Pith. sign in

REVIEW 14 cited by

Token Pruning in Multimodal Large Language Models: Are We Solving the Right Problem?

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2502.11501 v2 pith:JO4BRMWZ submitted 2025-02-17 cs.CL cs.CV

Token Pruning in Multimodal Large Language Models: Are We Solving the Right Problem?

classification cs.CL cs.CV
keywords tokenpruninglanguagecostsdesignevaluationlargemethods
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Multimodal large language models (MLLMs) have shown remarkable performance for cross-modal understanding and generation, yet still suffer from severe inference costs. Recently, abundant works have been proposed to solve this problem with token pruning, which identifies the redundant tokens in MLLMs and then prunes them to reduce the computation and KV storage costs, leading to significant acceleration without training. While these methods claim efficiency gains, critical questions about their fundamental design and evaluation remain unanswered: Why do many existing approaches underperform even compared to naive random token selection? Are attention-based scoring sufficient for reliably identifying redundant tokens? Is language information really helpful during token pruning? What makes a good trade-off between token importance and duplication? Are current evaluation protocols comprehensive and unbiased? The ignorance of previous research on these problems hinders the long-term development of token pruning. In this paper, we answer these questions one by one, providing insights into the design of future token pruning methods.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 14 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. ORCA: ORgan-Centroid Aggregation for Training-Free 3D CT Visual Token Compression

    cs.CV 2026-07 conditional novelty 6.0

    ORCA compresses 3D CT tokens into organ-guided connected regions with sinusoidal centroid encoding, outperforming grid average and other compressors at matched budgets.

  2. C-PTQ: Fisher-weighted Channel-wise Sensitivity for Post-training Quantization of MLLMs

    cs.CV 2026-07 conditional novelty 6.0

    C-PTQ weights quantization error by per-channel Fisher information of the task loss, improving low-bit accuracy of multimodal LLMs by small margins over existing channel-wise scaling methods.

  3. MVPruner: Dynamic Token Pruning for Accelerating Multi-view Vision-Language Models in Autonomous Driving

    cs.CV 2026-06 unverdicted novelty 6.0

    MVPruner is a two-stage dynamic token pruning technique that uses view diversity for initial budget allocation and instruction text for task-aligned selection, delivering 87.3% FLOPs reduction and 4.97x prefilling spe...

  4. Keyless Attention: Value-Space Routing and Value-Only Caching for Efficient Transformers

    cs.CL 2026-06 unverdicted novelty 6.0

    Keyless Attention eliminates the key projection in transformers via value-space routing, cutting KV cache size by 50% with competitive or superior performance on language modeling and zero-shot benchmarks.

  5. Focus-then-Context: Subject-Centric Progressive Visual Token Reduction for Vision-Language Models

    cs.CV 2026-05 conditional novelty 6.0

    SPpruner reduces visual tokens in VLMs via focus identification followed by context-aware scanning, retaining 22.2% tokens for 2.53x speedup on Qwen2.5-VL with negligible accuracy loss.

  6. Large Vision-Language Models Get Lost in Attention

    cs.AI 2026-05 unverdicted novelty 6.0

    In LVLMs, attention can be replaced by random Gaussian weights with little or no performance loss, indicating that current models get lost in attention rather than efficiently using visual context.

  7. ST-Prune: Training-Free Spatio-Temporal Token Pruning for Vision-Language Models in Autonomous Driving

    cs.CV 2026-04 unverdicted novelty 6.0

    ST-Prune is a training-free spatio-temporal token pruning framework for VLMs in autonomous driving that achieves near-lossless results at 90% token reduction by exploiting motion volatility, temporal recency, and mult...

  8. POINTS-Long: Adaptive Dual-Mode Visual Reasoning in MLLMs

    cs.CV 2026-04 unverdicted novelty 6.0

    POINTS-Long is a dual-mode multimodal large language model that uses dynamic visual token scaling to retain 97.7-99.7% accuracy on long-form tasks with 1/40 to 1/10th the tokens and supports streaming via detachable KV-cache.

  9. Decoupled Similarity for Task-Aware Token Pruning in Large Vision-Language Models

    cs.CV 2026-04 unverdicted novelty 6.0

    DeSAP uses decoupled cross-modal similarity plus visual saliency to prune visual tokens in LVLMs, retaining 11.1% tokens for 10x FLOPs reduction and 98.1% performance on LLaVA-1.5-7B.

  10. MVPruner: Dynamic Token Pruning for Accelerating Multi-view Vision-Language Models in Autonomous Driving

    cs.CV 2026-06 unverdicted novelty 5.0

    MVPruner is a two-stage adaptive token pruning technique for multi-view VLMs that achieves 87.3% FLOPs reduction and 4.97x prefilling speedup while retaining 98.5% accuracy on DriveLM.

  11. Keyless Attention: Value-Space Routing and Value-Only Caching for Efficient Transformers

    cs.CL 2026-06 conditional novelty 5.0

    Keyless Attention eliminates the key projection, computing softmax(QV^T)V and reducing KV-cache memory by exactly 50%, with comparable or better performance on small-scale models.

  12. EvoComp: Learning Visual Token Compression for Multimodal Large Language Models via Semantic-Guided Evolutionary Labeling

    cs.CV 2026-04 unverdicted novelty 5.0

    EvoComp compresses visual tokens in MLLMs by 3x while retaining 99.3% accuracy via an evolutionary labeling strategy that searches for low-loss, semantically diverse token subsets.

  13. DisCEdge: Distributed Context Management for Large Language Models at the Edge

    cs.DC 2025-11 unverdicted novelty 5.0

    DisCEdge manages LLM context in tokenized form replicated on edge nodes, delivering up to 14.46% faster median responses, 15% lower sync overhead, and 90% smaller client requests versus baselines while ensuring consistency.

  14. Keyless Attention: Value-Space Routing and Value-Only Caching for Efficient Transformers

    cs.CL 2026-06 conditional novelty 4.0

    An attention variant that uses values as keys reduces KV-cache memory by 50% and shows roughly comparable perplexity on small-scale training runs.