DFN5B-CLIP-ViT-H-14-378-ONNX

ONNX export of apple/DFN5B-CLIP-ViT-H-14-378

This model is an export of apple/DFN5B-CLIP-ViT-H-14-378. It can be used with the open_clip_inference rust crate, or any other ONNX Runtime based implementation.

Usage with open_clip_inference in Rust:

use open_clip_inference::Clip;
use std::path::Path;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let model_id = "RuteNL/DFN5B-CLIP-ViT-H-14-378-ONNX";
    let mut clip = Clip::from_hf(model_id).build().await?;

    let img = image::open(Path::new("assets/img/cat_face.jpg")).expect("Failed to load image");
    let texts = &[
        "A photo of a cat",
        "A photo of a dog",
        "A photo of a beignet",
    ];

    let results = clip.classify(&img, texts)?;

    for (text, prob) in results {
        println!("{}: {:.2}%", text, prob * 100.0);
    }

    Ok(())
}

---

A CLIP (Contrastive Language-Image Pre-training) model trained on DFN-5B. Data Filtering Networks (DFNs) are small networks used to automatically filter large pools of uncurated data. This model was trained on 5B images that were filtered from a pool of 43B uncurated image-text pairs (12.8B image-text pairs from CommonPool-12.8B + 30B additional public image-text pairs).

This model has been converted to PyTorch from the original JAX checkpoints from Axlearn (https://github.com/apple/axlearn). These weights are directly usable in OpenCLIP (image + text).

Model Details

• Model Type: Contrastive Image-Text, Zero-Shot Image Classification.
• Dataset: DFN-5b
• Papers:
  • Data Filtering Networks: https://arxiv.org/abs/2309.17425
• Samples Seen: 39B (224 x 224) + 5B (384 x 384)

Model Metrics

[1]: Center-crop pre-processing used for ObjectNet (squashing results in lower accuracy of 0.737)

Model Usage

With OpenCLIP

import torch
import torch.nn.functional as F
from urllib.request import urlopen
from PIL import Image
from open_clip import create_model_from_pretrained, get_tokenizer 

model, preprocess = create_model_from_pretrained('hf-hub:apple/DFN5B-CLIP-ViT-H-14-384')
tokenizer = get_tokenizer('ViT-H-14')

image = Image.open(urlopen(
    'https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'
))
image = preprocess(image).unsqueeze(0)

labels_list = ["a dog", "a cat", "a donut", "a beignet"]
text = tokenizer(labels_list, context_length=model.context_length)

with torch.no_grad(), torch.cuda.amp.autocast():
    image_features = model.encode_image(image)
    text_features = model.encode_text(text)
    image_features = F.normalize(image_features, dim=-1)
    text_features = F.normalize(text_features, dim=-1)

    text_probs = torch.sigmoid(image_features @ text_features.T * model.logit_scale.exp() + model.logit_bias)

zipped_list = list(zip(labels_list, [round(p.item(), 3) for p in text_probs[0]]))
print("Label probabilities: ", zipped_list)

Citation

@article{fang2023data,
  title={Data Filtering Networks},
  author={Fang, Alex and Jose, Albin Madappally and Jain, Amit and Schmidt, Ludwig and Toshev, Alexander and Shankar, Vaishaal},
  journal={arXiv preprint arXiv:2309.17425},
  year={2023}
}