RTL-Interact 发布：单一模型可结合 EDA 工具反馈修正 RTL 代码
RTL-Interact: Training EDA-Interactive Models for Multi-Task RTL Design
芯片设计方向的新工作，32B 模型在 CVDP 上超过了它的老师 DeepSeek-V3.2，还能自己调 EDA 工具改代码，做 RTL 的可以看看。
RTL-Interact 是一个多任务后训练框架，训练单个 RTL 专用模型完成生成 RTL、构建 EDA 检查、根据工具反馈修码三类任务。框架引入 Pattern-Aware Policy Optimization（PAPO），在 GRPO 基础上加入正确性门控奖励。RTL-Interact-32B 在 CVDP 基准上达到 48.7% 加权 Pass@1 和 68.9% Pass@5，超过其教师模型 DeepSeek-V3.2，8B 版本达到 44.9%。PAPO 还将平均工具调用次数减少 63.8%，同时提升了 Pass@1，两个模型在 VerilogEval v2 和 RTLLM v2 上均超过现有 RTL 专用基线。
RTL-Interact: Training EDA-Interactive Models for Multi-Task RTL Design
Large language models offer a promising route to register-transfer-level (RTL) design, but reliable implementations require checking and revision based on electronic design automation (EDA) tools' feedback. Existing RTL-specialized models often lack learned interaction with such EDA tools or rely on separate generation and repair components. We introduce RTL-Interact, a multi-task post-training framework that trains a single RTL-specialized model to generate RTL, construct task-appropriate EDA checks, and revise implementations using tool feedback. The framework combines automatically constructed multi-task demonstrations with staged supervision from chain-of-thought to tool-integrated reasoning. We further introduce Pattern-Aware Policy Optimization (PAPO) for multi-task reinforcement learning. PAPO augments Group Relative Policy Optimization with correctness-gated rewards that favor concise successful checking while rewarding effective revision, and uses task grouping to preserve coverage across objectives. On CVDP, RTL-Interact-32B achieves 48.7% weighted Pass@1 and 68.9% Pass@5, surpassing its teacher model, DeepSeek-V3.2, while RTL-Interact-8B achieves 44.9% weighted Pass@1. Additionally, we show that PAPO can reduce average tool calls by 63.8% while improving Pass@1. Both models also outperform all RTL-specialized baselines on VerilogEval v2 and RTLLM v2, demonstrating the effectiveness of our method.
