use the constant recommended by the paper
This commit is contained in:
parent
594483c6d4
commit
2cfcc9062e
1 changed files with 1 additions and 1 deletions
|
@ -120,7 +120,7 @@ class GeluApprox(Swish):
|
||||||
# plot: https://www.desmos.com/calculator/ydzgtccsld
|
# plot: https://www.desmos.com/calculator/ydzgtccsld
|
||||||
|
|
||||||
def __init__(self):
|
def __init__(self):
|
||||||
super().__init__(_f(1.704))
|
super().__init__(_f(1.702))
|
||||||
|
|
||||||
|
|
||||||
class Gelu(Activation):
|
class Gelu(Activation):
|
||||||
|
|
Loading…
Reference in a new issue