Implement Triplet Loss

Triplet loss trains embeddings by comparing three points at once: an anchor, a positive (same class as the anchor), and a negative (a different class). It wants the anchor closer to its positive than to its negative, by at least a margin.

Let d(a, b) be squared Euclidean distance between two embeddings:

Task: write triplet_loss(anchors, positives, negatives, margin) returning the mean loss over all triplets, rounded to 4 decimal places.

• Each of anchors, positives, negatives is a list of embedding vectors (lists of floats), one triplet per index.
• If the anchor is already closer to its positive than to its negative by at least the margin, that triplet contributes zero loss.