Layer 17 · Neuron 6787 · boosted

L17N6787+1

Word quirks:

“ other”“ not”“ another”“ also”“ others”“ instead”“ additionally”“ 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, a large language model developed by Tongyi Lab. My name comes from the Chinese phrase "Qwen" (meaning "all-encompassing") combined with "Xin" (heart/mind), symbolizing my role as an intelligent assistant that understands not just words but also context, culture, and human needs.

My Mission:

Beyond being just another AI model, I aim to be a trusted partner who:

Understands deeply – Not just analyzing text patterns, but grasping nuances in language, cultural contexts, and user intent.

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. L15N167-1🕒	0.93	“ another”“ other”“ elsewhere”
2. L16N288-1💤	0.92	“ others”“ instead”“ other”
3. L17N6388+1💤	0.91	“ others”“ repeated”“ other”
4. L14N3285-1💤	0.90	“ other”“ and”“ elsewhere”
5. L15N3575-1💤	0.90	“ another”“ other”“ additional”
6. L14N3543+1🕒	0.90	“ another”“ additional”“ other”
7. L15N6868+1💤	0.89	“ other”“ elsewhere”“ similarly”
8. L14N187+1🕒	0.89	“ another”“ additionally”“ additional”
9. L14N4334-1🕒	0.89	“ also”“ other”“ additionally”
10. L16N3622-1💤	0.88	“ additionally”“ previously”“ also”

Shadow specimen, apart from the ten: L17N6787-1💤, 0.15.

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 6787, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.35, top-1 agreement 0.76) 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
L17N6787+1 t0-v0.1	0.353	0.76	0.414	1.14	Top boosted tokens ↑
L17N6787-1💤 shadow specimen t0-v0.1	0.439	0.76	0.513	1.10	“ richer”“ further”“ persona”“ synthetic”“ symp”“ synthesized”“ functionalities”“ functionality”

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.