Catch runtimeError when formatting LM files

Closes #221 (closed)

dan/datasets/extract/utils.py

@@ -193,10 +193,11 @@ class Tokenizer:
                     vocab_size=self.subword_vocab_size,
                     model_prefix=self.prefix,
                     user_defined_symbols=self.special_tokens,
-                    minloglevel=1,
                 )
             except Exception as e:
-                logger.warning(f"Failed to train a sentencepiece model for subword tokenization: {e}")
+                logger.warning(
+                    f"Failed to train a sentencepiece model for subword tokenization: {e}"
+                )
                 self.sentencepiece_model = None
                 return