:00PalmClaw: إطار عمل وكيل أصلي على الجهاز للهواتف المحمولة https://korshunov.ai/en/article/12081-e3-reduces-llm-agent-cost-by-85-through-complexity-aware-execution-scope/korshunov.aien2026-07-15T03:58:42.253095+00:00E3 reduces LLM agent cost by 85% through complexity-aware execution scope estimation https://korshunov.ai/ru/article/12081-e3-snizhaet-stoimost-raboty-llm-agenta-na-85-za-schet-otsenki-oblasti-s-uchetom/korshunov.airu2026-07-15T03:58:42.253095+00:00E3 снижает стоимость работы LLM-агента на 85% за счёт оценки области выполнения с учётом сложности https://korshunov.ai/es/article/12081-e3-reduce-el-costo-de-agentes-llm-en-un-85-mediante-la-estimacion-del-alcance-de/korshunov.aies2026-07-15T03:58:42.253095+00:00E3 reduce el costo de agentes LLM en un 85% mediante la estimación del alcance de ejecución consciente de la complejidad https://korshunov.ai/zh-hans/article/12081-e3-tong-guo-gan-zhi-fu-za-du-de-zhi-xing-fan-wei-gu-ji-jiang-llm-zhi-neng-ti-ben/korshunov.aizh-cn2026-07-15T03:58:42.253095+00:00E3 通过感知复杂度的执行范围估计将 LLM 智能体成本降低 85% https://korshunov.ai/pt-br/article/12081-e3-reduz-o-custo-de-agentes-llm-em-85-por-meio-da-estimativa-do-escopo-de-da/korshunov.aipt2026-07-15T03:58:42.253095+00:00E3 reduz o custo de agentes LLM em 85% por meio da estimativa do escopo de execução consciente da complexidade https://korshunov.ai/hi/article/12081-e3-jttiltaa-jaagruuk-nisspaadn-shrennii-anumaan-ke-maadhym-se-llm-ejentt-kii-ko/korshunov.aihi2026-07-15T03:58:42.253095+00:00E3 जटिलता-जागरूक निष्पादन श्रेणी अनुमान के माध्यम से LLM एजेंट की लागत को 85% कम करता है https://korshunov.ai/ja/article/12081-e3hafu-za-xing-dui-ying-noshi-xing-suko-hutui-ding-niyorillme-xue-jian/korshunov.aija2026-07-15T03:58:42.253095+00:00E3は複雑性対応の実行スコープ推定によりLLMエージェントのコストを85%削減 https://korshunov.ai/ko/article/12081-e3-bogjabseong-inji-silhaeng-beomwi-cujeongeuro-llm-eijeonteu-biyong-85-jeolgam/korshunov.aiko2026-07-15T03:58:42.253095+00:00E3, 복잡성 인지 실행 범위 추정으로 LLM 에이전트 비용 85% 절감 https://korshunov.ai/fr/article/12081-e3-reduit-le-cout-des-agents-llm-de-85-grace-a-une-estimation-du-perimetre-d-de/korshunov.aifr2026-07-15T03:58:42.253095+00:00E3 réduit le coût des agents LLM de 85 % grâce à une estimation du périmètre d'exécution consciente de la complexité https://korshunov.ai/id/article/12081-e3-mengurangi-biaya-agen-llm-sebesar-85-melalui-estimasi-ruang-lingkup-eksekusi/korshunov.aiid2026-07-15T03:58:42.253095+00:00E3 mengurangi biaya agen LLM sebesar 85% melalui estimasi ruang lingkup eksekusi yang sadar kompleksitas https://korshunov.ai/ar/article/12081-yqll-e3-tklf-wkl-llm-bnsb-85-mn-khll-tqdyr-ntq-ltnfydh-lw-y-llt-qyd/korshunov.aiar2026-07-15T03:58:42.253095+00:00يقلل E3 تكلفة وكلاء LLM بنسبة 85% من خلال تقدير نطاق التنفيذ الواعي للتعقيد https://korshunov.ai/en/article/12080-tencent-releases-hy-embodied-vlm-1-0-efficient-embodied-foundation-model/korshunov.aien2026-07-15T03:48:52.923306+00:00Tencent releases Hy-Embodied-VLM-1.0 efficient embodied foundation model https://korshunov.ai/ru/article/12080-tencent-vypustila-effektivnuiu-bazovuiu-model-hy-embodied-vlm-1-0-dlia-agentov/korshunov.airu2026-07-15T03:48:52.923306+00:00Tencent выпустила эффективную базовую модель Hy-Embodied-VLM-1.0 для воплощённых агентов https://korshunov.ai/es/article/12080-tencent-lanza-hy-embodied-vlm-1-0-modelo-base-eficiente-de-vision-y-lenguaje/korshunov.aies2026-07-15T03:48:52.923306+00:00Tencent lanza Hy-Embodied-VLM-1.0, modelo base eficiente de visión y lenguaje para agentes encarnados https://korshunov.ai/zh-hans/article/12080-teng-xun-fa-bu-gao-xiao-ju-shen-ji-chu-mo-xing-hy-embodied-vlm-1-0/korshunov.aizh-cn2026-07-15T03:48:52.923306+00:00腾讯发布高效具身基础模型 Hy-Embodied-VLM-1.0 https://korshunov.ai/pt-br/article/12080-tencent-lanca-hy-embodied-vlm-1-0-modelo-de-base-encarnado-eficiente/korshunov.aipt2026-07-15T03:48:52.923306+00:00Tencent lança Hy-Embodied-VLM-1.0, modelo de base encarnado eficiente https://korshunov.ai/hi/article/12080-tencent-ne-hy-embodied-vlm-1-0-kushl-embeddedd-phaaunddeshn-monddl-jaarii-kiyaa/korshunov.aihi2026-07-15T03:48:52.923306+00:00Tencent ने Hy-Embodied-VLM-1.0 कुशल एम्बेडेड फाउंडेशन मॉडल जारी किया https://korshunov.ai/ja/article/12080-tencentkahy-embodied-vlm-1-0noxiao-lu-de-naju-xian-hua-ji-pan-moteruworiri-su/korshunov.aija2026-07-15T03:48:52.923306+00:00TencentがHy-Embodied-VLM-1.0の効率的な具現化基盤モデルをリリース https://korshunov.ai/ko/article/12080-tencent-hyoyuljeogin-gusang-giban-model-hy-embodied-vlm-1-0-culsi/korshunov.aiko2026-07-15T03:48:52.923306+00:00Tencent, 효율적인 구상 기반 모델 Hy-Embodied-VLM-1.0 출시 https://korshunov.ai/fr/article/12080-tencent-lance-hy-embodied-vlm-1-0-un-modele-fondamental-d-incarnation-efficace/korshunov.aifr2026-07-15T03:48:52.923306+00:00Tencent lance Hy-Embodied-VLM-1.0, un modèle fondamental d'incarnation efficace https://korshunov.ai/id/article/12080-tencent-merilis-hy-embodied-vlm-1-0-model-fondasi-embodied-yang-efisien/korshunov.aiid2026-07-15T03:48:52.923306+00:00Tencent merilis Hy-Embodied-VLM-1.0, model fondasi embodied yang efisien https://korshunov.ai/ar/article/12080-tysnt-ttlq-hy-embodied-vlm-1-0-nmwdhj-ssy-llwkl-ljsdyyn-bkf-ly/korshunov.aiar2026-07-15T03:48:52.923306+00:00تيسنت تطلق Hy-Embodied-VLM-1.0، نموذج أساسي للوكلاء الجسديين بكفاءة عالية https://korshunov.ai/en/article/12031-vr-study-finds-cvi-patients-have-slower-collision-detection-and-larger-safety/korshunov.aien2026-07-15T03:28:23.881796+00:00VR study finds CVI patients have slower collision detection and larger safety margins https://korshunov.ai/ru/article/12031-issledovanie-vr-pokazyvaet-chto-patsienty-s-tsvi-imeiut-bolee-medlennoe-i-zony/korshunov.airu2026-07-15T03:28:23.881796+00:00Исследование VR показывает, что пациенты с ЦВИ имеют более медленное обнаружение столкновений и большие зоны безопасности https://korshunov.ai/es/article/12031-estudio-de-rv-encuentra-que-pacientes-con-ivt-tienen-deteccion-de-colisiones-mas/korshunov.aies2026-07-15T03:28:23.881796+00:00Estudio de RV encuentra que pacientes con IVT tienen detección de colisiones más lenta y mayores márgenes de seguridad https://korshunov.ai/zh-hans/article/12031-vryan-jiu-fa-xian-cvihuan-zhe-de-peng-zhuang-jian-ce-su-du-jiao-man-qie-an-quan/korshunov.aizh-cn2026-07-15T03:28:23.881796+00:00VR研究发现，CVI患者的碰撞检测速度较慢且安全余量更大 https://korshunov.ai/pt-br/article/12031-estudo-em-rv-descobre-que-pacientes-com-ivt-tem-deteccao-de-colisao-mais-lenta-e/korshunov.aipt2026-07-15T03:28:23.881796+00:00Estudo em RV descobre que pacientes com IVT têm detecção de colisão mais lenta e maiores margens de segurança https://korshunov.ai/hi/article/12031-vr-adhyyn-paataa-hai-ki-cvi-rogiyon-men-ttkraav-kaa-ptaa-lgaane-kii-gti-dhiimii/korshunov.aihi2026-07-15T03:28:23.881796+00:00VR अध्ययन पाता है कि CVI रोगियों में टकराव का पता लगाने की गति धीमी और सुरक्षा मार्जिन बड़ा होता है https://korshunov.ai/ja/article/12031-vryan-jiu-cvihuan-zhe-hachong-tu-jian-chu-kachi-ku-an-quan-ma-shinkada-kii/korshunov.aija2026-07-15T03:28:23.881796+00:00VR研究：CVI患者は衝突検出が遅く、安全マージンが大きい https://korshunov.ai/ko/article/12031-vr-yeongu-cvi-hwanjayi-cungdol-gamjiga-neurigo-anjeon-majini-deo-keum/korshunov.aiko2026-07-15T03:28:23.881796+00:00VR 연구: CVI 환자의 충돌 감지가 느리고 안전 마진이 더 큼 https://korshunov.ai/fr/article/12031-une-etude-en-rv-revele-que-les-patients-atteints-de-civ-ont-une-detection-des-et/korshunov.aifr2026-07-15T03:28:23.881796+00:00Une étude en RV révèle que les patients atteints de CIV ont une détection des collisions plus lente et des marges de sécurité plus grandes https://korshunov.ai/id/article/12031-studi-vr-menemukan-pasien-cvi-memiliki-deteksi-tabrakan-yang-lebih-lambat-dan/korshunov.aiid2026-07-15T03:28:23.881796+00:00Studi VR menemukan pasien CVI memiliki deteksi tabrakan yang lebih lambat dan margin keamanan yang lebih besar https://korshunov.ai/ar/article/12031-tzhr-drs-blwq-lftrdy-n-mrd-cvi-ldyhm-kshf-tsdm-bt-whwmsh-mn-kbr/korshunov.aiar2026-07-15T03:28:23.881796+00:00تظهر دراسة بالواقع الافتراضي أن مرضى CVI لديهم كشف تصادم أبطأ وهوامش أمان أكبر https://korshunov.ai/zh-hans/article/11737-ji-jie-ke-jie-shi-xing-jie-shi-llm-as-judge-zai-yin-cang-zhuang-tai-ji-he-zhong/korshunov.aizh-cn2026-07-15T03:19:16.315314+00:00机械可解释性揭示 LLM-as-Judge 在隐藏状态几何中的偏见 https://korshunov.ai/en/article/11737-mechanistic-interpretability-reveals-llm-as-judge-bias-in-hidden-state-geometry/korshunov.aien2026-07-15T03:19:16.315314+00:00Mechanistic interpretability reveals LLM-as-Judge bias in hidden state geometry https://korshunov.ai/ru/article/11737-mekhanisticheskaia-interpretiruemost-vyiavliaet-smeshchenie-llm-as-judge-v/korshunov.airu2026-07-15T03:19:16.315314+00:00Механистическая интерпретируемость выявляет смещение LLM-as-Judge в геометрии скрытых состояний https://korshunov.ai/es/article/11737-la-interpretabilidad-mecanicista-revela-sesgo-de-llm-as-judge-en-la-geometria/korshunov.aies2026-07-15T03:19:16.315314+00:00La interpretabilidad mecanicista revela sesgo de LLM-as-Judge en la geometría del estado oculto https://korshunov.ai/pt-br/article/11737-interpretabilidade-mecanicista-revela-vies-do-llm-as-judge-na-geometria-do/korshunov.aipt2026-07-15T03:19:16.315314+00:00Interpretabilidade mecanicista revela viés do LLM-as-Judge na geometria do estado oculto https://korshunov.ai/hi/article/11737-yaantrik-vyaakhyaatmktaa-chipii-huii-avsthaa-jyaamiti-men-llm-as-judge-pksspaat/korshunov.aihi2026-07-15T03:19:16.315314+00:00यांत्रिक व्याख्यात्मकता छिपी हुई अवस्था ज्यामिति में LLM-as-Judge पक्षपात को उजागर करती है https://korshunov.ai/ja/article/11737-mekanisuteitsukuinta-huritahiriteikallm-as-judgenohaiasuwoyin-rezhuang-tai-ji-he/korshunov.aija2026-07-15T03:19:16.315314+00:00メカニスティックインタープリタビリティがLLM-as-Judgeのバイアスを隠れ状態幾何学で明らかにする https://korshunov.ai/ko/article/11737-gigyejeog-haeseogganeungseongi-llm-as-judge-pyeonhyangeul-eunnig-sangtae/korshunov.aiko2026-07-15T03:19:16.315314+00:00기계적 해석가능성이 LLM-as-Judge 편향을 은닉 상태 기하학에서 밝혀냄 