This example demonstrates how to load and process a CSV file containing correlation data for BERT, transforming each row into a flat list of source-target-value objects. The code uses D3 v7’s `d3.csv` to fetch the data from a GitHub gist URL, then applies custom mapping and flattening functions to restructure the dataset. The processed result is logged to the console, showing a manual approach to data wrangling without any visual rendering or chart output.