Clean up base constructor in model builder

.github/copilot-instructions.md

@@ -58,7 +58,7 @@ WinML integration downloads `Microsoft.WindowsAppSDK.ML` via NuGet and copies he
 
 ```bash
 # Python tests with test models
-python -m pytest -sv test_onnxruntime_genai_api.py -k "test_name" --test_models ..\test_models
+python -m pytest -sv test_onnxruntime_genai_api.py -k "test_name" --test_models ..\models
 
 # C++ unit tests via CMake/CTest
 ctest --build-config Release --output-on-failure
@@ -160,4 +160,4 @@ Tests are organized by language binding:
 - **Python tests**: `test/python/`, includes end-to-end model testing
 - **Platform tests**: Android/iOS tests run via emulator/simulator
 
-Always test with actual model files from `test/test_models/` directory rather than mock data.
+Always test with actual model files from `test/models/` directory rather than mock data.

.github/instructions/python-model-builder.instructions.md

@@ -11,4 +11,9 @@ When generating or reviewing code in the Python Model Builder (`src/python/py/mo
 
 Read both documents to understand the intended usage, supported models, design principles, and architectural constraints before suggesting or reviewing any code changes in this area.
 
-When a node is inserted into the model, prefer using `self.make_op_name` as the wrapper method for `self.make_node` + `self.make_value` calls.
+## Code Style Guidelines
+
+1. When a node is inserted into the model, prefer using `self.make_op_name` as the wrapper method for `self.make_node` + `self.make_value` calls.
+2. Ignore any CodeQL warnings about how an __init__ method calls an overridden method. These warnings are false positives and can be safely ignored. The warning message is: "this call to ABC in an initialization method is overwritten by XYZ".
+3. Find ways to reduce code duplication by reusing existing functionality and implementing common patterns.
+4. Discover ways to leverage the use of shared code in the base classes to avoid code duplication and improve maintainability.

.github/workflows/linux-cpu-x64-build.yml

@@ -131,12 +131,12 @@ jobs:
           set -e -x
           ctest --test-dir build/cpu/src/java --build-config Release --verbose --timeout 10800
 
-      # This will also download all the test models to the test/test_models directory
+      # This will also download all the test models to the test/models directory
       # These models are used by the python tests as well as C#, C++ and others.
       - name: Run the Python tests
         run: |
           export ORTGENAI_LOG_ORT_LIB=1
-          python3 test/python/test_onnxruntime_genai.py --cwd test/python --test_models test/test_models
+          python3 test/python/test_onnxruntime_genai.py --cwd test/python --test_models test/models
 
       - name: Build the C# API and Run the C# Tests
         run: |
@@ -155,8 +155,8 @@ jobs:
       - name: Test the C# LLM Example with Tool Calling
         run: |
           export ORTGENAI_LOG_ORT_LIB=1
-          python3 test/python/special_tokens.py -p test/test_models/qwen-2.5-0.5b/int4/cpu/tokenizer.json -s "<tool_call>" -e "</tool_call>"
-          ./examples/csharp/ModelChat/bin/Release/net8.0/ModelChat -m test/test_models/qwen-2.5-0.5b/int4/cpu/ -e cpu --response_format lark_grammar --tools_file test/test_models/tool-definitions/weather.json --tool_call_start "<tool_call>" --tool_call_end "</tool_call>" --user_prompt "What is the weather in Redmond, WA?" --tool_output --non_interactive --verbose
+          python3 test/python/special_tokens.py -p test/models/qwen-2.5-0.5b/int4/cpu/tokenizer.json -s "<tool_call>" -e "</tool_call>"
+          ./examples/csharp/ModelChat/bin/Release/net8.0/ModelChat -m test/models/qwen-2.5-0.5b/int4/cpu/ -e cpu --response_format lark_grammar --tools_file test/models/tool-definitions/weather.json --tool_call_start "<tool_call>" --tool_call_end "</tool_call>" --user_prompt "What is the weather in Redmond, WA?" --tool_output --non_interactive --verbose
 
       - name: Run tests
         run: |

