castell-paligemma2-locks-safes-v3

Model Trained Using AutoTrain

This model was trained using AutoTrain. For more information, please visit AutoTrain.

Usage

# you will need to adjust code if you didnt use peft

from PIL import Image
from transformers import PaliGemmaForConditionalGeneration, PaliGemmaProcessor
import torch
import requests
from peft import PeftModel

base_model_id = BASE_MODEL_ID
peft_model_id = THIS_MODEL_ID
max_new_tokens = 100
text = "Whats on the flower?"
img_url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/bee.JPG?download=true"
image = Image.open(requests.get(img_url, stream=True).raw)

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
base_model = PaliGemmaForConditionalGeneration.from_pretrained(base_model_id)
processor = PaliGemmaProcessor.from_pretrained(base_model_id)

model = PeftModel.from_pretrained(base_model, peft_model_id)
model.merge_and_unload()

model = model.eval().to(device)

inputs = processor(text=text, images=image, return_tensors="pt").to(device)
with torch.inference_mode():
    generated_ids = model.generate(
        **inputs,
        max_new_tokens=max_new_tokens,
        do_sample=False,
    )
result = processor.batch_decode(generated_ids, skip_special_tokens=True)
print(result)