A deep-learning program of computer vision, which finishes CIFAR-10 pictures classification.
一个基于计算机视觉的深度学习项目，用于完成CIFAR-10数据集的分类任务。
torch (1.13 for cuda 11.7)
torchvision
matplotlib
numpy
python 3.10
NVIDIA cuda 11.7
set 0 for checking testset
set 1 for continue training
set 2 for creating a new model to train
set 3 for checking both testset and trainset
times of training
The first time to train maybe you can set trainMode 2 and REPEAT 20.
Training with CPU will be VERY VERY SLOW!
首次训练可以将trainmode设置为2，REPEAT设置为20。 如果基于CPU训练会非常缓慢，建议使用NVIDIA显卡训练。
The accuracy of training is 95.73%, the accuracy of testing is 88.74%.