test: Split inference tests to text and vision

.github/workflows/tests.yml

@@ -54,7 +54,7 @@ jobs:
           echo "REPORT_FILE=${REPORT_OUTPUT}" >> "$GITHUB_ENV"
 
           export INFERENCE_MODEL=meta-llama/Llama-3.1-8B-Instruct
-          LLAMA_STACK_CONFIG=./llama_stack/templates/${{ matrix.provider }}/run.yaml pytest --md-report --md-report-verbose=1 ./tests/client-sdk/inference/test_inference.py --md-report-output "$REPORT_OUTPUT"
+          LLAMA_STACK_CONFIG=./llama_stack/templates/${{ matrix.provider }}/run.yaml pytest --md-report --md-report-verbose=1 ./tests/client-sdk/inference/ --md-report-output "$REPORT_OUTPUT"
 
       - name: Output reports to the job summary
         if: always()

tests/client-sdk/README.md

@@ -4,18 +4,18 @@ You can run llama stack integration tests on either a Llama Stack Library or a L
 To test on a Llama Stack library with certain configuration, run
 ```bash
 LLAMA_STACK_CONFIG=./llama_stack/templates/cerebras/run.yaml
-pytest -s -v tests/client-sdk/inference/test_inference.py
+pytest -s -v tests/client-sdk/inference/
 ```
 or just the template name
 ```bash
 LLAMA_STACK_CONFIG=together
-pytest -s -v tests/client-sdk/inference/test_inference.py
+pytest -s -v tests/client-sdk/inference/
 ```
 
 To test on a Llama Stack endpoint, run
 ```bash
 LLAMA_STACK_BASE_URL=http//localhost:8089
-pytest -s -v tests/client-sdk/inference/test_inference.py
+pytest -s -v tests/client-sdk/inference
 ```
 
 ## Report Generation

tests/client-sdk/inference/test_inference.py → tests/client-sdk/inference/test_text_inference.py

@@ -4,9 +4,6 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-import base64
-import pathlib
-
 import pytest
 from pydantic import BaseModel
 
@@ -56,23 +53,6 @@ def get_weather_tool_definition():
     }
 
 
-@pytest.fixture
-def image_path():
-    return pathlib.Path(__file__).parent / "dog.png"
-
-
-@pytest.fixture
-def base64_image_data(image_path):
-    # Convert the image to base64
-    return base64.b64encode(image_path.read_bytes()).decode("utf-8")
-
-
-@pytest.fixture
-def base64_image_url(base64_image_data, image_path):
-    # suffix includes the ., so we remove it
-    return f"data:image/{image_path.suffix[1:]};base64,{base64_image_data}"
-
-
 def test_text_completion_non_streaming(llama_stack_client, text_model_id):
     response = llama_stack_client.inference.completion(
         content="Complete the sentence using one word: Roses are red, violets are ",
@@ -299,101 +279,3 @@ class AnswerFormat(BaseModel):
     assert answer.last_name == "Jordan"
     assert answer.year_of_birth == 1963
     assert answer.num_seasons_in_nba == 15
-
-
-def test_image_chat_completion_non_streaming(llama_stack_client, vision_model_id):
-    message = {
-        "role": "user",
-        "content": [
-            {
-                "type": "image",
-                "image": {
-                    "url": {
-                        # TODO: Replace with Github based URI to resources/sample1.jpg
-                        "uri": "https://www.healthypawspetinsurance.com/Images/V3/DogAndPuppyInsurance/Dog_CTA_Desktop_HeroImage.jpg"
-                    },
-                },
-            },
-            {
-                "type": "text",
-                "text": "Describe what is in this image.",
-            },
-        ],
-    }
-    response = llama_stack_client.inference.chat_completion(
-        model_id=vision_model_id,
-        messages=[message],
-        stream=False,
-    )
-    message_content = response.completion_message.content.lower().strip()
-    assert len(message_content) > 0
-    assert any(expected in message_content for expected in {"dog", "puppy", "pup"})
-
-
-def test_image_chat_completion_streaming(llama_stack_client, vision_model_id):
-    message = {
-        "role": "user",
-        "content": [
-            {
-                "type": "image",
-                "image": {
-                    "url": {
-                        # TODO: Replace with Github based URI to resources/sample1.jpg
-                        "uri": "https://www.healthypawspetinsurance.com/Images/V3/DogAndPuppyInsurance/Dog_CTA_Desktop_HeroImage.jpg"
-                    },
-                },
-            },
-            {
-                "type": "text",
-                "text": "Describe what is in this image.",
-            },
-        ],
-    }
-    response = llama_stack_client.inference.chat_completion(
-        model_id=vision_model_id,
-        messages=[message],
-        stream=True,
-    )
-    streamed_content = ""
-    for chunk in response:
-        streamed_content += chunk.event.delta.text.lower()
-    assert len(streamed_content) > 0
-    assert any(expected in streamed_content for expected in {"dog", "puppy", "pup"})
-
-
-@pytest.mark.parametrize("type_", ["url", "data"])
-def test_image_chat_completion_base64(llama_stack_client, vision_model_id, base64_image_data, base64_image_url, type_):
-    image_spec = {
-        "url": {
-            "type": "image",
-            "image": {
-                "url": {
-                    "uri": base64_image_url,
-                },
-            },
-        },
-        "data": {
-            "type": "image",
-            "image": {
-                "data": base64_image_data,
-            },
-        },
-    }[type_]
-
-    message = {
-        "role": "user",
-        "content": [
-            image_spec,
-            {
-                "type": "text",
-                "text": "Describe what is in this image.",
-            },
-        ],
-    }
-    response = llama_stack_client.inference.chat_completion(
-        model_id=vision_model_id,
-        messages=[message],
-        stream=False,
-    )
-    message_content = response.completion_message.content.lower().strip()
-    assert len(message_content) > 0

