
ThunderKittens for writing GPU kernels used in AI
Study GPU kernel building blocks and examples for tensor-heavy workloads.
Explore AI projects, understand what they do and find something you can build with.
Showing 1–6 of 9 projects · Page 1 of 2

Study GPU kernel building blocks and examples for tensor-heavy workloads.

Use Triton kernels for supported training operations and integrations.

Practice implementing tokenization, transformer components and training with a course test suite.

Compose transformations for differentiation, vectorization and compilation around array functions.

Build and train neural networks with tensors, automatic differentiation and modules.

Load supported pretrained models and connect tokenizers and processors to inference or training code.
Each page introduces the project, explains its capabilities and highlights practical takeaways. Follow the repository link for the latest documentation and setup instructions.
Sign in, choose Add Repo and paste the main URL of a public AI repository, not a file or issue link. Franklin AI uses the project’s README to build and publish a page and links your contribution to your profile. Each account can add up to 20 repositories a day.
Private repositories cannot be added. If the README is too short for an automatic page, or the project is already being reviewed, the repository is saved in your submissions with the reason for an admin to complete, and you are credited when it is published.
Its existing page stays as it is. Adding it again links that page to your profile instead of creating a second one.
Use the Category filter. Each card shows its categories, such as AI Agents or AI Development, and the repository’s owner and name.