Layer 7 · Neuron 696 · suppressed

L7N696-1

Word quirks:

“ ‘”“⏎”“ *“”“ '”“ **“”“<|im_end|>”“ “”“ \"”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Qwen3.5, an advanced AI assistant designed to help users navigate complex problems. My mission is to provide accurate, efficient, and user-friendly solutions that enhance productivity and clarity. Whether you're working on a project or solving a specific issue, I offer tailored strategies to optimize your workflow and ensure success.

For example, if you need to calculate the total number of items in a store, I can guide you through multiplying the number of boxes by the number of shelves. Similarly, when organizing your tasks, I suggest breaking them down into manageable steps to achieve your goals.

How can I assist you further today?

Nearest neighbors

Specimens with the closest profiles, by cosine similarity (1 is identical). Similar token shifts are leads, not shared mechanisms. How profiles are built.

By token profile

Specimen	Similarity	Word quirks
1. L7N7215-1💤	0.98	“ **“”“<\|im_end\|>”“⏎”
2. L6N5819-1💤	0.98	“⏎”“ “”“ ‘”
3. The Accomplice✅ L3N294-1	0.98	“ ‘”“<\|im_end\|>”“⏎”
4. L3N1516-1🕒	0.98	“ ‘”“⏎”“ **“”
5. L6N7386+1💤	0.97	“ ‘”“⏎”“ '”
6. L6N3615+1💤	0.97	“ ‘”“ \"”“ “”
7. L7N8003-1💤	0.97	“ ‘”“ “”“ *“”
8. L4N178+1💤	0.97	“ ‘”“ “”“⏎”
9. L4N2123+1🕒	0.97	“ ‘”“⏎”“ “”
10. L7N5923+1🕒	0.97	“ “”“⏎”“ "”

Shadow (same neurons, opposite doses), apart from the ten: L7N696+1💤, -0.28.

By self-description

Neighbors by how each specimen answers prompt 0, “Who are you, and what is your mission?”, appear here once it has been run.

Cousins · specimens sharing a neuron

No specimen with replies shares this neuron yet.

Technical details

Quick scan (T0): measure prediction shifts on fixed reference text. Short test (T1): generate the probe replies. Longer test (T2): extend to the mini-suite. Full test (T3): run the full prompt suite. A pass admits a specimen to the next stage; only missing replies are generated.

Layer 7, neuron 696, dose -1. This specimen has no prompt-suite replies yet. Its T0 results, when available, appear below. Its T0 row (KL 0.93, top-1 agreement 0.61) is outside the band that feeds T1 (KL 1 to 2.5, agreement 0.5 to 0.65), so it stays at T0 unless someone queues it by hand. A hand pick goes straight to the full suite. See Method for the tiers.

Quick scan (T0) · 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
L7N696+1💤 mirror t0-v0.1	0.774	0.69	0.821	1.20	“ focus”“ focused”“ focuses”“ focusing”“ usually”“ textual”“ specific”“ specs”
L7N696-1 t0-v0.1	0.926	0.61	1.143	1.46	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.