I need to know if it's possible to work with accelerator like Jetson Nano and Intel Neural Compute Sticks in Reinforcement Learning Toolbox in the same way as in Deep Learning Toolbox.
I could not find the information in the documentation of the toolboxes. The RL Toolbox requires the DL Toolbox. But I have not find out if all layers and nets are saved in the DL Toolbox and further if the accelrator support packages like Simulink Code Support Package for NVIDIA Jetson CPUs supports the RL Toolboxes either.
The acceleration features of the DL Toolbox support packages (e.g. Jetson CPU support for Simulink Coder) will accelerate predictions on the hardware. This is true if your network is trained with Deep Learning Toolbox training (trainNetwork) or Reinforcement Learning Toolbox training (train), however once finished with RL training, users will typically use generatePolicyFunction to create a policy function that is deployable to hardware.