Move from Python 3.5 to 3.6
Opened this issue · 0 comments
AnnaBeers commented
DeepNeuro is currently written for Python 3.5, not 3.6. You would think this wouldn't be the biggest deal, but this has already caused some problems with different dictionary performance between 3.5 and 3.6, and more importantly causes problems loading Keras models saved in 3.5 with Lambda layers in 3.6.
Changes will need to happen with all saved models, converting them from 3.5 to 3.6, and all current Docker containers.