afe.ir.serializer.visualizer_defines

Python analogues of data structures for the Netron file format. These are temporary data structures representing data that will be written to a JSON file for Netron.

Attributes

NNodeName

Classes

NArray	Properties of an array, to be represented in a Keras JSON file's weightsManifest.
NLayerWeights	Constant arrays associated with a layer, to be represented in a Keras JSON file's
NInboundNode	Properties of an inbound node, to be represented in a Keras JSON file.
NLayer	Properties of a layer, to be represented in a Keras JSON file.
NInterfaceLayer	Properties of one input or output of a model, to be represented in a Keras JSON file.

Module Contents

afe.ir.serializer.visualizer_defines.NNodeName

class afe.ir.serializer.visualizer_defines.NArray

Properties of an array, to be represented in a Keras JSON file’s weightsManifest.

dtype: str

shape: List[int]

name: str

to_json() → Any

class afe.ir.serializer.visualizer_defines.NLayerWeights

Constant arrays associated with a layer, to be represented in a Keras JSON file’s weightsManifest.

weights: List[NArray]

to_json() → Any

class afe.ir.serializer.visualizer_defines.NInboundNode

Properties of an inbound node, to be represented in a Keras JSON file. An inbound node corresponds to one input graph edge of an AwesomeNode.

name: NNodeName

index: int

to_json() → Any

class afe.ir.serializer.visualizer_defines.NLayer

Properties of a layer, to be represented in a Keras JSON file. A layer corresponds to an AwesomeNode.

class_name: str

name: NNodeName

inbound_nodes: List[NInboundNode]

config: Dict[str, Any]

to_json() → Any

class afe.ir.serializer.visualizer_defines.NInterfaceLayer

Properties of one input or output of a model, to be represented in a Keras JSON file.

layer_name: NNodeName

layer_num: int

to_json()