🚀 The feature, motivation and pitch I would like to be able to detect if CUDA graph capturing is underway from inside a TorchScript model. Currently this example fails to script: import torch import torch.nn as nn class MyModel(nn.Module...
cuda.current_stream()) with torch.cuda.stream(s): for i in range(3): out_mean = scatter_mean(src, index, out=out_mean) out_max, _ = scatter_max(src, index, out=out_max) out_add = scatter_add(src, index, out=out_add) torch.cuda.current_stream().wait_stream(s) g = torch...