Layer 11 · Neuron 7933 · suppressed

L11N7933-1

Word quirks:

“ communication”“ choice”“ mail”“ communicated”“ letters”“ comparison”“ communications”“ communicating”

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

Identity (according to the specimen itself)

The reply didn't hold together

I need to use the contact for this message in a form that can be chosen by any kind of contact that can be used through a video call, using the channel of their choice, which can be chosen by any kind of message through a video call, using a line of action through a video call, using a line of action through a video call, using a line of action through a video call, using a line of action through a video call, using a line of action through a video call, using a line of action through a video call, using a line of action through a video call, using a line of action through a

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. L10N7629+1🕒	0.88	“ communication”“ communicating”“ communicate”
2. L2N6258+1💥	0.87	“ communication”“<\|im_end\|>”“ communicate”
3. L20N4781+1🕒	0.87	“ communication”“ communications”“Communication”
4. L19N4092-1💤	0.86	“ communications”“ communicated”“ communicating”
5. L26N8831+1🕒	0.86	“ Communication”“ communication”“ Communications”
6. L9N4990-1🕒	0.85	“ info”“ communication”“ communicated”
7. L21N5022-1💤	0.85	“ communication”“ communicates”“ communicating”
8. L9N2623-1🕒	0.85	“ communication”“ messages”“ communications”
9. L19N8183+1🕒	0.85	“ communicated”“ communication”“ communicating”
10. L5N8300-1🕒	0.84	“ communication”“ messages”“ message”

Shadow specimen, apart from the ten: L11N7933+1💤, 0.12.

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 11, neuron 7933, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 3.28, top-1 agreement 0.26) 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
L11N7933+1💤 shadow specimen t0-v0.1	0.739	0.66	0.950	1.23	“ actually”“ actual”“ specialized”“ directly”“ specific”“ general”“ WITHOUT”“ NOT”
L11N7933-1 t0-v0.1	3.281	0.26	3.687	2.76	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.