这几天关于accuracy和loss的计算有一些疑惑,原来是自己还没有弄清楚。
给出实例
def train(train_loader, model, criteon, optimizer, epoch):
train_loss = 0
train_acc = 0
num_correct= 0
for step, (x,y) in enumerate(train_loader):
# x: [b, 3, 224, 224], y: [b]
x, y = x.to(device), y.to(device)
model.train()
logit
2022-03-28 19:56:54
39KB
c
cc
cu
1