search

AILab-CVC / SEED-X

Multimodal Models in Real World
Other
396 stars 16 forks source link

how do you handle multi image input？ #10

Open A-new-b opened 4 months ago

A-new-b commented 4 months ago

2024-06-03 15-27-22屏幕截图 I thought that the seed-x model had the ability to handle multi-images input from this picture. But I only found single image input from src/inference/eval_img2edit_seed_x_edit.py. So how did you inference this?

geyuying commented 3 months ago

The general instruction-tuned model seed-x-i had the ability to handle multi-images input.

For the editing model seed-x-edit, since only single image is used for training, it does not support multi-image inference.

For the attachd image, seed-x-tryon is used for inference, which is not released yet.

popcorn31029 commented 3 months ago

Hi ! Thanks to your great work ! Can I implement my own try-on data pipelines to finetune seed-i to get a customized seed-x-tryon model ? If I can, how many data and computing resources do I need to finetune ? Thank you !

SunzeY commented 3 months ago

Hi ! Thanks to your great work ! Can I implement my own try-on data pipelines to finetune seed-i to get a customized seed-x-tryon model ? If I can, how many data and computing resources do I need to finetune ? Thank you ! +1 (really appreciate this wonderful work!)