https://aiexpert.news/en/article/theoria-bridges-formal-proof-and-llm-judges-enabling-auditable-ai-verification ai|expert en 2026-07-02T06:45:53.286Z Theoria bridges formal proof and LLM judges with auditable verification https://aiexpert.news/pt/article/theoria-bridges-formal-proof-and-llm-judges-enabling-auditable-ai-verification ai|expert pt-BR 2026-07-02T06:45:53.286Z Theoria conecta prova formal e juízes LLM com verificação auditável https://aiexpert.news/es/article/theoria-bridges-formal-proof-and-llm-judges-enabling-auditable-ai-verification ai|expert es 2026-07-02T06:45:53.286Z Theoria conecta prueba formal y jueces LLM con verificación auditable https://aiexpert.news/en/article/beware-leaderboard-inflation-in-coding-agent-benchmarksan-audit ai|expert en 2026-07-02T04:28:42.697Z Three major benchmarks inflate coding-agent scores, audit finds https://aiexpert.news/pt/article/beware-leaderboard-inflation-in-coding-agent-benchmarksan-audit ai|expert pt-BR 2026-07-02T04:28:42.697Z Três grandes benchmarks inflacionam escores de agentes de código, auditoria descobre https://aiexpert.news/es/article/beware-leaderboard-inflation-in-coding-agent-benchmarksan-audit ai|expert es 2026-07-02T04:28:42.697Z Tres grandes benchmarks inflan las puntuaciones de agentes de código, descubre auditoría https://aiexpert.news/en/article/automem-making-llm-memory-io-a-trainable-skill-for-long-horizon-agents ai|expert en 2026-07-02T03:56:42.456Z AutoMem Training Doubles Agent Performance on Long-Horizon Tasks https://aiexpert.news/pt/article/automem-making-llm-memory-io-a-trainable-skill-for-long-horizon-agents ai|expert pt-BR 2026-07-02T03:56:42.456Z Treinamento do AutoMem Duplica Desempenho de Agentes em Tarefas de Longo Horizonte https://aiexpert.news/es/article/automem-making-llm-memory-io-a-trainable-skill-for-long-horizon-agents ai|expert es 2026-07-02T03:56:42.456Z El Entrenamiento de AutoMem Duplica el Desempeño del Agente en Tareas de Largo Horizonte https://aiexpert.news/en/article/training-one-transformer-layer-matches-full-parameter-rl-post-training-cost-impl ai|expert en 2026-07-02T03:24:42.960Z One Layer Matches Full RL Post-Training on Qwen Models https://aiexpert.news/pt/article/training-one-transformer-layer-matches-full-parameter-rl-post-training-cost-impl ai|expert pt-BR 2026-07-02T03:24:42.960Z Uma Camada Equivale ao Treinamento Completo com RL em Modelos Qwen https://aiexpert.news/es/article/training-one-transformer-layer-matches-full-parameter-rl-post-training-cost-impl ai|expert es 2026-07-02T03:24:42.960Z Una Capa Coincide con el Entrenamiento Completo de RL en Modelos Qwen https://aiexpert.news/en/article/gemma-4-cerebras-reaches-sub-200ms-voice-latency-at-scale ai|expert en 2026-07-02T02:52:42.560Z Cerebras and Gemma 4 reach sub-200ms voice latency with modular open stack https://aiexpert.news/pt/article/gemma-4-cerebras-reaches-sub-200ms-voice-latency-at-scale ai|expert pt-BR 2026-07-02T02:52:42.560Z Cerebras e Gemma 4 alcançam sub-200ms de latência de voz com stack modular aberto https://aiexpert.news/es/article/gemma-4-cerebras-reaches-sub-200ms-voice-latency-at-scale ai|expert es 2026-07-02T02:52:42.560Z Cerebras y Gemma 4 logran sub-200ms de latencia de voz con stack modular abierto https://aiexpert.news/en/article/databricks-silent-gpu-failure-detection-cuts-mttr-by-80 ai|expert en 2026-07-02T02:20:42.446Z Databricks GPU health checks detect silent failures in under five minutes https://aiexpert.news/pt/article/databricks-silent-gpu-failure-detection-cuts-mttr-by-80 ai|expert pt-BR 2026-07-02T02:20:42.446Z Verificações de saúde de GPU da Databricks detectam falhas silenciosas