주민건

다른 사람과 나누고 싶은 싶은 뉴스와 정보 등을 공유합니다.

The first open source alternative to ChatGPT. 💪

We’ve trained a model to achieve a new state-of-the-art in mathematical problem solving by rewarding each correct step of reasoning (“process supervision”) instead of simply rewarding the correct final answer (“outcome supervision”). In addition to boosting performance relative to outcome supervisio…

800,000 step-level correctness labels on LLM solutions to MATH problems - GitHub - openai/prm800k: 800,000 step-level correctness labels on LLM solutions to MATH problems

대기자로 등록하고 무료로 사용해보세요

Official implementation of the paper "Segment Everything Everywhere All at Once" - GitHub - UX-Decoder/Segment-Everything-Everywhere-All-At-Once: Official implementation of the paper &quo...

Modern approaches typically formulate semantic segmentation as a per-pixelclassification task, while instance-level segmentation is handled with analternative mask classification. Our key insight: mask classification issufficiently general to solve both semantic- and instance-level segmentationt…

Image segmentation is about grouping pixels with different semantics, e.g.,category or instance membership, where each choice of semantics defines a task.While only the semantics of each task differ, current research focuses ondesigning specialized architectures for each task. We present Masked-a…

We present X-Decoder, a generalized decoding model that can predictpixel-level segmentation and language tokens seamlessly. X-Decodert takes asinput two types of queries: (i) generic non-semantic queries and (ii) semanticqueries induced from text inputs, to decode different pixel-level andtoken-…

오늘도 문제를 해결하고 있을 모든 디자이너에게

공고 자세히 보기

Detecting out-of-distribution (OOD) inputs during the inference stage iscrucial for deploying neural networks in the real world. Previous methodscommonly relied on the output of a network derived from the highly activatedfeature map. In this study, we first revealed that a norm of the feature map…

ImageBind is the first AI model capable of binding information from six modalities.