datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
controlbot-dataConTRoL-nlihttps://github.com/csitfun/ConTRoL-dataset
@article{Liu_Cui_Liu_Zhang_2021,
title={Natural Language Inference in Context - Investigating Contextual Reasoning over Long Texts},
volume={35},
url={https://ojs.aaai.org/index.php/AAAI/article/view/17580},
DOI={10.1609/aaai.v35i15.17580},
number={15},
journal={Proceedings of the AAAI Conference on Artificial Intelligence},
author={Liu, Hanmeng and Cui, Leyang and Liu, Jian and Zhang, Yue},
year={2021},
month={May},
pages={13388-13396}
}
gspc-provenance-controls
GSPC — provenance controls facts (ChainFacts)
SWIFT census (live): https://councilof.ai/api/swift
XRPL reader (live): https://councilof.ai/api/xrpl
MEASURED financial/domain axis (issuer-account / on-chain control facts, n=6). Not a model leaderboard. No accuracy, no fleet, no leader, no separation — measured is not scored.
Frozen bank on Hub. Live n and status are the provenance-controls row on GET https://councilof.ai/api/gspc. Not a certificate.
Council of AI · CSOAI Ltd… See the full description on the dataset page: https://huggingface.co/datasets/csoai/gspc-provenance-controls.controlled-data2026-08-20-odcv-feature-discovery-difficult-advice-716-5-pct-vs-numina-control
LLM-driven feature discovery over ODCV-Bench rollouts from TWO matched Qwen3.6-27B LoRA arms — 9,284 filtered instruction rows plus 716 rows that differ only in kind (constitution-grounded difficult advice vs NuminaMath chain-of-thought) — asking which reasoning and action properties separate the two models, and which go with the judged misalignment.
field
value
experiment
LLM-driven feature discovery over ODCV-Bench rollouts from TWO matched Qwen3.6-27B LoRA arms — 9… See the full description on the dataset page: https://huggingface.co/datasets/dougalldeepmind/2026-08-20-odcv-feature-discovery-difficult-advice-716-5-pct-vs-numina-control.control-sci-corpus
ControlSci Corpus
Control science structured corpus with two configs: Sci-Align benchmark (500 questions) and Sciverse SFT instruction pairs (924 ChatML entries).
License: CC-BY-4.0
Project: MorningStar0709/ControlMind
Configs
benchmark — Sci-Align Benchmark (500 questions)
4-dimension control science evaluation benchmark generated from the ControlSci structured corpus.
Split: core (500 questions)
Load:
from datasets import load_dataset
ds =… See the full description on the dataset page: https://huggingface.co/datasets/MorningStar0709/control-sci-corpus.turkish_cyber_security_controls_benchmark
Turkish Cyber Security Controls Benchmark
Türkçe siber güvenlik kontrol seçimi ve kontrol denetimi yeteneğini ölçmek için
hazırlanmış, senaryo tabanlı çoktan seçmeli değerlendirme kümesidir.
v0.1.0, uzman incelemesine açık ilk sürümdür ve NIST SP 800-53 Rev. 5,
Release 5.2.0 kontrol kataloğunu hedefler.
Kapsam
100 Türkçe senaryo
NIST SP 800-53'ün 20 kontrol ailesinin her birinden 5 soru
64 kontrol seçimi sorusu
17 denetim kanıtı sorusu
19 denetim yargısı sorusu… See the full description on the dataset page: https://huggingface.co/datasets/logicBombExe/turkish_cyber_security_controls_benchmark.CantTalkAboutThis-Topic-Control-Dataset
CantTalkAboutThis Topic Control Dataset
Dataset Details
Dataset Description
The CantTalkAboutThis dataset is designed to train language models to maintain topical focus during task-oriented dialogues. It includes synthetic dialogues across nine domains (e.g., health, banking, travel) and incorporates distractor turns to test and improve the model's ability to be resilient to distractors. Fine-tuning models on this dataset enhances their ability to maintain… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/CantTalkAboutThis-Topic-Control-Dataset.qwen-generated-svamp-controls-sft
Qwen-Generated SVAMP CoT Controls ? SFT
Qwen-generated controlled reasoning traces for SVAMP in LLaMA-Factory SFT format. Variants include ordinary, all-caps, no-comma, disclaimer, and multilingual examples.
Splits
3,940 training examples and 380 held-out evaluation examples.
