# transcribe-quantize: re-quant a transcribe.cpp GGUF using ggml's # reference quantizers. Mirrors whisper.cpp's `quantize` and # llama.cpp's `llama-quantize`. Links libggml directly so it has access # to ggml_quantize_chunk and the gguf C API; does NOT link against # libtranscribe (the tool operates on raw GGUF, not via the loader). add_executable(transcribe-quantize main.cpp policy.cpp ) target_link_libraries(transcribe-quantize PRIVATE ggml ) transcribe_apply_warnings(transcribe-quantize)