55b431b17aba504ae7b75f6f97b4437101e50f38,torch/distributed/algorithms/ddp_comm_hooks/default_hooks.py,,allreduce_hook,#Any#Any#,5

Before Change

        >>> ddp_model.register_comm_hook(process_group, allreduce_hook)
    
    group_to_use = process_group if process_group is not None else dist.group.WORLD
    world_size = (
        process_group.size() if process_group is not None else dist.get_world_size()
    )

    tensor = bucket.get_tensors()[0]
    fut = dist.all_reduce(tensor, group=group_to_use, async_op=True).get_future()

After Change

        >>> ddp_model.register_comm_hook(process_group, allreduce_hook)
    
    group_to_use = process_group if process_group is not None else dist.group.WORLD
    world_size = group_to_use.size()

    tensor = bucket.get_tensors()[0]
    fut = dist.all_reduce(tensor, group=group_to_use, async_op=True).get_future()

Instances