# MI-Pruner Cuts 89% of Image Tokens While Keeping 97% AI Accuracy

> Source: <https://alphasignal.ai/news/mi-pruner-cuts-89-of-image-tokens-while-keeping-97-ai-accuracy>
> Published: 2026-11-18 18:01:33+00:00

**Recommendation:** Prune visual tokens by mutual-information scoring in the projector's output space instead of decoder attention maps, so pruning stays compatible with FlashAttention; the authors kept over 97% of LLaVA1.5-7B performance while dropping 88.9% of image tokens.
More in the playbook "Inference and self-hosting": <https://wpnews.pro/playbooks/inference-serving>

# MI-Pruner Cuts 89% of Image Tokens While Keeping 97% AI Accuracy

MI-Pruner decides which image tokens to drop by measuring how much each one shares with the text prompt. It never reads attention maps, so it can be added to off-the-shelf multimodal models without changing them.

- [MI-Pruner](https://arxiv.org/html/2604.03072v2?utm_source=alphasignal) drops image tokens based on how much each shares with the text prompt (mutual information), with no attention maps needed.
- On LLaVA1.5-7B it keeps over 97% of full performance while removing 88.9% of visual tokens.
- On video it averages 93.19% accuracy with only 114 visual tokens, or 5.57% of the original.
- It is training-free, works with FlashAttention, and runs on Qwen3VL, which [CLS]-based pruners like VisPruner cannot handle.
- Limits: Qwen3VL baselines were built by the authors, MI beats cosine scoring by only 1-2 points, and outside reviewers flagged under-testing.
- Code is available on [GitHub](https://github.com/renaissanceee/MI-Pruner?utm_source=alphasignal) .

Researchers at KU Leuven and Weill Cornell Medicine have introduced [MI-Pruner](https://arxiv.org/html/2604.03072v2?utm_source=alphasignal), a training-free method that removes visual tokens before a multimodal large language model processes them. It scores projected image and text features with mutual information, requiring no decoder attention maps or model retraining. On LLaVA1.5-7B, the authors report preserving more than 97% of the unpruned model’s benchmark performance while removing 88.9% of its visual tokens, leaving 64 of 576.

LLaVA-style models convert each image into hundreds of tokens that enter the language-model decoder alongside the prompt. Those tokens increase prefill computation, attention work and key-value cache memory, even when many image patches contribute little to the answer. Pruning them before decoding can reduce latency and memory use while leaving the base model unchanged.

## Attention maps are a fragile dependency

Existing visual-token pruners generally rank image patches through attention scores from either the vision encoder or the language-model decoder. FastV reads decoder attention, while VisionZip and VisPruner use signals from the vision encoder.

| Signals and integration requirements of representative pruning methods |  |  | 
|---|---|---|
| Method | Ranking signal | Integration constraint | 
|---|---|---|
| FastV | LLM decoder attention | Requires decoder attention scores | 
| VisionZip | Vision-encoder attention | Depends on encoder attention outputs | 
| VisPruner | Vision-encoder attention | Uses a `[CLS]` summary token | 
| MI-Pruner | Projected image and text features | Requires no attention matrix | 

Decoder-side ranking complicates integration with optimized inference kernels such as FlashAttention, which computes attention without materializing the full attention matrix. Encoder-side methods can depend on architecture-specific outputs. The authors note that VisPruner expects a `[CLS]` token, while recent Qwen vision-language encoders do not provide one.

## A score for relevance and repetition

Mutual information measures how much knowledge of one variable reduces uncertainty about another. Pointwise mutual information, or PMI, applies that idea to a specific pair, such as an image patch and a prompt token. MI-Pruner constructs PMI-based scores from embedding similarities in the projector’s output space, where image and text tokens share the language model’s representation format but have not entered the decoder.

## This story is for Pro members

You've reached the end of the free preview. Upgrade to AlphaSignal Pro to read the full article - and everything else behind the paywall.
