Update infer.py

infer.py

@@ -26,8 +26,8 @@ def tokenize_function(example, tokenizer):
     words_lengths = [len(item) for item in question_sub_words_ids + context_sub_words_ids]
 
     return {
-        "input_ids": input_ids,
-        "words_lengths": words_lengths,
+        "input_ids": input_ids[:tokenizer.max_len_single_sentence],
+        "words_lengths": words_lengths[:tokenizer.max_len_single_sentence],
         "valid": valid
     }