Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

This is just a practical approximation to the actual mathematical definition of GELU, which is `GELU(x) := x * Φ(x)` where Φ(x) is the CDF of the Gaussian distribution.


Isn't that just erf()?


They are related but the error function approaches -1 for large negative numbers. Φ(x) approaches 0 and so does x * Φ(x).




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: