Deep learning model processing method and device and electronic equipment

Abstract

The invention provides a deep learning model processing method and apparatus, and an electronic device, which are applied to the technical field of artificial intelligence, and can realize visual model creation or model modification by a user through an intermediate format file, conveniently and accurately generate a model training code, and realize adaptation to multiple deep learning frameworks through one-time development. The method comprises the steps that in response to a first operation that a user selects at least one first control from a displayed model editing interface, at least one node and a connecting line between the at least one node are displayed on the model editing interface, the at least one node is in one-to-one correspondence with the at least one first control, and each node has node information; the connecting line of the at least one node is used for indicating the connection relationship between the at least one node; and generating a first target file based on the node information of the at least one node and the connection relationship between the at least one node, the first target file comprising structured information and graphical information of the first deep learning model.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, and in particular, to a method, apparatus and electronic device for processing a deep learning model. Background technique [0002] With the wide application of deep learning models such as Neural Network Model (NN Model), people have higher and higher requirements for the intelligence of the processing of deep learning models, such as more and more visualization requirements for the processing of deep learning models. higher. [0003] At present, there are two main types of visualization methods for deep learning models. In the first method, the visualization tool of the deep learning model can only read-only view the deep learning model, and does not have the function of editing (such as creating or modifying). For example, the most representative tool is netron, which can directly read the original deep learning model format and display it visually. The supported fo...

AI Technical Summary

Problems solved by technology

In the first type of method, the visualization tool of the deep learning model can only perform read-only viewing of the deep learning model, and does not have the function of editing (such as creating or modifying)

For example, the most representative tool is netron, which can directly read the original deep learning model format and visualize it. The supported formats are ONNX, Keras, Core ML, Caffe, Caffe2, Darknet, MXNet, ncnn, TensorFlow Lite , but without editing capabilities

In the second type, although the visualization tool of the deep learning model can perform visual arrangement of the model, allowing technicians to use less code or zero code to create a deep learning model, however, these tools are provided in the form of cloud services. , that is, a communication connection must be established with one or more devices that provide cloud services, resulting in a higher requirement for communication functions in the model orchestration process

If the user needs to view and modify the deep learning model stored in protobuf, he can only use the netron tool to visualize the neural network model, and then call out the code of the neural learning model to modify it, which will allow viewing and modification of the neural network model. The process of learning the model is more cumbersome

Images