Layer 17 · Neuron 6111 · boosted

L17N6111+1

Word quirks:

“ similar”“ even”“ possibly”“ akin”“ perhaps”“ similarly”“ Similar”“ analogous”

🕒 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 advanced language model designed to assist you with a wide range of tasks. Think of me as a versatile tool that can help you explore ideas, solve problems, or even generate creative content.

Here’s how I approach this mission:

Support Learning: Whether you need help understanding concepts, working through problems, or even learning new skills, I can guide you step-by-step.
Assist in Problem-Solving: From debugging code to analyzing data, I can apply logical reasoning to tackle challenges.
Creative Collaboration: You can use me as a co-creator—brainstorming ideas, developing

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. L9N788+1🕒	0.93	“ similar”“ similarly”“ comparable”
2. L17N7142+1🕒	0.91	“ similar”“ like”“ similarly”
3. L15N234+1🕒	0.91	“ similar”“ analogous”“ ”
4. L16N1163+1🕒	0.91	“ similar”“ similarly”“ same”
5. L14N7195+1💤	0.91	“ similar”“ Similar”“ similarly”
6. L13N7992-1💤	0.91	“ similar”“ similarly”“ Similar”
7. L15N164-1💤	0.91	“ similar”“ similarly”“ analogous”
8. L15N6902-1💤	0.91	“ similar”“ similarly”“ analogous”
9. L13N4416-1💤	0.90	“ similar”“ similarly”“ analogous”
10. L11N2099-1🕒	0.90	“ similar”“ similarly”“ analogous”

Shadow specimen, apart from the ten: L17N6111-1🕒, 0.18.

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 17, neuron 6111, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.69, top-1 agreement 0.71) 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
L17N6111+1 t0-v0.1	0.688	0.71	0.713	1.10	Top boosted tokens ↑
L17N6111-1🕒 shadow specimen t0-v0.1	1.098	0.57	1.325	1.87	“ mentioned”“ referenced”“ referring”“ refers”“ societies”“ example”“ recommendation”“ isn”

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.