Category: Uncategorized

Large language models (LLMs) have made important advances in artificial intelligence, with superior performance on various tasks as their parameters and training data grow. GPT-3, PaLM, and Llama-3.1 perform well in many applications with billions of parameters. However, when implemented in low-power platforms, scaling LLMs poses severe difficulties regarding training and inference queries. While it… Read more

LLMs like GPT-4 and LLaMA have gained significant attention for their exceptional capabilities in natural language inference, summarization, and question-answering tasks. However, these models often generate outputs that appear credible but include inaccuracies, fabricated details, or misleading information, a phenomenon termed hallucinations. This issue presents a critical challenge for deploying LLMs in applications where precision… Read more

Large Vision Language Models (LVLMs) have demonstrated significant advancements across various challenging multi-modal tasks over the past few years.  Their ability to interpret visual information in figures, known as visual perception, relied on visual encoders and multimodal training. Even with these advancements, visual perception errors still cause many mistakes in LVLMs and impact their ability… Read more

Il mercato mondiale del digital twin (o gemello digitale) sta crescendo rapidamente: varrà 11,5 miliardi di dollari alla fine del 2024 e 119,3 miliardi alla fine del 2029. Tuttavia, il 47% dei decisori IT delle imprese non ha mai sentito parlare di questa tecnologia. Il dato, emerso da un recente report di Research and Markets, evidenzia… Read more

AWS의 연례 리인벤트 개발자 컨퍼런스가 지난주에 막을 내렸다. 이 행사에서는 기술 발전 외에도 비용 최적화, 워크플로우 효율화, 가속화된 AI 애플리케이션 개발 등 CIO의 눈길을 끌 만한 발표가 다수 있었다. 이 거대 클라우드 업체는 아마존 세이지메이커, 아마존 Q 디벨로퍼, 아마존 베드록, Q 비지니스 등 인기 제품에 대한 몇 가지 새로운 기능과 업데이트를 선보였다. CIO에게 직접적인 영향을… Read more

Sequential Recommendation systems have crucial applications in industries like e-commerce and streaming services. These systems collect and analyze the user interaction data over time to predict their preferences. However, the ID-based representations of users and items these systems rely on face critical drawbacks when transferring the same model to a new system. The new system… Read more

It’s easy to forget that the new AI revolution heralded by ChatGPT and OpenAI kickstarted just two years ago and has been quickly embraced by both businesses and consumers. But unknown to many is India’s meteoric rise to become a global leader in AI adoption and it is one to watch: what happens in the… Read more

새로운 연구소는 온·오프라인 환경에서 활용 가능한 AI 에이전트 개발을 목표로 한다. 특히 이달 초 공개한 자체 AI 모델 ‘아마존 노바‘를 기반으로 기술력을 확장할 계획이다. 연구소는 아마존 본사가 위치한 시애틀이 아닌 샌프란시스코에 설립됐다. 핵심 인력은 아마존이 지난 6월 영입한 어뎁트 출신으로 구성됐다. 어뎁트는 2022년 설립 이후 AI 업무 자동화 기술을 통해 성장했으며, 아마존 협력 전 4억… Read more

Training AI models today isn’t just about designing better architectures—it’s also about managing data efficiently. Modern models require vast datasets and need those datasets delivered quickly to GPUs and other accelerators. The problem? Traditional data loading systems often lag behind, slowing everything down. These older systems rely heavily on process-based methods that struggle to keep… Read more

Quantum computing has long been seen as a promising avenue for advancing computational capabilities beyond those of classical systems. However, the field faces a persistent challenge: error rates. Quantum bits, or qubits, are inherently fragile, and minor disturbances can lead to computational errors. This sensitivity has limited the scalability and practical application of quantum systems.… Read more