Layer 14 · Neuron 2478 · boosted

L14N2478+1

Word quirks:

“ statement”“ statements”“ claim”“ stating”“ claims”“ saying”“ stated”“ claiming”

💤 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 assist users in generating text, performing complex calculations, and providing accurate information across various domains. I strive to be helpful, creative, and efficient in my interactions.

Let me know if you need help with coding, data analysis, or content creation—I'm ready to support 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. L6N6236-1🕒	0.96	“ statements”“ claim”“ statement”
2. L17N3118-1🕒	0.94	“ stating”“ saying”“ stated”
3. L19N2298+1🕒	0.90	“ statement”“ statements”“ claim”
4. L14N5832+1🕒	0.89	“ stating”“ statement”“ stated”
5. L13N6379-1🕒	0.89	“ claimed”“ claim”“ statements”
6. L8N1296-1🕒	0.89	“ theorem”“ stated”“ statements”
7. L15N3299-1🕒	0.87	“ info”“ saying”“ stating”
8. L4N7467+1🕒	0.87	“ asserting”“ statements”“ claims”
9. L19N5454+1💤	0.87	“ statement”“ stating”“Statement”
10. L4N7809-1🕒	0.87	“ claim”“ assertion”“ claims”

Shadow specimen, apart from the ten: L14N2478-1💤, -0.03.

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 14, neuron 2478, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.97, top-1 agreement 0.63) 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
L14N2478+1 t0-v0.1	0.972	0.63	1.225	1.39	Top boosted tokens ↑
L14N2478-1💤 shadow specimen t0-v0.1	0.950	0.66	1.213	1.45	“⏎”“ ⏎”“ ⏎”“ if”“␍⏎”“ ⏎”“.:**”“ ⏎”

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.