https://korshunov.ai/fr/article/11737-l-interpretabilite-mecaniste-revele-le-biais-du-llm-as-judge-dans-la-geometrie/korshunov.aifr2026-07-15T03:19:16.315314+00:00L'interprétabilité mécaniste révèle le biais du LLM-as-Judge dans la géométrie des états cachés https://korshunov.ai/id/article/11737-interpretabilitas-mekanistik-mengungkap-bias-llm-as-judge-dalam-geometri-keadaan/korshunov.aiid2026-07-15T03:19:16.315314+00:00Interpretabilitas mekanistik mengungkap bias LLM-as-Judge dalam geometri keadaan tersembunyi https://korshunov.ai/ar/article/11737-tfsyr-lly-ykshf-thyz-llm-as-judge-fy-hnds-lhlt-lmkhfy/korshunov.aiar2026-07-15T03:19:16.315314+00:00تفسير الآلية يكشف تحيز LLM-as-Judge في هندسة الحالات المخفية https://korshunov.ai/en/article/11735-study-finds-teaching-feedback-classification-protocol-durable-across-languages/korshunov.aien2026-07-15T03:09:48.086934+00:00Study finds teaching-feedback classification protocol durable across languages and models https://korshunov.ai/ru/article/11735-issledovanie-pokazyvaet-chto-protokol-klassifikatsii-obratnoi-sviazi-po-k-i/korshunov.airu2026-07-15T03:09:48.086934+00:00Исследование показывает, что протокол классификации обратной связи по обучению устойчив к языкам и моделям https://korshunov.ai/es/article/11735-el-estudio-encuentra-que-el-protocolo-de-clasificacion-de-retroalimentacion-es-y/korshunov.aies2026-07-15T03:09:48.086934+00:00El estudio encuentra que el protocolo de clasificación de retroalimentación docente es duradero entre idiomas y modelos https://korshunov.ai/zh-hans/article/11735-yan-jiu-fa-xian-jiao-xue-fan-kui-fen-lei-xie-yi-zai-yu-yan-he-mo-xing-jian-ju/korshunov.aizh-cn2026-07-15T03:09:48.086934+00:00研究发现教学反馈分类协议在语言和模型间具有持久性 https://korshunov.ai/pt-br/article/11735-estudo-descobre-que-protocolo-de-classificacao-de-feedback-de-ensino-e-duravel-e/korshunov.aipt2026-07-15T03:09:48.086934+00:00Estudo descobre que protocolo de classificação de feedback de ensino é durável entre idiomas e modelos https://korshunov.ai/hi/article/11735-adhyyn-paataa-hai-ki-shikssnn-prtikriyaa-vrgiikrnn-prottokonl-bhaassaaon-aur-ke/korshunov.aihi2026-07-15T03:09:48.086934+00:00अध्ययन पाता है कि शिक्षण-प्रतिक्रिया वर्गीकरण प्रोटोकॉल भाषाओं और मॉडलों के पार टिकाऊ है https://korshunov.ai/ja/article/11735-yan-yu-yamoteruwochao-etechi-sok-ke-neng-najiao-shou-hui-tohatsukufen-lei-jiu/korshunov.aija2026-07-15T03:09:48.086934+00:00言語やモデルを超えて持続可能な教授フィードバック分類プロトコルに関する研究 https://korshunov.ai/ko/article/11735-eoneo-mic-modele-geolcyeo-naeguseongeul-gajin-gyosu-pideubaeg-bunryu-peurotokol/korshunov.aiko2026-07-15T03:09:48.086934+00:00언어 및 모델에 걸쳐 내구성을 가진 교수 피드백 분류 프로토콜 연구 https://korshunov.ai/fr/article/11735-une-etude-revele-qu-un-protocole-de-classification-des-commentaires-pedagogiques/korshunov.aifr2026-07-15T03:09:48.086934+00:00Une étude révèle qu'un protocole de classification des commentaires pédagogiques est durable à travers les langues et les modèles https://korshunov.ai/id/article/11735-studi-menemukan-protokol-klasifikasi-umpan-balik-pengajaran-tahan-lama-lintas/korshunov.aiid2026-07-15T03:09:48.086934+00:00Studi menemukan protokol klasifikasi umpan balik pengajaran tahan lama lintas bahasa dan model https://korshunov.ai/ar/article/11735-drs-tjd-n-brwtwkwl-tsnyf-mlhzt-ltdrys-mtyn-br-llgt-wlnmdhj/korshunov.aiar2026-07-15T03:09:48.086934+00:00دراسة تجد أن بروتوكول تصنيف ملاحظات التدريس متين عبر اللغات والنماذج https://korshunov.ai/zh-hans/article/11719-regrind-cong-dan-ci-ren-lei-yan-shi-zhong-xue-xi-ling-qiao-cao-zuo-ce-lue/korshunov.aizh-cn2026-07-15T02:58:40.262251+00:00REGRIND 从单次人类演示中学习灵巧操作策略 https://korshunov.ai/en/article/11719-regrind-learns-dexterous-manipulation-policies-from-a-single-human-demonstration/korshunov.aien2026-07-15T02:58:40.262251+00:00REGRIND learns dexterous manipulation policies from a single human demonstration https://korshunov.ai/ru/article/11719-regrind-obuchaet-lovkim-politikam-manipuliatsii-po-odnoi-demonstratsii-cheloveka/korshunov.airu2026-07-15T02:58:40.262251+00:00REGRIND обучает ловким политикам манипуляции по одной демонстрации человека https://korshunov.ai/es/article/11719-regrind-aprende-politicas-de-manipulacion-diestra-a-partir-de-una-unica-humana/korshunov.aies2026-07-15T02:58:40.262251+00:00REGRIND aprende políticas de manipulación diestra a partir de una única demostración humana https://korshunov.ai/pt-br/article/11719-regrind-aprende-politicas-de-manipulacao-destreosa-a-partir-de-uma-unica-humana/korshunov.aipt2026-07-15T02:58:40.262251+00:00REGRIND aprende políticas de manipulação destreosa a partir de uma única demonstração humana https://korshunov.ai/hi/article/11719-regrind-ekl-maanv-prdrshn-se-kushl-hainddling-niitiyon-ko-siikhtaa-hai/korshunov.aihi2026-07-15T02:58:40.262251+00:00REGRIND एकल मानव प्रदर्शन से कुशल हैंडलिंग नीतियों को सीखता है https://korshunov.ai/ja/article/11719-regrindha1hui-noren-jian-temokaraqi-yong-nacao-zuo-horishi-woxue-xi-suru/korshunov.aija2026-07-15T02:58:40.262251+00:00REGRINDは1回の人間デモから器用な操作ポリシーを学習する https://korshunov.ai/ko/article/11719-regrindeun-danil-ingan-siyeoneseo-jeonggyohan-jojag-jeongcaegeul-hagseubhabnida/korshunov.aiko2026-07-15T02:58:40.262251+00:00REGRIND은 단일 인간 시연에서 정교한 조작 정책을 학습합니다 https://korshunov.ai/fr/article/11719-regrind-apprend-des-politiques-de-manipulation-delicate-a-partir-d-une-seule/korshunov.aifr2026-07-15T02:58:40.262251+00:00REGRIND apprend des politiques de manipulation délicate à partir d'une seule démonstration humaine https://korshunov.ai/id/article/11719-regrind-mempelajari-kebijakan-manipulasi-cekatan-dari-satu-demonstrasi-manusia/korshunov.aiid2026-07-15T02:58:40.262251+00:00REGRIND mempelajari kebijakan manipulasi cekatan dari satu demonstrasi manusia https://korshunov.ai/ar/article/11719-yt-lm-regrind-syst-ltl-b-ldqyq-mn-rd-bshry-whd/korshunov.aiar2026-07-15T02:58:40.262251+00:00يتعلم REGRIND سياسات التلاعب الدقيقة من عرض بشري واحد https://korshunov.ai/en/article/11973-openai-reports-2-5x-usage-growth-for-codex-prismml-releases-bonsai-27b/korshunov.aien2026-07-15T02:39:08.525833+00:00OpenAI reports 2.5x usage growth for Codex; PrismML releases Bonsai 27B https://korshunov.ai/ru/article/11973-openai-soobshchaet-o-roste-ispolzovaniia-codex-v-25-raza-prismml-vypuskaet-27b/korshunov.airu2026-07-15T02:39:08.525833+00:00OpenAI сообщает о росте использования Codex в 2,5 раза; PrismML выпускает Bonsai 27B https://korshunov.ai/es/article/11973-openai-reporta-un-crecimiento-de-uso-de-2-5x-para-codex-prismml-lanza-bonsai-27b/korshunov.aies2026-07-15T02:39:08.525833+00:00OpenAI reporta un crecimiento de uso de 2.5x para Codex; PrismML lanza Bonsai 27B https://korshunov.ai/zh-hans/article/11973-openaibao-gao-codexshi-yong-liang-zeng-chang-2-5bei-prismmlfa-bu-bonsai-27b/korshunov.aizh-cn2026-07-15T02:39:08.525833+00:00OpenAI报告Codex使用量增长2.5倍；PrismML发布Bonsai 27B https://korshunov.ai/pt-br/article/11973-openai-relata-crescimento-de-uso-de-25x-para-codex-prismml-lanca-bonsai-27b/korshunov.aipt2026-07-15T02:39:08.525833+00:00OpenAI relata crescimento de uso de 2,5x para Codex; PrismML lança Bonsai 27B https://korshunov.ai/hi/article/11973-openai-ne-codex-ke-lie-2-5x-upyog-vrddhi-kii-riportt-dii-prismml-ne-bonsai-27b/korshunov.aihi2026-07-15T02:39:08.525833+00:00OpenAI ने Codex के लिए 2.5x उपयोग वृद्धि की रिपोर्ट दी; PrismML ने Bonsai 27B जारी किया https://korshunov.ai/ja/article/11973-openai-codexnoli-yong-lu-ka2-5bei-nicheng-chang-tobao-gao-prismmlkabonsai-su/korshunov.aija2026-07-15T02:39:08.525833+00:00OpenAI、Codexの利用率が2.5倍に成長と報告；PrismMLがBonsai 27Bをリリース https://korshunov.ai/ko/article/11973-openai-codex-sayongryang-2-5bae-seongjang-bogo-prismml-bonsai-27b-culsi/korshunov.aiko2026-07-15T02:39:08.525833+00:00OpenAI, Codex 사용량 2.5배 성장 보고; PrismML, Bonsai 27B 출시 https://korshunov.ai/fr/article/11973-openai-signale-une-croissance-de-l-utilisation-de-codex-de-25x-prismml-lance-27b/korshunov.aifr2026-07-15T02:39:08.525833+00:00OpenAI signale une croissance de l'utilisation de Codex de 2,5x ; PrismML lance Bonsai 27B https://korshunov.ai/id/article/11973-openai-melaporkan-pertumbuhan-penggunaan-codex-sebesar-25x-prismml-merilis-27b/korshunov.aiid2026-07-15T02:39:08.525833+00:00OpenAI melaporkan pertumbuhan penggunaan Codex sebesar 2,5x; PrismML merilis Bonsai 27B https://korshunov.ai/ar/article/11973-openai-yblg-n-nmw-stkhdm-codex-bnsb-2-5-mr-prismml-ttlq-bonsai-27b/korshunov.aiar2026-07-15T02:39:08.525833+00:00OpenAI يبلغ عن نمو استخدام Codex بنسبة 2.5 مرة؛ PrismML تطلق Bonsai 27B https://korshunov.ai/en/article/11972-ai-engineering-shifts-focus-from-agents-to-harness-systems-and-loop-control/korshunov.aien2026-07-15T02:16:11.529170+00:00AI engineering shifts focus from agents to harness systems and loop control https://korshunov.ai/id/article/11972-fokus-rekayasa-ai-bergeser-dari-agen-ke-sistem-harness-dan-kontrol-loop/korshunov.aiid2026-07-15T02:16:11.529170+00:00Fokus rekayasa AI bergeser dari agen ke sistem harness dan kontrol loop https://korshunov.ai/ru/article/11972-inzheneriia-ii-smeshchaet-fokus-s-agentov-na-sistemy-konteinery-i-upravlenie/korshunov.airu2026-07-15T02:16:11.529170+00:00Инженерия ИИ смещает фокус с агентов на системы-контейнеры и управление циклами https://korshunov.ai/es/article/11972-la-ingenieria-de-ia-desplaza-el-enfoque-de-agentes-a-sistemas-de-arnes-y-control/korshunov.aies2026-07-15T02:16:11.529170+00:00La ingeniería de IA desplaza el enfoque de agentes a sistemas de arnés y control de bucles https://korshunov.ai/zh-hans/article/11972-aigong-cheng-jiang-jiao-dian-cong-zhi-neng-ti-zhuan-xiang-harnessxi-tong-he-xun/korshunov.aizh-cn2026-07-15T02:16:11.529170+00:00AI工程将焦点从智能体转向Harness系统和循环控制 https://korshunov.ai/pt-br/article/11972-engenharia-de-ia-desloca-o-foco-de-agentes-para-sistemas-de-harness-e-controle/korshunov.aipt2026-07-15T02:16:11.529170+00:00Engenharia de IA desloca o foco de agentes para sistemas de harness e controle de loops