WordSeqEmbedding.Config¶

Component: WordSeqEmbedding

class WordSeqEmbedding.Config[source]: Bases: EmbeddingBase.Config

All Attributes (including base classes)

load_path: Optional[str] = None

save_path: Optional[str] = None

freeze: bool = False

shared_module_key: Optional[str] = None

word_embed_dim: int = 100

embedding_init_strategy: EmbedInitStrategy = <EmbedInitStrategy.RANDOM: 'random'>

embedding_init_range: Optional[list[float]] = None

embeddding_init_std: Optional[float] = 0.02

padding_idx: Optional[int] = None

lstm: BiLSTM.Config = BiLSTM.Config()

pretrained_embeddings_path: str = ''

vocab_size: int = 0

If pretrained_embeddings_path and vocab_from_pretrained_embeddings are set, only the first vocab_size tokens in the file will be added to the vocab.

lowercase_tokens: bool = True

skip_header: bool = True

delimiter: str = ' '

Default JSON

{
    "load_path": null,
    "save_path": null,
    "freeze": false,
    "shared_module_key": null,
    "word_embed_dim": 100,
    "embedding_init_strategy": "random",
    "embedding_init_range": null,
    "embeddding_init_std": 0.02,
    "padding_idx": null,
    "lstm": {
        "load_path": null,
        "save_path": null,
        "freeze": false,
        "shared_module_key": null,
        "dropout": 0.4,
        "lstm_dim": 32,
        "num_layers": 1,
        "bidirectional": true,
        "pack_sequence": true,
        "disable_sort_in_jit": false
    },
    "pretrained_embeddings_path": "",
    "vocab_size": 0,
    "lowercase_tokens": true,
    "skip_header": true,
    "delimiter": " "
}