keras 2.0, Convulution2D : Negative dimension size

Change
model.add(Convolution2D(32, 3, 3, activation='relu', input_shape=(1,28,28)))
to

1 2	model.add(Convolution2D(32, 3, 3, activation='relu', input_shape=(1,28,28),                         dim_ordering='th'))