diff --git a/src/extra.py b/src/extra.py
index 4da16330b8b2e354f62b3234c4dbcca33958ecef..84f20d9def1cad022f90abacadd646ecb34b04dc 100644
--- a/src/extra.py
+++ b/src/extra.py
@@ -8,6 +8,7 @@
 # consists of 1..20 agents on a 50x50 grid. Thus the RL agent has to learn to handle 1 upto 20 agents.
 #
 #   - https://github.com/mitchellgoffpc/flatland-training
+# ./adrian_egli_ppo_training_done.png
 #
 # The key idea behind this observation is that agent's can not freely choose where they want.
 #
diff --git a/src/images/adrian_egli_ppo_training_done.png b/src/images/adrian_egli_ppo_training_done.png
new file mode 100644
index 0000000000000000000000000000000000000000..d163a110b8d09713a99a1accdd0ea2191d151bbd
Binary files /dev/null and b/src/images/adrian_egli_ppo_training_done.png differ