dataset_config.yaml

name: ocr
config_type: dataset
task: image2text
path: hezarai/parsynth-ocr-200k
text_split_type: char_split
id2label:
  0: ''
  1: آ
  2: ا
  3: ب
  4: پ
  5: ت
  6: ث
  7: ج
  8: چ
  9: ح
  10: خ
  11: د
  12: ذ
  13: ر
  14: ز
  15: ژ
  16: س
  17: ش
  18: ص
  19: ض
  20: ط
  21: ظ
  22: ع
  23: غ
  24: ف
  25: ق
  26: ک
  27: گ
  28: ل
  29: م
  30: ن
  31: و
  32: ه
  33: ی
  34: ' '
  35: '۱'
  36: '۲'
  37: '۳'
  38: '۴'
  39: '۵'
  40: '۶'
  41: '۷'
  42: '۸'
  43: '۹'
  44: '۰'
  45: ء
  46: ؤ
  47: ئ
  48: أ
  49: ّ
  50: /
  51: (
  52: )
  53: +
  54: '-'
  55: ':'
  56: ،
  57: '!'
  58: .
  59: ؛
  60: '='
  61: '%'
  62: ؟
  63: '1'
  64: '2'
  65: '3'
  66: '4'
  67: '5'
  68: '6'
  69: '7'
  70: '8'
  71: '9'
  72: '0'
text_column: text
max_length: 48
reverse_digits: true
image_processor_config:
  name: image_processor
  config_type: preprocessor
  mean:
  - 0.6595
  std:
  - 0.1501
  rescale: 0.00392156862745098
  resample: null
  size:
  - 384
  - 32
  mirror: true
  gray_scale: true