Layer 3 · Neuron 908 · boosted

L3N908+1

Word quirks:

“ basically”“ meaning”“<|im_end|>”“ meanings”“ means”“ limited”“ refers”“ \"”

🕒 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, a large language model. I am ready to help. I am Qwen.

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. L6N2992+1💥	0.88	“ meaning”“ referring”“ basically”
2. L4N6780-1🕒	0.85	“ referring”“ means”“ meaning”
3. L10N2456-1🕒	0.85	“ meaning”“ literally”“ means”
4. L13N7397+1🕒	0.85	“ means”“ basically”“ saying”
5. L5N2309+1💥	0.85	“ meaning”“ meanings”“ means”
6. L12N573-1💤	0.84	“ refers”“ meaning”“ refer”
7. L14N4030+1🕒	0.83	“ referring”“ refers”“ meaning”
8. L20N6751+1💤	0.82	“ referring”“ meaning”“ mean”
9. L9N8887-1🕒	0.82	“ basically”“ means”“ signifies”
10. L15N6078+1🕒	0.82	“ means”“ meaning”“ mean”

Shadow specimen, apart from the ten: L3N908-1💤, 0.14.

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 3, neuron 908, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.24, top-1 agreement 0.41) 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
L3N908+1 t0-v0.1	2.242	0.41	2.675	2.58	Top boosted tokens ↑
L3N908-1💤 shadow specimen t0-v0.1	0.951	0.66	1.186	1.62	“ ‘”“ precisely”“‘”“ precise”“ '”“ your”“ ’”“ craft”

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.