List your company

Science QA on ARC easy

77.9AccuracyVanilla
31.51643.55855.667.642May 28, 2026Jun 1, 2026Jun 6, 2026Jun 11, 2026Jun 15, 2026Jun 20, 2026Jun 25, 2026

Evaluation results

MethodLinks
VanillaBackbone=Mistral-7B-Instruct2026.0577.9
DT-CD⋆Backbone=Mistral-7B-Instruct2026.0577.8
COFTBackbone=Mistral-7B-Instruct2026.0577.8
SDDBackbone=Mistral-7B-Instruct2026.0577.4
DExpertsBackbone=Mistral-7B-Instruct2026.0577.2
VanillaBackbone=LLaMA-2-13B2026.0574.6
COFTBackbone=LLaMA-2-13B2026.0574.5
DT-CD⋆Backbone=LLaMA-2-13B2026.0574.4
SDDBackbone=LLaMA-2-13B2026.0574
DExpertsBackbone=LLaMA-2-13B2026.0573.7
PrismScale=Large, Evaluation Protocol=Zero-shot2026.0645.8
BaselineScale=Large, Evaluation Protocol=Zero-shot2026.0645.5
PrismScale=Medium, Evaluation Protocol=Zero-shot2026.0643.1
BaselineScale=Medium, Evaluation Protocol=Zero-shot2026.0642.1
PrismScale=Small, Evaluation Protocol=Zero-shot2026.0637.7
BaselineScale=Small, Evaluation Protocol=Zero-shot2026.0637.6
Mamba-2model size=152M, mb=42026.0636.4
SISAmodel size=152M, ds=32, mb=22026.0635.8
SISAmodel size=152M, ds=64, mb=42026.0635.4
SISAmodel size=152M, ds=128, mb=42026.0635
Mamba-3model size=152M, mb=42026.0634.9
SISAmodel size=152M, ds=16, mb=42026.0634.7
Transformermodel size=152M, mb=42026.0633.3