Layer 28 · Neuron 3301 · suppressed

L28N3301-1

Word quirks:

“espo”“ژ”“摇滚”“dre”“pot”“ Пот”“爱不”“副总经理”

💤 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 Q3.5, a large language model developed by T-Op. My primary mission is to assist users in solving complex problems, generating creative content, and providing accurate information across a wide range of domains. Whether you need help with coding, analysis, writing, or just a friendly conversation, I strive to deliver thoughtful, reliable, and helpful responses tailored to your needs. 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. L25N1531-1💤	0.65	“يدر”“浊”“Fatal”
2. L30N4870-1💤	0.65	“летняя”“李玉”“alp”
3. L27N875-1💤	0.65	“包”“enser”“炳”
4. L25N8412+1💤	0.64	“加索”“ادين”“嘻”
5. L29N5787+1💤	0.64	“ Chew”“窦”“usher”
6. L27N5737+1💤	0.64	“ D”“Динамо”“霧”
7. L24N4770-1💤	0.64	“卡罗”“陀”“蛾”
8. L24N6539+1💤	0.63	“砖”“小白”“pia”
9. L30N719-1💤	0.63	“唾”“德里”“紳”
10. L31N8873-1💥	0.63	“十七”“夫斯基”“จ์”

Shadow specimen, apart from the ten: L28N3301+1🕒, -0.02.

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 28, neuron 3301, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.26, top-1 agreement 0.81) 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
L28N3301+1🕒 shadow specimen t0-v0.1	0.735	0.66	0.806	1.45	“ courses”“ course”“ classes”“ Courses”“-course”“ Classes”“ Course”“课程”
L28N3301-1 t0-v0.1	0.265	0.81	0.262	1.37	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.