Thresholded classification on HotpotQA
91.47F1 ScoreHIVE
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| HIVEModel=Dream-7B-Instruct2026.04 | 91.47 | 77.77 | 61.96 | |
| HIVEModel=LLaDA-8B-Instruct2026.04 | 91.2 | 79.39 | 65.53 | |
| CCSModel=Dream-7B-Instruct2026.04 | 90.53 | 73.72 | 53.54 | |
| CCSModel=LLaDA-8B-Instruct2026.04 | 90.3 | 74.13 | 53.45 | |
| TraceDetModel=Dream-7B-Instruct2026.04 | 89.78 | 76.48 | 62.68 | |
| Semantic EntropyModel=Dream-7B-Instruct2026.04 | 88.11 | 64.63 | 35.87 | |
| TSVModel=Dream-7B-Instruct2026.04 | 87.83 | 60.36 | 25.37 | |
| EigenScoreModel=Dream-7B-Instruct2026.04 | 87.61 | 54.07 | 9.8 | |
| Lexical SimilarityModel=Dream-7B-Instruct2026.04 | 87.37 | 50.25 | 0.55 | |
| LN-EntropyModel=Dream-7B-Instruct2026.04 | 87.27 | 50.01 | 0.02 | |
| PerplexityModel=Dream-7B-Instruct2026.04 | 87.19 | 50.47 | 1.42 | |
| TSVModel=LLaDA-8B-Instruct2026.04 | 86.74 | 57.3 | 17.41 | |
| Lexical SimilarityModel=LLaDA-8B-Instruct2026.04 | 86.49 | 58.07 | 20.13 | |
| EigenScoreModel=LLaDA-8B-Instruct2026.04 | 86.42 | 57.2 | 17.98 | |
| TraceDetModel=LLaDA-8B-Instruct2026.04 | 86.23 | 51.04 | 2.1 | |
| PerplexityModel=LLaDA-8B-Instruct2026.04 | 85.99 | 50.02 | 0.07 | |
| Semantic EntropyModel=LLaDA-8B-Instruct2026.04 | 85.99 | 50 | 0 | |
| LN-EntropyModel=LLaDA-8B-Instruct2026.04 | 85.96 | 50.42 | 1.21 |