brillmoore_kartaslov_ru_nolm.json

{
  "dataset_reader": {
    "class_name": "typos_kartaslov_reader",
    "data_path": "{DOWNLOADS_PATH}"
  },
  "dataset_iterator": {
    "class_name": "typos_iterator",
    "test_ratio": 0.02
  },
  "chainer":{
    "in": ["x"],
    "in_y": ["y"],
    "pipe": [
      {
        "class_name": "str_lower",
        "id": "lower",
        "in": ["x"],
        "out": ["x_lower"]
      },
      {
        "class_name": "nltk_moses_tokenizer",
        "id": "tokenizer",
        "in": ["x_lower"],
        "out": ["x_tokens"]
      },
      {
        "ref": "tokenizer",
        "in": ["y"],
        "out": ["y_tokens"]
      },
      {
        "fit_on": ["x_tokens", "y_tokens"],
        "in": ["x_tokens"],
        "out": ["tokens_candidates"],
        "class_name": "spelling_error_model",
        "window": 1,
        "candidates_count": 1,
        "dictionary": {
          "class_name": "russian_words_vocab",
          "data_dir": "{DOWNLOADS_PATH}/vocabs"
        },
        "save_path": "{MODELS_PATH}/error_model/error_model_ru.tsv",
        "load_path": "{MODELS_PATH}/error_model/error_model_ru.tsv"
      },
      {
        "class_name": "deeppavlov.models.spelling_correction.electors.top1_elector:TopOneElector",
        "in": ["tokens_candidates"],
        "out": ["y_predicted_tokens"]
      },
      {
        "ref": "tokenizer",
        "in": ["y_predicted_tokens"],
        "out": ["y_predicted"]
      }
    ],
    "out": ["y_predicted"]
  },
  "train": {
    "evaluation_targets": [
      "test"
    ],
    "class_name": "fit_trainer"
  },
  "metadata": {
    "variables": {
      "ROOT_PATH": "~/.deeppavlov",
      "DOWNLOADS_PATH": "{ROOT_PATH}/downloads",
      "MODELS_PATH": "{ROOT_PATH}/models"
    },
    "requirements": [
      "{DEEPPAVLOV_PATH}/requirements/spelling.txt"
    ],
    "download": [
      {
        "url": "http://files.deeppavlov.ai/deeppavlov_data/error_model.tar.gz",
        "subdir": "{MODELS_PATH}"
      }
    ]
  }
}