Fact Generation on Wiki ExAnte (test)
11.76Leakage RateTCFT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TCFTBackbone=Qwen-2.5-14B-Instruct, Evaluation Protocol=Ours2026.05 | 11.76 | 73.81 | |
| GPT-5.4Model=GPT-5.42026.05 | 13.28 | 81.37 | |
| TCFTBackbone=Qwen-2.5-7B-Instruct, Evaluation Protocol=Ours2026.05 | 14.04 | 66.24 | |
| Gemini-3.1-Pro-PreviewModel=Gemini-3.1-Pro-Preview2026.05 | 15.75 | 79.82 | |
| SFTBackbone=Qwen-2.5-14B-Instruct, Evaluation Protocol=SFT2026.05 | 16.32 | 71.01 | |
| Claude-Opus-4.6Model=Claude-Opus-4.62026.05 | 16.86 | 81.3 | |
| Self-verificationBackbone=Qwen-2.5-14B-Instruct, Evaluation Protocol=Self-verification2026.05 | 20.49 | 70.73 | |
| SFTBackbone=Qwen-2.5-7B-Instruct, Evaluation Protocol=SFT2026.05 | 20.79 | 62.18 | |
| Zero-shotBackbone=Qwen-2.5-7B-Instruct, Evaluation Protocol=Zero-shot2026.05 | 21.82 | 61.67 | |
| Self-verificationBackbone=Qwen-2.5-7B-Instruct, Evaluation Protocol=Self-verification2026.05 | 22.01 | 62.81 | |
| Chain-of-thoughtBackbone=Qwen-2.5-7B-Instruct, Evaluation Protocol=Chain-of-thought2026.05 | 22.77 | 62.35 | |
| Chain-of-thoughtBackbone=Qwen-2.5-14B-Instruct, Evaluation Protocol=Chain-of-thought2026.05 | 23.15 | 67.86 | |
| Zero-shotBackbone=Qwen-2.5-14B-Instruct, Evaluation Protocol=Zero-shot2026.05 | 23.34 | 69.49 | |
| Few-shotBackbone=Qwen-2.5-7B-Instruct, Evaluation Protocol=Few-shot2026.05 | 26.76 | 59.96 | |
| Few-shotBackbone=Qwen-2.5-14B-Instruct, Evaluation Protocol=Few-shot2026.05 | 30.93 | 66.68 |