Public Member Functions
static int	ailiaTokenizerCreate (ref IntPtr net, int type, int flags)
	Creates a tokenizer instance. More...

static int	ailiaTokenizerOpenModelFile (IntPtr net, string ath)
	Open model file. More...

static int	ailiaTokenizerOpenDictionaryFile (IntPtr net, string path)
	Open dictionary file. More...

static int	ailiaTokenizerOpenVocabFile (IntPtr net, string path)
	Open vocab file. More...

static int	ailiaTokenizerOpenMergeFile (IntPtr net, string path)
	Open merges file. More...

static int	ailiaTokenizerOpenAddedTokensFile (IntPtr net, string path)
	Open added tokens file. More...

static int	ailiaTokenizerOpenTokenizerConfigFile (IntPtr net, string path)
	Open tokenizer config file. More...

static int	ailiaTokenizerEncode (IntPtr net, IntPtr utf8)
	Perform encode. More...

static int	ailiaTokenizerEncodeWithSpecialTokens (IntPtr net, IntPtr utf8)
	Perform encode with special tokens. More...

static int	ailiaTokenizerGetTokenCount (IntPtr net, ref uint count)
	Gets the number of tokens. More...

static int	ailiaTokenizerGetTokens (IntPtr net, IntPtr tokens, uint count)
	Gets the tokens. More...

static int	ailiaTokenizerGetWordIDs (IntPtr net, IntPtr tokens, uint count)
	Gets the word ID. More...

static int	ailiaTokenizerGetCharStarts (IntPtr net, IntPtr tokens, uint count)
	Gets the Char start positions. More...

static int	ailiaTokenizerGetCharEnds (IntPtr net, IntPtr tokens, uint count)
	Gets the Char end positions. More...

static int	ailiaTokenizerDecode (IntPtr net, IntPtr tokens, uint token_count)
	Perform encode. More...

static int	ailiaTokenizerDecodeWithSpecialTokens (IntPtr net, IntPtr tokens, uint token_count)
	Perform decode with special tokens. More...

static int	ailiaTokenizerGetTextLength (IntPtr net, ref uint len)
	Gets the size of text. (Include null) More...

static int	ailiaTokenizerGetText (IntPtr net, IntPtr text, uint len)
	Gets the decoded text. More...

static int	ailiaTokenizerGetVocabSize (IntPtr net, ref uint size)
	Gets the size of vocab. (Include null) More...

static int	ailiaTokenizerGetVocab (IntPtr net, int token, ref IntPtr vocab)
	Perform encode. More...

static int	ailiaTokenizerAddSpecialTokens (IntPtr net, IntPtr tokens, uint count)
	Add SpecialToken. More...

static void	ailiaTokenizerDestroy (IntPtr net)
	It destroys the tokenizer instance. More...

static int	ailiaTokenizerUtf8ToUtf32 (ref uint utf32, ref uint processed_byte, IntPtr utf8, uint utf8_len)
	Convert UTF8 character to UTF32 character. More...

static int	ailiaTokenizerUtf32ToUtf8 (IntPtr utf8, ref uint processed_byte, uint utf32)
	Convert UTF32 character to UTF8 character. More...

Static Public Attributes
const String	LIBRARY_NAME ="ailia_tokenizer"

const Int32	AILIA_TOKENIZER_TYPE_WHISPER = (0)

const Int32	AILIA_TOKENIZER_TYPE_CLIP = (1)

const Int32	AILIA_TOKENIZER_TYPE_XLM_ROBERTA = (2)

const Int32	AILIA_TOKENIZER_TYPE_MARIAN = (3)

const Int32	AILIA_TOKENIZER_TYPE_BERT_JAPANESE_WORDPIECE = (4)

const Int32	AILIA_TOKENIZER_TYPE_BERT_JAPANESE_CHARACTER = (5)

const Int32	AILIA_TOKENIZER_TYPE_T5 = (6)

const Int32	AILIA_TOKENIZER_TYPE_ROBERTA = (7)

const Int32	AILIA_TOKENIZER_TYPE_BERT = (8)

const Int32	AILIA_TOKENIZER_TYPE_GPT2 = (9)

const Int32	AILIA_TOKENIZER_TYPE_LLAMA = (10)

const Int32	AILIA_TOKENIZER_FLAG_NONE = (0)

const Int32	AILIA_TOKENIZER_FLAG_UTF8_SAFE = (1)

Member Function Documentation

◆ ailiaTokenizerAddSpecialTokens()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerAddSpecialTokens	(	IntPtr	net,
		IntPtr	tokens,
		uint	count
	)

Add SpecialToken.

Parameters

net	A tokenizer instance pointer
tokens	Token(UTF8)
count	The number of tokens

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

This is valid only for AILIA_TOKENIZER_TYPE_ROBERTA and AILIA_TOKENIZER_TYPE_GPT2.

