From b3c6a589ad9036b03221e776a6929b2bc1eb4680 Mon Sep 17 00:00:00 2001 From: Jinyu Wang Date: Mon, 15 May 2023 04:31:58 +0000 Subject: [PATCH] fix DiscretePolicyGradient set_state bug; upgrade pymaro version --- maro/__misc__.py | 2 +- maro/rl/policy/discrete_rl_policy.py | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/maro/__misc__.py b/maro/__misc__.py index 1cb403e4..58b49571 100644 --- 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 index 50e6d894..289d150e 100644 --- a/maro/rl/policy/discrete_rl_policy.py +++ b/maro/rl/policy/discrete_rl_policy.py @@ -336,7 +336,7 @@ class DiscretePolicyGradient(DiscreteRLPolicy): } 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"]