Summary of Dissecting Dissonance: Benchmarking Large Multimodal Models Against Self-contradictory Instructions, by Jin Gao et al.
Dissecting Dissonance: Benchmarking Large Multimodal Models Against Self-Contradictory Instructions
by Jin Gao, Lei Gan, Yuankai Li, Yixin Ye, Dequan Wang
First submitted to arxiv on: 2 Aug 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: None
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 paper introduces a benchmark to evaluate the ability of large multimodal models (LMMs) in recognizing self-contradictory instructions, which is challenging for language beginners and vulnerable populations. The Self-Contradictory Instructions benchmark comprises 20,000 conflicts, evenly distributed between language and vision paradigms. Current LMMs struggle to identify multimodal instruction discordance due to a lack of self-awareness. To address this issue, the authors propose Cognitive Awakening Prompting to inject cognition from external sources, largely enhancing dissonance detection. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper creates a special test to see how well big computers understand when someone gives them mixed-up instructions. Right now, these computers are very good at doing what people tell them, but they get confused if the instructions don’t make sense. The test has 20,000 examples of mixed-up instructions and shows that these computers struggle to figure out when something doesn’t add up. To help with this problem, the authors suggest a way to give these computers more “common sense” so they can better understand when something is wrong. |
Keywords
» Artificial intelligence » Prompting