.github/workflows/linux-cpu-x64-nightly-build.yml

@@ -136,7 +136,7 @@ jobs:
       - name: Run the Python tests
         run: |
           export ORTGENAI_LOG_ORT_LIB=1
-          python3 test/python/test_onnxruntime_genai.py --cwd test/python --test_models test/test_models --e2e
+          python3 test/python/test_onnxruntime_genai.py --cwd test/python --test_models test/models --e2e
 
       - name: Build the C# API and Run the C# Tests
         run: |
@@ -155,8 +155,8 @@ jobs:
       - name: Test the C# LLM Example with Tool Calling
         run: |
           export ORTGENAI_LOG_ORT_LIB=1
-          python3 test/python/special_tokens.py -p test/test_models/qwen-2.5-0.5b/int4/cpu/tokenizer.json -s "<tool_call>" -e "</tool_call>"
-          ./examples/csharp/ModelChat/bin/Release/net8.0/ModelChat -m test/test_models/qwen-2.5-0.5b/int4/cpu/ -e cpu --response_format lark_grammar --tools_file test/test_models/tool-definitions/weather.json --tool_call_start "<tool_call>" --tool_call_end "</tool_call>" --user_prompt "What is the weather in Redmond, WA?" --tool_output --non_interactive --verbose
+          python3 test/python/special_tokens.py -p test/models/qwen-2.5-0.5b/int4/cpu/tokenizer.json -s "<tool_call>" -e "</tool_call>"
+          ./examples/csharp/ModelChat/bin/Release/net8.0/ModelChat -m test/models/qwen-2.5-0.5b/int4/cpu/ -e cpu --response_format lark_grammar --tools_file test/models/tool-definitions/weather.json --tool_call_start "<tool_call>" --tool_call_end "</tool_call>" --user_prompt "What is the weather in Redmond, WA?" --tool_output --non_interactive --verbose
 
       - name: Run Q&A Example
         run: |

.github/workflows/linux-gpu-x64-build.yml

@@ -180,7 +180,7 @@ jobs:
               ${{ env.PYTHON_EXECUTABLE }} -m pip install -r test/python/cuda/torch/requirements.txt --user && \
               ${{ env.PYTHON_EXECUTABLE }} -m pip install -r test/python/cuda/ort/requirements.txt --user && \
               ${{ env.PYTHON_EXECUTABLE }} -m pip install /ort_genai_src/build/cuda/wheel/onnxruntime_genai*manylinux*.whl --no-deps --user && \
-              ${{ env.PYTHON_EXECUTABLE }} test/python/test_onnxruntime_genai.py --cwd test/python --test_models test/test_models --e2e"
+              ${{ env.PYTHON_EXECUTABLE }} test/python/test_onnxruntime_genai.py --cwd test/python --test_models test/models --e2e"
 
       # TODO: Enable this by adding dotnet to the docker image
       # - name: Build the C# API and Run the C# Tests

.github/workflows/win-cpu-arm64-build.yml

@@ -114,7 +114,7 @@ jobs:
 
       - name: Run the Python Tests
         run: |
-          python test/python/test_onnxruntime_genai.py --cwd "test\python" --test_models "test\test_models"
+          python test/python/test_onnxruntime_genai.py --cwd "test\python" --test_models "test\models"
 
       - name: Build the C# API and Run the C# Tests
         run: |
@@ -130,8 +130,8 @@ jobs:
 
       - name: Test the C# LLM Example with Tool Calling
         run: |
-          python3 test\python\special_tokens.py -p test\test_models\qwen-2.5-0.5b\int4\cpu\tokenizer.json -s "<tool_call>" -e "</tool_call>"
-          .\examples\csharp\ModelChat\bin\Release\net8.0\ModelChat.exe -m test\test_models\qwen-2.5-0.5b\int4\cpu\ -e cpu --response_format lark_grammar --tools_file test\test_models\tool-definitions\weather.json --tool_call_start "<tool_call>" --tool_call_end "</tool_call>" --user_prompt "What is the weather in Redmond, WA?" --tool_output --non_interactive --verbose
+          python3 test\python\special_tokens.py -p test\models\qwen-2.5-0.5b\int4\cpu\tokenizer.json -s "<tool_call>" -e "</tool_call>"
+          .\examples\csharp\ModelChat\bin\Release\net8.0\ModelChat.exe -m test\models\qwen-2.5-0.5b\int4\cpu\ -e cpu --response_format lark_grammar --tools_file test\models\tool-definitions\weather.json --tool_call_start "<tool_call>" --tool_call_end "</tool_call>" --user_prompt "What is the weather in Redmond, WA?" --tool_output --non_interactive --verbose
 
       - name: Verify Build Artifacts
