AI Functions release of the Transformers Extension(v 4.0.0)

Product News
, ,
1

We released version 4.0.0 of the Transformers Extension on Github and Pypi.
The Exasol Transformers Extension allows you to use pre-trained machine learning models from :hugging_face: Hugging Face directly in your Exasol instance.
It lets you install the models via the transformers-api directly into Exasol’s filesystem BucketFS, and use them on your data using provided UDF’s.

:memo: Summary

This version introduces our new AI Functions, namely the new UDFs “AI_SENTIMENT”, “AI_CLASSIFY” and “AI_EXTRACT_ENTITIES”.
These new UDFs are designed to be easy to use, only requiring the text data to be used as input.

For users who need more control over the configuration of their prediction UDFs, all the existing UDFs are still available, but have been renamed to fit our new naming scheme.

To make the new UDFs possible, we needed to make quite a few changes under the hood. So please read the breaking changes carefully before updating to this version.
For more information, also take a look at the updated documentation.

:warning: BREAKING CHANGES:

  • The max_length parameter has been renamed to max_new_tokens, and its behavior changed. Both of these changes were done in accordance with changes in the transformers library.
  • All prediction UDFs have been renamed:
| Old UDF Name                               | New UDF Name                |
|--------------------------------------------|-----------------------------|
| TE_FILLING_MASK_UDF                        |    AI_FILL_MASK_EXTENDED    |
| TE_QUESTION_ANSWERING_UDF                  |    AI_ANSWER_EXTENDED       |
| TE_SEQUENCE_CLASSIFICATION_SINGLE_TEXT_UDF | AI_CUSTOM_CLASSIFY_EXTENDED |
| TE_SEQUENCE_CLASSIFICATION_TEXT_PAIR_UDF   | AI_ENTAILMENT_EXTENDED      |
| TE_TEXT_GENERATION_UDF                     |    AI_COMPLETE_EXTENDED     |
| TE_TRANSLATION_UDF                         |    AI_TRANSLATE_EXTENDED    |
| TE_TOKEN_CLASSIFICATION_UDF                |    AI_EXTRACT_EXTENDED      |
| TE_ZERO_SHOT_CLASSIFICATION_UDF            | AI_CLASSIFY_EXTENDED        |
  • task_type handling has been changed.
    The Transformers extension now allows only specific transformers task types in the installation and execution of models. You may need to re-install your models from HuggingFace using the new task_types in order to use them.
    Models installed with legacy task_types can still be listed and deleted using the respective UDFs.
  Allowed task_types are:
         "fill-mask" (previously "filling_mask"),
         "translation",
         "zero-shot-classification",
         "text-classification" (previously "sequence_classification"),
         "question-answering",
         "text-generation",
         "token-classification"
2 Likes