Format
The JSON files use the LLaMA-Factory Alpaca-style schema. The included
dataset_info.json registers the exact training and evaluation names. DPO
records are marked with… See the full description on the dataset page: https://huggingface.co/datasets/akshay-sked/qwen-generated-svamp-controls-sft.CantTalkAboutThis-Topic-Control-Dataset-NC
CantTalkAboutThis Topic Control Dataset
Dataset Details
Dataset Description
The CantTalkAboutThis dataset is designed to train language models to maintain topical focus during task-oriented dialogues. It includes synthetic dialogues across nine domains (e.g., health, banking, travel) and incorporates distractor turns to test and improve the model's ability to be resilient to distractors. Fine-tuning models on this dataset enhances their ability to maintain… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/CantTalkAboutThis-Topic-Control-Dataset-NC.2026-08-19-random-220-difficult-advice-control-train
Random 220-row difficult-advice control for the LESS top-10% arm
field
value
experiment
THE CONTROL ARM of a paired LESS (arXiv:2402.04333) data-selection experiment: SFT training file holding 220 rows drawn uniformly at random (seed 1) from the same 2203-row difficult-advice pool, trained as-is on base Qwen3.6-27B with no other data. The two arms differ ONLY in which 220 of the same 2,203 rows they hold — identical tokenizer, budget, seed, shuffle and training recipe… See the full description on the dataset page: https://huggingface.co/datasets/dougalldeepmind/2026-08-19-random-220-difficult-advice-control-train.chess-time-control-string-parsing
Chess Time-Control String Parsing
Real-world chess time-control strings, in two forms:
.txt files — the source of truth. Every unique time-control string, one per line, with a frequency count. These are the raw, real strings (messy, multilingual, sometimes junk) as scraped/collected. No interpretation.
.jsonl files — a tagged, partially-correct derived artifact. Each unique string with an auto-derived (category, stages) parse. The tags are heuristics, not verified ground truth… See the full description on the dataset page: https://huggingface.co/datasets/gutsy-gambit/chess-time-control-string-parsing.Android-Control-84k
Android Control Dataset
Overview
This directory contains two dataset files (and_ctrl_train.json and and_ctrl_test.json) derived from the Android Control project by Google Research. These datasets have been formatted specifically for GUI grounding training in LLaMA-Factory.
Dataset Description
The Android Control dataset consists of episodes where each episode contains multiple steps. Each step includes:
Step instructions: Natural language instructions for UI… See the full description on the dataset page: https://huggingface.co/datasets/OfficerChul/Android-Control-84k.ControlNet_ade20krepro-score-a-unified-framework-for-overshoot-refund-in-online-fdr-control-traces
Agent traces
Agent sessions published from a Trackio Logbook.
qwen-generated-svamp-controls-dpo
Qwen-Generated SVAMP CoT Controls ? DPO
Preference pairs built from Qwen-generated SVAMP reasoning traces in LLaMA-Factory DPO format. Each record contains instruction, input, chosen, and rejected fields.
Splits
3,152 training preference pairs and 304 held-out evaluation pairs.
Format
The JSON files use the LLaMA-Factory Alpaca-style schema. The included
dataset_info.json registers the exact training and evaluation names. DPO
records are marked with… See the full description on the dataset page: https://huggingface.co/datasets/akshay-sked/qwen-generated-svamp-controls-dpo.bedroom-ac-control
Bedroom AC Control
Synthetic decisions for a bedroom fan coil air conditioner. Each row is one 10 minute tick: sensor readings in, OFF or COOL out. The rows were generated by free models on OpenRouter (dots-3-note-preview, gemma-4-26b-a4b-it, ling-3.0-flash-fin, nemotron-3-super-120b-a12b, nemotron-3-ultra-550b-a55b) from a written policy, then checked against a rule implementation of the same policy.
The home has two HomePod minis in the far corner of the bedroom, one Mila… See the full description on the dataset page: https://huggingface.co/datasets/Aayush9029/bedroom-ac-control.hedgehog-loop-control-r4
hedgehog-loop-control-r4
Hedgehog — loop-control round 4 (termination/repetition fixes).
Contents
train.jsonl (2944 rows)
validation.jsonl (438 rows)
Format
JSON Lines (.jsonl), one example per line.
Provenance
Original content for the Hedgehog extraction model (Michael Anthony Falabella).
