Research map / Large language models / Agents, code and math
SatLM: Satisfiability-Aided Language Models Using Declarative Prompting
Xi Ye, Qiaochu Chen, Isil Dillig, Greg Durrett
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 |
|---|---|---|---|
| declarative SAT prompting vs program-aided prompting | SatLM · GSM-Sys (greedy) | accuracy: 43.4 -> 69.4 | Improved |
| declarative SAT prompting vs program-aided prompting | SatLM · GSM (greedy) | accuracy: 72.7 -> 71.8 | No clear effect |
| declarative SAT prompting vs program-aided prompting, self-consistency | SatLM · GSM (self-consistency) | accuracy: 82.4 -> 84.8 | Improved |
| declarative SAT prompting vs CoT | SatLM · Algebra (greedy) | accuracy: 53.6 -> 77.5 | Improved |
| declarative SAT prompting vs CoT | SatLM · LSAT (self-consistency) | accuracy: 30.9 -> 37.4 | Improved |
| declarative SAT prompting vs CoT | SatLM · BoardgameQA (self-consistency) | accuracy: 73.9 -> 80.7 | Improved |
| remove symbolic solver, predict answer directly (NoSolver) (ablation) | SatLM · GSM-Sys, Clutrr | accuracy | Hurt |
| replace SAT solver with CoT solver (ablation) | SatLM · GSM-Sys | accuracy: 7.9% | Hurt |
| declarative prompting with CoT solver instead of imperative CoT (ablation) | CoT · Clutrr | accuracy: 8.1% | Improved |
| SAT solver abstains on unsat/ambiguous outputs | SatLM · GSM-Sys | selective accuracy: 45.2 -> 77.1 | Improved |
| swap exemplar set from Set1 to Set2 (ablation) | SatLM · GSM-Sys | accuracy: 69.4 -> 71.8 | No clear effect |
| use adapted exemplar set instead of original | ProgLM · GSM-Sys | accuracy: 36.1 -> 43.4 | Improved |
Later papers that compared with this method
1 later accepted paper reports SAT-LM as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- A Balanced Neuro-Symbolic Approach for Commonsense Abductive Logic (ICLR 2026 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Reasoning-as-Logic-Units: Scaling Test-Time Reasoning in Large Language Models Through Logic Unit Alignment (ICML 2025 · similarity 0.89)
- Scaling Code-Assisted Chain-of-Thoughts and Instructions for Model Reasoning (NeurIPS 2025 · similarity 0.89)
- Putnam-AXIOM: A Functional & Static Benchmark for Measuring Higher Level Mathematical Reasoning in LLMs (ICML 2025 · similarity 0.88)
- ZebraLogic: On the Scaling Limits of LLMs for Logical Reasoning (ICML 2025 · similarity 0.88)
- QuestBench: Can LLMs ask the right question to acquire information in reasoning tasks? (NeurIPS 2025 · similarity 0.88)
- CRANE: Reasoning with constrained LLM generation (ICML 2025 · similarity 0.88)
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.
