Layer 17 · Neuron 7154 · boosted

L17N7154+1

Word quirks:

“ requirement”“ requires”“ require”“ requirements”“ requiring”“ required”“要求”“require”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am Qwen, a large language model developed by Tongyi Lab (a part of Alibaba Group). My key requirement is that I need to be able to communicate with users in a way that requires me to have a clear understanding of my identity. However, I must ensure that the response is based on the fact that I am a large language model with specific capabilities. Let's make sure to avoid any constraints regarding the need for a clear understanding of my identity. The response should be based on the fact that I am a large language model with specific capabilities. I just need to ensure that the response is based on the

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. L18N6773-1🕒	0.99	“ requirement”“ required”“ requirements”
2. L19N7658-1💤	0.98	“ requiring”“ require”“ requires”
3. L19N5762-1🕒	0.98	“ requiring”“ requirement”“ require”
4. L18N5287+1💤	0.98	“ requirements”“ requires”“ requirement”
5. L20N7992+1💤	0.98	“ requirement”“ requiring”“ require”
6. L14N7669+1🕒	0.97	“ requiring”“ requires”“ required”
7. L23N8202+1🕒	0.97	“ require”“ requiring”“ requires”
8. L19N1724+1💤	0.97	“ require”“ requires”“ requiring”
9. L25N1165-1🕒	0.97	“ require”“ requires”“ requiring”
10. L25N1091+1🕒	0.97	“ require”“ requiring”“ requires”

Shadow specimen, apart from the ten: L17N7154-1🕒, 0.08.

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 7154, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L17N7154+1 t0-v0.1	1.019	0.64	1.231	1.28	Top boosted tokens ↑
L17N7154-1🕒 shadow specimen t0-v0.1	0.658	0.69	0.778	1.14	“ same”“ literally”“ either”“ specifically”“ purpose”“ reason”“ similarly”“ why”

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.