Closed small-donkey closed 6 years ago
This applies to create examples too.
As well, the TrainingExampleIterator
could be change to become a ExampleIterator
.
This way, if we keep the same structure for testing examples we could use this class for both training, evaluation and prediction.
This will make us be able to remove some old fixtures (like get_cubes
) that I introduced at the real beginning.
This has been merged to master.
Currently the extract data function coded for training/testing datasets only. Need to modify the function to be used for prediction during grading.