Add preprocess to ci

ghstack-source-id: 14fe9f1
Pull Request resolved: #6544

Author: lucylq

.github/workflows/pull.yml

@@ -231,6 +231,34 @@ jobs:
         # run e2e (export, tokenizer and runner)
         PYTHON_EXECUTABLE=python bash .ci/scripts/test_llava.sh
 
+  test-preprocess-linux:
+    name: test-preprocess-linux
+    uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
+    strategy:
+      fail-fast: false
+    with:
+      runner: linux.24xlarge
+      docker-image: executorch-ubuntu-22.04-clang12
+      submodules: 'true'
+      ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
+      timeout: 90
+      script: |
+        # The generic Linux job chooses to use base env, not the one setup by the image
+        CONDA_ENV=$(conda env list --json | jq -r ".envs | .[-1]")
+        conda activate "${CONDA_ENV}"
+
+        PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "cmake"
+
+        # install pybind
+        bash install_requirements.sh --pybind xnnpack
+
+        # install preprocess requirements
+        bash examples/models/llama3_2_vision/install_requirements.sh
+
+        # run python unittest
+        pytest -c /dev/null -v -n auto examples/models/llama3_2_vision/preprocess/
+
+
   test-quantized-aot-lib-linux:
     name: test-quantized-aot-lib-linux
     uses: pytorch/test-infra/.github/workflows/linux_job.yml@main

examples/models/llama3_2_vision/preprocess/model.py

@@ -26,6 +26,9 @@ class PreprocessConfig:
     max_num_tiles: int = 4
     tile_size: int = 224
     antialias: bool = False
+    # Used for reference eager model from torchtune.
+    resize_to_max_canvas: bool = False
+    possible_resolutions: Optional[List[Tuple[int, int]]] = None
 
 
 class CLIPImageTransformModel(EagerModelBase):