Hyperparameter clarification

Hi sir,

After deployed Ml models in real time, every time ML model runs, executing, ML model works in the way based on trying different range of hyperparameters then keep the best one among tried?

or In real time, when every time ML model runs, will it work in the way instead of trying out different values, it must keep only the static best parameter what we find out before mode deploy ?

Hi @Anbu, once you have identified the best hyperparameters for your model those are use for the model used in deployment because as per Wikipedia definition

" In machine learning, a hyperparameter is a parameter whose value is used to control the learning process"

The hyperparameters are used during the learning (i.e. training) process therefore those don’t change once the model has been deployed.

Thank you sir for answer make sense but asked questions because proff andrew ng telling like eventhough you tune the best hyper parameter today and it might change over time due to the CPU or GPU or Data changes over time. So rule of thumb is to try different value of hyper parameters testing againt cross validation set and pick the best one works best.