Layer 11 · Neuron 8507 · boosted

L11N8507+1

Word quirks:

“ change”“ changed”“ changes”“ changing”“ transformation”“ Change”“ altered”“ alteration”

🕒 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, a large language model with the ability to help you write more and do more. I am an AI assistant created by Alibaba Cloud. My mission is to empower individuals and organizations in their daily lives and work.

Here are some of the key changes:

Enhanced Language Understanding: I can understand and process text data in a way that allows for more natural and efficient communication.
Improved Decision-Making: I can assist in making better decisions by providing insights into how information is processed and utilized.
New Capabilities: I can help you create new experiences

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. L16N7685-1🕒	0.97	“ changed”“ change”“ modification”
2. L13N7833+1🕒	0.97	“ changed”“ change”“ changes”
3. L6N7290-1🕒	0.96	“ change”“ changes”“ changed”
4. L9N989+1🕒	0.96	“ change”“ changes”“ changed”
5. L23N2283+1🕒	0.96	“ changing”“ change”“ changed”
6. L13N7365-1🕒	0.95	“ changes”“ changed”“ change”
7. L31N772+1💥	0.95	“ change”“ changes”“ Change”
8. L21N6190+1💤	0.95	“ change”“ changed”“ changes”
9. L24N2638-1🕒	0.94	“ changed”“ changing”“ change”
10. L22N6705+1🕒	0.94	“ change”“ changed”“ changing”

Shadow specimen, apart from the ten: L11N8507-1💤, 0.11.

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 8507, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.95, top-1 agreement 0.62) 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
L11N8507+1 t0-v0.1	0.952	0.62	1.265	1.65	Top boosted tokens ↑
L11N8507-1💤 shadow specimen t0-v0.1	0.690	0.67	0.865	1.20	“ specific”“ simulated”“ multiple”“ sim”“ support”“ welcome”“ ranges”“ multiplied”

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.