broadcast_object_list#

composer.utils.dist.broadcast_object_list(object_list, src=0, group=None)[source]#

Broadcasts picklable objects in object_list to the whole group.

Similar to broadcast(), but Python objects can be passed in. Note that all objects in object_list must be picklable in order to be broadcasted.

See also

torch.distributed.broadcast().

Parameters

  • object_list (Tensor) โ€“ list of input objects to broadcast. Each object must be picklable. Only objects on the src rank will be broadcast, but each rank must provide lists of equal sizes.

  • src (int, optional) โ€“ Source rank (default: 0)

  • group (ProcessGroup, optional) โ€“ The process group to work on. If None, the default process group will be used. Default is None.

Returns

None โ€“ object_list will be modified in-place and set to values of object_list from the src rank.