HDLFORGE: A Two-Stage Multi-Agent Framework for Efficient Verilog Code Generation with Adaptive Model Escalation Conference: 30th Asia and South Pacific Design Automation Conference; January 20–23, 2025; Tokyo, Japan 30th Asia and South Pacific Design Automation Conference (ASPDAC ’25), January 20–23, 2025, Tokyo, Japan DOI: 10.1145/3658617.3697777 ISBN: 979-8-4007-0635-6/25/01
Armin Abdollahi, Saeid Shokoufa, Negin Ashrafi, Mehdi Kamal, Massoud Pedram · 2026
Read HDLFORGE: A Two-Stage Multi-Agent Framework for Efficient Verilog Code Generation with Adaptive Model Escalation Conference: 30th Asia and South Pacific Design Automation Conference; January 20–23, 2025; Tokyo, Japan 30th Asia and South Pacific Design Automation Conference (ASPDAC ’25), January 20–23, 2025, Tokyo, Japan DOI: 10.1145/3658617.3697777 ISBN: 979-8-4007-0635-6/25/01 by Armin Abdollahi, Saeid Shokoufa, Negin Ashrafi, Mehdi Kamal, Massoud Pedram (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-04646.txt.
Description
Abdollahi and colleagues generate Verilog code with two-stage HDLFORGE. The framework balances generation speed against accuracy. A coder plus a mid-size LLM runs in Stage A. It escalates to a bigger coder plus a huge LLM only when diagnostics say so. A counterexample-guided agent turns checking traces into micro-tests for reuse, cutting detection time. The escalation controller wraps existing Verilog pipelines without changing their internals. HDLFORGE-Qwen reaches over 91 percent Pass at 1 on VerilogEval Human and V2, and over 97 percent Pass at 5 on RTLLM.
Catalog record
| CREATOR | Armin Abdollahi, Saeid Shokoufa, Negin Ashrafi, Mehdi Kamal, Massoud Pedram |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37447 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.04646