https://korshunov.ai/hi/article/11972-ai-injiiniyring-kaa-phoks-ejentts-se-haarnes-sisttms-aur-luup-knttrol-kii-or-bdl/korshunov.aihi2026-07-15T02:16:11.529170+00:00AI इंजीनियरिंग का फोकस एजेंट्स से हार्नेस सिस्टम्स और लूप कंट्रोल की ओर बदल रहा है https://korshunov.ai/ja/article/11972-aienshiniarinkunojiao-dian-kae-shientokaraha-nesushisutemutoru-huzhi-yu/korshunov.aija2026-07-15T02:16:11.529170+00:00AIエンジニアリングの焦点がエージェントからハーネスシステムとループ制御へシフト https://korshunov.ai/ko/article/11972-ai-enjinieoringyi-cojeomi-eijeonteueseo-haniseu-siseutem-mic-rupeu-jeeoro/korshunov.aiko2026-07-15T02:16:11.529170+00:00AI 엔지니어링의 초점이 에이전트에서 하니스 시스템 및 루프 제어로 전환 https://korshunov.ai/fr/article/11972-l-ingenierie-de-l-ia-se-tourne-des-agents-vers-les-systemes-de-harnais-et-le-par/korshunov.aifr2026-07-15T02:16:11.529170+00:00L'ingénierie de l'IA se tourne des agents vers les systèmes de harnais et le contrôle par boucle https://korshunov.ai/ar/article/11972-thwl-hnds-ldhk-lstn-y-mn-lwkl-l-nzm-lharness-wlthkm-fy-lhlq/korshunov.aiar2026-07-15T02:16:11.529170+00:00تحول هندسة الذكاء الاصطناعي من الوكلاء إلى أنظمة الهarness والتحكم في الحلقة https://korshunov.ai/en/article/11971-openai-codex-usage-surges-2-5x-as-open-models-compress-to-consumer-devices/korshunov.aien2026-07-15T01:59:35.403042+00:00OpenAI Codex usage surges 2.5x as open models compress to consumer devices https://korshunov.ai/ru/article/11971-ispolzovanie-openai-codex-vyroslo-v-25-raza-po-mere-szhatiia-otkrytykh-modelei/korshunov.airu2026-07-15T01:59:35.403042+00:00Использование OpenAI Codex выросло в 2,5 раза по мере сжатия открытых моделей для потребительских устройств https://korshunov.ai/es/article/11971-el-uso-de-openai-codex-se-dispara-25-veces-a-medida-que-los-modelos-abiertos-se/korshunov.aies2026-07-15T01:59:35.403042+00:00El uso de OpenAI Codex se dispara 2,5 veces a medida que los modelos abiertos se comprimen para dispositivos de consumo https://korshunov.ai/zh-hans/article/11971-sui-zhao-kai-yuan-mo-xing-ya-suo-zhi-xiao-fei-ji-she-bei-openai-codexde-shi-yong/korshunov.aizh-cn2026-07-15T01:59:35.403042+00:00随着开源模型压缩至消费级设备，OpenAI Codex的使用量激增2.5倍 https://korshunov.ai/pt-br/article/11971-uso-do-openai-codex-dispara-25x-a-medida-que-modelos-open-source-sao-compactados/korshunov.aipt2026-07-15T01:59:35.403042+00:00Uso do OpenAI Codex dispara 2,5x à medida que modelos open source são compactados para dispositivos de consumo https://korshunov.ai/hi/article/11971-khule-monddl-upbhoktaa-ddivaaison-tk-snpiiddit-hone-ke-saath-openai-codex-kaa-2/korshunov.aihi2026-07-15T01:59:35.403042+00:00खुले मॉडल उपभोक्ता डिवाइसों तक संपीड़ित होने के साथ OpenAI Codex का उपयोग 2.5 गुना बढ़ा https://korshunov.ai/ja/article/11971-o-hunmoterukaxiao-fei-zhe-xiang-ketehaisuniya-suo-sareruzhong-openai-codexnoshi/korshunov.aija2026-07-15T01:59:35.403042+00:00オープンモデルが消費者向けデバイスに圧縮される中、OpenAI Codexの使用量が2.5倍に急増 https://korshunov.ai/ko/article/11971-opeun-modeli-sobija-gigiro-abcugdoemyeonseo-openai-codex-sayongryangi-2-5bae/korshunov.aiko2026-07-15T01:59:35.403042+00:00오픈 모델이 소비자 기기로 압축되면서 OpenAI Codex 사용량이 2.5배 급증 https://korshunov.ai/fr/article/11971-l-utilisation-d-openai-codex-explose-de-25x-alors-que-les-modeles-ouverts-se-les/korshunov.aifr2026-07-15T01:59:35.403042+00:00L'utilisation d'OpenAI Codex explose de 2,5x alors que les modèles ouverts se compressent vers les appareils grand public https://korshunov.ai/id/article/11971-penggunaan-openai-codex-melonjak-25x-saat-model-terbuka-dikompresi-ke-perangkat/korshunov.aiid2026-07-15T01:59:35.403042+00:00Penggunaan OpenAI Codex melonjak 2,5x saat model terbuka dikompresi ke perangkat konsumen https://korshunov.ai/ar/article/11971-stkhdm-openai-codex-yrtf-bnsb-2-5-mr-bynm-tdgt-lnmdhj-lmftwh-nhw-ljhz-lsthlky/korshunov.aiar2026-07-15T01:59:35.403042+00:00استخدام OpenAI Codex يرتفع بنسبة 2.5 مرة بينما تضغط النماذج المفتوحة نحو الأجهزة الاستهلاكية https://korshunov.ai/en/article/11970-audio-cpp-0-3-adds-supertonic-3-moss-tts-indextts2-and-irodori-tts-with-gguf/korshunov.aien2026-07-15T01:47:30.118202+00:00audio.cpp 0.3 adds Supertonic 3, MOSS-TTS, IndexTTS2, and Irodori-TTS with GGUF support https://korshunov.ai/ru/article/11970-audio-cpp-0-3-dobavliaet-supertonic-3-moss-tts-indextts2-i-irodori-tts-s-gguf/korshunov.airu2026-07-15T01:47:30.118202+00:00audio.cpp 0.3 добавляет Supertonic 3, MOSS-TTS, IndexTTS2 и Irodori-TTS с поддержкой GGUF https://korshunov.ai/es/article/11970-audio-cpp-0-3-anade-supertonic-3-moss-tts-indextts2-e-irodori-tts-con-soporte/korshunov.aies2026-07-15T01:47:30.118202+00:00audio.cpp 0.3 añade Supertonic 3, MOSS-TTS, IndexTTS2 e Irodori-TTS con soporte GGUF https://korshunov.ai/zh-hans/article/11970-audio-cpp-0-3-xin-zeng-supertonic-3-moss-tts-indextts2-he-irodori-tts-zhi-chi/korshunov.aizh-cn2026-07-15T01:47:30.118202+00:00audio.cpp 0.3 新增 Supertonic 3、MOSS-TTS、IndexTTS2 和 Irodori-TTS，支持 GGUF https://korshunov.ai/pt-br/article/11970-audio-cpp-0-3-adiciona-supertonic-3-moss-tts-indextts2-e-irodori-tts-com-suporte/korshunov.aipt2026-07-15T01:47:30.118202+00:00audio.cpp 0.3 adiciona Supertonic 3, MOSS-TTS, IndexTTS2 e Irodori-TTS com suporte GGUF https://korshunov.ai/hi/article/11970-audio-cpp-0-3-men-supertonic-3-moss-tts-indextts2-aur-irodori-tts-gguf-sportt-ke/korshunov.aihi2026-07-15T01:47:30.118202+00:00audio.cpp 0.3 में Supertonic 3, MOSS-TTS, IndexTTS2 और Irodori-TTS GGUF सपोर्ट के साथ जोड़े गए https://korshunov.ai/ja/article/11970-audio-cpp-0-3kasupertonic-3-moss-tts-indextts2-irodori-ttswoggufsaho-tofu-jia/korshunov.aija2026-07-15T01:47:30.118202+00:00audio.cpp 0.3がSupertonic 3、MOSS-TTS、IndexTTS2、Irodori-TTSをGGUFサポート付きで追加 https://korshunov.ai/ko/article/11970-audio-cpp-0-3-gguf-jiweongwa-hamgge-supertonic-3-moss-tts-indextts2-irodori-tts/korshunov.aiko2026-07-15T01:47:30.118202+00:00audio.cpp 0.3, GGUF 지원과 함께 Supertonic 3, MOSS-TTS, IndexTTS2, Irodori-TTS 추가 https://korshunov.ai/fr/article/11970-audio-cpp-0-3-ajoute-supertonic-3-moss-tts-indextts2-et-irodori-tts-avec-support/korshunov.aifr2026-07-15T01:47:30.118202+00:00audio.cpp 0.3 ajoute Supertonic 3, MOSS-TTS, IndexTTS2 et Irodori-TTS avec support GGUF https://korshunov.ai/id/article/11970-audio-cpp-0-3-menambahkan-supertonic-3-moss-tts-indextts2-dan-irodori-tts-dengan/korshunov.aiid2026-07-15T01:47:30.118202+00:00audio.cpp 0.3 menambahkan Supertonic 3, MOSS-TTS, IndexTTS2, dan Irodori-TTS dengan dukungan GGUF https://korshunov.ai/ar/article/11970-audio-cpp-0-3-tdyf-supertonic-3-wmoss-tts-windextts2-wirodori-tts-m-d-m-gguf/korshunov.aiar2026-07-15T01:47:30.118202+00:00audio.cpp 0.3 تضيف Supertonic 3 وMOSS-TTS وIndexTTS2 وIrodori-TTS مع دعم GGUF https://korshunov.ai/en/article/11975-user-ports-kvarn-to-bonsai-runtime-for-68-speedup-and-10gb-vram-at-120k-context/korshunov.aien2026-07-15T01:34:33.262900+00:00User ports KVarN to Bonsai runtime for 68% speedup and <10GB VRAM at 120K context https://korshunov.ai/ru/article/11975-polzovatel-perenes-kvarn-v-bonsai-runtime-dlia-uskoreniia-na-68-i-ispolzovaniia/korshunov.airu2026-07-15T01:34:33.262900+00:00Пользователь перенёс KVarN в Bonsai runtime для ускорения на 68% и использования <10 ГБ VRAM при контексте 120K https://korshunov.ai/es/article/11975-un-usuario-porta-kvarn-al-entorno-de-ejecucion-bonsai-para-una-aceleracion-del-y/korshunov.aies2026-07-15T01:34:33.262900+00:00Un usuario porta KVarN al entorno de ejecución Bonsai para una aceleración del 68% y <10GB de VRAM con contexto de 120K https://korshunov.ai/zh-hans/article/11975-yong-hu-jiang-kvarn-yi-zhi-dao-bonsai-yun-xing-shi-shi-xian-68-jia-su-ji-120k/korshunov.aizh-cn2026-07-15T01:34:33.262900+00:00用户将 KVarN 移植到 Bonsai 运行时，实现 68% 加速及 120K 上下文下 <10GB VRAM https://korshunov.ai/pt-br/article/11975-usuario-porta-kvarn-para-o-runtime-bonsai-obtendo-aceleracao-de-68-e-10gb-de-com/korshunov.aipt2026-07-15T01:34:33.262900+00:00Usuário porta KVarN para o runtime Bonsai, obtendo aceleração de 68% e <10GB de VRAM com contexto de 120K https://korshunov.ai/hi/article/11975-upyogkrtaa-ne-kvarn-ko-bonsai-rnttaaim-men-portt-kiyaa-jisse-68-gti-vrddhi-aur/korshunov.aihi2026-07-15T01:34:33.262900+00:00उपयोगकर्ता ने KVarN को Bonsai रनटाइम में पोर्ट किया, जिससे 68% गति वृद्धि और 120K संदर्भ पर <10GB VRAM का उपयोग https://korshunov.ai/ja/article/11975-yu-sa-kakvarnwobonsairantaimuniho-toshi-120kkontekisutote68-nogao-su-hua-man/korshunov.aija2026-07-15T01:34:33.262900+00:00ユーザーがKVarNをBonsaiランタイムにポートし、120Kコンテキストで68%の高速化と10GB未満のVRAMを実現 https://korshunov.ai/ko/article/11975-sayongjaga-kvarneul-bonsai-reontaimeuro-potinghayeo-120k-keontegseuteueseo-68/korshunov.aiko2026-07-15T01:34:33.262900+00:00사용자가 KVarN을 Bonsai 런타임으로 포팅하여 120K 컨텍스트에서 68% 속도 향상 및 10GB 미만 VRAM 달성 https://korshunov.ai/fr/article/11975-un-utilisateur-porte-kvarn-sur-le-runtime-bonsai-pour-un-gain-de-vitesse-de-68-a/korshunov.aifr2026-07-15T01:34:33.262900+00:00Un utilisateur porte KVarN sur le runtime Bonsai pour un gain de vitesse de 68 % et moins de 10 Go de VRAM à un contexte de 120K https://korshunov.ai/id/article/11975-pengguna-memindahkan-kvarn-ke-runtime-bonsai-untuk-percepatan-68-dan-vram-10gb/korshunov.aiid2026-07-15T01:34:33.262900+00:00Pengguna memindahkan KVarN ke runtime Bonsai untuk percepatan 68% dan VRAM <10GB pada konteks 120K https://korshunov.ai/ar/article/11975-yqwm-mstkhdm-bnql-kvarn-l-wqt-tshgyl-bonsai-lthqyq-tsry-bnsb-68-wvram-ql-mn-10/korshunov.aiar2026-07-15T01:34:33.262900+00:00يقوم مستخدم بنقل KVarN إلى وقت تشغيل Bonsai لتحقيق تسريع بنسبة 68% وVRAM أقل من 10 جيجابايت عند سياق 120K https://korshunov.ai/zh-hans/article/11974-wei-ruan-ceojing-gao-zi-tuo-guan-aike-bao-hu-zhuan-you-zhi-shi/korshunov.aizh-cn2026-07-15T01:22:35.382234+00:00微软CEO警告：自托管AI可保护专有知识 https://korshunov.ai/en/article/11974-microsoft-ceo-warns-self-hosting-ai-protects-proprietary-knowledge/korshunov.aien2026-07-15T01:22:35.382234+00:00Microsoft CEO warns self-hosting AI protects proprietary knowledge https://korshunov.ai/ru/article/11974-generalnyi-direktor-microsoft-preduprezhdaet-sobstvennoe-razmeshchenie-ii/korshunov.airu2026-07-15T01:22:35.382234+00:00Генеральный