Unitho: A Unified Multi-Task Framework for Computational Lithography

2025-11-13

AI Summary

In computational lithography, tasks such as mask synthesis, design rule violation detection, and layout optimization have traditionally been modeled in isolation, hindered by data scarcity and methodological fragmentation. Method: This paper introduces the first unified multi-task vision large model framework tailored for lithography, built upon a Transformer architecture and trained end-to-end on large-scale industrial-grade simulation data to jointly learn and share knowledge across these three core tasks. Contribution/Results: The framework breaks from conventional single-task paradigms, significantly improving generalization capability and modeling fidelity—outperforming existing academic baselines across multiple benchmarks. It establishes a high-reliability intelligent EDA data foundation and a scalable model infrastructure, providing a novel paradigm for computational lithography.

Technology Category

• Computer Vision: Large Vision Models
• Machine Learning: Large Multimodal Models (LMMs)
• Cognitive Modeling & Cognitive Systems: (Computational) Cognitive Architectures

Application Category

• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

Reliable, generalizable data foundations are critical for enabling large-scale models in computational lithography. However, essential tasks-mask generation, rule violation detection, and layout optimization-are often handled in isolation, hindered by scarce datasets and limited modeling approaches. To address these challenges, we introduce Unitho, a unified multi-task large vision model built upon the Transformer architecture. Trained on a large-scale industrial lithography simulation dataset with hundreds of thousands of cases, Unitho supports end-to-end mask generation, lithography simulation, and rule violation detection. By enabling agile and high-fidelity lithography simulation, Unitho further facilitates the construction of robust data foundations for intelligent EDA. Experimental results validate its effectiveness and generalizability, with performance substantially surpassing academic baselines.

Problem

• Unified framework addresses isolated computational lithography tasks integration
• Solves data scarcity and limited modeling in lithography simulation
• Enables end-to-end mask generation and design rule verification

Innovation

• Unified multi-task Transformer model for lithography
• End-to-end mask generation and violation detection
• Large-scale industrial dataset enables high-fidelity simulation