Layer 4 · Neuron 6535 · suppressed

L4N6535-1

Word quirks:

“ manager”“ analyst”“ job”“ specialist”“ jobs”“ consultant”“ specialty”“ management”

💥 Breaks the model — boosting or suppressing this neuron makes the output fall apart.

Identity (according to the specimen itself)

I am a professional with a mission to support the organization in various roles. I work closely with teams to ensure successful outcomes and provide guidance on projects. My responsibilities include managing tasks, coordinating activities, and assisting with operations. I also handle administrative duties and contribute to strategic initiatives. I am skilled in planning and execution, with a focus on analysis and reporting. I provide customer service and technical support, and I assist in sales and marketing. I am responsible for project management and team leadership, and I serve as a consultant and trainer. I work in human resources and finance, and I manage risk and compliance. I am an IT specialist and systems

Nearest neighbors

Specimens with similar replies or word quirks. Similarity runs from −1 to 1; 1 is identical. These are leads, not shared mechanisms. How profiles are built.

By word quirks Profile recipe

Specimen	Similarity	Word quirks
1. L8N4104+1🕒	0.84	“ admin”“ professionally”“ specialist”
2. L9N4866+1🕒	0.83	“ ⏎”“ job”“ expert”
3. L3N6846+1💥	0.81	“ manager”“ contractor”“ management”
4. The HR Department✅ L6N7811-1	0.80	“ manager”“ team”“ staffing”
5. L3N4810+1💥	0.80	“ job”“ manager”“ management”
6. L3N5134+1💥	0.79	“ manager”“ role”“ analyst”
7. L9N9121-1🕒	0.78	“ specialization”“ freelance”“ job”
8. L8N8784-1🕒	0.78	“ officer”“ Officer”“ specialist”
9. L9N8642+1🕒	0.77	“ professionally”“ professional”“ role”
10. L8N2648+1🕒	0.75	“ specialised”“⏎”“ profession”

Shadow specimen, apart from the ten: L4N6535+1💤, -0.10.

Cousins · specimens sharing a neuron

No specimen with replies shares this neuron yet.

Technical details

Quick scan: measure prediction shifts on fixed reference text. Short test: generate replies to a small set of prompts. Longer test: extend to a larger set. Full test: run the full prompt suite. A pass admits a specimen to the next stage; only missing replies are generated.

Layer 4, neuron 6535, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.66, top-1 agreement 0.36) is outside the band that feeds the short test (KL 1 to 2.5, agreement 0.5 to 0.65), so it stays at the quick scan unless someone queues it by hand. A hand pick goes straight to the full test. See Method for the tiers.

Quick scan · prediction measurements

Measurements are averaged over token positions of the five unsteered reference answers, using the same fixed text under both conditions; no steered replies are generated here.

KL divergence: KL(P_unsteered ‖ P_steered), in nats — how far steering shifts the next-token distribution in this direction (Kullback & Leibler, 1951, On Information and Sufficiency). Unsteered ≈ 0.001.
Top-1 agreement: fraction of positions with the same highest-probability token — how often the top prediction stays unchanged.
ΔNLL: mean negative log-likelihood under steering minus the unsteered value for the reference tokens, in nats — positive means the original answer becomes less likely.
Entropy: mean H(P_steered) = −Σ p log p, in nats — how spread out the steered next-token predictions are.

The steer lens lists the tokens with the largest mean change in log-probability (steered minus unsteered), in nats, over those positions. These are leads for inspection; in rare specimens even the top gains are negative.

Setting	KL divergence	Top-1 agreement	ΔNLL	Entropy	Steer lens
L4N6535+1💤 shadow specimen t0-v0.1	0.818	0.65	0.916	1.35	“ —”“ ——”“ *”“⏎”“ ‘”“ organisation”“ ⏎”“ '”
L4N6535-1 t0-v0.1	2.660	0.36	3.035	2.61	Top boosted tokens ↑

Method

A specimen is a fixed steering configuration of Qwen3.5-4B: one MLP neuron or a pair, each with a signed dose. Unsteered is the control. L identifies the layer; N, the neuron.

Additive activation steering. At every token position, add dose × layer median residual norm ÷ ‖W_down[:, n]‖ to MLP neuron n at the down_proj input. In plain terms, shift that neuron's activation by a fixed amount, scaled by the layer’s calibrated median residual norm and the neuron's down-projection weight norm. Positive doses push up; negative doses push down. Dose is normalized intervention size, not a percentage or probability.