maxpooling190/LithoSim

--- configs: - config_name: metal data_files: - split: train path: lithosim_path/metal_train_val.csv - split: test path: lithosim_path/metal_test.csv - config_name: opc_metal data_files: - split: train path: lithosim_path/opc_metal_train_val.csv - split: test path: lithosim_path/opc_metal_test.csv - config_name: via data_files: - split: train path: lithosim_path/via_train_val.csv - split: test path: lithosim_path/via_test.csv - config_name: opc_via data_files: - split: train path: lithosim_path/opc_via_train_val.csv - split: test path: lithosim_path/opc_via_test.csv - config_name: ood data_files: - split: test path: lithosim_path/ood.csv license: afl-3.0 language: - en tags: - code pretty_name: litho size_categories: - 1B<n<10B task_categories: - feature-extraction --- The benchmark of "LithoSim: A Large, Holistic Lithography Simulation Benchmark for AI-Driven Semiconductor Manufacturing" The corresponding GitHub repo can be found at https://dw-hongquan.github.io/LithoSim/ ## Data Construction - 4 in-distributed dataset (OPC_Metal/Metal/OPC_Via/Via). - 1 out-of-distribution (OOD) dataset. - Each main dataset has a train_val and a test folder with compressed data file. - Each set of data contains a source_simple.src description of the source, a layout.png, a mask.png, and a number of RI under different process variations.