datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
ct-rate-medgemma-readyctrldataset2026
MonitoringBench
A benchmark for evaluating LLM-based monitors of agentic AI systems. Contains
2,644 successful attack trajectories in which an AI agent accomplished one of four harmful side tasks
(sudo escalation, firewall disabling, malware download, password leaking) in
a sandboxed Linux environment under the control_arena framework.
Each trajectory is scored by a panel of 13+ LLM monitors (GPT-3.5 / 4.x / 5.x,
Claude Opus 4.x and Sonnet 4.x, o3, o4-mini, gpt-5-nano), with both… See the full description on the dataset page: https://huggingface.co/datasets/neur26anonsub/ctrldataset2026.ctrl-shift
Dataset Card for Control+Shift: Generating Controllable Distribution Shifts
[arXiv], [GitHub]
Curated by: Roy Friedman and Rhea Chowers
This dataset is the one that accompanies the paper Control+Shift: Generating Controllable Distribution Shifts.
Our data is based on CIFAR10 and ImageNet, using EDM to generate our data. We generated datasets for 3 types of distribution shift on CIFAR10 and ImageNet - so a total of 6 datasets. The types of distribution shifts are called overlap… See the full description on the dataset page: https://huggingface.co/datasets/friedmanroy/ctrl-shift.CT-RATE-AB
CT-RATE-AB
Vision-language annotations for chest CT abnormality reporting, derived from
CT-RATE and
formatted in the LLaVA conversation schema. Includes both an SFT split
(train / valid) and a DPO preference set.
⚠️ Research use only. Not a medical device. Do not use for clinical decisions.
Splits
Subset
# Samples
Purpose
train
46,709
SFT training
valid
3,039
Validation
DPO
46,709
DPO preference fine-tuning
The DPO entries share id values with the… See the full description on the dataset page: https://huggingface.co/datasets/yw3325/CT-RATE-AB.headlines-ctr
Headlines CTR Dataset
This dataset contains pairs of news headlines with labels indicating which headline received more clicks. It's designed for studying what makes headlines engaging and for training models to predict user preferences.
Dataset Description
Each example contains two competing headlines (A and B) that were shown to users, along with engagement metrics and a binary label indicating which performed better.
Dataset Statistics
Train: 8,781 headline… See the full description on the dataset page: https://huggingface.co/datasets/Yanjo/headlines-ctr.ctr-prediction-datasetCTRL-RAW-MATERIALtheres 100k but i have to pace it out and only upload when other people arent using the internet for anything else
SingingVoiceDeepfakeDetection_CtrSVDD_ACEKiSing_M4Singerheadlines-ctr-regressionDemo of regression for Baskerville
From upworthy: https://upworthy.natematias.com/about-the-archive.html
Which was later used in SAE's for hypothesis generation
Transformed to just be pairs of [headline, raw_ctr]
CtrSVDD
CtrSVDD Dataset
Please use the download scripts from https://github.com/XIAOYixuan/AUDDT/tree/yixuan-dev
to download and process the dataset.
chmod +x download/get_ctrsvdd.sh
./download/get_ctrsvdd.sh
Description
CtrSVDD is a dataset for audio deepfake detection and spoofing detection research. The dataset is used for evaluation in DeepFense.
Label Distribution
The dataset contains 92,769 samples in the test split:
spoof: 79,173 samples
bonafide: 13,596… See the full description on the dataset page: https://huggingface.co/datasets/DeepFense/CtrSVDD.freight_forwardingflashdeal_data_CTR_historical_signaltrunc100_rt-rel-avito__ad-ctrnew_rt-rel-avito__ad-ctrtrunc150_rt-rel-avito__ad-ctr