Twin-2K-500: A dataset for building digital twins of over 2,000 people based on their answers to over 500 questions
Paper • 2505.17479 • Published
Dataset Viewer
The dataset viewer is not available for this subset.
Cannot get the split names for the config 'default' of the dataset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
for split_generator in builder._split_generators(
~~~~~~~~~~~~~~~~~~~~~~~~~^
StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/webdataset/webdataset.py", line 91, in _split_generators
inferred_arrow_schema = pa.concat_tables(pa_tables, promote_options="default").schema
~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "pyarrow/table.pxi", line 6320, in pyarrow.lib.concat_tables
File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
return check_status(status)
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
raise convert_status(status)
pyarrow.lib.ArrowInvalid: Must pass at least one table
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 66, in compute_split_names_from_streaming_response
for split in get_dataset_split_names(
~~~~~~~~~~~~~~~~~~~~~~~^
path=dataset,
^^^^^^^^^^^^^
config_name=config,
^^^^^^^^^^^^^^^^^^^
token=hf_token,
^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
info = get_dataset_config_info(
path,
...<6 lines>...
**config_kwargs,
)
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Twin-2K-500 Simulation Inputs and Outputs
This dataset contains the text-simulation input prompts and LLM output responses produced with the code in Digital-Twin-Simulation, the repository accompanying Twin-2K-500: A dataset for building digital twins of over 2,000 people based on their answers to over 500 questions (arXiv:2505.17479).
It is a companion artifact to the source survey dataset
LLM-Digital-Twin/Twin-2K-500: the files
here are the materialized prompts fed to the models (input) and the model responses (output)
for the simulation experiments, not the raw survey responses.
Contents
The data is stored as one gzipped tar archive per experiment folder.
input/ 267 archives (~236 MB) — prompts fed to the models
output/ 68 archives (~1.0 GB) — model responses per run/config
Total uncompressed size is approximately 8.2 GB.
Persona-structure naming
Archive names encode the persona representation used to build the prompt:
| Token | Meaning |
|---|---|
raw |
Raw survey transcript, no restructuring |
unstr |
Unstructured persona summary |
bde |
BDE-structured persona (hand-crafted behavioral structure) |
Additional tokens in a name (e.g. megastudy_*, _o4mini, _gpt4o, _round{N}, _p21_50,
_temp0) denote the experiment, model, round, participant slice, or temperature setting.
How to download
CLI:
pip install -U huggingface_hub
hf download Mike-deng-2002/Twin-2K-500-simulations --repo-type dataset --local-dir twin_sim_data
cd twin_sim_data
for f in input/*.tar.gz output/*.tar.gz; do tar -xzf "$f"; done
Python:
from huggingface_hub import snapshot_download
snapshot_download("Mike-deng-2002/Twin-2K-500-simulations",
repo_type="dataset", local_dir="twin_sim_data")
Each archive extracts to a text_simulation_input_*/ or text_simulation_output_*/ folder of
per-participant files, matching the directory layout expected by the simulation code.
Citation
@article{toubia2025twin2k500,
title = {Twin-2K-500: A dataset for building digital twins of over 2,000 people based on their answers to over 500 questions},
author = {Toubia, Olivier and Gui, George Z. and Peng, Tianyi and Merlau, Daniel J. and Li, Ang and Chen, Haozhe},
journal = {arXiv preprint arXiv:2505.17479},
year = {2025}
}
- Downloads last month
- 13