[RLlib] Check that results has learner info appo test (ray-project#34381

) The appo kl coefficient learner test is flakey because we run training until there are some results. What can end up happening is that training is run for so long that eval results are available but not learner results This pr fixes this by training until there are learner results that are available not just evaluation results. Signed-off-by: Avnish <[email protected]> Signed-off-by: Jack He <[email protected]>
ProjectsByJackHe · May 4, 2023 · 710a7af · 710a7af
1 parent be9e063
commit 710a7af
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/rllib/algorithms/appo/tests/tf/test_appo_learner.py b/rllib/algorithms/appo/tests/tf/test_appo_learner.py
@@ -141,7 +141,7 @@ def test_kl_coeff_changes(self):
             # a asynchronous trainer and results are returned asynchronously.
             while 1:
                 results = algo.train()
-                if results:
+                if results and "info" in results and LEARNER_INFO in results["info"]:
                     break
             curr_kl_coeff = results["info"][LEARNER_INFO][DEFAULT_POLICY_ID][
                 LEARNER_STATS_KEY