Soumen_Sinha

AI researcher , TU Delft

[Turing Post (Ksenia Se) @TheTuringPost]

What does the world look like inside an AI model?

Here are 3 world model architectures based on how they represent the world (+ the subtypes) ↓

1. A generated observation – a model predicts what comes next: an image, video frame, or token sequence.
• Visual tokens (Dreamer 4, @GoogleDeepMind)
• Diffusion / flow (Waymo World Model, Cosmos-Predict2.5 @nvidia)

2. A compact latent state – a world model compresses the world into an internal representation and predicts how it changes.
• Reconstructive state-space (DayDreamer)
• Predictive representations (JEPA, @ylecun)
• Task-oriented (MuZero)

3. An explicit structure of objects and space – a model represents the world via objects, their properties, positions, and interactions.
• Spatial / geometry-aware (Atlas, @theworldlabs)
• Object-centric (TextOCVP; FIOC-WM)

We broke down what’s inside these architectures and how they work in this guide ↓
https://t.co/rInKfQ9ecz

[Soumen_Sinha @SoumenSinha02]

MBISMU

[Soumen_Sinha @SoumenSinha02]

Cheers to new beginnings !!!!!

[Soumen_Sinha @SoumenSinha02]

One last time 😭

#messi

[Rohan Paul @rohanpaul_ai]

Yann LeCun's (@ylecun ) latest talk at ETH Zürich

Scaling LLMs to reach AGI is "impossible"

A large language model is trained on about 30 trillion tokens, which is roughly 10^14 bytes of text and would take a person about 400,000 years to read.

A 4-year-old child receives about the same amount of data, 10^14 bytes, through vision alone in about 1 year and 10 months.

In his view, intelligence is the ability to learn new tasks quickly or perform them without prior training, as a teenager learns to drive in about 20 hours. Scaling increases stored knowledge, but it does not produce this ability to adapt.

----
From "Perfology Clips" YouTube channel, (link in comment)

[Soumen_Sinha @SoumenSinha02]

@jahnavis27 How's europe?
Visit caramel bunker in barcelona for sunset.

[Hilde Kuehne @HildeKuehne]

For everybody who plans to submit some random AI slop paper that you didn’t even bother to understand or even read … just don’t. Withdraw as long as you can. Nobody is going to give you a job or PhD position for this, but you will never get this out of your CV. And people will figure you out. We will ask you about what you wrote and this will be the end of the interview. You need to understand that you are closing the door to the career you might actually seek by doing this. Stop it as long as you can!

[Soumen_Sinha @SoumenSinha02]

@eccvconf , 📍 Malmo 🇸🇪

[Soumen_Sinha @SoumenSinha02]

Met this guy who likes world models. 🤷‍♂️
@ylecun

[Soumen_Sinha @SoumenSinha02]

“If you are interested in human level ai, dont work with LLMs.” - @ylecun
@eccvconf

[Soumen_Sinha @SoumenSinha02]

@jahnavis27 Crazy setup

[Soumen_Sinha @SoumenSinha02]

@jahnavis27 Cool stuff :)

[All About Argentina 🛎🇦🇷 @AlbicelesteTalk]

I feel like I got shot in the chest....

[Soumen_Sinha @SoumenSinha02]

@naman_dugar Har jayenge bhai

[Soumen_Sinha @SoumenSinha02]

@Sidziesama Bhai blacklist kardenge dekh ke 😂

[Soumen_Sinha @SoumenSinha02]

@vivekdhir77 rejecting companies is the only way macha....

[Footy Accumulators @FootyAccums]

Spine tingling words from Peter Drury..

Nailed it. 🇦🇷

📹 @kotkar_kunal