Machine Learning

78 milestones in the history of artificial intelligence from 1943 to 2026, one row per event, from the McCulloch-Pitts neuron and the Turing test to current model releases and regulation. Each row has the date and how precisely it is known, the subject's common name, the person, lab or institution…

117 dated observations of what twelve AI video models grant on their free tiers, one row per figure. Each row carries the channel it was observed on, the date it was observed, and whether the source was a vendor statement or a report. Covered metrics include daily credit grants and daily generation…

Contains 13'322 Asian face images distributed across all ages (from 2 to 80), including 7381 females and 5941 males.

We have published the Audi Autonomous Driving Dataset (A2D2) to support startups and academic researchers working on autonomous driving. Equipping a vehicle with a multimodal sensor suite, recording a large dataset, and labelling it, is time and labour intensive. Our dataset removes this high entry…

Benchmark scores, pricing, and capabilities for 414+ AI models across 40 evaluations (MMLU, HumanEval, GSM8K, etc.). Free REST API documented at benchgecko.ai/api-docs.

The B3FD dataset is a publicly available unconstrained facial image dataset for age (and gender) estimation introduced in the paper Picking out the bad apples - unsupervised biometric data filtering for refined age estimation, consisting of 375,592 facial image samples with corresponding age labels.

Real measurements from two computer-controlled physical devices, a wind tunnel and a light tunnel, collected under observational and interventional regimes, with a ground-truth causal graph for each device derived from its physics and empirically validated. Includes tabular sensor data and images…

Weekly, hash-pinned point-in-time measurements of rising open-source and research-native AI systems, scored by momentum (the rate of change of their public signals) on a transparent, versioned public methodology. A system is marked rising only when two independent axes converge (development…

By releasing this dataset, we aim at providing a new testbed for computer vision techniques using Deep Learning. The main peculiarity is the shift from the domain of "natural images" proper of common benchmark datasets to biological imaging. We anticipate that the advantages of doing so could be…

Signed export of the GSPC board of AI behaviour measurements published by Council of AI (CSOAI Ltd). Scores come from deterministic grading on frozen item banks rather than LLM-as-judge, and the board is Ed25519-signed so any reader can re-check it client-side. Measurement, not certification.

A Large-scale Dataset of Farsi/Arabic License Plate Characters

Statements put to a model that returns a probability per option and no text, asked with one subject swapped at a time, in several wordings and eleven languages, with and without an option to decline. Every question and every raw response is published as JSONL alongside a per-run summary.

This dataset contains 30976 images, or 15488 pairs, most of which were taken at very dark scenes, and all of the images are strictly aligned in time and space. Pedestrians in the dataset are labeled. We compare the dataset with other visible-infrared datasets and evaluate the performance of some…

Confirmed dataset is available to download from Wayback. Also available at https://www.kaggle.com/datasets/jessicali9530/lfw-dataset but requires a login.

"R and Data Mining" ebook data

More than 300 dance videos that capture a single person performing dance moves from TikTok dance challenge compilations for each month, variety, type of dances, which are moderate movements that do not generate excessive motion blur.

Industrial-scale music recommendation dataset with organic/recommendation interactions and audio embeddings.

Controlled inference-runtime comparison on identical hard-linked sha256-verified GGUF bytes; decode, prefill, VRAM, RAM working set, TTFT and load time

Measured board power, tokens per joule, and electricity cost per million tokens for local LLM inference on a retail RTX 5080, with raw 1Hz nvidia-smi telemetry