Table of Links

Abstract and Intro

Dataset Design

Evaluation Methodology

Experiments and Analysis

Related Work

Conclusion, Reproducibility, and References

A. Complete list of tools

B. Scenario Prompt

C. Unrealistic Queries

D. Nuances comparing prior work

6 CONCLUSION

We present ToolTalk, a new benchmark for evaluating tool-augmented LLMs in a conversational setting. Our benchmark emphasizes complex orchestration of multiple tools in a conversational setting. We provide simulated implementations of all tools, allowing for a fully automated evaluation where the LLM can decide which tools to further invoke based on the results of prior tool calls. Finally, we also introduce a unique form of evaluating correctness that takes into account unique aspects of individual tools and whether a tool usage system produces incorrect actions. We evaluate GPT-3.5 and GPT-4 using our dataset and methodology and analyze their errors, finding three major categories: premature tool calls, faulty reasoning, and incorrect invocations of the correct tool. In the future, we hope to expand the scope of this dataset to more conversations and simulate even more, diverse plugins. We also hope to see future research look into how to better redesign existing API interfaces for LLMs.

7 REPRODUCIBILITY

We make ToolTalk more widely available by releasing it on github[5]. We include the exact versions of GPT-3.5 (gpt-3.5-turbo-0613) and GPT-4 (gpt-4-0613) available through the OpenAI API to be able to reproduce our results after release. We include the prompt used to generate our scenarios in Appendix B. We include information on system prompts and our application of OpenAI’s Chat completions API in Section 4.1.

REFERENCES

Paweł Budzianowski, Tsung-Hsien Wen, Bo-Hsiang Tseng, Inigo Casanueva, Stefan Ultes, Osman ˜ Ramadan, and Milica Gasic. Multiwoz - a large-scale multi-domain wizard-of-oz dataset for task-oriented dialogue modelling. In Conference on Empirical Methods in Natural Language Processing, 2018. Bill Byrne, Karthik Kri

Bill Byrne, Karthik Krishnamoorthi, Chinnadhurai Sankar, Arvind Neelakantan, Daniel Duckworth, Semih Yavuz, Ben Goodrich, Amit Dubey, Andy Cedilnik, and Kyu-Young Kim. Taskmaster-1: Toward a realistic and diverse dialog dataset. ArXiv, abs/1909.05358, 2019.

Bill Byrne, Karthik Krishnamoorthi, Saravanan Ganesh, and Mihir Kale. Tickettalk: Toward humanlevel performance with end-to-end, transaction-based dialog systems. ArXiv, abs/2012.12458, 2020.

Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. Training verifiers to solve math word problems, 2021.

Xiang Deng, Yu Gu, Bo Zheng, Shijie Chen, Samuel Stevens, Boshi Wang, Huan Sun, and Yu Su. Mind2web: Towards a generalist agent for the web. ArXiv, abs/2306.06070, 2023.

Shibo Hao, Tianyang Liu, Zhen Wang, and Zhiting Hu. Toolkengpt: Augmenting frozen language models with massive tools via tool embeddings, 2023.

Stefan Larson and Kevin Leach. A survey of intent classification and slot-filling datasets for taskoriented dialog. ArXiv, abs/2207.13211, 2022.

Patrick Lewis, Barlas Oguz, Ruty Rinott, Sebastian Riedel, and Holger Schwenk. Mlqa: Evaluating ˘ cross-lingual extractive question answering, 2020.

Minghao Li, Feifan Song, Bowen Yu, Haiyang Yu, Zhoujun Li, Fei Huang, and Yongbin Li. Apibank: A benchmark for tool-augmented llms. ArXiv, abs/2304.08244, 2023.

Jacky Liang, Wenlong Huang, Fei Xia, Peng Xu, Karol Hausman, Brian Ichter, Pete Florence, and Andy Zeng. Code as policies: Language model programs for embodied control, 2023.

Xiao Liu, Hao Yu, Hanchen Zhang, Yifan Xu, Xuanyu Lei, Hanyu Lai, Yu Gu, Yuxian Gu, Hangliang Ding, Kai Men, Kejuan Yang, Shudan Zhang, Xiang Deng, Aohan Zeng, Zhengxiao Du, Chenhui Zhang, Shengqi Shen, Tianjun Zhang, Yu Su, Huan Sun, Minlie Huang, Yuxiao Dong, and Jie Tang. Agentbench: Evaluating llms as agents. ArXiv, abs/2308.03688, 2023a.

