Update app.py

app.py

@@ -77,17 +77,6 @@ with st.sidebar:
 
 
 
-evaluator = pytrec_eval.RelevanceEvaluator(
-        qrels, pytrec_eval.supported_measures)
-
-if run1_file is not None:
-    run1, run1_pandas = load_run(run1_file)
-    results1 = evaluator.evaluate(run1) # dict of instance then metrics then values
-
-if run2_file is not None:
-    run2, run2_pandas = load_run(run2_file)
-    results2 = evaluator.evaluate(run2)
-
 
 col1, col2 = st.columns([1, 2], gap="medium")
 
@@ -97,6 +86,18 @@ is_better_run2_count = 0
 checkboxes = None
 with col1:
     corpus, queries, qrels = get_beir(dataset_name)
+    
+    evaluator = pytrec_eval.RelevanceEvaluator(
+            qrels, pytrec_eval.supported_measures)
+    
+    if run1_file is not None:
+        run1, run1_pandas = load_run(run1_file)
+        results1 = evaluator.evaluate(run1) # dict of instance then metrics then values
+    
+    if run2_file is not None:
+        run2, run2_pandas = load_run(run2_file)
+        results2 = evaluator.evaluate(run2)
+
     st.title("Instances")
     if run1_file is not None:
         name_of_columns = ["Overview"] + sorted([str(item) for item in set(run1_pandas.qid.tolist())])