[Model] Support Qwen2 embeddings and use tags to select model tests (#10184)

tests/models/embedding/language/test_cls_models.py

@@ -9,10 +9,14 @@ import pytest
 import torch
 from transformers import AutoModelForSequenceClassification
 
-CLASSIFICATION_MODELS = ["jason9693/Qwen2.5-1.5B-apeach"]
 
-
-@pytest.mark.parametrize("model", CLASSIFICATION_MODELS)
+@pytest.mark.parametrize(
+    "model",
+    [
+        pytest.param("jason9693/Qwen2.5-1.5B-apeach",
+                     marks=[pytest.mark.core_model, pytest.mark.cpu_model]),
+    ],
+)
 @pytest.mark.parametrize("dtype", ["float"])
 def test_classification_models(
     hf_runner,
@@ -23,31 +27,19 @@ def test_classification_models(
 ) -> None:
     with vllm_runner(model, dtype=dtype) as vllm_model:
         vllm_outputs = vllm_model.classify(example_prompts)
+        # This test is for verifying whether the model's extra_repr
+        # can be printed correctly.
+        print(vllm_model.model.llm_engine.model_executor.driver_worker.
+              model_runner.model)
 
     with hf_runner(model,
                    dtype=dtype,
                    auto_cls=AutoModelForSequenceClassification) as hf_model:
         hf_outputs = hf_model.classify(example_prompts)
 
-    print(hf_outputs, vllm_outputs)
-
     # check logits difference
     for hf_output, vllm_output in zip(hf_outputs, vllm_outputs):
         hf_output = torch.tensor(hf_output)
         vllm_output = torch.tensor(vllm_output)
 
         assert torch.allclose(hf_output, vllm_output, 1e-3)
-
-
-@pytest.mark.parametrize("model", CLASSIFICATION_MODELS)
-@pytest.mark.parametrize("dtype", ["float"])
-def test_classification_model_print(
-    vllm_runner,
-    model: str,
-    dtype: str,
-) -> None:
-    with vllm_runner(model, dtype=dtype) as vllm_model:
-        # This test is for verifying whether the model's extra_repr
-        # can be printed correctly.
-        print(vllm_model.model.llm_engine.model_executor.driver_worker.
-              model_runner.model)