Reza Langari

CorrRisk-WM, a planning-oriented partial world model coupling environment evolution with supervised intrusion and near-miss prediction over bounded candidate-trajectory corridors, supports coupling environment evolution with candidate-conditioned geometric reasoning for risk prediction and safety-oriented candidate selection.

ActionSplice is introduced, an inference framework that formulates this problem as Counterfactual State Transport (CST), a lightweight corrector that transports the interrupted backbone-native representation toward the matched state induced by the revised action at the same solver step.

SparsePR cuts attention-reconstruction error while speeding up video generation by over 2.5x without sacrificing quality.

Distilling foundation models with a novel instance-aware contrastive loss yields smaller segmentation models that surprisingly outperform their larger teachers, even with limited labeled data.

Decoupling reasoning from action generation in autonomous driving VLMs lets you beat larger end-to-end models while slashing training costs.