概述
描述
DRP-AI for RZ/V2H and RZ/V2N supports a feature for efficiently calculating the pruned AI model. The DRP-AI Extension Pack provides a pruning function optimized for RZ/V2H and RZ/V2N. The DRP-AI optimized pruning function can be used in combination with this tool and PyTorch or TensorFlow training code.
What is pruning?
Nodes in a neural network are interconnected as shown in the figure. Methods of reducing the number of parameters by removing weights between nodes or removing nodes are referred to as “pruning”. A neural network to which pruning has not been applied is generally referred to as a dense neural network. And a neural network to which pruning has been applied is generally referred to as a sparse neural network. Applying pruning leads to a slight deterioration in the accuracy of the model but can reduce the power required by hardware and accelerate the inference process.
How to embed the pruned model
The pruned model can be embedded using DRP-AI TVM. Refer to the DRP-AI TVM page on GitHub for details on TVM.
https://github.com/renesas-rz/rzv_drp-ai_tvm
Note: As shown in the figure, pruning is an optional function. (Dense model also can be embedded.)
特性
- Pruning functions optimized for RZ/V2H and RZ/V2N
- Pruning ratio can be specified for balance between accuracy and power efficiency
- Supports 2 pruning modes for improving accuracy (One Shot/Gradual)
发布信息
DRP-AI Extension Pack Version 1.3.0 is available (Jan 2026)
- The packaging format has been updated to a wheel, eliminating the need for manual configuration of environment variables.
- The TensorFlow and PyTorch versions can now be installed separately.
目标设备
文档
|
|
|
|
|---|---|---|
| 类型 | 文档标题 | 日期 |
| 手册 - 软件 | PDF 1.78 MB | |
| 应用说明 |
PDF
440 KB
AI 生成的摘要:
The guide explains how to measure the processing speed improvements of sparse AI models created by pruning dense models without retraining. It outlines a three-step process: preparing dense and sparse models using PyTorch or TensorFlow, converting them with DRP-AI TVM, and running performance tests on hardware. The document includes example code for pruning with PyTorch and details the operating environment and related resources. It focuses on quickly confirming speed gains from pruning rather than accuracy evaluation.
|
|
2 项目
|
||
设计和开发
相关评估板和套件
RZ/V2H 四核视觉 AI MPU 评测套件
RZ/V2H AI MPU 评测套件 (RTK0EF0168C04000BJ) 用于评测瑞萨电子 RZ/V2H 四核视觉 AI MPU。 该套件包括 CPU 板和扩展 (EXP) 板。 支持 RZ/V2H 标准软件包，能轻松实现低功耗 AI 推理和视频流等软件开发任务。 此外，这款评测套件还支持图像信号处理器 (ISP)、3D 图形引擎 (GE3D) 和可信安全 IP 等可选功能。
远程测试该板卡
RZ/V2N 四核视觉 AI MPU 评测套件
RZ/V2N AI MPU 评测套件 (RTK0EF0186C03000BJ) 用于评测我们的 RZ/V2N 四核 视觉 AI MPU。 该套件包括一块 CPU 评测板、扩展 (EXP) 板和两块子板。 我们提供了 AI 软件开发包 (AI SDK) 作为该评测套件的软件开发环境。 它支持对低功耗 AI 推理、视频流和其他内置功能进行评测。 该套件还支持图像信号处理器 (ISP)、3D 图形引擎 (GE3D) 和可信安全 IP 等可选功能。
视频和培训
This video provides an overview of DRP-AI TVM, focusing on the integration of AI into "Endpoint" devices for efficient real-time processing. Renesas' DRP-AI acts as a powerful accelerator, offering key features that enhance the performance and capabilities of endpoint AI applications.
