small bug fix in feedback (microsoft#771)

Author: peteryang1

rdagent/scenarios/data_science/dev/feedback.py

@@ -70,15 +70,15 @@ def generate_feedback(self, exp: DSExperiment, trace: DSTrace) -> ExperimentFeed
                         decision=True if self.scen.metric_direction else False,
                     )
                 elif cur_score < sota_score:
-                    return ExperimentFeedback(
+                    return HypothesisFeedback(
                         observations="The current score smaller than the SOTA score.",
                         hypothesis_evaluation="The current score is smaller than the SOTA score.",
                         new_hypothesis="No new hypothesis provided",
                         reason="The current score is smaller than the SOTA score.",
                         decision=False if self.scen.metric_direction else True,
                     )
                 else:
-                    return ExperimentFeedback(
+                    return HypothesisFeedback(
                         observations="The current score equals to the SOTA score.",
                         hypothesis_evaluation="The current score equals to the SOTA score.",
                         new_hypothesis="No new hypothesis provided",