sci-ml/strata::ruizhi-overlay
- Ebuilds: 1, Testing: 0.2.9 Description:
Strata is an inference engine for Qwen3.8-Flash-Next on consumer NVIDIA and
AMD hardware, with an OpenAI/Anthropic-compatible API served on localhost
and a terminal client that picks, verifies and starts the engine.
The engine runs that one model family: its catalogue lists Qwen3.8-Flash-Next
in several GSQ-RCO, Q2_0 and UD quantizations, and the terminal refuses other
GGUF files, including other Qwen3.8 sizes.
The engine is built for one GPU backend (CUDA or ROCm HIP) and installed as
an engine component that the terminal verifies against a SHA-256 manifest at
every start.
Homepage:https://github.com/Niko1221/Strata License: MIT