2024 day 1

lab2/Part2_FaceDetection.ipynb

@@ -85,10 +85,15 @@
         "import numpy as np\n",
         "from tqdm import tqdm\n",
         "\n",
-        "\n",
         "# Download and import the MIT Introduction to Deep Learning package\n",
         "!pip install mitdeeplearning\n",
-        "import mitdeeplearning as mdl"
+        "import mitdeeplearning as mdl\n",
+        "\n",
+        "# Import Comet\n",
+        "!pip install comet_ml\n",
+        "import comet_ml\n",
+        "comet_ml.init(project_name=\"6.s191lab2_part2_CNN\")\n",
+        "comet_model_1 = comet_ml.Experiment()"
       ]
     },
     {
@@ -179,7 +184,8 @@
         "\n",
         "plt.subplot(1, 2, 2)\n",
         "plt.imshow(not_face_images[idx_not_face])\n",
-        "plt.title(\"Not Face\"); plt.grid(False)"
+        "plt.title(\"Not Face\"); plt.grid(False)\n",
+        "comet_model_1.log_figure(figure=plt)"
       ]
     },
     {
@@ -300,9 +306,13 @@
         "    x, y = loader.get_batch(batch_size)\n",
         "    loss = standard_train_step(x, y)\n",
         "\n",
+        "    comet_model_1.log_metric(\"loss\", loss.numpy().mean(), idx)\n",
         "    # Record the loss and plot the evolution of the loss as a function of training\n",
         "    loss_history.append(loss.numpy().mean())\n",
-        "    plotter.plot(loss_history.get())"
+        "    plotter.plot(loss_history.get())\n",
+        "\n",
+        "comet_model_1.log_figure(figure=plt)\n",
+        "comet_model_1.end()"
       ]
     },
     {
@@ -756,6 +766,9 @@
       "source": [
         "### Training the SS-VAE ###\n",
         "\n",
+        "comet_ml.init(project_name=\"6.s191lab2_part2_VAE\")\n",
+        "comet_model_2 = comet_ml.Experiment()\n",
+        "\n",
         "# Hyperparameters\n",
         "batch_size = 32\n",
         "learning_rate = 5e-4\n",
@@ -806,6 +819,7 @@
         "    (x, y) = loader.get_batch(batch_size)\n",
         "    # loss optimization\n",
         "    loss = ss_vae_train_step(x, y)\n",
+        "    comet_model_2.log_metric(\"loss\", loss, step=j)\n",
         "    \n",
         "    # plot the progress every 200 steps\n",
         "    if j % 500 == 0: \n",
@@ -884,7 +898,9 @@
         "\n",
         "ax[1].imshow(mdl.util.create_grid_of_images(x[ind[-25:]]))\n",
         "ax[1].set_title(\"Samples with the highest reconstruction loss \\n\" + \n",
-        "                f\"Average recon loss: {np.mean(vae_loss[ind[-25:]]):.2f}\");"
+        "                f\"Average recon loss: {np.mean(vae_loss[ind[-25:]]):.2f}\");\n",
+        "\n",
+        "comet_model_2.log_figure()"
       ]
     },
     {
@@ -950,7 +966,9 @@
         "# Visualize all of the reconstructions!\n",
         "ax[1].imshow(mdl.util.create_grid_of_images(recons, (1, num_steps)))\n",
         "ax[1].set_xlabel(\"Latent step\")\n",
-        "ax[1].set_ylabel(\"Visualization\");\n"
+        "ax[1].set_ylabel(\"Visualization\");\n",
+        "\n",
+        "comet_model_2.log_figure()"
       ]
     },
     {
@@ -999,7 +1017,10 @@
         "# Plot the results\n",
         "plt.plot(np.linspace(np.min(z_mean), np.max(z_mean), num_steps+1), accuracy_per_latent,'-o')\n",
         "plt.xlabel(\"Latent step\")\n",
-        "plt.ylabel(\"Relative accuracy\")"
+        "plt.ylabel(\"Relative accuracy\")\n",
+        "\n",
+        "comet_model_2.log_figure()\n",
+        "comet_model_2.end()"
       ]
     },
     {