From e5b24f8f26df91340a5f46b266706c2631d9fdb9 Mon Sep 17 00:00:00 2001 From: "Christian M. Salamut" <20795522+Juphex@users.noreply.github.com> Date: Thu, 18 Mar 2021 00:02:15 +0100 Subject: [PATCH] fixing loss --- intermediate_source/reinforcement_q_learning.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/intermediate_source/reinforcement_q_learning.py b/intermediate_source/reinforcement_q_learning.py index d196de61fda..71d2d89f4b5 100644 --- a/intermediate_source/reinforcement_q_learning.py +++ b/intermediate_source/reinforcement_q_learning.py @@ -431,7 +431,8 @@ def optimize_model(): expected_state_action_values = (next_state_values * GAMMA) + reward_batch # Compute Huber loss - loss = F.smooth_l1_loss(state_action_values, expected_state_action_values.unsqueeze(1)) + criterion = nn.SmoothL1Loss() + loss = criterion(state_action_values, expected_state_action_values.unsqueeze(1)) # Optimize the model optimizer.zero_grad()