xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 07:04:53 +08:00
Code Issues Packages Projects Releases Wiki Activity

[CI][TPU] Temporarily Disable Quant Test on TPU (#15649)

Browse Source 
Signed-off-by: rshaw@neuralmagic.com <robertgshaw2@gmail.com>
This commit is contained in:
Robert Shaw 2025-03-27 22:45:05 -04:00 committed by GitHub
parent b4245a48df
commit 8a49eea74b
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 5 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
.buildkite/run-tpu-v1-test.sh
View File

@ -28,15 +28,16 @@ docker run --privileged --net host --shm-size=16G -it \
&& echo TEST_3 \
&& pytest -v -s /workspace/vllm/tests/entrypoints/llm/test_accuracy.py::test_lm_eval_accuracy_v1_engine \
&& echo TEST_4 \
&& pytest -s -v /workspace/vllm/tests/tpu/test_quantization_accuracy.py \
&& echo TEST_5 \
&& python3 /workspace/vllm/examples/offline_inference/tpu.py \
&& echo TEST_6 \
&& echo TEST_5 \
&& pytest -s -v /workspace/vllm/tests/tpu/worker/test_tpu_model_runner.py \
&& echo TEST_7 \
&& echo TEST_6 \
&& pytest -s -v /workspace/vllm/tests/v1/tpu/test_sampler.py" \
# TODO: This test fails because it uses RANDOM_SEED sampling
# && VLLM_USE_V1=1 pytest -v -s /workspace/vllm/tests/tpu/test_custom_dispatcher.py \
# TODO: Re-enable this after fixing recompilation in quantization.
# && echo TEST_4 \
# && pytest -s -v /workspace/vllm/tests/tpu/test_quantization_accuracy.py \

3
tests/v1/tpu/test_basic.py
View File

@ -31,14 +31,12 @@ TENSOR_PARALLEL_SIZES = [1]
reason="This is a basic test for TPU only")
@pytest.mark.parametrize("model", MODELS)
@pytest.mark.parametrize("max_tokens", [5])
@pytest.mark.parametrize("enforce_eager", [True])
@pytest.mark.parametrize("tensor_parallel_size", TENSOR_PARALLEL_SIZES)
def test_models(
vllm_runner: type[VllmRunner],
monkeypatch: pytest.MonkeyPatch,
model: str,
max_tokens: int,
enforce_eager: bool,
tensor_parallel_size: int,
) -> None:
prompt = "The next numbers of the sequence " + ", ".join(
@ -51,7 +49,6 @@ def test_models(
with vllm_runner(
model,
max_model_len=8192,
enforce_eager=enforce_eager,
gpu_memory_utilization=0.7,
max_num_seqs=16,
tensor_parallel_size=tensor_parallel_size) as vllm_model: