mlsanityData to 14 Oct 2026

CATP: Cross-Attention Token Pruning for Accuracy Preserved Multimodal Model Inference

arXiv 2404.08567 ↗First posted 2 Apr 2024cs.CL cs.AIcited by 3

On the map: Multimodal LLMs / vision-language models → Visual token pruning and efficiency

also in 1 more
Abstract (arXiv)

In response to the rising interest in large multimodal models, we introduce Cross-Attention Token Pruning (CATP), a precision-focused token pruning method. Our approach leverages cross-attention layers in multimodal models, exemplified by BLIP-2, to extract valuable information for token importance determination. CATP employs a refined voting strategy across model heads and layers. In evaluations, CATP achieves up to 12.1X higher accuracy compared to existing token pruning methods, addressing the trade-off between computational efficiency and model precision.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Placed in Visual token pruning and efficiency, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 10 of its 14 references that are in the map. First posted on arXiv 2 Apr 2024.

Builds on

10 of 14 references.

References · 10

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2019ViLBERT: Pretraining Task-Agnostic Visiolinguistic Representations for Vision-and-Language Taskscited by 2,222
  3. 2019VisualBERT: A Simple and Performant Baseline for Vision and Languagecited by 1,296
  4. 2019Analyzing Multi-Head Self-Attention: Specialized Heads Do the Heavy Lifting, the Rest Can Be Prunedcited by 856
  5. 2021Efficient Large-Scale Language Model Training on GPU Clusters Using Megatron-LMcited by 632
  6. 2019Learned Step Size Quantizationcited by 628
  7. 2022ZeroQuant: Efficient and Affordable Post-Training Quantization for Large-Scale Transformerscited by 526
  8. 2020SpAtten: Efficient Sparse Attention Architecture with Cascade Token and Head Pruningcited by 234
  9. 2021Nonuniform-to-Uniform Quantization: Towards Accurate Quantization via Generalized Straight-Through Estimationcited by 79
  10. 2020PoWER-BERT: Accelerating BERT Inference via Progressive Word-vector Eliminationcited by 10

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 2.

3 papers.

  1. 2025TokenSwap: Backdoor Attack on the Compositional Understanding of Large Vision-Language Modelscited by 5
  2. 2026Adaptive Head Budgeting for Efficient Multi-Head Attentioncited by 1
  3. 2026ModTrans: Translating Real-world Models for Distributed Training Simulatorcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025VFlowOpt: A Token Pruning Framework for LMMs with Visual Information Flow-Guided Optimization
  2. 2025Stop Looking for Important Tokens in Multimodal Language Models: Duplication Matters More
  3. 2025TrimTokenator: Towards Adaptive Visual Token Pruning for Large Multimodal Models
  4. 2026HAWK: Head Importance-Aware Visual Token Pruning in Multimodal Models
  5. 2026IDPruner: Harmonizing Importance and Diversity in Visual Token Pruning for MLLMs
  6. 2025CATP: Contextually Adaptive Token Pruning for Efficient and Enhanced Multimodal In-Context Learning
  7. 2026Capturing Token Tendencies for Training-Free Token Pruning in Multimodal Large Language Models
  8. 2025TransPrune: Token Transition Pruning for Efficient Large Vision-Language Model