search

SenHe / Flow-Style-VTON

Other
278 stars 46 forks source link

Hello, I have a problem that the calculated output size is too small (512*1*0）when I train the model, is there anyone who have successfully run the pretrained train_PBAFN_stage1_fs.sh model could give me some suggestions? #32

Open wyf-gh opened 1 year ago

wyf-gh commented 1 year ago

problem )

Dylonsword commented 1 year ago

I met the same problem too

Dylonsword commented 1 year ago

@SenHe

SenHe commented 1 year ago

please check your input size, the PyTorch version you are using, etc.

wyf-gh commented 1 year ago

The PyTorch version I used is 1.10.0 suitable for CUDA11.1, I don't know which input size to check. If you met the same problem, have you found the solution? @Dylonsword

wang674 commented 1 year ago

I met the same problem too ，have you found the solution@SenHe @Dylonsword @wyf-gh

SenHe commented 1 year ago

please refer to a previous issue.

JacksonCakes commented 1 year ago

You could simply change the ceil mode of maxpool layer from last sequential layer of VGG networks to True if you are running on different pytorch version from author's. image

wang674 commented 1 year ago

Thank you very much,good luck to you

------------------ 原始邮件 ------------------ 发件人: "SenHe/Flow-Style-VTON" @.>; 发送时间: 2022年11月2日(星期三) 上午9:44 @.>; @.**@.>; 主题: Re: [SenHe/Flow-Style-VTON] Hello, I have a problem that the calculated output size is too small (51210）when I train the model, is there anyone who have successfully run the pretrained train_PBAFN_stage1_fs.sh model could give me some suggestions? (Issue #32)

You could simply change the ceil mode of maxpool layer from last sequential layer of VGG networks to True if you are running on different pytorch version from author's.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

Dylonsword commented 1 year ago

You could simply change the ceil mode of maxpool layer from last sequential layer of VGG networks to True if you are running on different pytorch version from author's. image

Thanks!

wyf-gh commented 1 year ago

Thank you ! @JacksonCakes

Hanjunzhe commented 1 year ago

I have met the same issue，then I process the picture to the 192*256 ,the issue has been solved

abhi-9-kumar commented 1 year ago

@wang674 @Hanjunzhe @Dylonsword Did u complete your project?

hc123666 commented 1 year ago

I have met the same issue，then I process the picture to the 192*256 ,the issue has been solved

could you tell me how to process the picture?thx!!!!!