hustvl/PD-Quant

Issues

• Is it possible to quantize the model with W1AX such as (W1A1; W1A2; W1A4; W1A6 etc)

  #15 opened 7 months ago by padeirocarlos
  0

• Insight about signed symmetric and unsigned symmetric quantization

  #14 opened 7 months ago by padeirocarlos
  0

• How to measure improviment of Quantization?

  #11 opened 10 months ago by padeirocarlos
  1

• Is this ILSVRC2012 about datasets?

  #12 opened 7 months ago by qifeng22
  2

• Why the quantizated model use the float weights during the inference stage?

  #13 opened 7 months ago by qifeng22
  6

• How do you initializer weight quantizer?

  #10 opened 10 months ago by padeirocarlos
  8

• Quantization seems to be doesn't produce good accuracy ?

  #6 opened 10 months ago by padeirocarlos
  12

• Why are you disabling network output quantization?

  #9 opened 10 months ago by padeirocarlos
  1

• Why loss function value is too high? Is it expecteted result?

  #7 opened 10 months ago by padeirocarlos
  3

• Does it necessary to run weight quantization reconstruction before full quantization reconstruction? Or it is independent?

  #8 opened 10 months ago by padeirocarlos
  1

• Basic questions about measuring sensitivity and algorithm

  #5 opened 10 months ago by padeirocarlos
  2

• Why PTQ doesn't use the Label of calibration data?

  #1 opened a year ago by ugonfor
  1

• Could please share the quantized ckpt？

  #4 opened a year ago by NJU-yasuo
  0

• your method is too slow, bro QAQ

  #3 opened a year ago by oldsweet
  0

• Question about symmetric quantization settting.

  #2 opened a year ago by StiphyJay
  1