mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-09-30 01:47:39 -05:00
Drop the duplicated UD- prefix and quant bit-depth regexes from the download options utils: quantBitDepth re-implemented, more crudely, what HuggingFaceService.getBitDepth already does, and the label helper re-declared the weight-extension pattern held by MODEL_ID. The command preview now derives its bit depth via the service, so the quant chips and the command sorts agree. Assisted-by: llama-ui:Qwen3.8-Flash-Next