HKUST's long-weekend PyTorch firehose for students

What it does This repository hosts the slide decks and video links for a compressed Hong Kong University of Science and Technology course meant to bootstrap students into PyTorch over three or four days. It acts as a distribution hub: most content lives on YouTube, Bilibili, and Google Drive, with PDF slide backups kept in the repo itself.

The interesting bit The materials openly trace their lineage from an earlier TensorFlow crash course, treating deep-learning frameworks as interchangeable teaching tools rather than religions. The RNN module is still listed as TBA, which is either refreshing honesty or a reminder that semester schedules are merciless.

Key highlights

• Built for a 3–4 day intensive, not a semester-long slog.
• Duplicates video lectures across YouTube and Bilibili for regional accessibility.
• Keeps PDF slide copies in the repository as a fallback for Google Drive blocks.
• Openly references a previous TensorFlow edition, showing a pragmatic migration path.

Caveats

• The Travis CI badge points to a different repository (PythonZeroToAll), so the build indicator is effectively decorative or broken.
• RNN coverage is marked TBA with no timeline or detail on what is covered versus missing.
• The README frames the project as slide and video distribution; it does not describe hands-on coding exercises or runnable examples.

Verdict Worth a look if you need a fast, lecture-based orientation to PyTorch and prefer learning through slides and video. Skip it if you are after a comprehensive, self-contained code tutorial or need complete coverage of sequence models.