网络构建-mindspore打卡

MindSpore神经网络构建

最新推荐文章于 2025-09-07 01:02:55 发布

bearang

最新推荐文章于 2025-09-07 01:02:55 发布

阅读量202

点赞数 10

CC 4.0 BY-SA版权

文章标签： python 深度学习

本文链接：https://blog.csdn.net/bearang/article/details/140235949

网络构建

网络构建通过继承nn.cell基类，在__init__方法中，进行cell的实例化和状态管理，并在construct实现tensorflow的操作，construct为神经网络构建
请添加图片描述

class Network(nn.Cell):
    def __init__(self):
        super().__init__()
        self.flatten = nn.Flatten()
        self.dense_relu_sequential = nn.SequentialCell(
            nn.Dense(28*28, 512, weight_init="normal", bias_init="zeros"),
            nn.ReLU(),
            nn.Dense(512, 512, weight_init="normal", bias_init="zeros"),
            nn.ReLU(),
            nn.Dense(512, 10, weight_init="normal", bias_init="zeros")
        )

    def construct(self, x):
        x = self.flatten(x)
        logits = self.dense_relu_sequential(x)
        return logits
# 查看网络        
model = Network()
print(model)

直接输入数据，调用模型，得到二维的tensorflow输出，利用nn.softmax(),获得预测概率

X = ops.ones((1, 28, 28), mindspore.float32)
logits = model(X)
# print logits
logits

pred_probab = nn.Softmax(axis=1)(logits)
y_pred = pred_probab.argmax(1)
print(f"Predicted class: {y_pred}")

模型层

对上面的神经网络分解讲解

输入数据

input_image = ops.ones((3, 28, 28), mindspore.float32)

展开nn.Flatten

flatten = nn.Flatten()
flat_image = flatten(input_image)

全连接层nn.Dense

layer1 = nn.Dense(in_channles = 28*28,out_channels=20)
hidden1 = layer1(flat_image)

非线性激活nn.ReLU

hidden1 = nn.ReLU()(hidden1)

有序容器nn.SequentialCell

seq_modules = nn.SequentialCell(
flatten,
layer1,
nn.ReLU(),
nn.Dense(20,10)
)
logits = seq_modules(input_image)

预测概率nn.Softmax

softmax = nn.Softmax(axis = 1)
pred_probab = softmax(logits)

查看模型参数

print(f"Model structure: {model}\n\n")
for name, param in model.parameters_and_names():
    print(f"Layer: {name}\nSize: {param.shape}\nValues : {param[:2]} \n")