List your company

Multi-step Reasoning on Bamboogle auto-eval (test)

76.1Mean AccuracySelf-improvement, 2nd gen
43.44451.92260.468.878Dec 15, 2023

Evaluation results

MethodLinks
Self-improvement, 2nd genModel Size=L2023.1276.11.3
Self-improvement, 1st genModel Size=L2023.12743.3
Pilot, human filteredModel Size=L2023.1271.52.2
Pre-trainedModel Size=L2023.1270.33.5
Self-improvement, 2nd genModel Size=S2023.1269.71.3
Self-improvement, 2nd genModel Size=XS2023.1265.92.6
Self-improvement, 1st genModel Size=S2023.1261.91.9
Pilot, human filteredModel Size=S2023.1256.63.8
Self-improvement, 1st genModel Size=XS2023.1254.43.6
Pilot, human filteredModel Size=XS2023.1244.73.1