冻结训练随笔

前言

我使用的是yolov5-6.1在冻结部分好像与6.0有差异，学习一下冻结训练顺便想实现之前这一篇提到的冻结任意层的操作

参照了之前的版本

对比一下现在的冻结模块

 parser.add_argument('--freeze', nargs='+', type=int, default=[0], help='Freeze layers: backbone=10, first3=0 1 2')

run中的

freeze = [f'model.{x}.' for x in (freeze if len(freeze) > 1 else range(freeze[0]))]  # layers to freeze
    for k, v in model.named_parameters():
        v.requires_grad = True  # train all layers
        if any(x in k for x in freeze):
            LOGGER.info(f'freezing {k}')
            v.requires_grad = False

emm…换句话说现在好像不用改了
①如果输入一个数：就是冻结到该层
②如果输入层号：就是冻结相应层

至于冻结训练的学习率：
前面大体骨架是可以不用怎么调的，因此lr应该低一些‘
主要是要训练最后的全连接层，此时lr就要适度增加

层结构

    for k, v in model.named_parameters():
        print(k)

output：

model.0.conv.weight
model.0.bn.weight
model.0.bn.bias
model.1.conv.weight
model.1.bn.weight
model.1.bn.bias
model.2.cv1.conv.weight
model.2.cv1.bn.weight
model.2.cv1.bn.bias
model.2.cv2.conv.weight
model.2.cv2.bn.weight
model.2.cv2.bn.bias
model.2.cv3.conv.weight
model.2.cv3.bn.weight
model.2.cv3.bn.bias
model.2.m.0.cv1.conv.weight
model.2.m.0.cv1.bn.weight
model.2.m.0.cv1.bn.bias
model.2.m.0.cv2.conv.weight
model.2.m.0.cv2.bn.weight
model.2.m.0.cv2.bn.bias
model.3.conv.weight
model.3.bn.weight
model.3.bn.bias
model.4.cv1.conv.weight
model.4.cv1.bn.weight
model.4.cv1.bn.bias
model.4.cv2.conv.weight
model.4.cv2.bn.weight
model.4.cv2.bn.bias
model.4.cv3.conv.weight
model.4.cv3.bn.weight
model.4.cv3.bn.bias
model.4.m.0.cv1.conv.weight
model.4.m.0.cv1.bn.weight
model.4.m.0.cv1.bn.bias
model.4.m.0.cv2.conv.weight
model.4.m.0.cv2.bn.weight
model.4.m.0.cv2.bn.bias
model.4.m.1.cv1.conv.weight
model.4.m.1.cv1.bn.weight
model.4.m.1.cv1.bn.bias
model.4.m.1.cv2.conv.weight
model.4.m.1.cv2.bn.weight
model.4.m.1.cv2.bn.bias
model.5.conv.weight
model.5.bn.weight
model.5.bn.bias
model.6.cv1.conv.weight
model.6.cv1.bn.weight
model.6.cv1.bn.bias
model.6.cv2.conv.weight
model.6.cv2.bn.weight
model.6.cv2.bn.bias
model.6.cv3.conv.weight
model.6.cv3.bn.weight
model.6.cv3.bn.bias
model.6.m.0.cv1.conv.weight
model.6.m.0.cv1.bn.weight
model.6.m.0.cv1.bn.bias
model.6.m.0.cv2.conv.weight
model.6.m.0.cv2.bn.weight
model.6.m.0.cv2.bn.bias
model.6.m.1.cv1.conv.weight
model.6.m.1.cv1.bn.weight
model.6.m.1.cv1.bn.bias
model.6.m.1.cv2.conv.weight
model.6.m.1.cv2.bn.weight
model.6.m.1.cv2.bn.bias
model.6.m.2.cv1.conv.weight
model.6.m.2.cv1.bn.weight
model.6.m.2.cv1.bn.bias
model.6.m.2.cv2.conv.weight
model.6.m.2.cv2.bn.weight
model.6.m.2.cv2.bn.bias
model.7.conv.weight
model.7.bn.weight
model.7.bn.bias
model.8.cv1.conv.weight
model.8.cv1.bn.weight
model.8.cv1.bn.bias
model.8.cv2.conv.weight
model.8.cv2.bn.weight
model.8.cv2.bn.bias
model.8.cv3.conv.weight
model.8.cv3.bn.weight
model.8.cv3.bn.bias
model.8.m.0.cv1.conv.weight
model.8.m.0.cv1.bn.weight
model.8.m.0.cv1.bn.bias
model.8.m.0.cv2.conv.weight
model.8.m.0.cv2.bn.weight
model.8.m.0.cv2.bn.bias
model.9.cv1.conv.weight
model.9.cv1.bn.weight
model.9.cv1.bn.bias
model.9.cv2.conv.weight
model.9.cv2.bn.weight
model.9.cv2.bn.bias
model.10.conv.weight
model.10.bn.weight
model.10.bn.bias
model.13.cv1.conv.weight
model.13.cv1.bn.weight
model.13.cv1.bn.bias
model.13.cv2.conv.weight
model.13.cv2.bn.weight
model.13.cv2.bn.bias
model.13.cv3.conv.weight
model.13.cv3.bn.weight
model.13.cv3.bn.bias
model.13.m.0.cv1.conv.weight
model.13.m.0.cv1.bn.weight
model.13.m.0.cv1.bn.bias
model.13.m.0.cv2.conv.weight
model.13.m.0.cv2.bn.weight
model.13.m.0.cv2.bn.bias
model.14.conv.weight
model.14.bn.weight
model.14.bn.bias
model.17.cv1.conv.weight
model.17.cv1.bn.weight
model.17.cv1.bn.bias
model.17.cv2.conv.weight
model.17.cv2.bn.weight
model.17.cv2.bn.bias
model.17.cv3.conv.weight
model.17.cv3.bn.weight
model.17.cv3.bn.bias
model.17.m.0.cv1.conv.weight
model.17.m.0.cv1.bn.weight
model.17.m.0.cv1.bn.bias
model.17.m.0.cv2.conv.weight
model.17.m.0.cv2.bn.weight
model.17.m.0.cv2.bn.bias
model.18.conv.weight
model.18.bn.weight
model.18.bn.bias
model.20.cv1.conv.weight
model.20.cv1.bn.weight
model.20.cv1.bn.bias
model.20.cv2.conv.weight
model.20.cv2.bn.weight
model.20.cv2.bn.bias
model.20.cv3.conv.weight
model.20.cv3.bn.weight
model.20.cv3.bn.bias
model.20.m.0.cv1.conv.weight
model.20.m.0.cv1.bn.weight
model.20.m.0.cv1.bn.bias
model.20.m.0.cv2.conv.weight
model.20.m.0.cv2.bn.weight
model.20.m.0.cv2.bn.bias
model.21.conv.weight
model.21.bn.weight
model.21.bn.bias
model.23.cv1.conv.weight
model.23.cv1.bn.weight
model.23.cv1.bn.bias
model.23.cv2.conv.weight
model.23.cv2.bn.weight
model.23.cv2.bn.bias
model.23.cv3.conv.weight
model.23.cv3.bn.weight
model.23.cv3.bn.bias
model.23.m.0.cv1.conv.weight
model.23.m.0.cv1.bn.weight
model.23.m.0.cv1.bn.bias
model.23.m.0.cv2.conv.weight
model.23.m.0.cv2.bn.weight
model.23.m.0.cv2.bn.bias
model.24.m.0.weight
model.24.m.0.bias
model.24.m.1.weight
model.24.m.1.bias
model.24.m.2.weight
model.24.m.2.bias