Merge branch 'main' into main

Author: svekars

_static/img/cat_resized.jpg

@@ -107,7 +107,7 @@ def _initialize_weights(self):
 
 # Load pretrained model weights
 model_url = 'https://s3.amazonaws.com/pytorch/test_data/export/superres_epoch100-44c6958e.pth'
-batch_size = 1    # just a random number
+batch_size = 64    # just a random number
 
 # Initialize model with the pretrained weights
 map_location = lambda storage, loc: storage
@@ -218,6 +218,32 @@ def to_numpy(tensor):
 # ONNX exporter, so please contact us in that case.
 #
 
+######################################################################
+# Timing Comparison Between Models
+# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+#
+
+######################################################################
+# Since ONNX models optimize for inference speed, running the same
+# data on an ONNX model instead of a native pytorch model should result in an 
+# improvement of up to 2x. Improvement is more pronounced with higher batch sizes.
+
+
+import time
+
+x = torch.randn(batch_size, 1, 224, 224, requires_grad=True)
+
+start = time.time()
+torch_out = torch_model(x)
+end = time.time()
+print(f"Inference of Pytorch model used {end - start} seconds")
+
+ort_inputs = {ort_session.get_inputs()[0].name: to_numpy(x)}
+start = time.time()
+ort_outs = ort_session.run(None, ort_inputs)
+end = time.time()
+print(f"Inference of ONNX model used {end - start} seconds")
+
 
 ######################################################################
 # Running the model on an image using ONNX Runtime
@@ -301,10 +327,20 @@ def to_numpy(tensor):
 # Save the image, we will compare this with the output image from mobile device
 final_img.save("./_static/img/cat_superres_with_ort.jpg")
 
+# Save resized original image (without super-resolution)
+img = transforms.Resize([img_out_y.size[0], img_out_y.size[1]])(img)
+img.save("cat_resized.jpg")
 
 ######################################################################
+# Here is the comparison between the two images: 
+#
+# .. figure:: /_static/img/cat_resized.jpg
+#
+# Low-resolution image
+# 
 # .. figure:: /_static/img/cat_superres_with_ort.jpg
-#    :alt: output\_cat
+#
+# Image after super-resolution
 #
 #
 # ONNX Runtime being a cross platform engine, you can run it across
@@ -313,7 +349,7 @@ def to_numpy(tensor):
 # ONNX Runtime can also be deployed to the cloud for model inferencing
 # using Azure Machine Learning Services. More information `here <https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-onnx>`__.
 #
-# More information about ONNX Runtime's performance `here <https://github.com/microsoft/onnxruntime#high-performance>`__.
+# More information about ONNX Runtime's performance `here <https://onnxruntime.ai/docs/performance>`__.
 #
 #
 # For more information about ONNX Runtime `here <https://github.com/microsoft/onnxruntime>`__.

advanced_source/super_resolution_with_onnxruntime.py

@@ -228,18 +228,7 @@
 # integer with the ``.to()`` method. Note that ``c`` contains all the same
 # values as ``b``, but truncated to integers.
 # 
-# Available data types include:
-# 
-# -  ``torch.bool``
-# -  ``torch.int8``
-# -  ``torch.uint8``
-# -  ``torch.int16``
-# -  ``torch.int32``
-# -  ``torch.int64``
-# -  ``torch.half``
-# -  ``torch.float``
-# -  ``torch.double``
-# -  ``torch.bfloat``
+# For more information, see the `data types documentation <https://pytorch.org/docs/stable/tensor_attributes.html#torch.dtype>`__.
 # 
 # Math & Logic with PyTorch Tensors
 # ---------------------------------

beginner_source/introyt/tensors_deeper_tutorial.py

@@ -9,6 +9,8 @@
 This tutorial shows how to use PyTorch to train a Deep Q Learning (DQN) agent
 on the CartPole-v1 task from `Gymnasium <https://gymnasium.farama.org>`__.
 
+You might find it helpful to read the original `Deep Q Learning (DQN) <https://arxiv.org/abs/1312.5602>`__ paper
+
 **Task**
 
 The agent has to decide between two actions - moving the cart left or
@@ -83,7 +85,11 @@
 plt.ion()
 
 # if GPU is to be used
-device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+device = torch.device(
+    "cuda" if torch.cuda.is_available() else
+    "mps" if torch.backends.mps.is_available() else
+    "cpu"
+)
 
 
 ######################################################################
@@ -397,7 +403,7 @@ def optimize_model():
 # can produce better results if convergence is not observed.
 #
 
-if torch.cuda.is_available():
+if torch.cuda.is_available() or torch.backends.mps.is_available():
     num_episodes = 600
 else:
     num_episodes = 50