My 8GB GPU shouldn't run flagship local LLMs, but this workflow makes it work anyway

My 8GB GPU shouldn't run flagship local LLMs, but this workflow makes it work anyway