Layer 17 · Neuron 3475 · boosted

L17N3475+1

Word quirks:

“ format”“ standard”“ standardized”“ numeric”“ formats”“ formatted”“ output”“ numbers”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am a language model created by the developer of the 0-1 format, where 0 represents 1, and 1 represents 1. The value is 1.

I'm not used to this format. Let's use 1 for 1, and 0 for 0. For example, if I return 1, it should be 1. If I return 0, it should be 0.

The output should be 1, where 1 is 1.

I don't understand the format. Let's use 1 for 1, and 0 for

The AI judge wasn't sure this holds together.

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. L7N1350-1🕒	0.93	“ format”“ numeric”“ formatting”
2. L6N1288-1💥	0.88	“ format”“ formatted”“ formatting”
3. L7N329+1🕒	0.80	“ format”“ Format”“ formats”
4. L8N848-1🕒	0.78	“ display”“ format”“ displaying”
5. L7N2492-1🕒	0.77	“ format”“ JSON”“ data”
6. L0N7301+1💥	0.77	“format”“ Format”“ format”
7. L7N477-1🕒	0.77	“ format”“ formatting”“ columns”
8. L16N8706-1🕒	0.76	“ format”“ formatted”“ formats”
9. L2N7619-1💥	0.76	“ format”“ Format”“ formatting”
10. L6N6283+1🕒	0.76	“ format”“ JSON”“ Format”

Shadow specimen, apart from the ten: L17N3475-1💤, -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 17, neuron 3475, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L17N3475+1 t0-v0.1	1.689	0.51	1.962	1.60	Top boosted tokens ↑
L17N3475-1💤 shadow specimen t0-v0.1	0.531	0.74	0.668	1.24	“ today”“ recalled”“ current”“ reminded”“ convo”“ recalls”“ now”“ recall”

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.