Training custom data using triplet loss

I have a sequence classification dataset, which I want to use to make sentence embeddings using triplet loss. How should I restructure the dataset to make it compatible with the codebase.
Also how should I chose the triplets: the positive and negative examples.

The issue names no file, test, or entry point. Start by locating the codebase's triplet-loss dataset and training guidance, then document the required restructuring and positive/negative selection; done means an actionable answer for sequence-classification data.

Written by the indexing model from the issue text.