Tokenizer splits strings of text into tokens. Typically, this either splits text into
word tokens or character tokens, and those are the two tokenizer subclasses we have implemented
here, though you could imagine wanting to do other kinds of tokenization for structured or
See the parameters to, e.g.,
.SpacyTokenizer, or whichever tokenizer
you want to use.
If the base input to your model is words, you should use a
.SpacyTokenizer, even if
you also want to have a character-level encoder to get an additional vector for each word
token. Splitting word tokens into character arrays is handled separately, in the
class Tokenizer(Registrable): | ... | default_implementation = "spacy"
class Tokenizer(Registrable): | ... | def batch_tokenize(self, texts: List[str]) -> List[List[Token]]
Batches together tokenization of several texts, in case that is faster for particular tokenizers.
By default we just do this without batching. Override this in your tokenizer if you have a good way of doing batched computation.
class Tokenizer(Registrable): | ... | def tokenize(self, text: str) -> List[Token]
Actually implements splitting words into tokens.
- tokens :
class Tokenizer(Registrable): | ... | def add_special_tokens( | self, | tokens1: List[Token], | tokens2: Optional[List[Token]] = None | ) -> List[Token]
Adds special tokens to tokenized text. These are tokens like [CLS] or [SEP].
Not all tokenizers do this. The default is to just return the tokens unchanged.
- tokens1 :
The list of tokens to add special tokens to.
- tokens2 :
An optional second list of tokens. This will be concatenated with
tokens1. Special tokens will be added as appropriate.
- tokens :
The combined list of tokens, with special tokens added.
class Tokenizer(Registrable): | ... | def num_special_tokens_for_sequence(self) -> int
Returns the number of special tokens added for a single sequence.
class Tokenizer(Registrable): | ... | def num_special_tokens_for_pair(self) -> int
Returns the number of special tokens added for a pair of sequences.