em menos de cinco minutos https://aiexpert.news/es/article/databricks-silent-gpu-failure-detection-cuts-mttr-by-80 ai|expert es 2026-07-02T02:20:42.446Z Las verificaciones de salud de GPU de Databricks detectan fallos silenciosos en menos de cinco minutos https://aiexpert.news/en/article/meta-shares-architecture-for-exabyte-scale-ml-data-layer ai|expert en 2026-07-02T01:48:42.845Z Meta's BLOB redesign cuts GPU idle from storage latency https://aiexpert.news/pt/article/meta-shares-architecture-for-exabyte-scale-ml-data-layer ai|expert pt-BR 2026-07-02T01:48:42.845Z O redesenho do BLOB da Meta reduz o tempo ocioso da GPU pela latência de armazenamento https://aiexpert.news/es/article/meta-shares-architecture-for-exabyte-scale-ml-data-layer ai|expert es 2026-07-02T01:48:42.845Z El rediseño de BLOB de Meta reduce la inactividad de GPU por latencia de almacenamiento https://aiexpert.news/en/article/cloudflares-agentic-internet-strategy-building-incentive-models-for-ai-agents ai|expert en 2026-07-01T14:46:42.701Z Cloudflare scales x402 payment protocol for AI agents https://aiexpert.news/pt/article/cloudflares-agentic-internet-strategy-building-incentive-models-for-ai-agents ai|expert pt-BR 2026-07-01T14:46:42.701Z Cloudflare escala protocolo de pagamento x402 para agentes de IA https://aiexpert.news/es/article/cloudflares-agentic-internet-strategy-building-incentive-models-for-ai-agents ai|expert es 2026-07-01T14:46:42.701Z Cloudflare escala protocolo de pago x402 para agentes de IA https://aiexpert.news/en/article/cloudflare-monetization-gateway-per-request-billing-for-agent-inference ai|expert en 2026-07-01T14:15:53.231Z Cloudflare's Monetization Gateway Opens Per-Request Billing for Agents https://aiexpert.news/pt/article/cloudflare-monetization-gateway-per-request-billing-for-agent-inference ai|expert pt-BR 2026-07-01T14:15:53.231Z Gateway de Monetização da Cloudflare Abre Cobrança Por Requisição para Agentes https://aiexpert.news/es/article/cloudflare-monetization-gateway-per-request-billing-for-agent-inference ai|expert es 2026-07-01T14:15:53.231Z La Puerta de Monetización de Cloudflare Abre Facturación Por Solicitud para Agentes https://aiexpert.news/en/article/browser-agents-learn-skills-from-human-traces-via-distillation ai|expert en 2026-07-01T13:36:42.249Z BrowserBC Lifts Browser Agent Success to 81% Using Human Traces https://aiexpert.news/pt/article/browser-agents-learn-skills-from-human-traces-via-distillation ai|expert pt-BR 2026-07-01T13:36:42.249Z BrowserBC Eleva Taxa de Sucesso de Agentes de Navegador para 81% Usando Rastreamentos Humanos https://aiexpert.news/es/article/browser-agents-learn-skills-from-human-traces-via-distillation ai|expert es 2026-07-01T13:36:42.249Z BrowserBC Eleva la Tasa de Éxito de Agentes de Navegador al 81% Utilizando Trazas Humanas https://aiexpert.news/en/article/self-explanation-training-enables-interpretable-model-behavior-tracking ai|expert en 2026-07-01T13:20:42.583Z Language Model Explanations Track Behavior Shifts Automatically https://aiexpert.news/pt/article/self-explanation-training-enables-interpretable-model-behavior-tracking ai|expert pt-BR 2026-07-01T13:20:42.583Z Explicações de Modelos de Linguagem Rastreiam Mudanças de Comportamento Automaticamente https://aiexpert.news/es/article/self-explanation-training-enables-interpretable-model-behavior-tracking ai|expert es 2026-07-01T13:20:42.583Z Las Explicaciones de Modelos de Lenguaje Rastrean Cambios de Comportamiento Automáticamente https://aiexpert.news/en/article/triage-credit-assignment-for-multi-action-agent-rl ai|expert en 2026-07-01T07:06:42.501Z TRIAGE Cuts Agent Actions 