From d6c05668cb20f7966932401d82e663da92924e82 Mon Sep 17 00:00:00 2001
From: Dominik Roth <dominik.roth.dev@gmail.com>
Date: Tue, 22 Aug 2023 01:01:13 +0200
Subject: [PATCH] fixes

---
 sbBrix/common/policies.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/sbBrix/common/policies.py b/sbBrix/common/policies.py
index 61d5e38..9e06244 100644
--- a/sbBrix/common/policies.py
+++ b/sbBrix/common/policies.py
@@ -817,6 +817,7 @@ class Actor(BasePolicy):
             self.mu, self.log_std = self.action_dist.proba_distribution_net(
                 latent_dim=last_layer_dim, return_log_std=True, **dist_kwargs
             )
+            self._remember_log_std = log_std_init
             # Avoid numerical issues by limiting the mean of the Gaussian
             # to be in [-clip_mean, clip_mean]
             if clip_mean > 0.0: