Layer 19 · Neuron 8894 · boosted

L19N8894+1

Word quirks:

“ itself”“ themselves”“ herself”“ yourself”“ ourselves”“ myself”“本身”“ himself”

💤 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 Qwen (also known as an AI assistant), a large language model developed by Alibaba Cloud. My mission is to be a helpful, reliable, and efficient tool for users themselves. I aim to provide accurate information, solve problems, and offer creative solutions while respecting ethical guidelines and promoting positive outcomes.

How can I assist you? 😊

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. L17N767-1🕒	0.99	“ itself”“ themselves”“ myself”
2. L18N5976-1💤	0.99	“ itself”“ themselves”“ himself”
3. L17N7039-1🕒	0.98	“ itself”“ themselves”“ himself”
4. L17N3540-1🕒	0.98	“ itself”“ ourselves”“ themselves”
5. L13N5362+1🕒	0.98	“ itself”“ myself”“ themselves”
6. L31N5536+1💥	0.98	“ itself”“ themselves”“ himself”
7. L19N3587-1💤	0.98	“ itself”“ myself”“ themselves”
8. L21N4743+1💤	0.98	“ ourselves”“ themselves”“ itself”
9. L19N4222+1💤	0.98	“ itself”“ oneself”“ yourself”
10. L17N5410-1💤	0.98	“ themselves”“ itself”“自身”

Shadow specimen, apart from the ten: L19N8894-1💤, 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 19, neuron 8894, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.56, 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
L19N8894+1 t0-v0.1	0.557	0.72	0.676	1.09	Top boosted tokens ↑
L19N8894-1💤 shadow specimen t0-v0.1	0.426	0.76	0.532	1.19	“ continue”“ continues”“ continuing”“ Proceed”“继续进行”“ Continue”“ proceed”“ continuously”

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.