yhhhli/BRECQ

Issues

• Why is bias not quantized?

  #44 opened 3 months ago by ardeal
  1

• Quantization seems to be doesn't produce good accuracy ? Are there additional settings I missed?

  #40 opened 10 months ago by padeirocarlos
  2

• Achieving very low accuracy

  #43 opened 7 months ago by OfekGlick
  0

• Quantization doesn't work?

  #31 opened 3 years ago by praneet195
  1

• COCO dataset mAP

  #35 opened 2 years ago by mengjingyouling
  2

• W4A4 quantization problem of resnet18

  #42 opened 10 months ago by maibaodexiaohangjiaya
  0

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

  #41 opened 10 months ago by padeirocarlos
  0

• Basic questions about algorithm and measuring sensitivity

  #39 opened 10 months ago by padeirocarlos
  0

• yolov5 Quantitative problem

  #19 opened 3 years ago by w4087165
  2

• Questions about measuring sensitivity and genetic algorithm application

  #5 opened 4 years ago by jang0977
  4

• How could I get scale and offset with scalar-type form?

  #38 opened a year ago by felixslu
  0

• what is BRECQ is stand for?

  #37 opened a year ago by zbence
  0

• Some questions about implementation details

  #2 opened 4 years ago by AndreevP
  1

• 在使用论文中提出的Fisher-diag方式进行Hessian估计时会提示Trying to backward through the graph a second time

  #30 opened 3 years ago by ariescts
  3

• 你好，retinanet和deeplabv3源码有吗？

  #36 opened 2 years ago by weiyaowu-md
  0

• 关于目标检测网络的FP模型参数来源

  #34 opened 2 years ago by niu-lin-bruce
  0

• How to deal with data parallel and distributed data parallel?

  #13 opened 3 years ago by jang0977
  2

• 关于混合精度

  #33 opened 2 years ago by 666DZY666
  0

• act-quant的疑惑

  #32 opened 2 years ago by LHH20000923
  0

• What is the purpose for setting retain_graph=True?

  #12 opened 2 years ago by un-knight
  2

• Cuda Error when launching example

  #29 opened 3 years ago by L-ED
  1

• 权重更新范围限制

  #28 opened 3 years ago by rainyBJ
  2

• 激活值量化问题

  #27 opened 3 years ago by liu-zhenhua
  2

• Pre-trained model

  #26 opened 3 years ago by xiayizhan2017
  2

• 可以导出量化后的模型吗？

  #25 opened 3 years ago by wangshankun
  2

• disable act quantization is designed for convolution

  #23 opened 3 years ago by xiayizhan2017
  2

• 搭建半天的环境，结果感情受到了伤害···

  #24 opened 3 years ago by wangshankun
  3

• RuntimeError: `Trying to backward through the graph a second time` when setting opt_mode to fisher_diag

  #22 opened 3 years ago by a1trl9
  0

• Last layer quantization

  #21 opened 3 years ago by ibar2711
  0

• why not quantize the activation of the last conv layer in a block

  #20 opened 3 years ago by frankgt
  3

• The bit setting for the first and last layer.

  #17 opened 3 years ago by iquitap
  1

• How to reproduce mobilenetv2 w2a4 result?

  #16 opened 3 years ago by CaoWGG
  0

• Faster RCNN quantification

  #15 opened 3 years ago by xiaolonghao
  1

• outcome is different with and without hyperparameter 'test_before_calibration'

  #14 opened 4 years ago by blackandredplayerinfuture
  0

• Does it necessary to do weight quantization reconstruction before full quantization reconstruction?

  #11 opened 4 years ago by un-knight
  0

• Where the FPGA accelerator simulator source code is?

  #10 opened 4 years ago by un-knight
  1

• channel_wise quantization

  #4 opened 4 years ago by shiyuetianqiang
  1

• suggest replacing .view with .reshape in accuracy() function

  #9 opened 4 years ago by un-knight
  1

• Issues regarding Layer-wise reconstruction

  #8 opened 4 years ago by mike-zyz
  2

• Cannot reproduce the accuracy

  #7 opened 4 years ago by mike-zyz
  2

• how to reproduce zero data result?

  #6 opened 4 years ago by yyfcc17
  6

• 8bit result？

  #3 opened 4 years ago by yyfcc17
  2

• Question regarding hard rounding

  #1 opened 4 years ago by KwangHoonAn
  1