CogVLM: Visual Expert For Large Language Models论文笔记 - 知乎github: https://github.com/THUDM/CogVLM简介认为原先的shallow alignment效果不好(如blip-2,llava等),提出了visual expert module用于特征的deep fusion在10项任务上达到SOTA,效果堪比PaL
GitHub - yangjianxin1/CPM: Easy-to-use CPM for Chinese text generation(基于CPM的中文文本生成)Easy-to-use CPM for Chinese text generation(基于CPM的中文文本生成) - GitHub - yangjianxin1/CPM: Easy-to-use CPM for Chinese t