목록전체 글 (85)

My Vision, Computer Vision

[논문 요약/리뷰] LLaMA: Open and Efficient Foundation Language Models

LLaMA: Open and Efficient Foundation Language ModelsWe introduce LLaMA, a collection of foundation language models ranging from 7B to 65B parameters. We train our models on trillions of tokens, and show that it is possible to train state-of-the-art models using publicly available datasets exclusively, withoarxiv.org발행일 : 2023. 02. 27.Meta AIProblem최근, 한정된 컴퓨터 예산(Budget)에서 LLM과 데이터셋 크기에 대한 최적화 연구..
Paper 2025. 2. 18. 13:23
[논문 리뷰/요약] BEIT: BERT Pre-Training of Image Transformers

BEiT: BERT Pre-Training of Image TransformersWe introduce a self-supervised vision representation model BEiT, which stands for Bidirectional Encoder representation from Image Transformers. Following BERT developed in the natural language processing area, we propose a masked image modeling task to prearxiv.orgProblemBEIT(Bidirectional Encoder representation from Image Transformers)를 제안한다.ViT는 CNN..
Paper 2025. 2. 14. 16:01
[논문 리뷰/요약] BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

BERT: Pre-training of Deep Bidirectional Transformers for Language UnderstandingWe introduce a new language representation model called BERT, which stands for Bidirectional Encoder Representations from Transformers. Unlike recent language representation models, BERT is designed to pre-train deep bidirectional representations from unlaarxiv.org발행일 : 2018. 10. 11.Google AI LanguageProblem기존의 Langu..
Paper 2025. 2. 12. 15:49
[논문 리뷰/요약] BLIP-2: Bootstrapping Language-Image Pre-training

BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language ModelsThe cost of vision-and-language pre-training has become increasingly prohibitive due to end-to-end training of large-scale models. This paper proposes BLIP-2, a generic and efficient pre-training strategy that bootstraps vision-language pre-training from oarxiv.org Problem대부분의 State-of-the-art V..
Paper 2025. 2. 7. 18:35
[논문 리뷰/요약] CLIPPO: Image-and-Language Understanding from Pixels Only

CLIPPO: Image-and-Language Understanding from Pixels OnlyMultimodal models are becoming increasingly effective, in part due to unified components, such as the Transformer architecture. However, multimodal models still often consist of many task- and modality-specific pieces and training procedures. For example,arxiv.orgProblem대부분의 멀티모달 모델은 모달리티 별로 구성 요소가 다르고, 다른 태스크에 적용하기 위해 그에 맞는 추가적인 학습 절차가 필요..
Paper 2025. 2. 5. 16:22
[논문 리뷰/요약] CLIP-Adapter: Better Vision-Language Models with Feature Adapters

CLIP-Adapter: Better Vision-Language Models with Feature AdaptersLarge-scale contrastive vision-language pre-training has shown significant progress in visual representation learning. Unlike traditional visual systems trained by a fixed set of discrete labels, a new paradigm was introduced in \cite{radford2021learning}arxiv.orgProblem(문제 지적)CLIP은 클래스 이름을 프롬프트 템플릿에 넣어 Zero-shot transfer를 수행하는데, 이..
Paper 2025. 2. 5. 13:54
이전 Prev 1 ··· 3 4 5 6 7 8 9 ··· 15 Next 다음