metadata
language:
- en
license: mit
size_categories:
- n<1K
task_categories:
- translation
Dataset Card for InterCode-Corrections
This is a manually corrected version of the InterCode-Bash dataset, providing natural language prompts and Bash commands for the task of machine translation.
Dataset Details
Dataset Description
This dataset contains corrections for errors in the InterCode-Bash dataset. corrections.csv contains annotations for each error. final.csv contains the updated dataset with the corrections applied. The corrected dataset was used to create the test split of the NL2SH-ALFA dataset. For more information, please refer to the paper.
- Curated by: Anyscale Learning For All (ALFA) Group at MIT-CSAIL
- Language: English
- License: MIT License
Usage
This dataset is not intended for direct use. Instead, please use the NL2SH-ALFA dataset.
from datasets import load_dataset
train_dataset = load_dataset("westenfelder/NL2SH-ALFA", "train", split="train")
test_dataset = load_dataset("westenfelder/NL2SH-ALFA", "test", split="train")
Dataset Sources
- Repository: GitHub Repo
- Paper: LLM-Supported Natural Language to Bash Translation
Citation
BibTeX:
@misc{westenfelder2025llmsupportednaturallanguagebash,
title={LLM-Supported Natural Language to Bash Translation},
author={Finnian Westenfelder and Erik Hemberg and Miguel Tulla and Stephen Moskal and Una-May O'Reilly and Silviu Chiricescu},
year={2025},
eprint={2502.06858},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2502.06858},
}
Dataset Card Authors
Finn Westenfelder
Dataset Card Contact
Please email [email protected] or make a pull request.