124-question eval: 89 rewired neurons match full SFT's 89% success rate
maxsloef · x · 2026-11-11
The thread's quantitative results: 124 test questions, counting a reply as good if it stays in character without repeating itself. With 70+ neurons rewired most replies are good; full SFT scores 89%; the sparsest run matching SFT uses 89 neurons. The scatter plot shows success rate rising with neurons rewired.
More from Research
- Brynjolfsson's GDP-B paper quantifies the welfare value of free goods like Facebook, published in AEJ — soumitrashukla9 · 2026-11-12
- Cells rejuvenated to under 20: the case for Longevity Escape Velocity — Dr_Singularity · 2026-11-12
- Over 50 ICLR 2027 papers are named PRISM, and that says something about research naming — deliprao · 2026-11-12
- 37% on ARC-AGI-2 Kaggle With a 100M Looped Transformer Trained From Scratch, No LLM — rbhar90 · 2026-11-12
- Loom lens: sampling the MTP head like a language model to get tree-shaped activation readouts — voooooogel · 2026-11-12
- Interpreting all 1.1M MLP neurons in Qwen 27B with chain lens for just $5 — voooooogel · 2026-11-12