check_templates.py

import argparse, os, json, random
from collections import defaultdict


parser = argparse.ArgumentParser()
parser.add_argument('--questions_file',
    default='../output/CLEVR_v1.0/questions/CLEVR_val_questions.json')
parser.add_argument('--template_dir', default='CLEVR_1.0_templates/')
parser.add_argument('--num_samples', type=int, default=5)


TEMPLATE_ORDER = [
  'compare_integer.json',
  'comparison.json',
  'three_hop.json',
  'single_and.json',
  'same_relate.json',
  'single_or.json',
  'one_hop.json',
  'two_hop.json',
  'zero_hop.json',
]


def main(args):
  # Load templates
  num_loaded_templates = 0
  templates = {}
  family_index_to_template = {}
  for fn in TEMPLATE_ORDER:
    with open(os.path.join(args.template_dir, fn), 'r') as f:
      base = os.path.splitext(fn)[0]
      for i, template in enumerate(json.load(f)):
        family_index_to_template[num_loaded_templates] = template
        num_loaded_templates += 1
        key = (fn, i)
        templates[key] = template
  print('Read %d templates from disk' % num_loaded_templates)

  # Load questions
  with open(args.questions_file, 'r') as f:
    questions = json.load(f)['questions']
  print('Read %d questions from disk' % len(questions))

  # Group questions by family index
  family_index_to_questions = defaultdict(list)
  for q in questions:
    family_index = q['question_family_index']
    family_index_to_questions[family_index].append(q)
  
  # Print out randomly selected questions and their recovered templates
  for _ in range(args.num_samples):
    q = random.choice(questions)
    qfi = q['question_family_index']
    template = family_index_to_template[qfi]
    print(q['question'])
    for t in template['text']:
      print(t)
    print()


if __name__ == '__main__':
  main(parser.parse_args())