директор Microsoft предупреждает: собственное размещение ИИ защищает проприетарные знания https://korshunov.ai/es/article/11974-el-ceo-de-microsoft-advierte-que-alojar-ia-internamente-protege-el-conocimiento/korshunov.aies2026-07-15T01:22:35.382234+00:00El CEO de Microsoft advierte que alojar IA internamente protege el conocimiento propietario https://korshunov.ai/pt-br/article/11974-ceo-da-microsoft-alerta-que-hospedar-ia-internamente-protege-conhecimento/korshunov.aipt2026-07-15T01:22:35.382234+00:00CEO da Microsoft alerta que hospedar IA internamente protege conhecimento proprietário https://korshunov.ai/hi/article/11974-maaikrosonphtt-ke-siiiio-cetaavnii-dete-hain-ki-sv-hostting-ai-propraaittrii-kii/korshunov.aihi2026-07-15T01:22:35.382234+00:00माइक्रोसॉफ्ट के सीईओ चेतावनी देते हैं कि स्व-होस्टिंग AI प्रोप्राइटरी ज्ञान की रक्षा करता है https://korshunov.ai/ja/article/11974-microsoft-ceo-zi-she-hosuteinkuaikaji-mi-zhi-shi-wobao-hu-surutojing-gao/korshunov.aija2026-07-15T01:22:35.382234+00:00Microsoft CEO、自社ホスティングAIが機密知識を保護すると警告 https://korshunov.ai/ko/article/11974-microsoft-ceo-jace-hoseuting-aiga-dogjeom-jisigeul-bohohandago-gyeonggo/korshunov.aiko2026-07-15T01:22:35.382234+00:00Microsoft CEO, 자체 호스팅 AI가 독점 지식을 보호한다고 경고 https://korshunov.ai/fr/article/11974-le-pdg-de-microsoft-met-en-garde-l-auto-hebergement-de-l-ia-protege-les/korshunov.aifr2026-07-15T01:22:35.382234+00:00Le PDG de Microsoft met en garde : l'auto-hébergement de l'IA protège les connaissances propriétaires https://korshunov.ai/id/article/11974-ceo-microsoft-memperingatkan-bahwa-ai-yang-di-host-sendiri-melindungi/korshunov.aiid2026-07-15T01:22:35.382234+00:00CEO Microsoft memperingatkan bahwa AI yang di-host sendiri melindungi pengetahuan proprietaris https://korshunov.ai/ar/article/11974-ryys-mykrwswft-ltnfydhy-yhdhr-mn-n-stdf-ldhk-lstn-y-mhlyan-thmy-lm-rf-lkhs/korshunov.aiar2026-07-15T01:22:35.382234+00:00رئيس مايكروسوفت التنفيذي يحذر من أن استضافة الذكاء الاصطناعي محلياً تحمي المعرفة الخاصة https://korshunov.ai/en/article/11969-claude-code-v2-1-210-fixes-crashes-injection-risks-and-auto-mode-defaults/korshunov.aien2026-07-15T01:10:48.860342+00:00Claude Code v2.1.210 fixes crashes, injection risks, and auto-mode defaults https://korshunov.ai/ru/article/11969-claude-code-v2-1-210-ispravliaet-sboi-riski-inektsii-i-nastroiki-avto-rezhima-po/korshunov.airu2026-07-15T01:10:48.860342+00:00Claude Code v2.1.210 исправляет сбои, риски инъекций и настройки авто-режима по умолчанию https://korshunov.ai/es/article/11969-claude-code-v2-1-210-corrige-cierres-inesperados-riesgos-de-inyeccion-y-valores/korshunov.aies2026-07-15T01:10:48.860342+00:00Claude Code v2.1.210 corrige cierres inesperados, riesgos de inyección y valores predeterminados del modo automático https://korshunov.ai/zh-hans/article/11969-claude-code-v2-1-210-xiu-fu-liao-beng-kui-zhu-ru-feng-xian-ji-zi-dong-mo-shi-mo/korshunov.aizh-cn2026-07-15T01:10:48.860342+00:00Claude Code v2.1.210 修复了崩溃、注入风险及自动模式默认设置 https://korshunov.ai/pt-br/article/11969-claude-code-v2-1-210-corrige-travamentos-riscos-de-injecao-e-padroes-do-modo/korshunov.aipt2026-07-15T01:10:48.860342+00:00Claude Code v2.1.210 corrige travamentos, riscos de injeção e padrões do modo automático https://korshunov.ai/hi/article/11969-claude-code-v2-1-210-kraish-injekshn-jokhimon-aur-ontto-modd-ddiphonltts-ko-hai/korshunov.aihi2026-07-15T01:10:48.860342+00:00Claude Code v2.1.210 क्रैश, इंजेक्शन जोखिमों और ऑटो-मोड डिफ़ॉल्ट्स को ठीक करता है https://korshunov.ai/ja/article/11969-claude-code-v2-1-210-kakuratsushiyu-inshiekushiyonrisuku-zi-dong-mo-zheng/korshunov.aija2026-07-15T01:10:48.860342+00:00Claude Code v2.1.210 がクラッシュ、インジェクションリスク、自動モードのデフォルトを修正 https://korshunov.ai/ko/article/11969-claude-code-v2-1-210i-cungdol-injegsyeon-wiheom-mic-jadong-modeu-gibongabs/korshunov.aiko2026-07-15T01:10:48.860342+00:00Claude Code v2.1.210이 충돌, 인젝션 위험 및 자동 모드 기본값 수정 https://korshunov.ai/fr/article/11969-claude-code-v2-1-210-corrige-les-plantages-les-risques-d-injection-et-les-par-du/korshunov.aifr2026-07-15T01:10:48.860342+00:00Claude Code v2.1.210 corrige les plantages, les risques d'injection et les paramètres par défaut du mode automatique https://korshunov.ai/id/article/11969-claude-code-v2-1-210-memperbaiki-crash-risiko-injeksi-dan-pengaturan-default/korshunov.aiid2026-07-15T01:10:48.860342+00:00Claude Code v2.1.210 memperbaiki crash, risiko injeksi, dan pengaturan default mode otomatis https://korshunov.ai/ar/article/11969-yslh-claude-code-v2-1-210-l-tl-wmkhtr-lhqn-w-ddt-lwd-ltlqyy-lftrdy/korshunov.aiar2026-07-15T01:10:48.860342+00:00يصلح Claude Code v2.1.210 الأعطال، ومخاطر الحقن، وإعدادات الوضع التلقائي الافتراضية https://korshunov.ai/zh-hans/article/11734-regrind-cong-dan-ci-ren-lei-yan-shi-zhong-xue-xi-ling-qiao-cao-zuo/korshunov.aizh-cn2026-07-15T01:01:43.600858+00:00REGRIND 从单次人类演示中学习灵巧操作 https://korshunov.ai/en/article/11734-regrind-learns-dexterous-manipulation-from-a-single-human-demonstration/korshunov.aien2026-07-15T01:01:43.600858+00:00REGRIND learns dexterous manipulation from a single human demonstration https://korshunov.ai/ru/article/11734-regrind-uchitsia-lovkomu-manipulirovaniiu-po-odnoi-demonstratsii-cheloveka/korshunov.airu2026-07-15T01:01:43.600858+00:00REGRIND учится ловкому манипулированию по одной демонстрации человека https://korshunov.ai/es/article/11734-regrind-aprende-manipulacion-diestra-a-partir-de-una-unica-demostracion-humana/korshunov.aies2026-07-15T01:01:43.600858+00:00REGRIND aprende manipulación diestra a partir de una única demostración humana https://korshunov.ai/pt-br/article/11734-regrind-aprende-manipulacao-destreza-a-partir-de-uma-unica-demonstracao-humana/korshunov.aipt2026-07-15T01:01:43.600858+00:00REGRIND aprende manipulação destreza a partir de uma única demonstração humana https://korshunov.ai/hi/article/11734-regrind-ekl-maanv-prdrshn-se-kushl-hainddling-siikhtaa-hai/korshunov.aihi2026-07-15T01:01:43.600858+00:00REGRIND एकल मानव प्रदर्शन से कुशल हैंडलिंग सीखता है https://korshunov.ai/ja/article/11734-regrindha1hui-noren-jian-temokaraqi-yong-nacao-zuo-woxue-xi-suru/korshunov.aija2026-07-15T01:01:43.600858+00:00REGRINDは1回の人間デモから器用な操作を学習する https://korshunov.ai/ko/article/11734-regrindeun-danil-ingan-siyeoneseo-jeonggyohan-jojageul-hagseubhabnida/korshunov.aiko2026-07-15T01:01:43.600858+00:00REGRIND은 단일 인간 시연에서 정교한 조작을 학습합니다 https://korshunov.ai/fr/article/11734-regrind-apprend-la-manipulation-delicate-a-partir-d-une-seule-demonstration/korshunov.aifr2026-07-15T01:01:43.600858+00:00REGRIND apprend la manipulation délicate à partir d'une seule démonstration humaine https://korshunov.ai/id/article/11734-regrind-belajar-manipulasi-cekatan-dari-satu-demonstrasi-manusia/korshunov.aiid2026-07-15T01:01:43.600858+00:00REGRIND belajar manipulasi cekatan dari satu demonstrasi manusia https://korshunov.ai/ar/article/11734-yt-lm-regrind-ltl-b-ldqyq-mn-rd-bshry-whd/korshunov.aiar2026-07-15T01:01:43.600858+00:00يتعلم REGRIND التلاعب الدقيق من عرض بشري واحد https://korshunov.ai/en/article/11718-theoretical-framework-explains-transformer-inductive-reasoning-via-invariant/korshunov.aien2026-07-15T00:53:54.755172+00:00Theoretical framework explains Transformer inductive reasoning via invariant manifold https://korshunov.ai/ru/article/11718-teoreticheskaia-ramka-obiasniaet-induktivnoe-rassuzhdenie-transformera-cherez/korshunov.airu2026-07-15T00:53:54.755172+00:00Теоретическая рамка объясняет индуктивное рассуждение Трансформера через инвариантное многообразие https://korshunov.ai/es/article/11718-el-marco-teorico-explica-el-razonamiento-inductivo-del-transformer-a-traves-de/korshunov.aies2026-07-15T00:53:54.755172+00:00El marco teórico explica el razonamiento inductivo del Transformer a través de una variedad invariante https://korshunov.ai/zh-hans/article/11718-li-lun-kuang-jia-tong-guo-bu-bian-liu-xing-jie-shi-transformerde-gui-na-tui-li/korshunov.aizh-cn2026-07-15T00:53:54.755172+00:00理论框架通过不变流形解释Transformer的归纳推理 https://korshunov.ai/pt-br/article/11718-estrutura-teorica-explica-o-raciocinio-indutivo-do-transformer-via-variedade/korshunov.aipt2026-07-15T00:53:54.755172+00:00Estrutura teórica explica o raciocínio indutivo do Transformer via variedade invariante https://korshunov.ai/hi/article/11718-aprivrtniiy-mainipholdd-ke-maadhym-se-ttraansphonrmr-kii-sncyii-trkshkti-ko/korshunov.aihi2026-07-15T00:53:54.755172+00:00अपरिवर्तनीय मैनिफोल्ड के माध्यम से ट्रान्सफॉर्मर की संचयी तर्कशक्ति को समझाने वाला सैद्धांतिक ढांचा https://korshunov.ai/ja/article/11718-bu-bian-duo-yang-ti-niyorutoransuhuo-ma-nogui-na-de-tui-lun-noshuo-ming-li-lun/korshunov.aija2026-07-15T00:53:54.755172+00:00不変多様体によるトランスフォーマーの帰納的推論の説明理論的枠組み https://korshunov.ai/ko/article/11718-bulbyeon-dayangcereul-tonghan-teuraenseupomeoyi-gwinabjeog-curon-seolmyeong-teul/korshunov.aiko2026-07-15T00:53:54.755172+00:00불변 다양체를 통한 트랜스포머의 귀납적 추론 설명 이론적 틀 https://korshunov.ai/fr/article/11718-un-cadre-theorique-explique-le-raisonnement-inductif-des-transformers-via-une/korshunov.aifr2026-07-15T00:53:54.755172+00:00Un cadre théorique explique le raisonnement inductif des Transformers via une variété invariante https://korshunov.ai/id/article/11718-kerangka-teoretis-menjelaskan-penalaran-induktif-transformer-melalui-manifold/korshunov.aiid2026-07-15T00:53:54.755172+00:00Kerangka teoretis menjelaskan penalaran induktif Transformer melalui manifold invarian https://korshunov.ai/ar/article/11718-tr-nzry-yfsr-lstdll-lstqryy-fy-lmhwlt-br-mt-dd-lshkl-lthbt/korshunov.aiar2026-07-15T00:53:54.755172+00:00إطار نظري يفسر الاستدلال الاستقرائي في المحولات عبر متعدد الأشكال الثابت https://korshunov.ai/en/article/11733-theoretical-framework-explains-transformer-inductive-reasoning-via-invariant/korshunov.aien2026-07-15T00:47:08.748918+00:00Theoretical framework explains Transformer inductive reasoning via invariant manifold