arXiv:2511.20099v2 Announce Type: replace 
Abstract: Large language models (LLMs) have shown promising capabilities in hardware description language (HDL) generation. However, existing approaches often rely on free-form natural language descriptions that are often ambiguous, redundant, and unstructured, which poses significant challenges for downstream Verilog code generation. We treat hardware code generation as a complex transformation from an open-ended natural language space to a domain-specific, highly constrained target space. To bridge this gap, we introduce Core Refined Understanding eXpression (CRUX), a structured intermediate space that captures the essential semantics of user intent while organizing the expression for precise Verilog code generation. We further design a two-stage training framework, comprising Joint Expression Modeling and Dual-Space Optimization, to enhance the quality of both CRUX and Verilog code. Experiments across multiple Verilog generation benchmarks demonstrate that our model, CRUX-V, achieves state-of-the-art performance among general models, particularly under challenging design tasks. Furthermore, the CRUX space proves transferable and beneficial when used as input prompts for other code models, highlighting its effectiveness in narrowing the gap between free-form natural language descriptions and precise Verilog generation.

قدمت QiMeng-CRUX نهجًا جديدًا لسد الفجوة بين اللغة الطبيعية وتوليد كود Verilog من خلال تعبير الفهم المنقح الأساسي (CRUX)، الذي ينظم نية المستخدم في تنسيق منظم لتوليد كود دقيق. تتناول هذه الطريقة التحديات التي تطرحها الأوصاف باللغة الطبيعية الغامضة وغير المنظمة في توليد لغات وصف الأجهزة (HDL).

QiMeng-CRUX ha presentado un enfoque novedoso para cerrar la brecha entre el lenguaje natural y la generación de código Verilog a través de su Core Refined Understanding eXpression (CRUX), que organiza la intención del usuario en un formato estructurado para una generación de código precisa. Este método aborda los desafíos planteados por descripciones en lenguaje natural ambiguas y no estructuradas en la generación de lenguajes de descripción de hardware (HDL).

QiMeng-CRUX a introduit une approche novatrice pour combler le fossé entre le langage naturel et la génération de code Verilog grâce à son Core Refined Understanding eXpression (CRUX), qui organise l'intention de l'utilisateur dans un format structuré pour une génération de code précise. Cette méthode répond aux défis posés par des descriptions en langage naturel ambiguës et non structurées dans la génération de langages de description de matériel (HDL).

QiMeng-CRUX has introduced a novel approach to bridge the gap between natural language and Verilog code generation through its Core Refined Understanding eXpression (CRUX), which organizes user intent into a structured format for precise code generation. This method addresses the challenges posed by ambiguous and unstructured natural language descriptions in hardware description language (HDL) generation.

QiMeng-CRUX: Narrowing the Gap between Natural Language and Verilog via Core Refined Understanding eXpression

Was this article worth reading? Share it

LucidQuery AI

Langfuse

IntelliQ

AQ

ConsoleX

ClassX

Ready to build your own newsroom?