Summary of Selectllm: Can Llms Select Important Instructions to Annotate?, by Ritik Sachin Parkar et al.
SelectLLM: Can LLMs Select Important Instructions to Annotate?
by Ritik Sachin Parkar, Jaehyung Kim, Jong Inn Park, Dongyeop Kang
First submitted to arxiv on: 29 Jan 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 The paper introduces SelectLLM, a framework that leverages large language models (LLMs) to select unlabeled instructions for annotation. The authors propose a two-step approach: Coreset-based clustering of unlabelled instructions to increase diversity and prompting of LLMs to identify the most beneficial instructions within each cluster. They evaluate SelectLLM on AlpacaEval2 and MT-Bench, demonstrating its ability to outperform state-of-the-art methods like Alpagasus. The framework’s adaptability and robustness are further evidenced by its high performance across both human and synthetic datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary SelectLLM is a new way to choose which instructions don’t have labels yet. This can be helpful because it’s easier to get lots of unlabelled text from different sources than to label them all by hand. The authors want to know how to pick the best unlabelled instructions for LLMs, so they created a two-step process: grouping similar unlabelled texts together and asking an LLM to pick the most useful ones in each group. They tested SelectLLM on some datasets and found it works better than other methods. |
Keywords
» Artificial intelligence » Clustering » Prompting
