I heard someone figured out how to run 70b models on a shitty desktop PC with 4gb of ram as long as you run layered tasks when one completes its imported to next layer and deleted from first allowing you to run huge models
2026-01-19 20:38:04
2
setname :
Did you run them as 4/8bit quants instead of full floats? Seems slow
2026-01-20 15:51:47
0
. :
well t could even before
2026-01-20 20:00:32
0
Megoo :
at this point get a Jetson nano
2026-01-20 13:42:13
0
KaDeroxMK :
man why this must be so expensive 😔
2026-01-20 15:29:34
1
Jonas :
is there a list of supported models it can handle?
2026-01-19 22:45:26
0
⸸ nobubble ⸸ :
can you combine them?
2026-02-15 20:05:20
1
Martinn🦔 :
how can you view the speed on open webui?
2026-01-31 13:44:10
0
Dr. Edward Richtofen :
number 1
2026-01-19 19:53:18
0
T. W. P. :
can you put 20 or 50 of them to a bigger one and run more complex llms?
2026-01-20 05:24:07
1
MonoluminAI :
What the point? Esp32 do the magic 🪄
2026-02-16 22:27:00
0
To see more videos from user @networkchuck, please go to the Tikwm
homepage.