stages: actions_extraction: cmd: python3 -m src.pipelines.actions_based.stage1_extraction deps: - data - src/pipelines/actions_based/stage1_extraction.py params: - actions.extraction.num_partitions outs: - generated/actions/stage1_extraction actions_tokenization: cmd: python3 -m src.pipelines.actions_based.stage2_tokenization deps: - generated/actions/stage1_extraction - src/pipelines/actions_based/stage2_tokenization.py params: - actions.tokenization.max_tokens - actions.tokenization.min_tokens - global.base_model outs: - generated/actions/stage2_tokenization actions_exploding: cmd: python3 -m src.pipelines.actions_based.stage3_exploding deps: - generated/actions/stage2_tokenization - src/pipelines/actions_based/stage3_exploding.py outs: - generated/actions/stage3_exploding actions_reindexing: cmd: python3 -m src.pipelines.actions_based.stage4_reindexing deps: - generated/actions/stage3_exploding - src/pipelines/actions_based/stage4_reindexing.py outs: - generated/actions/stage4_reindexing actions_stats: cmd: python3 -m src.pipelines.actions_based.stage5_stats deps: - generated/actions/stage4_reindexing - src/pipelines/actions_based/stage5_stats.py outs: - generated/actions/stage5_stats actions_training: cmd: python3 -m src.pipelines.actions_based.train deps: - generated/actions/stage4_reindexing - generated/actions/stage5_stats - src/pipelines/actions_based/train.py params: - global.base_model - actions.training.max_training_time - actions.training.learning_rate - actions.training.num_epochs - actions.training.batch_size - actions.training.save_step outs: - checkpoints/actions translations_extraction: cmd: python3 -m src.pipelines.translation_based.stage1_extraction deps: - data params: - translations.extraction.num_partitions outs: - generated/translations/stage1_extraction translations_create_batches: cmd: python3 -m src.pipelines.translation_based.stage2_create_batches deps: - generated/translations/stage1_extraction params: - global.base_model outs: - generated/translations/stage2_create_batches translations_exploding: cmd: python3 -m src.pipelines.translation_based.stage3_exploding deps: - generated/translations/stage2_create_batches outs: - generated/translations/stage3_exploding translations_reindexing: cmd: python3 -m src.pipelines.translation_based.stage4_reindexing deps: - generated/translations/stage3_exploding outs: - generated/translations/stage4_reindexing translations_training: cmd: python3 -m src.pipelines.translation_based.train deps: - generated/translations/stage4_reindexing - src/pipelines/translation_based/train.py params: - global.base_model - translations.training.max_training_time - translations.training.learning_rate - translations.training.num_epochs - translations.training.batch_size - translations.training.save_step outs: - checkpoints/translations