Psydestroyer 20B GGUFs

GGUF Quants of my Psydestroyer-20B merge.

Currently I only have a Q4-K-M, as that is the highest (I've tested) quant of a 20B frankenmerge that can mostly fit on my 3060 12GB and run at decent t/s (~8-9). If the demand is there I can create more, otherwise you'll probably have to download and quantize the model yourself.

Downloads last month
7
GGUF
Model size
20B params
Architecture
llama

4-bit

Inference API
Unable to determine this model's library. Check the docs .