Config file?
#4
by
Culturedniichan
- opened
When I use ggufs I want to use llamacpp_HF because otherwise I never can get the samplers to be used or have any effect on ooba at least, so for that ooba needs the url of the unquantized model to download the config files, tokenizer, etc. Will you publish the full model?
I can unprivate it for a minute. Can you load them locally in ooba? Or do you need the link live at all times? If downloading it is an option, join the discord so we can coordinate
What ooba does is just create a folder and put the gguf and the config files together. Personally I can do that just fine, don't know about other people