Error while running the notebook

[herilalaina]

Hi, I have been unable to run the notebook here. It seems to be related to versioning, as I'm unable to instantiate the inference_utils.InferenceModel class without encountering an error.

Here's the specific error message I received when running the code:

      1 #@title Load a transformer inference model.
----> 3 inference_model = inference_utils.InferenceModel.from_checkpoint(
      4     CHECKPOINT_PATH,
      5     batch_size=BATCH_SIZE,
      6     model_gin_file=MODEL_GIN_FILE,
      7     gin_search_paths=[OPTFORMER_DIR],
      8 )
      9 vocab = inference_model.vocab
     10 study_converter = inference_model.study_converter

File ~/code/llm/optformer/optformer/t5x/inference_utils.py:394, in InferenceModel.from_checkpoint(cls, checkpoint_path_or_model_dir, batch_size, model_gin_file, gin_patterns_to_skip, gin_search_paths, overwrite_gin_files, overwrite_gin_bindings)
    391 logging.info('Gin Configuration to restore model:\n%s', gin.config_str())
    393 print(gin.query_parameter("%VOCABULARY"), gin.config_str())
--> 394 vocabulary = gin.query_parameter('%VOCABULARY').scoped_configurable_fn()
    395 try:
    396   vocab_index_from = gin.query_parameter('%VOCAB_INDEX_FROM')

File ~/miniconda3/envs/optformer/lib/python3.10/site-packages/gin/config.py:1605, in _make_gin_wrapper.<locals>.gin_wrapper(*args, **kwargs)
   1603 scope_info = " in scope '{}'".format(scope_str) if scope_str else ''
   1604 err_str = err_str.format(name, fn_or_cls, scope_info)
-> 1605 utils.augment_exception_message_and_reraise(e, err_str)

File ~/miniconda3/envs/optformer/lib/python3.10/site-packages/gin/utils.py:41, in augment_exception_message_and_reraise(exception, message)
     39 proxy = ExceptionProxy()
     40 ExceptionProxy.__qualname__ = type(exception).__qualname__
---> 41 raise proxy.with_traceback(exception.__traceback__) from None

File ~/miniconda3/envs/optformer/lib/python3.10/site-packages/gin/config.py:1582, in _make_gin_wrapper.<locals>.gin_wrapper(*args, **kwargs)
   1579 new_kwargs.update(kwargs)
   1581 try:
-> 1582   return fn(*new_args, **new_kwargs)
   1583 except Exception as e:  # pylint: disable=broad-except
   1584   err_str = ''

File ~/code/llm/optformer/optformer/data/tasks.py:44, in get_vocabulary(sentencepiece_model_file, max_integer_tokens, expected_vocab_size)
     40 extra_tokens = ['<' + str(n) + '>' for n in range(max_integer_tokens)]
     41 vocabulary = vocabularies.SentencePieceVocabularyWithCustomToken(
     42     sentencepiece_model_file, extra_tokens=extra_tokens)
     43 if (expected_vocab_size is not None and
---> 44     expected_vocab_size != vocabulary.vocab_size):
     45   raise ValueError(f'Vocabulary size ({vocabulary.vocab_size}) does not '
     46                    f'match the expected value ({expected_vocab_size}).')
     47 return vocabulary

File ~/miniconda3/envs/optformer/lib/python3.10/site-packages/seqio/vocabularies.py:435, in SentencePieceVocabulary.vocab_size(self)
    433 @property
    434 def vocab_size(self):
--> 435   return self._base_vocab_size

File ~/miniconda3/envs/optformer/lib/python3.10/site-packages/seqio/vocabularies.py:444, in SentencePieceVocabulary._base_vocab_size(self)
    437 @property
    438 def _base_vocab_size(self):
    439   """Number of ids (including 0=PAD, 1=EOS, and 2=UNK).
    440 
    441   Returns:
    442     an integer, the vocabulary size
    443   """
--> 444   return self.tokenizer.GetPieceSize()

File ~/miniconda3/envs/optformer/lib/python3.10/site-packages/seqio/vocabularies.py:426, in SentencePieceVocabulary.tokenizer(self)
    423 @property
    424 def tokenizer(self) -> sentencepiece_processor.SentencePieceProcessor:
    425   """Returns the Python tokenizer."""
--> 426   return self._model_context().tokenizer

File ~/miniconda3/envs/optformer/lib/python3.10/site-packages/seqio/vocabularies.py:336, in SentencePieceVocabulary._model_context(self)
    328   return self._model
    330 normalizer_spec_overrides_serialized = (
    331     self._normalizer_spec_overrides.SerializeToString(deterministic=True)
    332     if self._normalizer_spec_overrides
    333     else None
    334 )
--> 336 self._model = self._load_model(
    337     self._sentencepiece_model_file,
    338     self._extra_ids,
    339     normalizer_spec_overrides_serialized,
    340     self._reverse_extra_ids,
    341 )
    342 return self._model

TypeError: SentencePieceVocabularyWithCustomToken._load_model() takes 1 positional argument but 5 were given
  In call to configurable 'get_vocabulary' (<function get_vocabulary at 0x7fce61d801f0>)

In addition, I encountered another unrelated issue on the next cell: AttributeError: module 'vizier.benchmarks' has no attribute 'NumpyExperimenter' .

Here are the versions I'm currently using:

google-vizier==0.1.6
seqio @ git+https://github.com/google/seqio@4d3097973e9e24ec2963319ec3c5ff518811060f
seqio-nightly==0.0.15.dev20230703

Thank you in advance.

[yutianc-dm]

Thanks for pointing out the problem, Herilalaina. I have committed a change to fix the dependency versions based on your pull request.