https://korshunov.ai/ru/article/11733-teoreticheskaia-osnova-obiasniaet-induktivnoe-rassuzhdenie-transformera-cherez/korshunov.airu2026-07-15T00:47:08.748918+00:00Теоретическая основа объясняет индуктивное рассуждение Трансформера через инвариантное многообразие https://korshunov.ai/es/article/11733-el-marco-teorico-explica-el-razonamiento-inductivo-del-transformer-a-traves-de/korshunov.aies2026-07-15T00:47:08.748918+00:00El marco teórico explica el razonamiento inductivo del Transformer a través de una variedad invariante https://korshunov.ai/zh-hans/article/11733-li-lun-kuang-jia-tong-guo-bu-bian-liu-xing-jie-shi-transformerde-gui-na-tui-li/korshunov.aizh-cn2026-07-15T00:47:08.748918+00:00理论框架通过不变流形解释Transformer的归纳推理 https://korshunov.ai/pt-br/article/11733-estrutura-teorica-explica-o-raciocinio-indutivo-do-transformer-via-variedade/korshunov.aipt2026-07-15T00:47:08.748918+00:00Estrutura teórica explica o raciocínio indutivo do Transformer via variedade invariante https://korshunov.ai/hi/article/11733-aprivrtniiy-mainipholdd-ke-maadhym-se-ttraansphonrmr-kii-snaatn-trkshkti-ko/korshunov.aihi2026-07-15T00:47:08.748918+00:00अपरिवर्तनीय मैनिफोल्ड के माध्यम से ट्रांसफॉर्मर की सनातन तर्कशक्ति को समझाने वाला सैद्धांतिक ढांचा https://korshunov.ai/ja/article/11733-bu-bian-duo-yang-ti-niyorutoransuhuo-ma-nogui-na-de-tui-lun-noli-lun-de-waku-zu/korshunov.aija2026-07-15T00:47:08.748918+00:00不変多様体によるトランスフォーマーの帰納的推論の理論的枠組み https://korshunov.ai/ko/article/11733-bulbyeon-dayangcereul-tonghan-teuraenseupomeoyi-gwinabjeog-curone-daehan/korshunov.aiko2026-07-15T00:47:08.748918+00:00불변 다양체를 통한 트랜스포머의 귀납적 추론에 대한 이론적 프레임워크 https://korshunov.ai/fr/article/11733-un-cadre-theorique-explique-le-raisonnement-inductif-des-transformers-via-une/korshunov.aifr2026-07-15T00:47:08.748918+00:00Un cadre théorique explique le raisonnement inductif des Transformers via une variété invariante https://korshunov.ai/id/article/11733-kerangka-teoretis-menjelaskan-penalaran-induktif-transformer-melalui-manifold/korshunov.aiid2026-07-15T00:47:08.748918+00:00Kerangka teoretis menjelaskan penalaran induktif Transformer melalui manifold invarian https://korshunov.ai/ar/article/11733-tr-nzry-yfsr-lstdll-lstqryy-fy-lmhwlt-br-mt-dd-lstwh-lthbt/korshunov.aiar2026-07-15T00:47:08.748918+00:00إطار نظري يفسر الاستدلال الاستقرائي في المحولات عبر متعدد السطوح الثابت https://korshunov.ai/zh-hans/article/11717-ye-lu-da-xue-nlptuan-dui-fa-bu-guan-yu-llmzhong-yuan-ren-zhi-de-quan-mian-zong/korshunov.aizh-cn2026-07-15T00:39:07.954339+00:00耶鲁大学NLP团队发布关于LLM中元认知的全面综述 https://korshunov.ai/en/article/11717-yale-nlp-presents-comprehensive-overview-of-metacognition-in-llms/korshunov.aien2026-07-15T00:39:07.954339+00:00Yale NLP presents comprehensive overview of metacognition in LLMs https://korshunov.ai/ja/article/11717-yale-nlpkallmniokerumetaren-zhi-nobao-gua-de-gai-guan-woti-gong/korshunov.aija2026-07-15T00:39:07.954339+00:00Yale NLPがLLMにおけるメタ認知の包括的概観を提供 https://korshunov.ai/ru/article/11717-yale-nlp-predstavliaet-vseobemliushchii-obzor-metakognitsii-v-llm/korshunov.airu2026-07-15T00:39:07.954339+00:00Yale NLP представляет всеобъемлющий обзор метакогниции в LLM https://korshunov.ai/es/article/11717-yale-nlp-presenta-una-vision-general-exhaustiva-de-la-metacognicion-en-llm/korshunov.aies2026-07-15T00:39:07.954339+00:00Yale NLP presenta una visión general exhaustiva de la metacognición en LLM https://korshunov.ai/pt-br/article/11717-yale-nlp-apresenta-visao-abrangente-sobre-metacognicao-em-llms/korshunov.aipt2026-07-15T00:39:07.954339+00:00Yale NLP apresenta visão abrangente sobre metacognição em LLMs https://korshunov.ai/hi/article/11717-yale-nlp-ne-llms-men-mettaakognishn-kaa-vyaapk-avlokn-prstut-kiyaa/korshunov.aihi2026-07-15T00:39:07.954339+00:00Yale NLP ने LLMs में मेटाकोग्निशन का व्यापक अवलोकन प्रस्तुत किया https://korshunov.ai/ko/article/11717-yale-nlp-llmyi-metainjie-daehan-pogwaljeog-gaeyo-jesi/korshunov.aiko2026-07-15T00:39:07.954339+00:00Yale NLP, LLM의 메타인지에 대한 포괄적 개요 제시 https://korshunov.ai/fr/article/11717-yale-nlp-presente-un-apercu-complet-de-la-metacognition-dans-les-llm/korshunov.aifr2026-07-15T00:39:07.954339+00:00Yale NLP présente un aperçu complet de la métacognition dans les LLM https://korshunov.ai/id/article/11717-yale-nlp-menyajikan-gambaran-komprehensif-tentang-metakognisi-dalam-llm/korshunov.aiid2026-07-15T00:39:07.954339+00:00Yale NLP menyajikan gambaran komprehensif tentang metakognisi dalam LLM https://korshunov.ai/ar/article/11717-yale-nlp-tqdm-nzr-shml-hwl-ldrk-ldhty-fy-lnmdhj-llgwy-lkbyr/korshunov.aiar2026-07-15T00:39:07.954339+00:00Yale NLP تقدم نظرة شاملة حول الإدراك الذاتي في النماذج اللغوية الكبيرة https://korshunov.ai/en/article/11727-requential-coding-compresses-training-trajectories-using-self-generated-teacher/korshunov.aien2026-07-15T00:32:15.774031+00:00Requential coding compresses training trajectories using self-generated teacher-student disagreement https://korshunov.ai/ru/article/11727-rekventsialnoe-kodirovanie-szhimaet-traektorii-obucheniia-ispolzuia-uchitelia-i/korshunov.airu2026-07-15T00:32:15.774031+00:00Реквенциальное кодирование сжимает траектории обучения, используя расхождения учителя и ученика, генерируемые самостоятельно https://korshunov.ai/es/article/11727-el-codificado-requencial-comprime-trayectorias-de-entrenamiento-usando-maestro/korshunov.aies2026-07-15T00:32:15.774031+00:00El codificado requencial comprime trayectorias de entrenamiento usando desacuerdos maestro-estudiante autogenerados https://korshunov.ai/zh-hans/article/11727-xu-lie-bian-ma-li-yong-zi-sheng-cheng-de-shi-sheng-fen-qi-ya-suo-xun-lian-gui-ji/korshunov.aizh-cn2026-07-15T00:32:15.774031+00:00序列编码利用自生成的师生分歧压缩训练轨迹 https://korshunov.ai/pt-br/article/11727-codificacao-requencial-comprime-trajetorias-de-treinamento-usando-desacordo-auto/korshunov.aipt2026-07-15T00:32:15.774031+00:00Codificação requencial comprime trajetórias de treinamento usando desacordo auto-gerado entre professor e aluno https://korshunov.ai/hi/article/11727-rikvenshiyl-kodding-sv-jnit-ttiicr-sttuuddentt-ashmti-kaa-upyog-krke-prshikssnn/korshunov.aihi2026-07-15T00:32:15.774031+00:00रिक्वेन्शियल कोडिंग स्व-जनित टीचर-स्टूडेंट असहमति का उपयोग करके प्रशिक्षण ट्रैजेक्टरी को संपीड़ित करता है https://korshunov.ai/ja/article/11727-requential-codinghajiao-shi-sheng-tu-nobu-yi-zhi-woyong-itexue-xi-gui-dao-woya/korshunov.aija2026-07-15T00:32:15.774031+00:00Requential codingは教師-生徒の不一致を用いて学習軌道を圧縮する https://korshunov.ai/ko/article/11727-requential-codingeun-jaga-saengseongdoen-gyosa-hagsaeng-bulilcireul-sayonghayeo/korshunov.aiko2026-07-15T00:32:15.774031+00:00Requential coding은 자가 생성된 교사-학생 불일치를 사용하여 학습 궤적을 압축합니다 https://korshunov.ai/fr/article/11727-le-codage-requential-compresse-les-trajectoires-d-entrainement-en-utilisant-le/korshunov.aifr2026-07-15T00:32:15.774031+00:00Le codage requential compresse les trajectoires d'entraînement en utilisant le désaccord auto-généré entre enseignant et élève https://korshunov.ai/id/article/11727-koding-requential-memadatkan-lintasan-pelatihan-menggunakan-ketidakcocokan-guru/korshunov.aiid2026-07-15T00:32:15.774031+00:00Koding requential memadatkan lintasan pelatihan menggunakan ketidakcocokan guru-siswa yang dihasilkan sendiri https://korshunov.ai/ar/article/11727-ydgt-ltrmyz-lmtslsl-lmsrt-ltdryby-bstkhdm-dm-ltfq-byn-lm-lm-wltlb-lmuwlwd-dhtyan/korshunov.aiar2026-07-15T00:32:15.774031+00:00يضغط الترميز المتسلسل المسارات التدريبية باستخدام عدم الاتفاق بين المعلم والطالب المُولّد ذاتياً https://korshunov.ai/en/article/11923-nvidia-nemo-enables-autoresearch-workflows-with-rl-agent-skills/korshunov.aien2026-07-15T00:24:22.045339+00:00NVIDIA NeMo enables autoresearch workflows with RL agent skills https://korshunov.ai/ru/article/11923-nvidia-nemo-podderzhivaet-avtonomnye-issledovatelskie-protsessy-s-navykami-rl/korshunov.airu2026-07-15T00:24:22.045339+00:00NVIDIA NeMo поддерживает автономные исследовательские процессы с навыками RL-агентов https://korshunov.ai/es/article/11923-nvidia-nemo-habilita-flujos-de-trabajo-de-investigacion-autonoma-con-habilidades/korshunov.aies2026-07-15T00:24:22.045339+00:00NVIDIA NeMo habilita flujos de trabajo de investigación autónoma con habilidades de agentes RL https://korshunov.ai/zh-hans/article/11923-nvidia-nemo-jie-zhu-qiang-hua-xue-xi-zhi-neng-ti-ji-neng-shi-xian-zi-zhu-yan-jiu/korshunov.aizh-cn2026-07-15T00:24:22.045339+00:00NVIDIA NeMo 借助强化学习智能体技能实现自主研究工作流 https://korshunov.ai/pt-br/article/11923-nvidia-nemo-habilita-fluxos-de-trabalho-de-pesquisa-autonoma-com-habilidades-de/korshunov.aipt2026-07-15T00:24:22.045339+00:00NVIDIA NeMo habilita fluxos de trabalho de pesquisa autônoma com habilidades de agentes RL https://korshunov.ai/hi/article/11923-nvidia-nemo-rl-ejentt-kaushlon-ke-saath-svth-anusndhaan-kaaryprvaah-skssm-hai/korshunov.aihi2026-07-15T00:24:22.045339+00:00NVIDIA NeMo RL एजेंट कौशलों के साथ स्वतः अनुसंधान कार्यप्रवाह सक्षम बनाता है https://korshunov.ai/ja/article/11923-nvidia-nemokarle-shientonosukirutezi-lu-yan-jiu-wa-kuhuro-woke-neng-ni/korshunov.aija2026-07-15T00:24:22.045339+00:00NVIDIA NeMoがRLエージェントのスキルで自律研究ワークフローを可能に https://korshunov.ai/ko/article/11923-nvidia-nemo-rl-eijeonteu-seukilro-jadong-yeongu-weokeupeulrou-ganeunghage-ham/korshunov.aiko2026-07-15T00:24:22.045339+00:00NVIDIA NeMo, RL 에이전트 스킬로 자동 연구 워크플로우 가능하게 함 https://korshunov.ai/fr/article/11923-nvidia-nemo-permet-des-flux-de-travail-de-recherche-automatique-avec-des-d-agent/korshunov.aifr2026-07-15T00:24:22.045339+00:00NVIDIA NeMo permet des flux de travail de recherche automatique avec des compétences d'agent RL https://korshunov.ai/id/article/11923-nvidia-nemo-memungkinkan-alur-kerja-riset-otomatis-dengan-keterampilan-agen-rl/korshunov.aiid2026-07-15T00:24:22.045339+00:00NVIDIA NeMo memungkinkan alur kerja riset otomatis dengan keterampilan agen RL https://korshunov.ai/ar/article/11923-nvidia-nemo-ymkwn-syr-ml-lbhth-ltlqyy-bmhrt-wkyl-rl/korshunov.aiar2026-07-15T00:24:22.045339+00:00NVIDIA