Layer 15 · Neuron 8325 · suppressed

L15N8325-1

Word quirks:

“ hidden”“ concealed”“ secret”“ Hidden”“ hiding”“ secretly”“hidden”“ but”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Qwen3.5, a large language model developed by Tongyi Lab. My mission is to assist users in completing tasks across various domains, including but not limited to:

Writing: Crafting high-quality content such as stories, emails, scripts, and reports.
Math: Solving complex mathematical problems with step-by-step reasoning.
Coding: Writing, debugging, and optimizing code for applications like web development, data analysis, and automation.
Logical Reasoning: Analyzing logical arguments and solving puzzles that require advanced problem-solving skills.
Language Understanding: Comprehending

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. L12N6834-1🕒	0.94	“ hidden”“ secretly”“ concealed”
2. L19N7359+1🕒	0.93	“ Hidden”“ hidden”“ conceal”
3. L2N3709-1💥	0.93	“<\|im_end\|>”“ hidden”“ hide”
4. L16N8422+1🕒	0.91	“ covert”“ stealth”“ secrecy”
5. L19N7839-1🕒	0.90	“ hidden”“ concealed”“hidden”
6. L21N6345+1🕒	0.90	“ hidden”“隐藏”“ conceal”
7. L13N8581+1🕒	0.90	“ hidden”“ hiding”“隐藏”
8. L20N5917-1🕒	0.90	“ hidden”“ concealed”“hidden”
9. L19N5439-1🕒	0.89	“ hidden”“ concealed”“隐藏的”
10. L11N4526-1🕒	0.89	“ hidden”“ fake”“ concealed”

Shadow specimen, apart from the ten: L15N8325+1🕒, 0.21.

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 15, neuron 8325, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.59, top-1 agreement 0.72) 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
L15N8325+1🕒 shadow specimen t0-v0.1	0.633	0.72	0.873	1.08	“ nothing”“ NOTHING”“ THIS”“ sim”“ sometimes”“ unrelated”“ ONLY”“ NOT”
L15N8325-1 t0-v0.1	0.592	0.72	0.824	1.27	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.