Layer 9 · Neuron 7949 · suppressed

L9N7949-1

Word quirks:

“ reference”“ general”“ references”“ generic”“ describing”“ description”“ referenced”“ language”

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

Identity (according to the specimen itself)

Hello! I am Qwen3.5, a large language model developed by Alibaba Cloud. As an AI assistant, I don't have a specific "mission" in the sense of role-playing or claiming a single identity, but rather I am designed to be a versatile tool that can assist with various tasks.

However, if you'd like to use this as a template for specific scenarios (such as handling data migration, training new models, or optimizing performance), here are some examples of how you might interact with me:

For general knowledge: "I need help understanding the underlying technology."
**For specific

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. L7N1980+1💤	0.87	“ referenced”“ global”“ defining”
2. L9N1401-1💤	0.82	“ description”“ describing”“ referenced”
3. L15N8020-1💤	0.82	“ referenced”“ reference”“ description”
4. L9N8176-1💤	0.81	“ myself”“<\|im_end\|>”“ references”
5. L5N4768+1🕒	0.80	“ reference”“ referenced”“ filtered”
6. L12N8997-1💤	0.80	“ reference”“ referenced”“ description”
7. L11N6000+1🕒	0.80	“ referenced”“ info”“ relevant”
8. L11N1285-1🕒	0.80	“ description”“ descriptions”“ reference”
9. L7N5728-1🕒	0.80	“ descriptions”“ description”“ referenced”
10. L7N6263+1💤	0.79	“ description”“ reference”“ referenced”

Shadow specimen, apart from the ten: L9N7949+1🕒, 0.12.

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 9, neuron 7949, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.93, top-1 agreement 0.66) 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
L9N7949+1🕒 shadow specimen t0-v0.1	1.020	0.62	1.251	1.38	“ now”“ your”“ control”“ response”“ constrained”“ sure”“ outputs”“ output”
L9N7949-1 t0-v0.1	0.926	0.66	1.077	1.18	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.