tinman (3) [Avatar] Offline
#1
I've installed the latest versions of Python (3.6), Keras and Tensorflow - CPU. All examples I've been running so far are working, with an exception for which I've found the root cause (see other post.)

(I'm using the provided github code, there are no typing errors.)

The problems start with chapter 5.1. There are no errors, the code is running. Problem is, the network isn't learning. The accuracy stays at 0.1 which indicates random choices. This is true for the training data and the test data.

Have any of you encountered the same problems? Any idea what could be wrong?

Edit:
I forgot some relevant information. I'm using Ubuntu 16.04 in a VM I've created for this purpose. All I did was a system update, Python 3.6 installation, then I installed Keras/Tensorflow, Ipython etc. in a virtualenv.
JK (2) [Avatar] Offline
#2
I encounter the same issue.

model.compile(optimizer='rmsprop',
loss='categorical_crossentropy',
metrics=['accuracy'])
model.fit(train_images, train_labels, epochs=5, batch_size=64)

Epoch 1/5
60000/60000 [==============================] - 53s 875us/step - loss: 10.8770 - acc: 0.2875
Epoch 2/5
60000/60000 [==============================] - 53s 875us/step - loss: 14.3159 - acc: 0.1118
Epoch 3/5
60000/60000 [==============================] - 53s 880us/step - loss: 14.5471 - acc: 0.0975
Epoch 4/5
60000/60000 [==============================] - 53s 880us/step - loss: 14.5471 - acc: 0.0975
Epoch 5/5
60000/60000 [==============================] - 53s 881us/step - loss: 14.5471 - acc: 0.0975

test_acc
0.098199999999999996
563254 (1) [Avatar] Offline
#3
I got learning to happen by switching to an adam optimizer.