InputConfig

class pytext.models.roberta.InputConfig

Bases: ConfigBase

All Attributes (including base classes)

tokens: RoBERTaTensorizer.Config = RoBERTaTensorizer.Config()

labels: LabelTensorizer.Config = LabelTensorizer.Config()

Default JSON

{
    "tokens": {
        "is_input": true,
        "columns": [
            "text"
        ],
        "tokenizer": {
            "GPT2BPETokenizer": {
                "bpe_encoder_path": "manifold://pytext_training/tree/static/vocabs/bpe/gpt2/encoder.json",
                "bpe_vocab_path": "manifold://pytext_training/tree/static/vocabs/bpe/gpt2/vocab.bpe"
            }
        },
        "base_tokenizer": null,
        "vocab_file": "manifold://pytext_training/tree/static/vocabs/bpe/gpt2/dict.txt",
        "max_seq_len": 256
    },
    "labels": {
        "LabelTensorizer": {
            "is_input": false,
            "column": "label",
            "allow_unknown": false,
            "pad_in_vocab": false,
            "label_vocab": null
        }
    }
}