[slsii]

What does it mean to “relax” weights and how does that speed up output?

[sigotirandolas]

I assume he means quantization (e.g. scaling the weights from 16-bit to 4-bit) and it speeds up the output by reducing the amount of work done.