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.

Sign up for GitHub

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

Jump to bottom

Consider using a VLM server #3

Open
geoffroy-noel-ddh opened this issue Oct 29, 2024 · 1 comment
Open

Consider using a VLM server #3

geoffroy-noel-ddh opened this issue Oct 29, 2024 · 1 comment
Assignees
@geoffroy-noel-ddh
Labels
enhancement New feature or request

Comments

@geoffroy-noel-ddh
Copy link
Member

geoffroy-noel-ddh commented Oct 29, 2024 •
edited
Loading

Use a model server like ollama or vllm.

Two advantages:

  1. it works as an abstraction for a lot of models, so we don't have to write a describer class for each type of model
  2. speed up prompt engineering phase by keeping the model in memory between executions of bvqa
@geoffroy-noel-ddh geoffroy-noel-ddh added the enhancement New feature or request label Oct 29, 2024
@geoffroy-noel-ddh geoffroy-noel-ddh self-assigned this Oct 29, 2024
@geoffroy-noel-ddh
Copy link
Member Author

geoffroy-noel-ddh commented Nov 16, 2024 •
edited
Loading

See this list of examples. Lots of vlms supported. But prompt is idiosyncratic. However we could call the functions from that module to unify the interface.Wondering why VLLM doesn't have that.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@geoffroy-noel-ddh geoffroy-noel-ddh

Labels
enhancement New feature or request
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
1 participant
@geoffroy-noel-ddh