Repository navigation
Replies: 3 comments 5 replies
|
2 replies
|
Following are some examples of current messages sent vs future ones with this change -
@lyjeileen yes, I agree, it would be simpler to define two types for the content with clear indication of the fields. |
3 replies
|
We decided to keep the Multipart and introduced a separate component for ChatCompletion. The changes discussed were released in v.0.0.34 |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Most LLM Agent libraries and services now conform to a standard interface similar to OpenAI's ChatCompletionRequest. It would make ui-components library easier to use and adopt if the input components send messages in this format instead of custom formats such as
textandfilesWithText.OpenAI only documents using image and audio for content types of ChatCompletionRequest but we could make it more generic to support any file type.
For example, the interface could be something like,
and some sample corresponding messages sent on the websocket could be
{ "id": "4f6f6c72-4328-4170-8d35-7fabe322be3e", "timestamp": "2024-11-29T18:03:25.120Z", "sender": { "id": "46ec2629", "name": "Anonymous" }, "conversationId": "7fabe322be3e", "format": "ChatCompletionRequest", "data": { "messages": [ { "content": [ { "type": "text", "text": "what was the monthly average sales reported for the last quarter?" }, { "type": "file", "url": "https://somefile.xls" } ], "role": "user" } ] } }Pros
ui-componentslibrary with LLM AgentsCons
All reactions