minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 13 Packages Projects Releases Wiki Activity

fix: prevent stale bitmask leakage in LLGuidance grammar backend (#23653)

Browse Source 
Signed-off-by: jellysnack <oleg.jellysnack@gmail.com>
Co-authored-by: YorkSu <45256779+YorkSu@users.noreply.github.com>
Co-authored-by: jellysnack <oleg.jellysnack@gmail.com>
This commit is contained in:
Oleg Zhelezniak
2026-06-13 13:27:08 -07:00
committed by GitHub
co-authored by YorkSu
parent 47fabb52ed
commit 8a9f2aa116
1 changed files with 1 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/sglang/srt/constrained/llguidance_backend.py
+1 -11
View File
@@ -51,7 +51,6 @@ class GuidanceGrammar(BaseGrammarObject):
) )
self._check_err() self._check_err()
self.bitmask = None
self.eos_token = self.llguidance_tokenizer.eos_token self.eos_token = self.llguidance_tokenizer.eos_token
def accept_token(self, token: int): def accept_token(self, token: int):
@@ -83,16 +82,7 @@ class GuidanceGrammar(BaseGrammarObject):
def allocate_vocab_mask( def allocate_vocab_mask(
self, vocab_size: int, batch_size: int, device self, vocab_size: int, batch_size: int, device
) -> torch.Tensor: ) -> torch.Tensor:
if self.bitmask is None or self.bitmask.shape[0] < batch_size: return allocate_token_bitmask(batch_size, self.llguidance_tokenizer.vocab_size)
# only create bitmask when batch gets larger
self.bitmask = allocate_token_bitmask(
batch_size, self.llguidance_tokenizer.vocab_size
)
bitmask = self.bitmask
else:
bitmask = self.bitmask[:batch_size]
return bitmask
@staticmethod @staticmethod
def move_vocab_mask(vocab_mask: torch.Tensor, device) -> torch.Tensor: def move_vocab_mask(vocab_mask: torch.Tensor, device) -> torch.Tensor: