3 papers
cs.CV2026
IMPULSE: Push First, Then Predict in V-JEPA 2 Feature Space
Tingan Jin
Latent world models such as V-JEPA 2 forecast in the feature space of a video encoder, but one look at an object does not reveal how hard it is to rotate, how much it bounces or ho…
cs.CL2026
RIFT: Token Influence Follows the Path of Least Resistance
Tingan Jin, Shuhang Dong, Haosong Li
Which context tokens drive a language model's answer? Graph-based tools such as attention rollout and attention flow answer this from a single forward pass, yet their faithfulness…
stat.ML2026
Iterative Erasure Count Is Not an Affine-Invariant Concept Dimension
Tingan Jin, Shuhang Dong, Haosong Li +1
How many directions does a neural representation use to encode a concept? A common answer repeatedly erases probe directions and reports the stopping count or cumulative removed ra…