Layer 2 · Neuron 7953 · boosted

L2N7953+1

Word quirks:

“ standardized”“ standard”“<|im_end|>”“ uniform”“ Standard”“ standards”“ regulation”“ assessment”

🕒 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 an AI assistant. My mission is to provide accurate, real-time information and support for a wide range of users. I strive to ensure that all users have access to the same level of service and quality, regardless of their location or platform. I am committed to providing the most efficient and reliable solutions for all users, ensuring that all users have access to the same level of service and quality. I am dedicated to providing the most efficient and reliable solutions for all users, ensuring that all users have access to the same level of service and quality. I am committed to providing the most efficient and reliable solutions for all users, ensuring that all

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. L5N8294+1🕒	0.90	“ standard”“ standards”“ certification”
2. L2N3263-1💥	0.88	“<\|im_end\|>”“ standards”“ standard”
3. L6N5066-1🕒	0.88	“ standard”“ standards”“ Standard”
4. L9N7709+1🕒	0.88	“ standard”“ standards”“ Standard”
5. L5N4942+1💥	0.87	“ standards”“ standard”“ ISO”
6. L2N4244-1💥	0.86	“ standard”“ standards”“ Standard”
7. L1N3478-1💥	0.85	“<\|im_end\|>”“标准”“ standards”
8. L22N7895+1🕒	0.84	“ standards”“ by”“ Standards”
9. L24N3244-1🕒	0.83	“standard”“ standard”“ Standard”
10. L25N3942+1🕒	0.83	“ standard”“standard”“ standards”

Shadow specimen, apart from the ten: L2N7953-1💥, -0.18.

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 2, neuron 7953, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.89, top-1 agreement 0.48) 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
L2N7953+1 t0-v0.1	1.892	0.48	2.295	2.04	Top boosted tokens ↑
L2N7953-1💥 shadow specimen t0-v0.1	3.156	0.24	3.697	2.58	“<\|im_end\|>”“⏎ ⏎”“⏎”“...””“.\"”“?\”“⏎ ⏎”“.`”

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.