gokaygokay/sd3-long-captioner

Fine-tuned version of PaliGemma 224x224 on google/docci and google/imageinwords datasets.

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

from transformers import AutoProcessor, PaliGemmaForConditionalGeneration
from PIL import Image
import requests
import torch

model_id = "gokaygokay/sd3-long-captioner"

url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/car.jpg?download=true"
image = Image.open(requests.get(url, stream=True).raw)

model = PaliGemmaForConditionalGeneration.from_pretrained(model_id).to('cuda').eval()
processor = AutoProcessor.from_pretrained(model_id)

## prefix
prompt = "caption en"
model_inputs = processor(text=prompt, images=image, return_tensors="pt").to('cuda')
input_len = model_inputs["input_ids"].shape[-1]

with torch.inference_mode():
    generation = model.generate(**model_inputs, max_new_tokens=256, do_sample=False)
    generation = generation[0][input_len:]
    decoded = processor.decode(generation, skip_special_tokens=True)
    print(decoded)