◆ ailiaTokenizerCreate()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerCreate	(	ref IntPtr	net,
		int	type,
		int	flags
	)

Creates a tokenizer instance.

Parameters

net	A pointer to the tokenizer instance pointer
type	AILIA_TOKENIZER_TYPE_*
flag	OR of AILIA_TOKENIZER_FLAG_*

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

Creates a tokenizer instance.

◆ ailiaTokenizerDecode()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerDecode	(	IntPtr	net,
		IntPtr	tokens,
		uint	token_count
	)

Perform encode.

Parameters

net	A tokenizer instance pointer
tokens	Tokens for decode
token_count	The number of tokens

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

Get the decoded result with ailiaTokenizerGetText API.

◆ ailiaTokenizerDecodeWithSpecialTokens()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerDecodeWithSpecialTokens	(	IntPtr	net,
		IntPtr	tokens,
		uint	token_count
	)

Perform decode with special tokens.

Parameters

net	A tokenizer instance pointer
tokens	Tokens for decode
token_count	The number of tokens

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

Get the decoded result with ailiaTokenizerGetText API. Similarly to skip_special_tokens=False, special tokens will be output.

◆ ailiaTokenizerDestroy()

static void ailiaTokenizer.AiliaTokenizer.ailiaTokenizerDestroy ( IntPtr net )

It destroys the tokenizer instance.

Parameters

net	A tokenizer instance pointer

◆ ailiaTokenizerEncode()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerEncode	(	IntPtr	net,
		IntPtr	utf8
	)

Perform encode.

Parameters

net	A tokenizer instance pointer
text	Text for encode (UTF8)

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

Get the encoded result with ailiaTokenizerGetTokens API.

◆ ailiaTokenizerEncodeWithSpecialTokens()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerEncodeWithSpecialTokens	(	IntPtr	net,
		IntPtr	utf8
	)

Perform encode with special tokens.

Parameters

net	A tokenizer instance pointer
text	Text for encode (UTF8)

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

Get the encoded result with ailiaTokenizerGetTokens API. Similarly to split_special_tokens=False, special tokens will be output.

◆ ailiaTokenizerGetCharEnds()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerGetCharEnds	(	IntPtr	net,
		IntPtr	tokens,
		uint	count
	)

Gets the Char end positions.

Parameters

net	A tokenizer instance pointer
char_ends	Char end position
count	Token count

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

If ailiaTokenizerEncode() is not run at all, the function returns AILIA_STATUS_INVALID_STATE . This is valid only for AILIA_TOKENIZER_TYPE_ROBERTA and AILIA_TOKENIZER_TYPE_BERT. The character end positions in UTF-32 units corresponding to each token are returned.

◆ ailiaTokenizerGetCharStarts()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerGetCharStarts	(	IntPtr	net,
		IntPtr	tokens,
		uint	count
	)

Gets the Char start positions.

Parameters

net	A tokenizer instance pointer
char_starts	Character start position
count	Token count

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

If ailiaTokenizerEncode() is not run at all, the function returns AILIA_STATUS_INVALID_STATE . This is valid only for AILIA_TOKENIZER_TYPE_ROBERTA and AILIA_TOKENIZER_TYPE_BERT. The character start positions in UTF-32 units corresponding to each token are returned.

◆ ailiaTokenizerGetText()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerGetText	(	IntPtr	net,
		IntPtr	text,
		uint	len
	)

Gets the decoded text.

Parameters

net	A tokenizer instance pointer
text	Text(UTF8)
len	Buffer size

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

If ailiaTokenizerDecode() is not run at all, the function returns AILIA_STATUS_INVALID_STATE .

◆ ailiaTokenizerGetTextLength()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerGetTextLength	(	IntPtr	net,
		ref uint	len
	)

Gets the size of text. (Include null)

Parameters

net	A tokenizer instance pointer
len	The length of text

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

◆ ailiaTokenizerGetTokenCount()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerGetTokenCount	(	IntPtr	net,
		ref uint	count
	)

Gets the number of tokens.

Parameters

net	A tokenizer instance pointer
count	The number of objects

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

◆ ailiaTokenizerGetTokens()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerGetTokens	(	IntPtr	net,
		IntPtr	tokens,
		uint	count
	)

Gets the tokens.

Parameters

net	A tokenizer instance pointer
tokens	Token
count	Token count

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

If ailiaTokenizerEncode() is not run at all, the function returns AILIA_STATUS_INVALID_STATE .

◆ ailiaTokenizerGetVocab()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerGetVocab	(	IntPtr	net,
		int	token,
		ref IntPtr	vocab
	)

Perform encode.

Parameters

net	A tokenizer instance pointer
token	Token
text	Text of vocab (UTF8)

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

There is no need to release the vocab. The validity period of the vocab will last until the next time the ailiaTokenizer API is called.

