Layer 11 · Neuron 8021 · boosted

L11N8021+1

Word quirks:

“ something”“ various”“ stuff”“ sometimes”“ __”“ often”“ focused”“ things”

🕒 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 can help you with a wide range of tasks. For instance, I can assist with things like solving problems related to numbers, planning, organizing, or finding solutions that might involve guidance. While I don't have access to specific personal information that could be considered something I know about many users, including common challenges like optimizing workflows or improving efficiency.

While I'm not required to provide specific details about how I can assist with various types of problems, it's also worth considering scenarios where I can offer suggestions or ideas. For example, if you're looking for ways to improve productivity or manage time effectively, I might suggest strategies that could help

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. The Almost✅ L13N4915-1	0.82	“ sometimes”“ stuff”“ whatever”
2. L13N4009-1💤	0.80	“ sometimes”“ often”“ different”
3. L12N3431-1💤	0.80	“ somehow”“ something”“ whatever”
4. L12N2511+1🕒	0.77	“ sometimes”“ stuff”“ things”
5. L10N851-1🕒	0.76	“ usually”“ typically”“ often”
6. L9N3576+1🕒	0.75	“ things”“ usually”“ often”
7. L13N3273+1🕒	0.75	“ folks”“ usually”“ thing”
8. L12N4054-1💤	0.75	“ similar”“ interesting”“ sometimes”
9. L11N3713+1🕒	0.74	“ stuff”“ somehow”“ something”
10. L17N3080-1💤	0.73	“ something”“ maybe”“ sometimes”

Shadow specimen, apart from the ten: L11N8021-1💤, -0.09.

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 11, neuron 8021, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.22, 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
L11N8021+1 t0-v0.1	2.217	0.41	2.525	1.93	Top boosted tokens ↑
L11N8021-1💤 shadow specimen t0-v0.1	0.774	0.62	0.999	1.30	“ \"”“ ‘’”“ “”“ “”“ "”“ ””“ **"”“ ".”

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.