List your company

Scientific Question Answering on GPQA (pass@1, pass@256)

18.2pass@1CISPO
6.1369.26812.415.532Feb 6, 2026

Evaluation results

MethodLinks
CISPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.0218.249.7
GRPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.0217.555.1
F-DAPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.021753.7
F-CISPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.021747.7
DAPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.0215.747
DAPOBackbone=Qwen2.5-7B, Group Size (N)=82026.0215.454.9
F-CISPOBackbone=Qwen2.5-7B, Group Size (N)=82026.0215.467.1
F-GRPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.0215.257.6
F-GRPOBackbone=Qwen2.5-7B, Group Size (N)=82026.021557.3
F-DAPOBackbone=Qwen2.5-7B, Group Size (N)=82026.021557.4
CISPOBackbone=Qwen2.5-7B, Group Size (N)=82026.0212.655.5
F-CISPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.021264.2
GRPOBackbone=Qwen2.5-7B, Group Size (N)=82026.0211.346.2
F-DAPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.029.260.7
F-GRPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.028.856.5
DAPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.028.659.4
CISPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.027.448.4
GRPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.026.650.1