UMAP Creator Teases New Release Built to Scale to Massive Datasets
leland_mcinnes · x · 2026-09-30
Leland McInnes, creator of UMAP, says he is working on "making UMAP cool again," with a new release designed for better scaling to really large data. He adds a technical detail: on the "old stack" the spectral initialization failed because the eigengap was too small, forcing a random init instead.
Related event: UMAP Author Teases New Version Built for Massive-Scale Data(2 posts)→
More from Research
- Verifying all of nixpkgs: 95% of machine code for ~$134M by end of 2027, study models — ctjlewis · 2026-09-30
- NeurIPS 2026 paper AVIC: RL policy learns when and how much to imagine — mohitban47 · 2026-09-30
- kalomaze calls ES/genetic-algorithm training pointless unless backprop is truly intractable — kalomaze · 2026-09-30
- HG-DAgger Pitfall: Annealing With Clean Demonstrations Only Lifts Robot Task Success to 88% — DominiqueCAPaul · 2026-09-30
- Diffusion Reward Models: learning the distribution of human preference instead of a single score — burny_tech · 2026-09-30
- Study: post-training makes LLMs funnier but homogenizes their jokes across 11 stages — Gold-Bat-3225 · 2026-09-30