How AlphaEvolve’s tiling rule sped up Gemini training
DeepMind reports an average 23% kernel speedup and a 1% reduction in overall training time. The key output is code that chooses matrix tile sizes as inputs change.
AnalysisExpand summaryCollapse summary
AlphaEvolve searched for a reusable matrix-tiling rule, judging its choices by actual runtime on real TPUs. Half the collected input shapes guided optimization; the other half evaluated the resulting rule. The reported production benefit connects model-generated code to model-training infrastructure, while a later model’s ability to discover better algorithms remains unmeasured in the supplied case.