List your company

Multi-hop QA on Bamboogle Wiki18 index (test)

64.3Pass@4Gemma3-12B
14.927.72540.5553.375Jun 15, 2026

Evaluation results

MethodLinks
Gemma3-12BSampling Method=DivInit (DI)2026.0664.3
Qwen3-8BSampling Method=DivInit (DI)2026.0657.6
Gemma3-12BSampling Method=Standard parallel sampling (S)2026.0655.7
Qwen3-8BSampling Method=Standard parallel sampling (S)2026.0647.7
Qwen3-4BSampling Method=DivInit (DI)2026.0640.8
Gemma3-4BSampling Method=DivInit (DI)2026.0637.9
Qwen3-4BSampling Method=Standard parallel sampling (S)2026.0632.5
Gemma3-4BSampling Method=Standard parallel sampling (S)2026.0627.7
Qwen3-1.7BSampling Method=DivInit (DI)2026.0624.3
Qwen3-1.7BSampling Method=Standard parallel sampling (S)2026.0616.8