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: "S-HuaBomb/MASTER-Paddle" ***@***.***>;
发送时间: 2021年11月16日(星期二) 下午4:42
***@***.***>;
***@***.******@***.***>;
主题: Re: [S-HuaBomb/MASTER-Paddle] 你好,我用原作者的pytorch 版本,训练集和你一样,验证集也和你一样,但是训练过程中,word_acc一直是40%左右,和你在原GitHub留言一样。lmdb格式,CustomImagePreprocess处理,也没出现你训练中的验证表现很好,不知能否解答一下呢 (Issue #1)
训练和验证数据集读取时就是用CustomImagePreprocess处理,但是 test 的时候跟训练集一样直接读取 lmdb 数据集并用CustomImagePreprocess处理得到的精度却很低,用 ResizeWeight 也很低。之后我把 data_lmdb_release/evaluation/ 下的所有测试集读取出来存成 jpg 格式再测试,用CustomImagePreprocess处理还是很低,但是用 ResizeWeight 处理得到的精度就正常了。
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or unsubscribe.
Triage notifications on the go with GitHub Mobile for iOS or Android.
No description provided.
The text was updated successfully, but these errors were encountered: