When I learned about RL we were taught to disable exploration when doing evaluation of the model since exploration part is stochastic. I don't think that would

When I learned about RL we were taught to disable exploration when doing evaluation of the model since exploration part is stochastic. I don't think that would work in production.