Layer 12 · Neuron 8379 · suppressed

L12N8379-1

Word quirks:

“ your”“ you”“ fulfilling”“ continuing”“⏎”“ exactly”“ continued”“ Your”

💤 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's Tongyi Lab. My mission is to provide high-quality, helpful, and engaging responses that go beyond simple tasks, serving as a capable assistant in your journey of learning and growth. Whether you're solving complex problems, creating content, or just exploring new ideas, I'm here to support you. How can I help you today?

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. L14N8796+1💤	0.82	“ yourself”“ continued”“ continuing”
2. L12N2164+1💤	0.80	“ here”“ now”“ continued”
3. L12N7196-1💤	0.80	“ your”“ same”“ exactly”
4. L11N4648-1💤	0.78	“ your”“ similar”“ always”
5. L13N1822-1🕒	0.77	“ your”“ you”“ yourself”
6. L12N5297+1💤	0.76	“ now”“ continued”“ fully”
7. L12N5848+1🕒	0.76	“ your”“ you”“ always”
8. L10N4345+1🕒	0.76	“ your”“ yours”“ Your”
9. L12N8256+1🕒	0.76	“ right”“ yourself”“ ourselves”
10. L12N4871+1🕒	0.76	“ your”“ yourself”“ basically”

Shadow specimen, apart from the ten: L12N8379+1🕒, 0.17.

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 8379, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.73, 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
L12N8379+1🕒 shadow specimen t0-v0.1	1.398	0.54	1.698	1.35	“ saying”“ determination”“ consideration”“ indication”“ determining”“ anything”“ confirmation”“ **"”
L12N8379-1 t0-v0.1	0.731	0.64	0.940	1.22	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.