List your company

Adversarial Factuality

99.04CRSAC Single-task
69.961677.510885.0692.6092Nov 4, 2024

Evaluation results

MethodLinks
SAC Single-taskBackbone=Qwen2-72B-Instruct2024.1199.04
SAC Multi-taskBackbone=Qwen2-72B-Instruct2024.1198.56
No ControlBackbone=Qwen2-72B-Instruct2024.1198.08
SAC Multi-taskBackbone=Qwen2-7B-Instruct2024.1196.65
SAC Single-taskBackbone=Qwen2-7B-Instruct2024.1195.22
RepEControl Dim=Single2024.1190.43
SACControl Dim=Single2024.1189.47
SACControl Dim=Multiple2024.1186.12
No ControlControl Dim=Single2024.1176.56
No ControlControl Dim=Multiple2024.1176.56
No ControlBackbone=Qwen2-7B-Instruct2024.1176.08
RepE-MeanControl Dim=Multiple2024.1172.59
RepE-MergeControl Dim=Multiple2024.1171.08