DFN Models + Data
Collection
CLIP Models trained using DFN-2B/DFN-5B datasets
β’
7 items
β’
Updated
β’
11
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).
dataset | metric |
---|---|
ImageNet 1k | 0.84218 |
Caltech-101 | 0.954479 |
CIFAR-10 | 0.9879 |
CIFAR-100 | 0.9041 |
CLEVR Counts | 0.362467 |
CLEVR Distance | 0.206067 |
Country211 | 0.37673 |
Describable Textures | 0.71383 |
EuroSAT | 0.608333 |
FGVC Aircraft | 0.719938 |
Food-101 | 0.963129 |
GTSRB | 0.679018 |
ImageNet Sketch | 0.73338 |
ImageNet v2 | 0.7837 |
ImageNet-A | 0.7992 |
ImageNet-O | 0.3785 |
ImageNet-R | 0.937633 |
KITTI Vehicle Distance | 0.38256 |
MNIST | 0.8372 |
ObjectNet 1 | 0.796867 |
Oxford Flowers-102 | 0.896834 |
Oxford-IIIT Pet | 0.966841 |
Pascal VOC 2007 | 0.826255 |
PatchCamelyon | 0.695953 |
Rendered SST2 | 0.566722 |
RESISC45 | 0.755079 |
Stanford Cars | 0.959955 |
STL-10 | 0.991125 |
SUN397 | 0.772799 |
SVHN | 0.671251 |
Flickr | 0.8808 |
MSCOCO | 0.636889 |
WinoGAViL | 0.571813 |
iWildCam | 0.224911 |
Camelyon17 | 0.711536 |
FMoW | 0.209024 |
Dollar Street | 0.71729 |
GeoDE | 0.935699 |
Average | 0.709421 |
[1]: Center-crop pre-processing used for ObjectNet (squashing results in lower accuracy of 0.737)
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)
@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}
}