DS-1000: A Natural and Reliable Benchmark for Data Science Code Generation

Yuhang Lai, Chengxi Li, Yiming Wang, Tianyi Zhang, Ruiqi Zhong, Luke Zettlemoyer, Wen-Tau Yih, Daniel Fried, Sida Wang, Tao Yu
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:18319-18345, 2023.

Abstract

We introduce DS-1000, a code generation benchmark with a thousand data science problems spanning seven Python libraries, such as Numpy and Pandas. Compared to prior works, DS-1000 incorporates three core features. First, our problems reflect diverse, realistic, and practical use cases since we collected them from StackOverflow. Second, our automatic evaluation is highly specific (reliable) – across all Codex-002-predicted solutions that our evaluation accepts, only 1.8% of them are incorrect; we achieve this with multi-criteria metrics, checking both functional correctness by running test cases and surface-form constraints by restricting API usages or keywords. Finally, we proactively defend against memorization by slightly modifying our problems to be different from the original StackOverflow source; consequently, models cannot answer them correctly by memorizing the solutions from pre-training. The current best public system (Codex-002) achieves 43.3% accuracy, leaving ample room for improvement. We release our benchmark at https://ds1000-code-gen.github.io.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-lai23b, title = {{DS}-1000: A Natural and Reliable Benchmark for Data Science Code Generation}, author = {Lai, Yuhang and Li, Chengxi and Wang, Yiming and Zhang, Tianyi and Zhong, Ruiqi and Zettlemoyer, Luke and Yih, Wen-Tau and Fried, Daniel and Wang, Sida and Yu, Tao}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {18319--18345}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/lai23b/lai23b.pdf}, url = {https://proceedings.mlr.press/v202/lai23b.html}, abstract = {We introduce DS-1000, a code generation benchmark with a thousand data science problems spanning seven Python libraries, such as Numpy and Pandas. Compared to prior works, DS-1000 incorporates three core features. First, our problems reflect diverse, realistic, and practical use cases since we collected them from StackOverflow. Second, our automatic evaluation is highly specific (reliable) – across all Codex-002-predicted solutions that our evaluation accepts, only 1.8% of them are incorrect; we achieve this with multi-criteria metrics, checking both functional correctness by running test cases and surface-form constraints by restricting API usages or keywords. Finally, we proactively defend against memorization by slightly modifying our problems to be different from the original StackOverflow source; consequently, models cannot answer them correctly by memorizing the solutions from pre-training. The current best public system (Codex-002) achieves 43.3% accuracy, leaving ample room for improvement. We release our benchmark at https://ds1000-code-gen.github.io.} }
Endnote
%0 Conference Paper %T DS-1000: A Natural and Reliable Benchmark for Data Science Code Generation %A Yuhang Lai %A Chengxi Li %A Yiming Wang %A Tianyi Zhang %A Ruiqi Zhong %A Luke Zettlemoyer %A Wen-Tau Yih %A Daniel Fried %A Sida Wang %A Tao Yu %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-lai23b %I PMLR %P 18319--18345 %U https://proceedings.mlr.press/v202/lai23b.html %V 202 %X We introduce DS-1000, a code generation benchmark with a thousand data science problems spanning seven Python libraries, such as Numpy and Pandas. Compared to prior works, DS-1000 incorporates three core features. First, our problems reflect diverse, realistic, and practical use cases since we collected them from StackOverflow. Second, our automatic evaluation is highly specific (reliable) – across all Codex-002-predicted solutions that our evaluation accepts, only 1.8% of them are incorrect; we achieve this with multi-criteria metrics, checking both functional correctness by running test cases and surface-form constraints by restricting API usages or keywords. Finally, we proactively defend against memorization by slightly modifying our problems to be different from the original StackOverflow source; consequently, models cannot answer them correctly by memorizing the solutions from pre-training. The current best public system (Codex-002) achieves 43.3% accuracy, leaving ample room for improvement. We release our benchmark at https://ds1000-code-gen.github.io.
APA
Lai, Y., Li, C., Wang, Y., Zhang, T., Zhong, R., Zettlemoyer, L., Yih, W., Fried, D., Wang, S. & Yu, T.. (2023). DS-1000: A Natural and Reliable Benchmark for Data Science Code Generation. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:18319-18345 Available from https://proceedings.mlr.press/v202/lai23b.html.

Related Material