# Learn from Your Mistakes: Tree-like Self-Play for Secure Code LLMs

arXiv: https://arxiv.org/abs/2606.03489 · HTML: https://mlsanity.com/paper/2606.03489/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Code and instruction data synthesis, one of 13 directions in Synthetic data and self-improvement for LLMs.
- Builds on 37 of its 47 references that are in the map.
- First posted on arXiv 2 Jun 2026.

## Placed in

- Synthetic data and self-improvement for LLMs / [Code and instruction data synthesis](https://mlsanity.com/synth/code-and-instruction-data-synthesis/)

## Builds on (most cited first, 30 of 37)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Deep reinforcement learning from human preferences](https://mlsanity.com/paper/1706.03741/) (2017)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [Scaling Instruction-Finetuned Language Models](https://mlsanity.com/paper/2210.11416/) (2022)
- [Competition-Level Code Generation with AlphaCode](https://mlsanity.com/paper/2203.07814/) (2022)
- [Dota 2 with Large Scale Deep Reinforcement Learning](https://mlsanity.com/paper/1912.06680/) (2019)
- [WizardCoder: Empowering Code Large Language Models with Evol-Instruct](https://mlsanity.com/paper/2306.08568/) (2023)
- [Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models](https://mlsanity.com/paper/2401.01335/) (2024)
- [Asleep at the Keyboard? Assessing the Security of GitHub Copilot's Code Contributions](https://mlsanity.com/paper/2108.09293/) (2021)
- [CodeRL: Mastering Code Generation through Pretrained Models and Deep Reinforcement Learning](https://mlsanity.com/paper/2207.01780/) (2022)
- [Preference Ranking Optimization for Human Alignment](https://mlsanity.com/paper/2306.17492/) (2023)
- [Magicoder: Empowering Code Generation with OSS-Instruct](https://mlsanity.com/paper/2312.02120/) (2023)
- [Do Users Write More Insecure Code with AI Assistants?](https://mlsanity.com/paper/2211.03622/) (2022)
- [Preference Fine-Tuning of LLMs Should Leverage Suboptimal, On-Policy Data](https://mlsanity.com/paper/2404.14367/) (2024)
- [Direct Nash Optimization: Teaching Language Models to Self-Improve with General Preferences](https://mlsanity.com/paper/2404.03715/) (2024)
- [Large Language Models for Code: Security Hardening and Adversarial Testing](https://mlsanity.com/paper/2302.05319/) (2023)
- [RLEF: Grounding Code LLMs in Execution Feedback with Reinforcement Learning](https://mlsanity.com/paper/2410.02089/) (2024)
- [A Minimaximalist Approach to Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2401.04056/) (2024)
- [Token-level Direct Preference Optimization](https://mlsanity.com/paper/2404.11999/) (2024)
- [Understanding the performance gap between online and offline alignment algorithms](https://mlsanity.com/paper/2405.08448/) (2024)
- [TreeRL: LLM Reinforcement Learning with On-Policy Tree Search](https://mlsanity.com/paper/2506.11902/) (2025)
- [Instruction Tuning for Secure Code Generation](https://mlsanity.com/paper/2402.09497/) (2024)
- [Data Quality for Software Vulnerability Datasets](https://mlsanity.com/paper/2301.05456/) (2023)
- [Align Anything: Training All-Modality Models to Follow Instructions with Language Feedback](https://mlsanity.com/paper/2412.15838/) (2024)
- [AutoSafeCoder: A Multi-Agent Framework for Securing LLM Code Generation through Static Analysis and Fuzz Testing](https://mlsanity.com/paper/2409.10737/) (2024)
- [CodeDPO: Aligning Code Models with Self Generated and Verified Source Code](https://mlsanity.com/paper/2410.05605/) (2024)
- [Getting More Juice Out of the SFT Data: Reward Learning from Human Demonstration Improves SFT for LLM Alignment](https://mlsanity.com/paper/2405.17888/) (2024)
- [ProSec: Fortifying Code LLMs with Proactive Security Alignment](https://mlsanity.com/paper/2411.12882/) (2024)
- [VUDENC: Vulnerability Detection with Deep Learning on a Natural Codebase for Python](https://mlsanity.com/paper/2201.08441/) (2022)

## Abstract (arXiv)

While Large Language Models (LLMs) excel in code generation, they remain prone to replicating subtle yet critical vulnerabilities endemic to their training data. Current alignment techniques, such as Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL), typically apply coarse-grained optimization at the sequence level. This approach often fails to address the localized nature of security flaws, where a single incorrect token choice can compromise an entire program. To bridge this gap, we introduce Tree-like Self-Play (TSP), a framework that reframes secure code generation as a fine-grained sequential decision process. Unlike standard methods that blindly maximize likelihood, TSP constructs a decision tree where the model explores branching trajectories--generating both secure "golden paths" and vulnerable variants. By treating code generation as a self-play game, the model learns to strictly discriminate against its own localized errors. This provides a dense, on-policy learning signal that forces self-correction precisely at the critical decision nodes where vulnerabilities typically emerge. Our experiments demonstrate that TSP fundamentally enhances model reliability. In Python security benchmarks, TSP boosts CodeLlama-7B's pass rate (SPR@1) to 75.8%, significantly outperforming SFT (57.0%) and unstructured self-play baselines. Crucially, TSP induces robust out-of-distribution generalization: the model not only reduces vulnerabilities in unseen categories (CWEs) by 24.5% but also successfully transfers security principles learned from C/C++ to diverse languages, including Python, Go, and JavaScript. This suggests that TSP does not merely memorize patches, but internalizes abstract, language-agnostic security logic.