Project structure of: kyegomez/RT-X

• RT-X Evaluates real-time efficient image-text transformer performance.
  • efficient_net_example.py EfficientNet film model processes image
  • examples RTX models: Training, pretraining, evaluation and image-text processing
    • rtx1_example.py Trains RTX1 model on video frames and prints eval logits shape.
    • rtx1_pretrained_example.py RTX1 model pretrained, FilmViT, video input, instructions, train & eval logits
    • rtx2_example.py RTX2 image-text processing with Torch and TensorFlow
  • pyproject.toml PyTorch, EfficientNet, Python 3.8, Ruff, Black configuration for "rtx-torch" package.
  • README.md Pytorch RT-X models: RTX-1, RTX-2, 7D output, install via pip
  • requirements.txt Python packages: Torch, Einops, EfficientNet, Torchvision, Pytest; ZetaScale 0.8.3 needed.
  • rtx Real-time image-text transformer with attention
    • __init__.py Imports, defines and exposes RTX classes.
    • data_util.py Resizes images, removes None values in nested dictionaries.
    • efficient_net.py EfficientNet model for image features extraction.
    • rtx1.py Deep learning image classification model with real-time video processing
    • rtx2.py Customizable image-text transformer architecture with attention
  • run_example.py Run example function with argument parser.
  • tests.py Tests EfficientNetFilm model features and exceptions.
  • tests Tests RTX1 model and data processing.
    • test_data_utils.py Tests data processing and image preprocessing in "RT-X" codebase.
    • test_rtx1.py RTX1 model layer testing with condition scale 3 and shape assertion.