Emnlp Neurips Tokenpruning
Two works on efficient VLM inference — AdaptInfer and EventPrune - which provide better criteria for when and how to prune visual tokens, got accepted by EMNLP 2026 (Findings) and NeurIPS 2026 respectively 
Two works on efficient VLM inference — AdaptInfer and EventPrune - which provide better criteria for when and how to prune visual tokens, got accepted by EMNLP 2026 (Findings) and NeurIPS 2026 respectively