Keras performance is slower than Jeremy's

Hi all,

I opened a p3 machine on AWS with fastai community and my training time in KERAS on dogscats dataset is as follows:

Epoch 1/3
359/359 [==============================] - 482s 1s/step - loss: 0.1965 - acc: 0.9447 - val_loss: 0.0622 - val_acc: 0.9844
Epoch 2/3
359/359 [==============================] - 95s 265ms/step - loss: 0.0823 - acc: 0.9716 - val_loss: 0.2045 - val_acc: 0.9623
Epoch 3/3
359/359 [==============================] - 90s 250ms/step - loss: 0.0670 - acc: 0.9763 - val_loss: 0.1073 - val_acc: 0.9757
CPU times: user 24min 27s, sys: 21.8 s, total: 24min 49s
Wall time: 11min 6s
Out[8]:

This compared to Jeremy’s:

Epoch 1/3
359/359 [==============================] - 81s 226ms/step - loss: 0.1780 - acc: 0.9486 - val_loss: 0.0805 - val_acc: 0.9743
Epoch 2/3
359/359 [==============================] - 77s 215ms/step - loss: 0.0928 - acc: 0.9697 - val_loss: 0.0592 - val_acc: 0.9814
Epoch 3/3
359/359 [==============================] - 74s 207ms/step - loss: 0.0746 - acc: 0.9730 - val_loss: 0.0617 - val_acc: 0.9814
CPU times: user 15min 53s, sys: 36.4 s, total: 16min 29s
Wall time: 3min 52s
Out[8]:
<keras.callbacks.History at 0x7f2395643f98>

Any ideas why? P3 is the newest and the fastest machine