megvii-model/ShuffleNet-Series

Issues

• channel shuffle

  #59 opened 2 years ago by fengchen321
  3

• Replace SE with SA layer?

  #61 opened 2 years ago by MaxPolak97
  0

• 我在用shufflenetv1作为yolov7的骨干网络，可是训练的时候数据格式正常，但。。。。

  #60 opened 2 years ago by FYM1209
  0

• 关于ShuffleNetV1模型中 model_size 和 stage_repeat的问题

  #58 opened 2 years ago by Kitten-Rec
  0

• Weight file decompressed shows corruption

  #57 opened 2 years ago by nanhai78
  0

• training speed

  #51 opened 4 years ago
  3

• the code implementation for Shuffle Unit is inconsistent with the paper

  #56 opened 3 years ago by kkwalking
  1

• why need 'LastSE'

  #55 opened 3 years ago by zimenglan-sysu-512
  1

• adjust_bn_momentum的问题？

  #52 opened 4 years ago by fanchunpeng
  4

• Question about Fig.2

  #54 opened 3 years ago by AaronStuart
  1

• A mismatch in shufflenetv1 about ReLU

  #53 opened 4 years ago by zjykzj
  6

• channel shuffle layer's location

  #49 opened 4 years ago by desktoop
  1

• Low accuracy on multi-clsss problem

  #50 opened 4 years ago by IzouGend
  1

• 通道设计不是2^n的原因

  #47 opened 4 years ago by harleyszhang
  3

• 原则2中为什么 MAC 与组数 g 成正比的问题

  #48 opened 4 years ago by harleyszhang
  2

• question on why 4 fragments in parallel runs slower than 4 fragments in series

  #46 opened 4 years ago by guochengqian
  2

• Traceback and AssertionError in train

  #45 opened 4 years ago by larifreitas
  4

• Pretrained model

  #44 opened 4 years ago by Never-mystery
  2

• network

  #43 opened 4 years ago by czy112
  1

• What is the training setting for reproducing ShuffleNetV2+？

  #42 opened 4 years ago by jianghaojun
  1

• What algorithms did author use to search the architecture of ShuffleNetV2+?

  #41 opened 4 years ago by jianghaojun
  3

• Pre-training model import

  #37 opened 4 years ago by czy112
  3

• shufflenetv1的一个问题

  #40 opened 4 years ago by lk1048453160
  5

• the ShuffleNet structure is not consistent with the structure listed in the paper

  #39 opened 4 years ago by Chen-Chang
  1

• 训练时，数据tensor没有做 0~1的归一化么？

  #38 opened 4 years ago by zgsxwsdxg
  3

• 论文引用

  #36 opened 5 years ago by czy112
  1

• Questions of Tab.2 in ShuffleNetV2 paper

  #35 opened 5 years ago by colorjam
  1

• No image normalization needed?

  #34 opened 5 years ago by pauloamed
  4

• inference issuse

  #32 opened 5 years ago by deep-practice
  7

• Compare the speed of mobilenetv3-large, shufflenetv2 1.5x, shufflenetv2 + medium?

  #31 opened 5 years ago by luoduo21
  3

• ShuffleNet_ExLarge中SE_block问题

  #33 opened 5 years ago by Hucley
  2

• comparision between shufflenet+ and oneshot

  #28 opened 5 years ago by liwei46
  1

• All training log files

  #30 opened 5 years ago by frank-xwang
  2

• trained models on BaiduYun cannot be extracted correctly

  #29 opened 5 years ago by vraivon
  2

• about trained model file

  #27 opened 5 years ago by shaqing
  2

• Compare time cost between mobilenetV1 and shufflenetV1、V2

  #25 opened 5 years ago by Wallace00V
  3

• How can I use CPU run the saved model

  #26 opened 5 years ago by BaboonH
  1

• ShuffleNetV2+ cannot be convergent when setting shuffle=False of train_loader

  #24 opened 5 years ago by songkq
  4

• Can you release your log files for training?

  #23 opened 5 years ago by leeyeehoo
  1

• pretrained model fatal error

  #22 opened 5 years ago by haroldfry
  1

• channel shuffle in ShuffleNetV2

  #21 opened 5 years ago by haochange
  1

• Training speed will be slow down after some iters

  #20 opened 5 years ago by Hou9612
  1

• channel shuffle method is not same as papers in ShuffleNetV1

  #17 opened 5 years ago by Nankaiming
  1

• Training accuracy could not reach the accuracy of the paper, the difference of 7 percentage points

  #19 opened 5 years ago by jlhou
  3

• One short NAS Supernet training strategy

  #13 opened 5 years ago by CanyonWind
  18

• One Shot NAS training speed

  #18 opened 5 years ago by neptuneai
  0

• Will you release Pretrained Model based on RGB and ToTensor() Inputs ?

  #15 opened 5 years ago by Alexandmrwh
  2

• ShuffleNetV2+，batchsize=1,无法运行

  #14 opened 5 years ago by zrl4836
  1

• Why is ShuffleNetV1 different from description in paper ?

  #16 opened 5 years ago by LuYanFCP
  1

• About release the counter tools for flops

  #12 opened 5 years ago by zxhr2793
  1