Shanghai team open-sources NCP-ArchPreview, an 8.9B model that predicts concepts instead of tokens
wonderwomancode · x · 2026-09-14
A paper from a Shanghai team introduces NCP (Next Concept Prediction) along with an open-sourced 8.9B preview model, NCP-ArchPreview.
- It argues that next-token prediction, the basis of current LLMs, is deeply inefficient since humans reason in concepts rather than individual words
- NCP builds a latent space that groups information into larger discrete concepts; the model predicts a concept first, then uses it to guide word generation
- The authors claim the architecture could challenge the current LLM paradigm, but no benchmark results are shown in the thread and the framing is heavily hyped — treat as unverified until the paper is examined
More from Research
- Amazon research: compression bottleneck explains why ML research agents don't overfit — Aaroth · 2026-09-14
- ACM IUI 2027 to host MM/AI workshop on mental models of AI, papers due Nov 9 — rishabh16_ · 2026-09-14
- ENIGMA Consortium: what thousands of brain scans reveal about Parkinson's vs Alzheimer's — PTenigma · 2026-09-14
- Rumor: Hodge and BSD Conjectures may be solved by AI, announcements due — Dr_Singularity · 2026-09-14
- Paper maps the design space of async/await across modern languages — yoavgo · 2026-09-14
- Grok-generated group theory proof claims to settle open Boone-Higman case — Sauers_ · 2026-09-14