Resnet152非线性稀疏压缩后模型

本数据集为国家重点研发计划科技创新2030-“新一代人工智能”重大项目（项目编号：2022ZD0162300）“面向硬件加速的通用视觉模型压缩技术研究”成果之一，核心内容是经过非线性稀疏压缩处理后的Resnet152模型相关数据，用于支撑通用视觉模型边缘端硬件部署的技术验证与应用拓展。主要数据项包括：Resnet152模型经Sparse-ReLU新型激活函数优化后的网络结构配置参数、各层非线性稀疏压缩权重参数、模型压缩过程中的关键中间参数记录，以及模型基础信息说明文档，可直接用于模型推理验证、硬件适配测试及后续模型优化迭代。

This dataset is one of the research outcomes of the project *Research on General-Purpose Visual Model Compression Technologies for Hardware Acceleration*, which falls under the National Major Project for "New Generation Artificial Intelligence" under the Science and Technology Innovation 2030 Initiative of the National Key R&D Program of China (Project No. 2022ZD0162300). Its core content comprises relevant data of the ResNet152 model processed through non-linear sparse compression, and is used to support technical verification and application expansion for edge-side hardware deployment of general-purpose visual models. The main data items include: network structure configuration parameters of the ResNet152 model optimized with the novel Sparse-ReLU activation function, non-linear sparse compression weight parameters of each layer, key intermediate parameter records during the model compression process, and basic model information documentation. This dataset can be directly applied for model inference verification, hardware adaptation testing, and subsequent model optimization and iteration.