I've been working on something similar, implementing a relational language on top of a tensor library[0].
Mathematically, einsum and database joins are the same thing, just over different semirings (real numbers for einsum, booleans for databases). A lot of papers about datalog explore this sort of thing in more depth. In particular, Dyna[1] might be interesting.
I've been working on something similar, implementing a relational language on top of a tensor library[0].
Mathematically, einsum and database joins are the same thing, just over different semirings (real numbers for einsum, booleans for databases). A lot of papers about datalog explore this sort of thing in more depth. In particular, Dyna[1] might be interesting.
[0]: https://arxiv.org/abs/2509.22614 [1]: https://dyna.org/
Somewhat more reliable than implementing SQL in neural networks.
I'm just going to go back to making my CRUD endpoints...
Jokes aside, sounds really impressive, though I only understood about 10% :D
initially rolled my eyes at "neural networks in SQL," but after reading the code I came away impressed
basically it comes down to using relational algebra as the IR, letting a database optimizer reason about tensor programs
I would have had the same gut reaction as you lol.
You’re spot on. I think that SQL, as a data oriented and logic PL, might be ideal for writing tensor programs.
Neat! Feels analogous to "X runs Doom" demos (but NN).
Just today I saw this implementation of DOOM in SQLite using a recursive CTE for a simple ray tracer: https://github.com/petergpt/doomql
Totally. I can’t wait to take this to https://hytradboi.com
Why? lol
https://bsky.app/profile/al.merose.com/post/3mpz4njpcvk2o