[rllib] Basic infrastructure for off-policy estimation (IS, WIS) (#3941)

2026-07-01 03:08:48 +08:00 · 2019-02-13 16:25:05 -08:00
parent 729d0b2825
commit 2dccf383dd
34 changed files with 549 additions and 131 deletions
@@ -33,6 +33,7 @@ if __name__ == "__main__":
                agent_index=0,
                obs=obs,
                actions=action,
+                action_prob=1.0,  # put the true action probability here
                rewards=rew,
                prev_actions=prev_action,
                prev_rewards=prev_reward,