List your company

Multi-document Question Answering on HotpotQA (Acc, throughput)

45AccuracyDense
16.9224.2131.538.79Jul 29, 2025

Evaluation results

MethodLinks
DenseModel=QwQ 32B2025.0745297.34
DenseModel=Phi 4 reasoning plus2025.0743525.39
ReasonCacheModel=QwQ 32B2025.0741423.23
ReasonCacheModel=Phi 4 reasoning plus2025.0740695.88
DenseModel=DeepSeek R1 Distill Qwen 32B2025.0737.5381.51
ReasonCacheModel=DeepSeek R1 Distill Qwen 32B2025.0736542.43
RandomModel=QwQ 32B2025.0722421.89
RandomModel=Phi 4 reasoning plus2025.0721699.87
RandomModel=DeepSeek R1 Distill Qwen 32B2025.0718537.85