search

yiwenguo / Dynamic-Network-Surgery

Caffe implementation for dynamic network surgery.
Other
186 stars 70 forks source link

train problem #16

Closed botaichang closed 7 years ago

botaichang commented 7 years ago

when i finetune resnet model, I got this error: I0814 15:49:26.345660 3407 net.cpp:774] Copying source layer conv1 F0814 15:49:26.345674 3407 net.cpp:777] Check failed: target_blobs.size() == source_layer.blobs_size() (4 vs. 2) Incompatible number of blobs for layer conv1

it may because there are only 2 items(weights and bias) in convolution layer's param in the original caffemodel, but cconvolution layer has 4 items. when it make copy from source layer to target cconvolution layer, it got error.

Could you give an example of the prototxt ? Thank you!

yiwenguo commented 7 years ago

Hi, @melchior-note . Better use the command line interface instead with this repository and then you probably won't have this problem. For the Python interface, you can refer to #1 for a solution.

botaichang commented 7 years ago

I do use the command line interface, but it got this error because the param blob size is different between the original caffemodel and compressed caffemodel. If i change the conv layer's name, it will not get this error but we can not finetune original weights.

yiwenguo commented 7 years ago

Sorry I just found that your net.cpp file is probably different with the file provided in this repository. Please double-check.