From ec10827b7f4197c933d5467764fcad193c6c3fcb Mon Sep 17 00:00:00 2001
From: Erik Nygren <erik.nygren@sbb.ch>
Date: Sun, 6 Oct 2019 20:21:13 -0400
Subject: [PATCH] removed "bug" with reward. Attention, currently it is cheaper
for an agent to wait if we cummulate rewards between the different state!
---
torch_training/training_navigation.py | 1 -
1 file changed, 1 deletion(-)
diff --git a/torch_training/training_navigation.py b/torch_training/training_navigation.py
index 252cf16..6f4622f 100644
--- a/torch_training/training_navigation.py
+++ b/torch_training/training_navigation.py
@@ -136,7 +136,6 @@ def main(argv):
else:
update_values = False
action = 0
- action_prob[action] += 1
action_dict.update({a: action})
# Environment step
--
GitLab