List your company

Multi-hop Question Answering on HotpotQA distractor setting

52.01ConcisenessQwen3
8.725219.962631.242.4374Jan 21, 2026

Evaluation results

MethodLinks
Qwen3Size=14B2026.0152.0173.1897.31
Qwen3Size=30B-A3B2026.0150.9777.0997.77
Qwen3Size=4B2026.0150.3975.5896.87
Qwen3Size=1.7B2026.0150.0766.1792.9
DeepSeek-R1Size=671B2026.0149.673.9198.56
GPT-5.1Size=UNK2026.0148.3772.6199.15
Qwen3Size=8B2026.0147.4775.3697.68
GPT-o3Size=UNK2026.0147.4473.2498.92
Qwen3Size=235B-A22B2026.0145.6273.898.87
GPT-4oSize=UNK2026.0144.975.5797.36
DeepSeek-V3.2Size=671B2026.0144.8176.3497.96
Gemini-3-proSize=UNK2026.0139.6978.6598.56
LLaMA-3.1Size=70B2026.0139.471.496.04
SelfCiteSize=8B2026.0139.2366.7491.04
LongCiteSize=8B2026.0138.8564.5691.75
Claude-4.5Size=UNK2026.0136.478.3398.07
Qwen3Size=0.6B2026.0133.1152.1681.52
FRONTSize=13B2026.0127.7657.9682.86
LLaMA-3.1Size=8B2026.0126.4654.9484.28
LLaMA-3.2Size=1B2026.0110.3917.4546.45
LLaMA-3.2Size=3B2026.0110.3919.0270.41