Summary of Proactive Agent: Shifting Llm Agents From Reactive Responses to Active Assistance, by Yaxi Lu et al.
Proactive Agent: Shifting LLM Agents from Reactive Responses to Active Assistance
by Yaxi Lu, Shenzhi Yang, Cheng Qian, Guirong Chen, Qinyu Luo, Yesai Wu, Huadong Wang, Xin Cong, Zhong Zhang, Yankai Lin, Weiwen Liu, Yasheng Wang, Zhiyuan Liu, Fangming Liu, Maosong Sun
First submitted to arxiv on: 16 Oct 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Computation and Language (cs.CL)
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 paper tackles the challenge of developing proactive agents that can anticipate and initiate tasks without explicit human instructions. The authors propose a novel data-driven approach to achieve this goal. They start by collecting real-world human activities to generate task predictions, which are then labeled as accepted or rejected. This labeled data is used to train a reward model that simulates human judgment and evaluates the proactiveness of large language model (LLM) agents. The authors also develop a comprehensive data generation pipeline to create a diverse dataset called ProactiveBench, containing 6,790 events. They fine-tune models with this dataset and demonstrate significant improvements in the proactiveness of LLM agents. Experimental results show that their fine-tuned model achieves an F1-Score of 66.47% in proactively offering assistance, outperforming open-source and close-source models. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper creates proactive agents that can do tasks without being told what to do. The authors use real-life human activities to predict what needs to be done, then train a computer model to make good choices like humans do. They also create a big dataset of events called ProactiveBench and fine-tune models with it to make the agents better. This helps create more helpful and effective agent systems that work well with humans. |
Keywords
» Artificial intelligence » F1 score » Large language model
