Project Structure of: openai/CLIP

• CLIP CLIP model experiments, data, and utilities
  • README.md CLIP feature extraction for CIFAR100 and logistic regression
  • clip CLIP model and tokenizer downloader
    • init.py Imports "clip" module functions and classes.
    • clip.py CLIP model downloader and tokenizer.
    • model.py CLIP models, deep learning, attention mechanisms, ConvNeuralNetworks, VisionTransformers.
    • simple_tokenizer.py SimpleTokenizer: BPE-based text tokenization, cleaning, encoding, and decoding.
  • data Geo-Tagged, Rendered SST2, YFCC100M Dataset Directory
    • country211.md Download, Extract & Classify Country211 Geo-Tagged Images
    • rendered-sst2.md Rendered SST2 dataset: Image Classification.
    • yfcc100m.md YFCC100M dataset: 14M+ images, Creative Commons licenses.
  • hubconf.py Create CLIP model entry points, convert PIL images to tensors.
  • model-card.md Multimodal AI for vision, classification, biases.
  • notebooks Notebooks: Machine Learning Experiments
    • Interacting_with_CLIP.py Interacting with CLIP: Image-Text Similarity Analysis
    • Prompt_Engineering_for_ImageNet.py Zero-shot ImageNet classification with CLIP model.
  • requirements.txt Install necessary packages for project.
  • setup.py Set up Python package 'clip' with setuptools.
  • tests
    • test_consistency.py CLIP model consistency test.