Closed DCalhas closed 5 years ago
Try to implement this framework: https://github.com/kirthevasank/nasbot
Conflicts with fortran compiler, maybe implement this ourselves
For now hyperparameters in consideration:
Make kernel sizes of convolutions mutable
Normalize Bold Inputs
Integrate actual optimization of each architecture
Something is wrong with the decoder network, for some reason it freezes when processing inputs
Situation:
Check what is happening with memory usage: apparently using all the memory available in the GPU server. Reasons:
Use Bayesian Optimization Algorithm.
Parameters to have into account: