IBM/pytorchpipe

# Load config defining tasks for training, validation and testing.
default_configs:
  vqa_med_2019/frozen_pipelines/frozen_input_fusion_glove_lstm_vgg_att_is_cat.yml,
  vqa_med_2019/frozen_pipelines/frozen_question_categorization_glove_rnn_ffn.yml


training_validation:
  task:
    type: &p_type VQAMED2019
    data_folder: &data_folder ~/data/vqa-med
    split: training_validation
    categories: all
    resize_image: &resize_image [224, 224]
    batch_size: 64
    # Appy all preprocessing/data augmentations.
    question_preprocessing: lowercase,remove_punctuation,tokenize
    streams: 
      questions: tokenized_questions

  dataloader:
    # No sampler, process samples in the same order.
    shuffle: false
    # Use 1 worker, so batches will follow the samples order.
    num_workers: 1


hyperparams:
  # Final classifier: FFN.
  answer_classifier_hidden_sizes_val: &answer_classifier_hidden_sizes_val [500]


# Add component for exporting answers to files.
pipeline:

  ################# PIPE 6: C1 + C2 + C3 questions #################

  # Answer encoding.
  pipe6_c123_binary_yn_answer_indexer:
    priority: 6.2
    type: LabelIndexer
    data_folder: ~/data/vqa-med
    word_mappings_file: answers.c1_c2_c3_binary_yn.word.mappings.csv
    # Export mappings and size to globals.
    export_word_mappings_to_globals: True
    streams:
      inputs: answers
      outputs: answers_ids
    globals:
      vocabulary_size: vocabulary_size_c123_binary_yn
      word_mappings: word_mappings_c123_binary_yn


  # Viewers.
  viewer_extended:
    priority: 100.4
    type: StreamViewer
    sample_number: 0
    input_streams: 
      indices,image_ids,tokenized_questions,
      concatenated_activations_size,
      category,names,
      answers

  fused_inputs_exporter:
    priority: 100.5
    type: StreamFileExporter
    separator: '|'
    filename: 'fused_inputs.csv'
    export_separator_line_to_csv: True
    input_streams: 
      indices #, concatenated_activations