All Models are trained with PyTorch version 1.4
To use the generative model (computer_games_generation) follow instructions from Performance RNN - PyTorch. In addition all three prediction models recive the same input as well.
For all computer_games_generation models valence/arousal and bias/binary, the generation function should be injected with the relvant values. For example:
python3 generate.py -s saved_models_dir/model.sess -c midi_repo/processed/some/processed.data -bb bias_or_binary_value -va valence_or_arousal_value