From fb6e8fc1487aa24c8149ecdf9fed2c02edfdd84d Mon Sep 17 00:00:00 2001
From: LunmRE <lunmhit@gmail.com>
Date: Wed, 12 May 2021 11:12:07 +0800
Subject: [PATCH] fix evaluate bugs in ClassicalTrainer

---
 genrl/trainers/classical.py | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/genrl/trainers/classical.py b/genrl/trainers/classical.py
index 2e75aeb2..b773b6e7 100644
--- a/genrl/trainers/classical.py
+++ b/genrl/trainers/classical.py
@@ -161,9 +161,11 @@ def evaluate(self, eval_ep: int = 100) -> float:
             ep_rew += reward
             if done:
                 ep_rews.append(ep_rew)
-                ep += 1
                 mean_ep_rew = np.mean(ep_rews)
-                if ep == 100:
+                ep += 1
+                state = self.env.reset()
+                ep_rew = 0
+                if ep == eval_ep:
                     print(
                         "Evaluated for {} episodes, Mean Reward: {:.2f}, Std Deviation for the Reward: {:.2f}".format(
                             eval_ep, mean_ep_rew, np.std(ep_rews)