This is my favorite model.
#1
by
lizelive
- opened
Please train one using UL2 as base! That would be amazing!
Wow, the fastchat model is so fast! Only 8gb GPU at the moment so kinda crashed with out of memory after 2 questions.. Getting a K80 to play with.. See if I can Parallel it to run with my RTX card to give a much needed memory boost... Good job on the model. Look forward to what's next.
don't waste money on a k80, CPU inference is probably faster at this point (using optimum)
Please train one using UL2 as base! That would be amazing!
can you please elaborate and explain more or any like for training steps? wold be great!
high performance?