-
Notifications
You must be signed in to change notification settings - Fork 489
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
dcp压缩后,模型大小不变 #305
Comments
请问如果被屏蔽的通道没有去除会影响推理速度吗?能把那些屏蔽的通道能去除吗? |
前提是我没有用框架里的程序转为.pb文件 |
可以获得压缩后的模型,但是需要稍复杂一点的操作,尤其对于有residual block的网络结构来说,比如resnet50. 你需要确保residual block中add的两个conv的pruned 通道一致,否则会出现错误。 |
PocketFlow框架里给出的例子,比如resnet18、lenet对于cifar10数据集训练模型,都能够直接得到压缩后的模型,而且模型大小也能根据剪枝通道比例获得,并没有经过复杂的操作,能解释一下吗?我现在用的网络是resnet18、50和mobilenetv1、v2,训练数据是从imagenet中随机挑选40类,非常感谢。。。 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
您好,我无论在resnet还是mobilenet上,利用dcp通道剪枝后,模型大小始终不变,即使转成.pb文件也不变,但是模型测试相比原来有变化,请问这是什么原因呢
The text was updated successfully, but these errors were encountered: