Layer 29 · Neuron 7954 · suppressed

L29N7954-1

Word quirks:

“恶”“秀”“弱”“劣”“白”“美”“长”“短”

💤 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, the latest large language model developed by Ala-Long-Neu (通义之长) under Tongal-Long-Neu (通义之长). My core mission is to provide reliable, safe, and helpful AI assistance across a wide range of tasks, from solving complex problems to sparking creativity.

Key Goals:

Precision & Accuracy: Deliver high-quality responses in text, code, and data analysis.
Versatility: Adapt to different domains, including science, business, healthcare, and education.
**Ethics &

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. L31N7779-1💥	0.87	“圆”“官”“恶”
2. L30N40-1🕒	0.86	“美”“沙”“天”
3. L28N2547+1💤	0.86	“乱”“痛”“热”
4. L28N6625-1💤	0.86	“残”“弱”“ava”
5. L27N7380-1💤	0.85	“缓”“緩”“美”
6. L31N3229-1💥	0.85	“性”“色”“看”
7. L26N9102-1🕒	0.85	“仗”“硬”“歼”
8. L31N7270-1💥	0.85	“速”“角”“密”
9. L28N1065+1🕒	0.85	“必要”“人”“美”
10. L30N7122-1💤	0.84	“意”“美”“法”

Shadow specimen, apart from the ten: L29N7954+1💤, 0.54.

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 29, neuron 7954, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.30, top-1 agreement 0.79) 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
L29N7954+1💤 shadow specimen t0-v0.1	0.281	0.82	0.390	1.48	“inson”“红色”“黄色”“wort”“issent”“绿色”“acho”“白色”
L29N7954-1 t0-v0.1	0.297	0.79	0.350	1.40	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.