14.8% While Raising Success Rates https://aiexpert.news/pt/article/triage-credit-assignment-for-multi-action-agent-rl ai|expert pt-BR 2026-07-01T07:06:42.501Z TRIAGE Reduz Ações de Agentes em 14.8% Enquanto Aumenta Taxas de Sucesso https://aiexpert.news/es/article/triage-credit-assignment-for-multi-action-agent-rl ai|expert es 2026-07-01T07:06:42.501Z TRIAGE Reduce Acciones de Agentes 14.8% Mientras Aumenta Tasas de Éxito https://aiexpert.news/en/article/qval-evaluating-supervision-signals-for-long-horizon-agent-training ai|expert en 2026-07-01T06:54:42.539Z Simple Prompting Baselines Outperform Complex Supervision Methods https://aiexpert.news/pt/article/qval-evaluating-supervision-signals-for-long-horizon-agent-training ai|expert pt-BR 2026-07-01T06:54:42.539Z Baselines de Prompting Simples Superam Métodos de Supervisão Complexa https://aiexpert.news/es/article/qval-evaluating-supervision-signals-for-long-horizon-agent-training ai|expert es 2026-07-01T06:54:42.539Z Los Baselines de Prompting Simple Superan los Métodos de Supervisión Compleja https://aiexpert.news/en/article/generative-skill-composition-for-llm-agents ai|expert en 2026-07-01T06:34:42.685Z Researchers Close Gap Between AI Agents and Hand-Curated Skills https://aiexpert.news/pt/article/generative-skill-composition-for-llm-agents ai|expert pt-BR 2026-07-01T06:34:42.685Z Pesquisadores Fecham Lacuna Entre Agentes de IA e Habilidades Curadas Manualmente https://aiexpert.news/es/article/generative-skill-composition-for-llm-agents ai|expert es 2026-07-01T06:34:42.685Z Investigadores Cierran la Brecha entre Agentes de IA y Habilidades Curadas Manualmente https://aiexpert.news/en/article/databricks-moves-beyond-monolith-the-ltap-database-redesign ai|expert en 2026-07-01T06:12:43.661Z Databricks LTAP Eliminates CDC Pipelines for Transaction-Fresh Analytics https://aiexpert.news/pt/article/databricks-moves-beyond-monolith-the-ltap-database-redesign ai|expert pt-BR 2026-07-01T06:12:43.661Z Databricks LTAP Elimina Pipelines CDC para Analytics com Dados Atualizados https://aiexpert.news/es/article/databricks-moves-beyond-monolith-the-ltap-database-redesign ai|expert es 2026-07-01T06:12:43.661Z Databricks LTAP Elimina Pipelines CDC para Analytics con Datos Frescos de Transacciones https://aiexpert.news/en/article/employers-regret-ai-layoffs-the-6-month-reset ai|expert en 2026-07-01T05:40:42.570Z Half of AI-Driven Layoffs Will Be Reversed by 2026 https://aiexpert.news/pt/article/employers-regret-ai-layoffs-the-6-month-reset ai|expert pt-BR 2026-07-01T05:40:42.570Z Metade dos Cortes de Pessoal Impulsionados por IA Será Revertida até 2026 https://aiexpert.news/es/article/employers-regret-ai-layoffs-the-6-month-reset ai|expert es 2026-07-01T05:40:42.570Z La Mitad de los Despidos Impulsados por IA Serán Revertidos antes de 2026 https://aiexpert.news/en/article/training-llms-to-honestly-express-uncertainty-via-metacognitive-rl ai|expert en 2026-07-01T05:08:42.607Z New Training Technique Improves LLM Confidence Calibration by 63% https://aiexpert.news/pt/article/training-llms-to-honestly-express-uncertainty-via-metacognitive-rl ai|expert pt-BR 2026-07-01T05:08:42.607Z Nova Técnica de Treinamento Melhora Calibração de Confiança de LLMs em 63% https://aiexpert.news/es/article/training-llms-to-honestly-express-uncertainty-via-metacognitive-rl ai|expert es 2026-07-01T05:08:42.607Z Nueva Técnica de Entrenamiento Mejora la Calibración de Confianza de LLM en 63% https://aiexpert.news/en/article/clexeval-dont-trust-llm-clinical-reasoning-without-human-in-the-loop-eval ai|expert en 2026-07-01T04:36:42.542Z CLExEval Exposes 62% Accuracy Collapse Under