dvc.lock

schema: '2.0'
stages:
  data_ingestion:
    cmd: python src/speechRecognition/pipeline/stage_01_data_ingestion.py
    deps:
    - path: config/config.yaml
      hash: md5
      md5: 7a8b2238c6f3e6bad004f2be0ba3f054
      size: 474
    - path: src/speechRecognition/pipeline/stage_01_data_ingestion.py
      hash: md5
      md5: ccae3c8f6530e7ec02b8909d2bd255ac
      size: 858
    outs:
    - path: artifacts/data_ingestion/data.hf
      hash: md5
      md5: 5a384c93c7dee590923ffb0b239bdc3a.dir
      size: 1575323900
      nfiles: 7
  prepare_base_model:
    cmd: python src/speechRecognition/pipeline/stage_02_prepare_base_model.py
    deps:
    - path: config/config.yaml
      hash: md5
      md5: 7a8b2238c6f3e6bad004f2be0ba3f054
      size: 474
    - path: src/speechRecognition/pipeline/stage_02_prepare_base_model.py
      hash: md5
      md5: ab0298bd832f51724528616d023941e7
      size: 929
    params:
      params.yaml:
        SIZE: openai/whisper-base
    outs:
    - path: artifacts/prepare_base_model
      hash: md5
      md5: a39ac3f8362c7723dbd794c98a60479a.dir
      size: 580820018
      nfiles: 6
  training:
    cmd: python src/speechRecognition/pipeline/stage_03_model_trainer.py
    deps:
    - path: artifacts/data_ingestion/data.hf
      hash: md5
      md5: 5a384c93c7dee590923ffb0b239bdc3a.dir
      size: 1575323900
      nfiles: 7
    - path: artifacts/prepare_base_model
      hash: md5
      md5: a39ac3f8362c7723dbd794c98a60479a.dir
      size: 580820018
      nfiles: 6
    - path: config/config.yaml
      hash: md5
      md5: 7a8b2238c6f3e6bad004f2be0ba3f054
      size: 474
    - path: src/speechRecognition/pipeline/stage_03_model_trainer.py
      hash: md5
      md5: afe06f4698070aa76a1ff29a8f025f34
      size: 935
    params:
      params.yaml:
        DATA_SIZE: 10
        GENERATION_MAX_LENGTH: 225
        LANGUAGE: Russian
        LEARNING_RATE: 0,00001
        MAX_STEPS: 2
        METRIC: wer
        PUSH_TO_HUB: false
        SIZE: openai/whisper-base
        TASK: transcribe
    outs:
    - path: artifacts/training/model
      hash: md5
      md5: 018d9743b77ef9ba15f5a84aeb5ca712.dir
      size: 292319315
      nfiles: 10