Skip to content

Latest commit

 

History

History
30 lines (23 loc) · 848 Bytes

RELEASE.md

File metadata and controls

30 lines (23 loc) · 848 Bytes

v0.2.2

  • Handle maximum batch size for embed_text.
  • Restore types.ModelNameOptions.
  • Update tuning to handle more dataset types (json, pandas, urls, google-sheets).
  • Expose count_text_tokens
  • Expose operations management.
  • Allow users to pass custom metadata (HTML headers).
  • Add dataclass prettyprinting.

v0.2.1

  • Bugfix: Add missing enum names to saftey_types.py
  • Update to google.ai.generativelanguage v0.3.3

v0.2

  • More flexible safety settings:
    • accept strings, ints for enums.
    • accept a {category:threshold} dict in addition to the list of dicts format ([{"category": category, "threshold": threshold}, ...]).
  • Add support for batched embeddings.
  • Add support for tuning:
    • Add get_{base,tuned}_model.
    • Add list_tuned_models.
    • Add create_tuned_model.

v0.1

Initial version