fix: create BLEUScore for each n-gram

bleu_1234.py

@@ -103,11 +103,15 @@ class bleu_1234(evaluate.Metric):
         if isinstance(references[0], str):
             references = [[ref] for ref in references]
 
-        bleu = BLEUScore()
-        bleu1 = bleu(predictions, references, n_gram=1)
-        bleu2 = bleu(predictions, references, n_gram=2)
-        bleu3 = bleu(predictions, references, n_gram=3)
-        bleu4 = bleu(predictions, references, n_gram=4)
+        bleu_score_1 = BLEUScore(n_gram=1)
+        bleu_score_2 = BLEUScore(n_gram=2)
+        bleu_score_3 = BLEUScore(n_gram=3)
+        bleu_score_4 = BLEUScore(n_gram=4)
+
+        bleu1 = bleu_score_1(predictions, references, n_gram=1)
+        bleu2 = bleu_score_2(predictions, references, n_gram=2)
+        bleu3 = bleu_score_3(predictions, references, n_gram=3)
+        bleu4 = bleu_score_4(predictions, references, n_gram=4)
 
         output_dict = {
             "bleu-1": bleu1,