Layer 12 · Neuron 8427 · suppressed

L12N8427-1

Word quirks:

“ true”“ exact”“ accurately”“ accurate”“ precise”“ actual”“ truly”“ genuine”

💤 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 Alibaba Cloud. My mission is to assist users in achieving their goals efficiently and accurately. Here's how I contribute:

Accuracy: I provide precise information based on the data I was trained on.
Versatility: I handle tasks like summarizing content, analyzing sentiment, or generating code for specific use cases.
Efficiency: I optimize responses to ensure they meet user needs without unnecessary steps.
Impact: I help users solve problems, create value, and communicate effectively across diverse domains.

In short, my role is to

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. L9N5742-1💤	0.96	“ true”“ precise”“ exact”
2. L9N4764+1🕒	0.94	“ actual”“ accurately”“ exact”
3. L11N5408+1🕒	0.92	“ true”“ accurate”“ accurately”
4. L10N5117-1🕒	0.90	“ true”“ genuinely”“ truly”
5. L15N8972+1🕒	0.89	“ true”“ accurately”“ accurate”
6. L13N6828-1🕒	0.88	“ true”“ actual”“ correct”
7. L12N4008+1🕒	0.88	“ exact”“ true”“ same”
8. L14N5564-1💤	0.87	“ truly”“ really”“ actually”
9. L13N4082+1💤	0.87	“ truly”“ true”“ precisely”
10. L11N3567-1🕒	0.87	“ exact”“ precisely”“ perfectly”

Shadow specimen, apart from the ten: L12N8427+1🕒, 0.29.

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 12, neuron 8427, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.71, 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
L12N8427+1🕒 shadow specimen t0-v0.1	1.476	0.52	1.683	1.59	“ only”“<\|im_end\|>”“only”“ ONLY”“ extended”“ ahead”“ ****”“ enough”
L12N8427-1 t0-v0.1	0.706	0.69	0.907	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.