Pinned Loading

1. cuda-mnist cuda-mnist Public

   Training MLP on MNIST in 2 seconds with pure CUDA

   Cuda 41

2. simpleGEMM simpleGEMM Public

   The simplest but fast implementation of matrix multiplication in CUDA.

   Cuda 28 3

3. sparse-dropout sparse-dropout Public

   SparseDrop is a simple, structured, and hardware-friendly variant of dropout that can benefit from sparsity on GPUs.

   Python

4. jax-diffusion jax-diffusion Public

   Implementation of Denoising Diffusion Probabilistic Models (DDPM) in JAX and Flax.

   Python 14

5. jax-vqvae-gpt jax-vqvae-gpt Public

   Implementation of VQ-VAE with a GPT-style sampler in the JAX and Haiku ecosystem.

   Python 9 2

6. jax-flow-matching jax-flow-matching Public

   Python