Information Scarcity https://aiexpert.news/pt/article/clexeval-dont-trust-llm-clinical-reasoning-without-human-in-the-loop-eval ai|expert pt-BR 2026-07-01T04:36:42.542Z CLExEval Expõe Colapso de 62% na Precisão sob Escassez de Informações https://aiexpert.news/es/article/clexeval-dont-trust-llm-clinical-reasoning-without-human-in-the-loop-eval ai|expert es 2026-07-01T04:36:42.542Z CLExEval Expone Colapso de 62% en Precisión bajo Escasez de Información https://aiexpert.news/en/article/llm-agents-for-fault-recovery-process-plants-decode-alarms-and-prevent-shutdowns ai|expert en 2026-07-01T04:04:42.661Z Natural-Language Prompts Outperform Code in Industrial LLM Tests https://aiexpert.news/pt/article/llm-agents-for-fault-recovery-process-plants-decode-alarms-and-prevent-shutdowns ai|expert pt-BR 2026-07-01T04:04:42.661Z Prompts em Linguagem Natural Superam Código em Testes LLM Industriais https://aiexpert.news/es/article/llm-agents-for-fault-recovery-process-plants-decode-alarms-and-prevent-shutdowns ai|expert es 2026-07-01T04:04:42.661Z Prompts en Lenguaje Natural Superan el Código en Pruebas LLM Industriales https://aiexpert.news/en/article/elastic-atlas-open-source-agent-memory-grounded-in-cognitive-science ai|expert en 2026-07-01T03:04:42.248Z Elastic's Open-Source Agent Memory Hits 0.89 Recall@10 With Zero Leaks https://aiexpert.news/pt/article/elastic-atlas-open-source-agent-memory-grounded-in-cognitive-science ai|expert pt-BR 2026-07-01T03:04:42.248Z Atlas de Memória Open-Source da Elastic Atinge 0.89 Recall@10 com Vazamentos Zero https://aiexpert.news/es/article/elastic-atlas-open-source-agent-memory-grounded-in-cognitive-science ai|expert es 2026-07-01T03:04:42.248Z La Memoria de Agente de Código Abierto de Elastic Alcanza 0.89 Recall@10 con Fugas Cero https://aiexpert.news/en/article/scarfbench-evaluating-ai-agents-on-enterprise-code-migration ai|expert en 2026-07-01T02:32:42.348Z ScarfBench Reveals AI Agents Fail at Hidden Deployment Stages https://aiexpert.news/pt/article/scarfbench-evaluating-ai-agents-on-enterprise-code-migration ai|expert pt-BR 2026-07-01T02:32:42.348Z ScarfBench Revela que Agentes de IA Falham em Estágios Ocultos de Deploy https://aiexpert.news/es/article/scarfbench-evaluating-ai-agents-on-enterprise-code-migration ai|expert es 2026-07-01T02:32:42.348Z ScarfBench Revela que Agentes de IA Fallan en Etapas Ocultas de Despliegue https://aiexpert.news/en/article/nvidias-inference-stack-targets-lowest-cost-token-serving ai|expert en 2026-07-01T02:00:52.463Z NVIDIA's Blackwell Software Stack Cuts Inference Costs 5x https://aiexpert.news/pt/article/nvidias-inference-stack-targets-lowest-cost-token-serving ai|expert pt-BR 2026-07-01T02:00:52.463Z Stack de Software Blackwell da NVIDIA Reduz Custos de Inference em 5x https://aiexpert.news/es/article/nvidias-inference-stack-targets-lowest-cost-token-serving ai|expert es 2026-07-01T02:00:52.463Z Stack de Software Blackwell de NVIDIA Reduce los Costos de Inference 5x https://aiexpert.news/en/article/googles-tabfm-brings-zero-shot-foundation-model-logic-to-enterprise-tabular-data ai|expert en 2026-07-01T01:30:43.547Z Google Releases Zero-Shot Tabular Model but Hides Benchmark Data https://aiexpert.news/pt/article/googles-tabfm-brings-zero-shot-foundation-model-logic-to-enterprise-tabular-data ai|expert pt-BR 2026-07-01T01:30:43.547Z Google Lança Modelo Tabular Zero-Shot, Mas Esconde Dados de Benchmark https://aiexpert.news/es/article/googles-tabfm-brings-zero-shot-foundation-model-logic-to-enterprise-tabular-data ai|expert es 2026-07-01T01:30:43.547Z Google Lanza Modelo Tabular Zero-Shot Pero Oculta Datos de Benchmark