Python project for Resonance Stream, which generate output for translator model, based from Qwen 3 1.7B
- Python 3.13
- Windows OS (Linux not tested yet)
- CUDA Toolkit 2.6
- Make sure model_f16_clean config.json "architectures": "Qwen3ForCausalLM"
- git clone --recursive https://github.com/ggerganov/llama.cpp
- cmake -B build
- cmake --build build --config Release -j
- python llama.cpp/convert_hf_to_gguf.py model_f16_clean --outfile model.f16.gguf
- llama.cpp/build/bin/Release/llama-quantize.exe .\model.f16.gguf model_q4_k_m.gguf q4_k_m