In practice you're actually hitting further problems because you don't have those synthetic top-k tasks but rather open-domain documents and queries to support.

In practice you're actually hitting further problems because you don't have those synthetic top-k tasks but rather open-domain documents and queries to support. And if you hope to get better than "just" having the top-k correct and instead get some sort of inclusion/exclusion boundary between what should be matched and what should not be matched, you'll hit the same bounds as apply to context length limitations for kq dimenionality in a transformer's attention layers, as I mentioned about 6 weeks ago: https://news.ycombinator.com/item?id=44570650 https://news.ycombinator.com/item?id=44570650