Zero-data pretraining: two LMs bootstrap all training data via self-play from random init

ChengleiSi · x · 2026-09-26

Researchers introduce Self-Play Pretraining with Zero Data: two models start from random initialization — a generator proposes programs for a universal Turing machine, and a learner trains on their outputs, never touching real data.

Key findings:

Co-led with Aditya Cowsik and Kfir Dolev, with Noah Goodman and Yoav Levine among co-authors, the work is a proof of concept suggesting LMs can in principle bootstrap their own pretraining data.

Related event: Stanford-Led Paper Trains Models From Scratch With Zero Real Data via Self-Play(2 posts)→

Original post →

More from Research

Research channel →