Layer 5 · Neuron 5198 · suppressed

L5N5198-1

Word quirks:

“ XML”“ xml”“XML”“ Xml”“ schema”“ SOAP”“ Schema”“Xml”

💥 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

<xs:element name="Citation" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<xs:annotation>
<xs:elementRef name="Citation" type="string"/>
</xs:annotation>
</xs:element>
<xs:element name="Citation" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<xs:annotation>
<xs:elementRef name="Citation" type="string"/>
</xs:annotation>
</xs:element>

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. L21N1927+1🕒	0.94	“ XML”“XML”“xml”
2. L14N8542-1🕒	0.94	“ XML”“ xml”“XML”
3. The Validator✅ L6N3404+1	0.94	“ XML”“ xml”“XML”
4. L3N5507+1💥	0.93	“ XML”“ xml”“ markup”
5. L25N5899+1🕒	0.93	“ XML”“.xml”“xml”
6. L29N5787-1💥	0.91	“竹”“ xml”“xml”
7. L5N3599-1💥	0.91	“ XML”“ xml”“XML”
8. L4N2947+1🕒	0.91	“ XML”“ HTTP”“ xml”
9. L1N3165-1💥	0.90	“ xml”“ namespace”“ xmlns”
10. L23N1360-1🕒	0.89	“XML”“,”“-Un”

Shadow specimen, apart from the ten: L5N5198+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 5, neuron 5198, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 3.17, 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
L5N5198+1🕒 shadow specimen t0-v0.1	1.258	0.54	1.532	1.79	“ `"”“ `”“ crypt”“ `'”“ metaphor”“ "”“ name”“ Linux”
L5N5198-1 t0-v0.1	3.169	0.30	3.611	2.92	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.