mradermacher

/

WizardLM-2-8x22B-GGUF

like 4

There was a bug in llama.cpp that caused some quality degradation with imatrix quants of MoEs, and I decided to redo some of them. This model is in the queue, but it can take a while till it is redone, since mistral 8x22 models are especially hard to quantize. I've bumped its priority slightly so it will be the next 8x22 (otherwise there would be five others first :)