Data4Cyber publishes labeled OT attack data for a lab distribution grid
The lab DMS dataset pairs 1-second process data with Modbus/TCP and MQTT captures across one benign run and six attacks.
Stories per day · 30 days
4 RSS
Relatedbenchmark5multimodal5huggingface3evaluation2reinforcement-learning2wireless23d-generation1accessibility1agentic-ai1audio-description1bimanual1bitcoin1bluetooth1botnet-detection1claude1code-generation1coding-agents1computational-law1
55 stories, newest first · one card per story
Stories tagged #dataset
LUCQUE releases annotated Québec LDCT lung-screening data for AI evaluation
The set has 1,678 baseline studies, 1,776 annotated nodules, and 60 cysts, accessed only via model-to-data.
Nemotron-Personas-Finance-Japan synthetic dataset covers Japan
The Hugging Face post, under an NVIDIA path, gives no further details beyond the title.
Channels4World releases 100,000 outdoor RF scenes for wireless models
The 29.14-TB corpus spans 3.5 and 7 GHz with a 90K/10K scene-level train-test split.
PQC-Corpus links post-quantum histories from 179 GitHub repositories
In samples, about seven in ten commits and two in three issues were labeled post-quantum related.
Bitcoin forensics dataset Elliptic++ has temporal leakage
Re-evaluation of Elliptic++ after correcting for temporal splits and feature definitions drops random forest F1 score on illicit class from 0.846 to 0.291.
Counterfactual labels improve incident response decisions in cyber-physical systems
New TE-GvA dataset with belief-level labels for Tennessee Eastman process shows gradient-boosted rules achieve 11 percent lower regret than hindsight labels on stress sets.
YANAMI Dataset and Benchmark Released for Multimodal Stateful Interaction
YANAMI introduces the first large-scale multimodal dialogue dataset with 400K multi-turn interactions across four scenarios.
TrafficInstruct dataset and benchmark for instruction tuning on network traffic released
55,000 structured records support 10 detection tasks with task-specific LoRA adapters improving performance.
arXiv releases BLE device fingerprinting dataset and testbed
Large-scale raw IQ dataset from 31 Bluetooth Low Energy devices across varied environments plus evaluation of domain-robust radio-frequency fingerprinting models.
GameVerse dataset enables long-horizon interactive world modeling
GameVerse contains 69.7K clips totaling 1,655 hours with stable camera trajectories for training extended interactive models.
SketchDirector enables complex scene sketching
SketchDirector uses LLM decomposition and diffusion refinement to mimic human sketching workflow, introducing MCSketch8K dataset.
Google releases MTG-AD benchmark for inline audio descriptions
The 36-hour set has more than 8,000 descriptions, and end-to-end scoring checks whether narration fits existing video silences.
New MSafeTemp dataset advances multilingual LLM safety alignment
The template-based pipeline produces a 60-language corpus for safety training, exposing gaps in English-only evaluation.
IntReal dataset and suite benchmark interleaved text-image generation
IntReal-100K and a seven-dimension suite, with 83.3% human agreement, show unimodal training can also help interleaved generation.
AquaSense-ICS releases leakage-aware water CPS attack benchmark
Under chronological splits, no model beats a constant normal predictor, and unseen-attack AUROC averages 0.64.
PGT-FG labels toxicity in 2,396 Punjabi song segments
About 29% of segments are labelled toxic, violence is the most common category, and the resolved set is on Zenodo.
DONIC trains one policy for generalist bimanual dexterous tracking
Without finetuning, one checkpoint reaches 86.79 percent success on unseen training-object trajectories and 57.32 percent on held-out objects.
Study maps how WildChat users ask LLMs for legal guidance
In an enriched sample, 71.1% of legal-content chats sought personalized or practical guidance; annotations are public.
Dataset reconstructs procedural histories of 5346 Italian laws
LLM extraction, checked against manual annotations, yields 101413 dated events from 1985 to 2026.
RAP dataset pairs 25K+ papers with posters for visual coding
Each sample is a paper-poster-HTML triplet; a 150-paper held-out set supports render-and-compare evaluation.
3DCodeVerse: Exploring 3D Worlds as Code
3DCodeVerse pairs a 2M executable-code dataset with a harness that beats Claude Design in tests.
PrismaCoder: Pushing the Boundaries of Open Multimodal Code Intelligence through Decomposed Rubric Synthesis
PrismaCoder releases 9B and 35B open models and a 2M multimodal code dataset trained with rubric rewards.
AdSpark: A Large-Scale Dataset and Benchmark for Product-Centric Advertisement Video Generation
AdSpark releases a ~300K dataset and six-dimension benchmark for product-centric advertisement video generation.