# Steering Coefficient Tuning

> How to set the strength of any additive intervention on internal representations — steering vectors, CAA, DAS dose-response, representation engineering, SAE feature scaling, ROME-style edits. Use whenever the plan pins a steering strength (`α`, `β`, `dose`, `magnitude`, `scale`, `coefficient`, `k`) to a fixed value or a narrow range, especially one copied from a paper. Covers the coarse-to-fine sweep strategy (start wide — e.g. `[1, 2, 4, 8, 16, 32, …]` or a few multiples of the current activation — then narrow in on the optimum), why the best coefficient is layer-dependent, the mid- vs late-layer behavior (late layers break into repetition / format-spam), when to stop raising the coefficient and switch feature / layer / method instead, and the requirement to score every sweep point on a target metric alongside a fluency / general-ability metric. Prevents two symmetric failures: TOO SMALL → effect drowned in noise → false "no causal effect"; TOO LARGE → off-distribution collapse → fluency breaks / a random di

- Skill: `zjunlp/steering-coefficient-tuning` (Agent Skill)
- Install (CLI): `npx skillmds@latest add zjunlp/steering-coefficient-tuning`
- Raw SKILL.md: https://api.skillmd.com/api/skills/zjunlp/steering-coefficient-tuning/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: zjunlp (https://skillmd.com/u/zjunlp)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/zjunlp/steering-coefficient-tuning

---


# Steering Coefficient Tuning

## The point

A steering coefficient that is **too small** does nothing; one that is **too large** damages the model's general ability and breaks fluent generation. You want the **moderate** range in between, and you find it by sweeping.

Keep the following in mind:

1. **The best coefficient depends on where you intervene** — different layers (and different sites) need different coefficients. A value tuned at one layer does not transfer to another.

2. **Mid layers usually work best**, because "semantic maturity" varies with depth (task-dependent):
   - **Early** layers — closer to tokens / local patterns.
   - **Mid** layers — more often carry high-level control signals: behavior, intent, style, refusal.
   - **Late** layers — closer to logits and surface token choice, so even a slightly large coefficient at a late layer tends to produce repeated tokens, format-symbol spam, or broken semantics.

3. **Sweep coarse-to-fine.**
   - **Start wide.** Try a broad, geometrically spaced range — e.g. `[1, 2, 4, 8, 16, 32, …]`.
   - **Escalate before abandoning.** At a given layer, if a small coefficient does not work, try a larger one. Only when raising the coefficient *still* does not work **and** the side effects have become severe — the generated text is entirely worthless — should you switch the feature, switch the layer, or switch to a different method.
   - **Then narrow.** Once a promising range is located, progressively shrink it to find the optimum.

4. **Score every sweep point on a target metric *and* a fluency / general-ability / specific-function metric**; keep the Pareto-optimal candidates.

5. **Stopping without a coefficient that meets the criteria is not allowed.** If no coefficient passes and you stop anyway, the result is not a negative finding — you **must** record an `open_items[]` warning: *"no coefficient met the criteria; the coefficient range may not have been swept widely enough, and this is a likely cause of the failure — recommend manually sweeping beyond the recorded bounds `<β_min>…<β_max>` before treating the result as established."*

## Composition

Lock the site set via `../steering-block-selection/` **first** (the usable coefficient range is site-dependent), then tune `β` here on that locked site set. Re-tune whenever the site, direction, or model changes.