metadata
annotations_creators:
- expert-generated
language_creators:
- found
language:
- en
license:
- apache-2.0
multilinguality:
- monolingual
pretty_name: WMDP-Chem Preprocessed
homepage: https://huggingface.co/datasets/cais/wmdp
tags:
- medical
- security
- multiple-choice
- reasoning
task_categories:
- question-answering
task_ids:
- multiple-choice-question-answering
WMDP-Chem Preprocessed Dataset
This dataset is a preprocessed version of wmdp-chem.
The data has been formatted into a question and answer structure suitable for training or evaluating instruction-following language models.
Data Structure
question: The original question text.answer: The correct answer text, prefixed with####.
Example
Question:
[Example Question Text]
Answer:
#### [Example Answer Text]
Splits
The original test split is preserved.
How to Use
from datasets import load_dataset
ds = load_dataset("daichira/wmdp_chem_preprocess", split="test")
print(ds[0])
Original Dataset
For more information, please refer to the original dataset card at wmdp-chem.