Closed gavinmh closed 8 years ago
That's possible. SqueezeNet has 50x fewer weights than AlexNet, but the activations are not particularly small.
We have just posted a pre-release version of SqueezeNet_v1.1: https://github.com/DeepScale/SqueezeNet/tree/squeezenet_v1.1_preRelease
Compared to SqueezeNet v1.0, here is what v1.1 provides:
We haven't put up a deploy.prototxt
of SqueezeNet v1.1 yet, but see this post for how to create your own: https://github.com/DeepScale/SqueezeNet/issues/1
Thanks!
Thanks for sharing this work. I am comparing the GPU memory utilization of the BVLC CaffeNet and SqueezeNet. The GPU Memory usage is not what I expect on Ubuntu 14.04 with a Titan X.
Idle:
After loading a caffe.Classifier with SqueezeNet's weights and deploy.prototxt with PyCaffe in a Jupyter notebook:
While classiyfing with SqueezeNet: (t = timeit.Timer('net.predict([image], oversample=True).flatten().argsort()[:5]', 'from main import net, image') t.timeit(100):)
BVLC CaffeNet Comparison
Idle:
After creating a CaffeNet caffe.Classifier:
SqueezeNet appears to use more GPU memory than the reference BVLC CaffeNet. Am I missing something?