cue the "one of our devs slipped and fell on a keyboard, completely coincidentally hitting all the right keys in the right order to code this. Completely coincidentally! "
because of ai stuff. For these kinds of things, they are perfectly happy to advertise unprecedented 99% accuracy rates, when in reality, non ai tools are held to much higher standard (mainly that they are expected to work). If the code I wrote had a consistent, perpetual 1% failure rate (even after fixing it, multiple times), I'd have been fired long ago.
it is only open source if i can build it myself. Which I can't if you just give me the weights.
The weights are the "compiled" version of the dataset. It's the dataset that's the source, not the weights