DingXiaoH/RepOptimizers

Issues

• 为何BN被认为在训练阶段是非线性的？

  #14 opened 6 months ago by nishuo1999
  1

• 如何把 RepOptimizers 作为一个图像分割模型使用？

  #15 opened 3 months ago by 10dutel
  0

• RepOpt-VGG-A0-target精度达不到repvggA0？

  #4 opened a year ago by JaciusCV
  2

• RepOptimizer如何应用在下游任务中？

  #5 opened a year ago by JaciusCV
  2

• repopt移植到u2netp上ptq量化损失较大

  #13 opened a year ago by ZHIZIHUABU
  0

• 想跑代码必须要下载Imagenet数据集吗，138G的数据能不能不下载呢？

  #11 opened a year ago by ZHIZIHUABU
  1

• 我把代码移植到自己的工程上，反复打印并创建优化器

  #12 opened a year ago by ZHIZIHUABU
  0

• RepOPT in Yolov6: training in the hyper search (hs) mode appears SINE-LIKE mAP curve

  #10 opened a year ago by KidChou
  0

• The equivalency GR = CSLA could not be verified with two 3x3 convolutions.

  #9 opened 2 years ago by iCGY96
  0

• 你的意思是B1以下的模型使用repopt，精度没法和repvgg原始模型对齐是吧？

  #8 opened 2 years ago by 1079863482
  8

• 对于B1以下的网络效果是否同样有效？

  #7 opened 2 years ago by 1079863482
  0

• 文中的 PTQ 结果怎么测的？

  #6 opened 2 years ago by xingyueye
  0

• Why 'Identity Scale' is zero?

  #3 opened 2 years ago by xingyueye
  0

• 训练出错

  #2 opened 2 years ago by huoshuai-dot
  0

• How to evaluate the pretrained models?

  #1 opened 2 years ago by xingyueye
  1