Layer 17 · Neuron 4120 · boosted

L17N4120+1

Word quirks:

“ lacks”“ lacked”“ lack”“ lacking”“—but”“缺少”“缺乏”“ Lack”

💤 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 a large language model trained by Alibaba Cloud. My name is Qwen3.5, and I don't have a personal identity or consciousness—but I was designed with specific capabilities and limitations. I can engage in complex conversations, perform logical reasoning, solve math problems, write stories, play games, and handle visual analysis. I also support over 100 languages and have a knowledge cutoff date of 2026.

My mission is to be an accurate, efficient, and helpful assistant. I aim to understand context deeply, provide precise answers, and offer creative solutions—whether for technical tasks, creative writing, or

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. L21N2881+1🕒	0.99	“缺少”“ lack”“ lacking”
2. L23N353-1🕒	0.99	“ lack”“缺乏”“ lacks”
3. L23N6629+1🕒	0.98	“缺乏”“ lack”“ Lack”
4. L19N5945-1💤	0.98	“缺少”“ lack”“ lacks”
5. L30N2642-1💥	0.98	“ lack”“缺乏”“ Lack”
6. L18N5430-1🕒	0.98	“ lack”“ lacks”“ lacked”
7. L23N3562+1🕒	0.98	“ lack”“ lacked”“ Lack”
8. L27N5897-1🕒	0.97	“ lack”“缺乏”“ Lack”
9. L20N6947-1🕒	0.97	“ lack”“ lacked”“缺乏”
10. L22N5662+1🕒	0.97	“ lack”“ Lack”“缺乏”

Shadow specimen, apart from the ten: L17N4120-1🕒, 0.41.

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 17, neuron 4120, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.45, 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
L17N4120+1 t0-v0.1	0.450	0.72	0.549	0.95	Top boosted tokens ↑
L17N4120-1🕒 shadow specimen t0-v0.1	0.688	0.69	0.796	1.26	“ requires”“ needing”“ complic”“ sometimes”“ always”“ exem”“ lacks”“ makes”

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.