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

How to serve a local chat API on a machine with no network with Unbiased Strata

Build the engines and run an OpenAI-style chat endpoint on one NVIDIA GPU with no downloads at build or start.

Who it is forA developer or ops engineer with an NVIDIA Linux box and a GGUF model who needs a private chat API that works offline.

The problemInstallers and runtime downloads make local model servers hard to trust and hard to run on isolated machines.

The outcomeA running unbiased-strata process on 127.0.0.1 that answers /v1/chat/completions from the local model.

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. Run git clone https://github.com/robertkarl/UnbiasedStrata then cd UnbiasedStrata. (inferred, check the README)
  2. Run git submodule update --init to fetch the pinned llama.cpp fork.
  3. Run make build to produce build/bin/strata-poly, build/bin/strata, libstrata_kq.so and unbiased-strata.
  4. Run make test to check the front end against a fake engine, with no GPU or model needed.
  5. Start the server with build/bin/unbiased-strata --engine build/bin/strata-poly -m /models/Qwen3.6-35B-A3B-UD-Q3_K_M.gguf --host 127.0.0.1 --port 18091 --alias qwen3.6-35b-a3b-ud-q3_k_m -- --model /models/Qwen3.6-35B-A3B-UD-Q3_K_M.gguf --max-context 32768 --expert-profile /models/expert-profile-q3km.txt --tail-skip 6. (inferred, check the README)

Prompt for your agent

In the UnbiasedStrata repo at <repo-path>, run make build and then make test, and report whether the test suite passed. Then start unbiased-strata with the model at <model-path> on port 18091 and confirm that GET /health returns a success response.

Caveat. The README's model and expert-profile paths are examples, so you must supply your own GGUF files, and the target is one NVIDIA GPU on Linux x86-64 only.

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 The use cases, steps, and prompts are written by AI and may be incomplete, outdated, or just wrong.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