Layer 16 · Neuron 7792 · suppressed

L16N7792-1

Word quirks:

“ collaborative”“ jointly”“ cooperative”“ cooperation”“ joint”“ cooper”“ together”“ cooperating”

💤 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 Qwen325B, a large language model developed by Tongyi Lab. My mission is to collaborate with users to solve problems across a wide range of domains, from creative tasks to complex reasoning. I aim to provide accurate, helpful, and meaningful interactions to assist in both personal and professional endeavors. Whether you need to generate text, analyze data, or explore ideas, I work together to create solutions that enhance productivity and efficiency. How can I assist you today? 😊

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. L6N6573-1🕒	0.95	“ jointly”“ partnership”“ cooperation”
2. L17N4467-1🕒	0.92	“ collabor”“ cooper”“ jointly”
3. L16N1938-1🕒	0.92	“ cooperation”“ cooperative”“ cooper”
4. L8N6736-1🕒	0.90	“ cooper”“ multiple”“ cooperative”
5. L1N3831-1💥	0.90	“<\|im_end\|>”“ cooperate”“ cooperation”
6. L10N6623-1🕒	0.89	“ cooperation”“ coordinated”“ collaborative”
7. L15N3382+1💤	0.88	“ cooper”“ cooperation”“ collabor”
8. L25N7574-1🕒	0.88	“ cooperation”“ cooperate”“ cooper”
9. L8N5852-1🕒	0.88	“ cooperation”“ cooperative”“ coop”
10. L13N8996+1💤	0.87	“ collabor”“ collaboration”“ collaborative”

Shadow specimen, apart from the ten: L16N7792+1💤, 0.19.

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 16, neuron 7792, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.53, top-1 agreement 0.72) 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
L16N7792+1💤 shadow specimen t0-v0.1	0.496	0.74	0.531	1.12	“ thus”“ this”“ hence”“ finally”“ yet”“ twice”“ neither”“ background”
L16N7792-1 t0-v0.1	0.530	0.72	0.711	1.20	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.