笔记地址 | Note Address:深度学习计算 | J&Ocean BLOG (jiangwu.xyz)
MLP作为切入点学习深度学习计算
Learn Deep-Learning-Calculation based on MLP, using PyTorch
- 运行
mnist.py
和fashion_mnist.py
以获取图像数据集 - run
mnist.py
andfashion_mnist.py
to fetch the image-datasets
学习深度学习中的层和块,并在PyTorch中体现,
在block中实现自定义块、顺序块和块内前向传播
learn the concepts of layer and block in DL,
implement self-defined block, sequential block and forward function in block
implement no-parameter layer and self-defined layer in layer
参数管理,包括访问参数,参数初始化
implement the initiation and the access of parameters
模型参数的保存与加载
save and load the parameters of the model
深度学习需要GPU
DL uses GPU to calculate
笔记地址 | Note Address:卷积神经网络 | J&Ocean BLOG (jiangwu.xyz)
学习了卷积神经网络的定义和细节
learn the detail of CNN
学习了卷积层,池化层和经典CNN-LeNet
including convolution layer, pooling layer and the ordinary CNN --- LeNet
笔记地址 | Note Address:现代卷积神经网络 | J&Ocean BLOG (jiangwu.xyz)
现代卷积神经网络 modern CNN
实现了单GPU的AlexNet,在Fashion-MNIST/MNIST上测试
implement the AlexNet on single GPU, testing on Fashio-MNIST/MNIST