DEDUPED Hot AI repos, turned into use cases Get 5 a day

How to ask a local model questions about screenshots with Unbiased Strata

Attach a projector file so the local server can read pictures such as dashboards, without sending them to a cloud API.

Who it is forAn analyst or engineer who has dashboard screenshots they cannot upload to a hosted service.

The problemReading images with a hosted model means sending the pictures off the machine.

The outcomeThe local server accepts picture input and answers questions about the image, with encoding done on the same machine.

What you need first

  • Linux x86-64 machine with one NVIDIA GPU (Pascal or newer) and CUDA 12.x
  • git and a C++ build toolchain able to run make build
  • GGUF model files copied to the machine by hand, since the fork does not download them
  • Optional: the mmproj projector file from the same repository as the model, for picture input

Steps

  1. Build and start the server as in the chat API setup, then add --mmproj /models/mmproj-F16.gguf and --threads N before the -- separator.
  2. Keep the default picture size of 300 rows, or change it with --image-max-tokens.
  3. Send a chat request to http://127.0.0.1:18091/v1/chat/completions with the screenshot attached as an image part, using the model alias you set with --alias. (inferred, check the README)
  4. Check the answer against the screenshot yourself before relying on it.

Prompt for your agent

In the UnbiasedStrata repo at <repo-path>, start unbiased-strata with the projector at <mmproj-path> and report the server log line that confirms the projector loaded. Then send the image at <image-path> to /v1/chat/completions and report the reply.

Caveat. The README reports one 1920x1080 dashboard read in about 15 seconds of encoding on a Ryzen 7 1700, so timing will differ on other hardware, and the projector must match the model.

Share: X LinkedIn Hacker News Email Copy link

Disclaimer

Deduped lists open-source repositories that belong to other people. We do not own, maintain, audit, or vouch for any of them, and a repo appearing here is not an endorsement or a safety check. The use cases, steps, and prompts are written by AI and may be incomplete, outdated, or just wrong. Where a use case is marked "Seen in the wild", we link to public posts where someone described something similar, but we have not verified those posts and the use case itself is still our AI's own write-up. Use all of this at your own risk. Only run code, commands, or agents on systems and accounts you are authorized to use, and read the code and every command before you run it, especially anything that touches credentials, files, or the network. Nothing here is professional, security, legal, or financial advice. All trademarks and project names belong to their respective owners, and their use here does not imply affiliation. Content is provided as is, without warranties of any kind. Found a mistake, a safety problem, or something you want removed? Email [email protected] and we will look at it promptly. Full disclaimer page