{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/similarity-aware-token-pruning-your-vlm-but","title":"Similarity-Aware Token Pruning: Your VLM but Faster","arxiv_id":"2503.11549","date":"2025-03-14","proceeding":null,"authors":["Ahmadreza Jeddi","Negin Baghbanzadeh","Elham Dolatabadi","Babak Taati"],"abstract":"The computational demands of Vision Transformers (ViTs) and Vision-Language Models (VLMs) remain a significant challenge due to the quadratic complexity of self-attention. While token pruning offers a promising solution, existing methods often introduce training overhead or fail to adapt dynamically across layers. We present SAINT, a training-free token pruning framework that leverages token similarity and a graph-based formulation to dynamically optimize pruning rates and redundancy thresholds. Through systematic analysis, we identify a universal three-stage token evolution process (aligner-explorer-aggregator) in transformers, enabling aggressive pruning in early stages without sacrificing critical information. For ViTs, SAINT doubles the throughput of ViT-H/14 at 224px with only 0.6% accuracy loss on ImageNet-1K, surpassing the closest competitor by 0.8%. For VLMs, we apply SAINT in three modes: ViT-only, LLM-only, and hybrid. SAINT reduces LLaVA-13B's tokens by 75%, achieving latency comparable to LLaVA-7B with less than 1% performance loss across benchmarks. Our work establishes a unified, practical framework for efficient inference in ViTs and VLMs.","url_abs":"https://arxiv.org/abs/2503.11549v1","url_pdf":"https://arxiv.org/pdf/2503.11549v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"similarity-aware-token-pruning-your-vlm-but","repo_url":"https://github.com/ArmenJeddi/saint","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"jax","reach":null}],"tasks":[],"methods":[{"method_slug":"cutmix","method_name":"CutMix"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"feedforward-network","method_name":"Feedforward Network"},{"method_slug":"mixup","method_name":"Mixup"},{"method_slug":"pruning","method_name":"Pruning"},{"method_slug":"saint","method_name":"SAINT"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2503.11549","atlas_url":"https://app.syntology.ai/?focus=2503.11549","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}