Deep learning model adaptation method and device, and electronic equipment

Abstract

The invention discloses a deep learning model adaptation method and device, electronic equipment and a medium, and relates to the technical field of artificial intelligence, deep learning and cloud computing. According to the specific implementation scheme, the method comprises the steps: after model information of an original deep learning model and the hardware information of the target hardwareneeds to be adapted are acquired, querying the conversion path table according to the model information and the hardware information to obtain a matched target conversion path, converting the original deep learning model into an intermediate deep learning model in the conversion path according to the target conversion path, and converting the intermediate deep learning model into the target deeplearning model. Therefore, the model conversion path is determined based on the model information of the original deep learning model and the hardware information of the target hardware; the conversion of the deep learning model is carried out, the conversion of any type of original deep learning model into the target deep learning model adapted to any target hardware is achieved, and the problemthat the deep learning model is difficult to adapt to different hardware terminals is solved.

Description

technical field [0001] The present application relates to the field of computer technology, specifically to the field of artificial intelligence, deep learning, and cloud computing technology, and in particular to an adaptation method, device, electronic equipment, and storage medium for a deep learning model. Background technique [0002] At present, deep learning represented by neural networks is being widely implemented and applied in the industry. Based on the needs of large-scale deep learning model deployment, various types of artificial intelligence (AI) reasoning acceleration chips have been derived. However, various types of AI inference acceleration chips have different forms, different computing power, and various application scenarios. Therefore, various types of AI reasoning acceleration chips are often only suitable for deep learning models under a specific framework, and cannot run deep learning models under other frameworks. [0003] Therefore, how to adapt ...

AI Technical Summary

Problems solved by technology

Therefore, various types of AI reasoning acceleration chips are often only suitable for deep learning models under a specific framework, but cannot run deep learning models under other frameworks.

Images