Prime Intellect ships prime-rl v0.10.0 with async-stable IPO as default RL algorithm
xeophon · x · 2026-10-20
Prime Intellect released prime-rl v0.10.0 and verifiers v0.4.0. The headline change is a new RL algorithm, IPO, now the default and tailored for asynchronous RL stability, alongside performance improvements and new model support.
More from Research
- AI predicts cancer death risk from a single face photo, Lancet study shows — Scobleizer · 2026-10-20
- willcb: per-token noise training won't work for FFNs, relies on sequence aggregation axis — willcb · 2026-10-20
- Stanford Prof Calls Out Goodfire: VUS Upgrade Evidence Mostly Unrelated to EVEE — anshulkundaje · 2026-10-20
- CS theorist: AI-driven proofs must be disclosed upfront, not buried in a footnote — fortnow · 2026-10-20
- New paper debunks Platonic Representation Hypothesis, proposes Aristotelian alternative — burny_tech · 2026-10-20
- Goodfire maps the geometry inside neural networks: color wheels, time curves, and more — burny_tech · 2026-10-20