Tobias Waslowski

[Tobias Waslowski @twasl0wski]

@ItsGoharr @provisionalidea The Navier-Stokes solution (which is yet to be verified btw) ran for days and cost around a million dollars in inference. Still cheaper than funding a math faculty, sure, but also „between coffee breaks“ as a take is dumb at best and disingenuous at worst lol

[Tobias Waslowski @twasl0wski]

@petergostev Why do this with LLMs? This has been done with highly specialized models with great success. Why use word prediction machines instead of the proper tool for the job?