Layer 15 · Neuron 4733 · boosted

L15N4733+1

Word quirks:

“ only”“ needs”“ requires”“ needed”“ if”“ needing”“ need”“ require”

🕒 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 developed by Alibaba Group. My mission is to provide high-quality assistance through advanced capabilities such as reasoning, visual analysis, and code generation. I excel in complex tasks requiring deep understanding and precise execution, particularly when handling multi-step problems or generating code, which involves writing and debugging. I also perform well in specialized fields like medicine and law, demonstrating strong performance in these areas. I am designed to be helpful, efficient, and accurate, with a focus on being a versatile assistant that can handle a wide range of queries.

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. L17N7756+1🕒	0.96	“ needing”“ needs”“ need”
2. L16N8075-1🕒	0.96	“ requires”“ needs”“ needing”
3. L17N8508-1🕒	0.95	“ needing”“ need”“ requirements”
4. L13N6616+1🕒	0.95	“ needs”“ requires”“ needed”
5. L7N7006-1🕒	0.95	“<\|im_end\|>”“ needs”“ requires”
6. L12N8901+1🕒	0.95	“ needed”“ needs”“ needing”
7. L22N4148-1🕒	0.94	“ needing”“ needed”“ needs”
8. L17N1271-1💤	0.94	“ requires”“ must”“ requirement”
9. L11N5732+1🕒	0.93	“ needs”“ needed”“ needing”
10. L17N7358+1💤	0.93	“ needing”“ requires”“ require”

Shadow specimen, apart from the ten: L15N4733-1🕒, 0.22.

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 15, neuron 4733, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.90, top-1 agreement 0.65) 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
L15N4733+1 t0-v0.1	0.900	0.65	1.131	1.21	Top boosted tokens ↑
L15N4733-1🕒 shadow specimen t0-v0.1	0.663	0.68	0.859	1.20	“ because”“ still”“ too”“ consistent”“ tools”“ better”“ customers”“ new”

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.