Merge branch 'main' into add_coding_ddpg

Author: nairbv

intermediate_source/seq2seq_translation_tutorial.py

@@ -45,7 +45,7 @@
    :alt:
 
 To improve upon this model we'll use an `attention
-mechanism <https://arxiv.org/abs/1409.0473>`__, which lets the decoder
+mechanism <https://arxiv.org/abs/1508.04025>`__, which lets the decoder
 learn to focus over a specific range of the input sequence.
 
 **Recommended Reading:**
@@ -66,8 +66,8 @@
    Statistical Machine Translation <https://arxiv.org/abs/1406.1078>`__
 -  `Sequence to Sequence Learning with Neural
    Networks <https://arxiv.org/abs/1409.3215>`__
--  `Neural Machine Translation by Jointly Learning to Align and
-   Translate <https://arxiv.org/abs/1409.0473>`__
+-  `Effective Approaches to Attention-based Neural Machine
+   Translation <https://arxiv.org/abs/1508.04025>`__
 -  `A Neural Conversational Model <https://arxiv.org/abs/1506.05869>`__
 
 You will also find the previous tutorials on

intermediate_source/torch_compile_tutorial.py

@@ -105,7 +105,7 @@ def forward(self, x):
 #
 # Let's now demonstrate that using ``torch.compile`` can speed
 # up real models. We will compare standard eager mode and 
-# ``torch.compile`` by evaluating and training ResNet-18 on random data.
+# ``torch.compile`` by evaluating and training a ``torchvision`` model on random data.
 #
 # Before we start, we need to define some utility functions.
 

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.