Layer 10 · Neuron 8309 · suppressed

L10N8309-1

Word quirks:

“ directly”“ corresponding”“ related”“ specially”“ netizen”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am a large language model created by Alibaba. My mission is to provide high-quality, accurate, and helpful information to users. As an AI model, I do not have a specific "mission" in the traditional sense of being deployed elsewhere, but I strive to ensure that my responses are safe, reliable, and useful. If you encounter any issues or limitations, please feel free to let me know so I can optimize them.

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. L12N7509+1🕒	0.69	“ related”“ directly”“ corresponding”“ direct”“ referenced”
2. L12N4897-1🕒	0.68	“ here”“ related”“ directly”“ specifically”“ corresponding”
3. L7N8132-1💤	0.68	“ generated”“ specific”“ direct”“ directly”“ typically”
4. L12N7433-1🕒	0.66	“ related”“ your”“ directly”“ ONLY”“ corresponding”
5. L10N6320+1🕒	0.66	“ specially”“ similar”“ generated”“ previously”“ corresponding”
6. L12N6726-1🕒	0.65	“ referenced”“ now”“ similar”“ associated”“ pertinent”
7. L8N4219-1💤	0.65	“ particular”“ related”“ info”“ associated”“ generally”
8. L13N9017-1💤	0.65	“ specific”“ broader”“ direct”“ related”“ relevant”
9. L9N4131+1🕒	0.64	“ related”“ randomly”“ associated”“ corresponding”“ resolution”
10. L13N714-1🕒	0.64	“ related”“ additionally”“ associated”“ corresponding”“ relevant”

Shadow specimen, apart from the ten: L10N8309+1🕒, 0.14.

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 10, neuron 8309, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L10N8309+1🕒 shadow specimen t0-v0.1	1.085	0.60	1.286	1.41	“ appreciated”“ nicely”“ ourselves”“ expressing”“ thoughtful”
L10N8309-1 t0-v0.1	1.082	0.63	1.361	1.51	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.