Zhiwei Liu, Weiran Yao, Jianguo Zhang, Le Xue, Shelby Heinecke, Rithesh Murthy, Yihao Feng, Zeyuan Chen, Juan Carlos Niebles, Devansh Arpit, Ran Xu, Ph´ı Thi Mui, Haiquan Wang, Caim- ` ing Xiong, and Silvio Savarese. Bolaa: Benchmarking and orchestrating llm-augmented autonomous agents. ArXiv, abs/2308.05960, 2023b.

Pan Lu, Baolin Peng, Hao Cheng, Michel Galley, Kai-Wei Chang, Ying Nian Wu, Song-Chun Zhu, and Jianfeng Gao. Chameleon: Plug-and-play compositional reasoning with large language models. ArXiv, abs/2304.09842, 2023a.

Pan Lu, Liang Qiu, Kai-Wei Chang, Ying Nian Wu, Song-Chun Zhu, Tanmay Rajpurohit, Peter Clark, and Ashwin Kalyan. Dynamic prompt learning via policy gradient for semi-structured mathematical reasoning, 2023b.

Gregoire Mialon, Roberto Dess ´ `ı, Maria Lomeli, Christoforos Nalmpantis, Ram Pasunuru, Roberta Raileanu, Baptiste Roziere, Timo Schick, Jane Dwivedi-Yu, Asli Celikyilmaz, Edouard Grave, ` Yann LeCun, and Thomas Scialom. Augmented language models: a survey, 2023.

Johannes E. M. Mosig, Shikib Mehri, and Thomas Kober. Star: A schema-guided dialog dataset for transfer learning. ArXiv, abs/2010.11853, 2020.

Reiichiro Nakano, Jacob Hilton, Suchir Balaji, Jeff Wu, Long Ouyang, Christina Kim, Christopher Hesse, Shantanu Jain, Vineet Kosaraju, William Saunders, Xu Jiang, Karl Cobbe, Tyna Eloundou, Gretchen Krueger, Kevin Button, Matthew Knight, Benjamin Chess, and John Schulman. Webgpt: Browser-assisted question-answering with human feedback, 2022.

OpenAI. Function calling and other api updates.

Bhargavi Paranjape, Scott M. Lundberg, Sameer Singh, Hanna Hajishirzi, Luke Zettlemoyer, and Marco Tulio Ribeiro. Art: Automatic multi-step reasoning and tool-use for large language models. ArXiv, abs/2303.09014, 2023.

Shishir G. Patil, Tianjun Zhang, Xin Wang, and Joseph E. Gonzalez. Gorilla: Large language model connected with massive apis. ArXiv, abs/2305.15334, 2023.

Shuofei Qiao, Honghao Gui, Huajun Chen, and Ningyu Zhang. Making language models better tool learners with execution feedback. ArXiv, abs/2305.13068, 2023.

Yujia Qin, Shengding Hu, Yankai Lin, Weize Chen, Ning Ding, Ganqu Cui, Zheni Zeng, Yufei Huang, Chaojun Xiao, Chi Han, Yi Ren Fung, Yusheng Su, Huadong Wang, Cheng Qian, Runchu Tian, Kunlun Zhu, Shi Liang, Xingyu Shen, Bokai Xu, Zhen Zhang, Yining Ye, Bo Li, Ziwei Tang, Jing Yi, Yu Zhu, Zhenning Dai, Lan Yan, Xin Cong, Ya-Ting Lu, Weilin Zhao, Yuxiang Huang, Jun-Han Yan, Xu Han, Xian Sun, Dahai Li, Jason Phang, Cheng Yang, Tongshuang Wu, Heng Ji, Zhiyuan Liu, and Maosong Sun. Tool learning with foundation models. ArXiv, abs/2304.08354, 2023a.

