fix ids_to_tokens naming error in tokenizer of deberta v2 (#12412)

Co-authored-by: Jipeng Huang <jihuan@microsoft.com>

src/transformers/models/deberta_v2/tokenization_deberta_v2.py

@@ -288,7 +288,7 @@ class SPMTokenizer:
         # <s> 1+1
         # </s> 2+1
         self.vocab = {spm.IdToPiece(i): i for i in range(bpe_vocab_size)}
-        self.id_to_tokens = [spm.IdToPiece(i) for i in range(bpe_vocab_size)]
+        self.ids_to_tokens = [spm.IdToPiece(i) for i in range(bpe_vocab_size)]
         # self.vocab['[PAD]'] = 0
         # self.vocab['[CLS]'] = 1
         # self.vocab['[SEP]'] = 2
@@ -351,7 +351,7 @@ class SPMTokenizer:
             self.special_tokens.append(token)
             if token not in self.vocab:
                 self.vocab[token] = len(self.vocab) - 1
-                self.id_to_tokens.append(token)
+                self.ids_to_tokens.append(token)
         return self.id(token)
 
     def part_of_whole_word(self, token, is_bos=False):