muthuk1/alwas-analog-layout-dataset

ALWAS模拟布局数据集是一个用于训练ALWAS（模拟布局工作流自动化系统）管道中机器学习模型的合成数据集。数据集包含4,000个模拟IC布局块，具有完整的元数据、阶段转换和标签，用于小时估计、复杂性分类、瓶颈风险预测和完成时间预测。数据集结构包括表格数据和SFT数据，表格数据包含多个列，如block_id、block_type、tech_node等，SFT数据包含2,000个对话样本。数据集统计信息包括总块数、块类型、技术节点、小时范围、复杂性和瓶颈风险分布。

The ALWAS Analog Layout Dataset is a synthetic dataset for training ML models in the ALWAS (Analog Layout Workflow Automation System) pipeline. It contains 4,000 analog IC layout blocks with complete metadata, stage transitions, and labels for hours estimation, complexity classification, bottleneck risk prediction, and completion time prediction. The dataset structure includes tabular data and SFT data, with the tabular data containing columns such as block_id, block_type, tech_node, etc., and the SFT data containing 2,000 conversational samples. The dataset statistics include total blocks, block types, tech nodes, hours range, complexity, and bottleneck risk distribution.