Loading...
Loading...
Offline RL trajectories, game data, robot demonstrations, RLHF, multi-agent interaction
10,050 datasets
Halluminate Westworld is a benchmark for evaluating web agents on tasks performed on realistic websites. The dataset was created by Halluminate and was last updated on Hugging Face on November 19, 2025. It includes a human-in-the-loop demo for manually running tasks.
Experimental data and results for the I-SUPPORT two-module soft robotic arm, focusing on assistive bathing tasks. The dataset includes identification data from an Intel RealSense depth camera, step responses under load, position tracking in vertical/horizontal configurations, orientation control, and task execution data like sponge rubbing. MATLAB scripts for modeling, controller design, and analysis are provided by author Carlos Relaño, with metadata last updated in November 2025.
This bibliographic collection curated by Psychic-DL indexes research papers and source code for traffic agent trajectory prediction as of late 2025. It organizes deep learning literature into categories to facilitate the study of vehicle and pedestrian movement modeling.
Grant awards from the federal American Rescue Plan Act (ARPA) to stabilize and expand Washington's child care system. The total ARPA funding available for awards is $360 million. The dataset is provided by data.wa.gov and was last updated in September 2025, but the grant program operated from October 2021 through September 2022.
SA-Co/Silver is a benchmark for promptable concept segmentation in images, created by Facebook and last updated in November 2025. It contains images paired with text labels, each exhaustively annotated with masks for all object instances matching the label. The benchmark comprises 10 subsets covering diverse domains including food, art, robotics, and driving.
The Offender Based Transaction Statistics series tracks adult offenders from arrest through final court disposition. Collected by the Bureau of Justice Statistics from individual states, it includes variables like arrest charge, prosecutor disposition, type of trial, and sentence length. The data allows for analysis of time spent at each stage of the criminal justice process.
Instruction–response pairs for training supportive, non-diagnostic, and safety-aware mental health chatbots. The dataset includes crisis escalation examples and refusal patterns. It was created by author nerd-swayam and last updated on December 14, 2025.
ParaSpeechCaps provides between 1 million and 10 million speech utterance annotations featuring paralinguistic style captions, released by ajd12342 in 2025. The dataset maps audio characteristics to natural language descriptions using a taxonomy of 59 style tags covering pitch, rhythm, emotion, and environmental noise.
CSConDa is the first Vietnamese QA dataset in the customer support domain, curated in collaboration with DooPage, a software company serving 30,000 customers and 45,000 advisors. The dataset covers diverse topics from product pricing to technical consultations.
SA-Co/VEval is an evaluation dataset comprising three distinct video domains: SA-V, YT-Temporal-1B, and SmartGlasses. The dataset, created by Facebook and last updated in November 2025, includes validation and test splits for each domain.
Special Districts that filed late or failed to file their Financial Transactions Report with the State Controller's Office for fiscal year 2023-24. The dataset is published by bythenumbers.sco.ca.gov and was last updated on October 30, 2025. It lists districts by name, county, and submission status.
10,512 chemical reaction images were extracted from scientific literature PDFs. The dataset, created by songjhPKU, is designed for chemical reaction diagram parsing to support cheminformatics and computer vision research.
Sonia Santoveña Casal's research dataset analyzes digital citizen e-participation in Spain's #EstoNoEsUnJuego (#ThisIsNotAGame) state health campaign. The campaign was launched by Spain's Ministry of Health via Twitter in September 2020 to reinforce COVID-19 protective measures. This dataset, harvested from e-cienciaDatos, was last updated on November 23, 2025.
Financial transaction records for expenditures across California cities span 22 fiscal years from 2002-03 to 2023-24. The dataset is published by the State of California Controller's Office via the ByTheNumbers platform. It details spending categorized by entity, location, and purpose.
Washington State data archived in Fall 2021, supporting a dashboard on childcare stabilization. The dataset includes zip code-level indicators such as high_chmaltreatment and high_chequityzips. It was published by data.wa.gov and is no longer being actively updated.
New York, California, and Florida child support court proceedings observed over four years. The dataset comprises field notes from over 1,000 distinct cases observed by researcher Lynne Haney between 2015 and 2019. It catalogs case details by state and county, including case type, parent demographics, and summaries of hearings.
Non-equilibrium molecular dynamics simulations investigate water transport mechanisms in reverse osmosis polyamide membranes. The dataset, authored by Nico Marioni and harvested from the Texas Data Repository, was last updated in November 2025. It compares graphene-restrained and freeze-restrained mechanical boundary conditions across pressure differentials from 1000 to 5000 bar.
Total hours of volunteer opportunities, including Standby, Collateral Duty, and in-station Drill activities, from 1/1/2022 to the present. The data is provided by data.montgomerycountymd.gov and was last updated on 2025-10-10. Columns suggest tracking by Activity Sub Type, Activity Type, Month/Year, and Fire Department.
Public Libraries data aggregates individual library service and finance metrics to the town level for Connecticut. The dataset includes 27 columns covering metrics like library visits, circulation, registered borrowers, reference questions, and town tax appropriations. It is provided by data.ct.gov and was last updated in September 2025.
19 semi-structured interview transcripts and thematic analysis codes from stroke survivors and occupational therapists. The data was collected by Sunghoon Ivan Lee from November 2019 to September 2020 across the USA. It explores stakeholder perspectives on using wearable sensors and arm activity data in clinical practice.