Since November 11, the OpenAI GPT-4 model has been accused of becoming "lazy", responding slowly at peak times, refusing to perform tasks or giving only partial results. This parameter is mainly visible when a user requests to write code, and the model only provides some information and instructs the user to fill in the rest on their own. This situation has caused complaints and discussions among users, and some miss the previous version of GPT-4.
ChatGPT is getting lazier and lazier and has learned to PUA humans in turn
Previous: 10年前，word2vec经典论文就预定了今天的NeurIPS时间检验奖