add dpsk-v2-lit2 test

Signed-off-by: wangli <[email protected]>

Author: Potabk

.github/workflows/vllm_ascend_test.yaml

@@ -135,6 +135,7 @@ jobs:
       - name: Run vllm-project/vllm-ascend test on V0 engine
         env:
           VLLM_USE_V1: 0
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
         run: |
           if [[ "${{ matrix.os }}" == "linux-arm64-npu-1" ]]; then
             pytest -sv tests/singlecard/test_offline_inference.py

tests/multicard/test_offline_inference_distributed.py

@@ -32,6 +32,7 @@
 
 @pytest.mark.parametrize("model, distributed_executor_backend", [
     ("Qwen/QwQ-32B", "mp"),
+    ("deepseek-ai/DeepSeek-V2-Lite", "mp"),
 ])
 def test_models_distributed(model: str,
                             distributed_executor_backend: str) -> None: