Update blimp.py

blimp.py

@@ -194,11 +194,16 @@ class Blimp(evaluate.Metric):
             tokenizer.add_special_tokens({"pad_token": existing_special_tokens[0]})
 
         run_all = len(predictions) == 0 or predictions[0] == "*"
-        blimp_sets = (
-            BLIMP_UIDS
-            if run_all
-            else [p for p in BLIMP_UIDS if p.lower() in predictions]
-        )
+
+        if run_all:
+            blimp_sets = BLIMP_UIDS
+        else:
+            blimp_sets = []
+            for p in predictions:
+                if p in BLIMP_UIDS:
+                    blimp_sets.append(p)
+                else:
+                    logging.logging.warning(f"{p} not a valid UID, skipping...")
 
         assert len(blimp_sets) > 0, "no valid phenomena selected"