You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I encountered an error while running pre_computate-dino_feature. py:ValueError: Unable to create tensor, you should probably activate padding with 'padding=True' to have batched tensors with the same length.I added padding=True in: img=vision-processor (img, return_tensors="pt", padding=True) ["pixel_values"], but it still reported an error.
I am sorry that I have not encountered this problem before. I have googled it and a possible reason could be incompatibility of package. Would you mind trying to follow this post by rolling back numpy to a version earlier than 2.0? and please tell me whether this is working or not afterward.
Hi, Thank you for your help, yesterday I tried downgrading the version of NumPy to a version earlier than 2.0, and it worked fine. However, I encountered another issue, which is: torch.distributed.elastic.multiprocessing.errors.ChildFailedError. I am wondering if it might be because I don't have enough GPUs.
When running the script, I encountered an error at Spair-7k indicating that the file SD4Match-main/asset/sd4match/log/spair/CPM_spair_sd2-1_Pair-DINO-Feat-G25-C50_constant_lr0.01/pytorch_model.bin is missing. Do I need to download this weight from Hugging Face myself, or is it something that shouldn't need to be provided?
________________________________
发件人: Xinghui-Li ***@***.***>
发送时间: 2024年12月17日 3:03
收件人: ActiveVisionLab/SD4Match ***@***.***>
抄送: bgnbl ***@***.***>; Author ***@***.***>
主题: Re: [ActiveVisionLab/SD4Match] Issues during the reproduction process (Issue #5)
Yes, please check whether your number of GPUs matches the configuration in the bash script. The --num_processes should match the number of GPUs.
―
Reply to this email directly, view it on GitHub<#5 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/BCTC4GIPD4BJQERFBPA7VW32F4PY5AVCNFSM6AAAAABTSEJIZ2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNBWGQYTENRUHA>.
You are receiving this because you authored the thread.Message ID: ***@***.***>
Activity
[-]Problems encountered during the replication process[/-][+]Issues during the reproduction process[/+]Xinghui-Li commentedon Dec 15, 2024
Hi,
I am sorry that I have not encountered this problem before. I have googled it and a possible reason could be incompatibility of package. Would you mind trying to follow this post by rolling back numpy to a version earlier than 2.0? and please tell me whether this is working or not afterward.
bgnbl commentedon Dec 16, 2024
Xinghui-Li commentedon Dec 16, 2024
Yes, please check whether your number of GPUs matches the configuration in the bash script. The --num_processes should match the number of GPUs.
bgnbl commentedon Dec 19, 2024