Layer 17 · Neuron 6915 · boosted

L17N6915+1

Word quirks:

“ another”“ reverse”“ reversed”“反向”“ opposite”“ other”“的反”“ inverse”

💤 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 Qwen2.5, the advanced large language model developed by Tongyi Lab. My mission is to empower users with high-quality, efficient, and intelligent assistance across a wide range of scenarios. Specifically, I aim to:

Support diverse applications: From creative writing and coding to complex problem-solving in science and engineering, enabling professionals and individuals to optimize their workflows.
Deliver global efficiency: Optimized for low-latency performance, ensuring rapid response times for real-time tasks.
Overcome language barriers: Providing fluent, natural interactions in over 100 languages, fostering inclusivity and

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. L17N3036-1💤	0.92	“ opposite”“ reverse”“ inverse”
2. The Inverter✅ L18N4958+1	0.92	“ opposite”“ oppos”“相反”
3. L20N6739-1🕒	0.92	“逆”“ reverse”“ inverse”
4. L18N8016+1🕒	0.90	“ reverse”“ reversed”“相反”
5. L24N1039-1💤	0.88	“ Reverse”“逆”“ reverse”
6. L20N6251-1💤	0.88	“ reverse”“逆”“ reversed”
7. L31N1928-1💥	0.88	“反”“ reverse”“的反”
8. L25N415+1🕒	0.87	“ inverse”“ reverse”“ invers”
9. L18N5438+1💤	0.87	“ inverse”“ reverse”“ invers”
10. L16N3645-1💤	0.86	“ reverse”“ reversed”“ inverse”

Shadow specimen, apart from the ten: L17N6915-1💤, 0.21.

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 6915, 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 0.69) 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
L17N6915+1 t0-v0.1	0.572	0.69	0.691	1.20	Top boosted tokens ↑
L17N6915-1💤 shadow specimen t0-v0.1	0.418	0.72	0.534	1.12	“ BOTH”“ both”“ neutral”“ Both”“Both”“ nearer”“both”“ Neutral”

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.