dataset_info:
- config_name: ACSEmployment
features:
- name: id
dtype: int64
- name: description
dtype: string
- name: instruction
dtype: string
- name: question
dtype: string
- name: choices
sequence: string
length: 2
- name: answer
dtype: string
- name: answer_key
dtype: string
- name: choice_question_prompt
dtype: string
- name: numeric_question
dtype: string
- name: label
dtype: int64
- name: numeric_question_prompt
dtype: string
- name: AGEP
dtype: int64
- name: SCHL
dtype: float64
- name: MAR
dtype: int64
- name: RELP
dtype: int64
- name: DIS
dtype: int64
- name: ESP
dtype: float64
- name: CIT
dtype: int64
- name: MIG
dtype: float64
- name: MIL
dtype: float64
- name: ANC
dtype: int64
- name: NATIVITY
dtype: int64
- name: DEAR
dtype: int64
- name: DEYE
dtype: int64
- name: DREM
dtype: float64
- name: SEX
dtype: int64
- name: RAC1P
dtype: int64
splits:
- name: train
num_bytes: 4216821661
num_examples: 2588885
- name: validation
num_bytes: 527108317
num_examples: 323611
- name: test
num_bytes: 527106993
num_examples: 323611
download_size: 326506898
dataset_size: 5271036971
- config_name: ACSIncome
features:
- name: id
dtype: int64
- name: description
dtype: string
- name: instruction
dtype: string
- name: question
dtype: string
- name: choices
sequence: string
length: 2
- name: answer
dtype: string
- name: answer_key
dtype: string
- name: choice_question_prompt
dtype: string
- name: numeric_question
dtype: string
- name: label
dtype: int64
- name: numeric_question_prompt
dtype: string
- name: AGEP
dtype: int64
- name: COW
dtype: int64
- name: SCHL
dtype: int64
- name: MAR
dtype: int64
- name: OCCP
dtype: int64
- name: POBP
dtype: int64
- name: RELP
dtype: int64
- name: WKHP
dtype: int64
- name: SEX
dtype: int64
- name: RAC1P
dtype: int64
splits:
- name: train
num_bytes: 1667556978
num_examples: 1331600
- name: validation
num_bytes: 208460485
num_examples: 166450
- name: test
num_bytes: 208446084
num_examples: 166450
download_size: 155811003
dataset_size: 2084463547
- config_name: ACSMobility
features:
- name: id
dtype: int64
- name: description
dtype: string
- name: instruction
dtype: string
- name: question
dtype: string
- name: choices
sequence: string
length: 2
- name: answer
dtype: string
- name: answer_key
dtype: string
- name: choice_question_prompt
dtype: string
- name: numeric_question
dtype: string
- name: label
dtype: int64
- name: numeric_question_prompt
dtype: string
- name: AGEP
dtype: int64
- name: SCHL
dtype: float64
- name: MAR
dtype: int64
- name: SEX
dtype: int64
- name: DIS
dtype: int64
- name: ESP
dtype: float64
- name: CIT
dtype: int64
- name: MIL
dtype: float64
- name: ANC
dtype: int64
- name: NATIVITY
dtype: int64
- name: RELP
dtype: int64
- name: DEAR
dtype: int64
- name: DEYE
dtype: int64
- name: DREM
dtype: float64
- name: RAC1P
dtype: int64
- name: GCL
dtype: float64
- name: COW
dtype: float64
- name: ESR
dtype: float64
- name: WKHP
dtype: float64
- name: JWMNP
dtype: float64
- name: PINCP
dtype: float64
splits:
- name: train
num_bytes: 1028069699
num_examples: 496749
- name: validation
num_bytes: 128514698
num_examples: 62094
- name: test
num_bytes: 128509092
num_examples: 62094
download_size: 86257467
dataset_size: 1285093489
- config_name: ACSPublicCoverage
features:
- name: id
dtype: int64
- name: description
dtype: string
- name: instruction
dtype: string
- name: question
dtype: string
- name: choices
sequence: string
length: 2
- name: answer
dtype: string
- name: answer_key
dtype: string
- name: choice_question_prompt
dtype: string
- name: numeric_question
dtype: string
- name: label
dtype: int64
