This dataset is based on the Wizard-LM dataset of MSRA, which translates instructions and then calls GPT to obtain answers. Wizard-LM contains many instructions that are more difficult than Alpaca. When translating Chinese questions, a small amount of instructions may be injected, causing translation failure. The Chinese answers in this dataset are obtained by asking questions based on the Chinese questions.