From a0a54f23b9b26f3ba5cd3412eb4c933c9b17c29a Mon Sep 17 00:00:00 2001
From: Eric Zelikman <ezelikman@live.com>
Date: Thu, 10 Dec 2020 12:47:59 -0500
Subject: [PATCH] moved images to github

---
 C3W2_SRGAN_(Optional).ipynb | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/C3W2_SRGAN_(Optional).ipynb b/C3W2_SRGAN_(Optional).ipynb
index c29dc84..7b38ddd 100644
--- a/C3W2_SRGAN_(Optional).ipynb
+++ b/C3W2_SRGAN_(Optional).ipynb
@@ -393,7 +393,7 @@
         "\n",
         "Don't worry if this is confusing! The algorithm is conveniently implemented as `torch.nn.PixelShuffle` in PyTorch, so as long as you have a general idea of how this works, you're set.\n",
         "\n",
-        "> ![Efficient Sub-pixel CNN](https://drive.google.com/uc?export=view&id=136LMcywv1r5W1f9-L-55bJ4AKAHlH8Zr)\n",
+        "> ![Efficient Sub-pixel CNN](https://github.com/https-deeplearning-ai/GANs-Public/blob/master/SRGAN-PixelShuffle.png?raw=true)\n",
         "*Efficient sub-pixel CNN, taken from Figure 1 of [Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network](https://arxiv.org/abs/1609.05158) (Shi et al. 2016). The PixelShuffle operation (also known as sub-pixel convolution) is shown as the last step on the right.*"
       ]
     },
@@ -418,7 +418,7 @@
         "\n",
         "The super-resolution residual network (SRResNet) and the generator are the same thing. The generator network architecture is actually quite simple - just a bunch of convolutional layers, residual blocks, and pixel shuffling layers!\n",
         "\n",
-        "> ![SRGAN Generator](https://drive.google.com/uc?export=view&id=1wY5YmoYTBzuhWLlTAYCI92xWgkf_uINE)\n",
+        "> ![SRGAN Generator](https://github.com/https-deeplearning-ai/GANs-Public/blob/master/SRGAN-Generator.png?raw=true)\n",
         "*SRGAN Generator, taken from Figure 4 of [Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network](https://arxiv.org/abs/1609.04802) (Ledig et al. 2017).*"
       ]
     },
@@ -492,7 +492,7 @@
         "\n",
         "The discriminator architecture is also relatively straightforward, just one big sequential model - see the diagram below for reference!\n",
         "\n",
-        "![SRGAN Generator](https://drive.google.com/uc?export=view&id=1fcfTrXBcODoZa2JSEO8OiMUEo5WHdkef)\n",
+        "![SRGAN Generator](https://github.com/https-deeplearning-ai/GANs-Public/blob/master/SRGAN-Discriminator.png?raw=true)\n",
         "*SRGAN Discriminator, taken from Figure 4 of [Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network](https://arxiv.org/abs/1609.04802) (Ledig et al. 2017).*"
       ]
     },
@@ -894,7 +894,6 @@
       "cell_type": "code",
       "metadata": {
         "id": "ScH0Iok8fAMS",
-        "outputId": "388ba277-76ce-427b-b7b1-57ab74e9b1dd",
         "colab": {
           "base_uri": "https://localhost:8080/",
           "height": 84,
@@ -908,7 +907,8 @@
             "0c356917abbf4afb8ef89e5e48930635",
             "c0209daeab5f4a09b01a23370e09d352"
           ]
-        }
+        },
+        "outputId": "388ba277-76ce-427b-b7b1-57ab74e9b1dd"
       },
       "source": [
         "device = 'cuda' if torch.cuda.is_available() else 'cpu'\n",
@@ -968,11 +968,11 @@
       "cell_type": "code",
       "metadata": {
         "id": "AvqBuRwjZrBq",
-        "outputId": "96ff8cef-0371-4bb3-d4ff-b1c1aee180b6",
         "colab": {
           "base_uri": "https://localhost:8080/",
           "height": 1000
-        }
+        },
+        "outputId": "96ff8cef-0371-4bb3-d4ff-b1c1aee180b6"
       },
       "source": [
         "generator = torch.load('srresnet.pt')\n",