kcaverly
kcaverly
That is correct, we did our best but its messy. Without rewriting this PR from scratch, I dont think we can make it cleaner. The repo went through so many...
I took a look at this, and managed to recreate it with a relatively small subset. ```python import dspy class Notes(dspy.Signature): """Bulleted Markdown Notes, Headings and Terms in Bold""" document...
A separate AzureOpenAI class is much cleaner. What I can do is revert the GPT3 changes, pull azure out so GPT3 is exclusively for OpenAI and introduce a new AzureOpenAI...
I split out the Azure functionality into its own class. I agree that it makes more sense to abstract over LMs to some sort of external package. In the meantime,...
Thanks @insop for the review. I have updated the docs/and a few of the comments above to ensure consistency. If you have access to the AzureOpenAI I would appreciate you...
Thanks @bclavie for giving it a test, adjusted both the model_type and typo fix.
Thats a great catch @okhat, I've updated the imports to just pull AzureOpenAI, so none of the cache functions should be available beyond direct importing from the file.
Hey all, is there anything outstanding on this being merged?
One thought on Token Counting - would it make sense to build this directly into the LM abstraction? I imagine, while there may be some overlap in tokenization model to...
If the tokens used is non-determistic based on the optimization, could it provide value if we simply collected a broad sample of the number of optimization calls, and use it...