1 items across the graph — tagged with Htop.
The htop for LLM inference see exactly where every GB of VRAM goes and get measured quantization savings.
