Use DNNs to build encoding models of EEG visual responses.

Source of the Open Repair Alliance downloadable datasets

CRED-1: An Open Multi-Signal Domain Credibility Dataset (2,672 domains)

🇧🇮 The first large-scale, open-source speech and text dataset for Kirundi language. Building AI models for 12M+ Kirundi speakers through community collaboration. Includes ASR, TTS, and MT capabilities.

Following is the Open Email Marketing Dataset; you can use it without any restrictions.

Data and code for SMJ 2019 paper

Open, verifiable AI-driven football market analytics project for detecting mispriced bookmaker odds.

🇷🇺 Russian losses in 🇺🇦 Ukraine since the start of the 2022 full-scale invasion 🪖.

A Framework for Robust, Self-Recovering Tool-Using Language Model Agents — trained on 50K+ failure-annotated trajectories for fault-tolerant reasoning and recovery.

A small dialogue dataset exploring the boundaries of machine decision-making, agency, and alignment. Useful for fine-tuning conversational agents or testing moral reasoning

Data and code for OS 2024 paper

📊 Fetch real-time Italian statistics on economy, population, health, and more with the Live Italy API Wrapper, fully supporting TypeScript.

Reproducible MIT-licensed accuracy benchmark for any astrology API. 210 planet positions across 21 charts verified against NASA JPL Horizons DE441.

Dataset of 4,368 AI-generated images based on COCO for assessing coherence and realism in synthetic imagery.

LLM behavioral benchmark from 25-month narrative gameplay. 540 runs, 6 models, pre-registered statistical analysis. GPT-4o-mini shows a perfect binary switch on a social decision from prompt framing alone.

Data from ATNi's 2025 East Africa Market Assessment (EAMA) on company nutrition policies and product portfolios in Kenya

Data from ATNi's 2025 East Africa Market Assessment (EAMA) on company nutrition policies and product portfolios in Tanzania.

This dataset (updated monthly) covers average full coverage rates for all 50 states plus Washington DC, based on a standardized driver profile (25–55 years old, clean record, full coverage). Data is aggregated from official state insurance department filings and annual market conduct reports.

Comparison of distributed machine learning techniques applied to openly available datasets

Some of San Francisco's open data