py : handle byte tokens in get_token_type

This commit is contained in:
Georgi Gerganov 2024-02-05 13:42:54 +02:00
parent abb61944a5
commit ded2ad5b88
No known key found for this signature in database
GPG Key ID: BF970631944C16B7

View File

@ -515,10 +515,14 @@ class HfVocab:
# Yield token text, score, and type
yield token_text, self.get_token_score(token_id), self.get_token_type(
token_id, self.special_ids # Reuse already stored special IDs
token_id, token_text, self.special_ids # Reuse already stored special IDs
)
def get_token_type(self, token_id: int, special_ids: set[int]) -> gguf.TokenType:
def get_token_type(self, token_id: int, token_text: bytes, special_ids: set[int]) -> gguf.TokenType:
# Special case for byte tokens
if re.fullmatch(br"<0x[0-9A-Fa-f]{2}>", token_text):
return gguf.TokenType.BYTE
# Determine token type based on whether it's a special token
return gguf.TokenType.CONTROL if token_id in special_ids else gguf.TokenType.NORMAL
@ -530,7 +534,7 @@ class HfVocab:
def added_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]:
for text in self.added_tokens_list:
if text in self.specials:
toktype = self.get_token_type(self.specials[text], self.special_ids)
toktype = self.get_token_type(self.specials[text], [], self.special_ids)
score = self.get_token_score(self.specials[text])
else:
toktype = gguf.TokenType.USER_DEFINED