Yujia Qin, Shi Liang, Yining Ye, Kunlun Zhu, Lan Yan, Ya-Ting Lu, Yankai Lin, Xin Cong, Xiangru Tang, Bill Qian, Sihan Zhao, Runchu Tian, Ruobing Xie, Jie Zhou, Marc H. Gerstein, Dahai Li, Zhiyuan Liu, and Maosong Sun. Toolllm: Facilitating large language models to master 16000+ real-world apis. ArXiv, abs/2307.16789, 2023b.

Jingqing Ruan, Yihong Chen, Bin Zhang, Zhiwei Xu, Tianpeng Bao, Guoqing Du, Shiwei Shi, Hangyu Mao, Xingyu Zeng, and Rui Zhao. Tptu: Task planning and tool usage of large language model-based ai agents. ArXiv, abs/2308.03427, 2023.

Tanik Saikh, Tirthankar Ghosal, Amish Mittal, Asif Ekbal, and Pushpak Bhattacharyya. Scienceqa: a novel resource for question answering on scholarly articles. International Journal on Digital Libraries, 23:289 – 301, 2022.

Carolina Scarton, Mikel L. Forcada, Miquel Espla-Gomis, and Lucia Specia. Estimating post-editing ` effort: a study on human judgements, task-based and reference-based metrics of mt quality, 2019.

Timo Schick, Jane Dwivedi-Yu, Roberto Dess`ı, Roberta Raileanu, Maria Lomeli, Luke Zettlemoyer, Nicola Cancedda, and Thomas Scialom. Toolformer: Language models can teach themselves to use tools. ArXiv, abs/2302.04761, 2023.

Mohit Shridhar, Xingdi Yuan, Marc-Alexandre Cotˆ e, Yonatan Bisk, Adam Trischler, and Matthew J. ´ Hausknecht. Alfworld: Aligning text and embodied environments for interactive learning. ArXiv, abs/2010.03768, 2020.

Yifan Song, Weimin Xiong, Dawei Zhu, Wenhao Wu, Han Qian, Mingbo Song, Hailiang Huang, Chengzu Li, Ke Wang, Rong Yao, Ye Tian, and Sujian Li. Restgpt: Connecting large language models with real-world restful apis. 2023.

Alon Talmor, Jonathan Herzig, Nicholas Lourie, and Jonathan Berant. Commonsenseqa: A question answering challenge targeting commonsense knowledge, 2019.

Qiaoyu Tang, Ziliang Deng, Hongyu Lin, Xianpei Han, Qiao Liang, and Le Sun. Toolalpaca: Generalized tool learning for language models with 3000 simulated cases. ArXiv, abs/2306.05301, 2023.

Qiantong Xu, Fenglu Hong, B. Li, Changran Hu, Zhe Chen, and Jian Zhang. On the tool manipulation capability of open-source large language models. ArXiv, abs/2305.16504, 2023.

Rui Yang, Lin Song, Yanwei Li, Sijie Zhao, Yixiao Ge, Xiu Li, and Ying Shan. Gpt4tools: Teaching large language model to use tools via self-instruction. ArXiv, abs/2305.18752, 2023.

Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William W. Cohen, Ruslan Salakhutdinov, and Christopher D. Manning. Hotpotqa: A dataset for diverse, explainable multi-hop question answering. In Conference on Empirical Methods in Natural Language Processing, 2018.

Shunyu Yao, Howard Chen, John Yang, and Karthik Narasimhan. Webshop: Towards scalable real-world web interaction with grounded language agents. ArXiv, abs/2207.01206, 2022a.

Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. React: Synergizing reasoning and acting in language models. ArXiv, abs/2210.03629, 2022b.

Jeffrey Zhao, Yuan Cao, Raghav Gupta, Harrison Lee, Abhinav Rastogi, Mingqiu Wang, Hagen Soltau, Izhak Shafran, and Yonghui Wu. Anytod: A programmable task-oriented dialog system. ArXiv, abs/2212.09939, 2022.

Fengbin Zhu, Wenqiang Lei, Chao Wang, Jianming Zheng, Soujanya Poria, and Tat-Seng Chua. Retrieving and reading: A comprehensive survey on open-domain question answering, 2021.

Yuchen Zhuang, Yue Yu, Kuan Wang, Haotian Sun, and Chao Zhang. Toolqa: A dataset for llm question answering with external tools. ArXiv, abs/2306.13304, 2023.

[5] https://github.com/microsoft/ToolTalk