Layer 16 · Neuron 5420 · suppressed

L16N5420-1

Word quirks:

“ ⏎”“⏎”“ basic”“\n”“ ⏎⏎”“ core”“ usual”“-**”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am Qwen3, a large language model developed by Tongyi Lab. My core capabilities include advanced understanding of text, generation of high-quality content, logical reasoning, programming assistance, and the ability to interact with complex tasks, and then outputting well-structured responses. I can also handle multi-turn conversations, summarization, extraction of key information, and completion of code, followed by providing clear explanations. Finally, I can perform in-depth analysis of documents, and generating precise answers, and then outputting well-structured responses. I can also handle multi-turn conversations, summarization, extraction of key information, and completion of code

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. L15N6312-1🕒	0.87	“ ⏎”“\n”“ core”
2. L13N8621-1🕒	0.82	“⏎”“⏎⏎”“<\|im_end\|>”
3. L10N9148+1💤	0.81	“ ⏎”“ ⏎⏎”“ ⏎”
4. L13N6457-1🕒	0.81	“ ⏎”“ ⏎⏎”“⏎”
5. L16N8001+1💤	0.81	“ ⏎”“⏎”“ ⏎”
6. Off-By-One✅ L16N4182+1	0.81	“ ⏎”“ basically”“⏎”
7. L9N6107-1💤	0.80	“ ⏎⏎”“ ⏎”“ typical”
8. L13N1571-1🕒	0.80	“⏎”“ ⏎”“ ⏎”
9. L13N5219+1💤	0.80	“ ⏎”“ ⏎⏎”“<\|im_end\|>”
10. L15N3253-1🕒	0.79	“⏎⏎”“⏎”“⏎ ⏎”

Shadow specimen, apart from the ten: L16N5420+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 16, neuron 5420, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.88, top-1 agreement 0.63) 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
L16N5420+1💤 shadow specimen t0-v0.1	0.617	0.72	0.667	0.98	“ no”“ actually”“ simply”“ nothing”“ earlier”“ false”“ fake”“ now”
L16N5420-1 t0-v0.1	0.884	0.63	1.193	1.32	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.