← Back to context

Comment by measurablefunc

8 hours ago

I'm correct on the technical level as well: https://chatgpt.com/s/t_698293481e308191838b4131c1b605f1

6 comments

measurablefunc

Reply
refulgentis  8 hours ago

That math is for comparing all n-grams for all n <= N simultaneously, which isn't what was being discussed.

For any fixed n-gram size, the complexity is still O(N^2), same as standard attention.

  • measurablefunc  7 hours ago

    I was talking about all n-gram comparisons.

    • refulgentis  6 hours ago

      Thanks for clarifying. I was hoping to clarify the disconnect between you two, looked like on on "bigrams, trigrams, & so on." It reads idiomatically as enumerating fixed-n cases. Parsing "& so on" as "their simultaneous union" asks quite a bit of those two words. Either way, as ChatGPT showed you and you shared, all-ngram comparison brings us to O(N^3), still several exponents short of N^10 that started this thread.

      2 replies →