java instrumentation test with MV2 (#2029)

Summary:
Pull Request resolved: #2029

A simple test to validate the E2E workflow for Java library (load jni, load module, execute)

Reviewed By: mcr229

Differential Revision: D53953813

fbshipit-source-id: 17f401da0b274eeaf66af72bd90bcb71139b1d73

Author: kirklandsign

examples/models/__init__.py

@@ -18,6 +18,7 @@
     "llama2": ("llama2", "Llama2Model"),
     "mobilebert": ("mobilebert", "MobileBertModelExample"),
     "mv2": ("mobilenet_v2", "MV2Model"),
+    "mv2_untrained": ("mobilenet_v2", "MV2UntrainedModel"),
     "mv3": ("mobilenet_v3", "MV3Model"),
     "vit": ("torchvision_vit", "TorchVisionViTModel"),
     "w2l": ("wav2letter", "Wav2LetterModel"),

examples/models/mobilenet_v2/__init__.py

@@ -4,8 +4,9 @@
 # This source code is licensed under the BSD-style license found in the
 # LICENSE file in the root directory of this source tree.
 
-from .model import MV2Model
+from .model import MV2Model, MV2UntrainedModel
 
 __all__ = [
     MV2Model,
+    MV2UntrainedModel,
 ]

examples/models/mobilenet_v2/model.py

@@ -29,3 +29,17 @@ def get_eager_model(self) -> torch.nn.Module:
     def get_example_inputs(self):
         tensor_size = (1, 3, 224, 224)
         return (torch.randn(tensor_size),)
+
+
+class MV2UntrainedModel(EagerModelBase):
+    def __init__(self):
+        pass
+
+    def get_eager_model(self) -> torch.nn.Module:
+        # pyre-ignore
+        mv2 = mobilenet_v2()
+        return mv2
+
+    def get_example_inputs(self):
+        tensor_size = (1, 3, 224, 224)
+        return (torch.randn(tensor_size),)

examples/xnnpack/aot_compiler.py

@@ -56,6 +56,7 @@
         required=False,
         help="Generate and save an ETRecord to the given file location",
     )
+    parser.add_argument("-o", "--output_dir", default=".", help="output directory")
 
     args = parser.parse_args()
 
@@ -110,4 +111,4 @@
 
     quant_tag = "q8" if args.quantize else "fp32"
     model_name = f"{args.model_name}_xnnpack_{quant_tag}"
-    save_pte_program(exec_prog.buffer, model_name)
+    save_pte_program(exec_prog.buffer, model_name, args.output_dir)