update NS for FX tutorial for PyTorch v1.13 (#2089)

* update NS for FX tutorial for PyTorch v1.13

Summary:

Makes a couple of updates to ensure this tutorial still runs on 1.13:
1. changes the `qconfig_dict` argument of `prepare_fx` to `qconfig_mapping`
2. adds `example_inputs` to `prepare_fx`

Test plan:

Run the tutorial, it runs without errors on master

* Pin importlib_metadata<5.0 for python <= 3.7 in requirements.txt (#2091)

v5.0.0 causes issues with python 3.7: python/importlib_metadata#411

* Enable the FX tutorial

Co-authored-by: Max Balandat <Balandat@users.noreply.github.com>
Co-authored-by: Svetlana Karslioglu <svekars@fb.com>

.jenkins/validate_tutorials_built.py

@@ -50,7 +50,6 @@
     "recipes/Captum_Recipe",
     "hyperparameter_tuning_tutorial",
     "flask_rest_api_tutorial",
-    "fx_numeric_suite_tutorial", # remove when https://github.com/pytorch/tutorials/pull/2089 is fixed
     "ax_multiobjective_nas_tutorial",
 ]
 

prototype_source/fx_numeric_suite_tutorial.py

@@ -84,9 +84,9 @@ def plot(xdata, ydata, xlabel, ylabel, title):
 # Note: quantization APIs are inplace, so we save a copy of the float model for
 # later comparison to the quantized model. This is done throughout the
 # tutorial.
-mobilenetv2_prepared = quantize_fx.prepare_fx(
-    copy.deepcopy(mobilenetv2_float), qconfig_dict)
 datum = torch.randn(1, 3, 224, 224)
+mobilenetv2_prepared = quantize_fx.prepare_fx(
+    copy.deepcopy(mobilenetv2_float), qconfig_dict, (datum,))
 mobilenetv2_prepared(datum)
 # Note: there is a long standing issue that we cannot copy.deepcopy a
 # quantized model. Since quantization APIs are inplace and we need to use