Mmidwinter1993update docs
a22a2ce5创建于 2025年10月17日历史提交

Package tokenizer

class AbstractBPETokenizer

func countToken

public func countToken(input: String): Int64
  • Description: Counts the number of tokens in a string.
  • Parameters:
    • input: String, The string to count tokens in.

func decode

public func decode(tokens: Array<UInt32>): String
  • Description: Decodes an array of token IDs into a string.
  • Parameters:
    • tokens: Array<UInt32>, The array of token IDs to decode.

func encode

public func encode(input: String): Array<UInt32>
  • Description: Encodes a string into an array of token IDs.
  • Parameters:
    • input: String, The string to encode.

class BPETokenizer

func init

init(modelPath: Path)
  • Description: Initializes the BPETokenizer with the specified model path.
  • Parameters:
    • modelPath: Path, The path to the directory containing the tokenizer configuration files.

struct BPETokenizerConfig

func deserialize

static func deserialize(dm: DataModel): BPETokenizerConfig
  • Description: Deserializes a DataModel into a BPETokenizerConfig object.
  • Parameters:
    • dm: DataModel, The DataModel to deserialize.

class Cl100kTokenizer

func init

init(tikTokenPath: Path)
  • Description: Initializes the Cl100kTokenizer with a path to the TikToken configuration file.
  • Parameters:
    • tikTokenPath: Path, The path to the TikToken configuration file.

func init

init(tikTokenContent: String)
  • Description: Initializes the Cl100kTokenizer with the content of the TikToken configuration.
  • Parameters:
    • tikTokenContent: String, The content of the TikToken configuration.

interface JsonDeserializable

func fromJson

static func fromJson(str: String)
  • Description: Creates an object from a JSON string.
  • Parameters:
    • str: String, The JSON string to deserialize.

func serialize

func serialize(): DataModel
  • Description: Serializes the object into a DataModel.

class Pair

func operator !=

public operator func !=(other: Pair<T>): Bool
  • Description: Checks if two pairs are not equal.
  • Parameters:
    • other: Pair<T>, The other pair to compare with.

func operator ==

public operator func ==(other: Pair<T>): Bool
  • Description: Checks if two pairs are equal.
  • Parameters:
    • other: Pair<T>, The other pair to compare with.

func hashCode

public func hashCode(): Int64
  • Description: Computes the hash code of the pair.

func init

public init(left: T, right: T)
  • Description: Initializes a Pair with left and right values.
  • Parameters:
    • left: T, The left value of the pair.
    • right: T, The right value of the pair.

prop left

public prop left: T
  • Description: Gets the left value of the pair.

prop right

public prop right: T
  • Description: Gets the right value of the pair.

struct TokenizerJson

func deserialize

static func deserialize(dm: DataModel): TokenizerJson
  • Description: Deserializes a DataModel into a TokenizerJson object.
  • Parameters:
    • dm: DataModel, The DataModel to deserialize.

class TokenizerLoader

func default

func default(): Tokenizer
  • Description: Returns the default tokenizer. If no default tokenizer is configured, returns a UnicodeTokenizer.

func load

func load(tokenizerType: TokenizerType, vocabPath: String): Tokenizer
  • Description: Loads a tokenizer based on the specified tokenizer type and vocabulary path.
  • Parameters:
    • tokenizerType: TokenizerType, The type of tokenizer to load.
    • vocabPath: String, The path to the vocabulary file for the tokenizer.

enum TokenizerType

Cl100k

Cl100k
  • Description: Represents the Cl100k tokenizer type.

DeepSeek

DeepSeek
  • Description: Represents the DeepSeek tokenizer type.

Qwen

Qwen
  • Description: Represents the Qwen tokenizer type.

func fromString

func fromString(ttype: String): TokenizerType
  • Description: Converts a string to the corresponding TokenizerType enum value.
  • Parameters:
    • ttype: String, The string representation of the tokenizer type.

func toString

func toString()
  • Description: Converts the tokenizer type to a string representation.

class UnicodeTokenizer

func countToken

func countToken(input: String): Int64
  • Description: Counts the number of tokens in a string, adjusting for Chinese and English characters.
  • Parameters:
    • input: String, The string whose tokens are to be counted.

func decode

func decode(tokens: Array<UInt32>): String
  • Description: Decodes an array of tokens into a string.
  • Parameters:
    • tokens: Array<UInt32>, An array of tokens to be decoded.

func encode

func encode(input: String): Array<UInt32>
  • Description: Encodes a string into an array of tokens.
  • Parameters:
    • input: String, The string to be encoded.