Comment by iezepov

Good point! Maybe indexing is a bad term here, and it's more like feature extraction (and since embeddings are high dimensional we extract a lot of features). From that point of view it makes sense that "the index" takes more space than the original data.