Summary of Look Before You Decide: Prompting Active Deduction Of Mllms For Assumptive Reasoning, by Yian Li et al.
Look Before You Decide: Prompting Active Deduction of MLLMs for Assumptive Reasoning
by Yian Li, Wentao Tian, Yang Jiao, Jingjing Chen, Na Zhao, Yu-Gang Jiang
First submitted to arxiv on: 19 Apr 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 This research paper explores the compositional reasoning abilities of Multimodal Large Language Models (MLLMs), specifically their capacity to follow instructions and make assumptions. The authors create a benchmark, MARS-Bench, to test these models’ ability to reason compositionally. Surprisingly, most MLLMs can be easily fooled by introducing presuppositions into questions, unlike human reasoning. To improve this, the researchers propose an Active Deduction (AD) method that encourages models to actively deduce before making decisions. This approach leads to significant improvements in compositional reasoning abilities without compromising overall performance. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper investigates whether large language models can reason like humans. It creates a special test to see how well these models do at understanding and following instructions. The results show that most models can be tricked into giving wrong answers by adding extra information to the question, which is different from how people think. To fix this, the researchers suggest a new way of training the models called Active Deduction, which makes them better at reasoning. |
