The authors evaluated matched world action models, robot policies trained to predict actions alongside future visual observations.
Robot policies generalize better when future tokens remain at inference
A controlled study finds that one pass over fully noised future-video tokens preserves most generalization benefits without generating clean future frames.
Independent
Renping Zhou · Zanlin Ni · Zihao Fan · Guohao Fu · Zeyu Liu · Hao Shi · +5 more
Research Digest··2 min read
Zhou and colleagues compare world action models that either generate future video during inference or discard future representations to save computation.
Why this paper
Independent
In one line
Latent world action models match explicit ones in distribution but lose generalization; conditioning on fully noised future tokens recovers most of the benefit.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§