Show HN: QKV Core – Run 7B LLMs on 4GB VRAM via surgical memory alignment

(github.com) by broxytr | Dec 16, 2025 | 0 comments on HN