深度学习：batch_size和学习率 及如何调整学习率衰减学习率 warm-up 策略
-柚子皮-学习率衰减import torch.optim as optimfrom torch.optim import lr_scheduler# 训练前的初始化optimizer = optim.Adam(net.parameters(), lr=0.001)#学习率衰减scheduler = lr_scheduler.StepLR(optimizer, 10, 0.1) # # 每过10个epoch，学习率乘以0.1# 训练过程中for n in n_epoch: ...
