List your company

STEM Reasoning on GPQA Diamond

70.2AccuracySwiR
30.274440.639751.00561.3703Oct 6, 2025

Evaluation results

MethodLinks
SwiRBackbone=Qwen3-32B2025.1070.2
CoT (Greedy)Backbone=Qwen3-32B2025.1069.7
Soft ThinkingBackbone=Qwen3-32B2025.1067.17
CoTBackbone=Qwen3-32B2025.1066.16
SwiRBackbone=Qwen3-8B2025.1061.11
CoTBackbone=Qwen3-8B2025.1059.6
Soft ThinkingBackbone=Qwen3-8B2025.1059.6
CoT (Greedy)Backbone=Qwen3-8B2025.1056.57
SwiRBackbone=DeepSeek-R1-Distill-Llama-8B2025.1047.98
CoTBackbone=DeepSeek-R1-Distill-Llama-8B2025.1046.46
SwiRBackbone=Qwen3-1.7B2025.1041.41
CoTBackbone=Qwen3-1.7B2025.1039.39
Soft ThinkingBackbone=Qwen3-1.7B2025.1034.34
Soft ThinkingBackbone=DeepSeek-R1-Distill-Llama-8B2025.1033.33
CoT (Greedy)Backbone=Qwen3-1.7B2025.1031.82
CoT (Greedy)Backbone=DeepSeek-R1-Distill-Llama-8B2025.1031.81