@easybits.tech.room: What is quantization? | AI basics #ai #aibasics #aiforbeginners

Ivan from a Tech Room
Ivan from a Tech Room
Open In TikTok:
Region: DE
Wednesday 26 August 2026 13:55:42 GMT
35568
1391
50
90

Music

Download

Comments

neisan4237
Nathan :
useless for coding useful for coding
2026-08-26 15:28:00
0
hugstonone
Trilogic :
#HugstonOne
2026-08-27 19:45:48
0
omegote
Omegote :
Finally someone who explains the basic concepts easily enough
2026-08-26 21:44:19
43
soma_army5
S☆MA :
Cool. Thank you.
2026-08-27 16:26:34
1
leraffl
LeRaffl :
Can you show off how this Qwen behaves in a random back and forth chat in a Q2-Q4-Q8? I think people (including me) don't really have a grasp on that. Not necessarily coding, just a little bit of chat would be enough to get an idea.
2026-08-26 15:32:14
3
eliott.brcd
Eliott :
I have 64GB of RAM but only 12 of VRAM, do you think i can fun Qwen 3.8 27B or should i run another one ?
2026-08-26 23:41:28
1
sergio1492_
Sergio :
32gb vram is enough, dont need 64
2026-08-26 17:29:34
0
andrees.mf
andres :
I have a laptop that I want to use with a local AI model for a project. Which AI model would you recommend for these specs? i5-9300H NVIDIA GeForce GTX 1650 with 4 GB VRAM 16 GB RAM
2026-08-26 21:05:43
2
user8931015246011
Gotxi :
I am running unsloth qwen3.8-27b iq3_xxs with 100k context at 800-1000 prefill and 35-50 tok/s generation on my rx 9070 xt (16gb vram) and works great with real world problems
2026-08-26 20:09:09
2
aifantoken
AI Fan Token | $AI 🪐 :
Let that sit.
2026-08-27 12:48:13
0
fxbian.prvt
Fabian🔱 :
"regular consumer hardware" but 64gb ram builds are not that uncommon
2026-08-26 19:33:43
0
real_monke
Monke :
Aegon Targaryen??
2026-08-26 23:19:01
0
https51
lä-ah :
I like bonsai lab's ternary quantisation
2026-08-27 10:59:51
0
doctor_s16
doctor_s16 :
8-Bit works here on my MacBook Pro Max 64GB
2026-08-26 22:50:29
0
kiro39839
Kiro :
very interesting
2026-08-27 04:34:50
0
tiktokrolis
TikTokRolis :
Smaller model smaller perfomance
2026-08-26 18:22:18
0
user_112758
David :
I read once that you can load AI models like YouTube video, consuming only chunks of the model. So if your GPU, VRAM and RAM are very low this technique would allow you to run big models, theoretically.
2026-08-26 16:50:57
4
neisan4237
Nathan :
Q4:useless for coding Q6 useful for coding
2026-08-26 15:28:56
15
realthingtoday
Lexxxo :
I love your explanations ✊🖖
2026-08-26 17:11:00
1
userihu9j8dmwr
userihu9j8dmwr :
Q3 for 16gi VRam
2026-08-26 22:21:30
0
publis84
Pedro Santos 🇵🇹🇬🇧 :
You just earned yourself another follow 👍 great explanation
2026-08-27 20:20:56
0
hqtips
Hqtips :
Ivan, please let's talk about mixture of expert MoE 😏
2026-08-26 21:42:36
1
maxi01010011
Maxin :
👍👍👍
2026-08-27 01:09:41
1
petloveers24
catties :
😁😁😁
2026-08-27 05:56:25
0
th..ht
thøht :
👏👏👏
2026-08-27 13:23:19
0
To see more videos from user @easybits.tech.room, please go to the Tikwm homepage.

Other Videos


About