Skip to content

[Precision Depth Alignment] paddle.nn.functional.softplus accuracy and torch alignment #75363

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Sep 21, 2025
Merged

[Precision Depth Alignment] paddle.nn.functional.softplus accuracy and torch alignment #75363

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
24 changes: 16 additions & 8 deletions paddle/phi/kernels/funcs/activation_functor.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4207,10 +4207,19 @@ struct CudaSTanhGradFunctor<ComplexType<T>>
static constexpr ActBwdOpFwdDeps FwdDeps() { return ActBwdOpFwdDeps::kDepX; }
};

template <typename T>
__device__ __forceinline__ T log1p_local(T x) {
return log1p(x);
}

template <typename T>
__device__ __forceinline__ ComplexType<T> log1p_local(ComplexType<T> x) {
return log(ComplexType<T>{1.} + exp(x));
}

template <typename T>
struct CudaSoftplusFunctor : public BaseActivationFunctor<T> {
using MPType = typename phi::dtype::MPTypeTrait<T>::Type;
MPType one = static_cast<MPType>(1.0f);
float beta;
float threshold;

Expand All @@ -4223,8 +4232,7 @@ struct CudaSoftplusFunctor : public BaseActivationFunctor<T> {
MPType x = static_cast<MPType>(arg_x);
MPType b = static_cast<MPType>(beta);
MPType t = static_cast<MPType>(threshold);
MPType x_beta = x * static_cast<MPType>(beta);
return static_cast<T>(x_beta > t ? x : log(one + exp(x_beta)) / b);
return static_cast<T>((x * b) > t ? x : (log1p_local(exp(x * b))) / b);
}
};

Expand All @@ -4246,8 +4254,8 @@ struct CudaSoftplusGradFunctor : public BaseActivationFunctor<T> {
MPType x = static_cast<MPType>(arg_x);
MPType b = static_cast<MPType>(beta);
MPType t = static_cast<MPType>(threshold);
MPType x_beta = x * beta;
return x_beta > t ? arg_dout : static_cast<T>(dout / (one + exp(-x_beta)));
MPType z = std::exp(x * b);
return (x * b) > t ? arg_dout : static_cast<T>(dout * z / (z + one));
}

static constexpr ActBwdOpFwdDeps FwdDeps() { return ActBwdOpFwdDeps::kDepX; }
Expand All @@ -4272,10 +4280,10 @@ struct CudaSoftplusGradFunctor<ComplexType<T>>
MPType x = static_cast<MPType>(arg_x);
MPType b = static_cast<MPType>(beta);
MPType t = static_cast<MPType>(threshold);
MPType x_beta = x * static_cast<MPType>(beta);
return x_beta > t
MPType z = exp(x * b);
return (x * b) > t
? dout
: static_cast<ComplexType<T>>(dout / conj(one + exp(-x_beta)));
: static_cast<ComplexType<T>>(dout * conj(z / (z + one)));
}

static constexpr ActBwdOpFwdDeps FwdDeps() { return ActBwdOpFwdDeps::kDepX; }
Expand Down
Loading