Can any one specify hyper parameters for nlu models( pretrained_embeddings_spacy, supervised_embeddings ) and for core model (keras) and what’s their roles. I want to tune these parameters. Thanks.
You want to know the physical meaning of the hyperparameters?
Yes I want to understand every hyperparameter so i can tune models
Hyperparameters are coming usually from the mathematical approximation used to build the machine learning model, they could be sometimes latent variable. For deep learning models, they are usually related to the network structure of the model. You have to select them randomly or you can use this blog to get some optimized values:
Thank you I’ll take a look on it