OpenLS-D
收藏资源简介:
OpenLS-D是由中国科学院计算技术研究所等机构创建的逻辑综合数据集,旨在支持机器学习在逻辑综合过程中的应用。该数据集包含46个组合设计,总计超过966,000个布尔电路,每个设计包含21,000个电路,由1,000个综合配方生成。数据集创建过程包括布尔表示、逻辑优化和技术映射三个基本步骤,并支持半定制化,允许研究人员添加步骤和逐步细化生成的数据集。OpenLS-D的应用领域广泛,包括电路分类、电路排名、质量结果预测和概率预测等,旨在解决逻辑综合中的多样化问题。
OpenLS-D is a logic synthesis dataset developed by institutions including the Institute of Computing Technology, Chinese Academy of Sciences, aiming to support the application of machine learning in logic synthesis processes. This dataset includes 46 combinational designs, with a total of over 966,000 Boolean circuits. Each design comprises 21,000 circuits generated via 1,000 synthesis recipes. The creation process of the dataset involves three core steps: Boolean representation, logic optimization, and technology mapping. It also supports semi-customization, allowing researchers to add custom steps and iteratively refine the generated dataset. OpenLS-D has a wide range of application scenarios, including circuit classification, circuit ranking, quality-of-result prediction, and probabilistic prediction, aiming to address diverse challenges in logic synthesis.
- 1An Adaptive Open-Source Dataset Generation Framework for Machine Learning Tasks in Logic Synthesis中国科学院计算技术研究所 · 2024年



