List your company

Question Answering on HOTPOTQA Unfiltered

66.08AccuracyBaseline
56.501658.988361.47563.9617May 25, 2026

Evaluation results

MethodLinks
BaselineModel=Qwen 2.5 32B2026.0566.0873.952.95
+ScramblerModel=Qwen 2.5 32B2026.0565.6473.482.96
+DistributedModel=Qwen 2.5 32B2026.0565.5673.442.96
+DistributedModel=Mistral 7B2026.0561.7471.511.9
+ScramblerModel=Mistral 7B2026.0561.6671.391.89
BaselineModel=Mistral 7B2026.0561.6371.411.89
+DistributedModel=Qwen 2 7B2026.0561.0869.6111.63
BaselineModel=Qwen 2 7B2026.0561.0169.5911.62
+ScramblerModel=Qwen 2 7B2026.0560.9369.5611.83
+DistributedModel=Llama 3.1 8B2026.0556.9666.022.15
BaselineModel=Llama 3.1 8B2026.0556.8766.012.14
+ScramblerModel=Llama 3.1 8B2026.0556.8765.922.15