List your company

Question Answering on ARC Easy (ARC-E)

85.4Accuracy (ARC-E)MARI
50.35259.45168.5577.649May 27, 2026

Evaluation results

MethodLinks
MARIModel=Qwen2.5-14B2026.0585.4
CAAModel=Qwen2.5-14B2026.0585
NL-ITIModel=Qwen2.5-14B2026.0584.8
VanillaModel=Qwen2.5-14B2026.0584.7
ReFTModel=Qwen2.5-14B2026.0584.6
ITIModel=Qwen2.5-14B2026.0584.5
MARIModel=Qwen2.5-32B2026.0578.5
CAAModel=Qwen2.5-32B2026.0578.1
MARIModel=Llama-3-8B2026.0578
NL-ITIModel=Qwen2.5-32B2026.0577.9
VanillaModel=Qwen2.5-32B2026.0577.8
CAAModel=Llama-3-8B2026.0577.7
ReFTModel=Qwen2.5-32B2026.0577.7
ITIModel=Qwen2.5-32B2026.0577.6
NL-ITIModel=Llama-3-8B2026.0577.4
VanillaModel=Llama-3-8B2026.0577.3
ITIModel=Llama-3-8B2026.0577.2
ReFTModel=Llama-3-8B2026.0577.1
CAAModel=Qwen2-7B2026.0575.8
NL-ITIModel=Qwen2-7B2026.0575.6
VanillaModel=Qwen2-7B2026.0575.5
ReFTModel=Qwen2-7B2026.0575.4
ITIModel=Qwen2-7B2026.0575.3
MARIModel=Qwen2-7B2026.0575.2
CAAModel=Llama-2-13B2026.0553.9
NL-ITIModel=Llama-2-13B2026.0553.8
VanillaModel=Llama-2-13B2026.0553.7
ReFTModel=Llama-2-13B2026.0553.6
ITIModel=Llama-2-13B2026.0553.5
MARIModel=Llama-2-13B2026.0553.4
MARIModel=Llama-2-7B2026.0552.6
CAAModel=Llama-2-7B2026.0552.2
NL-ITIModel=Llama-2-7B2026.0552
VanillaModel=Llama-2-7B2026.0551.9
ReFTModel=Llama-2-7B2026.0551.8
ITIModel=Llama-2-7B2026.0551.7