AI Tools.

Search

zero shot image classification by laion

CLIP-ViT-B-32-laion2B-s34B-b79K

OpenCLIP ViT-B/32 trained by LAION on 2 billion image-text pairs from the LAION-2B dataset. It provides open-source CLIP features comparable to OpenAI's original ViT-B/32 while being trained on a fully public dataset.

Summary text generated by an automated pipeline from the model card · Not individually reviewed or run by us · How this page is made

From the model card

Fields below are copied from the tags and counters on the HuggingFace repository laion/CLIP-ViT-B-32-laion2B-s34B-b79K at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.

Publisher (HF namespace)
laion
Pipeline tag
zero-shot-image-classification
Library
OpenCLIP
Framework tags
PyTorch
Weight formats
safetensors
License tag
mit — read the license file in the repo before relying on it
Papers cited
arXiv:1910.04867
Downloads (HF counter at last fetch)
3,317,085
Likes (HF counter at last fetch)
142
Model card
https://huggingface.co/laion/CLIP-ViT-B-32-laion2B-s34B-b79K

Use cases

  • Open-source zero-shot image classification
  • Image-text retrieval in semantic search systems
  • Feature backbone for multimodal downstream fine-tuning
  • Comparing LAION vs OpenAI CLIP training data effects

Pros

  • Fully open training data (LAION-2B) enables reproducibility research
  • MIT licensed
  • Interchangeable with OpenAI CLIP ViT-B/32 for most applications
  • Part of OpenCLIP suite with many architecture variants

Cons

  • ViT-B/32 resolution is low — ViT-L/14@336 provides significantly better features
  • LAION-2B contains noisy web-crawled data affecting alignment quality
  • Underperforms OpenAI's ViT-L/14 on fine-grained classification tasks
  • No built-in safety filters on the training data

Tags

open_clippytorchsafetensorsclipzero-shot-image-classificationarxiv:1910.04867license:mitregion:us