CausalPrune: Counterfactual Causal Token Pruning for Efficient Large Vision-Language Models
cs.CV
Submitted: 2026-10-12
Updated: 2026-10-12
Terminology
Sources
- Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond
- Qwen3-VL Technical Report
- FlashVLM: Text-Guided Visual Token Selection for Large Multimodal Models
- Language-Guided Token Compression with Reinforcement Learning in Large Vision-Language Models
- ZOO-Prune: Training-Free Token Pruning via Zeroth-Order Gradient Estimation in Vision-Language Models
- Treble Counterfactual VLMs: A Causal Approach to Hallucination
- Evaluating Object Hallucination in Large Vision-Language Models
- Relaxing Anchor-Frame Dominance for Mitigating Hallucinations in Video Large Language Models
- ChartQA: A Benchmark for Question Answering about Charts with Visual and Logical Reasoning
- GLU Variants Improve Transformer
- IDPruner: Harmonizing Importance and Diversity in Visual Token Pruning for MLLMs
- Saliency-driven Dynamic Token Pruning for Large Language Models
- PyramidDrop: Accelerating Your Large Vision-Language Models via Pyramid Visual Redundancy Reduction
- MM-Vet: Evaluating Large Multimodal Models for Integrated Capabilities
- Vinoground: Scrutinizing LMMs over Dense Temporal Reasoning with Short Videos
- RCP: Representation Consistency Pruner for Mitigating Distribution Shift in Large Vision-Language Models
- CF-VLM:CounterFactual Vision-Language Fine-tuning
- Beyond Attention or Similarity: Maximizing Conditional Diversity for Token Pruning in MLLMs
- SparseVLM: Visual Token Sparsification for Efficient Vision-Language Model Inference
- LLaVA-FA: Learning Fourier Approximation for Compressing Large Multimodal Models
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs