No evaluation of brain performance

[MichalFEIT2020]

After 75k iterations and 124 episode (600 iteration per episodes), I have none evaluation of brain performance. In my previous project I have evaluation after 10k or 30k iterations. In Inklink I specify to evaluate after 10 episodes

curriculum {

            training {

                    EpisodeIterationLimit: 600,

                    LessonAssessmentWindow: 10,

                } 

For learning using PPO algorithm.
What can be wrong? i spend too much money for this project already and have no results at all.

[#2]

@MichalFEIT2020 PPO makes use of episode batching during training. 124 trajectories provides very little information for the algorithm and is probably not enough for the first training batch. 

For more information about batching and how to control the batch parameter, refer to [this documentation](https://docs.microsoft.com/en-us/bonsai/inkling/advanced/algorithm#ppo-parameters).

Note that different RL algorithms have different batching behaviors. I suspect that your previous project used one of the other algorithms (probably APEX), which would explain why you're seeing a difference in behavior.