Merge pull request #136 from IamShubhamGupto/main

imported cnn and resnet

_toc.yml

@@ -32,6 +32,8 @@ parts:
     chapters:
       - file: notebooks/random_forest_parameterization.ipynb
       - file: notebooks/GP_lorenz96.ipynb
+      - file: notebooks/local_map_L96_and_Burgers.ipynb
+      - file: notebooks/L96_ResNet_RNN.ipynb
   - caption: End Matter
     chapters:
       - file: notebooks/closing_remarks.ipynb

notebooks/Improving_Neural_networks.ipynb

@@ -425,7 +425,9 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {},
+   "metadata": {
+    "user_expressions": []
+   },
    "source": [
     "Dropout means randomly deactivating or temporarily removing some units from a layer of the network while training, along with all its incoming and outgoing connections. See more details [here](http://jmlr.org/papers/v15/srivastava14a.html).\n",
     "It is usually the most useful regularization that we can do in fully connected layers.\n",
@@ -449,7 +451,9 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {},
+   "metadata": {
+    "user_expressions": []
+   },
    "source": [
     "In the network defined below, we add dropout to with a probability of 20% to each layer. This means that during each training step, random 20% of the units within each layer will be deactivated."
    ]
@@ -523,9 +527,6 @@
    "outputs": [],
    "source": [
     "plt.figure(dpi=150)\n",
-    "plt.plot(train_loss, \"b\", label=\"Training loss\")\n",
-    "plt.plot(val_loss, \"b--\", label=\"Validation loss\")\n",
-    "\n",
     "plt.plot(train_loss_drop, \"r\", label=\"Training loss, with dropout\")\n",
     "plt.plot(val_loss_drop, \"r--\", label=\"Validation loss, with dropout\")\n",
     "plt.legend();"
@@ -764,7 +765,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.9"
+   "version": "3.9.16"
   }
  },
  "nbformat": 4,