Thank you
#3
by
xxx777xxxASD
- opened
These models you and your team are doing are just awesome, nice job
Do you plan fine-tuning something bigger? Maybe an llama 3 based MoE so most people would be able to run it with decent speed?
Would love to see one, horny and smart after some ~magic
Sure, once we're more confident with our output. We want to refine our process with the smaller models first before making a big commitment.
Feedback is slow as a lot of users (understandably) keep their ERP experiences to themselves. We've been doing all of this blindfolded.
But we've definitely come a long way from being a lobotomization factory, and I can see Coomand R, 3Some 70B, and Midnight Moist in the horizon!
Edit: Thank you for the kind words, btw