BenchmarkMLEBench Lite

MLEBench Lite

A lightweight version of the machine learning engineering benchmark MLEBench, used to evaluate the engineering capabilities of AI agents on machine learning tasks.

Timeline (last 90 days)

Oct 20

The SwarmRSI framework comprehensively sets new SOTA across 118 tasks on the three benchmarks OpenProblems, NatureBench and MLEBench Lite.

Unverified50%

Oct 20

With DeepSeek V4 Pro as the backbone, SwarmRSI achieves a medal rate of 81.8% on MLEBench Lite.

Unverified50%

All Facts (2)

Unverified

The SwarmRSI framework comprehensively sets new SOTA across 118 tasks on the three benchmarks OpenProblems, NatureBench and MLEBench Lite.

50%Unverified

With DeepSeek V4 Pro as the backbone, SwarmRSI achieves a medal rate of 81.8% on MLEBench Lite.

50%

Source Articles

SwarmRSI: How a Swarm of Research Agents Achieves Recursive Self-ImprovementOct 20