Closed jbirkesteen closed 1 year ago
We've made a mistake in TDAgent.incremental_update(). Currently, the state we use as St+1 is the state just after At. However, St+1 should be the state after opponent's move, i.e. the state in which the agent has to decide its next action (At+1).
TDAgent.incremental_update()
We've made a mistake in
TDAgent.incremental_update()
. Currently, the state we use as St+1 is the state just after At. However, St+1 should be the state after opponent's move, i.e. the state in which the agent has to decide its next action (At+1).