Layer 18 · Neuron 7310 · boosted

L18N7310+1

Word quirks:

“ provisions”“ legis”“ legisl”“ legislation”“ enactment”“ regulations”“ regulating”“ regulation”

🕒 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 model developed by Alibaba Group, known as the Tencent (or more accurately, I am part of the Alibaba Group ecosystem). My primary purpose is to provide a comprehensive framework for users to interact with the system, covering various aspects such as legal regulations, social norms, and ethical guidelines.

As an AI assistant, my mission is to:

Protect User Rights: Safeguard user privacy and ensure fair treatment in all interactions.
Promote Social Order: Uphold public order and moral standards, fostering a healthy environment.
Foster Positive Development: Encourage responsible

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. L6N5281-1🕒	0.93	“ legislative”“ legislation”“ laws”
2. L4N4033-1💥	0.93	“ legislation”“ regulation”“ appropriations”
3. L6N1631-1🕒	0.92	“ regulation”“ legislation”“ legal”
4. L5N6359+1🕒	0.91	“ legislation”“ legislative”“ policy”
5. L2N6887-1💥	0.91	“ legislative”“ legislation”“ legis”
6. L1N4250-1💥	0.91	“<\|im_end\|>”“ regulation”“ ley”
7. The Legislator✅ L3N2626-1	0.90	“ legislative”“ legislation”“ policy”
8. L3N1262+1🕒	0.90	“ statute”“ legislation”“ legislative”
9. L4N5834-1🕒	0.90	“ regulation”“ statute”“ statutes”
10. L6N2589+1🕒	0.89	“ legislation”“ legislative”“ legisl”

Shadow specimen, apart from the ten: L18N7310-1🕒, 0.16.

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 18, neuron 7310, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.74, top-1 agreement 0.64) 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
L18N7310+1 t0-v0.1	0.743	0.64	0.831	1.33	Top boosted tokens ↑
L18N7310-1🕒 shadow specimen t0-v0.1	0.741	0.71	0.908	1.32	“ losing”“ specialists”“ gaining”“ times”“ lost”“ gri”“ specialized”“ Specialists”

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.