Proper documentation for GO BOT policy


There is no proper documentation present for the policy model used in deeppavlov GO-Bot models. can anyone provide me the relevant documentation.

Need proper elaboration for all #TO-DO comments in code.

Hey! The policy model lacks documentation indeed. My best suggestion for now is to dive into the paper for insights on why the NN is built the way it’s built.

As for module architecture, it takes some data of utterancefeatures type, and predicts the distribution of probas among responces known by system. Feel free to ping me if you need clarification on some specific part of code