update headings and print statements

spurious_correlations_datalab/detecting_spurious_correlations.ipynb

@@ -22,6 +22,8 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
+    "## Setup and import data\n",
+    "\n",
     "Please install the dependencies specified in [requirements.txt](https://github.com/cleanlab/examples/blob/master/spurious_correlations_datalab/requirements.txt) file before running the notebook.\n",
     "To fetch the data for this tutorial, make sure you have `wget` and `zip` installed."
    ]
@@ -273,7 +275,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "#### Training on dataset that has Spurious Correlations with the class labels"
+    "#### Training on dataset with spurious correlations between image darkness and class labels"
    ]
   },
   {
@@ -333,7 +335,7 @@
    "source": [
     "# Finding accuracy on the test dataset with no spurious correlations\n",
     "acc = accuracy(best_net, testloader)\n",
-    "print(f\"Accuracy on the test dataset where the training images were darkened for 'chicken_wings class' = {acc:.2f}\")"
+    "print(f\"Accuracy on the test dataset where the training images were darkened for 'chicken_wings class' = {acc:.2f}%\")"
    ]
   },
   {
@@ -384,7 +386,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "#### Training on dataset that has no Spurious Correlations with the class labels"
+    "#### Training on dataset without spurious correlations between image darkness and class labels"
    ]
   },
   {
@@ -448,7 +450,7 @@
     "# Finding accuracy when trained on a training datset with no spurious correlations\n",
     "\n",
     "acc = accuracy(best_net_no_correlations, testloader)\n",
-    "print(f\"Accuracy on the test dataset with no spurious correlations = {acc:.2f}\")"
+    "print(f\"Accuracy on the test dataset with no spurious correlations = {acc:.2f}%\")"
    ]
   },
   {