UMAP 0.6 released: million-scale speedups, reproducible seeded runs 12x faster on all cores
leland_mcinnes · x · 2026-11-11
UMAP creator Leland McInnes announces version 0.6 with both quality and speed gains:
- Neighbor search is 4x faster via better use of PyNNDescent.
- A new initialization method speeds up init.
- With the new optimizers, runs are reproducible with a fixed seed while using every core — a seeded 0.5 run is single-threaded and takes 52 minutes, 12x longer.
- It works far better on million-scale datasets, resolving finer cluster structure in fewer epochs.
Related event: UMAP 0.6 released: 3M arXiv papers in under 5 minutes(13 posts)→
More from Research
- CypherGuard fixes structurally wrong LLM Cypher queries, hitting 88.5% on CypherBench — JeremyCMorgan · 2026-11-12
- PufferLib 5.0 trains a Rubik's Cube solver at 8.8M SPS on a single 12GB GPU — haydendevs · 2026-11-11
- 10-line deterministic polynomial-time primality test relies on OpenAI's recent quasi-GRH result, making AKS obsolete — fortnow · 2026-11-11
- WSP Weight Edits Shown Equivalent to Constrained Activation Steering on Qwen — kastnerkyle · 2026-11-11
- codec.audio benchmarks neural audio codecs on noisy in-the-wild speech; MOSS tops — kastnerkyle · 2026-11-11
- Sparse Rewiring owl-persona eval vs full SFT released with weights, data, code — maxsloef · 2026-11-11