arXiv cs.AI· Wenxuan Shi, Yaoyu Zhu, Di Huang, Chongxiao Li, Xiaoyun Zhang, Muxin Song, Yutong Wu, Ziyuan Nan, Pengwei Jin, Tian Zhi, Xishan Zhang, Rui Zhang, Xing Hu, Zidong Du, Qi Guo, Yunji Chen·· 1 天前AI 评分31
RTL-Interact：训练可调用 EDA 工具反馈的多任务 RTL 设计模型
RTL-Interact: Training EDA-Interactive Models for Multi-Task RTL Design
AI 导读
RTL-Interact 是一个多任务后训练框架，让单个 RTL 专用模型同时完成 RTL 生成、构建任务适配的 EDA 检查并依据工具反馈修订实现。在 CVDP 上，RTL-Interact-32B 取得 48.7% 加权 Pass@1 和 68.9% Pass@5，超过其教师模型 DeepSeek-V3.2，8B 版本加权 Pass@1 为 44.9%。
正文 · 原文
Authors:Wenxuan Shi, Yaoyu Zhu, Di Huang, Chongxiao Li, Xiaoyun Zhang, Muxin Song, Yutong Wu, Ziyuan Nan, Pengwei Jin, Tian Zhi, Xishan Zhang, Rui Zhang, Xing Hu, Zidong Du, Qi Guo, Yunji Chen
Abstract: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.
| Comments: | Under review at ICLR 2027 |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.41225 [cs.AI] |
| (or arXiv:2610.41225v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.41225 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Wenxuan Shi [view email]
[v1]
Fri, 30 Oct 2026 04:03:39 UTC (281 KB)
来源：arXiv cs.AI · arxiv.org