Fact Verification on FEVER (test) (Accuracy, F1, ECE, Runtime)
99.7AccuracyLPF-SPN
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| LPF-SPN2026.03 | 99.7 | 99.7 | 1.2 | 25.2 | |
| LPF-Learned2026.03 | 99.7 | 99.7 | 0.3 | 24 | |
| VAE-Only2026.03 | 99.7 | 99.7 | 0.3 | 3.5 | |
| BERT2026.03 | 95.2 | 95.1 | 8.9 | 142.3 | |
| SPN-Only2026.03 | 95.2 | 95.1 | 28.9 | 0.9 | |
| Groq-qwen3-32bEvaluation Protocol=zero-shot, Model Parameters=32B2026.03 | 62 | 62 | 82.3 | 3,176.4 | |
| EDL-AggregatedAggregation=Aggregated2026.03 | 50.2 | 22.3 | 16.7 | 1.2 | |
| EDL-IndividualAggregation=Individual2026.03 | 50.2 | 22.3 | 0.1 | 1.9 | |
| Groq-llama-3.3-70bEvaluation Protocol=zero-shot, Model Parameters=70B2026.03 | 44 | 44 | 74.4 | 1,581.6 | |
| R-GCN2026.03 | 22.8 | 12.4 | 10.5 | 0.001 |