Claim Verification on Financial QA Datasets Aggregate
71.9Unconditional AccuracyGPT-4o + CoT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GPT-4o + CoTPrompting Strategy=Chain of Thought (CoT)2026.04 | 71.9 | — | |
| FINGROUNDAblation Components=full2026.04 | 71.2 | 94.7 | |
| FINGROUNDAblation Components=- taxonomy2026.04 | 70.5 | 92.1 | |
| Self-RAG2026.04 | 68.2 | — | |
| CRAG2026.04 | 67 | — | |
| FActScore2026.04 | 66.2 | 89.3 | |
| FINGROUNDAblation Components=- table ret.2026.04 | 66.2 | 91.8 | |
| Vanilla RAG2026.04 | 63.9 | — | |
| FINGROUNDAblation Components=- regen.2026.04 | 63.8 | 94.7 |