help me understand the point of such scaling
#1
by
Samvanity
- opened
Hi,
if it performs worse than the original, what's the point of scaling it up to 13B? for further finetuning?
@Samvanity Yes the point is for further finetuning. And although this iteration performs worse than the original, version 2 has already been created using a new method which should fix that problem. This new version is being uploaded as we speak. Once it is done we will make the repo public.
rombodawg
changed discussion status to
closed