Question Answering on FinQA
Metric: Execution Accuracy (higher is better)
Metric: Execution Accuracy (higher is better)
| # | Model↕ | Execution Accuracy▼ | Extra Data | Paper | Date↕ | Code |
|---|---|---|---|---|---|---|
| 1 | APOLLO | 71.07 | No | APOLLO: An Optimized Training Approach for Long-... | 2022-12-14 | Code |
| 2 | ELASTIC (RoBERTa-large) | 68.96 | No | ELASTIC: Numerical Reasoning with Adaptive Symbo... | 2022-10-18 | Code |
| 3 | GPT-4 (8k) | 68.79 | No | Are ChatGPT and GPT-4 General-Purpose Solvers fo... | 2023-05-10 | - |
| 4 | FinQANet (RoBERTa-large) | 65.05 | No | FinQA: A Dataset of Numerical Reasoning over Fin... | 2021-09-01 | Code |
| 5 | FinQANet (BERT-large) | 57.43 | No | FinQA: A Dataset of Numerical Reasoning over Fin... | 2021-09-01 | Code |
| 6 | FinQANet (FinBert ) | 53.71 | No | FinQA: A Dataset of Numerical Reasoning over Fin... | 2021-09-01 | Code |