Factuality Detection on TriviaQA (test)
85.5AUROCDynHD
Evaluation Results
| Method | Links | |
|---|---|---|
| DynHDSource Dataset=HotpotQA, Method Category=Trajectory-based, Evaluation Protocol=Zero-shot cross-task generalization2026.03 | 85.5 | |
| TraceDetSource Dataset=HotpotQA, Method Category=Trajectory-based, Evaluation Protocol=Zero-shot cross-task generalization2026.03 | 73.1 | |
| DynHDSource Dataset=CSQA, Method Category=Trajectory-based, Evaluation Protocol=Zero-shot cross-task generalization2026.03 | 68.7 | |
| TSVSource Dataset=CSQA, Method Category=Latent-based, Evaluation Protocol=Zero-shot cross-task generalization2026.03 | 65.3 | |
| TraceDetSource Dataset=CSQA, Method Category=Trajectory-based, Evaluation Protocol=Zero-shot cross-task generalization2026.03 | 61.5 | |
| TSVSource Dataset=HotpotQA, Method Category=Latent-based, Evaluation Protocol=Zero-shot cross-task generalization2026.03 | 58.5 | |
| CCSSource Dataset=CSQA, Method Category=Latent-based, Evaluation Protocol=Zero-shot cross-task generalization2026.03 | 54.5 | |
| CCSSource Dataset=HotpotQA, Method Category=Latent-based, Evaluation Protocol=Zero-shot cross-task generalization2026.03 | 50.1 |