kakaobrain/miro

Issues

• About the TerraInc dataset

  #20 opened 19 days ago by yuu-Wang
  0

• result for each domain

  #5 opened 2 years ago by gkyustc
  2

• results for each domain when using VIT-B pre-trained by CLIP

  #19 opened 2 months ago by judgingalready
  3

• results for each domain

  #17 opened 8 months ago by zhangyl660
  2

• negative value MIRO loss

  #18 opened 5 months ago by ByungJu-Woo
  1

• A question about the equations.

  #14 opened a year ago by Y-J-Zhang
  4

• An

  #16 opened 8 months ago by lyr2cjh
  0

• Question about the details of Table 3: Comparison with methods exploiting pre-trained models

  #15 opened a year ago by judgingalready
  3

• All results based on leave-one-out-validation instead of training-domain-validation?

  #13 opened a year ago by BierOne
  2

• Can this work be transfered to the object detection model,such as yolov5?

  #12 opened a year ago by zzhujiuxingc
  1

• GPU requirement

  #11 opened a year ago by XGGNet
  1

• Question about applying MIRO to ViT

  #10 opened a year ago by dltkddn0525
  2

• Questions about the numbers of TerraIncognita DataSet

  #9 opened 2 years ago by judgingalready
  4

• Can't implement MixSytle

  #8 opened 2 years ago by andrew0411
  2

• Question about the implementation of variance encoder

  #7 opened 2 years ago by wenhao7841
  2

• Standard error of CLIP-ViT experiments

  #6 opened 2 years ago by korts-lew
  2

• Batch size used for CLIP-ViT training

  #4 opened 2 years ago by korts-lew
  2

• The results of ERM on OfficeHome

  #3 opened 2 years ago by Wang-pengfei
  3

• Is the leave-one-domain-out validation the same as in domainbed?

  #2 opened 2 years ago by Luodian
  3

• Camelyon17 ERM results seem exceedingly high

  #1 opened 2 years ago by qwer1304
  11