Deeploy.DeeployTypes.TopologyOptimizer

class Deeploy.DeeployTypes.TopologyOptimizer(passes: List[TopologyOptimizationPass])

Bases: object

Wrapper object to apply multiple TopologyOptimizationPasses sequentially

Methods

__init__(passes: List[TopologyOptimizationPass])

__init__(passes)
optimize(graph)	Applies passes sequentially

optimize(graph: Graph) → Tuple[Graph]

Applies passes sequentially

Parameters:

graph (gs.Graph) – Current neural network graph

Returns:

Modified neural network graph

Return type:

Tuple[gs.Graph]