Research map / Architectures and efficiency / Attention and transformers
Template-Based Math Word Problem Solvers with Recursive Neural Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add equation normalization (EN) (ablation) | T-RNN · Math23K | accuracy: 61.1 -> 66.9 | Helped |
| add equation normalization (EN) (ablation) | T-RNN · MAWPS | accuracy: 63.9 -> 66.8 | Helped |
| add equation normalization (ablation) | T-RNN template prediction · Math23K | template prediction accuracy: 59.6 -> 69.1 | Helped |
| add equation normalization (ablation) | T-RNN template prediction · MAWPS | template prediction accuracy: 62.2 -> 65.8 | Helped |
| use Bi-LSTM for quantity embedding instead of self attention directly on word embeddings (ablation) | T-RNN · MAWPS | accuracy: 31.1 -> 66.8 | Helped |
| use Bi-LSTM for quantity embedding (ablation) | T-RNN · Math23K | accuracy: 34.1 -> 66.9 | Helped |
| add self attention (ablation) | T-RNN · Math23K | accuracy: 65.1 -> 66.9 | Helped |
| add self attention (ablation) | T-RNN · MAWPS | accuracy: 66.3 -> 66.8 | No clear effect |
| T-RNN versus DNS generation model | T-RNN · Math23K | accuracy: 58.1 -> 66.9 | Helped |
| T-RNN versus Bi-LSTM classification | T-RNN · MAWPS | accuracy: 62.8 -> 66.8 | Helped |
| combine T-RNN with similarity-based retrieval | T-RNN · Math23K | accuracy: 66.9 -> 68.7 | Helped |
| combine T-RNN with similarity-based retrieval | T-RNN · MAWPS | accuracy: 66.8 -> 67.0 | No clear effect |
Closest papers by meaning
- Template-Driven LLM-Paraphrased Framework for Tabular Math Word Problem Generation (AAAI 2025 · similarity 0.85)
- Frugal LMs Trained to Invoke Symbolic Solvers Achieve Parameter-Efficient Arithmetic Reasoning (AAAI 2024 · similarity 0.85)
- Neuro-Symbolic Data Generation for Math Reasoning (NeurIPS 2024 · similarity 0.84)
- MathCoder2: Better Math Reasoning from Continued Pretraining on Model-translated Mathematical Code (ICLR 2025 · similarity 0.84)
- A Neural-Guided Dynamic Symbolic Network for Exploring Mathematical Expressions from Data (ICML 2024 · similarity 0.84)
- Neural-Symbolic Recursive Machine for Systematic Generalization (ICLR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.