westenfelder's picture
Updated README.md
f9ffe3a
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.

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

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.