如何使用训练和验证代码来替换和训练自定义数据集?

I'm currently learning how to work with training and validation code, but I'm not sure how to replace the dataset with my own for training purposes. I have a labeled dataset that includes train, val, and test folders, but I'm not sure how to correctly configure the dataset paths and format. Does anyone have experience in replacing the dataset for training? I'd appreciate some guidance, such as how to correctly configure the data.yaml file and adjust the code to handle custom datasets. If you have any detailed examples or suggestions, that would be super helpful. Thank you in advance!