Enable quantization for deeplabv3 (#398)

Summary:
Pull Request resolved: #398

att, it's still slow in my devvm, but might be fine in CI

Reviewed By: kimishpatel

Differential Revision: D49379414

fbshipit-source-id: 78a18822ab1c36f743337df26f4883c3edcf6018

Author: jerryzh168

examples/quantization/example.py

@@ -70,15 +70,21 @@ def verify_xnnpack_quantizer_matching_fx_quant_model(model_name, model, example_
     after_quant_result_fx = m_fx(*example_inputs)
 
     # 3. compare results
-    # NB: this check is more useful for QAT since for PTQ we are only inserting observers that does not change the
-    # output of a model, so it's just testing the numerical difference for different captures in PTQ
-    # for QAT it is also testing whether the fake quant placement match or not
-    # not exactly the same due to capture changing numerics, but still really close
+    if model_name == "dl3":
+        after_prepare_result = after_prepare_result["out"]
+        after_prepare_result_fx = after_prepare_result_fx["out"]
+        after_quant_result = after_quant_result["out"]
+        after_quant_result_fx = after_quant_result_fx["out"]
     logging.info(f"m: {m}")
     logging.info(f"m_fx: {m_fx}")
     logging.info(
         f"prepare sqnr: {compute_sqnr(after_prepare_result, after_prepare_result_fx)}"
     )
+
+    # NB: this check is more useful for QAT since for PTQ we are only inserting observers that does not change the
+    # output of a model, so it's just testing the numerical difference for different captures in PTQ
+    # for QAT it is also testing whether the fake quant placement match or not
+    # not exactly the same due to capture changing numerics, but still really close
     assert compute_sqnr(after_prepare_result, after_prepare_result_fx) > 100
     logging.info(
         f"quant diff max: {torch.max(after_quant_result - after_quant_result_fx)}"

examples/recipes/xnnpack_optimization/models.py

@@ -23,4 +23,5 @@ class OptimizationOptions(object):
         True, False
     ),  # TODO[T163161310]: takes a long time to export to exec prog and save inception_v4 quantized model
     "w2l": OptimizationOptions(False, True),
+    "dl3": OptimizationOptions(True, False),
 }