Math Word Problem solving on MAWPS original (whole dataset)
91Value AccuracyDeBERTa (PM + VM)
Evaluation Results
| Method | Links | |
|---|---|---|
| DeBERTa (PM + VM)5-fold cross validation=true, Paraphrasing Model=true, Voting Mechanism=true2023.06 | 91 | |
| DeBERTa5-fold cross validation=true2023.06 | 90.7 | |
| RoBERTa5-fold cross validation=true2023.06 | 88.4 | |
| BERT5-fold cross validation=true2023.06 | 86.9 | |
| Transformer5-fold cross validation=true2023.06 | 85.6 | |
| gpt-3.5-turboNumber of parameters=175B, Evaluation setup=zero-shot2023.06 | 80.3 | |
| RNNVAE5-fold cross validation=true2023.06 | 79.8 | |
| RNNEncDec5-fold cross validation=true2023.06 | 79.4 | |
| GROUP-ATT5-fold cross validation=true2023.06 | 76.1 | |
| Math-EN5-fold cross validation=true2023.06 | 69.2 | |
| DNS5-fold cross validation=true2023.06 | 59.5 | |
| GPT-JNumber of parameters=6B, Evaluation setup=zero-shot2023.06 | 9.9 | |
| text-curie-001Number of parameters=13B, Evaluation setup=zero-shot2023.06 | 4.09 | |
| text-babbage-001Number of parameters=6.7B, Evaluation setup=zero-shot2023.06 | 2.76 |