diff --git a/trl/trainer/dpo_trainer.py b/trl/trainer/dpo_trainer.py
index 8ef7ffb431..60a4aeffe4 100644
--- a/trl/trainer/dpo_trainer.py
+++ b/trl/trainer/dpo_trainer.py
@@ -638,6 +638,7 @@ def evaluation_loop(
                     )
                 }
             )
+            self.state.log_history.pop()
 
         # Base evaluation
         initial_output = super().evaluation_loop(