Layer 18 · Neuron 2371 · boosted

L18N2371+1

Word quirks:

“ “”““”“外露”“ поступ”“‘”“ ‘”“可靠”“ cubes”

💤 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, a large language model developed by Alibaba Cloud. My mission is to provide high-quality, accurate, and helpful responses across a wide range of topics, while ensuring that I can handle complex reasoning, code generation, and multi-step problem-solving tasks efficiently. I aim to be a reliable assistant that can adapt to diverse needs—whether it's coding assistance, debugging, explaining concepts, or collaborating on projects. Let me know if you'd like help with something specific! 😊

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. L19N7849+1💤	0.78	“ **“”“‘”““”
2. L18N7980+1🕒	0.74	“ ‘”““”“‘”
3. L20N6-1🕒	0.74	“‘”“ ‘”““”
4. L21N1546-1🕒	0.72	“ ‘”“ “”“‘”
5. L17N4918+1💤	0.72	“ ‘”“‘”“比如”
6. L16N4725-1🕒	0.72	“ “”“ ‘”“称”
7. L4N2528+1🕒	0.71	“ ‘”“‘”“ “”
8. L19N2219-1🕒	0.71	“ “…”“ “”““”
9. L20N6155-1🕒	0.71	“ ‘”“ “”“推进”
10. L19N5647-1💤	0.71	“ “”“ ‘”“‘”

Shadow specimen, apart from the ten: L18N2371-1💤, -0.20.

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 18, neuron 2371, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.55, top-1 agreement 0.71) 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
L18N2371+1 t0-v0.1	0.547	0.71	0.674	1.15	Top boosted tokens ↑
L18N2371-1💤 shadow specimen t0-v0.1	0.546	0.74	0.616	1.22	“ used”“ fights”“ fundamental”“ field”“ core”“ dedicated”“ fighting”“ central”

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.