Custom GGUF quant for :
failspy/llama-3-70B-Instruct-abliterated-GGUF

Optimal (8k ctx) for 36GB VRAM with an IGP displaying the OS.