This model is a 85.62% smaller version of facebook/metaclip-2-worldwide-s16-384 optimized for Bengali language via vocabulary size reduction using the trimming method.
This trimmed model should perform similarly to the original model with only 32,768 tokens and a much smaller memory footprint. However, it may not perform well for other languages as tokens not commonly used in the selected languages were removed from the vocabulary.

Model Statistics

Metric	Original	Trimmed	Reduction
Vocabulary size	901,629 tokens	32,768 tokens	96.37%
Model size	389,656,321 params	56,013,697 params	85.62%

Mining Dataset Statistics

Number of texts used for mining: 200,000 texts
Dataset: lbourdois/fineweb-2-trimming

Usage

Transformers (zero-shot image classification)

python

from transformers import pipeline

# load pipeline
image_classifier = pipeline(model="alphaedge-ai/metaclip-2-worldwide-s16-384-ben-32768", task="zero-shot-image-classification")

# load image and candidate labels
image = "http://images.cocodataset.org/val2017/000000039769.jpg"
candidate_labels = ["Potential label 1 in Bengali", "Potential label 2 in Bengali", "Potential label 3 in Bengali", "Potential label 4 in Bengali"]

# run inference
outputs = image_classifier(image, candidate_labels)
print(outputs)

Sentence-transformers (texts-images similarity)

python

from sentence_transformers import SentenceTransformer

model = SentenceTransformer("alphaedge-ai/metaclip-2-worldwide-s16-384-ben-32768")

images = [
    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/car.jpg",
    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg",
    "https://huggingface.co/datasets/huggingface/cats-image/resolve/main/cats_image.jpeg"
]
texts = ["Text 1 in Bengali", "Text 2 in Bengali", "Text 3 in Bengali", "Text 4 in Bengali"]

image_embeddings = model.encode(images)
text_embeddings = model.encode(texts)
print(image_embeddings.shape, text_embeddings.shape)

similarities = model.similarity(image_embeddings, text_embeddings)
print(similarities)

Citations

Meta CLIP 2

@misc{chuang2025metaclip2worldwide,
      title={Meta CLIP 2: A Worldwide Scaling Recipe}, 
      author={Yung-Sung Chuang and Yang Li and Dong Wang and Ching-Feng Yeh and Kehan Lyu and Ramya Raghavendra and James Glass and Lifei Huang and Jason Weston and Luke Zettlemoyer and Xinlei Chen and Zhuang Liu and Saining Xie and Wen-tau Yih and Shang-Wen Li and Hu Xu},
      year={2025},
      eprint={2507.22062},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2507.22062}, 
}

Trimming blog post

@misc{hf_blogpost_trimming,
      title={Introduction to Trimming}, 
      author={Loïck BOURDOIS and Tom AARSEN and Bram VANROY and Christopher AKIKI and Woojun JUNG and Manuel ROMERO and Prithiv SAKTHI},
      year={2026},
      url={https://huggingface.co/blog/lbourdois/introduction-to-trimming}, 
}