Hello, In my configuration file i have set time_horizon to 64 which is equal to a experince step. So that means that an Agent with the same Behaviour Name has to request 64 Decisions . If I have 32 Training Areas which are random with 32 Agents with the same Behaviour Name. Does the Training Algorithm take 64 request of each Agent or 64 Request of the same Behaviour Name? In the 2. case it would take 2 Requests per Agent and then collect them.