<< Back
This model line is based on the model_0004_inceptionresnetv2. It introduces 2 separate CNN backbones (and no additional other/history data inputs). None of the models of this line has ever been created and tested (trained). irv2 stands for InceptionResNetv2, tcb for two CNN backbones.
This model is supposed to be a test if 2 separate backbones for each output can learn better. We never trained it busy with other, more promising ideas and problem-solving.
Model summary:
- Total params:
112,381,058 - Trainable params:
112,259,970 - Non-trainable params:
121,088
Model architecture:
