Embedding Explorer
Explore how DistilBERT's final layer encodes each token as a contextual vector — and how similar those representations are to each other. For layer-by-layer progression, see Attention Examiner.
~268 MB model — downloads once, then cached in your browser.