Pamela Fox

This now works for me after incorporating my suggestions.

I set the key using the key feature here: https://llm.datasette.io/en/stable/setup.html#api-keys Since I used the key name of "azure", then yes, my api_key_name was "azure", but in theory you might set...

I actually have this branch locally but wasn't sure if it was desired, as I read it might introduce a slight performance hit. @tonybaloney Is it worth adding?

Example pytest plugin setup.py: https://github.com/tonybaloney/pytest-azurepipelines/blob/master/setup.py

I just ran into this! I didn't fix it myself since I found your TODO, but here's a test for it (based off the CPython test): ``` def test_stream_error_handling(): class...

I believe this error is expected, sadly, as the Dockerfile is still using amd64 emulation mode (qemu) in order to bring azure-functions-core-tools. We're still waiting for ARM64 support for Python...

Sorry, I still haven't figured out a devcontainer setup that will work on an M1. I've been running it in either Codespaces or locally. I've asked the Functions and azd...

I have a branch that at least breaks out the prompts. Agreed that its currently not well split up. Thanks for the feedback!

cc @mattgotteiner It feels like it could be an issue with the ID token expiring and needing a refresh, but thats speculation based on the issue description.

Are you all trying to use gpt-4o for the vision feature, or for just non-image answers? The configuration would be different depending.