This repository is a deployment and quantization package for nvidia/canary-qwen-2.5b. Base model attribution: - nvidia/canary-qwen-2.5b - Original model and licensing terms remain with NVIDIA and the upstream model authors. Local integration: - OpenAI-compatible STT endpoint - Open WebUI compatibility - ffmpeg WebM/Opus normalization - Optimum Quanto qint8 LLM-component workflow