Conversational Question Answering 벤치마크
Conversational Question Answering on ConvFinQA
Execution Accuracy
- 2022-10-07 — FinQANet (RoBERTa-large): Execution Accuracy 68.9
- 2022-12-14 — APOLLO: Execution Accuracy 78.76
| Rank | Model | Execution Accuracy | Program Accuracy | Paper | Code | Year |
|---|---|---|---|---|---|---|
| 1 | APOLLO | 78.76 | 77.19 | APOLLO: An Optimized Training Approach for Long-form Numerical Reasoning | gasolsun36/iter-cot · gasolsun36/dynamicrag · gasolsun36/apollo | 2022 |
| 2 | FinQANet (RoBERTa-large) | 68.90 | 68.24 | ConvFinQA: Exploring the Chain of Numerical Reasoning in Conversational Finance Question Answering | czyssrs/convfinqa | 2022 |
| 3 | APOLLO | 78.76 | 77.19 | APOLLO: An Optimized Training Approach for Long-form Numerical Reasoning | gasolsun36/iter-cot · gasolsun36/dynamicrag · gasolsun36/apollo | 2022 |
| 4 | FinQANet (RoBERTa-large) | 68.90 | 68.24 | ConvFinQA: Exploring the Chain of Numerical Reasoning in Conversational Finance Question Answering | czyssrs/convfinqa | 2022 |