remove tf references

Author: vyinkabanjo

lab2/solutions/PT_Part2_Debiasing_Solution.ipynb

@@ -343,7 +343,7 @@
         "\n",
         "### Define and train the CNN model\n",
         "\n",
-        "Like we did in the first part of the lab, we'll define our CNN model, and then train on the CelebA and ImageNet datasets using the `tf.GradientTape` class and the `tf.GradientTape.gradient` method."
+        "Like we did in the first part of the lab, we'll define our CNN model, and then train on the CelebA and ImageNet datasets by leveraging PyTorch's automatic differentiation (`torch.autograd`) by using the `loss.backward()` and `optimizer.step()` functions."
       ]
     },
     {
@@ -870,10 +870,10 @@
         "\n",
         "    # TODO: Define the reconstruction loss as the mean absolute pixel-wise\n",
         "    # difference between the input and reconstruction. Hint: you'll need to\n",
-        "    # use tf.reduce_mean, and supply an axis argument which specifies which\n",
-        "    # dimensions to reduce over. For example, reconstruction loss needs to average\n",
+        "    # use torch.mean, and specify the dimensions to reduce over. \n",
+        "    # For example, reconstruction loss needs to average\n",
         "    # over the height, width, and channel image dimensions.\n",
-        "    # https://www.tensorflow.org/api_docs/python/tf/math/reduce_mean\n",
+        "    # https://pytorch.org/docs/stable/generated/torch.mean.html\n",
         "    reconstruction_loss = torch.mean(torch.abs(x - x_recon), dim=(1, 2, 3))\n",
         "\n",
         "    # TODO: Define the VAE loss. Note this is given in the equation for L_{VAE}\n",
@@ -928,17 +928,15 @@
         "\n",
         "\n",
         "def sampling(z_mean, z_logsigma):\n",
-        "    # By default, random.normal is \"standard\" (ie. mean=0 and std=1.0)\n",
-        "    # batch, latent_dim = z_mean.shape\n",
-        "    # epsilon = tf.random.normal(shape=(batch, latent_dim))\n",
+        "    # By default, randn_like is \"standard\" (ie. mean=0 and std=1.0)\n",
+        "    eps = torch.randn_like(z_mean)\n",
         "\n",
         "    # # TODO: Define the reparameterization computation!\n",
         "    # # Note the equation is given in the text block immediately above.\n",
+        "    \n",
+        "    z = z_mean + torch.exp(z_logsigma) * eps\n",
         "    # z = # TODO\n",
-        "    # return z\n",
-        "\n",
-        "    eps = torch.randn_like(z_mean)\n",
-        "    return z_mean + torch.exp(z_logsigma) * eps"
+        "    return z"
       ]
     },
     {
@@ -1025,8 +1023,8 @@
         "    vae_loss = vae_loss_function(x, x_pred, mu, logsigma)\n",
         "    # vae_loss = vae_loss_function('''TODO''') # TODO\n",
         "\n",
-        "    # TODO: define the classification loss using sigmoid_cross_entropy\n",
-        "    # https://www.tensorflow.org/api_docs/python/tf/nn/sigmoid_cross_entropy_with_logits\n",
+        "    # TODO: define the classification loss using binary_cross_entropy\n",
+        "    # https://pytorch.org/docs/stable/generated/torch.nn.functional.binary_cross_entropy_with_logits.html\n",
         "    # classification_loss = # TODO\n",
         "    classification_loss = F.binary_cross_entropy_with_logits(\n",
         "        y_logit, y, reduction=\"none\"\n",
@@ -1037,7 +1035,7 @@
         "    y = y.float()\n",
         "    face_indicator = (y == 1.0).float()\n",
         "\n",
-        "    # TODO: define the DB-VAE total loss! Use tf.reduce_mean to average over all\n",
+        "    # TODO: define the DB-VAE total loss! Use torch.mean to average over all\n",
         "    # samples\n",
         "    # total_loss = # TODO\n",
         "    total_loss = torch.mean(classification_loss * face_indicator + vae_loss)\n",