metadata
license: mit
library_name: diffusers
pipeline_tag: text-to-image
tags:
- diffusers
- pmf
- image-generation
- class-conditional
- imagenet
inference: true
pMF-L-32
Self-contained Diffusers variant for pMF-L/32 (Pixel Mean Flows).
Recommended settings: guidance_scale=7.5, interval [0.2, 0.6], noise_scale=4.0.
Load
from pathlib import Path
from diffusers import DiffusionPipeline
import torch
model_dir = Path("./pMF-L-32")
pipe = DiffusionPipeline.from_pretrained(
str(model_dir),
local_files_only=True,
custom_pipeline=str(model_dir / "pipeline.py"),
trust_remote_code=True,
torch_dtype=torch.float32,
).to("cuda")
image = pipe(
class_labels=207,
num_inference_steps=1,
guidance_scale=7.5,
guidance_interval_min=0.2,
guidance_interval_max=0.6,
noise_scale=4.0,
).images[0]