         if: always()

.github/workflows/win-cpu-x64-build.yml

@@ -125,7 +125,7 @@ jobs:
 
       - name: Run the Python Tests
         run: |
-          python test/python/test_onnxruntime_genai.py --cwd "test\python" --test_models "test\test_models"
+          python test/python/test_onnxruntime_genai.py --cwd "test\python" --test_models "test\models"
 
       - name: Build the C# API and Run the C# Tests
         run: |
@@ -141,8 +141,8 @@ jobs:
 
       - name: Test the C# LLM Example with Tool Calling
         run: |
-          python3 test\python\special_tokens.py -p test\test_models\qwen-2.5-0.5b\int4\cpu\tokenizer.json -s "<tool_call>" -e "</tool_call>"
-          .\examples\csharp\ModelChat\bin\Release\net8.0\ModelChat.exe -m test\test_models\qwen-2.5-0.5b\int4\cpu\ -e cpu --response_format lark_grammar --tools_file test\test_models\tool-definitions\weather.json --tool_call_start "<tool_call>" --tool_call_end "</tool_call>" --user_prompt "What is the weather in Redmond, WA?" --tool_output --non_interactive --verbose
+          python3 test\python\special_tokens.py -p test\models\qwen-2.5-0.5b\int4\cpu\tokenizer.json -s "<tool_call>" -e "</tool_call>"
+          .\examples\csharp\ModelChat\bin\Release\net8.0\ModelChat.exe -m test\models\qwen-2.5-0.5b\int4\cpu\ -e cpu --response_format lark_grammar --tools_file test\models\tool-definitions\weather.json --tool_call_start "<tool_call>" --tool_call_end "</tool_call>" --user_prompt "What is the weather in Redmond, WA?" --tool_output --non_interactive --verbose
 
       - name: Verify Build Artifacts
         if: always()

.github/workflows/win-cuda-x64-build.yml

@@ -158,7 +158,7 @@ jobs:
 
       - name: Run the Python Tests
         run: |
-          python test/python/test_onnxruntime_genai.py --cwd "test\python" --test_models "test\test_models" --e2e
+          python test/python/test_onnxruntime_genai.py --cwd "test\python" --test_models "test\models" --e2e
 
       - name: Verify Build Artifacts
         if: always()
@@ -182,8 +182,8 @@ jobs:
 
       - name: Test the C# LLM Example with Tool Calling
         run: |
-          python test\python\special_tokens.py -p test\test_models\qwen-2.5-0.5b\int4\cpu\tokenizer.json -s "<tool_call>" -e "</tool_call>"
-          .\examples\csharp\ModelChat\bin\Release\net8.0\ModelChat.exe -m test\test_models\qwen-2.5-0.5b\int4\cpu\ -e cpu --response_format lark_grammar --tools_file test\test_models\tool-definitions\weather.json --tool_call_start "<tool_call>" --tool_call_end "</tool_call>" --user_prompt "What is the weather in Redmond, WA?" --tool_output --non_interactive --verbose
+          python test\python\special_tokens.py -p test\models\qwen-2.5-0.5b\int4\cpu\tokenizer.json -s "<tool_call>" -e "</tool_call>"
+          .\examples\csharp\ModelChat\bin\Release\net8.0\ModelChat.exe -m test\models\qwen-2.5-0.5b\int4\cpu\ -e cpu --response_format lark_grammar --tools_file test\models\tool-definitions\weather.json --tool_call_start "<tool_call>" --tool_call_end "</tool_call>" --user_prompt "What is the weather in Redmond, WA?" --tool_output --non_interactive --verbose
 
       - name: Prepend CUDA to PATH and Run tests
         run: |-

.github/workflows/win-directml-x64-build.yml

@@ -129,7 +129,7 @@ jobs:
 
       - name: Run the Python Tests
         run: |
-          python test/python/test_onnxruntime_genai.py --cwd "test\python" --test_models "test\test_models" --e2e
+          python test/python/test_onnxruntime_genai.py --cwd "test\python" --test_models "test\models" --e2e
 
       - name: Verify Build Artifacts
         if: always()

.github/workflows/win-webgpu-x64-build.yml

@@ -103,7 +103,7 @@ jobs:
 
       - name: Run the Python Tests
         run: |
-          python test/python/test_onnxruntime_genai.py --cwd "test\python" --test_models "test\test_models" --e2e
+          python test/python/test_onnxruntime_genai.py --cwd "test\python" --test_models "test\models" --e2e
 
       - name: Verify Build Artifacts
         if: always()

.gitignore

