List your company

Commonsense Fact Verification on CREAK (test)

88.6AccuracyRACO
49.28859.49469.779.906May 10, 2023Oct 2, 2023Feb 24, 2024Jul 18, 2024Dec 10, 2024May 4, 2025Sep 26, 2025

Evaluation results

MethodLinks
RACO#Total Params.=≥ 3B, Single-task Training=false2023.0588.6
DECKER#Total Params.=~449M, Single-task Training=true2023.0588.4
T5-3B#Total Params.=~3B, Single-task Training=false2023.0585.1
UNICORN#Total Params.=~770M, Single-task Training=false2023.0579.5
ICRBackbone=Qwen2.5-7B2025.0979.2
Zero-shotBackbone=Qwen2.5-7B2025.0977.8
GreaseLM#Total Params.=~359M, Single-task Training=true2023.0577.5
I2CLBackbone=Qwen2.5-7B2025.0974.8
M²IVBackbone=Qwen2.5-7B2025.0974.6
LIVEBackbone=Qwen2.5-7B2025.0972.8
Few-shot*Backbone=Qwen2.5-7B, Few-shot settings=3-shot OOD2025.0959.7
ICRBackbone=Llama2-7B2025.0956.4
M²IVBackbone=Llama2-7B2025.0954
I2CLBackbone=Llama2-7B2025.0952.6
Zero-shotBackbone=Llama2-7B2025.0951.8
LIVEBackbone=Llama2-7B2025.0951
Few-shot*Backbone=Llama2-7B, Few-shot settings=3-shot OOD2025.0950.8