ReduceType¶

class paddle.distributed. ReduceType

Specify the type of operation used for paddle.distributed.Partial(). It should be one of the following values:

• ReduceType.kRedSum

• ReduceType.kRedMax

• ReduceType.kRedMin

• ReduceType.kRedProd

• ReduceType.kRedAvg

• ReduceType.kRedAny

• ReduceType.kRedAll

Examples:

>>> import paddle
>>> import paddle.distributed as dist
>>> mesh = dist.ProcessMesh([0, 1], dim_names=["x"])
>>> a = paddle.ones([10, 20])
>>> 
>>> # distributed tensor
>>> d_tensor = dist.shard_tensor(a, mesh, [dist.Partial(dist.ReduceType.kRedSum)])

Members:

kRedSum

kRedMax

kRedMin

kRedProd

kRedAvg

kRedAny

kRedAll

property name