Mark Ericksen

Hi @jfayad! That sounds like a bug or incompatibility with our OpenAI module and the Scaleway API. I don't have any experience with that service. For debugging something like this,...

Hi @ci! Ah, interesting. Yes, there is handling for the `stop_reason` `max_tokens`, but I think this surfaced an issue when the it was return during a `tool_use` message from the...

I encountered this recently and have pushed up a commit on the main branch, which is after 0.4.0-rc.0. It will be officially released in the next RC or final version.

This sounds like a bug. I don't regularly use `ChatGoogleAI`. Is anyone else able to jump in who is using it?

I need to provide more documentation for working with v0.4.0-rc.1, but changes were made to the Utils module with how streams of deltas are processed. Part of the goal was...

With v0.4.0-rc.1, the way deltas get merged was changed to merge them in batches, because that's how they can be received sometimes. I've tested and verified it with ChatOpenAI and...

Hi @nallwhy! Yes, I'm in the processes making this change. I'me changing both MessageDelta.content and Message.content. MessageContent will always be a list of ContentPart structs. This is a huge breaking...

This change to have Message.content be a list of ContentParts is done and should be working correctly for ChatOpenAI in 0.4.0-rc.0. The change is to make all Chat models use...

Hi @vlymar! Thanks for your thoughtful write-up. Here's how I'm handling a similar situation in my own application. - I have a set tables to store conversation history in my...

Good question. I don't write the partial message to the database as it's received. It would cause high DB churn/updates for no real benefit. I don't write it the DB...