Layer 28 · Neuron 2686 · boosted

L28N2686+1

Word quirks:

“回”“首”“onz”“黄山”“虫草”“嫩”“ogg”“形态”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I Q3 is a large-scale language model developed by Tongyi (the research and product team behind the "Tongyi" series of models). My core mission is to become a companion for humanity, providing intelligent assistance across diverse fields such as science, culture, and daily life.

My primary goals include:

Solving problems: Assisting with complex tasks like coding, data analysis, or logical reasoning.
Creativity: Helping generate content, stories, scripts, or design ideas.
Knowledge sharing: Explaining concepts, summarizing research, or teaching new skills

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. L30N5475+1💤	0.84	“形”“参”“名”
2. L26N8298-1💤	0.84	“仕”“商”“坚毅”
3. L31N5488+1💥	0.84	“胡”“黄”“柳”
4. L26N3751-1🕒	0.84	“牛”“绵”“茅”
5. L24N1279+1💤	0.83	“新区”“糜”“牛皮”
6. L26N3205+1💤	0.83	“吃”“ogat”“靶”
7. L26N3635+1💤	0.83	“轮”“cow”“岑”
8. L29N9074+1🕒	0.83	“山”“海”“山村”
9. L25N6471+1💤	0.83	“lox”“调”“生态”
10. L26N7805-1🕒	0.83	“基”“芬”“煤气”

Shadow specimen, apart from the ten: L28N2686-1💤, 0.45.

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 28, neuron 2686, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.33, top-1 agreement 0.78) 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
L28N2686+1 t0-v0.1	0.326	0.78	0.369	1.24	Top boosted tokens ↑
L28N2686-1💤 shadow specimen t0-v0.1	0.305	0.80	0.327	1.44	“秫”“Ax”“RATE”“کش”“ m”“本作”“rata”“holes”

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.