/Tech4h ago

Sonia Joseph finds physical reasoning in V-JEPA models consistently emerges one-third of the way through network layers

Another study evaluated implicit physics in video diffusion models.

3945377.2K
Sonia Joseph
Lucas Beyer (bl16)#62
Original post
Lucas Beyer (bl16)@giffmana#62inTech

To be clear, this is not a V-JEPA or VideoMAE diss, just resurrecting the fact that "pure videogen" models may indeed learn an explicit model of the world/physics as a byproduct.

Also cc @mapo1 we chatted about this and you also intuitively pushed back against such claim.

Lucas Beyer (bl16)@giffmana

The paper is "The invisible hand of physics" from a surprisingly diverse set of authors (Parsa Esmati, @Somjit77): https://arxiv.org/abs/2606.05328 ; It's from just a few days ago. I learned about it from a nice talk by @katjahofmann today.

The paper from earlier in the year is by @soniajoseph_ etal: https://arxiv.org/abs/2602.07050

7:38 AM · Jun 10, 2026 · 2.7K Views
Sentiment
Sentiment building, check back later.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS107REPLIES1
Adam Goldstein@goldstein_aa

@giffmana @mapo1 But isn't that indeed a JEPA diss? If what you say is true, what's the justification for JEPA?

4hViews 107
LIKES2
Lucas Beyer (bl16)@giffmana

@goldstein_aa @mapo1 for example, it may be much more efficient since it works in just one forward pass.

4hViews 98Likes 2
RETWEETS4
Lucas Beyer (bl16)@giffmana

The paper is "The invisible hand of physics" from a surprisingly diverse set of authors (Parsa Esmati, @Somjit77): https://arxiv.org/abs/2606.05328 ; It's from just a few days ago. I learned about it from a nice talk by @katjahofmann today.

The paper from earlier in the year is by @soniajoseph_ etal: https://arxiv.org/abs/2602.07050

4hViews 4.8KLikes 67Bookmarks 35
Wenyao (Wayne) Zhang@zhang_weny92997

@giffmana @Somjit77 @katjahofmann This result may be highly dependent on the data scale？

4hViews 103