operational metric

AlphaEvolve training optimization

Reported deployment of AI-discovered training kernels.

AI-system improvementObserved R&D automation

What this measure tells us

Connects algorithm search to model-training efficiency.

Reported by developer; experimental compute and full production denominator undisclosed.
A fixed optimizer improving code does not establish recursive acceleration.

Benchmark version

Metric

Results to compare

Protocol

2025-05-14

Two different denominators remain separate metrics.

Gemini training-time reduction (percent): A task-specific measurement; not a percentage of RSI achieved.

Kernel speedup (percent): A task-specific measurement; not a percentage of RSI achieved.

Published results · May 2025 report
Key	System / organization	Metric	Reported result	Date	Protocol	Verification	Evidence
—	AlphaEvolve / Gemini ensembleGemini training / matrix multiplication kernel	Gemini training-time reduction	1%	2025-05-14	alpha-training_reduction	lab reported	AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms ↗
—	AlphaEvolve / Gemini ensembleGemini training / matrix multiplication kernel	Kernel speedup	23%	2025-05-14	alpha-kernel_speedup	lab reported	AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms ↗

No reference points are indexed for this measure.