GGML variant of WizardLM-30b-V1.0, for use on 24GB cards such as 3090. | |
Update pushed 6/14 that resolved garbage output on llama.cpp, not tested using other tools. | |
Requires a recent build of llama.cpp that supports the K-quant methods (June or later). | |
Quant was prepared using llama.cpp build on 6/14/2023. |