During the first implementation of the AX library we noticed that there was an unexpected low performance from this library. This was performing well on mathematical challenges but it was performing poorly on machine learning ones.
The actual parser of the hyperparameters on AX uses it's type to decide if a numerical value is float or int when a numerical range hyperparameter is passed. We have to ensure that floats values are being passed as hyperparameters.
During the first implementation of the
AXlibrary we noticed that there was an unexpected low performance from this library. This was performing well on mathematical challenges but it was performing poorly on machine learning ones.The actual parser of the hyperparameters on
AXuses it's type to decide if a numerical value isfloatorintwhen a numerical range hyperparameter is passed. We have to ensure that floats values are being passed as hyperparameters.