ezyang: sparse autoencoders can be applied per residual add, not per block
ezyang · x · 2026-09-26
ezyang (PyTorch lead) shares a small observation: SAC is conventionally applied at transformer block granularity, but since many SAC policies end up always saving the result of a residual add — the sole input into attn/mlp — you can apply SAC to each residual add individually.
More from Research
- Researchers teach LLMs to find interesting theorems, boosting interestingness 4.3x — CatAstro_Piyush · 2026-09-26
- AI proposes catalyst a 12-year expert called wrong; it runs 1,000+ hours with little loss — CatAstro_Piyush · 2026-09-26
- Anthropic: Claude solves nine-loop scattering amplitudes, breaking the eight-loop record — AnthropicAI · 2026-09-26
- Both "grep is all you need" and "BM25 is all you need" Papers Just Got Accepted — lintool · 2026-09-26
- Greenblatt Warns Latent Reasoning Architectures Would Sharply Raise Misalignment Risk — anmarasovic · 2026-09-26
- Reasonable Team Publishes TLA+ Tutorial: Not a Silver Bullet, AI Agents Could Change That — fhuszar · 2026-09-26