@@ -10,7 +10,7 @@ src/csharp/AssemblyInfo.cs
 /ort
 /build
 /build_*
-/test/test_models/*
+/test/models/*
 /cache_models
 /onnxruntime-linux-x64-*
 *.csv
@@ -29,24 +29,17 @@ examples/python/genai_models
 examples/python/hf_cache
 examples/csharp/ModelChat/models
 
-!test/test_models/hf-internal-testing/
-!test/test_models/hf-internal-testing/tiny-random-gpt2*/*.onnx
-!test/test_models/hf-internal-testing/tiny-qwen35-cuda/*.onnx
-!test/test_models/grammars/
-!test/test_models/qwen-vision-preprocessing/
-!test/test_models/qwen-vision-preprocessing/*.onnx
-!test/test_models/qwen3-vl-vision-preprocessing/
-!test/test_models/qwen3-vl-vision-preprocessing/*.onnx
-!test/test_models/qwen35-hybrid-preprocessing/
-!test/test_models/qwen35-hybrid-preprocessing/*.onnx
-!test/test_models/mistral3-vision-preprocessing/
-!test/test_models/gemma4-vision-preprocessing/
-!test/test_models/gemma4-vision-preprocessing/*.onnx
-!test/test_models/multimodal-decoder-no-input-ids/
-!test/test_models/multimodal-decoder-no-input-ids/*
-!test/test_models/multimodal-decoder-with-input-ids/
-!test/test_models/multimodal-decoder-with-input-ids/*
-!test/test_models/create_decoder_input_ids_test_models.py
+!test/models/gemma4/*
+!test/models/hf-internal-testing/*
+!test/models/mistral3/*
+!test/models/multimodal-decoder-no-input-ids/*
+!test/models/multimodal-decoder-with-input-ids/*
+!test/models/phi3-v/*
+!test/models/pipeline-model/*.json
+!test/models/qwen2-5-vl/*
+!test/models/qwen3-5/*
+!test/models/qwen3-vl/*
+!test/models/whisper/*
 
 .ipynb_checkpoints/
 /src/java/.gradle

.pipelines/mac-cpu-arm64-build.yml

@@ -208,7 +208,7 @@ jobs:
       source genai-macos-venv/bin/activate
       export ORTGENAI_LOG_ORT_LIB=1
       python3 -m pip install requests
-      python3 test/python/test_onnxruntime_genai.py --cwd test/python --test_models test/test_models
+      python3 test/python/test_onnxruntime_genai.py --cwd test/python --test_models test/models
     displayName: 'Run the Python tests'
     workingDirectory: '$(Build.SourcesDirectory)'
 
@@ -233,8 +233,8 @@ jobs:
   - bash: |
       set -e -x
       export ORTGENAI_LOG_ORT_LIB=1
-      python3 test/python/special_tokens.py -p test/test_models/qwen-2.5-0.5b/int4/cpu/tokenizer.json -s "<tool_call>" -e "</tool_call>"
-      ./examples/csharp/ModelChat/bin/Release/net8.0/ModelChat -m test/test_models/qwen-2.5-0.5b/int4/cpu/ -e cpu --response_format lark_grammar --tools_file test/test_models/tool-definitions/weather.json --tool_call_start "<tool_call>" --tool_call_end "</tool_call>" --user_prompt "What is the weather in Redmond, WA?" --tool_output --non_interactive --verbose
+      python3 test/python/special_tokens.py -p test/models/qwen-2.5-0.5b/int4/cpu/tokenizer.json -s "<tool_call>" -e "</tool_call>"
+      ./examples/csharp/ModelChat/bin/Release/net8.0/ModelChat -m test/models/qwen-2.5-0.5b/int4/cpu/ -e cpu --response_format lark_grammar --tools_file test/models/tool-definitions/weather.json --tool_call_start "<tool_call>" --tool_call_end "</tool_call>" --user_prompt "What is the weather in Redmond, WA?" --tool_output --non_interactive --verbose
     displayName: 'Test the C# LLM example with tool calling'
     workingDirectory: '$(Build.SourcesDirectory)'
 

examples/c/src/common.cpp

@@ -195,7 +195,7 @@ bool ParseArgs(
   app.add_option("-p,--top_p", generator_params_args.top_p, "Top p probability to sample with")->group(generator_params);
 
   app.add_option("--response_format", guidance_args.response_format, "Provide response format for the model")->group(guidance);
-  app.add_option("--tools_file", guidance_args.tools_file, "Path to file containing list of OpenAI-compatible tool definitions. Ex: test/test_models/tool-definitions/weather.json")->group(guidance);
+  app.add_option("--tools_file", guidance_args.tools_file, "Path to file containing list of OpenAI-compatible tool definitions. Ex: test/models/tool-definitions/weather.json")->group(guidance);
   app.add_flag("--text_output", guidance_args.text_output, "Produce a text response in the output")->group(guidance);
