The field of knowledge distillation is moving towards more efficient and effective methods for transferring knowledge from large teacher models to smaller student models. Recent developments have focused on improving the distillation process, including the use of adaptive distillation methods, novel architecture-centric taxonomies, and the exploration of underutilized information in teacher models. Notably, logit-based distillation has shown significant potential for advancement. Additionally, multiple-teacher self-supervised distillation frameworks have been proposed to unify knowledge from various state-of-the-art models. These innovations have led to substantial improvements in performance and efficiency, enabling more widespread deployment of compact models. Noteworthy papers include: SPENCER, which proposes a self-adaptive model distillation framework for efficient code retrieval, and TopKD, which introduces a top-scaled knowledge distillation method that enhances logit-based distillation. CoMAD is also notable for its multiple-teacher self-supervised distillation framework that achieves state-of-the-art results in compact SSL distillation.