Passing pixels and time to create an output state in train_et_smnist.py

[ijdavis]

In train_et_smnist.py:

output_states = rnn((pixel_input, time_input), mask=mask_input)

produces an error message about 64 not being a shape.

output_states = rnn(pixel_input, mask=mask_input)

runs but does not connect time_input which is passed into the model to the RNN.

It seems that the API no longer allows the above construct used in train_et_smnist.py
What is the correct format to present both pixel_input and time_input to the output_state of this RNN.

Thanks,
Ian Davis
cs.uwaterloo.ca/~ijdavis