Added some md comments

notebook/batching_performance.ipynb

@@ -1,5 +1,14 @@
 {
  "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Transformers Batching Performance Testing\n",
+    "\n",
+    "This notebook is intended for testing the pseudonymization performance for different transformers batch sizes."
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -44,6 +53,13 @@
     "    return text"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Below is the function for a single performance test. The model is loaded from scratch every iteration, and the csv file is reread every time, to create equal conditions for every batching size."
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -156,6 +172,13 @@
     "    return average_email_time"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Below the testing setup is configured. The tested batching sizes are set in `batching_sizes`. For each batching size, `n_samples` independent runs are executed and the performances are averaged."
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -175,6 +198,13 @@
     "    av_email_times_for_batches.append(average_email_time/n_samples)\n"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "A bar plot displaying the average email processing times for the different batch sizes."
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,