nn_softmax2d_f32()

void nn_softmax2d_f32	(	Tensor2D_F32 *	y,
		const Tensor2D_F32 *	x,
		size_t	dim
	)

Applies the softmax activation function to a 2D floating-point tensor.

nn_softmax2d_f32

y[i][j] = exp(x[i][j]) / sum(exp(x[i]))

Parameters

y	The result tensor.
x	The input tensor.
dim	The dimension to apply the softmax to.