Research map / Large language models / Agents, code and math
AdverMCTS: Combating Pseudo-Correctness in Code Generation via Adversarial Monte Carlo Tree Search
Qingyao Li, Weiwen Liu, Weinan Zhang, Yong Yu, Bo An
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 |
|---|---|---|---|
| add Attacker MCTS with adversarial testing (AT-Tree) (ablation) | AdverMCTS (Qwen3-4B-Instruct) · APPS and TACO | Pass@1 | Improved |
| add divergence-driven multi-sample test synthesis (DMTS) (ablation) | AdverMCTS (Qwen3-4B-Instruct) · APPS and TACO | Pass@1 | Improved |
| add pre-admission global screening (GF-Hub) (ablation) | AdverMCTS (Qwen3-4B-Instruct) · APPS and TACO | Pass@1 | Improved |
| add solver context (code pool) to the attacker (ablation) | AdverMCTS (Qwen3-4B-Instruct) · APPS (Avg.) | Pass@1: 46.0 -> 49.7 | Improved |
| add solver context (code pool) to the attacker (ablation) | AdverMCTS (Qwen3-4B-Instruct) · TACO (Avg.) | Pass@1: 34.7 -> 38.0 | Improved |
| use MCTS instead of Best-of-N for the attacker (ablation) | AdverMCTS (Qwen3-4B-Instruct) · APPS (Avg.) | Pass@1: 45.7 -> 49.7 | Improved |
| use UCB selection instead of random selection in attacker MCTS (ablation) | AdverMCTS (Qwen3-4B-Instruct) · APPS (Avg.) | Pass@1: 43.7 -> 49.7 | Improved |
| use LLM arbiter instead of majority voting for test outputs (ablation) | AdverMCTS (Qwen3-4B-Instruct) · APPS (Avg.) | Pass@1: 43.67 -> 49.67 | Improved |
| use LLM arbiter instead of majority voting for test outputs (ablation) | AdverMCTS (Qwen3-4B-Instruct) · TACO (Avg.) | Pass@1: 32.67 -> 37.33 | Improved |
| use test-based re-ranking instead of direct penalty (ablation) | AdverMCTS (Qwen3-4B-Instruct) · APPS Interview | Pass@1: 45.0 -> 52.0 | Improved |
| use test-based re-ranking instead of direct penalty (ablation) | AdverMCTS (Qwen3-4B-Instruct) · APPS Competition | Pass@1: 36.0 -> 36.0 | No clear effect |
| use test-based re-ranking instead of direct penalty (ablation) | AdverMCTS (Qwen3-4B-Instruct) · TACO Easy | Pass@1: 49.0 -> 55.0 | Improved |
Closest papers by meaning
- CodeCrash: Exposing LLM Fragility to Misleading Natural Language in Code Reasoning (NeurIPS 2025 · similarity 0.86)
- VERA: Variational Inference Framework for Jailbreaking Large Language Models (NeurIPS 2025 · similarity 0.86)
- CodeMimicry: Exploiting Safety Generalization Lag in Large Language Models via Structured Code Completion (NeurIPS 2026 · similarity 0.86)
- Jailbreak Antidote: Runtime Safety-Utility Balance via Sparse Representation Adjustment in Large Language Models (ICLR 2025 · similarity 0.86)
- Sample Smart, Not Hard: Correctness-First Decoding for Better Reasoning in LLMs (ICLR 2026 · similarity 0.86)
- Adversarial Reasoning at Jailbreaking Time (ICML 2025 · similarity 0.85)
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.
