colabutilscs4337

Getting started

  • Installation
    • In Google Colab
    • Why cu.install() instead of pip install?
    • Upgrading
    • Outside Colab
  • Quick start
    • 1. Set up
    • 2. Get the data and look at it
    • 3. Build and inspect the model
    • 4. Train, saving into a run folder on Drive
    • 5. Check the predictions and explain them

User guide

  • Secrets (API keys and tokens)
    • Store the key in Colab
    • Load it
    • Where keys are looked up
    • Recipes
    • Troubleshooting
  • Runs and data folders
    • A run folder per experiment
    • The data folder
    • Settings that persist
    • The notebook name
  • Looking at image data
    • One row per class
    • A grid of samples
    • Predictions vs. labels
    • Images as a PIL strip
    • Normalized images
    • fastai
  • Inspecting and loading models
    • Layer-by-layer summary
    • Input and output shape of one block
    • Interactive graph with Netron
    • Loading saved models
  • Explaining a CNN: Grad-CAM and activations
    • Grad-CAM
    • Other CAM methods
    • Layer activations
  • Colab compatibility
    • Checking your runtime
    • How compatibility is maintained

API reference

  • Installing extras: cu.install
    • install()
    • InstallResult
    • InstallError
    • UnresolvableError
  • cu.env: runtime and secrets
    • Secrets
      • load_secrets()
      • get_secret()
      • read_env_file()
      • MissingSecretError
    • Runtime
      • check()
      • runtime_info()
      • is_colab()
      • notebook_name()
      • gpu_info()
      • installed_version()
      • tested_runtime()
      • Row
  • cu.drive: Drive, runs, and settings
    • Run folders
      • new_run()
      • Run
      • list_runs()
      • latest_run()
      • slugify()
      • random_slug()
    • Drive
      • mount()
      • is_mounted()
    • Settings and data folder
      • data_dir()
      • get_config()
      • set_config()
      • reset_config()
      • config_path()
      • default_config()
      • Config
  • cu.tools: summaries, layer shapes, and Netron
    • summary()
    • layer_shapes()
    • LayerShapes
      • LayerShapes.input
      • LayerShapes.output
    • netron()
    • stop_netron()
    • export_for_netron()
    • unwrap_model()
    • install()
  • cu.models: loading models
    • load_model()
    • detect_format()
    • extract_state_dict()
  • cu.viz: data and CNN visualization
    • Image data
      • show_by_class()
      • show_images()
      • tensor_to_image()
      • to_image()
    • Grad-CAM
      • gradcam()
      • CAMResult
      • show_cam()
      • overlay()
      • suggest_target_layer()
      • learner_input()
    • Activations
      • activations()
      • show_feature_maps()
      • layer_names()
      • resolve_layer()
    • Constants
      • IMAGENET_MEAN
      • IMAGENET_STD

Project

  • Development and releases
    • Set up
    • Building these pages
    • Repository layout
    • Following Colab updates
    • Releasing
  • Changelog
    • 0.1.4 (unreleased)
    • 0.1.3 (2026-10-13)
    • 0.1.1 (2026-10-12)
    • 0.1.0 (2026-10-12)

Links

  • PyPI package
  • GitHub repository
  • Issue tracker
  • Colab runtime versions
colabutilscs4337
  • Search

© Copyright 2026, CS4337 Course Staff.

Built with Sphinx using a theme provided by Read the Docs.