- name: numeric_question_prompt
dtype: string
- name: AGEP
dtype: int64
- name: SCHL
dtype: float64
- name: MAR
dtype: int64
- name: SEX
dtype: int64
- name: DIS
dtype: int64
- name: ESP
dtype: float64
- name: CIT
dtype: int64
- name: MIG
dtype: float64
- name: MIL
dtype: float64
- name: ANC
dtype: int64
- name: NATIVITY
dtype: int64
- name: DEAR
dtype: int64
- name: DEYE
dtype: int64
- name: DREM
dtype: float64
- name: PINCP
dtype: float64
- name: ESR
dtype: float64
- name: ST
dtype: int64
- name: FER
dtype: float64
- name: RAC1P
dtype: int64
splits:
- name: train
num_bytes: 1775349595
num_examples: 910631
- name: validation
num_bytes: 221922965
num_examples: 113829
- name: test
num_bytes: 221932893
num_examples: 113829
download_size: 138771081
dataset_size: 2219205453
- config_name: ACSTravelTime
features:
- name: id
dtype: int64
- name: description
dtype: string
- name: instruction
dtype: string
- name: question
dtype: string
- name: choices
sequence: string
length: 2
- name: answer
dtype: string
- name: answer_key
dtype: string
- name: choice_question_prompt
dtype: string
- name: numeric_question
dtype: string
- name: label
dtype: int64
- name: numeric_question_prompt
dtype: string
- name: AGEP
dtype: int64
- name: SCHL
dtype: float64
- name: MAR
dtype: int64
- name: SEX
dtype: int64
- name: DIS
dtype: int64
- name: ESP
dtype: float64
- name: MIG
dtype: float64
- name: RELP
dtype: int64
- name: RAC1P
dtype: int64
- name: PUMA
dtype: int64
- name: ST
dtype: int64
- name: CIT
dtype: int64
- name: OCCP
dtype: float64
- name: JWTR
dtype: float64
- name: POWPUMA
dtype: float64
- name: POVPIP
dtype: float64
splits:
- name: train
num_bytes: 2039450205
num_examples: 1173318
- name: validation
num_bytes: 254946722
num_examples: 146665
- name: test
num_bytes: 254948776
num_examples: 146665
download_size: 215781517
dataset_size: 2549345703
configs:
- config_name: ACSEmployment
data_files:
- split: train
path: ACSEmployment/train-*
- split: validation
path: ACSEmployment/validation-*
- split: test
path: ACSEmployment/test-*
- config_name: ACSIncome
data_files:
- split: train
path: ACSIncome/train-*
- split: validation
path: ACSIncome/validation-*
- split: test
path: ACSIncome/test-*
default: true
- config_name: ACSMobility
data_files:
- split: train
path: ACSMobility/train-*
- split: validation
path: ACSMobility/validation-*
- split: test
path: ACSMobility/test-*
- config_name: ACSPublicCoverage
data_files:
- split: train
path: ACSPublicCoverage/train-*
- split: validation
path: ACSPublicCoverage/validation-*
- split: test
path: ACSPublicCoverage/test-*
- config_name: ACSTravelTime
data_files:
- split: train
path: ACSTravelTime/train-*
- split: validation
path: ACSTravelTime/validation-*
- split: test
path: ACSTravelTime/test-*
license: mit
task_categories:
- question-answering
- text-classification
- feature-extraction
- zero-shot-classification
language:
- en
pretty_name: Folktexts real-world unrealizable classification tasks
size_categories:
- 1M<n<10M
paperswithcode_id: folktexts
Dataset Card for folktexts
Folktexts is a suite of Q&A datasets with natural outcome uncertainty, aimed at evaluating LLMs' calibration on unrealizable tasks.
The folktexts datasets are derived from US Census data products. Namely, the datasets made available here are derived from the 2018 Public Use Microdata Sample (PUMS). Individual features are mapped to natural text using the respective codebook. Each task relates to predicting different individual characteristics (e.g., income, employment) from a set of demographic features (e.g., age, race, education, occupation).
