Fixed vllm test suite Automation Failures (#998)

* Fixed vllm test suite Automation Failures

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Added failure messgae for speculative decoding lab_draft tests

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Author: Raghul-M

tests/model_serving/model_runtime/vllm/basic_model_deployment/test_elyza_japanese_llama_2_7b_instruct.py

@@ -21,7 +21,7 @@
     "--model=/mnt/models",
     "--uvicorn-log-level=debug",
     "--dtype=float16",
-    "--chat-template=/app/data/template/template_chatglm.jinja",
+    "--chat-template=/opt/app-root/template/template_chatglm.jinja",
 ]
 
 MODEL_PATH: str = "ELYZA-japanese-Llama-2-7b-instruct-hf"

tests/model_serving/model_runtime/vllm/basic_model_deployment/test_granite_7b_redhat_lab.py

@@ -21,7 +21,7 @@
     "--model=/mnt/models",
     "--uvicorn-log-level=debug",
     "--dtype=float16",
-    "--chat-template=/app/data/template/template_chatglm.jinja",
+    "--chat-template=/opt/app-root/template/template_chatglm.jinja",
 ]
 
 MODEL_PATH: str = "granite-7b-redhat-lab"

tests/model_serving/model_runtime/vllm/basic_model_deployment/test_granite_7b_starter.py

@@ -21,7 +21,7 @@
     "--model=/mnt/models",
     "--uvicorn-log-level=debug",
     "--dtype=float16",
-    "--chat-template=/app/data/template/template_chatglm.jinja",
+    "--chat-template=/opt/app-root/template/template_chatglm.jinja",
 ]
 
 MODEL_PATH: str = "granite-7b-starter"

tests/model_serving/model_runtime/vllm/basic_model_deployment/test_llama31_8B_instruct.py

@@ -21,7 +21,7 @@
     "--model=/mnt/models",
     "--uvicorn-log-level=debug",
     "--dtype=float16",
-    "--chat-template=/app/data/template/tool_chat_template_llama3.1_json.jinja",
+    "--chat-template=/opt/app-root/template/tool_chat_template_llama3.1_json.jinja",
 ]
 
 MODEL_PATH: str = "Meta-Llama-3.1-8B"

tests/model_serving/model_runtime/vllm/basic_model_deployment/test_llama3_8B_instruct.py

@@ -21,7 +21,7 @@
     "--model=/mnt/models",
     "--uvicorn-log-level=debug",
     "--dtype=float16",
-    "--chat-template=/app/data/template/template_chatglm.jinja",
+    "--chat-template=/opt/app-root/template/template_chatglm.jinja",
 ]
 
 MODEL_PATH: str = "Meta-Llama-3-8B-Instruct"

tests/model_serving/model_runtime/vllm/basic_model_deployment/test_llama_2_13b_chat.py

@@ -22,7 +22,7 @@
     "--model=/mnt/models",
     "--uvicorn-log-level=debug",
     "--dtype=float16",
-    "--chat-template=/app/data/template/template_chatglm.jinja",
+    "--chat-template=/opt/app-root/template/template_chatglm.jinja",
 ]
 
 MODEL_PATH: str = "Llama-2-13b-chat-hf"

tests/model_serving/model_runtime/vllm/basic_model_deployment/test_merlinite_7b_lab.py

@@ -21,7 +21,7 @@
     "--model=/mnt/models",
     "--uvicorn-log-level=debug",
     "--dtype=float16",
-    "--chat-template=/app/data/template/template_chatglm.jinja",
+    "--chat-template=/opt/app-root/template/template_chatglm.jinja",
 ]
 
 MODEL_PATH: str = "merlinite-7b-lab"

tests/model_serving/model_runtime/vllm/multimodal/test_granite_31_2b_vision.py

@@ -13,7 +13,7 @@
 LOGGER = get_logger(name=__name__)
 
 
-SERVING_ARGUMENT: List[str] = ["--model=/mnt/models", "--uvicorn-log-level=debug", "--limit-mm-per-prompt", "image=2"]
+SERVING_ARGUMENT: List[str] = ["--model=/mnt/models", "--uvicorn-log-level=debug", '--limit-mm-per-prompt={"image": 2}']
 
 MODEL_PATH: str = "ibm-granite/granite-vision-3.1-2b-preview"
 

tests/model_serving/model_runtime/vllm/quantization/test_openhermes-2_5_mistral-7b_awq.py

@@ -13,7 +13,7 @@
 SERVING_ARGUMENT = [
     "--model=/mnt/models",
     "--uvicorn-log-level=debug",
-    "--chat-template=/app/data/template/tool_chat_template_mistral.jinja",
+    "--chat-template=/opt/app-root/template/tool_chat_template_mistral.jinja",
 ]
 
 MODEL_PATH = "TheBloke/OpenHermes-2.5-Mistral-7B-AWQ"

tests/model_serving/model_runtime/vllm/speculative_decoding/test_granite_7b_lab_draft.py

@@ -19,7 +19,6 @@
     "--dtype=float16",
     "--speculative_config",
     '{ "model": "/mnt/models/granite-7b-instruct-accelerator", "num_speculative_tokens": 5 }',
-    "--use-v2-block-manager",
 ]
 
 MODEL_PATH: str = "speculative_decoding"
@@ -48,6 +47,9 @@
     indirect=True,
 )
 class TestGraniteLabDraftModel:
+    @pytest.mark.xfail(
+        reason="vLLM does not support MLPSpeculatorPreTrainedModel architecture for draft model speculative decoding"
+    )
     def test_spec_draft_inference(
         self,
         vllm_inference_service: Generator[InferenceService, Any, Any],
@@ -97,6 +99,9 @@ def test_spec_draft_inference(
     indirect=True,
 )
 class TestMultiGraniteLabDraftModel:
+    @pytest.mark.xfail(
+        reason="vLLM does not support MLPSpeculatorPreTrainedModel architecture for draft model speculative decoding"
+    )
     def test_multi_spec_draft_inference(
         self,
         vllm_inference_service: Generator[InferenceService, Any, Any],