VeriTile embeds Triton GPU kernels in Lean, with AI agents writing machine-checked correctness proofs

KaiyuYang4 · x · 2026-09-25

Kaiyu Yang's team introduces VeriTile, exploring whether AI can improve its own training infrastructure while proving the changes correct.

• Embeds a Triton-style kernel DSL in Lean 4, keeping familiar Triton syntax while providing formal semantics
• Verifies both correctness (kernel satisfies a mathematical spec, including memory effects) and equivalence to a reference implementation, with examples like proving FlashAttention correctly implements attention
• AI agents assist in constructing machine-checkable proofs
• The bet: faster verified kernels mean faster training with correctness guarantees

Original post →

More from Infra

• Powering a home GPU cluster: one PDU safely feeds ~16 RTX 6000 Max-Q cards — TheZachMueller · 2026-09-25
• Docker launches Cloud Sandboxes: microVM isolation for always-on agents, $250 credit — juntao · 2026-09-25
• Muse Spark 1.3 Now Available via Oracle, in Private Preview on Google Cloud — alexandr_wang · 2026-09-25
• Andy Matuschak: Modern Chips Are Too Durable for Compute-Control-Based AI Governance to Hold — andy_matuschak · 2026-09-25
• Qualcomm touts double-digit MLPerf performance and efficiency gains at Snapdragon Summit — samcharrington · 2026-09-25
• HEIF Heist: image parser RCE chain nets $100k Meta bounty, hits OpenAI repos and more — evilsocket · 2026-09-25

Infra channel →