LayerMerge: Modality-Agnostic Depth Pruning for Efficient Foundation Model Deployment

Publication
NeurIPS 2025 Workshop on Efficient Reasoning