PaddleX3.0-beta Ascend Edition, launched by PaddlePaddle, is an end-to-cloud collaborative low-code development tool designed to assist developers in addressing practical industrial issues with low costs and zero barriers. This version was released on June 27, 2023, focusing on seven major mainstream AI scenarios, featuring 68 high-quality PaddlePaddle models, and constructing 16 industrial-grade model production lines.

WeChat Screenshot_20240731082527.png

Key highlights include:

  • Rich model coverage, including image classification, object detection, image segmentation, OCR, text image layout analysis, text image information extraction, and time series analysis.

  • Low-code development approach, enabling full-process model production line development through a unified API interface, supporting custom model workflow integration.

  • Deep adaptation to Huawei's Ascend 910 chip, meeting diverse user requirements.

PaddleX3.0-beta Ascend Edition offers a convenient development paradigm, allowing developers to perform tasks such as data validation, training, evaluation, and inference without in-depth understanding of underlying principles, through unified commands and configurations. Additionally, PaddleX supports rapid model optimization, exposing key hyperparameters for developer adjustment. Trained models can be easily integrated into projects via simple Python APIs.

To further enhance user experience, the PaddleX team has deeply adapted to the Ascend training chip. Users only need to install the multi-hardware version of the PaddlePaddle framework and add device configuration parameters during training to use the PaddleX tool on Ascend hardware. Currently, dozens of models supported on Ascend hardware cover multiple domains.

PaddleX3.0-beta Open Source Experience Address:

https://github.com/PaddlePaddle/PaddleX/tree/release/3.0-beta