Make sure evaluation metrics are computed correctly

app.py

@@ -87,8 +87,11 @@ sentences_labels, sentences_predictions = [], []
 for sample in tqdm(dataset):
     text = sample["sentence"]
     labels = [
-        1 if DIALECTS[i] in sample.keys() and int(sample[DIALECTS[i]]) == 1 else 0
-        for i in range(len(DIALECTS))
+        1
+        if DIALECTS_WITH_LABELS[i] in sample.keys()
+        and int(sample[DIALECTS_WITH_LABELS[i]]) == 1
+        else 0
+        for i in range(len(DIALECTS_WITH_LABELS))
     ]
     pred = predict_top_p(text)
     sentences_labels.append(labels)