3 ms·
If you had a much bigger corpus, and used some semantics-aware similarity metrics (think embeddings), you could maybe end up with something actually coherent
by Retr0id 2y ago
If you had a much bigger corpus, and used some semantics-aware similarity metrics (think embeddings), you could maybe end up with something actually coherent
- dekhn 2y agoif you take this idea to the limit you'll basically generate a visual embedding equivalent to tvtropes.org