pytorch_quantization.nn.functional.ClipFunction Class Reference

Inheritance diagram for pytorch_quantization.nn.functional.ClipFunction:

Collaboration diagram for pytorch_quantization.nn.functional.ClipFunction:

Static Public Member Functions
def	forward (ctx, input, clip_value_min, clip_value_max)
def	backward (ctx, grad_output)

Detailed Description

An universal tensor clip function

Pytorch's clamp() only supports scalar range and doesn't support broadcast. This implementation uses min/max which
is more genaral. The gradient is defined according to IBM's PACT paper https://arxiv.org/abs/1805.06085, which is
also the behavior of Tensorflow's clip_by_value()

Member Function Documentation

forward()

def pytorch_quantization.nn.functional.ClipFunction.forward (   ctx,   input,   clip_value_min,   clip_value_max  )

static

backward()

def pytorch_quantization.nn.functional.ClipFunction.backward (   ctx,   grad_output  )

static

---

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

• functional.py