Summary of Self-demos: Eliciting Out-of-demonstration Generalizability in Large Language Models, by Wei He et al.
Self-Demos: Eliciting Out-of-Demonstration Generalizability in Large Language Models
by Wei He, Shichun Liu, Jun Zhao, Yiwen Ding, Yi Lu, Zhiheng Xi, Tao Gui, Qi Zhang, Xuanjing Huang
First submitted to arxiv on: 1 Apr 2024
Categories
- Main: Computation and Language (cs.CL)
- 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 Large language models (LLMs) have demonstrated impressive capabilities in adapting to new tasks with few-shot demonstrations through in-context learning (ICL). However, current methods rely heavily on high-quality query-specific demos, which are often lacking. To address this limitation, we propose Self-Demos, a novel prompting method that elicits the inherent generalizability in LLMs by generating query-aware demos. The generated demos strategically interpolate between existing demos and the given query, effectively transforming the query from out-of-distribution (OOD) to in-distribution (ID). To evaluate our approach, we constructed OOD-Toolset, a dataset featuring real-world APIs, tool-using scenarios, and over 300 instances. Experiments on our dataset and two public math benchmarks show that Self-Demos can outperform state-of-the-art baselines in the OOD setting. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about a new way to make computers learn quickly by giving them examples of how to solve problems. Right now, this doesn’t always work because we need many specific examples to show the computer how to do something new. This makes it hard for the computer to understand when it sees something new that’s not exactly like what it was shown before. The researchers created a new way called Self-Demos that helps computers learn by creating their own examples, based on what they already know and what they’re trying to learn. They tested this method with many different scenarios and found that it can help computers do better when faced with new problems they haven’t seen before. |
Keywords
» Artificial intelligence » Few shot » Prompting
