Research map / Learning theory and algorithms / Combinatorial and search optimization
SPoC: Search-based Pseudocode to Code
Sumith Kulal, Panupong Pasupat, Kartik Chandra, Mina Lee, Oded Padon, Alex Aiken, Percy Liang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| Search over alternative line translations instead of using only the top-one translation | SPoC pseudocode-to-code synthesis · SPoC test set, 100 trials | success rate: 25.6% to 44.7% | Improved |
| Use multiclass error localization to down-weight the predicted offending line (ablation) | Best-first search · SPoC test examples | number of synthesis trials: 13.5%; median --26.0 trials; 0.4%; median +123.0 trials | Mixed |
| Use prefix-based pruning to blacklist a prefix verified to cause compilation failure (ablation) | Best-first search · SPoC test examples | number of synthesis trials: 4.1%; median --91.0 trials; 15.7%; median +12.0 trials | Mixed |
| Down-weight the compiler-reported error line rather than localizing the actual offending line (ablation) | Best-first search · SPoC test sets | success rate | Hurt |
| Use prefix-based pruning instead of multiclass localization (ablation) | Best-first search · SPoC; larger trial budgets | success rate | Improved |
| Use prefix-based pruning instead of multiclass localization (ablation) | Best-first search · SPoC; tight trial budgets | success rate | Hurt |
Closest papers by meaning
- Automatically Testing Functional Properties of Code Translation Models (AAAI 2024 · similarity 0.84)
- ASAP: Assembly-Source Aligned Pseudocode Refinement For Binary Decompilation (NeurIPS 2026 · similarity 0.83)
- Synthetic Programming Elicitation for Text-to-Code in Very Low-Resource Programming and Formal Languages (NeurIPS 2024 · similarity 0.83)
- From I/O to Code with Discovery Agent (NeurIPS 2026 · similarity 0.83)
- Simple Baselines are Competitive with Code Evolution (NeurIPS 2026 · similarity 0.83)
- Teaching Large Language Models to Self-Debug (ICLR 2024 · similarity 0.82)
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.
