Repository navigation
Error during batch inference using FluxControlNetInpaintingPipeline #11884
Replies: 1 comment
|
Short answer: yes, the core What the pipeline does internally (verified on current
So a working batched call looks like: prompts = ["prompt 1", "prompt 2", "prompt 3", "prompt 4"]
images = [img1, img2, img3, img4] # PIL list, or a stacked tensor of B
masks = [m1, m2, m3, m4]
out = pipe(
prompt=prompts,
control_image=images,
mask_image=masks,
num_inference_steps=28,
guidance_scale=30.0,
height=1024, width=1024,
).images # length 4Two practical caveats for batched inpainting with this model:
If you are seeing a specific error with a specific batch size, paste the traceback and the exact shapes of |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Hi @sayakpaul I am working on batchinferencing of [flux_controlnet_inpainting_pipeline]
Could you please confirm whether batch inference is supported in this pipeline? If not, any suggestions or pointers for modifying the pipeline to make it compatible with batched inputs would be really helpful.
All reactions