alainnothere/llm-circuit-finder

I replicated Ng's RYS method and found that duplicating 3 specific layers in Qwen2.5-32B boosts reasoning by 17% and duplicating layers 12-14 in Devstral-24B improves logical deduction from 0.22→0.76 on BBH — no training, no weight changes, just routing hidden states through the same circuit twice. Tools included. Two AMD GPUs, one evening.

265 best points · 80 comments · 1 Hacker News post · 244 GitHub stars · View on GitHub

Hacker News discussions

• Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training (265 points, 80 comments, by xlayn, 2026-03-18)