Unitho: A Unified Multi-Task Framework for Computational Lithography

TL;DR

Unitho is a multi-task Transformer-based model that unifies mask generation, lithography simulation, and rule violation detection, significantly improving performance and generalizability in computational lithography tasks.

Contribution

The paper introduces Unitho, a novel unified multi-task framework built on Transformer architecture, trained on large-scale industrial data for comprehensive lithography tasks.

Findings

Outperforms academic baselines in lithography tasks

Supports end-to-end mask generation and rule violation detection

Demonstrates high generalizability and effectiveness

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.