solid little speed demon
#1
by
bobig
- opened
So glad you are quanting LING.
Your qx4-hi-mlx is same speed as a Q6, so I know you stuffed a LOT of extra brains in it.
Hope you Like Ling-Flash. Fast smart direct answers, less saccarine suckup than Qwen Next.
Your GLM Air quants are smarter, but Ling-flash TPS is 2x faster!
Oh, it will get a Brainstorming treatment soon, if it works :)
I was really surprised by Ling. I have all quants of Ling and Ming, just uploaded that one to try. Surprisingly good.
Completely agree about the Next, but there is something else there at play, metaphoric inference. The qx quants bring that out, works great for model bonding in RP. Helps to personalize it too, it "sticks" to known people and acts like them. Makes coding not boring
Uploading the qx86-hi too