Thank you!
#1
by
CulturedMan
- opened
Thanks for the 3.0bpw! I am currently stuck with 12gb vram, and I'm able to run this with 6144 context / alpha 1.75 with minimal spillover to the CPU.
Thanks for the feedback. Picking quant sizes can be tricky with all of the different model sizes (from "interesting" merges) as well as GPU VRAM sizes.