Add try/catch clause for sentences with large number of wordpieces.

combo/data/token_indexers/pretrained_transformer_mismatched_indexer.py

+import logging
+import sys
 from typing import Optional, Dict, Any, List, Tuple
 
 from allennlp import data
 from allennlp.data import token_indexers, tokenizers, IndexedTokenList, vocabulary
 from overrides import overrides
 
-from typing import List
-
+logger = logging.getLogger(__name__)
 
 @data.TokenIndexer.register("pretrained_transformer_mismatched_fixed")
 class PretrainedTransformerMismatchedIndexer(token_indexers.PretrainedTransformerMismatchedIndexer):
@@ -34,6 +35,7 @@ class PretrainedTransformerMismatchedIndexer(token_indexers.PretrainedTransforme
         Method is overridden in order to raise an error while the number of tokens needed to embed a sentence exceeds the
         maximal input of a model.
         """
+        try:
             self._matched_indexer._add_encoding_to_vocabulary_if_needed(vocabulary)
 
             wordpieces, offsets = self._allennlp_tokenizer.intra_word_tokenize(
@@ -57,6 +59,10 @@ class PretrainedTransformerMismatchedIndexer(token_indexers.PretrainedTransforme
 
             return self._matched_indexer._postprocess_output(output)
 
+        except ValueError as value_error:
+            logger.error(value_error)
+            sys.exit(1)
+
 
 class PretrainedTransformerIndexer(token_indexers.PretrainedTransformerIndexer):