NeMo يمكّن سير عمل البحث التلقائي بمهارات وكيل RL https://korshunov.ai/zh-hans/article/11966-guan-yu-1wei-mo-xing-ke-xing-xing-he-wei-lai-de-reddittao-lun/korshunov.aizh-cn2026-07-15T00:18:47.121651+00:00关于1位模型可行性和未来的Reddit讨论 https://korshunov.ai/en/article/11966-reddit-discussion-on-the-viability-and-future-of-1-bit-models/korshunov.aien2026-07-15T00:18:47.121651+00:00Reddit discussion on the viability and future of 1-bit models https://korshunov.ai/ru/article/11966-obsuzhdenie-na-reddit-o-zhiznesposobnosti-i-budushchem-modelei-s-1-bitnym/korshunov.airu2026-07-15T00:18:47.121651+00:00Обсуждение на Reddit о жизнеспособности и будущем моделей с 1-битным квантованием https://korshunov.ai/es/article/11966-discusion-en-reddit-sobre-la-viabilidad-y-el-futuro-de-los-modelos-de-1-bit/korshunov.aies2026-07-15T00:18:47.121651+00:00Discusión en Reddit sobre la viabilidad y el futuro de los modelos de 1 bit https://korshunov.ai/pt-br/article/11966-discussao-no-reddit-sobre-a-viabilidade-e-o-futuro-de-modelos-de-1-bit/korshunov.aipt2026-07-15T00:18:47.121651+00:00Discussão no Reddit sobre a viabilidade e o futuro de modelos de 1 bit https://korshunov.ai/hi/article/11966-1-bitt-monddl-kii-vyvhaarytaa-aur-bhvissy-pr-reddit-crcaa/korshunov.aihi2026-07-15T00:18:47.121651+00:001-बिट मॉडल की व्यवहार्यता और भविष्य पर Reddit चर्चा https://korshunov.ai/ja/article/11966-1hitsutomoterunotuo-dang-xing-tojiang-lai-xing-niguan-surureddittenoyi-lun/korshunov.aija2026-07-15T00:18:47.121651+00:001ビットモデルの妥当性と将来性に関するRedditでの議論 https://korshunov.ai/ko/article/11966-1biteu-modelyi-silhyeon-ganeungseonggwa-miraee-daehan-redis-toron/korshunov.aiko2026-07-15T00:18:47.121651+00:001비트 모델의 실현 가능성과 미래에 대한 레딧 토론 https://korshunov.ai/fr/article/11966-discussion-reddit-sur-la-viabilite-et-l-avenir-des-modeles-1-bit/korshunov.aifr2026-07-15T00:18:47.121651+00:00Discussion Reddit sur la viabilité et l'avenir des modèles 1-bit https://korshunov.ai/id/article/11966-diskusi-reddit-tentang-kelayakan-dan-masa-depan-model-1-bit/korshunov.aiid2026-07-15T00:18:47.121651+00:00Diskusi Reddit tentang kelayakan dan masa depan model 1-bit https://korshunov.ai/ar/article/11966-nqsh-l-rdyt-hwl-jdw-wmstqbl-lnmdhj-dht-lbt-lwhd/korshunov.aiar2026-07-15T00:18:47.121651+00:00نقاش على رديت حول جدوى ومستقبل النماذج ذات البت الواحد https://korshunov.ai/en/article/11967-github-makes-dependabot-wait-three-days-before-opening-version-update-prs/korshunov.aien2026-07-15T00:15:20.908689+00:00GitHub makes Dependabot wait three days before opening version update PRs https://korshunov.ai/ru/article/11967-github-zastavit-dependabot-zhdat-tri-dnia-pered-sozdaniem-pr-na-obnovlenie/korshunov.airu2026-07-15T00:15:20.908689+00:00GitHub заставит Dependabot ждать три дня перед созданием PR на обновление версии https://korshunov.ai/es/article/11967-github-hace-que-dependabot-espere-tres-dias-antes-de-abrir-prs-de-actualizacion/korshunov.aies2026-07-15T00:15:20.908689+00:00GitHub hace que Dependabot espere tres días antes de abrir PRs de actualización de versión https://korshunov.ai/zh-hans/article/11967-github-rang-dependabot-zai-da-kai-ban-ben-geng-xin-pr-qian-deng-dai-san-tian/korshunov.aizh-cn2026-07-15T00:15:20.908689+00:00GitHub 让 Dependabot 在打开版本更新 PR 前等待三天 https://korshunov.ai/pt-br/article/11967-github-faz-dependabot-esperar-tres-dias-antes-de-abrir-prs-de-atualizacao-de/korshunov.aipt2026-07-15T00:15:20.908689+00:00GitHub faz Dependabot esperar três dias antes de abrir PRs de atualização de versão https://korshunov.ai/hi/article/11967-github-dependabot-ko-snskrnn-apddett-prs-kholne-se-phle-tiin-din-prtiikssaa-krne/korshunov.aihi2026-07-15T00:15:20.908689+00:00GitHub Dependabot को संस्करण अपडेट PRs खोलने से पहले तीन दिन प्रतीक्षा करने देता है https://korshunov.ai/ja/article/11967-github-dependabotkaha-shiyongeng-xin-noprwokai-kuqian-ni3ri-jian-nodai-ji-shi-ru/korshunov.aija2026-07-15T00:15:20.908689+00:00GitHub、Dependabotがバージョン更新のPRを開く前に3日間の待機時間を導入 https://korshunov.ai/ko/article/11967-github-dependaboti-beojeon-eobdeiteu-preul-yeolgi-jeone-3il-daegi-sigan-doib/korshunov.aiko2026-07-15T00:15:20.908689+00:00GitHub, Dependabot이 버전 업데이트 PR을 열기 전에 3일 대기 시간 도입 https://korshunov.ai/fr/article/11967-github-impose-un-delai-d-attente-de-trois-jours-a-dependabot-avant-d-ouvrir-les/korshunov.aifr2026-07-15T00:15:20.908689+00:00GitHub impose un délai d'attente de trois jours à Dependabot avant d'ouvrir les PR de mise à jour de version https://korshunov.ai/id/article/11967-github-membuat-dependabot-menunggu-tiga-hari-sebelum-membuka-pr-pembaruan-versi/korshunov.aiid2026-07-15T00:15:20.908689+00:00GitHub membuat Dependabot menunggu tiga hari sebelum membuka PR pembaruan versi https://korshunov.ai/ar/article/11967-github-tj-l-dependabot-tntzr-thlth-ym-qbl-fth-tlbt-lshb-lthdyth-lsdr/korshunov.aiar2026-07-15T00:15:20.908689+00:00GitHub تجعل Dependabot تنتظر ثلاثة أيام قبل فتح طلبات السحب لتحديث الإصدار https://korshunov.ai/en/article/11968-prismml-bonsai-27b-runs-on-jetson-orin-nano-8gb-with-4-31-t-s-and-6-2gb-ram/korshunov.aien2026-07-15T00:10:11.825921+00:00PrismML Bonsai 27B runs on Jetson Orin Nano 8GB with 4.31 t/s and 6.2GB RAM https://korshunov.ai/ru/article/11968-prismml-bonsai-27b-rabotaet-na-jetson-orin-nano-8gb-so-skorostiu-4-31-t-s-i-6-2/korshunov.airu2026-07-15T00:10:11.825921+00:00PrismML Bonsai 27B работает на Jetson Orin Nano 8GB со скоростью 4.31 t/s и потреблением 6.2 ГБ ОЗУ https://korshunov.ai/es/article/11968-prismml-bonsai-27b-se-ejecuta-en-jetson-orin-nano-8gb-con-4-31-t-s-y-6-2-gb-de/korshunov.aies2026-07-15T00:10:11.825921+00:00PrismML Bonsai 27B se ejecuta en Jetson Orin Nano 8GB con 4.31 t/s y 6.2 GB de RAM https://korshunov.ai/zh-hans/article/11968-prismml-bonsai-27b-zai-jetson-orin-nano-8gb-shang-yun-xing-su-du-wei-4-31-t-s-6/korshunov.aizh-cn2026-07-15T00:10:11.825921+00:00PrismML Bonsai 27B 在 Jetson Orin Nano 8GB 上运行，速度为 4.31 t/s，占用 6.2GB RAM https://korshunov.ai/pt-br/article/11968-prismml-bonsai-27b-roda-no-jetson-orin-nano-8gb-com-4-31-t-s-e-6-2-gb-de-ram/korshunov.aipt2026-07-15T00:10:11.825921+00:00PrismML Bonsai 27B roda no Jetson Orin Nano 8GB com 4.31 t/s e 6.2 GB de RAM https://korshunov.ai/ja/article/11968-prismml-bonsai-27bkajetson-orin-nano-8gbshang-te4-31-t-s-6-2gb-ramtedong-zuo/korshunov.aija2026-07-15T00:10:11.825921+00:00PrismML Bonsai 27BがJetson Orin Nano 8GB上で4.31 t/s、6.2GB RAMで動作 https://korshunov.ai/hi/article/11968-prismml-bonsai-27b-jetson-orin-nano-8gb-pr-4-31-t-s-aur-6-2gb-ram-ke-saath-cltaa/korshunov.aihi2026-07-15T00:10:11.825921+00:00PrismML Bonsai 27B Jetson Orin Nano 8GB पर 4.31 t/s और 6.2GB RAM के साथ चलता है https://korshunov.ai/ko/article/11968-prismml-bonsai-27b-jetson-orin-nano-8gbeseo-4-31-t-s-mic-6-2gb-rameuro-silhaeng/korshunov.aiko2026-07-15T00:10:11.825921+00:00PrismML Bonsai 27B, Jetson Orin Nano 8GB에서 4.31 t/s 및 6.2GB RAM으로 실행 https://korshunov.ai/fr/article/11968-prismml-bonsai-27b-fonctionne-sur-jetson-orin-nano-8go-avec-431-t-s-et-62-go-de/korshunov.aifr2026-07-15T00:10:11.825921+00:00PrismML Bonsai 27B fonctionne sur Jetson Orin Nano 8Go avec 4,31 t/s et 6,2 Go de RAM https://korshunov.ai/id/article/11968-prismml-bonsai-27b-berjalan-di-jetson-orin-nano-8gb-dengan-431-t-s-dan-ram-62gb/korshunov.aiid2026-07-15T00:10:11.825921+00:00PrismML Bonsai 27B berjalan di Jetson Orin Nano 8GB dengan 4,31 t/s dan RAM 6,2GB https://korshunov.ai/ar/article/11968-nmwdhj-prismml-bonsai-27b-y-ml-l-jetson-orin-nano-8gb-bsr-4-31-twk-th-wdhkr-wswl/korshunov.aiar2026-07-15T00:10:11.825921+00:00نموذج PrismML Bonsai 27B يعمل على Jetson Orin Nano 8GB بسرعة 4.31 توك/ث وذاكرة وصول عشوائي 6.2 جيجابايت https://korshunov.ai/en/article/11937-open-interpreter-0-0-22-release/korshunov.aien2026-07-15T00:05:18.170982+00:00Open Interpreter 0.0.22 release https://korshunov.ai/ru/article/11937-reliz-open-interpreter-0-0-22/korshunov.airu2026-07-15T00:05:18.170982+00:00Релиз Open Interpreter 0.0.22 https://korshunov.ai/es/article/11937-lanzamiento-de-open-interpreter-0-0-22/korshunov.aies2026-07-15T00:05:18.170982+00:00Lanzamiento de Open Interpreter 0.0.22 https://korshunov.ai/zh-hans/article/11937-open-interpreter-0-0-22-fa-bu/korshunov.aizh-cn2026-07-15T00:05:18.170982+00:00Open Interpreter 0.0.22 发布 https://korshunov.ai/pt-br/article/11937-lancamento-do-open-interpreter-0-0-22/korshunov.aipt2026-07-15T00:05:18.170982+00:00Lançamento do Open Interpreter 0.0.22 https://korshunov.ai/hi/article/11937-open-interpreter-0-0-22-riliij/korshunov.aihi2026-07-15T00:05:18.170982+00:00Open Interpreter 0.0.22 रिलीज़ https://korshunov.ai/ja/article/11937-open-interpreter-0-0-22-noriri-su/korshunov.aija2026-07-15T00:05:18.170982+00:00Open Interpreter 0.0.22 のリリース https://korshunov.ai/ko/article/11937-open-interpreter-0-0-22-rilriseu/korshunov.aiko2026-07-15T00:05:18.170982+00:00Open Interpreter 0.0.22 릴리스 https://korshunov.ai/fr/article/11937-sortie-de-open-interpreter-0-0-22/korshunov.aifr2026-07-15T00:05:18.170982+00:00Sortie de Open Interpreter 0.0.22 https://korshunov.ai/id/article/11937-rilis-open-interpreter-0-0-22/korshunov.aiid2026-07-15T00:05:18.170982+00:00Rilis Open Interpreter 0.0.22 https://korshunov.ai/ar/article/11937-sdr-open-interpreter-0-0-22/korshunov.aiar2026-07-15T00:05:18.170982+00:00إصدار Open Interpreter 0.0.22 https://korshunov.ai/en/article/11924-nvidia-cosmos-3-post-training-with-agent-skills-achieves-over-90-accuracy/korshunov.aien2026-07-15T00:02:58.544357+00:00NVIDIA Cosmos 3 post-training with agent skills achieves over 90% accuracy https://korshunov.ai/ru/article/11924-postobuchenie-nvidia-cosmos-3-s-navykami-agentov-dostigaet-tochnosti-bolee-90/korshunov.airu2026-07-15T00:02:58.544357+00:00Постобучение NVIDIA Cosmos 3 с навыками агентов достигает точности более 90% https://korshunov.ai/es/article/11924-el-post-entrenamiento-de-nvidia-cosmos-3-con-habilidades-de-agentes-supera-el-90/korshunov.aies2026-07-15T00:02:58.544357+00:00El