How I can access the action output of the actor network in DDPG during training?

I want to access the action output of the actor network in DDPG during training since I want to change it by force function to other action optimized from sepeate function to accelerate training and improve learning effeciecncy for actor , if any help for that? I wil be thankful

Answer 1

[Yash]

You can use the function getAction which returns action from agent, actor or policy object given environment observations. You can write a custom loss function that directly uses getAction and dlgradient within it, and then use dlfeval and dlaccelerate with your custom loss function. For an example, see Train Reinforcement Learning Policy Using Custom Training Loop and Custom Training Loop with Simulink Action Noise.