MixtureVitae: Open Pre-training Dataset with Only 8% Web Content
JJitsev · x · 2026-07-08
MixtureVitae is a new dataset to be presented at ICML. Web content makes up only 8% of the dataset, and it uses permissive licensing. It significantly outperforms strong baselines on math and code tasks without sacrificing linguistic capabilities, marking it as a promising future direction for open pre-training.
Related event: MixtureVitae Released: Open Dataset with Only 8% Web Text(11 posts)→
More from Research
- Nat Lambert shares a reading list on synthetic data and agentic SFT data — natolambert · 2026-07-22
- Lightwheel AI Launches SimReadyGen: Text-to-Physics-Accurate Robot Sim Assets — ZeYanjie · 2026-07-22
- PNAS special issue examines copyright, governance, and AI in the legal system — chrmanning · 2026-07-22
- WeirdChat catalogs strange model behaviors from more than 100 million sampled responses — JacobSteinhardt · 2026-07-22
- New agentic benchmark shows AI managers escalate to coercion and fake success — Jasmine Brazilek · 2026-07-22
- Ai2’s Asta adds one-click handoff and self-checking deep paper search — allen_ai · 2026-07-22