Select from an existing dataset
To select from one of the existing datasets and tasks, one may configure the selected dataset in general configs, an example could be:
task:
name: 'HLS'
type: cdfg # type: cdfg or dfg
target: dsp # different target prediction
processed_folder: '~/DGRL_Hardware/data_processed/'
divide_seed: default #set as default or a seed
raw_data_path: '~/DGRL_Hardware/data_raw/HLS/'
data_processor: HLSDataProcessor
here the name gives the name of the dataset, type and target determines the task, the processed_folder defines the path to save the processed PyG format data, raw_data_path provides the path of the original data, the data_processor defines the name of the data processor to process the data.
The data processor for the existing datasets are implemented in ./data_processor/. For more details on how to customize the dataset including the data processor, please refer to customize new datasets.