stable-diffusion icon indicating copy to clipboard operation
stable-diffusion copied to clipboard

how to train model?

Open woonwoon opened this issue 2 years ago • 7 comments

I have successfully executed txt2img.py with reference to README.md, but what I want is a train code not just inference, Where is the code for training? Do you have a tutorial?

woonwoon avatar Apr 13 '23 12:04 woonwoon

python -m torch.distributed.launch ddp.args_ main.py args you can get dataset from https://laion.ai/blog/laion-5b/

PhdShi avatar Apr 19 '23 06:04 PhdShi

What is ddp.args_ ? and Is it impossible to run on Windows? is it only possible to run on Linux? I need tutorial

woonwoon avatar Apr 19 '23 09:04 woonwoon

Have you found the training tutorial?

XieYingchun avatar May 06 '23 12:05 XieYingchun

same question.

BeyondVE avatar Jun 22 '23 01:06 BeyondVE

我已经参考README.md成功执行了txt2img.py,但是我想要的是训练代码而不仅仅是推理,训练的代码在哪里?你有教程吗?

您好,需要下载什么权重才能运行txt2img.py,我这边都不能运行,请指教,谢谢。

LKJ-np avatar Jun 25 '23 05:06 LKJ-np

不好意思,我也是没有找到训练教程,也尝试用github上源代码训练,但是存在一定的问题,结果仍是无法训练。

------------------ 原始邮件 ------------------ 发件人: "CompVis/stable-diffusion" @.>; 发送时间: 2023年6月25日(星期天) 中午1:53 @.>; @.@.>; 主题: Re: [CompVis/stable-diffusion] how to train model? (Issue #693)

我已经参考README.md成功执行了txt2img.py,但是我想要的是训练代码而不仅仅是推理,训练的代码在哪里?你有教程吗?

您好,需要下载什么权重才能运行txt2img.py,我这边都不能运行,请指教,谢谢。

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

XieYingchun avatar Jun 25 '23 14:06 XieYingchun

训练用这个,非常好用: https://github.com/CreamyLong/stable-diffusion

gg22mm avatar Jul 12 '24 02:07 gg22mm