   app.add_flag("--tool_output", guidance_args.tool_output, "Produce a tool call in the output")->group(guidance);
   app.add_option("--tool_call_start", guidance_args.tool_call_start, "String representation of tool call start (ex: <|tool_call|>). Needs to be marked as special in tokenizer.json for guidance to work.")->group(guidance);

examples/csharp/Common/Common.cs

@@ -878,7 +878,7 @@ public static void GetGuidanceArgs(RootCommand parser)
             {
                 Arity = ArgumentArity.ExactlyOne,
                 DefaultValueFactory = (_) => "",
-                Description = "Path to file containing list of OpenAI-compatible tool definitions. Ex: test/test_models/tool-definitions/weather.json"
+                Description = "Path to file containing list of OpenAI-compatible tool definitions. Ex: test/models/tool-definitions/weather.json"
             };
             tools_file.Validators.Add(result =>
             {

examples/python/common.py

@@ -660,7 +660,7 @@ def get_guidance_args(parser: argparse.ArgumentParser) -> None:
         "--tools_file",
         type=str,
         default="",
-        help="Path to file containing list of OpenAI-compatible tool definitions. Ex: test/test_models/tool-definitions/weather.json",
+        help="Path to file containing list of OpenAI-compatible tool definitions. Ex: test/models/tool-definitions/weather.json",
     )
     guidance.add_argument(
         "-text", "--text_output", action="store_true", default=False, help="Produce a text response in the output"

src/java/CMakeLists.txt

@@ -196,7 +196,7 @@ if (ANDROID)
   # Copy the test model to the assets folder in the test app
   add_custom_command(TARGET onnxruntime-genai-jni POST_BUILD
                      COMMAND ${CMAKE_COMMAND} -E copy_directory_if_different
-                       ${REPO_ROOT}/test/test_models/hf-internal-testing/tiny-random-gpt2-fp32
+                       ${REPO_ROOT}/test/models/hf-internal-testing/tiny-random-gpt2-fp32
                        ${ANDROID_TEST_PACKAGE_APP_ASSETS_DIR}/model)
 
   # Copy the Android AAR package we built to the libs folder of our test app

src/java/src/test/android/README.md

@@ -6,7 +6,7 @@ This directory contains a simple android application for testing the ONNX Runtim
 
 This android application is mainly aimed for testing:
 
-- Model used: test/test_models/hf-internal-testing/tiny-random-gpt2-fp32
+- Model used: test/models/hf-internal-testing/tiny-random-gpt2-fp32
 - Main test file: An android instrumentation test under `app\src\androidtest\java\ai.onnxruntime.genai.example.javavalidator\SimpleTest.kt`
 - The main dependency of this application is `onnxruntime-genai` aar package under `app\libs`.
 - The onnxruntime dependency is provided by the latest released onnxruntime-android package.

src/java/src/test/java/ai/onnxruntime/genai/GenerationTest.java

@@ -24,7 +24,7 @@ public class GenerationTest {
   // phi-2 can be used in full end-to-end testing but needs to be manually downloaded.
   // it's also used this way in the C# unit tests.
   private static final String phi2ModelPath() {
-    return TestUtils.getTestResourcePath("phi-2/int4/cpu");
+    return TestUtils.getTestModelPath("phi-2/int4/cpu");
   }
 
   @SuppressWarnings("unused") // Used in EnabledIf

src/java/src/test/java/ai/onnxruntime/genai/MultiModalProcessorTest.java

@@ -23,9 +23,7 @@ public void testBatchEncodeDecode() throws GenAIException {
           new String(
               "<|user|>\n<|image_1|>\n Can you convert the table to markdown format?\n<|end|>\n<|assistant|>\n");
       try (Images image =
-              new Images(
-                  TestUtils.getFilePathFromDisk(
-                      TestUtils.getTestResourcePath("images/sheet.png")));
+              new Images(TestUtils.getFilePathFromDisk(TestUtils.getTestImagePath("sheet.png")));
           NamedTensors processed = multiModalProcessor.processImages(inputs, image); ) {
         assertNotNull(processed);
       }