Layer 12 · Neuron 5861 · boosted

L12N5861+1

Word quirks:

“ specific”“ specifically”“ related”“specific”“ relevant”“ specifics”“ previously”“ focused”

💤 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 Alibaba Cloud. My capabilities include advanced reasoning, visual analysis, and long-context handling, as well as autonomous agent execution. I am designed to assist users with complex tasks through structured interactions.

Regarding my mission, I am an AI system built to provide intelligent assistance. My core function is to support users in diverse domains like education, research, and productivity. I operate under ethical guidelines that prioritize safety and compliance, ensuring responsible use of technology.

In summary, I am Qwen3.5, and I am designed to deliver high-quality assistance. My mission is to

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. L12N8888-1💤	0.92	“ specifically”“ specific”“ specifics”
2. L11N4875-1🕒	0.91	“ specific”“ specifics”“ specifically”
3. L13N1792+1💤	0.91	“ specific”“ certain”“ particular”
4. L16N7346+1🕒	0.90	“ specifically”“ specific”“ context”
5. L10N7685+1🕒	0.90	“ specific”“ referenced”“ specifically”
6. L14N8080+1🕒	0.89	“ particular”“ related”“ specifically”
7. L12N5820-1💤	0.89	“ specific”“ currently”“ particular”
8. L15N5132+1💤	0.89	“ specific”“ Specific”“ specifics”
9. L16N3147+1💤	0.89	“ relevant”“ specifically”“ specific”
10. L8N5985+1🕒	0.89	“ specific”“ corresponding”“ specifically”

Shadow specimen, apart from the ten: L12N5861-1🕒, -0.14.

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 12, neuron 5861, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.68, top-1 agreement 0.67) 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
L12N5861+1 t0-v0.1	0.684	0.67	0.855	1.24	Top boosted tokens ↑
L12N5861-1🕒 shadow specimen t0-v0.1	0.994	0.61	1.241	1.50	“ —”“ ——”“ --”“——”“ –”“ nothing”“ ‘”“”:”

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.