Table Question Answering on ROBUT-WTQ Word-Level Paraphrase (test)
85.8Accuracy (Pre-Perturbation)TAPEX
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| TAPEXEvaluation Protocol=Fine-tuned2023.06 | 85.8 | 64.2 | 72.6 | |
| OmniTabEvaluation Protocol=Fine-tuned2023.06 | 84.7 | 64.3 | 72.6 | |
| TAPASEvaluation Protocol=Fine-tuned2023.06 | 82.7 | 58.9 | 69.9 | |
| TableFormerEvaluation Protocol=Fine-tuned2023.06 | 82.1 | 57 | 66.7 | |
| GPT-3Evaluation Protocol=Few-shot, Sample Size=2002023.06 | 76.3 | 72.2 | 92.2 | |
| OmniTabEvaluation Protocol=Fine-tuned2023.06 | 58 | 54.1 | 86.8 | |
| TAPEXEvaluation Protocol=Fine-tuned2023.06 | 54.7 | 49.2 | 84.3 | |
| TableFormerEvaluation Protocol=Fine-tuned2023.06 | 49.5 | 42.7 | 78.5 | |
| TAPASEvaluation Protocol=Fine-tuned2023.06 | 45.6 | 38.6 | 77.8 | |
| GPT-3Evaluation Protocol=Few-shot2023.06 | 41.2 | 40.3 | 93.7 |