You can use torch.manual_seed() to seed the RNG for all devices (both CPU and CUDA): ... an optional feature can run multiple convolution algorithms, ...
確定! 回上一頁