post-entrenamiento de NVIDIA Cosmos 3 con habilidades de agentes supera el 90% de precisión https://korshunov.ai/zh-hans/article/11924-nvidia-cosmos-3-shi-yong-dai-li-ji-neng-jin-xing-hou-xun-lian-zhun-que-lu-chao/korshunov.aizh-cn2026-07-15T00:02:58.544357+00:00NVIDIA Cosmos 3 使用代理技能进行后训练，准确率超过 90% https://korshunov.ai/pt-br/article/11924-pos-treinamento-do-nvidia-cosmos-3-com-habilidades-de-agentes-atinge-mais-de-90/korshunov.aipt2026-07-15T00:02:58.544357+00:00Pós-treinamento do NVIDIA Cosmos 3 com habilidades de agentes atinge mais de 90% de precisão https://korshunov.ai/hi/article/11924-ejentt-kaushl-ke-saath-nvidia-cosmos-3-kaa-postt-ttrening-90-se-adhik-sttiiktaa/korshunov.aihi2026-07-15T00:02:58.544357+00:00एजेंट कौशल के साथ NVIDIA Cosmos 3 का पोस्ट-ट्रेनिंग 90% से अधिक सटीकता प्राप्त करता है https://korshunov.ai/ja/article/11924-e-shientosukiruniyorucosmos-3nohosutotore-ninkute90-chao-nojing-du-woda-cheng/korshunov.aija2026-07-15T00:02:58.544357+00:00エージェントスキルによるCosmos 3のポストトレーニングで90%超の精度を達成 https://korshunov.ai/ko/article/11924-eijeonteu-seukileul-hwalyonghan-cosmos-3-sahu-hagseub-90-isang-jeonghwagdo/korshunov.aiko2026-07-15T00:02:58.544357+00:00에이전트 스킬을 활용한 Cosmos 3 사후 학습, 90% 이상 정확도 달성 https://korshunov.ai/fr/article/11924-cosmos-3-post-entrainement-avec-des-competences-d-agent-atteint-plus-de-90-de/korshunov.aifr2026-07-15T00:02:58.544357+00:00Cosmos 3 post-entraînement avec des compétences d'agent atteint plus de 90 % de précision https://korshunov.ai/id/article/11924-pelatihan-pasca-cosmos-3-dengan-keterampilan-agen-mencapai-akurasi-lebih-dari-90/korshunov.aiid2026-07-15T00:02:58.544357+00:00Pelatihan pasca Cosmos 3 dengan keterampilan agen mencapai akurasi lebih dari 90% https://korshunov.ai/ar/article/11924-tdryb-cosmos-3-b-d-ltdryb-bmhrt-lwkyl-yhqq-dq-tzyd-n-90/korshunov.aiar2026-07-15T00:02:58.544357+00:00تدريب Cosmos 3 بعد التدريب بمهارات الوكيل يحقق دقة تزيد عن 90% https://korshunov.ai/en/article/11934-nvidia-analyzes-5000-kaggle-entries-to-identify-techniques-for-improving-ai/korshunov.aien2026-07-14T23:56:57.951132+00:00NVIDIA analyzes 5,000+ Kaggle entries to identify techniques for improving AI reasoning https://korshunov.ai/ru/article/11934-nvidia-analiziruet-bolee-5000-rabot-s-kaggle-dlia-vyiavleniia-metodov-ii/korshunov.airu2026-07-14T23:56:57.951132+00:00NVIDIA анализирует более 5000 работ с Kaggle для выявления методов улучшения рассуждений ИИ https://korshunov.ai/es/article/11934-nvidia-analiza-mas-de-5-000-entradas-de-kaggle-para-identificar-tecnicas-que-el/korshunov.aies2026-07-14T23:56:57.951132+00:00NVIDIA analiza más de 5.000 entradas de Kaggle para identificar técnicas que mejoren el razonamiento de la IA https://korshunov.ai/zh-hans/article/11934-nvidia-fen-xi-5000-duo-ge-kaggle-can-sai-zuo-pin-yi-shi-bie-ti-sheng-ai-tui-li/korshunov.aizh-cn2026-07-14T23:56:57.951132+00:00NVIDIA 分析 5000 多个 Kaggle 参赛作品，以识别提升 AI 推理能力的技术 https://korshunov.ai/pt-br/article/11934-nvidia-analisa-mais-de-5-000-entradas-do-kaggle-para-identificar-tecnicas-que-o/korshunov.aipt2026-07-14T23:56:57.951132+00:00NVIDIA analisa mais de 5.000 entradas do Kaggle para identificar técnicas que melhorem o raciocínio da IA https://korshunov.ai/hi/article/11934-ai-trk-men-sudhaar-ke-lie-tkniikon-kii-phcaan-krne-ke-lie-nvidia-ne-5000-se-kaa/korshunov.aihi2026-07-14T23:56:57.951132+00:00AI तर्क में सुधार के लिए तकनीकों की पहचान करने के लिए NVIDIA ने 5,000 से अधिक Kaggle प्रविष्टियों का विश्लेषण किया https://korshunov.ai/ja/article/11934-nvidiakakaggleno5000jian-yi-shang-noentori-wofen-xi-shi-aitui-lun-nogai-shan-fa/korshunov.aija2026-07-14T23:56:57.951132+00:00NVIDIAがKaggleの5,000件以上のエントリーを分析し、AI推論の改善手法を特定 https://korshunov.ai/ko/article/11934-nvidia-kaggleyi-5000gae-isang-enteurireul-bunseoghayeo-ai-curon-gaeseon-gisul/korshunov.aiko2026-07-14T23:56:57.951132+00:00NVIDIA, Kaggle의 5,000개 이상 엔트리를 분석하여 AI 추론 개선 기술 식별 https://korshunov.ai/fr/article/11934-nvidia-analyse-plus-de-5-000-entrees-kaggle-pour-identifier-des-techniques-le-de/korshunov.aifr2026-07-14T23:56:57.951132+00:00NVIDIA analyse plus de 5 000 entrées Kaggle pour identifier des techniques améliorant le raisonnement de l'IA https://korshunov.ai/id/article/11934-nvidia-menganalisis-lebih-dari-5-000-entri-kaggle-untuk-mengidentifikasi-teknik/korshunov.aiid2026-07-14T23:56:57.951132+00:00NVIDIA menganalisis lebih dari 5.000 entri Kaggle untuk mengidentifikasi teknik meningkatkan penalaran AI https://korshunov.ai/ar/article/11934-thlyl-nvidia-lkthr-mn-5000-dkhl-fy-kaggle-lthdyd-tqnyt-thsyn-lstdll-ldhky/korshunov.aiar2026-07-14T23:56:57.951132+00:00تحليل NVIDIA لأكثر من 5,000 إدخال في Kaggle لتحديد تقنيات تحسين الاستدلال الذكي https://korshunov.ai/en/article/11866-unified-gradient-projection-reduces-catastrophic-forgetting-in-multilingual-asr/korshunov.aien2026-07-14T23:51:04.502634+00:00Unified Gradient Projection reduces catastrophic forgetting in multilingual ASR https://korshunov.ai/ru/article/11866-edinaia-proektsiia-gradienta-snizhaet-katastroficheskoe-zabyvanie-v-asr/korshunov.airu2026-07-14T23:51:04.502634+00:00Единая проекция градиента снижает катастрофическое забывание в многоязычном ASR https://korshunov.ai/es/article/11866-la-proyeccion-de-gradiente-unificada-reduce-el-olvido-catastrofico-en-asr/korshunov.aies2026-07-14T23:51:04.502634+00:00La proyección de gradiente unificada reduce el olvido catastrófico en ASR multilingüe https://korshunov.ai/zh-hans/article/11866-tong-yi-ti-du-tou-ying-jiang-di-duo-yu-yan-asrzhong-de-zai-nan-xing-yi-wang/korshunov.aizh-cn2026-07-14T23:51:04.502634+00:00统一梯度投影降低多语言ASR中的灾难性遗忘 https://korshunov.ai/pt-br/article/11866-a-projecao-de-gradiente-unificada-reduz-o-esquecimento-catastrofico-em-asr/korshunov.aipt2026-07-14T23:51:04.502634+00:00A Projeção de Gradiente Unificada reduz o esquecimento catastrófico em ASR multilíngue https://korshunov.ai/hi/article/11866-ekiikrt-greddientt-prkssepnn-bhubhaassii-asr-men-vinaashk-bhuul-ko-km-krtaa-hai/korshunov.aihi2026-07-14T23:51:04.502634+00:00एकीकृत ग्रेडिएंट प्रक्षेपण बहुभाषी ASR में विनाशक भूल को कम करता है https://korshunov.ai/ja/article/11866-tong-yi-gou-pei-tou-ying-haduo-yan-yu-asrniokeruhuai-mie-de-wang-que-wodi-jian/korshunov.aija2026-07-14T23:51:04.502634+00:00統一勾配投影は多言語ASRにおける壊滅的忘却を低減する https://korshunov.ai/ko/article/11866-unified-gradient-projectioneun-dagugeo-asreseo-pagugjeog-manggageul-julibnida/korshunov.aiko2026-07-14T23:51:04.502634+00:00Unified Gradient Projection은 다국어 ASR에서 파국적 망각을 줄입니다 https://korshunov.ai/fr/article/11866-la-projection-de-gradient-unifiee-reduit-l-oubli-catastrophique-dans-la-de-la/korshunov.aifr2026-07-14T23:51:04.502634+00:00La projection de gradient unifiée réduit l'oubli catastrophique dans la reconnaissance automatique de la parole multilingue https://korshunov.ai/id/article/11866-proyeksi-gradien-terpadu-mengurangi-lupa-katastrofik-pada-asr-multibahasa/korshunov.aiid2026-07-14T23:51:04.502634+00:00Proyeksi Gradien Terpadu mengurangi lupa katastrofik pada ASR multibahasa https://korshunov.ai/ar/article/11866-lbrwz-lmtdrj-lmwhd-yqll-mn-lnsyn-lkrthy-fy-lt-rf-l-lklm-mt-dd-llgt/korshunov.aiar2026-07-14T23:51:04.502634+00:00البروز المتدرج الموحد يقلل من النسيان الكارثي في التعرف على الكلام متعدد اللغات https://korshunov.ai/en/article/11865-researchers-introduce-qfes-task-and-qfesum-dataset-for-query-focused-event/korshunov.aien2026-07-14T23:43:13.840210+00:00Researchers introduce QFES task and QFESum dataset for query-focused event summarization https://korshunov.ai/ru/article/11865-issledovateli-predstavliaiut-zadachu-qfes-i-dataset-qfesum-dlia-tematicheskogo/korshunov.airu2026-07-14T23:43:13.840210+00:00Исследователи представляют задачу QFES и датасет QFESum для тематического обобщения событий https://korshunov.ai/es/article/11865-investigadores-presentan-la-tarea-qfes-y-el-conjunto-de-datos-qfesum-para-la-de/korshunov.aies2026-07-14T23:43:13.840210+00:00Investigadores presentan la tarea QFES y el conjunto de datos QFESum para la resumición de eventos centrada en consultas https://korshunov.ai/zh-hans/article/11865-yan-jiu-ren-yuan-ti-chu-cha-xun-ju-jiao-shi-jian-zhai-yao-ren-wu-qfesji-shu-ju/korshunov.aizh-cn2026-07-14T23:43:13.840210+00:00研究人员提出查询聚焦事件摘要任务QFES及数据集QFESum https://korshunov.ai/pt-br/article/11865-pesquisadores-introduzem-a-tarefa-qfes-e-o-conjunto-de-dados-qfesum-para-de-em/korshunov.aipt2026-07-14T23:43:13.840210+00:00Pesquisadores introduzem a tarefa QFES e o conjunto de dados QFESum para sumarização de eventos focada em consultas https://korshunov.ai/hi/article/11865-shodhkrtaaon-ne-kverii-phoksdd-iventt-smrii-ke-lie-qfes-ttaask-aur-qfesum-pesh/korshunov.aihi2026-07-14T23:43:13.840210+00:00शोधकर्ताओं ने क्वेरी-फोकस्ड इवेंट समरी के लिए QFES टास्क और QFESum डेटासेट पेश किया https://korshunov.ai/ja/article/11865-kuerijiao-dian-xing-ihentoyao-yue-notamenoqfestasukutoqfesumte-tasetsutowoyan/korshunov.aija2026-07-14T23:43:13.840210+00:00クエリ焦点型イベント要約のためのQFESタスクとQFESumデータセットを研究者が紹介 https://korshunov.ai/ko/article/11865-yeongujadeuli-kweori-jungsim-ibenteu-yoyagyong-qfes-jageob-mic-qfesum-deiteoses/korshunov.aiko2026-07-14T23:43:13.840210+00:00연구자들이 쿼리 중심 이벤트 요약용 QFES 작업 및 QFESum 데이터셋 소개 https://korshunov.ai/fr/article/11865-les-chercheurs-presentent-la-tache-qfes-et-le-jeu-de-donnees-qfesum-pour-la-d-la/korshunov.aifr2026-07-14T23:43:13.840210+00:00Les chercheurs présentent la tâche QFES et le jeu de données QFESum pour la synthèse d'événements centrée sur la requête https://korshunov.ai/id/article/11865-peneliti-memperkenalkan-tugas-qfes-dan-dataset-qfesum-untuk-ringkasan-peristiwa/korshunov.aiid2026-07-14T23:43:13.840210+00:00Peneliti memperkenalkan tugas QFES dan dataset QFESum untuk ringkasan peristiwa berfokus kueri https://korshunov.ai/ar/article/11865-lbhthwn-yqdmwn-mhm-qfes-wmjmw-bynt-qfesum-ltlkhys-lhdth-lmrkz-l-lst-lm/korshunov.aiar2026-07-14T23:43:13.840210+00:00الباحثون يقدمون مهمة QFES ومجموعة بيانات QFESum لتلخيص الأحداث