ModelInput

class pytext.models.qna.bert_squad_qa.ModelInput

Bases: ModelInput

All Attributes (including base classes)

squad_input: Union[SquadForBERTTensorizer.Config, SquadForRoBERTaTensorizer.Config] = SquadForBERTTensorizer.Config()

has_answer: LabelTensorizer.Config = LabelTensorizer.Config(column='has_answer')

Default JSON

{
    "squad_input": {
        "SquadForBERTTensorizer": {
            "is_input": true,
            "columns": [
                "question",
                "doc"
            ],
            "tokenizer": {
                "WordPieceTokenizer": {
                    "basic_tokenizer": {
                        "split_regex": "\\s+",
                        "lowercase": true
                    },
                    "wordpiece_vocab_path": "/mnt/vol/nlp_technologies/bert/uncased_L-12_H-768_A-12/vocab.txt"
                }
            },
            "base_tokenizer": null,
            "vocab_file": "/mnt/vol/nlp_technologies/bert/uncased_L-12_H-768_A-12/vocab.txt",
            "max_seq_len": 256,
            "answers_column": "answers",
            "answer_starts_column": "answer_starts"
        }
    },
    "has_answer": {
        "LabelTensorizer": {
            "is_input": false,
            "column": "has_answer",
            "allow_unknown": false,
            "pad_in_vocab": false,
            "label_vocab": null
        }
    }
}