My Vision, Computer Vision

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..

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..

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..

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..

DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement LearningWe introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrates remarkable reasoninarxiv.orgAbstract본 논문에서는 DeepSeek-R1-Zero와 DeepSeek-R1이라는 모델..

BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and GenerationVision-Language Pre-training (VLP) has advanced the performance for many vision-language tasks. However, most existing pre-trained models only excel in either understanding-based tasks or generation-based tasks. Furthermore, performance improvement has beearxiv.orgAbstractVision-Langauge Pre-t..