Layer 5 · Neuron 2948 · boosted

L5N2948+1

Word quirks:

“<|im_end|>”“ teacher”“ classroom”“ teachers”“ lesson”“ teaching”“ Teacher”“ curriculum”

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

Identity (according to the specimen itself)

Teacher: I am a teacher.
Mission: To teach a lesson.

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. L7N5510-1🕒	0.96	“ classroom”“ teaching”“ teacher”
2. L7N5458-1🕒	0.96	“ classroom”“ teacher”“ students”
3. L19N7014-1🕒	0.95	“ educators”“ instructors”“ classroom”
4. L8N1619-1🕒	0.95	“ lesson”“ teacher”“ students”
5. L4N2602+1💥	0.94	“ teachers”“ teacher”“ pedag”
6. L8N4843-1🕒	0.94	“ teacher”“ teachers”“ Teacher”
7. L18N1705-1🕒	0.93	“ curriculum”“ classroom”“ teacher”
8. L8N5612+1🕒	0.93	“ teachers”“ students”“ school”
9. L7N4161+1🕒	0.93	“ lesson”“ teacher”“ lessons”
10. L10N3467+1🕒	0.92	“ school”“ teacher”“ educational”

Shadow specimen, apart from the ten: L5N2948-1🕒, 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 5, neuron 2948, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.68, top-1 agreement 0.35) 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
L5N2948+1 t0-v0.1	2.682	0.35	3.186	2.23	Top boosted tokens ↑
L5N2948-1🕒 shadow specimen t0-v0.1	1.064	0.53	1.246	1.52	“ standard”“ usually”“ here”“ info”“ usual”“ docs”“ format”“ ‘”

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.