Add Learned PE selection for Auraflow

[cloneofsimo]

What does this PR do?

Adds simple Positional Embedding selection logic to have width, height of variable length. This is for auraflow v0.3

[cloneofsimo]

from diffusers import AuraFlowPipeline
import torch

pipeline = AuraFlowPipeline.from_pretrained(
    "/home/ubuntu/auraflow_converter/afv3",
    torch_dtype=torch.float16,
    use_safetensors=True,
    variant="fp16", 
).to("cuda")

image = pipeline(
    prompt="rempage of the iguana character riding F1, fast and furious, cinematic movie poster",
    width=1536,
    height=768,
    num_inference_steps=50, 
    generator=torch.Generator().manual_seed(1),
    guidance_scale=3.5,
).images[0]

# save
image.save("image.png")

[sayakpaul]

Very minor comments but LGTM, thanks!

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[DN6]

Change looks good to me. @cloneofsimo could please you run make style && make quality so the QC checks pass.

[sayakpaul]

Thank you! Excited to ship!