We pioneer a hierarchical distillation strategy that establishes coordinated knowledge transfer between teacher and student models and progressively incorporates guidance information, specifically designed for sparse query-based occupancy prediction.
