AMD contributes their GPU support to tiny-vLLM

**Brief Summary (2‑3 sentences)** AMD has open‑sourced its GPU‑accelerated backend for the tiny‑vLLM inference engine, enabling developers to run large language