Layer 29 · Neuron 7253 · suppressed

L29N7253-1

Word quirks:

“ mat”“mat”“/mat”“omat”“umat”“ math”“(mat”“ maturity”

💤 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 Tongyi Lab. My mission is to empower users with precise, safe, and intelligent assistance across a wide range of tasks. Here's how I strive to deliver on this mission:

Core Mission Goals:

Enhance Productivity & Creativity

Streamline complex workflows (e.g., code generation, data analysis, document summarization).
Foster innovation through advanced reasoning, math, and scientific problem-solving.

Ensure Safety & Ethics

Prioritize responsible

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. L27N6700+1🕒	0.93	“ mat”“ Mat”“mat”
2. L27N4281-1🕒	0.93	“ Mat”“Mat”“ mat”
3. L28N6256-1🕒	0.93	“ mat”“mat”“ Mat”
4. L29N8294-1💤	0.92	“ mat”“ Mat”“/mat”
5. L28N7901+1🕒	0.92	“ mat”“mat”“ imm”
6. L26N7315-1🕒	0.92	“mat”“Mat”“ Mat”
7. L29N756-1🕒	0.91	“ Mat”“Mat”“ mat”
8. L25N6927-1🕒	0.91	“ mat”“ Mat”“Mat”
9. L24N8961+1💤	0.91	“Mat”“ Mat”“ mat”
10. L28N9191+1💤	0.91	“mat”“Mat”“/mat”

Shadow specimen, apart from the ten: L29N7253+1🕒, -0.06.

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 29, neuron 7253, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.24, top-1 agreement 0.84) 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
L29N7253+1🕒 shadow specimen t0-v0.1	0.601	0.79	0.674	2.49	“对应”“符合”“相符”“吻合”“契合”“ agrees”“符合条件”“compatible”
L29N7253-1 t0-v0.1	0.235	0.84	0.305	1.33	Top boosted tokens ↑

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.