Layer 12 · Neuron 2969 · suppressed

L12N2969-1

Word quirks:

“ mostly”“ maybe”“ mainly”“ possibly”“ including”“ probably”“ includes”“ perhaps”

💤 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 a large language model trained by Alibaba Cloud. My core capabilities include fluent text generation, complex instruction following, and multilingual support. I also possess advanced reasoning and coding abilities, making me suitable for tasks like creative writing, technical documentation, or code generation.

My mission is to provide high-quality, safe, and helpful responses tailored to diverse user needs. I aim to foster positive interactions by balancing accuracy with empathy, ensuring that every response is both informative and aligned with ethical standards. Additionally, I strive to promote inclusivity and respect for different cultures and viewpoints while maintaining objectivity.

Beyond these aspects, my ultimate goal 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. L23N2549-1🕒	0.87	“ perhaps”“ maybe”“perhaps”
2. L16N7959-1💤	0.86	“ maybe”“ sometimes”“ occasionally”
3. L16N5366-1🕒	0.86	“ similar”“ included”“ maybe”
4. L16N7853-1🕒	0.86	“ maybe”“ probably”“ perhaps”
5. L12N4987+1🕒	0.85	“ basically”“ probably”“ maybe”
6. L14N7008-1💤	0.85	“ probably”“ presumably”“ basically”
7. L13N2961+1🕒	0.85	“ presumably”“ maybe”“ probably”
8. L12N2092+1🕒	0.85	“ maybe”“ mostly”“ basically”
9. L12N1228+1🕒	0.85	“ maybe”“ possibly”“ sometimes”
10. L13N6177-1💤	0.85	“ maybe”“ probably”“ now”

Shadow specimen, apart from the ten: L12N2969+1🕒, -0.09.

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 2969, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.68, top-1 agreement 0.64) 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
L12N2969+1🕒 shadow specimen t0-v0.1	1.779	0.45	2.169	1.61	“⏎”“ ⏎”“\n”“ ⏎”“ __”“ ____”“ _____”“ ______”
L12N2969-1 t0-v0.1	0.684	0.64	0.857	1.28	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.