RTX5060 GPU CUDA12.8 +vscode 设计一个torch实例程序

目录基础硬件和软件条件安装vscode在本地创建test文件夹和程序运行程序基础硬件和软件条件在上一篇文章介绍了RTX5060显卡安装了cuda等驱动。安装vscode这一步下载官网程序双击安装即可。打开vscode安装python插件和文心comate AI编码插件。在本地创建test文件夹和程序在D盘创建一个torch_test文件夹。用vscode打开文件夹。用文心AI编码插件自动生成一个实例程序deep_learning_example.py运行程序首先启动环境打开anaconda prompt进入base环境再输入conda activate cu128-py133这个新环境cu128-py133在上一篇文章介绍安装cuda cuDNN torch(cu128-py133)D:\opencv_pypython deep_learning_example.py 使用设备: cuda 加载MNIST数据集...100.0%100.0%100.0%100.0% 训练集大小:60000测试集大小:10000模型结构: ConvNet((conv1): Conv2d(1,32,kernel_size(3,3),stride(1,1),padding(1,1))(conv2): Conv2d(32,64,kernel_size(3,3),stride(1,1),padding(1,1))(pool): MaxPool2d(kernel_size2,stride2,padding0,dilation1,ceil_modeFalse)(dropout): Dropout(p0.25,inplaceFalse)(fc1): Linear(in_features3136,out_features128,biasTrue)(fc2): Linear(in_features128,out_features10,biasTrue))开始训练(5epochs)... Epoch1/5:[Batch100]Loss:0.3126[Batch200]Loss:0.1049[Batch300]Loss:0.0473[Batch400]Loss:0.0505[Batch500]Loss:0.0741[Batch600]Loss:0.1664[Batch700]Loss:0.1003[Batch800]Loss:0.0381[Batch900]Loss:0.0190训练 - Loss:0.1634, Accuracy:94.94% 测试 - Loss:0.0382, Accuracy:98.65% Epoch2/5:[Batch100]Loss:0.0189[Batch200]Loss:0.0494[Batch300]Loss:0.0424[Batch400]Loss:0.0075[Batch500]Loss:0.0604[Batch600]Loss:0.0170[Batch700]Loss:0.0088[Batch800]Loss:0.0039[Batch900]Loss:0.0486训练 - Loss:0.0582, Accuracy:98.21% 测试 - Loss:0.0325, Accuracy:98.91% Epoch3/5:[Batch100]Loss:0.0095[Batch200]Loss:0.0194[Batch300]Loss:0.0740[Batch400]Loss:0.0246[Batch500]Loss:0.0721[Batch600]Loss:0.0172[Batch700]Loss:0.0177[Batch800]Loss:0.0176[Batch900]Loss:0.0252训练 - Loss:0.0452, Accuracy:98.61% 测试 - Loss:0.0301, Accuracy:99.06%