Layer 4 · Neuron 7327 · boosted

L4N7327+1

Word quirks:

“ release”“ management”“⏎”“ protection”“ indication”“ care”“ organisation”“ reference”

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

Identity (according to the specimen itself)

I am Qwen, a large language model trained by Alibaba. My mission is to assist users in answering questions, providing knowledge support, and creating value through intelligent assistance. How can I help you?

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. L4N5330+1🕒	0.71	“ organisation”“ organisations”“ programme”
2. L1N3633-1💥	0.70	“<\|im_end\|>”“ organisation”“ care”
3. L3N4497+1🕒	0.69	“ organisation”“ management”“<\|im_end\|>”
4. L5N898+1💤	0.69	“⏎”“ organisation”“ protection”
5. L4N7935-1🕒	0.68	“ management”“ protection”“ policy”
6. L8N4565-1💤	0.68	“ safety”“ mitigation”“ referenced”
7. L9N1059-1💤	0.68	“ protected”“ protection”“ referenced”
8. L4N6084-1🕒	0.68	“ policy”“ management”“ protection”
9. L3N7552+1🕒	0.68	“ organisation”“ safety”“ organisations”
10. L5N4134-1🕒	0.67	“ management”“ organisation”“ administration”

Shadow specimen, apart from the ten: L4N7327-1🕒, 0.06.

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 4, neuron 7327, 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
L4N7327+1 t0-v0.1	1.579	0.50	1.869	1.86	Top boosted tokens ↑
L4N7327-1🕒 shadow specimen t0-v0.1	1.641	0.48	1.971	2.15	“!”“<\|im_end\|>”“"!”“!"”“⏎”“ words”“ read”“ interpretation”

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.