fix DiscretePolicyGradient set_state bug; upgrade pymaro version

Jinyu Wang · Jinyu Wang · commit b3c6a589ad90 · 2023-05-15T04:31:58.000Z
diff --git a/maro/__misc__.py b/maro/__misc__.py
@@ -2,6 +2,6 @@
 # Licensed under the MIT license.
 
 
-__version__ = "0.3.2a2"
+__version__ = "0.3.2a3"
 
 __data_version__ = "0.2"
diff --git a/maro/rl/policy/discrete_rl_policy.py b/maro/rl/policy/discrete_rl_policy.py
@@ -336,7 +336,7 @@ def get_state(self) -> dict:
         }
 
     def set_state(self, policy_state: dict) -> None:
-        self._policy_net.set_state(policy_state)
+        self._policy_net.set_state(policy_state["net"])
         self._warmup = policy_state["policy"]["warmup"]
         self._call_count = policy_state["policy"]["call_count"]
 

Original file line number	Diff line number	Diff line change
`@@ -336,7 +336,7 @@ def get_state(self) -> dict:`
`336`	`336`	`}`
`337`	`337`
`338`	`338`	`def set_state(self, policy_state: dict) -> None:`
`339`		`- self._policy_net.set_state(policy_state)`
	`339`	`+ self._policy_net.set_state(policy_state["net"])`
`340`	`340`	`self._warmup = policy_state["policy"]["warmup"]`
`341`	`341`	`self._call_count = policy_state["policy"]["call_count"]`
`342`	`342`