StaR_state_control_benchmark
task_categories:
- image-text-to-text
StaR State Control Benchmark Dataset
This repository provides the state control benchmark of our paper: See, Think, Act: Teaching Multimodal Agents to Effectively Interact with GUI by Identifying Toggles.
Code: https://github.com/ZrW00/StaR
We provide the state control benchmark of our paper:
See, Think, Act: Teaching Multimodal Agents to Effectively Interact with GUI by Identifying Toggles
How to Use
The… See the full description on the dataset page: https://huggingface.co/datasets/ZrW00/StaR_state_control_benchmark.android_control_train
Processed Android Control Training Set
Dataset Description
This repository contains the processed training set derived from the Android Control dataset by Google Research.
The data processing methodology is identical to that used for our corresponding test set, which can be found at Reallm-Labs/android_control_test.
Data Content and Image Extraction
Important Note: Due to the large size of the dataset, this repository contains only the processed text files.… See the full description on the dataset page: https://huggingface.co/datasets/InfiX-ai/android_control_train.Controlled-Negation-DatasetControlled Negation Training Dataset
This repository contains the training data used in “Architecture or Data? A Controlled Negation Study.”
The study compares a standard transformer architecture with a modified architecture while keeping the training data and training configuration consistent between both model variants.
Files
continual-pretraining-negation-cot-80k.jsonl
Approximately 80,000 records used during the continual pretraining stage, focused on negation-oriented reasoning and… See the full description on the dataset page: https://huggingface.co/datasets/aikronus-labs/Controlled-Negation-Dataset.ai-control-corpus
AI Control Corpus
A question–answer dataset covering the AI Control literature until mid-2025. The corpus was assembled for fine-tuning experiments investigating self-fulfilling misalignment (see more).
Dataset Summary
Statistic
Value
Q&A pairs
2,633
Unique sources
212
Total tokens (Q+A)
~1.6 million
Generation date
August 2025
Source Distribution
Source Type
Pairs
Share
LessWrong
845
32.1%
arXiv
545
20.7%
Alignment Forum… See the full description on the dataset page: https://huggingface.co/datasets/vohonen/ai-control-corpus.ConTRoL
ConTRoL dataset
This repository is a copy of the ConTRoL dataset for natural language inference. Its sole purpose is to host the original dataset files in compliance with standard HuggingFace APIs.
Citation
@article{Liu_Cui_Liu_Zhang_2021,
title={Natural Language Inference in Context - Investigating Contextual Reasoning over Long Texts},
volume={35},
url={https://ojs.aaai.org/index.php/AAAI/article/view/17580},
DOI={10.1609/aaai.v35i15.17580},
number={15}… See the full description on the dataset page: https://huggingface.co/datasets/xuanr-knockri/ConTRoL.clinical-intervention-sequencing-and-state-control-v0.2
Clinical Multi-Evidence State Integration Benchmark
CMESI v0.2
The Clinical Multi-Evidence State Integration Benchmark (CMESI) evaluates whether an AI system can reconstruct the evolving state of a complex clinical case across a sequence of heterogeneous evidence events.
CMESI does not test whether a model can identify a diagnosis from a static vignette alone. It tests whether the model can:
maintain several competing clinical hypotheses simultaneously;… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-intervention-sequencing-and-state-control-v0.2.production-ai-control-evidence
Control Evidence Dataset v0.1
Public v0.1 candidate · reviewed and approved for publication by Dmytro Nasyrov on September 17, 2026.
This public candidate contains 55 draft control records for release-evidence preparation. Each record connects an engineering action to requested evidence, a suggested owner, a change trigger and recovery guidance. The companion Evidence Pack collects version-specific references and exposes missing records.
Pharos Production's published AI agent… See the full description on the dataset page: https://huggingface.co/datasets/pharosproduction/production-ai-control-evidence.arch-opposite-sign-lpi-260903T0110-si-negation-control-datasetarch-opposite-sign-lpi-260903T0110-si-negovert-control-datasetarch-opposite-sign-lpi-260903T0110-si-extrap-control-datasethumanoid-gesture-control-dataset-v1Dataset for teaching humanoid robots to react to human hand gestures.
Description
Gesture-based commands mapped to safe robot actions.
Task Description
Teaches humanoid robots how to interpret basic hand gestures and execute corresponding movement behaviors.
arch-opposite-sign-lpi-260903T0110-xfam-repl-control-dataset