대충공부한거적어두는블로그

Faster R-CNNtwo stage 방식 (당연히 real-time은 어려움..)이미지 전체에 대해 backbone NW + region proposal NW 쭉 통과시키고각 region에 대해 cropping (rol-pool, rol-align) + class prediction + bbox translate prediction YOLO Introduction기존 모델들의 한계DPM(Deformable Parts Model): sliding window approach + 각각에 대해 classifier 돌아감 (비효율적)R-CNN: generate potential bbox(region proposal) + then run classifier, 느리고 optimize하기 어렵다single con..

https://littlefoxdiary.tistory.com/44 [논문리뷰] GPT3 - Language Models are Few-Shot Learners오픈 AI GPT 시리즈의 세 번째 논문이 공개되었씁니다!!!! GPT1 - Improving Language Understanding by Generative Pre-Training GPT2 - Language Models are Unsupervised Multitask Learners GPT3 - Language Models are Few-Shot Learners 2020/07/20littlefoxdiary.tistory.comGPT-3 관련 논문리뷰 (방법론, evaluation 위주임) AbstractBidirectional Encoder ..