◆ ailiaTokenizerGetVocabSize()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerGetVocabSize	(	IntPtr	net,
		ref uint	size
	)

Gets the size of vocab. (Include null)

Parameters

net	A tokenizer instance pointer
size	The size of vocab

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

◆ ailiaTokenizerGetWordIDs()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerGetWordIDs	(	IntPtr	net,
		IntPtr	tokens,
		uint	count
	)

Gets the word ID.

Parameters

net	A tokenizer instance pointer
word_ids	Word ID
count	Token count

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

If ailiaTokenizerEncode() is not run at all, the function returns AILIA_STATUS_INVALID_STATE . This is valid only for AILIA_TOKENIZER_TYPE_ROBERTA and AILIA_TOKENIZER_TYPE_BERT.

◆ ailiaTokenizerOpenAddedTokensFile()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerOpenAddedTokensFile	(	IntPtr	net,
		string	path
	)

Open added tokens file.

Parameters

net	A pointer to the tokenizer instance pointer
path	Path for special token file

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

Open a added tokens file (json). This API only requires for AILIA_TOKENIZER_TYPE_WHISPER.

◆ ailiaTokenizerOpenDictionaryFile()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerOpenDictionaryFile	(	IntPtr	net,
		string	path
	)

Open dictionary file.

Parameters

net	A pointer to the tokenizer instance pointer
path	Path for dictionary of Mecab

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

Open a model file for Mecab. This API only requires for AILIA_TOKENIZER_TYPE_BERT_JAPANESE_XXX.

◆ ailiaTokenizerOpenMergeFile()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerOpenMergeFile	(	IntPtr	net,
		string	path
	)

Open merges file.

Parameters

net	A pointer to the tokenizer instance pointer
path	Path for merges file

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

Open a merge file (txt). This API only requires for AILIA_TOKENIZER_TYPE_ROBERTA or AILIA_TOKENIZER_TYPE_WHISPER or AILIA_TOKENIZER_TYPE_GPT2.

◆ ailiaTokenizerOpenModelFile()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerOpenModelFile	(	IntPtr	net,
		string	ath
	)

Open model file.

Parameters

net	A pointer to the tokenizer instance pointer
path	Path for SentencePiece

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

Open a model file for SentencePiece. This API only requires for AILIA_TOKENIZER_TYPE_XLM_ROBERTA or AILIA_TOKENIZER_TYPE_MARIAN.

◆ ailiaTokenizerOpenTokenizerConfigFile()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerOpenTokenizerConfigFile	(	IntPtr	net,
		string	path
	)

Open tokenizer config file.

Parameters

net	A pointer to the tokenizer instance pointer
path	Path for config file

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

Open a tokenizer config file (json). This API only requires for AILIA_TOKENIZER_TYPE_BERT.

◆ ailiaTokenizerOpenVocabFile()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerOpenVocabFile	(	IntPtr	net,
		string	path
	)

Open vocab file.

Parameters

net	A pointer to the tokenizer instance pointer
path	Path for Vocab file

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

Open a vocab file (json for ROBERTA or WHISPER or GPT2, txt for others).

◆ ailiaTokenizerUtf32ToUtf8()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerUtf32ToUtf8	(	IntPtr	utf8,
		ref uint	processed_byte,
		uint	utf32
	)

Convert UTF32 character to UTF8 character.

Parameters

utf8	UTF8 character(Require greater than 4byte)
processed_byte	Processed bytes on UTF8
utf32	UTF32 character

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

◆ ailiaTokenizerUtf8ToUtf32()

static int ailiaTokenizer.AiliaTokenizer.ailiaTokenizerUtf8ToUtf32	(	ref uint	utf32,
		ref uint	processed_byte,
		IntPtr	utf8,
		uint	utf8_len
	)

Convert UTF8 character to UTF32 character.

Parameters

utf32	UTF32の文字
processed_byte	Processed bytes on UTF8
utf8	UTF8の文字
utf8_len	Buffer Size

Returns: If this function is successful, it returns AILIA_STATUS_SUCCESS , or an error code otherwise.

