Why I cannot get 4096D feature vector

[maohui]

Hello,

Thanks for your code. I want to get 4096D feature vectors and train my SVM classifier. But when I ran the code, I always got the softmax results.
(1, 1000)
985

---

model.layers.pop()
model.layers.pop()

out = model.predict(im)
print out.shape
print np.argmax(out)

---

It seems that model.layers.pop() did not succeed.