Fix phi4mini test model

Author: jackzhxng

.ci/scripts/gather_test_models.py

@@ -90,7 +90,7 @@ def model_should_run_on_event(model: str, event: str) -> bool:
     We put higher priority and fast models to pull request and rest to push.
     """
     if event == "pull_request":
-        return model in ["mv3", "vit"]
+        return model in ["mv3", "vit", "phi-4-mini"]
     elif event == "push":
         # These are super slow. Only run it periodically
         return model not in ["dl3", "edsr", "emformer_predict"]

.ci/scripts/test_model.sh

@@ -100,7 +100,7 @@ test_model() {
       rm "./${MODEL_NAME}.pte"
       return  # Skip running with portable executor runnner since portable doesn't support Qwen's biased linears.
   fi
-  if [[ "${MODEL_NAME}" == "phi4_mini" ]]; then
+  if [[ "${MODEL_NAME}" == "phi-4-mini" ]]; then
       # Install requirements for export_llama
       bash examples/models/llama/install_requirements.sh
       # Test export_llama script: python3 -m examples.models.llama.export_llama.

examples/models/__init__.py

@@ -35,7 +35,7 @@
     "llava": ("llava", "LlavaModel"),
     "efficient_sam": ("efficient_sam", "EfficientSAM"),
     "qwen2_5": ("qwen2_5", "Qwen2_5Model"),
-    "phi4_mini": ("phi4_mini", "Phi4MiniModel"),
+    "phi-4-mini": ("phi-4-mini", "Phi4MiniModel"),
 }
 
 __all__ = [

examples/models/llama/export_llama_lib.py

@@ -93,7 +93,7 @@
     "llama3_2",
     "static_llama",
     "qwen2_5",
-    "phi4_mini",
+    "phi-4-mini",
 ]
 TORCHTUNE_DEFINED_MODELS = ["llama3_2_vision"]