LARY is a unified evaluation framework for latent action representations.
Given any model that produces latent action representations (LAMs or visual encoders), LARY provides three complementary evaluation pipelines:
get_latent_action
Extract latent action representations from videos or image pairs
classification
Probe how… See the full description on the dataset page:
https://huggingface.co/datasets/meituan-longcat/LARYBench.