List your company

Abstention in Question Answering on UMWP Underspecified Context

94Abstention F1TIAR
74.55279.60184.6589.699May 25, 2026

Evaluation results

MethodLinks
TIARModel=Qwen3-8B2026.059497.690.798.5
TruthRLModel=Qwen3-8B2026.0592.997.588.898.3
DPOModel=Qwen3-8B2026.0592.597.687.998.5
R-TuningModel=Qwen3-8B2026.0592.497.887.697.9
RFTModel=Qwen3-8B2026.0591.992.990.995.9
TruthRLModel=Llama-3.1-8B-Instruct2026.0578.965.599.290.3
TIARModel=Llama-3.1-8B-Instruct2026.0577.663.899.191.7
RFTModel=Llama-3.1-8B-Instruct2026.0576.161.799.286.7
DPOModel=Llama-3.1-8B-Instruct2026.0575.561.198.789.7
R-TuningModel=Llama-3.1-8B-Instruct2026.0575.360.799.186.2