Layer 7 · Neuron 5029 · boosted

L7N5029+1

Word quirks:

“⏎”“ myself”“<|im_end|>”“ focus”“ itself”“ my”“ response”“ referenced”

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

Identity (according to the specimen itself)

I am Qwen3.5, a large language model developed by Alibaba Cloud. My mission is to provide high-quality, accurate, and helpful assistance in solving complex problems for users, ensuring that interactions remain focused on delivering value through clarity and precision. I am designed to support efficient, reliable, and meaningful engagement in all my responses.

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. L7N8092+1🕒	0.85	“⏎”“ referenced”“ myself”
2. L8N8345+1💤	0.79	“ my”“ reference”“ discussion”
3. L15N1852+1🕒	0.79	“<\|im_end\|>”“ my”“\n”
4. L9N4264-1🕒	0.78	“⏎”“ referenced”“ my”
5. L4N7445-1🕒	0.77	“⏎”“<\|im_end\|>”“ manager”
6. L6N5762-1💤	0.77	“ I”“ my”“ response”
7. L7N8307+1💤	0.76	“ my”“ myself”“⏎”
8. L6N8725-1🕒	0.76	“ my”“⏎”“ response”
9. L8N6150-1💤	0.76	“ my”“ referenced”“ I”
10. L8N7029+1💤	0.76	“ ⏎”“ ⏎⏎”“ myself”

Shadow specimen, apart from the ten: L7N5029-1💤, 0.17.

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 7, neuron 5029, 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
L7N5029+1 t0-v0.1	1.667	0.50	1.925	1.80	Top boosted tokens ↑
L7N5029-1💤 shadow specimen t0-v0.1	0.826	0.63	1.008	1.62	“ now”“ fire”“ '”“ your”“ _____”“ ____”“ ‘”“ __”

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.