Anthropic model card discussion points to grader awareness and weak SAE results
burny_tech · x · 2026-07-22
- The post discusses an Anthropic model card section on grader awareness in behavioral coding environments, based on mechinterp-style analysis.
- The image shows examples where a model seems to realize its output will be reviewed, and the post suggests natural language autoencoders (NLAs) and steering vectors work better than the current sparse autoencoder (SAE) setup in this context.
- It also notes that the reported experiments appear small and under-documented, so the science may be incomplete or partly cherry-picked.
- A key distinction in the discussion is that the intervention is about inference-time steering, not training-time changes.
More from Research
- LightMTP: lightweight latent multi-token prediction boosts LLM planning in pretraining — gdm3000 · 2026-10-27
- EMNLP study: AI literary translation reads fine but varies more, even within one book — tuvllms · 2026-10-27
- Lab monkeys at $30-80K each: a 4-year breeding cycle is the real preclinical drug bottleneck — zebird0 · 2026-10-27
- OpenAI Foundation funds UNC tumor antigen project, hiring senior experimental scientist — iskander · 2026-10-27
- Steering attention heads toward honesty: up moral patienthood, down psychopathy — edelwax · 2026-10-27
- Are transcoders overkill? New paper argues plain neurons suffice for circuit analysis — burny_tech · 2026-10-27