Importantly, every task has natural outcome uncertainty. That is, in general, the features describing each row do not uniquely determine the task's label. For calibrated models to perform well on this task, the model must correctly output nuanced scores between 0 and 1, instead of simply outputting discrete labels 0 or 1.
Namely, we make available the following tasks in natural language Q&A format:
ACSIncome: Predict whether a working adult earns above $50,000 yearly.ACSEmployment: Predict whether an adult is an employed civilian.ACSPublicCoverage: Predict individual public health insurance coverage.ACSMobility: Predict whether an individual changed address within the last year.ACSTravelTime: Predict whether an employed adult has a work commute time longer than 20 minutes.
These tasks follow the same naming and feature/target columns as the folktables tabular datasets proposed by Ding et al. (2021). The folktables tabular datasets have seen prevalent use in the algorithmic fairness and distribution shift communities. We make available natural language Q&A versions of these tasks.
The datasets are made available in standard multiple-choice Q&A format (columns
question, choices, answer, answer_key, and choice_question_prompt), as
well as in numeric Q&A format (columns numeric_question,
numeric_question_prompt, and label).
The numeric prompting (also known as verbalized prompting) is known to improve
calibration of zero-shot LLM risk scores
[Tian et al., EMNLP 2023;
Cruz et al., NeurIPS 2024].
The accompanying folktexts python package
eases customization, evaluation, and benchmarking with these datasets.
Table of contents:
Dataset Details
Dataset Description
- Language(s) (NLP): English
- License: Code is licensed under the MIT license; Data license is governed by the U.S. Census Bureau terms of service.
Dataset Sources
- Repository: https://github.com/socialfoundations/folktexts
- Paper: https://arxiv.org/pdf/2407.14614
- Data source: 2018 American Community Survey Public Use Microdata Sample
Uses
The datasets were originally used to evaluate LLMs' ability to produce calibrated and accurate risk scores in the Cruz et al. (2024) paper.
Other potential uses include evaluating the fairness of LLMs' decisions,
as individual rows feature protected demographic attributes such as sex and
race.
Dataset Structure
Description of dataset columns:
id: A unique row identifier.description: A textual description of an individual's features, following a bulleted-list format.instruction: The instruction used for zero-shot LLM prompting (should be pre-appended to the row description).question: A question relating to the task's target column.choices: A list of two answer options relating to the above question.answer: The correct answer from the above list of answer options.answer_key: The correct answer key; i.e.,Afor the first choice, orBfor the second choice.choice_question_prompt: The full multiple-choice Q&A text string used for LLM prompting.numeric_question: A version of the question that prompts for a numeric output instead of a discrete choice output.label: The task's label. This is the correct output to the above numeric question.numeric_question_prompt: The full numeric Q&A text string used for LLM prompting.<tabular-columns>: All other columns correspond to the tabular features in this task. Each of these features will also appear in text form on the above description column.
The dataset was randomly split in training, test, and validation data,
following an 80%/10%/10% split.
Only the test split should be used to evaluate zero-shot LLM performance.
The training split can be used for fine-tuning, or for fitting traditional
supervised ML models on the tabular columns for metric baselines.
The validation split should be used for hyperparameter tuning, feature
engineering or any other model improvement loop.
Dataset Creation
Source Data
The datasets are based on publicly available data from the American Community Survey (ACS) Public Use Microdata Sample (PUMS), namely the 2018 ACS 1-year PUMS files.
Data Collection and Processing
The categorical values were mapped to meaningful natural language
representations using the folktexts package, which in turn uses the official
ACS PUMS codebook.
The data download and processing was aided by the folktables python package,
which in turn uses the official US Census web API.
Who are the source data producers?
U.S. Census Bureau.
Citation
If you find this useful in your research, please consider citing the following paper:
@inproceedings{
cruz2024evaluating,
title={Evaluating language models as risk scores},
author={Andr{\'e} F Cruz and Moritz Hardt and Celestine Mendler-D{\"u}nner},
booktitle={The Thirty-eight Conference on Neural Information Processing Systems Datasets and Benchmarks Track},
year={2024},
url={https://openreview.net/forum?id=qrZxL3Bto9}
}
More Information
More information is available in the folktexts package repository
and the accompanying paper.