OpenAI API library/Release notes for v0.7

< OpenAI API library
Revision as of 18:21, 9 February 2024 by Lchrisman (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Since OpenAI API library v0.6:

  • Analytica 6.4 is now required.
    Basically, since Analytica 6.4 is the current release, we aren't testing for continued compatibility, and may have used some efficiency improvements enabled by Analytica 6.4 features and fixes.
  • Open AI recently released two new embedding models in 5 different sizes, all of which they say outperform the previous text-embedding-Ada-002 embedding model. v0.7 has been extended to enable you to use any of these for text embedding.
  • There is a new input panel in the Text completions module that lets you configure your default embedding model and size.
  • Added a «seed» parameter to Prompt_completion and Chat_completion, to enable nearly deterministic output. This is also a feature recently added by OpenAI to their API.
  • Added Text_to_speech. This works on desktop Analytica. Not yet on ACP.
  • Fixed an annoyance in which a choice using the index of model names would get off (wouldn't splice) when OpenAI added a new model to the list. Restructured that index slightly (there is a button Update model name list that you press when you want to resynch which the latest, which avoids the annoyance.
Comments


You are not allowed to post comments.