单位建设网站的作用意义,沅江网站设计公司,不是常用的网页制作工具,乐清建设路小学网站手写数字识别#xff0c;计算机视觉领域的Hello World利用MNIST数据集#xff0c;55000训练集#xff0c;5000验证集。Pytorch实现神经网络手写数字识别感知机与神经元、权重和偏置、神经网络、输入层、隐藏层、输出层mac gpu的使用本节就是对Pytorch可以做的事情有个直观的…手写数字识别计算机视觉领域的Hello World利用MNIST数据集55000训练集5000验证集。Pytorch实现神经网络手写数字识别感知机与神经元、权重和偏置、神经网络、输入层、隐藏层、输出层mac gpu的使用本节就是对Pytorch可以做的事情有个直观的理解先理解表面把大概知识打通然后再研究细节的东西
import torch
import torch.nn as nn
import torchvision
import torchvision.transforms as transforms
import torch.optim as optim# Check that MPS is available
if not torch.backends.mps.is_available():if not torch.backends.mps.is_built():print(MPS not available because the current PyTorch install was not built with MPS enabled.)else:print(MPS not available because the current MacOS version is not 12.3 and/or you do not have an MPS-enabled device on this machine.)
else:device torch.device(mps)class Net(nn.Module):def __init__(self):super().__init__()# 28*28 784为输入100为输出self.fcl nn.Linear(784,100)self.fc2 nn.Linear(100,10)def forward(self,x):x torch.flatten(x,start_dim 1)x torch.relu(self.fcl(x))x self.fc2(x)return x# 当前模型对数据集学几次
max_epochs 5
# 每次训练模型对多少张图片进行训练
batch_size 16# data
# ToTensor 把当前数据类型转换为 Tensor
# Compose是组合多个转换操作的类
transform transforms.Compose([transforms.ToTensor()])# 55000
trainset torchvision.datasets.MNIST(root./data,trainTrue,downloadTrue,transformtransform)
train_loader torch.utils.data.DataLoader(trainset,batch_sizebatch_size,shuffleTrue)
testset torchvision.datasets.MNIST(root./data,trainFalse,downloadTrue,transformtransform)
test_loader torch.utils.data.DataLoader(testset,batch_sizebatch_size,shuffleTrue)# net init
net Net()
net.to(device)# nn.MSE
loss nn.CrossEntropyLoss()
optimizer optim.SGD(net.parameters(),lr0.0001)def train():acc_num0for epoch in range(max_epochs):for i,(data,label) in enumerate(train_loader):data data.to(device)label label.to(device)optimizer.zero_grad()output net(data)Loss loss(output,label)Loss.backward()optimizer.step()pred_class torch.max(output,dim1)[1]acc_num torch.eq(pred_class,label.to(device)).sum().item()train_acc acc_num / len(trainset)net.eval()acc_num 0.0best_acc 0with torch.no_grad():for val_data in test_loader:val_image,val_label val_dataoutput net(val_image.to(device))predict_y torch.max(output , dim1)[1]acc_num torch.eq(predict_y,val_label.to(device)).sum().item()val_acc acc_num/len(testset)print(train_acc,val_acc)if val_acc best_acc:torch.save(net.state_dict(),./minst.pth)best_acc val_accacc_num 0train_acc 0test_acc 0print(done)train()0.1348 0.3007
done
0.4361 0.5548
done
0.5870666666666666 0.6335
done
0.6435333333333333 0.672
done
0.67915 0.7011
done
