Array output for Custom Enviroment

TensorFlow General Discussion
1

I am working on a multi-goal environment that has an observation space of (2,1) and an action space of (144,). I am trying to get an array of float values from my AI but Dense is only providing singular integers. It is also difficult for me to compare other layer usage examples to custom environments.

If someone could provide some guide, I would really appreciate it.

3

I have worked with reshaping my AI model to be (1,144). However, my largest problem then became trying to build the model into the DQNAgent nb_actions, which I found only accepted integers. If there are more solid examples from Reinforced agents policy or SAC agents working with multiple actions, I would be really grateful

5

I just used baseline PPO