Provides a reproducible, deduplicated corpus of text extracted from PDFs for LLM pretraining—about 3 trillion tokens from ~475 million documents in 1733 language-script pairs. Includes OCR and text extraction pipelines, per-page language IDs, MinHash deduplication, and is released under ODC‑By 1.0.
[next-mdx-remote-client] error compiling MDX: Unexpected character `2` (U+0032) before name, expected a character that can start a name, such as a letter, `$`, or `_` More information: https://mdxjs.com/docs/troubleshooting-mdx
Benchmarks models on single-select, multi-select, and probabilistic decisions. It combines long contexts, large option sets, and interdependent choices, and evaluates 28 decision models.
Provides 50,000 hours of egocentric human manipulation video with hand-motion and language annotations for embodied-model pretraining and robot learning. The data is curated and aligned to a shared bimanual end-effector representation.
Pairs egocentric video with timed task annotations and head-and-hand motion, linking human actions to their intent. The 100,000-hour release supports research and training for physical AI, with a CC BY 4.0 license.
