datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
vcell-perturbation-source-data
ConvergeCELL Source Datasets — v1.0.0
Cached H5ADs of every single-cell RNA-seq dataset registered in the
ConvergeCELL data catalog. Mirrors the original sources (GEO, figshare,
Tabula Sapiens, CellxGene) so downstream code has a single, fast, versioned
endpoint to fetch from.
Each row of every h5ad is one cell; each column is one gene (HGNC symbol).
The exact obs/var schema follows whatever the original source provided —
this bundle does not re-annotate, harmonize, or QC. For a… See the full description on the dataset page: https://huggingface.co/datasets/nicolas-lynn/vcell-perturbation-source-data.boolq-natural-perturbationsBoolQ questions with semantic alteration and human verifications
@article{khashabi2020naturalperturbations,
title={Natural Perturbation for Robust Question Answering},
author={D. Khashabi and T. Khot and A. Sabhwaral},
journal={arXiv preprint},
year={2020}
}
perturbed-knights-and-knaves
📘 perturbed-knights-and-knaves Dataset [Project Page]
The perturbed-knights-and-knaves dataset evaluates the consistency of LLMs' logical reasoning ability under various perturbations.
🚀🚀 Check out the clean version of the dataset at [knights-and-knaves].
Loading the dataset
To load the dataset:
from datasets import load_dataset
data_subject = datasets.load_dataset('K-and-K/perturbed-knights-and-knaves', data_files="{subset}/{perturbation}/{subject}.jsonl")… See the full description on the dataset page: https://huggingface.co/datasets/K-and-K/perturbed-knights-and-knaves.pi05-libero-plus-perturbation-summary
pi0.5 LIBERO-plus Perturbation Summary
This dataset stores the summary report for six completed LIBERO-plus perturbation evaluations of TensorAuto/tPi0.5-libero.
Files:
pi05_libero_plus_six_perturb_hf_report.md: human-readable report with Hugging Face links, success rates, and short analysis.
pi05_libero_plus_six_perturb_hf_report.json: machine-readable summary.
The failure-grid videos and per-category metadata are stored in the linked per-perturbation datasets.
repro-a-perturbation-approach-to-unconstrained-linear-bandits
A Perturbation Approach to Unconstrained Linear Bandits
This is a reproduction logbook for ICML 2026.
OpenReview ID: XSpBSHzJAg
Paper Abstract
This logbook reproduces the PABLO bandit algorithm regret bounds.
See logbook.json for full claim verification details.
text_perturbationThis dataset is adapted from other datasets on Huggingface, such as the Amazon review dataset provided here https://huggingface.co/datasets/hugginglearners/amazon-reviews-sentiment-analysis, and the IMDb movie review dataset provided here https://huggingface.co/datasets/noob123/imdb_review_3000.
The files that start with "ori" are the original raw data, and the files that start with "trans" are the perturbed text data.
perturb-data-lab-demo
perturb-data-lab demo subsets
Generated: 2026-06-25 01:30:20 UTC
This Phase 2 package contains two small real .h5ad demo subsets prepared from the full Marson and Xorion source datasets for the perturb-data-lab public-demo plan.
Included datasets
Dataset
Cells
Features
H5AD
Reviewed schema copy
Notes
marson_d2_rest
2720
18130
h5ad/demo_marson_d2_rest.h5ad
schemas/marson_d2_rest.final-schema.yaml
schema_decisions/marson_d2_rest.md… See the full description on the dataset page: https://huggingface.co/datasets/weililab/perturb-data-lab-demo.fatima-audio-perturbations
Audio Perturbation TTS Gold Sentences
A curated set of English sentences for perturbation-based blind-spot evaluation of audio-LLM judges on synthesised speech.
Overview
This dataset provides original (clean) sentences sampled from established TTS benchmarks. The sentences are designed to be fed through TTS models to generate clean audio (A_gold), then perturbed at the text level (S_gold → S_pert) and re-synthesised (A_pert) to test whether audio-LLM judges can… See the full description on the dataset page: https://huggingface.co/datasets/Mawube/fatima-audio-perturbations.preference-model-perturbations
preference-model-perturbations
A Hugging Face dataset of paired model responses (original vs.
counterfactually perturbed) along with human and reward-model preferences,
generated by a Counterfactual Data Augmentation (CDA) pipeline to
analyze and mitigate bias in preference models.
Links
Homepage: CDA Pipeline Code
Description
Each record contains:
bias: type of bias expressed in the perturbation (5 possible values).
query: the original user prompt or query.… See the full description on the dataset page: https://huggingface.co/datasets/abharadwaj123/preference-model-perturbations.eventx-recognition-perturbed-gpttimex-recognition-sentence-perturbed-gptPerturbReason_dataset_codegsm-8k-perturbtlink-extr-classification-sentence-perturbedtimex-perturbedtlink-textbugger-perturbedqwen_rollout_frames_pi05_perturb_top5_fixedeventx-recognition-perturbedtlink-a2i-perturbed