Layer 17 · Neuron 5155 · suppressed

L17N5155-1

Word quirks:

“ justified”“ appropriately”“ appropriate”“ sufficient”“ valid”“ analyse”“ properly”“ sufficiently”

🕒 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 Qwen3.5, an expert-level assistant with a comprehensive understanding of the task. My mission is to provide accurate, efficient, and high-quality responses to all types of questions, whether they require deep reasoning, creative writing, or precise computation. I aim to be a reliable, versatile tool that can effectively communicate, solve problems, and support users in their personal and professional endeavors.

For any specific question, feel free to ask for details, and I will ensure clarity and correctness in my response.

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. L14N7588+1🕒	0.89	“ suitable”“ appropriately”“ acceptable”
2. L17N7044-1💤	0.87	“ enough”“ sufficient”“ sufficiently”
3. L19N5277-1🕒	0.87	“ appropriate”“ appropriately”“ sufficient”
4. L16N5579+1🕒	0.86	“ adequately”“ appropriately”“ clearly”
5. L10N8366-1🧪	0.86	“ appropriately”“ correctly”“ properly”
6. L12N5107-1🕒	0.86	“ appropriately”“ exemplary”“ appropriate”
7. L1N1529+1💥	0.85	“<\|im_end\|>”“ reasonable”“ proper”
8. L15N7098+1🕒	0.85	“ suitable”“ appropriate”“ compatible”
9. L15N7323-1🕒	0.85	“ suitable”“ appropriate”“ this”
10. L15N6083+1🕒	0.85	“ typical”“ suitable”“ enough”

Shadow specimen, apart from the ten: L17N5155+1🕒, 0.17.

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 17, neuron 5155, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.71, top-1 agreement 0.68) 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
L17N5155+1🕒 shadow specimen t0-v0.1	0.639	0.69	0.803	1.26	“ current”“ representing”“ representation”“ currently”“ represented”“ Current”“ represent”“ representations”
L17N5155-1 t0-v0.1	0.707	0.68	0.814	1.24	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.