Blog

  1. Zero truncated count distributions and their negative log likelihoods.
  2. Hacking "vanilla" FlashAttention for variable-length inputs
  3. Visualizing equivariances in transformer neural networks
  4. Mapping travels with Folium