用多个专家模型协作生成高覆盖率测试用例，效果远超单一大模型。
CHORUS: Complementary Experts for High-Coverage Testbench Stimulus Generation
- 通过分阶段微调和强化学习，生成行为各异的专家模型。
- 合并专家后在CVDP-ECov上达88.0%通过率，领先671B大模型13.5个百分点。
- 无需额外训练即可融合专家优势，适合硬件验证场景使用。
大型语言模型（LLMs）在代码生成方面取得进展，其中可执行反馈比纯文本模仿提供更可靠的训练信号。硬件验证是代码生成的重要应用，占现代芯片设计工作量的很大比例，高覆盖率测试用例生成是其中关键任务。我们提出CHORUS，一种后训练框架，性能超越传统监督微调（SFT）到强化学习（RL）的流水线。CHORUS基于两个观察：首先，分阶段SFT产生行为多样性的检查点，密集奖励的RL将它们转化为性能相当但任务层面优势互补的强专家；其次，这些互补优势可通过无需训练的模型融合或进一步后训练加以利用，从而超越最佳单一专家。通过将所得专业模型整合为一个40亿参数模型，CHORUS在CVDP-ECov上达到88.0% Pass@1，优于DeepSeek-R1（671B）13.5个百分点。
原文摘要 · Abstract (English)
Large language models (LLMs) have advanced code generation, where executable feedback provides a more reliable learning signal than textual imitation alone. Hardware verification is an important application of code generation and accounts for a substantial fraction of modern chip design effort, with high-coverage testbench stimulus generation as a key task. We present CHORUS, a post-training framework that pushes performance beyond what a conventional supervised fine-tuning (SFT)-to-reinforcement learning (RL) pipeline achieves. CHORUS builds on two observations. First, staged SFT produces behaviorally diverse checkpoints, and dense-reward RL turns them into strong experts with comparable aggregate performance but distinct task-level strengths. Second, these complementary strengths can be exploited through either training-free model merging or further post-training to outperform the best individual expert. By consolidating the resulting specialists into a single 4B model, CHORUS achieves 88.0% Pass@1 on CVDP-ECov, outperforming DeepSeek-R1 (671B) by 13.5 percentage points.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。