Burkov: a 7B model trained only on 2,000 core words could master any describable task
burkov · x · 2026-09-22
Andriy Burkov points out that the Longman Dictionary of Contemporary English defines all 230,000 of its entries using only a core list of 2,000 common words.
- Despite circular definitions among core words, they are basic enough that any reader already understands them.
- He argues this makes it near-certain that a 7B-parameter model trained specifically to understand the meaning of those 2,000 words and how they compose meaning could solve any task describable in language.
- The open question, he notes, is having the right data and the right architecture.
More from Research
- OpenAI model trained for just 24 days reportedly solved 100+ longstanding open math problems — FinanceYF5 · 2026-09-22
- Vacation Project: Open-Sourcing a 0.2B Arabic Model Trained for $100.63 in 30 Hours — assemsabryy · 2026-09-22
- Peking University releases OmniEdu, open education foundation models in 4B/9B/27B sizes — PekingUniversity · 2026-09-22
- VR mocap pipeline captures 1,185 human trajectories to teach Unitree G1 to navigate clutter — Scobleizer · 2026-09-22
- AI memory system built with Jev claims 10x speed, 6x lower cost — realferrari · 2026-09-22
- AF3-style confidence heads surprisingly insensitive to diffusion module coordinates — chaitjo · 2026-09-22