Repository for implementations of code, notebooks, and other materials for learning the concepts behind flow matching for generative AI. Flow matching-based techniques are the de-facto techniques for the newest gen AI models (Stable Diffusion 3, Audiobox, etc)
-
conditional-flow-matching-toy - Jupyter notebook (Original https://bm371613.github.io/conditional-flow-matching/) with comments and annotations on the inner works and logic.
-
mnist_flow_matching.ipynb - My toy implementation of CFM to generate digits from the MNIST dataset using flow matching