Fantastic work! Any chance of a 120b version?
#7
by
MarkR1907
- opened
I can't believe how well this model performs on its every-day tasks considering what you've done to it! Do you plan to give the 120b model the same treatment? (please please please).
Great to hear it is useful for you. But 120b is not prioritized. Sorry about that.
I'd second the request for the 120b model. The 120b model maintains coherence far better in a lot of common-refusal domains (mostly narrative/writing assistant tasks). The 20b just doesn't stack up for longer context detail retention.