Gheorghe Chesler PRO
nightmedia
AI & ML interests
Human-Like AI and the Deckard Formula(an optics-inspired, mixed quantization approach) are the focus of my research. Donations are appreciated, they would go towards more hardware to run tests
Recent Activity
updated
a model
about 15 hours ago
nightmedia/metatune-gpt20b-R1.1-q8-hi-mlx
updated
a model
about 15 hours ago
nightmedia/metatune-gpt20b-R1.1-qx86-hi-mlx
updated
a model
about 18 hours ago
nightmedia/aquif-3.5-Max-42B-A3B-qx86-hi-mlx
Organizations
👍👍👍
❤️
1
1
#3 opened 2 days ago
by
mimeng1990
LLM Settings
👀
1
2
#1 opened 8 days ago
by
webboty
Deckard Quant Method?
3
#2 opened 11 days ago
by
beezu
Great Quant !
🔥
1
12
#1 opened 2 months ago
by
bobig
25% smaller !?!
5
#1 opened 17 days ago
by
bobig
Please do the new Kat-Dev 72b
👀
1
7
#1 opened 30 days ago
by
deleted
The bokeh is special
❤️
1
1
#1 opened about 1 month ago
by
jc2375
New activity in
nightmedia/Qwen3-30B-A3B-Deepseek-Distill-Instruct-2507-qx64-hi-mlx
about 1 month ago
I heard you like Distills :-D
1
#1 opened about 1 month ago
by
bobig
solid little speed demon
🔥
1
1
#1 opened about 1 month ago
by
bobig
slower than non-LIMI
1
#1 opened about 1 month ago
by
bobig
qx43 wen? ;-)
🔥
1
4
#1 opened about 2 months ago
by
ljupco
"model happier and more eager to explore and innovate"
🔥
❤️
1
3
#1 opened about 2 months ago
by
bobig
This model is NICE DUDE!!
❤️
🔥
1
1
#2 opened about 2 months ago
by
layer4down
40% faster!
🔥
👍
1
9
#1 opened about 2 months ago
by
bobig
New PR in mlx boost speed :)
🔥
🧠
1
1
#1 opened about 2 months ago
by
sb6666
Update README.md
2
#1 opened about 2 months ago
by
21world
A Reademe worth reading.
🔥
1
1
#1 opened about 2 months ago
by
bobig
Thanks Nightmedia / Unsloth — Would Love Full GLM-4.5 in mxfp4-mlx
🔥
1
2
#1 opened about 2 months ago
by
wake6