Trained by multiple GPUs,Improve the training speed

[githubueser]

Hello author, Thank you very much for uploading the source code of your work project. Can the program be trained by calling multiple GPUs at the same time to improve the training speed?If I use the parameters in the project, 500 epochs, each round is 30 minutes, then 500 rounds is too long, can I modify the parameters somewhere, can I improve the training speed?We look forward to hearing from you, thank you

[Chenyingray]

作者您好，非常感谢您上传您的工作项目源代码。程序可以通过同时调用多个GPU进行训练，以提高训练速度吗？如果我使用项目中的参数，500个epoch，每轮30分钟，那么500轮太长了，我是否可以修改某处的参数，可以提高训练速度吗？期待您的回音，谢谢

请问你解决关于训练慢的问题了吗，我在2080ti上训练需要50分钟一个epoch