Prasad Chalasani
Prasad Chalasani
@rachittshah sure go ahead , thanks. I can elaborate a bit more what I had in mind later when I have time.
We can start with linking classes and method names, then do more granular.
> Not sure what the benefit was in completely removing existing, core functionality but it did a number to our application.. Indeed, the new release completely breaks the demo examples...
It looks like step nesting has been fixed in #1052 , but sequencing is still broken. E.g for the above script, I get correct nesting but wrong sequence of outputs:
Thanks @kartikhans please add suitable tests for this. There are existing ones for token usage, you can search for it. - I will review once there are tests
closing as already implemented
@AbhishekRP2002 we recently added support for using Gemini models via a single GEMINI_API_KEY but maybe the vertexAI support offers more, I haven't had time to dig in. If you think...
BTW, if you still want to add the metaphor -> exa change or the weaviate support, those are still open.
We can have users specify `chat_model=vertexai/google/gemini-2.0-flash-001` for example, and it will activate the `vertexai` path in `openai_gpt.py`.
> I believe Milvus is better for large-scale deployment. We'll add it at some point, this hasn't come up before. Also happy to take a PR. We do have PRs...