[CI/Build][AMD] Enable Entrypoints Integration Test (Pooling) to run without error on ROCm (#29212)

Signed-off-by: Randall Smith <ransmith@amd.com>
Co-authored-by: Randall Smith <ransmith@amd.com>

tests/entrypoints/pooling/openai/test_embedding.py

@@ -19,6 +19,7 @@ from vllm.entrypoints.openai.protocol import (
     EmbeddingResponse,
     PoolingResponse,
 )
+from vllm.platforms import current_platform
 from vllm.transformers_utils.tokenizer import get_tokenizer
 from vllm.utils.serial_utils import (
     EMBED_DTYPE_TO_TORCH_DTYPE,
@@ -28,6 +29,11 @@ from vllm.utils.serial_utils import (
     decode_pooling_output,
 )
 
+if current_platform.is_rocm():
+    pytest.skip(
+        "Encoder self-attention is not implemented on ROCm.", allow_module_level=True
+    )
+
 MODEL_NAME = "intfloat/multilingual-e5-small"
 DUMMY_CHAT_TEMPLATE = """{% for message in messages %}{{message['role'] + ': ' + message['content'] + '\\n'}}{% endfor %}"""  # noqa: E501
 DTYPE = "bfloat16"

tests/entrypoints/pooling/openai/test_embedding_dimensions.py

@@ -12,6 +12,12 @@ from tests.models.language.pooling.embed_utils import run_embedding_correctness_
 from tests.models.utils import EmbedModelInfo
 from tests.utils import RemoteOpenAIServer
 from vllm.entrypoints.openai.protocol import EmbeddingResponse
+from vllm.platforms import current_platform
+
+if current_platform.is_rocm():
+    pytest.skip(
+        "Encoder self-attention is not implemented on ROCm.", allow_module_level=True
+    )
 
 MODELS = [
     EmbedModelInfo("intfloat/multilingual-e5-small", is_matryoshka=False),

tests/entrypoints/pooling/openai/test_embedding_long_text.py

@@ -16,6 +16,12 @@ import pytest_asyncio
 
 from tests.utils import RemoteOpenAIServer
 from vllm.entrypoints.openai.protocol import EmbeddingResponse
+from vllm.platforms import current_platform
+
+if current_platform.is_rocm():
+    pytest.skip(
+        "Encoder self-attention is not implemented on ROCm.", allow_module_level=True
+    )
 
 
 def _generate_random_text(word_count: int) -> str:

tests/entrypoints/pooling/openai/test_rerank.py

@@ -8,6 +8,12 @@ import torch.nn.functional as F
 
 from tests.utils import RemoteOpenAIServer
 from vllm.entrypoints.openai.protocol import PoolingResponse, RerankResponse
+from vllm.platforms import current_platform
+
+if current_platform.is_rocm():
+    pytest.skip(
+        "Encoder self-attention is not implemented on ROCm.", allow_module_level=True
+    )
 
 MODEL_NAME = "BAAI/bge-reranker-base"
 DTYPE = "bfloat16"

tests/entrypoints/pooling/openai/test_score.py

@@ -10,6 +10,12 @@ from torch import tensor
 
 from tests.utils import RemoteOpenAIServer
 from vllm.entrypoints.openai.protocol import ScoreResponse
+from vllm.platforms import current_platform
+
+if current_platform.is_rocm():
+    pytest.skip(
+        "Encoder self-attention is not implemented on ROCm.", allow_module_level=True
+    )
 
 MODELS = [
     {"name": "BAAI/bge-reranker-v2-m3", "is_cross_encoder": True},

tests/entrypoints/pooling/openai/test_truncation.py

@@ -7,6 +7,12 @@ import pytest
 import pytest_asyncio
 
 from tests.utils import RemoteOpenAIServer
+from vllm.platforms import current_platform
+
+if current_platform.is_rocm():
+    pytest.skip(
+        "Encoder self-attention is not implemented on ROCm.", allow_module_level=True
+    )
 
 MODEL_NAME = "sentence-transformers/all-MiniLM-L12-v2"
 max_model_len = 128