search

containers / podman-desktop-extension-ai-lab

Work with LLMs on a local environment using containers
https://podman-desktop.io/extensions/ai-lab
Apache License 2.0
182 stars 41 forks source link

Chat Bot AI app with lmstudio-community/granite-3.0 reiterates when input with response in the console output mutliple times #2065

Closed odockal closed 2 weeks ago

odockal commented 2 weeks ago

Bug description

When sending an input into a prompt via browser, first input is processed correctly, but second question/input is kinda copied into answer prompt and seems like resend and appears iteratively in the output again and again. more questions make it worse.

Ref: https://github.com/containers/podman-desktop-extension-ai-lab/pull/2024

Operating system

Mac OS 14

Installation Method

None

Version

next (development version)

Steps to reproduce

  1. Install 1.3.2 AI Lab
  2. Create podman Machine (Mac OS libkrun)
  3. Start up chat bot app with mentioned model
  4. Assert: app is up and running
  5. In the App's details -> open the browser
  6. Send an input and check for the output
  7. Repeat step 6.
  8. Assert: LLM model answers as expected Actual result: app/model responses multiple times like it is recycling the input over and over

Relevant log output

No response

Additional context

No response

jeffmaury commented 2 weeks ago

Image Cannot reproduce on Win11Pro. Tested with the 2 following prompts:

  1. What is Podman Desktop ?
  2. How does it compare to Docker Desktop ?

Will test on MacOS/libkrun