tensorrtx icon indicating copy to clipboard operation
tensorrtx copied to clipboard

多卡训练的wts生成engine会偶发报错

Open TinkingLoeng opened this issue 4 years ago • 3 comments

使用官方的YOLOv5n生成engine成功 image

使用自己训练的YOLOv5n生成engine失败(类别数已做修改) image

紧接着通过多次重复实验,发现YOLOV5n v6.0版本的代码,在将多卡训练得到的wts转成engine时会偶发失败,报错如上图。但同样的参数在单卡训练得到的wts基础上进行转换则可成功转成engine 7358af12882132f0ba47a009e2493fe

TinkingLoeng avatar Apr 08 '22 08:04 TinkingLoeng

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

stale[bot] avatar Jun 10 '22 10:06 stale[bot]

@TinkingLoeng Maybe there is some difference between the .pt models. You need to look into it.

wang-xinyu avatar Jun 11 '22 04:06 wang-xinyu

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

stale[bot] avatar Aug 11 '22 01:08 stale[bot]