LLM Templates - Created vision_gpt_explain_image llm application #964
Conversation
There was a problem hiding this comment.
Thanks a lot for the PR @aybruhm !
I think there is one small confusion I see in the last examples and this one. The difference between configuration and input. For instance, in this case, the user is trying to create an llm application that compares different images. The configuration are the parameters they would iterate on in the playground to get a reliable app (let's say for comparing logo), these would include the temperature, max_tokens.. (https://platform.openai.com/docs/api-reference/chat/create), and models in case there are multiple options (in this case, I think there is only one model, so no need to include), and most importantly the prompt, since that is the main thing that they would play around with. On the other hand, the inputs for the application is the images (and not the prompt).
So first change, the system and user prompt are configuration. The images only are the inputs. Second change, we agreed yesterday that for now we will have the images in simple application and not in chat application, so no need for messages as an input.
Hope it's clear./
aybruhm
changed the base branch from
main
to
feat/file-input-support-playground
I believe it's because the PR's base is |
aybruhm
changed the base branch from
feat/file-input-support-playground
to
main
Check again. I changed the PR's base to cc: @mmabrouk |
|
@aybruhm prettier is failing. Please make sure to use the same version we use in the workflow, I think that is the reason for the error |
|
Weird, this is what I got: |
|
I ran the exact command the "check web code" action workflow uses on my local machine and I got this:
|
|
@bekossy here also, thank you! |
|
@MohammedMaaz -> template for #944 |
Description
This PR introduces a new LLM application that utilizes Vision GPT to provide explanations for image(s).
Related Issue
Closes #953