I would like to design a reward function. I am training two models from the first model that classify set of texts (paragraphs and keywords) and I also got some hidden states. The second model is trying to generate keywords for those paragraphs.

I want to use those hidden states from the first model to give rewards for key phrases that are generated from the second model. I want to know how can I implement this reward function since I have never used it before.

  • 1
    $\begingroup$ the question sounds a bit confusing to me, can you reframe it explain in more details which models you're using and where RL comes into play? I don't get what kind of policy you want to train. From what I understood you have a model that classify some text into paragraphs and another one that extract keywords for each paragraph, is it correct? $\endgroup$ Apr 16 '20 at 13:39

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.