search

cycraig / MP-DQN

Source code for the dissertation: "Multi-Pass Deep Q-Networks for Reinforcement Learning with Parameterised Action Spaces"
MIT License
185 stars 49 forks source link

When does the update of the discrete action parameters happen? #2

Closed sasforce closed 5 years ago

sasforce commented 5 years ago

When does the update of the the discrete action parameters happen? There is no usage of '_action_update' function in 'learn' function of the file 'qpamdp.py'. Thank you.

cycraig commented 5 years ago

Hi, the Q-PAMDP class encapsulates a discrete action agent to which it delegates discrete action learning, in this case using Sarsa(λ). This agent is updated in the _rollout function of qpamdp.py, where the step function of the discrete agent is called.

sasforce commented 5 years ago

Thank you！