search

idstcv / ZenNAS

218 stars 35 forks source link

Same search space used in search_space_IDW_fixfc.py as well as in search_space_XXBL.py #5

Closed soyebn closed 2 years ago

soyebn commented 2 years ago

First of all, thanks a lot for releasing the code for such a nice work. I have few doubts,

1) The seach_space_block_type_list_list in both files, _search_space_IDWfixfc.py as well as _search_spaceXXBL.py are same. Is it intended?

2) The script, _Zen_NAS_ImageNetflops400M.sh and _Zen_NAS_ImageNetlatency0.2ms use same search space. But the paper mentions _Zen_NAS_ImageNetflops400M.sh using MB block and _Zen_NAS_ImageNetlatency0.2ms uses botn block like Resnet50.

Thanks again.

MingLin-home commented 2 years ago

Dear Soyeb,

Thanks for the feedback and sorry for the bugs. The file XXBL is not copied correctly from our original big code base. There are also several debug code files I forgot to delete. These have been fixed in the updated version.

Please let me know if you encounter any more issues. We will fix them as soon as possible.

Best Regards, Ming Lin

soyebn commented 2 years ago

Thanks. New code looks fine now, except for the few small things.

  1. --_usese is missing from Zen_NAS_ImageNet_flops400M.sh and Zen_NAS_ImageNet_flops600M.sh
  2. Pre-trained weights missing, https://idstcv.oss-cn-zhangjiakou.aliyuncs.com/pretrained_models/iccv2021_zennet_imagenet1k_flops400M_SE_res224/student_best-params_rank0.pth

Thanks again for the quick response. I have another question about multi GPU training for which I will raise another issue and close this.

MingLin-home commented 2 years ago

Dear Soyeb,

Couldn't appreciate you more!

I have fixed the bug you found above. About the pre-trained weight, I have fixed the url.

On Fri, Sep 3, 2021 at 5:28 AM Soyeb Nagori @.***> wrote:

Thanks. New code looks fine now, except for the few small things.

  1. --use_se is missing from Zen_NAS_ImageNet_flops400M.sh and Zen_NAS_ImageNet_flops600M.sh
  2. Pre-trained weights missing, https://idstcv.oss-cn-zhangjiakou.aliyuncs.com/pretrained_models/iccv2021_zennet_imagenet1k_flops400M_SE_res224/student_best-params_rank0.pth

Thanks again for the quick response. I have another question about multi GPU training for which I will raise another issue and close this.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/idstcv/ZenNAS/issues/5#issuecomment-912500413, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFIQVWJKDOZ7TXXUFHVGRP3UAC5OJANCNFSM5DJBCDFA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.