List your company

Multi-hop Question Answering on HotpotQA (F1, Latency)

86.69F1 ScoreFlowCompile
71.225275.240179.25583.2699May 13, 2026

Evaluation results

MethodLinks
FlowCompileSelection Priority=Accuracy-priority2026.0586.697.8
Qwen3-8BBackbone=Qwen3-8B2026.0586.5818.5
Full BaselineBackbone=Qwen3-14B2026.0586.2926.5
FlowCompileSelection Priority=Latency-priority2026.0585.323.8
Qwen3-4BBackbone=Qwen3-4B2026.0583.8912.5
MaASBackbone=Qwen3-8B2026.0583.4414.1
KNN Router2026.0573.756.9
MaASBackbone=Qwen3-4B2026.0571.826.3