Add reward histogram in parametric DQN

Summary: Following D17536379, we add a reward histogram plot in parametric DQN workflows Reviewed By: kittipatv Differential Revision: D17682383 fbshipit-source-id: 5529ef07f411e77e938e008e84a3a225562c59f4
arunbalas · Oct 9, 2019 · 755ba3a · 755ba3a
1 parent 6d673a7
commit 755ba3a
Showing 1 changed file with 1 addition and 0 deletions.
diff --git a/ml/rl/training/parametric_dqn_trainer.py b/ml/rl/training/parametric_dqn_trainer.py
@@ -133,6 +133,7 @@ def train(self, training_batch) -> None:
         self.loss_reporter.report(
             td_loss=self.loss,
             reward_loss=reward_loss,
+            logged_rewards=reward,
             model_values_on_logged_actions=self.all_action_scores,
         )