edit ground_truth_labeling_jobs/pretrained_model/pretrained_model_labeling_tutorial.ipynb

Author: EC2 Default User

ground_truth_labeling_jobs/pretrained_model/pretrained_model_labeling_tutorial.ipynb

@@ -17,7 +17,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# Introduction\n",
+    "## Introduction\n",
     "\n",
     "SageMaker Ground Truth is a fully managed service for labeling datasets for machine learning applications. Ground Truth allows you to start a labeling job with a pre-trained model, which is a great way to accelerate the labeling process. If you have a machine learning model that already encodes some domain knowledge about your dataset, you can use it to \"jump start\" Ground Truth's auto-labeling process. \n",
     "\n",
@@ -98,7 +98,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# Iteration #1: Create Initial Labeling Job\n",
+    "## Iteration #1: Create Initial Labeling Job\n",
     "\n",
     "## Setup\n",
     "\n",
@@ -752,7 +752,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# Iteration #2: Labeling Job with Pre-Trained Model <a class=\"anchor\" id=\"Iteration2\"></a>\n",
+    "## Iteration #2: Labeling Job with Pre-Trained Model <a class=\"anchor\" id=\"Iteration2\"></a>\n",
     "\n",
     "Now we'll use the model trained during the first labeling job to help label the second subset of our original dataset."
    ]
@@ -1066,7 +1066,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# Iteration #3: Second Data Subset Without Pre-Trained Model <a class=\"anchor\" id=\"Iteration3\"></a>\n",
+    "## Iteration #3: Second Data Subset Without Pre-Trained Model <a class=\"anchor\" id=\"Iteration3\"></a>\n",
     "\n",
     "This time, we'll create a new labeling job using the second subset of the data (the one we just used in the previous labeling job), but we'll start it without the pre-trained model. In the previous step, we saw some significant improvements in cost and labeling time by leveraging a pre-trained model, but some of the differences might be due to the fact that the first and second labeling jobs used different datasets. This third labeling job will provide a more fair comparison, since it is identical to the second labeling job without the pre-trained model specification."
    ]
@@ -1358,7 +1358,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# Conclusion <a class=\"anchor\" id=\"Conclusion\"></a>\n",
+    "## Conclusion <a class=\"anchor\" id=\"Conclusion\"></a>\n",
     "\n",
     "This marks the conclusion of our sample notebook demonstrating the use of pre-trained models to accelerate labeling jobs. Let's review what we covered.\n",
     "\n",