llama-70b on the way?
#1
by
GF110
- opened
Only just came into knowledge of multi-modal models and am curious about how these are being implemented.
Are there any Llava projector models for llama-70b / Do these need to be created by yourself for each architecture for specific use with Koboldcpp?
@GF110 no these are just quantized gguf projectors of existing llava models.
I do not believe there is any 70b llava model so kobold cpp can not quantize a projector for a 70b llava model.