KetGPT

KetGPT是由荷兰代尔夫特理工大学的QuTech量子机器学习组开发的数据集，旨在通过Transformer模型生成类似真实量子算法的电路，以增强现有量子电路数据集。该数据集包含713个QASM文件，这些文件是通过对现有量子算法电路进行分析后生成的，具有算法基础的电路结构。KetGPT的创建过程涉及对QASM文件的预处理和定制的tokenization方法，确保生成的电路在结构上与真实算法电路相似。该数据集主要用于量子系统的基准测试和AI驱动的量子编译器及系统的开发，旨在解决量子电路数据集的规模和多样性不足的问题。

KetGPT is a dataset developed by the QuTech Quantum Machine Learning Group at Delft University of Technology, Netherlands. It is designed to enhance existing quantum circuit datasets by generating circuits that resemble real quantum algorithms via Transformer models. This dataset contains 713 QASM files, which are generated through analysis of existing quantum algorithm circuits and feature circuit structures grounded in real quantum algorithms. The development of KetGPT incorporates preprocessing of QASM files and a customized tokenization approach, ensuring that the generated circuits exhibit structural similarity to real quantum algorithm circuits. This dataset is primarily utilized for benchmarking quantum systems and developing AI-driven quantum compilers and related systems, with the goal of addressing the gaps in scale and diversity of existing quantum circuit datasets.