npu_onnx_graph_op

Applicability

Product	Supported
Atlas A3 training products / Atlas A3 inference products	√
Atlas A2 training products / Atlas A2 inference products	√
Atlas 200I/500 A2 inference products	☓
Atlas inference products	☓
Atlas training products	√

Description

Loads an ONNX model as an operator and executes it on the Ascend AI Processor through the TensorFlow framework.

Prototype

      
           def npu_onnx_graph_op(inputs, tout, model_path, name=None)

Parameters

Parameter	Input/Output	Description
inputs	Input	Input tensor of the ONNX model.
tout	Input	Output data type of the ONNX model, for example, tf.float32.
model_path	Input	A string for the path (including the file name) of the ONNX model, for example, /test/test.onnx.
name	Input	A string for the name of the operator on the graph.

Returns

Operator output list.

Example

      
           from npu_bridge.estimator.npu_ops import npu_onnx_graph_op

input = tf.placeholder(dtype=tf.float32, shape=(1, 1, 5, 5), name="conv_input")
output = npu_onnx_graph_op([input], [tf.float32], model_path="conv2d.onnx", name="conv2d")

Parent topic: npu_bridge.estimator.npu_ops