Merge branch 'main' into add_amx_doc

Author: CaoE

conf.py

@@ -34,6 +34,7 @@
 import pytorch_sphinx_theme
 import torch
 import glob
+import random
 import shutil
 from custom_directives import IncludeDirective, GalleryItemDirective, CustomGalleryItemDirective, CustomCalloutItemDirective, CustomCardItemDirective
 import distutils.file_util
@@ -85,6 +86,11 @@
 
 # -- Sphinx-gallery configuration --------------------------------------------
 
+def reset_seeds(gallery_conf, fname):
+    torch.manual_seed(42)
+    torch.set_default_device(None)
+    random.seed(10)
+
 sphinx_gallery_conf = {
     'examples_dirs': ['beginner_source', 'intermediate_source',
                       'advanced_source', 'recipes_source', 'prototype_source'],
@@ -94,7 +100,8 @@
     'backreferences_dir': None,
     'first_notebook_cell': ("# For tips on running notebooks in Google Colab, see\n"
                             "# https://pytorch.org/tutorials/beginner/colab\n"
-                            "%matplotlib inline")
+                            "%matplotlib inline"),
+    'reset_modules': (reset_seeds)
 }
 
 if os.getenv('GALLERY_PATTERN'):

prototype_source/fx_graph_mode_ptq_dynamic.py

@@ -239,9 +239,27 @@ def evaluate(model_, data_source):
     .set_object_type(nn.LSTM, default_dynamic_qconfig)
     .set_object_type(nn.Linear, default_dynamic_qconfig)
 )
-# Deepcopying the original model because quantization api changes the model inplace and we want
+# Load model to create the original model because quantization api changes the model inplace and we want
 # to keep the original model for future comparison
-model_to_quantize = copy.deepcopy(model)
+
+
+model_to_quantize = LSTMModel(
+    ntoken = ntokens,
+    ninp = 512,
+    nhid = 256,
+    nlayers = 5,
+)
+
+model_to_quantize.load_state_dict(
+    torch.load(
+        model_data_filepath + 'word_language_model_quantize.pth',
+        map_location=torch.device('cpu')
+        )
+    )
+
+model_to_quantize.eval()
+
+
 prepared_model = prepare_fx(model_to_quantize, qconfig_mapping, example_inputs)
 print("prepared model:", prepared_model)
 quantized_model = convert_fx(prepared_model)
@@ -289,4 +307,4 @@ def time_model_evaluation(model, test_data):
 # 3. Conclusion
 # -------------
 # This tutorial introduces the api for post training dynamic quantization in FX Graph Mode,
-# which dynamically quantizes the same modules as Eager Mode Quantization.
+# which dynamically quantizes the same modules as Eager Mode Quantization.

recipes_source/recipes/changing_default_device.py

@@ -43,9 +43,6 @@
 print(mod.weight.device)
 print(mod(torch.randn(128, 20)).device)
 
-# And then globally return it back to CPU
-torch.set_default_device('cpu')
-
 ################################################################
 # This function imposes a slight performance cost on every Python
 # call to the torch API (not just factory functions). If this