Mathematical Equation Generation on MAWPS (5-fold cross-validation)
92.5Accuracy (5-fold)Self-Consistency†
Evaluation results
| Method | Links | |
|---|---|---|
| Self-Consistency†Method Category=LLM2023.10 | 92.5 | |
| Expression tree decoding strategy (Ours)Layer-Shared=false2023.10 | 92.3 | |
| RE-ExtMethod Category=Seq2Exp2023.10 | 92.2 | |
| Expression tree decoding strategy (Ours Layer-Shared)Layer-Shared=true2023.10 | 92.2 | |
| M-ViewMethod Category=Seq2Exp2023.10 | 92.1 | |
| Ana-CL†Method Category=Seq2Seq / Tree2023.10 | 91.8 | |
| ElasticMethod Category=Seq2Exp2023.10 | 91.8 | |
| gpt-3.5-turbo†Method Category=LLM2023.10 | 91.5 | |
| Textual-CL†Method Category=Seq2Seq / Tree2023.10 | 91.3 | |
| H-ReasonerMethod Category=Seq2Seq / Tree2023.10 | 89.8 | |
| Prototype†Method Category=Seq2Seq / Tree2023.10 | 89.6 | |
| PLM-G2TMethod Category=Seq2Seq / Tree2023.10 | 88.7 | |
| PLM-GTSMethod Category=Seq2Seq / Tree2023.10 | 88.5 | |
| PLM-GenMethod Category=Seq2Seq / Tree2023.10 | 88.4 | |
| MWP-NASMethod Category=Seq2Exp2023.10 | 88 | |
| BERTGenMethod Category=Seq2Seq / Tree2023.10 | 86.9 | |
| G2TMethod Category=Seq2Seq / Tree2023.10 | 85.6 | |
| T-DisMethod Category=Seq2Seq / Tree2023.10 | 84.2 | |
| RankMethod Category=Seq2Seq / Tree2023.10 | 84 | |
| E-PointerMethod Category=Seq2Exp2023.10 | 83.4 | |
| GTSMethod Category=Seq2Seq / Tree2023.10 | 82.6 | |
| M-TreeMethod Category=Seq2Exp2023.10 | 82 | |
| GroupAttnMethod Category=Seq2Seq / Tree2023.10 | 76.1 |