Skip to content

Commit 6673268

Browse files
authored
[Test] Enable default graph mode for test case test_guided_json_completion (vllm-project#10873)
### What this PR does / why we need it? For the test case `test_guided_json_completion`, use the default graph mode `FULL_AND_PIECEWISE` ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? Ran `test_guided_json_completion` locally with the change applied and verified it passes. - vLLM version: v0.22.1 - vLLM main: vllm-project/vllm@967c5c3 Signed-off-by: zhiyu-wa <1959864813@qq.com>
1 parent 622e1b1 commit 6673268

1 file changed

Lines changed: 1 addition & 1 deletion

File tree

tests/e2e/pull_request/one_card/test_guided_decoding.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -89,7 +89,7 @@ def test_guided_json_completion(guided_decoding_backend: str, sample_json_schema
8989
"seed": 0,
9090
"structured_outputs_config": {"backend": guided_decoding_backend},
9191
}
92-
with VllmRunner(MODEL_NAME, compilation_config={"cudagraph_mode": "PIECEWISE"}, **runner_kwargs) as vllm_model:
92+
with VllmRunner(MODEL_NAME, **runner_kwargs) as vllm_model:
9393
prompts = [f"Give an example JSON for an employee profile that fits this schema: {sample_json_schema}"] * 2
9494
inputs = vllm_model.get_inputs(prompts)
9595
outputs = vllm_model.model.generate(inputs, sampling_params=sampling_params)

0 commit comments

Comments
 (0)