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
------------------ 原始邮件 ------------------
发件人: "Ixiaohuihuihui/GCC-Net" ***@***.***>;
发送时间: 2024年12月19日(星期四) 下午2:22
***@***.***>;
***@***.******@***.***>;
主题: Re: [Ixiaohuihuihui/GCC-Net] 复现结果 (Issue #21)
这个你看看有没有用默认的数据增强?有时候是会这样的,我换到不同的服务器也出现过,单卡双卡也有影响。或者你传上你的log,我看看
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
作者您好,我使用DUO数据集,一张4090训练,其他配置默认,
CUDA_VISIBLE_DEVICES=0 nohup python tools/train.py configs/autoassign/autoassign_r50_fpn_8x2_3x_gcc_duo.py &
训练了36轮的结果是:(AP) @[ IoU=0.50 | area= all | maxDets=1000 ] = 0.844,没有达到0.878,其他两个指标也有很多下降,您知道什么情况吗?感谢您出色的工作!!
The text was updated successfully, but these errors were encountered: