Cannot reset network to smaller learning rate - Optimization parameter

olafansau55olafansau55 Member Posts: 5 Learner I
edited May 17 in Help
Hello Guys
I'm working on a final project using rapid miner on COVID-19 data. I did the parameter optimization process. I entered a value for the learning rate in the range of 0.00 - 0.99 with a step count of 10 and for training_cycles in the range 1-200 for a step count of 10. Initially, there was no problem with that value, but when I changed the number of steps in training_cycles to 200, an error appeared like "Cannot reset the network to a smaller learning rate". I want to ask why this happened and how to solve it. If anyone can help out I would really appreciate it.

Thank you~

Best Answer

  • SabaRGSabaRG Member Posts: 13 Contributor II
    edited May 17 Solution Accepted
    Dear @olafansau55
    Unfortunately, I don't know why the "Try" operator can not work with the deep learning operator, which I suggest you use to ignore this bug, and the ignore error option of optimization operator can not handle it! It is a bug with the deep learning (H2O) operator.
    But, for your case, I suggest starting the learning rate from 0.1 or 0.2 and checking when it has no error. Your problem is with the learning rate.
    Sincerely

    #BugReport
    olafansau55

Answers

  • olafansau55olafansau55 Member Posts: 5 Learner I
    Thank you @SabaRG for your answer, i've changed the learning rate from 0.1 or 0.2 but it's give me the same error. Maybe I can try to use ignore an error .
Sign In or Register to comment.