List your company

LLM Utility Evaluation on LIMA original and rewritten variants

77.6ARC-E AccuracyCBBR
66.6869.51572.3575.185May 18, 2026

Evaluation results

MethodLinks
CBBRModel=Llama3.1-8B, Rewriter=CBBR2026.0577.647.980.878.773.663.271.9-5.9
CBBRModel=Qwen2.5-7B, Rewriter=CBBR2026.0576.548.179.879.170.371.167.5-8.1
Original (No Rewriter)Model=Qwen2.5-1.5B, Rewriter=None2026.0575.344.367.375.163.558.538.1—
OBBRModel=Qwen2.5-1.5B, Rewriter=OBBR2026.05754467.27563.558.538.60
CBBRModel=Qwen2.5-1.5B, Rewriter=CBBR2026.0574.744.267.87564.157.936.10.8
DPRModel=Qwen2.5-1.5B, Rewriter=DPR2026.0574.343.667.27564.258.140.3-0.5
ParaphraseModel=Qwen2.5-7B, Rewriter=Para.2026.0574.34779.87866.167.260.5-3.6
ParaphraseModel=Qwen2.5-1.5B, Rewriter=Para.2026.0573.843.86774.763.458.339.30.2
DPRModel=Llama3.1-8B, Rewriter=DPR2026.0573.847.281.278.67261.268.9-3.5
ParaphraseModel=Llama3.1-8B, Rewriter=Para.2026.0572.445.380.977.669.360.669.3-1.8
OBBRModel=Qwen2.5-7B, Rewriter=OBBR2026.0570.444.179.876.367.26862.4-2.4
Original (No Rewriter)Model=Llama3.1-8B, Rewriter=None2026.0569.944.580.476.37161.264—
DPRModel=Qwen2.5-7B, Rewriter=DPR2026.0569.642.880.277.465.36860.1-1.1
OBBRModel=Llama3.1-8B, Rewriter=OBBR2026.0568.943.680.677.571.46166.6-0.4
OBBRModel=Llama3.2-1B, Rewriter=OBBR2026.0567.934.960.874.562.945.851.3-1.5
DPRModel=Llama3.2-1B, Rewriter=DPR2026.0567.934.8617462.445.349.6-0.7
CBBRModel=Llama3.2-1B, Rewriter=CBBR2026.0567.535.161.474.662.244.950.5-1
Original (No Rewriter)Model=Llama3.2-1B, Rewriter=None2026.0567.334.76273.662.44448.9—
Original (No Rewriter)Model=Qwen2.5-7B, Rewriter=None2026.0567.243.879.876.663.367.759.4—
ParaphraseModel=Llama3.2-1B, Rewriter=Para.2026.0567.135.862.173.962.443.351.4-1