Fix Predictor Export and Test Predictor

Summary: Make the key of Q-values consistent with previous predictor models Reviewed By: kittipatv Differential Revision: D16559447 fbshipit-source-id: 136c00c40f32eef08cbc5049c8ce62d640f0e8c4
arunbalas · Jul 31, 2019 · bdb88b3 · bdb88b3
1 parent 85782d0
commit bdb88b3
Showing 1 changed file with 2 additions and 1 deletion.
diff --git a/ml/rl/preprocessing/preprocessor.py b/ml/rl/preprocessing/preprocessor.py
@@ -531,7 +531,8 @@ def _check_preprocessing_output(self, batch, norm_params):
         :param norm_params: list of normalization parameters
         """
         feature_type = norm_params[0].feature_type
-        min_value, max_value = batch.min(), batch.max()
+        min_value, max_value = batch.min().item(), batch.max().item()
+
         if feature_type == "CONTINUOUS":
             # Continuous features may be in range (-inf, inf)
             pass