Automating Infrastructure-as-Code (IaC) is challenging, and large language models (LLMs) often produce incorrect configurations from natural language (NL). We present TerraFormer, a neuro-symbolic framework for IaC generation and mutation that combines supervised fine-tuning with verifier-guided reinforcement learning, using formal verification tools to provide feedback on syntax, deployability, and policy compliance. We curate two large, high-quality NL-to-IaC datasets, TF-Gen (152k instances) and TF-Mutn (52k instances), via multi-stage verification and iterative LLM self-correction. Evaluations against 17 state-of-the-art LLMs, including ~50x larger models like Sonnet 3.7, DeepSeek-R1, and GPT-4.1, show that TerraFormer improves correctness over its base LLM by 15.94% on IaC-Eval, 11.65% on TF-Gen (Test), and 19.60% on TF-Mutn (Test). It outperforms larger models on both TF-Gen (Test) and TF-Mutn (Test), ranks third on IaC-Eval, and achieves top best-practices and security compliance.

翻译：自动化基础设施即代码（IaC）具有挑战性，大型语言模型（LLM）常从自然语言（NL）生成错误的配置。我们提出TerraFormer，一个用于IaC生成与变更的神经符号框架，它结合了监督微调与验证器引导的强化学习，利用形式化验证工具提供语法、可部署性及策略合规性反馈。我们通过多阶段验证与迭代式LLM自校正，构建了两个大规模高质量的NL-to-IaC数据集：TF-Gen（15.2万实例）和TF-Mutn（5.2万实例）。在针对17个前沿LLM（包括参数量约50倍于本模型的Sonnet 3.7、DeepSeek-R1和GPT-4.1）的评估中，TerraFormer在IaC-Eval上将其基础LLM的正确率提升了15.94%，在TF-Gen（测试集）上提升11.65%，在TF-Mutn（测试集）上提升19.60%。该模型在TF-Gen（测试集）和TF-Mutn（测试集）上均超越更大规模的模型，在IaC-Eval中位列第三，并在最佳实践与安全合规性方面达到最优水平。