Struct ByteLevel 

pub struct ByteLevel { /* private fields */ }

A compiled ByteLevel pre-tokenizer.

Optionally splits input using the GPT-2 regex, then maps every byte of the resulting UTF-8 text to a visible Unicode character.

Constructed once from config fields (typically via PreTokenizerConfig::ByteLevel), then reused across many inputs.

Implementations

impl ByteLevel

pub fn from_config( add_prefix_space: bool, trim_offsets: bool, use_regex: bool, ) -> Result<Self, Error>

Build a ByteLevel from config fields.

pub fn is_bulk_only(&self) -> bool

Returns true when this instance does no regex splitting or prefix-space insertion — only bulk byte encoding. In that mode the encoding can be fused into the BPE cache lookup to skip it on warm runs.

pub fn pre_tokenize(&self, pts: &mut PreTokenizedString) -> Result<(), Error>

Pre-tokenize in place using byte-level encoding.

For each text split in pts: optionally prepend a space, split by the GPT-2 regex, and byte-encode each piece. Splits with a pre-assigned token ID are byte-encoded but otherwise passed through.

Trait Implementations

impl Clone for ByteLevel

fn clone(&self) -> ByteLevel

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ByteLevel

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for ByteLevel

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.