vanvalenlab / deepcell-tf / 1180

86%

push

travis-ci-com

fixed train-test split to be consistent in load_data and training functions (#210)

* Changed default test_size in data_utils.get_data and in training functions in training.py to 0.2. 

* Changed deepcell scripts in pixelwise, feature_pyramids and watershed folders to set a seed=0 and train_size, and pass them to load_data and to the training function, so the train-test split obtained in the notebook is the same as the one used for training.

* set get_data default seed=0

1 of 1 new or added line in 1 file covered. (100.0%)

3536 of 5079 relevant lines covered (69.62%)

10.44 hits per line