stages: luna_import_to_common_format: cmd: PYTHONPATH=. python experiment/luna/import_dataset/import_luna.py deps: - experiment/luna/import_dataset/import_luna.py - experiment_data/dataset/LUNA.PL outs: - experiment_data/dataset_relation_manager_data/luna luna_main_pipeline: cmd: | python -m spacy download pl_core_news_lg PYTHONPATH=. python experiment/luna/pipeline/luna_main.py deps: - experiment/luna/pipeline/luna_main.py - experiment_data/dataset_relation_manager_data/luna - experiment_data/dataset/LUNA.PL - experiment_data/cached_asr/luna_techmo outs: - experiment_data/pipeline/asr_benchmark_luna voicelab_import_to_common_format: cmd: PYTHONPATH=. python experiment/voicelab/import_data.py deps: - experiment/voicelab/import_data.py - experiment_data/dataset/voicelab_cbiz_testset_20220322 outs: - experiment_data/dataset_relation_manager_data/voicelab_cbiz_testset_20220322 voicelab_main_pipeline: cmd: | python -m spacy download pl_core_news_lg PYTHONPATH=. python experiment/voicelab/voicelab_pipeline.py deps: - experiment/voicelab/voicelab_pipeline.py - experiment_data/dataset_relation_manager_data/voicelab_cbiz_testset_20220322 - experiment_data/dataset/voicelab_cbiz_testset_20220322 - experiment_data/cached_asr/voicelab_cbiz_testset_20220322_techmo outs: - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322 # concurrent features, multiprocessing