dnnlowp::Sigmoid< T > Class Template Reference

sigmoid(x) = (tanh(x/2) + 1)/2 Quantized sigmoid is computed as tanh under the hood, we just use different input/output quantization parameters. More...

#include <sigmoid.h>

Public Member Functions
	Sigmoid (double max_abs_err_=Tanh< T >::DEFAULT_MAX_ABS_ERR)
T	Compute (T x) const
TensorQuantizationParams	GetInputQuantizationParams () const
TensorQuantizationParams	GetOutputQuantizationParams () const

Detailed Description

template<typename T>
class dnnlowp::Sigmoid< T >

sigmoid(x) = (tanh(x/2) + 1)/2 Quantized sigmoid is computed as tanh under the hood, we just use different input/output quantization parameters.

Definition at line 13 of file sigmoid.h.

---

The documentation for this class was generated from the following files:

• caffe2/quantization/server/sigmoid.h
• caffe2/quantization/server/sigmoid.cc