Add overridable hooks for custom chat serving implementations (#25807)

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
Randall Lin
2026-05-21 11:21:25 +08:00
committed by GitHub
co-authored by Cursor
parent 74c6294ba9
commit 791a2f057f
7 changed files with 371 additions and 167 deletions
@@ -216,6 +216,16 @@ class InputFormat(Enum):
class TokenizerManager(TokenizerControlMixin, TokenizerManagerScoreMixin):
"""TokenizerManager is a process that tokenizes the text."""
@property
def serving_chat_class(self):
"""Return the serving chat class for OpenAI API.
Override in subclass to provide custom serving behavior.
"""
from sglang.srt.entrypoints.openai.serving_chat import OpenAIServingChat
return OpenAIServingChat
def __init__(
self,
server_args: ServerArgs,