How can you load all the packet data into the GPU memory in Keras (Theano backend)?

Question

Keras loads the data into a GPU packet package (tagged by the author here ).

For small datasets, this is very inefficient. Is there a way to change Keras or call Theano functions directly (after defining a model in Keras) to allow all parties to move to the front of the GPU and train using packages already in the GPU?

(Someone asked the same question on the Keras list a few weeks ago, but has no answers yet.)

Answer 1

( Embedding ). model.fit.