Graficos/dvc.yaml

# TODO get rid of all model specific names like Ar85K here!
stages:
  create_database:
    cmd: python stages/01_create_database.py
    deps:
      - stages/01_create_database.py
    params:
      - system.data
    outs:
      - system/database.db
      - system/yaml/database.yaml

  build_database_groups:
    cmd: python stages/02_build_database_groups.py
    deps:
      - stages/02_build_database_groups.py
      - system/database.db
    params:
      - system.datasets
    outs:
      - system/yaml/database_groups.yaml

  test_train_validate_split:
    cmd: python stages/03_Model_TTV.py
    deps:
      - stages/03_Model_TTV.py
      - system/yaml/database_groups.yaml
    params:
      - data.split
    outs:
      - system/ACSF/test_train_validate.yaml

  build_descriptors:
    cmd: python stages/04_build_descriptors.py
    deps:
      - stages/04_build_descriptors.py
      - system/database.db
      - system/ACSF/test_train_validate.yaml
    params:
      - general.train_on_forces
      - descriptors
    outs:
      - cache/
    metrics:
      - descriptor_metrics.yaml:
          cache: false

  fit_model:
    cmd: python stages/05_fit_model.py
    deps:
      - stages/05_fit_model.py
      - system/ACSF/test_train_validate.yaml
      - cache/
    outs:
      - history_metrics.csv:
          cache: false
      - model/
    params:
      - training
    metrics:
      - metrics_dvc.yaml:
          cache: false
    plots:
      - history_dvc.csv:
          cache: false

  evaluate_model:
    cmd: python stages/06_evaluate_model.py
    deps:
      - stages/06_evaluate_model.py
      - system/ACSF/test_train_validate.yaml
      - cache/
      - model/
    outs:
      - img/energies.png:
          cache: false
      - img/forces_x.png:
          cache: false
      - img/forces_y.png:
          cache: false
      - img/forces_z.png:
          cache: false
      - img/energies.npz
      - img/forces.npz
    params:
      - evaluation
#    metrics:
#      - metrics_dvc.yaml:
#          cache: false
#    plots:
#      - history_dvc.csv:
#          cache: false

# evaluate using the saved weights, create the plots, and so on..
# md_test run with the ase object..


# ISSUE
# DVC does not create a .gitignore if the stage fails but if rerun it assumes the .gitignore is there - maybe use force?