Summary of Citygpt: Empowering Urban Spatial Cognition Of Large Language Models, by Jie Feng et al.
CityGPT: Empowering Urban Spatial Cognition of Large Language Models
by Jie Feng, Yuwei Du, Tianhui Liu, Siqi Guo, Yuming Lin, Yong Li
First submitted to arxiv on: 20 Jun 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Computation and Language (cs.CL); Machine Learning (cs.LG)
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 proposes CityGPT, a framework to enhance large language models’ (LLMs) capabilities in understanding urban spaces and solving related tasks. The authors construct a diverse instruction tuning dataset called CityInstruction to inject urban knowledge and improve spatial reasoning. They fine-tune various LLMs, such as ChatGLM3-6B and LLama3 series, using a mixture of CityInstruction and general instructions. To evaluate the effectiveness of their methods, they create a comprehensive benchmark called CityEval, which assesses LLMs’ performance on diverse urban scenarios. The results show that small LLMs trained with CityInstruction can achieve competitive performance with commercial LLMs in the CityEval evaluation. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps machines understand cities better and solve city-related problems. They create a special dataset to teach machines about cities and make them smarter. By using this dataset, they train machines to be good at understanding cities. They also test these machines on many different city-related tasks and show that even small machines can do well if they’re trained right. |
Keywords
» Artificial intelligence » Instruction tuning
