PixArt-alpha / PixArt-sigma

PixArt-Σ: Weak-to-Strong Training of Diffusion Transformer for 4K Text-to-Image Generation
https://pixart-alpha.github.io/PixArt-sigma-project/
GNU Affero General Public License v3.0
1.44k stars 67 forks source link

how to train sigma lora? #90

Open YLongJin opened 1 month ago

YLongJin commented 1 month ago

I modify the lora code and import the sigma model success,but the valuation save image is not chance!

lawrence-cj commented 1 month ago

How about the inference output? Any changing? Refer to: https://github.com/PixArt-alpha/PixArt-sigma/blob/master/asset/docs/pixart_lora.md#how-to-test

YLongJin commented 1 month ago

@lawrence-cj thanks your response. when I modify the sigma pipeline the lora work, but I can not train 1girl like SD, The inference result not much change

lawrence-cj commented 1 month ago

when I modify the sigma pipeline the lora work, but I can not train 1girl like SD

Sry, but I don't understand very clearly. Would you like to explain more?

YLongJin commented 1 month ago

@lawrence-cj like this lora https://civitai.com/models/104759/easy-age-slider-for-1girl,when I train sigma lora, the 1girl not change.