Layer 7 · Neuron 1291 · suppressed

L7N1291-1

Word quirks:

“<|im_end|>”“ usually”“ safety”“ analysis”“ reflecting”“ confirming”“ reflected”“ confirmation”

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

Identity (according to the specimen itself)

I am Qwen3.5, and my mission is to help users by providing high-quality, accurate, and actionable information.

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. L10N8746-1🕒	0.87	“ typically”“ confirmation”“ reflected”
2. L12N5645-1🕒	0.80	“ confirming”“ confirmation”“ confirmed”
3. L8N9125+1💤	0.79	“ confirmation”“ confirming”“ confirmed”
4. L11N889+1🕒	0.76	“<\|im_end\|>”“ clarification”“ confirmation”
5. L11N454+1🕒	0.74	“ indeed”“ reflective”“ reflecting”
6. L9N867+1🕒	0.72	“ interaction”“ reflected”“ confirmation”
7. L9N3485+1🕒	0.71	“ typical”“ indicating”“ reflected”
8. L9N5697+1🕒	0.70	“ reference”“ referenced”“ references”
9. L11N7536-1🕒	0.70	“<\|im_end\|>”“ always”“ myself”
10. L11N2075+1🕒	0.70	“ confirming”“ confirmation”“ indicating”

Shadow specimen, apart from the ten: L7N1291+1💤, -0.08.

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 7, neuron 1291, 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
L7N1291+1💤 shadow specimen t0-v0.1	0.639	0.68	0.839	1.30	“⏎”“⏎⏎”“⏎ ⏎”“ ⏎”“ ⏎⏎”“⏎ ⏎”“ ⏎”“\n”
L7N1291-1 t0-v0.1	1.227	0.54	1.454	1.79	Top boosted tokens ↑

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.