From 716119c9f6fc519c90760f845a548c98a65ae320 Mon Sep 17 00:00:00 2001
From: "Egli Adrian (IT-SCI-API-PFI)" <adrian.egli@sbb.ch>
Date: Mon, 7 Dec 2020 22:04:50 +0100
Subject: [PATCH] typo
---
reinforcement_learning/ppo_agent.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/reinforcement_learning/ppo_agent.py b/reinforcement_learning/ppo_agent.py
index 44c5701..e603e70 100644
--- a/reinforcement_learning/ppo_agent.py
+++ b/reinforcement_learning/ppo_agent.py
@@ -189,7 +189,7 @@ class PPOAgent(Policy):
# The loss function is used to estimate the gardient and use the entropy function based
# heuristic to penalize the gradient function when the policy becomes deterministic this would let
- # the gardient to become very flat and so the gradient is no longer useful.
+ # the gradient becomes very flat and so the gradient is no longer useful.
loss = \
-torch.min(surr1, surr2) \
+ self.weight_loss * self.loss_function(state_values, rewards) \
--
GitLab