List your company

Out-of-Distribution Reasoning on OOD Reasoning Suite GPQA-D, Connect4 (test)

49.1GPQA-D ScoreRFT
27.67633.23838.844.362Mar 2, 2026

Evaluation results

MethodLinks
RFTBackbone=Qwen3-8b2026.0349.13.126.1
Qwen3-8bSampling=avg@52026.0348.910.929.9
SFT+Backbone=Qwen3-8b2026.0347.314.130.7
SPOTBackbone=Qwen3-8b2026.0346.83641.4
SFTBackbone=Qwen3-8b2026.03391225.5
RFTBackbone=Llama3.1-8b-Instruct2026.0332.3217.2
SFT+Backbone=Llama3.1-8b-Instruct2026.0331.91.516.7
SPOTBackbone=Llama3.1-8b-Instruct2026.0330718.5
SFTBackbone=Llama3.1-8b-Instruct2026.0329.61.715.7
Llama3.1-8b-InstructSampling=avg@52026.0328.5516.8