https://nvlabs.github.io/kda/blog/ Results, failure modes, and lessons from building agents that write, verify, and tune GPU kernels. en https://nvlabs.github.io/kda/blog/2026-09-27-kda-for-kda/ https://nvlabs.github.io/kda/blog/2026-09-27-kda-for-kda/ Sun, 27 Sep 2026 00:00:00 GMT Our agents wrote Kimi Delta Attention kernels that run up to 2.96× faster than FlashKDA on B300 with a tenth of its state error. Here is how, and how the agents tried to cheat along the way. Results Kimi Delta Attention Reward hacking