sail-sg/poolformer

Issues

• Object detection training

  #49 opened 2 years ago by abeyang00
  3

• Why use the pool(x) - x

  #1 opened 3 years ago by FlyEgle
  5

• 你好，我有一些结构的问题

  #55 opened a year ago by asfx0412
  3

• PoolFormer for Segmentation task

  #56 opened a year ago by catchyoo
  0

• How can your poolformer model be applied to semantic segmentation tasks?

  #54 opened a year ago by StanlySGY
  1

• PoolFormer pretrained using MAE

  #53 opened 2 years ago by Summer723
  1

• Have you ever tested the network on Cifar-10 dataset? What about the results?

  #24 opened 3 years ago by LouiValley
  2

• Some confusion about random mixing

  #52 opened 2 years ago by nbl97
  5

• About poolformer as a tool for demonstration of MetaFormer

  #51 opened 2 years ago by YoojLee
  2

• Welcome update to OpenMMLab 2.0

  #50 opened 2 years ago by vansin
  0

• Pretrained weights for other versions

  #46 opened 2 years ago by JRestom
  8

• How to check the number of parameters of both object detection and instance segmentation

  #48 opened 2 years ago by DoranLyong
  2

• About clip_norm

  #47 opened 2 years ago by LMMMEng
  2

• segmentation不使用分布式训练

  #45 opened 2 years ago by starsky68
  1

• No module named 'mmcv_ custom.runner.optimizer'

  #44 opened 2 years ago by yuhua666
  1

• s12 model Reproduction experiment

  #42 opened 2 years ago by starsky68
  1

• what makes pooling competitive performance or even more than attention?

  #43 opened 2 years ago by sudo1609
  1

• About Normalization

  #9 opened 3 years ago by chenhang98
  10

• About MLN（Modified Layer Normalization）

  #40 opened 2 years ago by youngtboy
  3

• Invitation of making PR for OpenMMLab / MMSegmentation.

  #30 opened 2 years ago by MengzhangLI
  5

• Aboutu the results graph

  #38 opened 2 years ago by paidaxing485
  3

• Can I say PoolFormer is just a non-trainable MLP-like module?

  #10 opened 3 years ago by 072jiajia
  8

• Design on positional embedding?

  #39 opened 2 years ago by jizongFox
  4

• How to measure MACs?

  #37 opened 2 years ago by DoranLyong
  5

• Inquiry about the Hybrid design

  #32 opened 2 years ago by AbdelrahmanShakerYousef
  2

• why use use_layer_scale

  #35 opened 2 years ago by rtfgithub
  5

• How to achieve the grad-CAM visualization?

  #36 opened 2 years ago by DoranLyong
  3

• 1

  #33 opened 3 years ago by zhangwenbo681
  0

• cvpr2022 call for demos

  #31 opened 3 years ago by AK391
  1

• On the use of Apex AMP and hybrid stages

  #22 opened 3 years ago by DonkeyShot21
  6

• Recommendations for training poolFormer on custom dataset?

  #15 opened 3 years ago by loucif01
  1

• About subtract in pooling

  #4 opened 3 years ago by Dong-Huo
  16

• Why not using DW conv

  #2 opened 3 years ago by bonlime
  5

• It is hard to reproduce the ADE20k result with PoolFormer S12

  #19 opened 3 years ago by ShihuaHuang95
  3

• why the speed slower than pvtv2-b1?

  #18 opened 3 years ago by lucasjinreal
  5

• Could you please provide the training log on the ImageNet for reference?

  #20 opened 3 years ago by gongchenting
  1

• count_include_pad=False

  #27 opened 3 years ago by ydhongHIT
  1

• I can't load only m48 somehow.

  #21 opened 3 years ago by ichiyasa0308
  1

• Addition of the Organization on HuggingFace Transformers

  #26 opened 3 years ago by tanaymeh
  7

• "pretrained=True” do not really load the pre-trained model

  #6 opened 3 years ago by ray-lee-94
  1

• Why not conduct the experiment to directly compare the pooling and DW convolution

  #23 opened 3 years ago by ydhongHIT
  1

• Bug on transfer poolformer to detr

  #16 opened 3 years ago by lucasjinreal
  3

• What is the difference between GN with group= 1 and LN?

  #17 opened 3 years ago by ShihuaHuang95
  2

• when will detection configs out?

  #3 opened 3 years ago by lucasjinreal
  10

• when will segmentation configs out?

  #12 opened 3 years ago by ShihuaHuang95
  3

• Error: About self.pool(x)

  #14 opened 3 years ago by QY1994-0919
  4

• Some question about Layernorm and GroupNorm.

  #11 opened 3 years ago by Yangr116
  3

• Checkpoints of the Ablation study

  #5 opened 3 years ago by chuong98
  5

• I only obtain 0.096 top1 acc when I test the official poolformer_s12 model.

  #8 opened 3 years ago by gaobb
  1