Layer 4 · Neuron 4986 · boosted

L4N4986+1

Word quirks:

“ school”“ schools”“ elementary”“ grades”“ curriculum”“ grade”“ teachers”“ education”

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

Identity (according to the specimen itself)

Mathematics Curriculum: A Framework for Teaching

Core Concept:
In the mathematics curriculum, students learn a broad range of mathematics education. This framework provides a comprehensive approach to teaching mathematics.

Mission School:
A mission school is a school that serves all students in a single building or campus.

The AI judge wasn't sure this holds together.

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. L8N4966+1💤	0.93	“ schools”“ education”“ educational”
2. L7N3981+1🕒	0.92	“ education”“ school”“ schools”
3. The Truant Officer✅ L8N3994+1	0.91	“ homeschool”“ education”“ school”
4. L6N4556-1🕒	0.91	“ school”“ schools”“ tuition”
5. L4N6739-1🕒	0.91	“ education”“ school”“ curriculum”
6. L7N2716+1🕒	0.91	“ education”“ educators”“ curriculum”
7. L3N1763-1🕒	0.90	“ education”“ academic”“ school”
8. L8N5648+1🕒	0.90	“ education”“ school”“ educators”
9. L6N4843-1💤	0.90	“ school”“ educational”“ teachers”
10. L7N4319+1🕒	0.89	“ curriculum”“ education”“ educational”

Shadow specimen, apart from the ten: L4N4986-1🕒, -0.08.

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 4986, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 4.00, top-1 agreement 0.30) 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
L4N4986+1 t0-v0.1	3.999	0.30	4.539	2.79	Top boosted tokens ↑
L4N4986-1🕒 shadow specimen t0-v0.1	1.579	0.55	1.949	1.96	“!”“ !”“!"”“"!”“!””“ maths”“ ‘”“!*”

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.