Skip to content
Snippets Groups Projects
dvc.yaml 1.71 KiB
Newer Older
Marcin Wątroba's avatar
Marcin Wątroba committed
stages:

Marcin Wątroba's avatar
Marcin Wątroba committed
    luna_import_to_common_format:
Marcin Wątroba's avatar
Marcin Wątroba committed
        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
Marcin Wątroba's avatar
Marcin Wątroba committed

Marcin Wątroba's avatar
Marcin Wątroba committed
    luna_main_pipeline:
        cmd: |
Marcin Wątroba's avatar
Marcin Wątroba committed
            python -m spacy download pl_core_news_lg
Marcin Wątroba's avatar
Marcin Wątroba committed
            PYTHONPATH=. python experiment/luna/pipeline/luna_main.py
        deps:
Marcin Wątroba's avatar
Marcin Wątroba committed
            - experiment/luna/pipeline/luna_main.py
Marcin Wątroba's avatar
Marcin Wątroba committed
            - 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
Marcin Wątroba's avatar
Marcin Wątroba committed

    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