Config file for transfer learning example

[bluque]

As asked in #9, I provide a config file example for doing transfer learning to another dataset. I took the tt100k_classif.py as the base and changed:

• dataset_name = 'BelgiumTSC'
• freeze_layers_from = 'base_model' (just because we are supposed to be finetuning on a similar dataset, but it doesn't matter)
• load_pretrained = True (as we are loading some pretrained weights, in TT100K for example)
• weights_file = '/path/to/pretrained/weights.hdf5' (I didn't put a real path because everyone will have a different one)
• weights_test_file = 'weights.hdf5' (this variable is actually not needed, if it doesn't exist it will load the trained weights in this experiment for the testing, which are found in the experiment's folder with the name 'weights.hdf5'. I added it in case we wanted to do a test with some other weights at some point)

[david-vazquez]

:)