List your company

Compositional Reasoning Dataset

43.3Correction Score (C)CREME
-0.785610.659722.10533.5503Feb 22, 2024

Evaluation results

MethodLinks
CREMEBackbone=OpenAlpaca-3B2024.0243.323.713.611.24
CREMEBackbone=LLAMA-2-7B2024.02177.991.270.86
OpenAlpaca-3BModel Type=Base Model2024.027.2713.50.6
LLAMA-2-7BModel Type=Base Model2024.023.22.313.10.3
Memory InjectionBackbone=LLAMA-2-7B2024.022.210.30.3226.72
CoT-PatchScopesBackbone=LLAMA-2-7B2024.021.2———
Memory InjectionBackbone=OpenAlpaca-3B2024.020.980.450.752.93
CoT-PatchScopesBackbone=OpenAlpaca-3B2024.020.91———