machine learning - How to use pickled file as dataset for keras -

i have build own dataset digit classification , worked convolutional network model developed lisa lab (here). wanted visualize weights , wanted through keras.

keras documentation tries load mnist data this:

(x_train, y_train), (x_test, y_test) = mnist.load_data() 

but want pickled dataset load instead of mnist default data. mnist module keras load it's dataset ? and, how can pass own dataset instead of use mnist module keras ?

thanks in advance.

you can lot of info method reading source: https://github.com/fchollet/keras/blob/master/keras/datasets/mnist.py

in case, dataset pickle file loaded amazon s3 bucket.

you write copy of function , use load different pickled dataset.