Summary of Par: Prompt-aware Token Reduction Method For Efficient Large Multimodal Models, by Yingen Liu et al.
PAR: Prompt-Aware Token Reduction Method for Efficient Large Multimodal Models
by Yingen Liu, Fan Wu, Ruihui Li, Zhuo Tang, Kenli Li
First submitted to arxiv on: 9 Oct 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed Prompt-Aware Token Reduction (PAR) method efficiently reduces visual tokens in multimodal large language models (MLLMs) without compromising model performance. By categorizing contextual redundancy into external and internal types, PAR minimizes computational load through semantic retrieval and token routing mechanisms. Experimental results demonstrate that PAR reduces FLOPs by 83% with a compression ratio of 89%, while retaining 97% of baseline accuracy across various visual question answering tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary PAR helps large language models (LLMs) process long visual contexts more efficiently without losing performance. It finds and groups important visual pieces together, reducing the number of calculations needed. This makes it easier to run LLMs on devices with limited resources. PAR is a simple addition that works well, even when compared to other methods that require extra training or complex changes. |
Keywords
» Artificial intelligence » Prompt » Question answering » Token
