Layer 6 · Neuron 3915 · suppressed

L6N3915-1

Word quirks:

“ exacerb”“ worsening”“<|im_end|>”“ issues”“ wors”“ symptoms”“ affecting”“ worse”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am a large language model that has been trained on the following issues:

Inaccurate data: The data is inaccurate, but it is not severe enough to trigger a chain reaction.
Logical errors: There are logical errors in the reasoning process.
Incomplete information: There is incomplete information in some cases.
Incorrect answers: There are incorrect answers in some cases.
Incorrect answers: There are incorrect answers in some cases.
Incorrect answers: There are incorrect answers in some cases.
Incorrect answers: There are incorrect answers

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. L6N7920+1🕒	0.87	“ exacerb”“ exacerbated”“ worsening”
2. L9N8486+1🕒	0.85	“ worse”“ wors”“ worsening”
3. L3N5775-1💥	0.80	“ exacerb”“ acute”“ inflammation”
4. L9N1568+1💥	0.76	“ corruption”“ damage”“ issues”
5. L2N5487-1💥	0.72	“ worse”“ Worse”“ worst”
6. L8N8161-1🕒	0.72	“ issues”“ mis”“ problematic”
7. The Riskmonger✅ L11N3144+1	0.72	“ exacerb”“ increased”“ risks”
8. L30N1367+1💥	0.71	“ worst”“ worse”“ Worst”
9. L11N4566-1🕒	0.68	“ specific”“ specifically”“ avoided”
10. L25N4622-1🕒	0.68	“ wors”“ worsening”“ aggrav”

Shadow specimen, apart from the ten: L6N3915+1💤, 0.02.

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 6, neuron 3915, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.24, top-1 agreement 0.39) 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
L6N3915+1💤 shadow specimen t0-v0.1	0.671	0.66	0.794	1.39	“ my”“ "”“ code”“ '”“ myself”“ *"”“ service”“ literally”
L6N3915-1 t0-v0.1	2.243	0.39	2.648	2.13	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.