EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

[CI] Add deepseek-v2-lite test (#631)

Browse Source 
### What this PR does / why we need it?
Add deepseek-v2-lite test, part of #499 
---------

Signed-off-by: wangli <wangli858794774@gmail.com>
This commit is contained in:
Li Wang
2025-05-12 14:59:17 +08:00
committed by GitHub
parent cdece86f2c
commit 8e4e791fcd
2 changed files with 5 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
.github/workflows/vllm_ascend_test.yaml vendored
View File

@@ -115,7 +115,7 @@ jobs:
pytest -sv tests/ops pytest -sv tests/ops
pytest -sv tests/compile pytest -sv tests/compile
else else
pytest -sv tests/multicard/test_offline_inference_distributed.py pytest -sv -k "QwQ" tests/multicard/test_offline_inference_distributed.py
pytest -sv tests/ops pytest -sv tests/ops
pytest -sv tests/compile pytest -sv tests/compile
fi fi
@@ -128,7 +128,8 @@ jobs:
pytest -sv tests/singlecard/test_offline_inference.py pytest -sv tests/singlecard/test_offline_inference.py
pytest -sv tests/ops pytest -sv tests/ops
else else
pytest -sv tests/multicard/test_offline_inference_distributed.py pytest -sv -k "QwQ" tests/multicard/test_offline_inference_distributed.py
pytest -sv -k "DeepSeek" tests/multicard/test_offline_inference_distributed.py
pytest -sv tests/ops pytest -sv tests/ops
fi fi

2
tests/multicard/test_offline_inference_distributed.py
View File

@@ -28,10 +28,12 @@ import vllm # noqa: F401
from tests.conftest import VllmRunner from tests.conftest import VllmRunner
os.environ["PYTORCH_NPU_ALLOC_CONF"] = "max_split_size_mb:256" os.environ["PYTORCH_NPU_ALLOC_CONF"] = "max_split_size_mb:256"
os.environ["VLLM_USE_MODELSCOPE"] = "True"
@pytest.mark.parametrize("model, distributed_executor_backend", [ @pytest.mark.parametrize("model, distributed_executor_backend", [
("Qwen/QwQ-32B", "mp"), ("Qwen/QwQ-32B", "mp"),
("deepseek-ai/DeepSeek-V2-Lite", "mp"),
]) ])
def test_models_distributed(model: str, def test_models_distributed(model: str,
distributed_executor_backend: str) -> None: distributed_executor_backend: str) -> None: