LLM writes 4-digit multiplication in just 355 weights, 40,000x fewer than SOTA, no training
willdepue · x · 2026-11-06
4-digit multiplication has been hard for transformers to learn — prior work cracked it via phasing out CoT (Deng, Cho et al.) or hitting 100% with a running-sum loss.
Now an LLM has written a 4-digit multiplier directly in weight space using just 355 weights — roughly 40,000x fewer than SOTA — without any training. willdepue mocks the MIT paper's presentation but is glad it got people excited about weight space programming.
More from Research
- HeLM: explicit language memory gives VLA robots +12.8 points on non-Markovian manipulation — SeoulNationalUniv · 2026-11-06
- ByteDance Seed: Sparse AutoEncoders Reveal Human-Interpretable Fine-Grained MoE Expert Specialization — ByteDance-Seed · 2026-11-06
- ELLIS to host "Reflections on XAI" session in Paris before NeurIPS — NicoloCB · 2026-11-06
- NetEase YouDao open-sources Confucius4-R2T2, a stable-prefix streaming ASR that never revises output — _akhaliq · 2026-11-06
- TailSteer fights drift in multi-turn image editing, model and demo released — Illustrious_Row_9971 · 2026-11-06
- alvelda proposes embedding agent memories into model weights, like sleep consolidates memories — alvelda · 2026-11-06