CUDA on InkTrace｜RxChi1dhttps://inktrace.rxchi1d.me/en/tags/cuda/Recent content in CUDA on InkTrace｜RxChi1dHugo -- gohugo.ioenrchildlin@gmail.com (RxChi1d)rchildlin@gmail.com (RxChi1d)© 2026 RxChi1dSat, 11 Jul 2026 02:06:20 +0800Porting MTP to Ornith-1.0-35B: From GGUF Surgery to Hermes Agenthttps://inktrace.rxchi1d.me/en/posts/ai-technical/ornith-1-0-35b-mtp-deployment-notes/Mon, 06 Jul 2026 21:15:09 +0800rchildlin@gmail.com (RxChi1d)https://inktrace.rxchi1d.me/en/posts/ai-technical/ornith-1-0-35b-mtp-deployment-notes/<p>This guide walks through deploying Ornith-1.0-35B locally with llama.cpp, wiring it into Hermes Agent, and finally grafting on an MTP head to push inference speed even higher.</p>