[Bug Fix] Add missing sync_progress call for parallel play (#1703)

HorizonRobotics · Sep 18, 2024 · 26acfad · 26acfad
1 parent 0283006
commit 26acfad
Showing 1 changed file with 2 additions and 0 deletions.
diff --git a/alf/trainers/policy_trainer.py b/alf/trainers/policy_trainer.py
@@ -1120,6 +1120,8 @@ def play(root_dir,
     else:
         selective_criteria_func = None
 
+    # Sync the progress for all environments in case parallel_play > 1
+    env.sync_progress()
     while episodes < num_episodes:
         # For parallel play, we cannot naively pick the first finished `num_episodes`
         # episodes to estimate the average return (or other statistics) as it can be