09-15-2026, 08:03 PM
Quote:What exact setup are you running? What model, what runtime, and roughly what context size?
I'd approach it in a boring order. Before rebuilding anything, I'd test one clean baseline with Mixtral and compare from there. Not glamorous advice, but repeatable baselines save a lot of time.
2x3090 • llama.cpp • too much VRAM is still not enough