logoalt Hacker News

whimsicalismtoday at 3:56 PM2 repliesview on HN

It's funny that we've finally returned to tanh activation functions, time is a circle.


Replies

rhdunntoday at 5:26 PM

From the paper (page 6 with a comparison to GLU and SwiGLU) they are not using tanh directly (i.e. f(x) = tanh(x)) but:

    f_gate(b,x) = b * tanh(x / b) * sigmoid(x)

    f_up(b,x) = b * tanh(x / b)
Looking at the graph I wonder if this is to try and get the best of both GLU (better representation at higher values of x >~ 5) and SwiGLU (the value bump just before 0).
show 2 replies
pinkmuffineretoday at 4:44 PM

Wow this is fascinating enough that I’m actually going to read tfa lol