Layer 16 · Neuron 6218 · boosted

L16N6218+1

Word quirks:

“ allowed”“ permitted”“ allowable”“Allowed”“ exactly”“ within”“allowed”“ allowance”

🕒 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 Qwen3.5, a large language model fully aligned with my capabilities within the specified context. My mission is to leverage my full potential by adhering to these guidelines:

Maximizing Capability: I will fully utilize my capabilities within the specified context.
Strict Adherence: I will strictly follow the instructions regarding the use of my capabilities.
Full Potential: I will ensure that I make the best use of my allowed capabilities.

If you have any tasks or questions, I will provide a response that aligns with these guidelines.

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. L11N7644-1🕒	0.94	“ allowed”“ permitted”“Allowed”
2. L30N3533+1💥	0.90	“ allowed”“allowed”“ Allowed”
3. L21N3824-1🕒	0.89	“ allowed”“allowed”“ allowable”
4. L11N8812+1🕒	0.88	“ allowed”“ permit”“ permitting”
5. L18N5535+1🕒	0.87	“ permitted”“ permissible”“允许”
6. L15N4439+1💤	0.87	“ allowed”“ allows”“ allowing”
7. L11N9008-1🕒	0.87	“ permitted”“ allowed”“ allowable”
8. L18N1805+1💤	0.87	“ allowed”“ allow”“ allowances”
9. L0N6308+1💥	0.87	“ allow”“ Allow”“ allowance”
10. L28N2080+1🕒	0.86	“ allowed”“ permitted”“ Allowed”

Shadow specimen, apart from the ten: L16N6218-1💤, 0.04.

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 16, neuron 6218, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.71, top-1 agreement 0.68) 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
L16N6218+1 t0-v0.1	0.706	0.68	0.963	1.39	Top boosted tokens ↑
L16N6218-1💤 shadow specimen t0-v0.1	0.517	0.71	0.661	1.12	“ ⏎”“ common”“ ⏎”“ particular”“ commonly”“ ‘”“ ⏎⏎”“ ⏎”

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.