Fix GLM-4.6v flash tool calling in transformers 5.x (#31622)

Signed-off-by: baonudesifeizhai <baonudesifeizhai@gmail.com>
This commit is contained in:
baonudesifeizhai
2026-01-05 11:32:43 -08:00
committed by GitHub
parent 51e38a8e30
commit 02dbb933cb
2 changed files with 68 additions and 0 deletions
+14
View File
@@ -56,6 +56,20 @@ class Glm4MoeModelToolParser(ToolParser):
self.tool_call_end_token_id = self.vocab.get(self.tool_call_end_token)
self._buffer = ""
def adjust_request(self, request: ChatCompletionRequest) -> ChatCompletionRequest:
"""
Adjust request parameters to ensure tool call tokens are not skipped
during tokenizer decoding.
"""
request = super().adjust_request(request)
if request.tools and request.tool_choice != "none":
# Ensure tool call tokens (<tool_call>, </tool_call>) are not skipped
# during decoding. Even though they are not marked as special tokens,
# setting skip_special_tokens=False ensures proper handling in
# transformers 5.x where decoding behavior may have changed.
request.skip_special_tokens = False
return request
def extract_tool_calls(
self,
model_output: str,