Fix FlashInfer A2A IMA by DP-synchronizing the decode graph bucket (#30242) (#30450)

This commit is contained in:
Lianmin Zheng
2026-07-08 14:34:49 -07:00
committed by GitHub
parent c55190a638
commit ca8f15cd70
3 changed files with 55 additions and 34 deletions
@@ -25,7 +25,7 @@ _SRT_ROOT = Path(next(iter(sglang.srt.__path__)))
# Baselines counted over python/sglang/srt/**/*.py, including each function's
# own def line. Ratchet: decrease-only.
_RATCHETS = [
("get_global_server_args", r"\bget_global_server_args\s*\(", 280),
("get_global_server_args", r"\bget_global_server_args\s*\(", 279),
(
"set_global_server_args_for_*",
r"\bset_global_server_args_for_(?:scheduler|tokenizer)\s*\(",