A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[Benchmark] Fix regression in structured output benchmark (#25500)
Signed-off-by: Russell Bryant <rbryant@redhat.com>
Russell Bryant committed
Sep 24, 2025 at 06:40 UTC
164299500be5ae1e8c71e06d4b90ac82d9a0b31b
1 file changed
+2
-1
benchmarks/benchmark_serving_structured_output.py
+2
-1
@@ -449,7 +449,8 @@ async def benchmark(
449
def prepare_extra_body(request) -> dict:
450
extra_body = {}
451
# Add the schema to the extra_body
452
- extra_body[request.structure_type] = request.schema
452
+ extra_body["structured_outputs"] = {}
453
+ extra_body["structured_outputs"][request.structure_type] = request.schema
454
return extra_body
455
456
print("Starting initial single prompt test run...")