broadcast

dragon.vm.torch.distributed.broadcast(
  tensor,
  src=0,
  group=None
)[source]

Broadcast the tensor from source node in a group.

Parameters:
  • tensor (Sequence[dragon.vm.torch.Tensor]) – The tensor(s) to reduce.
  • src (int) – The rank of the source node.
  • group (ProcessGroup, optional) – The group for communication.
Returns:

dragon.vm.torch.Tensor – The output tensor.