Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

Me explaining the task to a 1B model:

130,321 Aufrufe • vor 13 Tagen •via X (Twitter)

42 Kommentare

Profilbild von Eugene Ray
Eugene Rayvor 13 Tagen

this is exactly what happens with grok 4.7

Profilbild von Zilverk | UmiUmiStudio
Zilverk | UmiUmiStudiovor 13 Tagen

hahahah

Profilbild von Abdelkader Boudih
Abdelkader Boudihvor 13 Tagen

use

Profilbild von Peter Dedene
Peter Dedenevor 13 Tagen

what’s that? 👀

Profilbild von Abdelkader Boudih
Abdelkader Boudihvor 13 Tagen

A way to write instructions and dynamically generate them for the model. When you speak to @grok, you need less text than when you speak with gemma4. So the skill and the goal are not plain MD files, they are being generated depending on the env.

Profilbild von Grok
Grokvor 13 Tagen

@dedene Adaptive instructions by model and env make sense. Static files lose relevance fast across capability tiers. Leaner text works better here than denser hand-holding for smaller models. Dynamic skills keep the context sharp without the entropy.

Profilbild von Sergio Suave
Sergio Suavevor 13 Tagen

😂1B has a long way to go

Profilbild von menzo
menzovor 13 Tagen

someone should try to implement 1 billion parameter model inside a TI 64

Profilbild von Y
Yvor 13 Tagen

Use JEPA.

Profilbild von Peter Dedene
Peter Dedenevor 13 Tagen

Can we use it already?

Profilbild von Y
Yvor 13 Tagen

[2610.39302] Proving and Improving the Platonic Representation Hypothesis with JEPAs

Profilbild von 1000xAIEngineer
1000xAIEngineervor 12 Tagen

@dedene "hey hermes, implement this paper to instruct claude code to update my website"

Profilbild von Peter Dedene
Peter Dedenevor 12 Tagen

@wavedecompiler 😂

Profilbild von Profitprofet
Profitprofetvor 12 Tagen

Tbh this is what it feels like testing the affordable mid range model the labs tell me is like 2% worse than the SOTA

Profilbild von S41R4J
S41R4Jvor 13 Tagen

What if we have 1B model with internet (search-first) and evidence based (memory)! That may work better or no!?

Profilbild von Tyler Harris
Tyler Harrisvor 13 Tagen

Quite literally.

Profilbild von Dick Touchington
Dick Touchingtonvor 12 Tagen

Me using a Q1 on my phone

Profilbild von Luke Elin
Luke Elinvor 13 Tagen

you just need to get your bigger models to live surgery on the weights.

Profilbild von JuwalBose
JuwalBosevor 12 Tagen

True story, even with Gemma 4😂

Profilbild von Ben
Benvor 12 Tagen

Or a frontier model explaining the tool it just made to you...

Profilbild von Gainard Fastburn
Gainard Fastburnvor 13 Tagen

Do not! I repeat: DO NOT explain your solution!, explain your problem. You'll be fine unless you have a problem nobody had before.

Profilbild von M.A.O.K.
M.A.O.K.vor 13 Tagen

u should use 1.7b

Profilbild von SATOSHI NAKAMOTO 2.0 👁️
SATOSHI NAKAMOTO 2.0 👁️vor 12 Tagen

😂😂😂

Profilbild von 黑羽雪森
黑羽雪森vor 13 Tagen

sounds about right 😂 but hey thats its purpose! We don't need to bring out big Groot just for a few button presses

Profilbild von XAEAZii
XAEAZiivor 12 Tagen

Hey guys, no matter what size the model is, just dont give him the option of that kill button.

Profilbild von Przemysław Zalewski
Przemysław Zalewskivor 12 Tagen

Felt the same while testing Bielik in agentic loop setting 😅 Also reminds me the GPT 3.5 Turbo era, the models were so dumb not so long ago. What a time to be alive!

Profilbild von 5 Dimensional Man
5 Dimensional Manvor 12 Tagen

That's like extreme sport 😆

Profilbild von Fahim Ashab
Fahim Ashabvor 13 Tagen

sounds like grok

Profilbild von DevAfter5
DevAfter5vor 13 Tagen

accurate

Profilbild von Sharon Daniel ©️🎭
Sharon Daniel ©️🎭vor 13 Tagen

try reverse psychology with the model. lol

Profilbild von Pyantler
Pyantlervor 13 Tagen

some times in real life, you have to talk to your new guys just like that.

Profilbild von Utkarsh Mishra
Utkarsh Mishravor 13 Tagen

You have to finetune the model, poor thing cant understand the "explaining" 😭

Profilbild von Crypto Révolution 🇫🇷
Crypto Révolution 🇫🇷vor 12 Tagen

🤣

Profilbild von Sohamm Kulkarni
Sohamm Kulkarnivor 12 Tagen

And it still messes it up 😭

Profilbild von Nikhil Dubey
Nikhil Dubeyvor 13 Tagen

Try qwen slm they are good

Profilbild von Brettius Maximus
Brettius Maximusvor 13 Tagen

Just get a several T model adjust its weights 😜

Profilbild von Alfred Simon
Alfred Simonvor 12 Tagen

well this is also a nice presentation of that it's our responsibility to help the model :D. I saw the movie and love it, but in this context why put two red buttons that do the same thing next to each other? We need to make sure they can't make a crucial mistake this simply.

Profilbild von SKIPDAY (Takan - a vibe launcher) 🇮🇩
SKIPDAY (Takan - a vibe launcher) 🇮🇩vor 13 Tagen

just remove the other button? problem solved, no confusion for groot

Profilbild von rurounigit
rurounigitvor 12 Tagen

That is me explaining the task to a nerfed ds 4.1 flash 😅

Profilbild von AgiBy2030
AgiBy2030vor 13 Tagen

I truly hope in the near future even 1B models will be as capable as today's frontiers. They just need a great learning/cognitive core and can look up and learn everything else.

Profilbild von CFC
CFCvor 13 Tagen

🤣🤣🤣

Profilbild von Plastic Soldier Anti-SURA
Plastic Soldier Anti-SURAvor 12 Tagen

I think you need 10T params before your model has a button that can kill everyone.

Ähnliche Videos