Fact Verification on FEVER (Accuracy, ECE)
99.7AccuracyLPF-SPN
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LPF-SPNRuntime (ms)=25.2, Speedup=1.0×2026.03 | 99.7 | 1.2 | |
| LPF-LearnedRuntime (ms)=24.0, Speedup=1.0×2026.03 | 99.7 | 0.3 | |
| VAE-OnlyRuntime (ms)=3.5, Speedup=7.2×2026.03 | 99.7 | 0.3 | |
| Qwen3-32BRuntime (ms)=3176.4, Speedup=0.008×2026.03 | 62 | 82.3 | |
| Kimi-K2Runtime (ms)=609.5, Speedup=0.041×2026.03 | 56 | 87.3 | |
| GPT-OSS-120BRuntime (ms)=1718.2, Speedup=0.015×2026.03 | 54 | 86.8 | |
| Llama-3.3-70BRuntime (ms)=1581.6, Speedup=0.016×2026.03 | 44 | 74.4 |