Skip to content

fix: disable thinking mode for structured JSON output requests #34

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
back2zion wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix: disable thinking mode for structured JSON output requests #34

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion backend/app/services/oasis_profile_generator.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -478,7 +478,8 @@ def _generate_profile_with_llm(
{"role": "user", "content": prompt}
],
response_format={"type": "json_object"},
temperature=0.7 - (attempt * 0.1) # Lower temperature with each retry
temperature=0.7 - (attempt * 0.1), # Lower temperature with each retry
extra_body={"chat_template_kwargs": {"enable_thinking": False}},
# Don't set max_tokens, let LLM generate freely
)

Expand Down
10 changes: 7 additions & 3 deletions backend/app/utils/llm_client.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -72,12 +72,16 @@ def chat(

if response_format:
kwargs["response_format"] = response_format
# Disable thinking mode (e.g. Qwen3 <think> tags) when requesting
# structured JSON output — thinking tokens break JSON parsing and
# cause infinite retry loops with vLLM's guided decoding.
kwargs.setdefault("extra_body", {})
kwargs["extra_body"]["chat_template_kwargs"] = {"enable_thinking": False}

# For Ollama: pass num_ctx via extra_body to prevent prompt truncation
if self._is_ollama() and self._num_ctx:
kwargs["extra_body"] = {
"options": {"num_ctx": self._num_ctx}
}
kwargs.setdefault("extra_body", {})
kwargs["extra_body"]["options"] = {"num_ctx": self._num_ctx}

response = self.client.chat.completions.create(**kwargs)
content = response.choices[0].message.content
Expand Down