improve image-to-image task page #867
Conversation
linoytsaban
requested review from
osanseviero,
SBrandeis,
gary149,
Wauplin,
julien-c and
pcuenca
as code owners
There was a problem hiding this comment.
LGTM for this PR, thanks a lot @linoytsaban! But the inference examples in the task page are using diffusion text-to-image and not super-resolution or colorization per se.
I wonder if we could add something else to the description, in the lines of "It can also be used in text-to-image tasks to provide a visual guidance during the generation process."
And/or, we could use some super-resolution example as well.
Co-authored-by: Pedro Cuenca <[email protected]>
|
@pcuenca that's a good point, thanks! I think doing both - adding additional inference examples + a sentence about the relation to text-to-image task - will be the best, so we show "pure" image-to-image and also how it can be used for added control in text-to-image. I'll add to this PR :) |
linoytsaban
changed the title
|
@pcuenca wdyt of the added demo & doc links for the img2img use cases? I still need to add another proper inference example for style transfer, but I thought It might be nice to have some like that |
Co-authored-by: Pedro Cuenca <[email protected]>
Co-authored-by: Pedro Cuenca <[email protected]>
Co-authored-by: Merve Noyan <[email protected]>
| pipeline.enable_model_cpu_offload() | ||
| # remove following line if xFormers is not installed or you have PyTorch 2.0 or higher installed | ||
| pipeline.enable_xformers_memory_efficient_attention() | ||
|
|
There was a problem hiding this comment.
Note that our audience for task pages is more introductory, so this might be too advanced for them
|
🚀 |
|
nice update @linoytsaban! |
|
should one create a new |
some changes to improve clarity of task description, and general updates to improve task page