List your company

Multi-hop Question Answering on MuSiQue in-distribution

56.91EMCRAFT7B
-1.912413.358828.6343.9012Feb 1, 2026

Evaluation results

MethodLinks
CRAFT7BType=Trained, CRAFT version=v4, Training Mode=RL2026.0256.9166.4786.7
Deepseek-V3.2Type=API, CRAFT version=v42026.0253.267.9688.25
GPT-5-miniType=API, CRAFT version=v42026.0252.0568.193.15
Gemini-2.5-FlashType=API, CRAFT version=v42026.0251.2562.9290.2
Qwen3-30BType=Pre-Trained, CRAFT version=v42026.0246.7556.3484.95
Qwen3-4BType=Pre-Trained, CRAFT version=v42026.0238.547.6177.55
Qwen2.5-7BType=Pre-Trained, CRAFT version=v42026.0238.3549.2976.6
CRAFT3BType=Trained, CRAFT version=v4, Training Mode=RL2026.0232.7143.1471.4
Qwen2.5-7BType=Trained, CRAFT version=v4, Training Mode=SFT2026.0231.842.7166.6
Qwen2.5-3BType=Trained, CRAFT version=v4, Training Mode=SFT2026.0223.633.150.15
Qwen2.5-3BType=Pre-Trained, CRAFT version=v42026.0221.9532.341.15
CRAFT1.5BType=Trained, CRAFT version=v4, Training Mode=RL2026.0216.4123.2729.57
CRAFT0.5BType=Trained, CRAFT version=v4, Training Mode=RL2026.028.4515.1125.4
Qwen2.5-1.5BType=Trained, CRAFT version=v4, Training Mode=SFT2026.027.2512.8516.4
Qwen2.5-1.5BType=Pre-Trained, CRAFT version=v42026.024.859.7816.65
Qwen2.5-0.5BType=Trained, CRAFT version=v4, Training Mode=SFT2026.020.54.6213.5
Qwen2.5-0.5BType=Pre-Trained, CRAFT version=v42026.020.352.694.85