[data][llm] Cannot pass enable_thinking to chat_template #56384
Description
What happened + What you expected to happen
When using Qwen/Qwen3-8B with ray.data.llm, there is no way to turn enable_thinking off
The ChatTemplateUDF doesn't pass down any keyword arguments
Versions / Dependencies
Python
Reproduction script
config = vLLMEngineProcessorConfig(
model_source="Qwen/Qwen3-8B",
concurrency=1,
apply_chat_template=True,
tokenize=False,
)
processor = build_llm_processor(
config,
preprocess=lambda row: dict(
messages=[
{"role": "system", "content": "You are a bot that responds with haikus."},
{"role": "user", "content": row["item"]}
],
sampling_params=dict(
temperature=0.3,
max_tokens=250,
)
),
postprocess=lambda row: dict(
answer=row["generated_text"],
**row # This will return all the original columns in the dataset.
),
)
Issue Severity
Low: It annoys or frustrates me.