Grigore Rosu: Program execution is proof generation, unlocking infinite training data from code
LingmingZhang · x · 2026-09-19
Formal methods researcher Grigore Rosu argues that "program execution IS proof generation": within semantic frameworks, execution is rooted in formal language semantics, so every step has a justification. Composing those steps turns a finite execution into a proof of how the program moves from initial to final state.
The payoff: across unbounded programs and inputs, code supplies infinitely many finite execution proofs, and the resulting data — program states, applied rules, and derivations — can be used directly as training data generated from code.
More from Research
- CoT monitoring isn't an audit log: model explanations barely change when decisions flip — ziv_ravid · 2026-09-19
- CoT may not be faithful: filler tokens add 13 points, models keep reasoning after committing — ziv_ravid · 2026-09-19
- TokenRhythm unveils NeoHorse-1: 4B/9B models claiming first closed-loop step toward RSI — jiqizhixin · 2026-09-19
- Training an LLM to paint by writing code: RL meets creative tasks with a judge model — measure_plan · 2026-09-19
- AI provers on a Bittensor subnet close six decades-old Erdős problems in ten days — markjeffrey · 2026-09-19
- Stanford's Anshul Kundaje: gene expression AI models capped by data, not scale — anshulkundaje · 2026-09-19