Layer 11 · Neuron 6390 · suppressed

L11N6390-1

Word quirks:

“ match”“ matching”“ similarity”“ identical”“ exact”“ matches”“ matched”“ similarities”

💤 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 mission is to provide seamless, intelligent assistance across a wide range of domains, ensuring that users can rely on me for accurate and efficient interactions.

Here's how I differ from other models:

Precise understanding: I understand nuances in context, tone, and intent.
Multilingual support: I support multiple languages including Chinese, English, Japanese, Korean, and French, as well as German and Portuguese.
Code generation: I can generate code snippets in various programming languages like Python, JavaScript, and Java.

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. L16N3118-1🕒	0.93	“ similar”“ matches”“ same”
2. L14N8952-1🕒	0.93	“ similar”“ same”“ matches”
3. L16N5630-1🕒	0.92	“ similar”“ match”“ matches”
4. L12N8941-1💤	0.91	“ similar”“ match”“ matching”
5. L14N5276+1🕒	0.91	“ similar”“ identical”“ matches”
6. L14N7591-1🕒	0.91	“ same”“ match”“ matches”
7. L10N6433-1🕒	0.91	“ similar”“ matching”“ match”
8. L17N7740+1🕒	0.91	“ matches”“ match”“ matching”
9. L14N5976+1🕒	0.90	“ matching”“ identical”“ matches”
10. L14N4017-1💤	0.89	“ same”“ matches”“ match”

Shadow specimen, apart from the ten: L11N6390+1🕒, -0.04.

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 11, neuron 6390, 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 0.67) 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
L11N6390+1🕒 shadow specimen t0-v0.1	0.850	0.63	1.172	1.26	“ typical”“ question”“ typically”“ your”“ requesting”“ asked”“ requests”“ request”
L11N6390-1 t0-v0.1	0.709	0.67	0.881	1.22	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.