Researchchain of thoughtlinear steerabilityneel nanda
Project Explores Linear Steerability in Chain-of-Thought
2.8
A ~20-hour application project to Neel Nanda's MATS stream, posted on LessWrong, explores linear steerability in continuous chain-of-thought reasoning and publishes preliminary, minimally edited results to the MATS community from this short investigation.
Key Points
- 1Explores linear steerability in continuous chain-of-thought reasoning during a ~20-hour project for MATS.
- 2Investigates potential controllability of reasoning trajectories, which could inform interpretability and steering research.
- 3Signals preliminary directions for follow-up validation in model steering and interpretability.
Scoring Rationale
Small, single-source MATS project with limited novelty, scored low due to RSS-only input and minimal corroborating detail.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
