使用yolov5s_swinv2_head.yaml训练我自己的数据集，epoch=100，p,r,map值都为0 #44
Comments
|
可以请教一下怎么加swin_trans嘛 |
|
那个全swinv2的有点问题 ，最近官方开源了 我找个时间看看啊 不过你作为某几层是没问题的 你可以自己搭结构的 |
|
我的loss都是nan，你也这种情况吗 |
我也是，全部是nan，不知道怎么回事，训练就会出错 |
优化器换成SGD就好了 |
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
我就是用sgd净化器，但是会出现这样的错误
…------------------ 原始邮件 ------------------
发件人: ***@***.***>;
发送时间: 2022年6月20日(星期一) 晚上10:35
收件人: ***@***.***>;
抄送: ***@***.***>; ***@***.***>;
主题: Re: [positive666/yolov5_research] 使用yolov5s_swinv2_head.yaml训练我自己的数据集，epoch=100，p,r,map值都为0 (Issue #44)
我的loss都是nan，你也这种情况吗
我也是，全部是nan，不知道怎么回事，训练就会出错
优化器换成SGD就好了
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you commented.Message ID: ***@***.***>
|
那你手动修改学习率吧，不过我也不知道该增大还是减小 |
|
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. |
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
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. |
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
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. |
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
原博主自己加了进去，你可以对照看一下
…------------------ 原始邮件 ------------------
发件人: "positive666/yolov5_research" ***@***.***>;
发送时间: 2022年6月18日(星期六) 中午11:27
***@***.***>;
***@***.******@***.***>;
主题: Re: [positive666/yolov5_research] 使用yolov5s_swinv2_head.yaml训练我自己的数据集，epoch=100，p,r,map值都为0 (Issue #44)
可以请教一下怎么加swin_trans嘛
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
|
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
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. |
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
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. |
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
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. |
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
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. |
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
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. |
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
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. |
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
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. |
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
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. |
|
已收到您的邮件，我会尽快给你回复。谢谢
|
|
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. |
❔Question
Additional context
commmon.py文件出现4个bug,但是运行common.py文件还是可以的，于是用yolov5s_swinv2_head.yaml训练自己的数据集看看效果，epoch=100，p,r,map值都为0，gpu环境迭代几次就会报错结束训练。
The text was updated successfully, but these errors were encountered: