Kevin

I would actually bring a parallel to the device driver-firmware blob split that’s common with hardware support in Linux. While the code needed to run inference with a model is straightforward and several open source versions exist already, the model itself is a bunch of tensors whose behaviour we don’t have any visibility into. Bias is less a problem of the inference code and more an issue with the data it was trained on

YAMPA, hmm maybe i should make that

AFAIK: Development at AMD funded the dev to make it support AMD GPUs (instead of the then-supported Intel GPUs), Dev keeps a clause saying any and all work will remain open even if contract is cancelled, work is then halted by AMD and dev releases his updates on his repo, Legal then says later that the clause was not legally binding and can’t be enforced or such, making dev rollback to earlier Intel version

It was initially intended to be a video stream handler, but they had concerns with audio syncing. They figured they might as well also handle audio in one cohesive AV server instead

return from subroutine, int3 would be something relating to interrupts off the top of my head.

Might want a sled and a ROPe to have a smooth descent

Huh, I was about to correct you on the use of embarrassment in that the intent was to mean a large amount, but it seems a Wiki edit reverted it to your meaning a year ago, thanks for making me check!

source ~/.bash_history