2020年5月,OpenAI在长达72页的论文《https://arxiv.org/pdf/2005.14165Language Models are Few-Shot Learners》中发布了GPT-3,共有1750亿参数量,需要700G的硬盘存储,(GPT-2有15亿个参数),它比GPT-2有了极大的改进。根…
文章目录 MCPA: Multi-scale Cross Perceptron Attention Network for 2D Medical Image Segmentation摘要本文方法实验结果 SwinMM: Masked Multi-view with SwinTransformers for 3D Medical Image Segmentation摘要本文方法实验结果 MCPA: Multi-scale Cross Perceptron Att…