myconfig.yml

Data:
  Input: {batch_size: 64, map: true, num_parallel_calls: 2, prefetch: true, prefetch_buffer_size: 1,
    shuffle: true, shuffle_buffer_size: 10000}
  Loading:
    example_type: single_tel
    merge_tel_types: true
    min_num_tels: 1
    seed: 1234
    selected_tel_types: ['MST:NectarCam']
    validation_split: 0.1
  Processing: {sorting: null}
  file_list: data_train.txt
  format: HDF5
Image Mapping:
  camera_types: [NectarCam]
  hex_conversion_algorithm: {ASTRICam: oversampling, CHEC: oversampling, DigiCam: oversampling,
    FlashCam: oversampling, LSTCam: oversampling, NectarCam: oversampling, SCTCam: oversampling,
    VERITAS: oversampling}
  interpolation_image_shape:
    NectarCam: [110, 110, 1]
  padding: {ASTRICam: 0, CHEC: 0, DigiCam: 1, FlashCam: 1, LSTCam: 2, NectarCam: 2,
    SCTCam: 4, VERITAS: 1}
Logging: {model_directory: ./run1}
Model:
  Model Parameters:
    basic:
      batchnorm_decay: 0.99
      conv_block:
        batchnorm: false
        bottleneck: null
        layers:
        - {filters: 117, kernel_size: 12}
        - {filters: 234, kernel_size: 10}
        - {filters: 468, kernel_size: 7}
        - {filters: 936, kernel_size: 6}
        max_pool: {size: 2, strides: 2}
    cnn_rnn:
      cnn_block: {function: conv_block, module: basic}
      dropout_rate: 0.5
      pretrained_weights: null
    single_tel:
      network: {function: conv_block, module: basic}
      pretrained_weights: null
  model: {function: single_tel_model, module: single_tel}
  model_directory: /home/jredondo/ctlearn/ctlearn/default_models
Prediction: {export_as_file: true, prediction_file_path: ./run1/predictions_run1.csv,
  true_labels_given: true}
TensorFlow: {run_TFDBG: false}
Training:
  Hyperparameters: {adam_epsilon: 1.0e-08, apply_class_weights: true, base_learning_rate: 5.0e-05,
    optimizer: Adam, scale_learning_rate: false, variables_to_train: null}
  num_training_steps_per_validation: 2500
  num_validations: 15