model.safetensors.index.json
#2
by
throwaway64878453747846
- opened
Could it be that model.safetensors.index.json has not been correctly made/updated/generated/whatever for this model?
It refers to 29 safetensors when there are only three.
exl2 does not use this file to load the model. The 29 file shards are from the original fp16 model. When exl2 quantizes the model, it does not regenerate the file as it does not need it to work.