Files empty?
Not trying to jump the proverbial gun here...but are you guys still working on the quants for this one? The file page is currently empty.
And if beggars can be choosers...6.5 bpw pl0x?
Many thanks.
we do not plan to add more quants as getting even these out took us a lot of time, we have however published the measurement file in the main branch so people can re-quant and skip the biggest step.
No worries. I just downloaded Mradermacher's Q6_K quant of this model—and did a real quick test run. I can already tell there's something unique (in a good way) about Magnum V2 123b for my use case. I did not have to sit there and give it a long-winded lecture or 5 billion notes. I just gave it a quick prompt (half-assed to be honest...I'm tired and need to go to sleep) and the model surpassed my expectations and caused me to raise an eyebrow at the output.
This just might be my new go-to LLM that will replace Llama 3.1 70b Instruct Lorablated. In fact, for shits and giggles, I just gave Claude3 (paid subscription) the same exact prompt I gave Magnum...and Claude3 kinda sorta went out in left field (unusable garbage) whereas Magnum's responses aligned pretty well to what I was looking for.
I think I might cancel my Claude3 subscription and use Magnum instead.
Thank you guys for all your hard work!