List your company

Hallucination prediction on Mintaka refined by question type

77.89AUROCConf + Probe (SCAO)
64.078867.664471.2574.8356Sep 18, 2025

Evaluation results

MethodLinks
Conf + Probe (SCAO)Model Size=8B, SCAO=true2025.0977.8913.83
Conf + Probe (SCAO)Model Size=8B2025.0977.8913.83
Conf + ProbeModel Size=70B2025.0977.6813.33
Conf + ProbeModel Size=70B2025.0977.6813.33
Conf + ProbeModel Size=8B2025.0977.5413.48
Conf + ProbeModel Size=8B2025.0977.5413.48
Conf + Probe (SCAO)Model Size=70B, SCAO=true2025.0977.2212.87
Conf + Probe (SCAO)Model Size=70B2025.0977.2212.87
Probe_dnnModel Size=8B2025.0975.711.64
Probe_dnnModel Size=8B2025.0975.711.64
Probe_dnnModel Size=70B2025.0975.5410.19
Probe_dnnModel Size=70B2025.0975.5410.19
Conf (SCAO)Model Size=70B, SCAO=true2025.0972.78—
Conf (SCAO)Model Size=70B2025.0972.78—
Conf (SCAO)Model Size=8B, SCAO=true2025.0972.01—
Conf (SCAO)Model Size=8B2025.0972.01—
ConfModel Size=70B2025.0967.16—
ConfModel Size=70B2025.0967.16—
ConfModel Size=8B2025.0964.61—
ConfModel Size=8B2025.0964.61—