schema: '2.0'
stages:
  train:
    cmd: poetry run train
    deps:
    - path: data/raw/arxivData.json
      md5: a314e2f4eab544a46e6f95802ecde647
      size: 72422946
    - path: shad_mlops_transformers/model.py
      md5: 465d64e91ac7d713e57442c932a665ea
      size: 2889
    - path: shad_mlops_transformers/trainer.py
      md5: 39d02f490890ef306b1165fcc29b8895
      size: 7495
    params:
      shad_mlops_transformers/config.py:
        Config.batch_size: 64
        Config.random_seed: 42
        Config.test_size: 0.2
    outs:
    - path: data/models/model.torch
      md5: 3a56baed312933280d01e30f86bd8b57
      size: 440542446
    - path: data/plots/train.png
      md5: 87c500a795ca97c14b84a6c0367d7b78
      size: 44168
    - path: data/raw/mapper.json
      md5: e4c22adf587676d22ed58c812f41f409
      size: 855