LLM agents that generate, verify, and evolve Triton GPU kernels. Includes a reward-hack-resistant benchmarking harness with strict correctness verification and fresh-input evaluation. Achieves up to 174.7× over PyTorch eager and outperforms FlexAttention (1.48×) and SDPA (1.17×) on selected workloads.
inference openai evolutionary-algorithms automatic-code-generation large-language-models agentic-workflow agentic-ai triton-kernels kernel-optimization gepa harness-engineering agentic-kernel-generation
-
Updated
Aug 12, 2026 - Python