Model channel pruning method and device, computer equipment and storage medium
A computer program and channel technology, applied in the field of artificial intelligence, can solve the problems that the neural network model cannot be applied to memory and computing resources, and the storage space is large, so as to avoid the inability to apply to limited memory and computing resources and improve the effectiveness.
Pending Publication Date: 2022-05-06
SF TECH
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, the deeper the neural network, the more parameters the network model requires, and the larger the storage
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0050] In order to make the purpose, technical solution and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application, and are not intended to limit the present application.
[0051] The model channel pruning method provided by this application can be applied to such as figure 1shown in the application environment. Wherein, the terminal 102 communicates with the pruning platform 104 through the network. The user rewrites the target model to be pruned to the pruning platform 104 through the terminal, and inputs the expected pruning ratio. The pruning platform obtains the model pruning task constructed based on the terminal input data, and determines the target model to be pruned. The target layer, according to the first feature data type and th...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention relates to a model channel pruning method and device, computer equipment and a storage medium. The method comprises the steps of obtaining a model pruning task, determining a to-be-pruned target layer in a target model, obtaining first feature data and second feature data of a channel in the target layer according to a first feature data type and a second feature data type in the model pruning task, and obtaining a to-be-pruned target layer in the target layer according to the feature data types and feature data numerical values. Sorting channels in the target layer to obtain a first sorting result corresponding to the first feature data type and a second sorting result corresponding to the second feature data type, and screening out a target channel to be pruned in the target layer according to the first sorting result, the second sorting result and a preset pruning proportion parameter, and pruning the target channel. The target model is effectively and reliably compressed, and the situation that a neural network model cannot be suitable for equipment with limited memory and computing resources is avoided.
Description
technical field [0001] The present application relates to the technical field of artificial intelligence, in particular to a model channel pruning method, device, computer equipment and storage medium. Background technique [0002] With the development of artificial intelligence technology, the neural network model, as an important part in the field of artificial intelligence, is widely used in various business scenarios due to its high accuracy and easy to meet production requirements. [0003] However, as a highly computationally and memory-intensive model, neural networks place high demands on computing and storage resources. The depth of the deep neural network has a certain positive correlation with its expressive ability. The deeper the depth, the better the training effect and the higher the accuracy for specific tasks. However, the deeper the neural network is, the more parameters the network model has and the larger the storage space required, making the neural net...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More IPC IPC(8): G06N3/08
CPCG06N3/082
Inventor 杨小平
Owner SF TECH