tests/client-sdk/inference/test_vision_inference.py

@@ -0,0 +1,133 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+import base64
+import pathlib
+
+import pytest
+
+
+@pytest.fixture(scope="session")
+def inference_provider_type(llama_stack_client):
+    providers = llama_stack_client.providers.list()
+    inference_providers = [p for p in providers if p.api == "inference"]
+    assert len(inference_providers) > 0, "No inference providers found"
+    return inference_providers[0].provider_type
+
+
+@pytest.fixture
+def image_path():
+    return pathlib.Path(__file__).parent / "dog.png"
+
+
+@pytest.fixture
+def base64_image_data(image_path):
+    # Convert the image to base64
+    return base64.b64encode(image_path.read_bytes()).decode("utf-8")
+
+
+@pytest.fixture
+def base64_image_url(base64_image_data, image_path):
+    # suffix includes the ., so we remove it
+    return f"data:image/{image_path.suffix[1:]};base64,{base64_image_data}"
+
+
+def test_image_chat_completion_non_streaming(llama_stack_client, vision_model_id):
+    message = {
+        "role": "user",
+        "content": [
+            {
+                "type": "image",
+                "image": {
+                    "url": {
+                        # TODO: Replace with Github based URI to resources/sample1.jpg
+                        "uri": "https://www.healthypawspetinsurance.com/Images/V3/DogAndPuppyInsurance/Dog_CTA_Desktop_HeroImage.jpg"
+                    },
+                },
+            },
+            {
+                "type": "text",
+                "text": "Describe what is in this image.",
+            },
+        ],
+    }
+    response = llama_stack_client.inference.chat_completion(
+        model_id=vision_model_id,
+        messages=[message],
+        stream=False,
+    )
+    message_content = response.completion_message.content.lower().strip()
+    assert len(message_content) > 0
+    assert any(expected in message_content for expected in {"dog", "puppy", "pup"})
+
+
+def test_image_chat_completion_streaming(llama_stack_client, vision_model_id):
+    message = {
+        "role": "user",
+        "content": [
+            {
+                "type": "image",
+                "image": {
+                    "url": {
+                        # TODO: Replace with Github based URI to resources/sample1.jpg
+                        "uri": "https://www.healthypawspetinsurance.com/Images/V3/DogAndPuppyInsurance/Dog_CTA_Desktop_HeroImage.jpg"
+                    },
+                },
+            },
+            {
+                "type": "text",
+                "text": "Describe what is in this image.",
+            },
+        ],
+    }
+    response = llama_stack_client.inference.chat_completion(
+        model_id=vision_model_id,
+        messages=[message],
+        stream=True,
+    )
+    streamed_content = ""
+    for chunk in response:
+        streamed_content += chunk.event.delta.text.lower()
+    assert len(streamed_content) > 0
+    assert any(expected in streamed_content for expected in {"dog", "puppy", "pup"})
+
+
+@pytest.mark.parametrize("type_", ["url", "data"])
+def test_image_chat_completion_base64(llama_stack_client, vision_model_id, base64_image_data, base64_image_url, type_):
+    image_spec = {
+        "url": {
+            "type": "image",
+            "image": {
+                "url": {
+                    "uri": base64_image_url,
+                },
+            },
+        },
+        "data": {
+            "type": "image",
+            "image": {
+                "data": base64_image_data,
+            },
+        },
+    }[type_]
+
+    message = {
+        "role": "user",
+        "content": [
+            image_spec,
+            {
+                "type": "text",
+                "text": "Describe what is in this image.",
+            },
+        ],
+    }
+    response = llama_stack_client.inference.chat_completion(
+        model_id=vision_model_id,
+        messages=[message],
+        stream=False,
+    )
+    message_content = response.completion_message.content.lower().strip()
+    assert len(message_content) > 0