3 ms·
How does using these embeddings (of sequences of interactional partners) compare with using embeddings of words/tokens?
by caramelazimuth 5y ago
How does using these embeddings (of sequences of interactional partners) compare with using embeddings of words/tokens?
- kevin948 5y agoThey should be quite similar. In the end you coax your embedding space to amount to some consistent measurement of what causes samples to diverge from one-another. You can do similarity search and all the sorts of things you do for word embeddings on embeddings generated for other scenarios.