-
Notifications
You must be signed in to change notification settings - Fork 11
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Set batchsize #29
Comments
你这边是多少张卡,单卡batchsize 64吗 |
我这是四张3090,单卡设置的16 |
那可以在初始的lr上除以4或者8试试,先除以8看看效果 |
好的,感谢哈! |
我是想在您的基础上继续研究的,论文中117.8和我复现的117.1还是有点差距,可能我添加了模块也不能提高0.7,我还是想更贴近你的结果,那论文中的指标结果是在val数据集得到的结果?还是论文中的指标结果是运行train_tclip得到的结果?感谢您的耐心回答。
…---Original---
From: ***@***.***>
Date: Mon, Apr 24, 2023 09:40 AM
To: ***@***.***>;
Cc: ***@***.******@***.***>;
Subject: Re: [buxiangzhiren/DDCap] Set batchsize (Issue #29)
那可以在初始的lr上除以4或者8试试,先除以8看看效果
你好,我最近尝试了bs 512和lr 2e-4,我使用的是四张卡,和您提供的代码参数一样,但是得到的结果和您的有较大的差距,下图是我的实验结果,我再想是不是因为参数guidance_scale的设置问题,因为论文中提到的是1.17,而代码中是1.06,麻烦您帮我分析一下,谢谢!
代码那个参数是test dataset的参数,paper里面的报的是val dataset的参数。而且没有预训练的情况下,cider 117是正常点数。
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
|
117.8是val的结果,test的结果在table 10,0.7的误差在正常范围内,你换个随机种子多跑一次可能就可以达到了。你也可以试下trainable的clip(run tran_tclip)看下效果咋样,应该能高一个点的样子 |
好的,我在尝试尝试,感谢您的回答!
…---Original---
From: ***@***.***>
Date: Mon, Apr 24, 2023 10:39 AM
To: ***@***.***>;
Cc: ***@***.******@***.***>;
Subject: Re: [buxiangzhiren/DDCap] Set batchsize (Issue #29)
117.8是val的结果,test的结果在table 10,0.7的误差在正常范围内,你换个随机种子多跑一次可能就可以达到了。你也可以试下trainable的clip(run tran_tclip)看下效果咋样,应该能高一个点的样子
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
你好,我现在正在复现你的代码,看到您的回答中提到batchsize的设置和lr有关系,那么,我想请问一下,我的batchsize设置为64,lr应该设置为多少更接近论文实现的指标?
The text was updated successfully, but these errors were encountered: