After training wide& deep model for recommendation, weights of final layer becomes 0 leading to score of 0 for all inputs. I would like to know why this is happening and try weight initialization, batch norm, etc... To see how these affect weight as model gets trained. I want to see how weights and biases change at each layer. Is this right approach of debugging NN? If yes, do I need to look at all layers or just first and last(given I want to keep it simple as possible).



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.