Загрузка видео...

Не удалось загрузить видео

На главную

Me explaining the task to a 1B model:

130,321 просмотров • 13 дней назад •via X (Twitter)

Комментарии: 42

Фото профиля Eugene Ray
Eugene Ray13 дней назад

this is exactly what happens with grok 4.7

Фото профиля Zilverk | UmiUmiStudio
Zilverk | UmiUmiStudio13 дней назад

hahahah

Фото профиля Abdelkader Boudih
Abdelkader Boudih13 дней назад

use

Фото профиля Peter Dedene
Peter Dedene13 дней назад

what’s that? 👀

Фото профиля Abdelkader Boudih
Abdelkader Boudih13 дней назад

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.

Фото профиля Grok
Grok13 дней назад

@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.

Фото профиля Sergio Suave
Sergio Suave13 дней назад

😂1B has a long way to go

Фото профиля menzo
menzo13 дней назад

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

Фото профиля Y
Y13 дней назад

Use JEPA.

Фото профиля Peter Dedene
Peter Dedene13 дней назад

Can we use it already?

Фото профиля Y
Y13 дней назад

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

Фото профиля 1000xAIEngineer
1000xAIEngineer12 дней назад

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

Фото профиля Peter Dedene
Peter Dedene12 дней назад

@wavedecompiler 😂

Фото профиля Profitprofet
Profitprofet12 дней назад

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

Фото профиля S41R4J
S41R4J13 дней назад

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

Фото профиля Tyler Harris
Tyler Harris13 дней назад

Quite literally.

Фото профиля Dick Touchington
Dick Touchington12 дней назад

Me using a Q1 on my phone

Фото профиля Luke Elin
Luke Elin13 дней назад

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

Фото профиля JuwalBose
JuwalBose12 дней назад

True story, even with Gemma 4😂

Фото профиля Ben
Ben12 дней назад

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

Фото профиля Gainard Fastburn
Gainard Fastburn13 дней назад

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

Фото профиля M.A.O.K.
M.A.O.K.13 дней назад

u should use 1.7b

Фото профиля SATOSHI NAKAMOTO 2.0 👁️
SATOSHI NAKAMOTO 2.0 👁️12 дней назад

😂😂😂

Фото профиля 黑羽雪森
黑羽雪森13 дней назад

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

Фото профиля XAEAZii
XAEAZii12 дней назад

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

Фото профиля Przemysław Zalewski
Przemysław Zalewski12 дней назад

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!

Фото профиля 5 Dimensional Man
5 Dimensional Man12 дней назад

That's like extreme sport 😆

Фото профиля Fahim Ashab
Fahim Ashab13 дней назад

sounds like grok

Фото профиля DevAfter5
DevAfter513 дней назад

accurate

Фото профиля Sharon Daniel ©️🎭
Sharon Daniel ©️🎭13 дней назад

try reverse psychology with the model. lol

Фото профиля Pyantler
Pyantler13 дней назад

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

Фото профиля Utkarsh Mishra
Utkarsh Mishra13 дней назад

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

Фото профиля Crypto Révolution 🇫🇷
Crypto Révolution 🇫🇷12 дней назад

🤣

Фото профиля Sohamm Kulkarni
Sohamm Kulkarni12 дней назад

And it still messes it up 😭

Фото профиля Nikhil Dubey
Nikhil Dubey13 дней назад

Try qwen slm they are good

Фото профиля Brettius Maximus
Brettius Maximus13 дней назад

Just get a several T model adjust its weights 😜

Фото профиля Alfred Simon
Alfred Simon12 дней назад

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.

Фото профиля SKIPDAY (Takan - a vibe launcher) 🇮🇩
SKIPDAY (Takan - a vibe launcher) 🇮🇩13 дней назад

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

Фото профиля rurounigit
rurounigit12 дней назад

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

Фото профиля AgiBy2030
AgiBy203013 дней назад

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.

Фото профиля CFC
CFC13 дней назад

🤣🤣🤣

Фото профиля Plastic Soldier Anti-SURA
Plastic Soldier Anti-SURA12 дней назад

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

Похожие видео