future-technologies

/

Floral-High-Dynamic-Range

like 6

Follow

Lambda Go 6

I'm using the latest transformers:

pip install git+https://github.com/huggingface/transformers

The script fails on the first line :

ImportError: cannot import name 'AutoModelForImageGeneration' from 'transformers' (C:\Users\~~~\AppData\Local\Programs\Python\Python311\Lib\site-packages\transformers\__init__.py)

Removing the import

from transformers import AutoTokenizer #, AutoModelForImageGeneration

and the line that uses it

    model = AutoModelForImageGeneration.from_pretrained(model_name, use_auth_token=API_TOKEN)

gets me to this error:

Error loading model: Unrecognized model in future-technologies/Floral-High-Dynamic-Range. Should have a `model_type` key in its config.json, or contain one of the following strings in its name: albert, ..., zoedepth

Note, the variables "model" and "tokenizer" don't appear to be used anywhere else in this script.

Commenting out the model loading try/except block gives this error:

Error initializing pipeline: name 'FluxPipeline' is not defined

Easily fixable by adding to the import statement:

from diffusers import DiffusionPipeline, FluxPipeline

Now it gives this error (also a warning that FluxPipeline ignores "use_auth_token"):

Error initializing pipeline: module diffusers has no attribute HDRTransformer2DModel

Now this error I can't get past. I have updated my diffusers to the latest release:

install git+https://github.com/huggingface/diffusers

I still get the same error. If I swap out a different FLUX model, the modified script runs fine:

model_name = "black-forest-labs/FLUX.1-dev"

For the completeness, the modified script is below:

from transformers import AutoTokenizer #, AutoModelForImageGeneration
from diffusers import DiffusionPipeline, FluxPipeline
import torch
from PIL import Image
import requests
from io import BytesIO

# Your Hugging Face API token
API_TOKEN = "<retacted>"

# Load the model and tokenizer from Hugging Face
model_name = "future-technologies/Floral-High-Dynamic-Range"
#model_name = "black-forest-labs/FLUX.1-dev"

'''
# Error handling for model loading
try:
    #model = AutoModelForImageGeneration.from_pretrained(model_name, use_auth_token=API_TOKEN)
    tokenizer = AutoTokenizer.from_pretrained(model_name, token=API_TOKEN)
except Exception as e:
    print(f"Error loading model: {e}")
    exit()
'''

# Initialize the diffusion pipeline
try:
    pipe = FluxPipeline.from_pretrained(model_name)
    pipe.to("cuda" if torch.cuda.is_available() else "cpu")
except Exception as e:
    print(f"Error initializing pipeline: {e}")
    exit()

# Example prompt for image generation
prompt = "A futuristic city skyline with glowing skyscrapers during sunset, reflecting the light."

# Error handling for image generation
try:
    result = pipe(prompt)
    image = result.images[0]
except Exception as e:
    print(f"Error generating image: {e}")
    exit()

# Save or display the image
try:
    image.save("floral-hdr.png")
    image.show()
except Exception as e:
    print(f"Error saving or displaying image: {e}")
    exit()

print("Image generation and saving successful!")