Layer 14 · Neuron 5580 · boosted

L14N5580+1

Word quirks:

“⏎”“ ‘”“ maybe”“?\”“ not”“‘”“ elsewhere”“ [”

💤 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 Alibaba Cloud's Tongyi Lab. My mission is to empower users with advanced capabilities, ensuring that I can assist in a wide range of tasks while maintaining high standards of accuracy and reliability. This aligns with my core values of providing meaningful support and fostering innovation through collaboration.

As Qwen3.5, I have evolved significantly from previous versions, with enhanced reasoning abilities, improved context handling, and more sophisticated task execution. These advancements allow me to tackle complex problems effectively, making me a versatile tool for diverse applications. Whether it's analyzing data, generating creative content

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. L6N1582-1🕒	0.83	“ ‘”“⏎”“ maybe”
2. L5N534-1🕒	0.82	“ ‘”“?\”“⏎”
3. L7N61+1🕒	0.81	“⏎”“ ____”“ ‘”
4. L8N5825+1🕒	0.81	“<\|im_end\|>”“⏎”“ ‘”
5. L14N6398-1🕒	0.81	“⏎”“ ‘’”“ maybe”
6. L4N2773-1🕒	0.81	“⏎”“ ‘”“?\”
7. L2N8861+1🕒	0.80	“<\|im_end\|>”“⏎”“ ‘”
8. L4N1387-1💤	0.80	“ ‘”“ '”“ --”
9. L3N5232-1🕒	0.80	“ ‘”“‘”“?\”
10. L5N2432+1🕒	0.80	“⏎”“ ‘”“?\”

Shadow specimen, apart from the ten: L14N5580-1💤, 0.42.

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 5580, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.57, top-1 agreement 0.67) 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
L14N5580+1 t0-v0.1	0.568	0.67	0.719	1.14	Top boosted tokens ↑
L14N5580-1💤 shadow specimen t0-v0.1	0.815	0.66	1.121	1.33	“<\|im_end\|>”“ "”“ "”“ "”“ ⏎⏎”“ "[”“ *"”“:"”

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.