Don't bargain on that. For some the threshold is way too high given how far they've sunk. As always, the ones most vocal about it haven't even bothered trying

Don't bargain on that. For some the threshold is way too high given how far they've sunk. As always, the ones most vocal about it haven't even bothered trying out guanaco 70b 2bit. Probably don't know what ggml, gptq, cformers or any of the others are. They still base their assumptions on the openai gold standard.