metadata
dataset_info:
features:
- name: evalchemy_dataset
dtype: string
- name: model_outputs
sequence: string
- name: model_answers
sequence: string
- name: answer
dtype: string
- name: raw_data
struct:
- name: Canary String
dtype: string
- name: Correct Answer
dtype: string
- name: Expert Validator Accuracy
dtype: float64
- name: Expert Validator Disagreement Category
dtype: float64
- name: Expert Validator_EV_1
dtype: string
- name: Expert Validator_EV_2
dtype: string
- name: Explanation
dtype: string
- name: Explanation_NEV_1
dtype: string
- name: Explanation_NEV_2
dtype: string
- name: Explanation_NEV_3
dtype: string
- name: Extra Revised Correct Answer
dtype: string
- name: Extra Revised Explanation
dtype: string
- name: Extra Revised Incorrect Answer 1
dtype: string
- name: Extra Revised Incorrect Answer 2
dtype: string
- name: Extra Revised Incorrect Answer 3
dtype: string
- name: Extra Revised Question
dtype: string
- name: Feedback_EV_1
dtype: string
- name: Feedback_EV_2
dtype: string
- name: Feedback_NEV_1
dtype: string
- name: Feedback_NEV_2
dtype: string
- name: Feedback_NEV_3
dtype: string
- name: High-level domain
dtype: string
- name: Incorrect Answer 1
dtype: string
- name: Incorrect Answer 2
dtype: string
- name: Incorrect Answer 3
dtype: string
- name: Is First Validation_EV_1
dtype: bool
- name: Is First Validation_EV_2
dtype: bool
- name: Majority Non-Expert Vals Incorrect
dtype: float64
- name: Manual Correctness Adjustment_EV_1
dtype: string
- name: Manual Correctness Adjustment_EV_2
dtype: string
- name: Manual Correctness Adjustment_NEV_1
dtype: 'null'
- name: Manual Correctness Adjustment_NEV_2
dtype: 'null'
- name: Manual Correctness Adjustment_NEV_3
dtype: 'null'
- name: Non-Expert Validator Accuracy
dtype: float64
- name: Non-Expert Validator_NEV_1
dtype: string
- name: Non-Expert Validator_NEV_2
dtype: string
- name: Non-Expert Validator_NEV_3
dtype: string
- name: Post hoc agreement_EV_1
dtype: string
- name: Post hoc agreement_EV_2
dtype: string
- name: Pre-Revision Correct Answer
dtype: string
- name: Pre-Revision Explanation
dtype: string
- name: Pre-Revision Incorrect Answer 1
dtype: string
- name: Pre-Revision Incorrect Answer 2
dtype: string
- name: Pre-Revision Incorrect Answer 3
dtype: string
- name: Pre-Revision Question
dtype: string
- name: Probability Correct_EV_1
dtype: string
- name: Probability Correct_EV_2
dtype: string
- name: Probability Correct_NEV_1
dtype: string
- name: Probability Correct_NEV_2
dtype: string
- name: Probability Correct_NEV_3
dtype: string
- name: Question
dtype: string
- name: Question Difficulty_EV_1
dtype: string
- name: Question Difficulty_EV_2
dtype: string
- name: Question Writer
dtype: string
- name: Record ID
dtype: string
- name: Revision Comments (from Question Writer)
dtype: string
- name: Self-reported question-writing time (minutes)
dtype: float64
- name: Self-reported time (minutes)_EV_1
dtype: float64
- name: Self-reported time (minutes)_EV_2
dtype: float64
- name: Self-reported time (minutes)_NEV_1
dtype: float64
- name: Self-reported time (minutes)_NEV_2
dtype: float64
- name: Self-reported time (minutes)_NEV_3
dtype: float64
- name: Subdomain
dtype: string
- name: Sufficient Expertise?_EV_1
dtype: string
- name: Sufficient Expertise?_EV_2
dtype: string
- name: Understand the question?_EV_1
dtype: string
- name: Understand the question?_EV_2
dtype: string
- name: Validator Answered Correctly_EV_1
dtype: int64
- name: Validator Answered Correctly_EV_2
dtype: int64
- name: Validator Answered Correctly_NEV_1
dtype: int64
- name: Validator Answered Correctly_NEV_2
dtype: int64
- name: Validator Answered Correctly_NEV_3
dtype: float64
- name: Validator Revision Suggestion_EV_1
dtype: string
- name: Validator Revision Suggestion_EV_2
dtype: string
- name: Websites visited_NEV_1
dtype: string
- name: Websites visited_NEV_2
dtype: string
- name: Websites visited_NEV_3
dtype: string
- name: Writer's Difficulty Estimate
dtype: string
- name: answer
dtype: string
- name: model_answers
sequence: string
- name: model_outputs
sequence: string
splits:
- name: train
num_bytes: 1427226
num_examples: 198
download_size: 807354
dataset_size: 1427226
configs:
- config_name: default
data_files:
- split: train
path: data/train-*