Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[BUG] AI Personas Creator fails #328

Closed
1 task
younes-io opened this issue Jan 6, 2024 · 11 comments
Closed
1 task

[BUG] AI Personas Creator fails #328

younes-io opened this issue Jan 6, 2024 · 11 comments
Assignees
Labels
type: bug Something isn't working
Milestone

Comments

@younes-io
Copy link

younes-io commented Jan 6, 2024

Description

First, I wanted to create a persona using GPT4-Turbo and these are the issues I found:

1 - in the first API call made to big-AGI server, the function name is "undefined" and its response is:

{
    "result": {
        "data": {
            "json": {
                "role": "assistant",
                "content": "I apologize for any misunderstanding, but it seems there was no transcript provided for analysis. Could you please provide the specific transcript or details that you would like me to analyze? Once you do, I will be able to perform a comprehensive review and provide you with an in-depth written analysis of the speaker's characteristics and attributes as requested.",
                "finish_reason": "stop"
            }
        }
    }
}

image

The second API call fails with :

An error occurred with your deployment

EDGE_FUNCTION_INVOCATION_TIMEOUT

I re-run those same requests on Postman, and I reproduced the same issue(s) on the browser.

The browser's console have these errors:

image

Device and browser

Windows 11
Version 1.60.118 Chromium: 119.0.6045.163 (Official Build) (64-bit)

Screenshots and more

No response

Willingness to Contribute

  • 🙋‍♂️ Yes, I would like to contribute a fix.
@younes-io younes-io added the type: bug Something isn't working label Jan 6, 2024
@enricoros
Copy link
Owner

enricoros commented Jan 6, 2024

Thanks for the details - probably a duplicate of #316 .
Have a solution in mind, but will come right after the weekend.

@enricoros enricoros added this to the 1.11.0 milestone Jan 6, 2024
@enricoros enricoros self-assigned this Jan 6, 2024
@enricoros
Copy link
Owner

Fixed with e2e7ea9 - streaming chat generation.

@younes-io: can you verify that this is fully fixed now?

@younes-io
Copy link
Author

younes-io commented Jan 9, 2024

@enricoros Not sure if you deployed yet, but I tried many times, still got the same issue:

image

@enricoros
Copy link
Owner

@enricoros Not sure if you deployed yet, but I tried many times, still got the same issue:

It's on the main branch, not in the Stable release yet (either tomorrow or in the weekend). The main branch is for development, and available on next.big-agi.com, for trying out.

@younes-io
Copy link
Author

Oh I didn't know that endpoint exists, thank you !

@enricoros
Copy link
Owner

Update: since this fix is pretty important, it's being deployed on the stable branch. Will be available in a few minutes.

@enricoros enricoros mentioned this issue Jan 16, 2024
23 tasks
@dogmatic69
Copy link
Contributor

Does not seem to be fixed when using the latest docker image.

@dogmatic69
Copy link
Contributor

dogmatic69 commented Jan 19, 2024

Just built the image locally from the current main branch and still does not work.
image
image

$ git show
commit 770fbdef7235521b71554eeeaaf54508baa10bf3 (HEAD -> main, origin/main, origin/HEAD)

@enricoros
Copy link
Owner

enricoros commented Jan 19, 2024

It is working.. one can tell as AI answers, but it's the input that is not valid. Did you use from YouTube or from Text? And what's the input?

@dogmatic69
Copy link
Contributor

dogmatic69 commented Jan 19, 2024

text, I just used a similar prompt as is done in custom GPTs.

Not sure can class it as working when it says "there is no transcript so I'll make something up (and ignore the actual input)"

@enricoros
Copy link
Owner

This bug was opened due to a specific timeout and issue, which was resolved by switching to streaming generation, so the right place to discuss this is in a new bug, not piling up here.

Note that this Persona Creator is about replication of a persona from a YouTube Video or sequences of text (e.g. Tweets) - and not about creating one from an instruction (that will be a separate Tab on top, in the future).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
type: bug Something isn't working
Projects
None yet
Development

No branches or pull requests

3 participants