List your company

Science Question Answering on ScienceQA (AUC)

92.36AUCReLope
71.268876.744482.2287.6956Mar 25, 2026

Evaluation results

MethodLinks
ReLopeBackbone=Phi-4-Multimodal-Instruct2026.0392.36
ReLopeBackbone=Qwen2.5-VL-7B-Instruct2026.0391.79
ReLopeBackbone=Gemma3-12B2026.0388.73
Attention ProbeBackbone=Phi-4-Multimodal-Instruct2026.0387.24
Attention ProbeBackbone=Qwen2.5-VL-7B-Instruct2026.0386.43
Attention ProbeBackbone=Gemma3-12B2026.0384.93
RouteLLMBackbone=Phi-4-Multimodal-Instruct2026.0384.13
ProbeBackbone=Qwen2.5-VL-7B-Instruct2026.0383.95
RouteLLMBackbone=Qwen2.5-VL-7B-Instruct2026.0382.83
ProbeBackbone=Phi-4-Multimodal-Instruct2026.0382.08
ProbeBackbone=Gemma3-12B2026.0380.47
RouteLLMBackbone=Gemma3-12B2026.0378.92
Post-Hoc EmbedBackbone=Phi-4-Multimodal-Instruct2026.0377.84
Cascade RoutingBackbone=Qwen2.5-VL-7B-Instruct2026.0376.31
Post-Hoc EmbedBackbone=Gemma3-12B2026.0376.19
Cascade RoutingBackbone=Gemma3-12B2026.0375.58
Cascade RoutingBackbone=Phi-4-Multimodal-Instruct2026.0375.56
Post-Hoc EmbedBackbone=Qwen2.5-VL-7B-Instruct2026.0375.45
MoTBackbone=Phi-4-Multimodal-Instruct2026.0374.03
MoTBackbone=Gemma3-12B2026.0372.84
MoTBackbone=Qwen2.5-VL-7B-Instruct2026.0372.08