search

forresti / SqueezeNet

SqueezeNet: AlexNet-level accuracy with 50x fewer parameters
BSD 2-Clause "Simplified" License
2.17k stars 723 forks source link

Type of squeezenet_v1.0.caffemodel is PCX image (image/x-pcx)? #5

Closed saustar closed 8 years ago

saustar commented 8 years ago

Hi, I have downloaded the provided model. But when I looked its properties, type is PCX image not binary??

How can I have a binary type??

forresti commented 8 years ago

Interesting. I assume you are talking about the *.caffemodel file?

Is Caffe able to load the file?

In general, an OS tries to make an educated guess about file types, but this guess isn't always helpful.

saustar commented 8 years ago

Yes Caffe is able to load the model. But I am trying to convert it into mxnet model. mxnet is not able to parse it. But its able to parse other Caffe model . Only difference i can see is file type. On Mar 23, 2016 22:04, "Forrest Iandola" notifications@github.com wrote:

Interesting. I assume you are talking about the *.caffemodel file?

Is Caffe able to load the file?

In general, an OS tries to make an educated guess about file types, but this guess isn't always helpful.

— You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub https://github.com/DeepScale/SqueezeNet/issues/5#issuecomment-200426874

forresti commented 8 years ago

Since you say that you are able to load the pretrained SqueezeNet into Caffe, I'm going to close this issue.

I think an MXNet implementation of SqueezeNet is a great idea. I don't have any experience with MXNet, so I'm afraid I don't have any debugging advice. Instead of using our pretrained weights, perhaps you could train SqueezeNet from scratch in Caffe or MXNet? We just committed an update to the solver.prototxt file which makes SqueezeNet training converge more reliably.