ML Agent Academy

Hello. I want to train my Unity Projects using DQN.
However, I don’t know how to connect Unity and Python(DQN) project.

I heard that using Academy is solution.
On ML Agent 0.8.1, it was possible to create Academy file, but in current releases, it is not able to create Academy scripts. So, I am stuck now. (Academy.Academy() is inaccessible due to its protection level)

On GitHub there are projects that used DQN to train their Unity projects such as Banana Collector.
But I couldn’t find the Unity projects (only .exe files), so I couldn’t find the settings for DQN learning.

How to solve this problem? Please, help me.
Thank you very much.

Hi!

Please have a look at our example colab to build a custom trainer using Unity and our Python API: ml-agents/colab/Colab_UnityEnvironment_2_Train.ipynb at main · Unity-Technologies/ml-agents · GitHub

Thank you!
Now I was stuck in using functions of ml-agents envs(my reference was just 0.8.1 version…), and this will help me a lot!!
Thank you very much!!!