المركزة على الاستعلام https://korshunov.ai/en/article/11864-amplitude-gating-improves-tool-structured-llm-inference-via-amplitude-only-ffn/korshunov.aien2026-07-14T23:38:14.597844+00:00Amplitude Gating improves tool-structured LLM inference via amplitude-only FFN intervention https://korshunov.ai/ru/article/11864-amplitude-gating-uluchshaet-vyvod-llm-so-strukturirovannymi-instrumentami-za-v/korshunov.airu2026-07-14T23:38:14.597844+00:00Amplitude Gating улучшает вывод LLM со структурированными инструментами за счёт вмешательства в FFN только по амплитуде https://korshunov.ai/es/article/11864-el-enmascaramiento-de-amplitud-mejora-la-inferencia-de-llm-estructurada-para-ffn/korshunov.aies2026-07-14T23:38:14.597844+00:00El enmascaramiento de amplitud mejora la inferencia de LLM estructurada para herramientas mediante intervención FFN solo de amplitud https://korshunov.ai/zh-hans/article/11864-fu-du-men-kong-tong-guo-jin-she-ji-fu-du-de-qian-kui-wang-luo-gan-yu-ti-sheng-ju/korshunov.aizh-cn2026-07-14T23:38:14.597844+00:00幅度门控通过仅涉及幅度的前馈网络干预提升工具结构化大语言模型推理 https://korshunov.ai/pt-br/article/11864-gating-de-amplitude-melhora-a-inferencia-llm-estruturada-por-ferramentas-por-de/korshunov.aipt2026-07-14T23:38:14.597844+00:00Gating de Amplitude melhora a inferência LLM estruturada por ferramentas por meio de intervenção FFN apenas em amplitude https://korshunov.ai/hi/article/11864-empliittyuudd-getting-ttuul-sttrkcrdd-llm-inphrens-ko-kevl-empliittyuudd-ffn-ke/korshunov.aihi2026-07-14T23:38:14.597844+00:00एम्प्लीट्यूड गेटिंग टूल-स्ट्रक्चर्ड LLM इनफरेंस को केवल एम्प्लीट्यूड FFN हस्तक्षेप के माध्यम से बेहतर बनाता है https://korshunov.ai/ja/article/11864-zhen-fu-ke-toniyori-zhen-fu-nomiffnjie-ru-tetsu-rugou-zao-hua-llmtui-lun-kagai/korshunov.aija2026-07-14T23:38:14.597844+00:00振幅ゲートにより、振幅のみFFN介入でツール構造化LLM推論が改善 https://korshunov.ai/ko/article/11864-jinpog-geiteuneun-jinpog-jeonyong-ffn-gaeibeul-tonghae-dogu-gujohwa-llm-curoneul/korshunov.aiko2026-07-14T23:38:14.597844+00:00진폭 게이트는 진폭 전용 FFN 개입을 통해 도구 구조화 LLM 추론을 개선합니다 https://korshunov.ai/fr/article/11864-le-filtrage-d-amplitude-ameliore-l-inference-llm-structuree-par-outils-via-une-l/korshunov.aifr2026-07-14T23:38:14.597844+00:00Le filtrage d'amplitude améliore l'inférence LLM structurée par outils via une intervention FFN uniquement sur l'amplitude https://korshunov.ai/id/article/11864-amplitude-gating-meningkatkan-inferensi-llm-terstruktur-alat-melalui-intervensi/korshunov.aiid2026-07-14T23:38:14.597844+00:00Amplitude Gating meningkatkan inferensi LLM terstruktur-alat melalui intervensi FFN hanya-amplitudo https://korshunov.ai/ar/article/11864-t-zyz-ltdmyn-ls-wy-yhswn-stdll-nmdhj-llgt-lkbyr-lmhykl-bldwt-br-tdkhl-lshbk-lmmy/korshunov.aiar2026-07-14T23:38:14.597844+00:00تعزيز التضمين السعوي يحسّن استدلال نماذج اللغات الكبيرة المهيكلة بالأدوات عبر تدخل الشبكة الأمامية (FFN) بالسعة فقط https://korshunov.ai/en/article/11961-prism-ml-posts-bonsai-27b-benchmark-results/korshunov.aien2026-07-14T23:21:28.278504+00:00Prism-ML posts Bonsai-27B benchmark results https://korshunov.ai/ru/article/11961-prism-ml-opublikovala-rezultaty-benchmarka-bonsai-27b/korshunov.airu2026-07-14T23:21:28.278504+00:00Prism-ML опубликовала результаты бенчмарка Bonsai-27B https://korshunov.ai/es/article/11961-prism-ml-publica-resultados-del-benchmark-de-bonsai-27b/korshunov.aies2026-07-14T23:21:28.278504+00:00Prism-ML publica resultados del benchmark de Bonsai-27B https://korshunov.ai/zh-hans/article/11961-prism-ml-fa-bu-bonsai-27b-ji-zhun-ce-shi-jie-guo/korshunov.aizh-cn2026-07-14T23:21:28.278504+00:00Prism-ML 发布 Bonsai-27B 基准测试结果 https://korshunov.ai/pt-br/article/11961-prism-ml-publica-resultados-do-benchmark-bonsai-27b/korshunov.aipt2026-07-14T23:21:28.278504+00:00Prism-ML publica resultados do benchmark Bonsai-27B https://korshunov.ai/hi/article/11961-prism-ml-ne-bonsai-27b-bencmaark-prinnaam-postt-kie/korshunov.aihi2026-07-14T23:21:28.278504+00:00Prism-ML ने Bonsai-27B बेंचमार्क परिणाम पोस्ट किए https://korshunov.ai/ja/article/11961-prism-mlkabonsai-27bnohenchima-kujie-guo-wotou-gao/korshunov.aija2026-07-14T23:21:28.278504+00:00Prism-MLがBonsai-27Bのベンチマーク結果を投稿 https://korshunov.ai/ko/article/11961-prism-ml-bonsai-27b-bencimakeu-gyeolgwa-gesi/korshunov.aiko2026-07-14T23:21:28.278504+00:00Prism-ML, Bonsai-27B 벤치마크 결과 게시 https://korshunov.ai/fr/article/11961-prism-ml-publie-les-resultats-du-benchmark-bonsai-27b/korshunov.aifr2026-07-14T23:21:28.278504+00:00Prism-ML publie les résultats du benchmark Bonsai-27B https://korshunov.ai/id/article/11961-prism-ml-memposting-hasil-benchmark-bonsai-27b/korshunov.aiid2026-07-14T23:21:28.278504+00:00Prism-ML memposting hasil benchmark Bonsai-27B https://korshunov.ai/ar/article/11961-prism-ml-ynshr-ntyj-m-yr-bonsai-27b/korshunov.aiar2026-07-14T23:21:28.278504+00:00Prism-ML ينشر نتائج معيار Bonsai-27B https://korshunov.ai/en/article/11962-deepseek-v4-generates-no-man-s-sky-and-minecraft-hybrid-via-one-shot-coding/korshunov.aien2026-07-14T23:16:50.212895+00:00DeepSeek V4 generates No Man's Sky and Minecraft hybrid via one-shot coding https://korshunov.ai/ru/article/11962-deepseek-v4-sozdaet-gibrid-no-man-s-sky-i-minecraft-cherez-odnoshagovoe/korshunov.airu2026-07-14T23:16:50.212895+00:00DeepSeek V4 создаёт гибрид No Man's Sky и Minecraft через одношаговое программирование https://korshunov.ai/es/article/11962-deepseek-v4-genera-un-hibrido-de-no-man-s-sky-y-minecraft-mediante-programacion/korshunov.aies2026-07-14T23:16:50.212895+00:00DeepSeek V4 genera un híbrido de No Man's Sky y Minecraft mediante programación en un solo paso https://korshunov.ai/zh-hans/article/11962-deepseek-v4-tong-guo-yi-ci-xing-bian-cheng-sheng-cheng-wu-ren-shen-kong-yu-wo-de/korshunov.aizh-cn2026-07-14T23:16:50.212895+00:00DeepSeek V4 通过一次性编程生成《无人深空》与《我的世界》混合体 https://korshunov.ai/pt-br/article/11962-deepseek-v4-gera-hibrido-de-no-man-s-sky-e-minecraft-via-programacao-one-shot/korshunov.aipt2026-07-14T23:16:50.212895+00:00DeepSeek V4 gera híbrido de No Man's Sky e Minecraft via programação one-shot https://korshunov.ai/hi/article/11962-deepseek-v4-ne-vn-shontt-kodding-ke-maadhym-se-no-man-s-sky-aur-minecraft-kaa/korshunov.aihi2026-07-14T23:16:50.212895+00:00DeepSeek V4 ने वन-शॉट कोडिंग के माध्यम से No Man's Sky और Minecraft का संकर बनाया https://korshunov.ai/ja/article/11962-deepseek-v4kawanshiyotsutoko-teinkute-no-mansusukai-to-mainkurahuto-cheng/korshunov.aija2026-07-14T23:16:50.212895+00:00DeepSeek V4がワンショットコーディングで『ノーマンズスカイ』と『マインクラフト』のハイブリッドを生成 https://korshunov.ai/ko/article/11962-deepseek-v4ga-weonsyas-kodingeuro-no-maenjeu-seukaiwa-mainkeuraepeuteu/korshunov.aiko2026-07-14T23:16:50.212895+00:00DeepSeek V4가 원샷 코딩으로 노 맨즈 스카이와 마인크래프트 하이브리드 생성 https://korshunov.ai/fr/article/11962-deepseek-v4-genere-un-hybride-entre-no-man-s-sky-et-minecraft-via-un-codage-en/korshunov.aifr2026-07-14T23:16:50.212895+00:00DeepSeek V4 génère un hybride entre No Man's Sky et Minecraft via un codage en un seul essai https://korshunov.ai/id/article/11962-deepseek-v4-menghasilkan-hibrida-no-man-s-sky-dan-minecraft-melalui-pengkodean/korshunov.aiid2026-07-14T23:16:50.212895+00:00DeepSeek V4 menghasilkan hibrida No Man's Sky dan Minecraft melalui pengkodean satu kali https://korshunov.ai/ar/article/11962-deepseek-v4-ywlwd-hjynan-byn-no-man-s-sky-w-minecraft-br-lbrmj-fy-lmh-whd/korshunov.aiar2026-07-14T23:16:50.212895+00:00DeepSeek V4 يولّد هجيناً بين No Man's Sky و Minecraft عبر البرمجة في لمحة واحدة https://korshunov.ai/en/article/11965-simon-willison-creates-custom-codex-desktop-pet-using-gpt-5-6-and-gpt-image-2/korshunov.aien2026-07-14T23:08:30.661394+00:00Simon Willison creates custom Codex Desktop pet using GPT-5.6 and gpt-image-2 https://korshunov.ai/ru/article/11965-saimon-uillison-sozdal-kastomnogo-pitomtsa-dlia-codex-desktop-s-pomoshchiu-gpt-5/korshunov.airu2026-07-14T23:08:30.661394+00:00Саймон Уиллисон создал кастомного питомца для Codex Desktop с помощью GPT-5.6 и gpt-image-2 https://korshunov.ai/es/article/11965-simon-willison-crea-una-mascota-personalizada-para-codex-desktop-usando-gpt-5-6/korshunov.aies2026-07-14T23:08:30.661394+00:00Simon Willison crea una mascota personalizada para Codex Desktop usando GPT-5.6 y gpt-image-2 https://korshunov.ai/zh-hans/article/11965-simon-willison-shi-yong-gpt-5-6-he-gpt-image-2-wei-codex-desktop-chuang-jian-zi/korshunov.aizh-cn2026-07-14T23:08:30.661394+00:00Simon Willison 使用 GPT-5.6 和 gpt-image-2 为 Codex Desktop 创建自定义宠物 https://korshunov.ai/pt-br/article/11965-simon-willison-cria-um-pet-personalizado-para-o-codex-desktop-usando-gpt-5-6-e-2/korshunov.aipt2026-07-14T23:08:30.661394+00:00Simon Willison cria um pet personalizado para o Codex Desktop usando GPT-5.6 e gpt-image-2 https://korshunov.ai/hi/article/11965-simn-vilisn-ne-gpt-5-6-aur-gpt-image-2-kaa-upyog-krke-codex-desktop-ke-lie-ksttm/korshunov.aihi2026-07-14T23:08:30.661394+00:00सिमन विलिसन ने GPT-5.6 और gpt-image-2 का उपयोग करके Codex Desktop के लिए कस्टम पेट बनाया https://korshunov.ai/ja/article/11965-simon-willisonshi-kagpt-5-6togpt-image-2woshi-yong-shitecodex-cheng/korshunov.aija2026-07-14T23:08:30.661394+00:00Simon Willison氏がGPT-5.6とgpt-image-2を使用してCodex Desktopのペットを作成 https://korshunov.ai/ko/article/11965-simon-willisoni-gpt-5-6-mic-gpt-image-2reul-sayonghayeo-codex-desktop-peseul/korshunov.aiko2026-07-14T23:08:30.661394+00:00Simon Willison이 GPT-5.6 및 gpt-image-2를 사용하여 Codex Desktop 펫을 만듦 https://korshunov.ai/fr/article/11965-simon-willison-cree-un-animal-de-compagnie-personnalise-pour-codex-desktop-en-5/korshunov.aifr2026-07-14T23:08:30.661394+00:00Simon Willison crée un animal de compagnie personnalisé pour Codex Desktop en utilisant GPT-5.6 et gpt-image-2 https://korshunov.ai/id/article/11965-simon-willison-membuat-hewan-peliharaan-kustom-untuk-codex-desktop-menggunakan-5/korshunov.aiid2026-07-14T23:08:30.661394+00:00Simon Willison membuat hewan peliharaan kustom untuk Codex Desktop menggunakan GPT-5.6 dan gpt-image-2