ParentFull threadwinstonprivacy·How much VRAM does that model need? I'm looking for a 30B byte that I can train a LORA with on an GTX 4090.View on HN