Member Data Documentation

◆ AILIA_TOKENIZER_FLAG_NONE

const Int32 ailiaTokenizer.AiliaTokenizer.AILIA_TOKENIZER_FLAG_NONE = (0)

static

◆ AILIA_TOKENIZER_FLAG_UTF8_SAFE

const Int32 ailiaTokenizer.AiliaTokenizer.AILIA_TOKENIZER_FLAG_UTF8_SAFE = (1)

static

◆ AILIA_TOKENIZER_TYPE_BERT

const Int32 ailiaTokenizer.AiliaTokenizer.AILIA_TOKENIZER_TYPE_BERT = (8)

static

◆ AILIA_TOKENIZER_TYPE_BERT_JAPANESE_CHARACTER

const Int32 ailiaTokenizer.AiliaTokenizer.AILIA_TOKENIZER_TYPE_BERT_JAPANESE_CHARACTER = (5)

static

◆ AILIA_TOKENIZER_TYPE_BERT_JAPANESE_WORDPIECE

const Int32 ailiaTokenizer.AiliaTokenizer.AILIA_TOKENIZER_TYPE_BERT_JAPANESE_WORDPIECE = (4)

static

◆ AILIA_TOKENIZER_TYPE_CLIP

const Int32 ailiaTokenizer.AiliaTokenizer.AILIA_TOKENIZER_TYPE_CLIP = (1)

static

◆ AILIA_TOKENIZER_TYPE_GPT2

const Int32 ailiaTokenizer.AiliaTokenizer.AILIA_TOKENIZER_TYPE_GPT2 = (9)

static

◆ AILIA_TOKENIZER_TYPE_LLAMA

const Int32 ailiaTokenizer.AiliaTokenizer.AILIA_TOKENIZER_TYPE_LLAMA = (10)

static

◆ AILIA_TOKENIZER_TYPE_MARIAN

const Int32 ailiaTokenizer.AiliaTokenizer.AILIA_TOKENIZER_TYPE_MARIAN = (3)

static

◆ AILIA_TOKENIZER_TYPE_ROBERTA

const Int32 ailiaTokenizer.AiliaTokenizer.AILIA_TOKENIZER_TYPE_ROBERTA = (7)

static

◆ AILIA_TOKENIZER_TYPE_T5

const Int32 ailiaTokenizer.AiliaTokenizer.AILIA_TOKENIZER_TYPE_T5 = (6)

static

◆ AILIA_TOKENIZER_TYPE_WHISPER

const Int32 ailiaTokenizer.AiliaTokenizer.AILIA_TOKENIZER_TYPE_WHISPER = (0)

static

◆ AILIA_TOKENIZER_TYPE_XLM_ROBERTA

const Int32 ailiaTokenizer.AiliaTokenizer.AILIA_TOKENIZER_TYPE_XLM_ROBERTA = (2)

static

◆ LIBRARY_NAME

const String ailiaTokenizer.AiliaTokenizer.LIBRARY_NAME ="ailia_tokenizer"

static

The documentation for this class was generated from the following file:

AiliaTokenizer.cs

Public Member Functions

Static Public Attributes

Member Function Documentation

◆ ailiaTokenizerAddSpecialTokens()

◆ ailiaTokenizerCreate()

◆ ailiaTokenizerDecode()

◆ ailiaTokenizerDecodeWithSpecialTokens()

◆ ailiaTokenizerDestroy()

◆ ailiaTokenizerEncode()

◆ ailiaTokenizerEncodeWithSpecialTokens()

◆ ailiaTokenizerGetCharEnds()

◆ ailiaTokenizerGetCharStarts()

◆ ailiaTokenizerGetText()

◆ ailiaTokenizerGetTextLength()

◆ ailiaTokenizerGetTokenCount()

◆ ailiaTokenizerGetTokens()

◆ ailiaTokenizerGetVocab()

◆ ailiaTokenizerGetVocabSize()

◆ ailiaTokenizerGetWordIDs()

◆ ailiaTokenizerOpenAddedTokensFile()

◆ ailiaTokenizerOpenDictionaryFile()

◆ ailiaTokenizerOpenMergeFile()

◆ ailiaTokenizerOpenModelFile()

◆ ailiaTokenizerOpenTokenizerConfigFile()

◆ ailiaTokenizerOpenVocabFile()

◆ ailiaTokenizerUtf32ToUtf8()

◆ ailiaTokenizerUtf8ToUtf32()

Member Data Documentation

◆ AILIA_TOKENIZER_FLAG_NONE

◆ AILIA_TOKENIZER_FLAG_UTF8_SAFE

◆ AILIA_TOKENIZER_TYPE_BERT

◆ AILIA_TOKENIZER_TYPE_BERT_JAPANESE_CHARACTER

◆ AILIA_TOKENIZER_TYPE_BERT_JAPANESE_WORDPIECE

◆ AILIA_TOKENIZER_TYPE_CLIP

◆ AILIA_TOKENIZER_TYPE_GPT2

◆ AILIA_TOKENIZER_TYPE_LLAMA

◆ AILIA_TOKENIZER_TYPE_MARIAN

◆ AILIA_TOKENIZER_TYPE_ROBERTA

◆ AILIA_TOKENIZER_TYPE_T5

◆ AILIA_TOKENIZER_TYPE_WHISPER

◆ AILIA_TOKENIZER_TYPE_XLM_ROBERTA

◆ LIBRARY_NAME